Untitled

SUPERCOLLIDER 4
SUPERCOLLIDER 4
Edited by
John Nonte
Lockheed Engineering and Science Company
Superconducting Super Collider Laboratory
Dallas, Texas
Springer Science+Business Media, LLC

Llbrary of Congress Cataloglng-In-Publlcatlon Data
Supercnlllder 4 I ed'ted by ~ohn Nonte.

p. ca.
·Prnceedlngs nf rhe Fourth InternatIOnal Industrlal Symposlu. On
the Super Colilder. held March 4-6. 1992. In ArlanTa. Georgla·--T.p.
verso.
Includes blbllographlcal references and Index.
ISBN 978-1-4613-6529-7 ISBN 978-1-4615-3454-9 (eBook)
DOI 10.1007/978-1-4615-3454-9
1. Superconductlng Super COlllder--Congresses. 1. Nonte. ~Dhn.
11. InternatIonal Industrlal SymposIum on rhe Supercollider (4th,
1992, Ne" Orleans. La.) Ur. Tltle, Supercolllder four.
OC7B7 • P7S8B67 1992
539.7-3--dc20 92-30221
CIP
Proceedings of the Fourth International Industrial Symposium on Tbe Super Collider,

held March 4-6, 1992, in New Orleans, Louisiana
ISBN 978·1-4613·6529·7
© 1992 Springer Scienoe+Business Media New York

Originlllly published by Plenum Press, New York in 1992
Softcover reprint of the hardcover 1st edition 1992
All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means. electronic, mecbanical, photocopying, microfilming,
recording, or otherwise, without written permission from the Publisher
Preface
The fourth annual International Industrial Symposium on the Super Collider, rrssc held
March 4-6, 1992, in New Orleans was a great success. Present at this year's conference
were 839 attendees representing 24 universities and colleges, 34 states, 13 countries,
17 national laboratories, 11 research centers, many government entities at the local, state and
federal levels, and 235 businesses and companies. This year's symposium also included
101 exhibits by 78 organizations. In all categories, this year's participation exceeded the
totals of previous years and is an example of the growing support for the Superconducting
Super Collider Program.
This year's program had many highlights. One of the best was a message from
President George Bush, read by Linda Stuntz, Acting Deputy Secretary, Department of
Energy. President Bush said that each of us "can be proud of the role that you are playing in
building the Collider and in setting the stage for a new era of research and discovery in high
energy physics."
The 1992 IISSC's theme was "SSC-Discovering the Future." This theme was chosen
in commemoration of the SOOth anniversary of Columbus's voyage of discovery and the
relationship of the SSC with discovery. This theme was articulated by all the speakers in the
opening plenary session. Progress on the program was also very evident at this year's
symposium. In the pictorial session, 66 photographs from all over the world were displayed
to highlight progress in making the SSC a reality. In addition, we had the participation of over
100 local school children at the "Discovering for Education" luncheon, where Nobel laureate
Sheldon Glashow showed how discovery and physics can also be fun. The Congressional
speakers discussed this year's funding challenges and the inputs needed from all of us to
make the program succeed. The quality and diversity of this year's sessions and speakers
allowed the IISSC to meet its goals of promoting a fruitful exchange of ideas among the
SSC's community.
This year's plenary session speakers included:
Andrew J. Jarabak, Westinghouse Electric Corporation

Joseph Cipriano, DOE Project Manager for the sse
Dr. Edward Bingler, Executive Director, Texas National Research Laboratory
Commission
Dr. Roy F. Schwitters, Director of SSCL
Edward Siskin, General Manager of SSCL
George Robertson, Deputy Project Manager of SSCL
Dr. Edward W. Kolb, Institute for Theoretical Physics, University of California
\I
Dr. Fred Gilman, Associate Director, Physics Research Division, SSCL
Dr. Volker Soergel, DESY
Dr. Sheldon L. Glashow, Lyman Laboratory, Harvard University
Linda Stuntz, Acting Deputy Secretary of Energy, DOE
Dr. John Toll, President, Universities Research Association, Inc.
Dr. Alvin W. Trivelpiece, Director, Oak Ridge National Laboratory
Carleton H. Jones, Manager, Research and Development Space and
Telecommunications, Bechtel Corporation
Dr. John B. Yasinsky, President, Power Systems, Westinghouse Electric Corp.
Honorable Tom Bevill, Congressman from Alabama, Chairman of the House
Appropriations Subcommittee for Energy and Water Development
Congressional Panel:
Honorable Joe Barton, Congressman from Texas
Honorable Jim Chapman, Congressman from Texas
Honorable Martin Frost, Congressman from Texas
Honorable James Hayes, Congressman from Louisiana
The IISSC presented its annual award to Senator J. Bennett Johnston of Louisiana,
Chairman Senate Appropriations Subcommittee for Energy and Water Development. The
award was presented to Senator Johnston for his leadership in obtaining Congressional
support for the funding increase obtained in FY 92. Senator Johnston has played a key part
in the SSC's success and is well deserving of this year's award.
The Symposium included a series of parallel technical sessions offering papers on the
following topics: education, superconductors, environment, safety, health, magnets,
detectors, cryogenics, quality assurance, technology transfer, accelerators, and special
sessions on SSC for the Layman and Small Disadvantaged Business Opportunities.
This year's program committee deserves special recognition. It was chaired by
Dr. Thomas Bush, Associate Director, Magnet Division, SSCL. His committee consisted of
the following members from the IISSC Board of Directors:
Mr. Charles E. Anderson

Dr. William R. Frisken
Dr. Eric Gregory
Mr. T. Scott Kreilick
Dr. Paul Mantsch
Dr. Uriel Nauenberg
Dr. Satoshi Ozaki
Mr. Maurice M. Sabado
Dr. Clyde Taylor
Ms. Gully Walter
The program committee's effort was also supplemented by numerous other associates
from government, industry, and national laboratories.
For the fourth consecutive year, Pamela Patterson served as conference manager.
Through her attention to detail and expert management, the symposium's arrangements
proceeded smoothly and without problems. For this she is to be congratulated considering the
large size and scope of the IISSC.
The following companies, organizations, societies, and agencies assisted in producing
IISSC 1992:
vi
ABB Technology Company
Air Products and Chemicals, Inc.
Ansaldo Componenti S.p.A.
Argonne National Laboratory
Armco, Inc.
Babcock & Wilcox Government Group
Brookhaven National Laboratory
C. Hoh Pipe & Tube, Inc.
CRSS, Inc.
CVI Incorporated
EG&G Inc.
E.!. duPont de Nemours & Company
Fermi National Accelerator Laboratory
Furukawa Electric Technologies, Inc.
General Dynamics Space Systems Division
Grumman Space Systems
Hamamatsu Corporation
Hitachi, Ltd.
Hudson International Conductors
IGC Advanced Superconductors, Inc.
Intermagnetics General Corporation
Ishikawajima-Harima Heavy Industries Company, Ltd.
Lawrence Berkeley Laboratory
Lockheed Engineering & Sciences Company
Martin Marietta Corporation
Morrison-Knudsen Company, Inc.
MVE Cryogenics
New England Electric Wire Corporation
Nippon Steel USA, Inc.
Noell, Inc.
NRC, Inc.
Oxford Superconducting Technology
Parsons Brinckerhoff Quade & Douglas, Inc.
Science Applications International Corporation
SiemensAG
Southeastern Universities Research Association
SSC Laboratory
Sulzer Bros, Inc.
Sverdrup Corporation
Teledyne Japan K.K.
Tempel Steel Company
The PBIMK Team
Texas National Research Laboratory Commission
Thevenet et CleIjounie
University of Colorado
U.S. Department of Energy
Westinghouse Electric Corporation
York University
Those listed above provided support in many ways, including financial, in kind,
volunteers, or combinations of all three. This year a total of $44,710 in funds were either
contributed or granted as operating funds for the nSSc.
vii
The 1992 International Industrial Symposium on the Super Collider is a private non-
profit corporation, whose Board of Directors serves as volunteers. The IISSC exists to
promote the SSC program by sponsoring an annual symposium. Attendees can exchange
ideas and learn about the possibilities created by the SSC program in furthering science,
technology, and education. The 1992 Board was composed of the following members from
industry, universities, government, and national laboratories.
Member Affiliation
Charles E. Anderson Air Products and Chemicals Inc.

Mr. Robert Baldi General Dynamics Space Systems
Dr. Edward C. Bingler Texas National Research Laboratory
Ms. Regina Borchard Martin Marietta Science Systems
Dr. Tom Bush SSC Laboratory
Mr. Tony Favale Grumman Space Systems
Dr. William R. Frisken Lawrence Berkeley Laboratory
Mr. Paul H. Gilbert The PBIMK Team
Mr. James H. Richardson Gonzales SSC Laboratory
Dr. Eric Gregory IGC Advanced Superconductors, Inc.
Ms. Phyllis Hale SSC Laboratory
Mr. Andy Jarabak Westinghouse Electric Corporation
Dr. Thomas B. W. Kirk Argonne National Laboratory
Mr. T. Scott Kreilick Hudson International Conductors
Dr. Paul Mantsch Fermilab
Dr. Uriel Nauenberg University of Colorado
Mr. John Nonte Lockheed Engineering & Sciences Co.
Dr. Satoshi Ozaki Brookhaven National Laboratory
Mr. Maurice M. Sabado Science Applications International Corp.
Dr. Giuseppe Scarfi ANSALDO Componenti S.p.A.
Dr. Clyde Taylor Lawrence Berkeley Laboratory
Dr. Robert K. Tener Sverdrup Corporation
Mr. Kuniyasu Toga Hitachi, Ltd.
Dr. William A. Wallenmeyer Southeastern Universities Research Assoc.
Ms. Gully Walter Tempel Steel Company
Mr. Kenneth F. Wilson CVI Incorporated
The Board of Directors met in New Orleans on March 4, 1992, prior to the start of the
symposium, to review preparations for the 1992 symposium and to elect new officers and
members of IISSC, Inc. for the coming year.
President and Conference Chairman Charles Anderson

Vice President Tom Kirk
Secretary Gully Walter
Treasurer Kenneth Wilson
Administrative Officer Phyllis Hale
Program Committee Chairman Regina Borchard
Also, at that time the site and date of IISSC 1993 was chosen to be San Francisco,
May 5-8, 1993, at the San Francisco Hilton.
viii
To assure the success of SSC in the coming years the Board of Directors urges all the
attendees to renew their commitment to this premier research program. By attending the
symposium, we can grow with the SSC, show support, and display to the world all that has
been accomplished. Remember, however, the lessons learned in Columbus's voyage; that
great discoveries occur from events that cannot be predicted. Let us all now set sail on the
ship of SSC in "Discovering the Future."
On behalf of the IISSC's Board of Directors, we thank all those who attended and
contributed to IISSC 1992. We cordially invite all to attend the 1993 symposium in San
Francisco and I look forward to seeing you there. I am certain it will be another memorable
event.
/2:.k~"?~
, Andrew J. Jarabak
Chairman
IISSC 1992
ix
THE WHITE 1-10 E
WA III T N
March 4, 1992
I am delighted to extend warm greetings to all

those who are gathered in New Orleans for the
Fourth Annual International Industrial Symposium
on the superconducting Super Collider.
This impressive gathering of researchers,

educators, public officials, and business and
industry leaders underscores not only the progress
that we are making with the Super Collider but
also the scope and intensity of interest in this
exciting project. Truly, it is an international
undertaking -- one that will benefit people around
the world as it yields greater insight into the
fundamental nature of matter and energy. Each of
you can be proud of the role that you are playing
in building the Collider and in setting the stage
for a new era of research and discovery in high
energy physics.
The year ahead poses great opportunities and

challenges as we work to uphold America's
leadership in this important area of science
and technology. It is critical that we maintain
the momentum and the technical progress that we
have achieved, and I commend you for your efforts
to help ensure that we keep moving forward in the
development of the SSC.
You have my best wishes for a productive

conference.
xi
Contents
A Partnership (Program Status)

Chairman: Thomas O. Bush
Superconducting Super CoIlider Laboratory
Opening Remarks, International Industrial Symposium on the

Superconducting Super Collider
A. larabak .................................................................................................................. 1
SSC Education Initiatives

Chairman: Thomas Gadsden
Superconducting Super CoIlider Laboratory
An Assessment of the SSC Impact on the Training and Employment

Needs of North Central Texas
C. Orsak, P. McGlohen, and L. lenkens .................................................................... 5
Building Blocks of the Universe
M. Bleadon, W. Cooper, T. Delong, P. Fales, R. Lindsay,
E. Malamud, C. O'Connell, C. O'Connor, R. Pasquinelli,
M. Peshkin, G. Snow, M. Stanley, S. Trombly-Freytag, and l. Wisne ..................... 13
SSC Education: Science to Capture the Imagination
T. Gadsden and S. Kivlighn ...................................................................................... 21
Superconductors I
Chairman: John D. Scudiere
Oxford Superconducting Technologies
Summary of the Performance of Superconducting Cable Produced for

the Accelerator System String Test Program
D. Christopherson, D. Capone II, l. Seuntjens, C. Hannaford,
S. Graham, and D. Pollock ...................................................................................... 25
Development of NbTi Superconductor for the SSC by
Sumitomo Electric Industries
S. Saito, T. Sashida, M. Ohmatsu, Y. Yamada,
M. Koganeya, A. Mikumo, M. Nagata, and M. Yokota ............................................ 33
xiii
A Novel Approach to Make Fine Filament Superconductors
H. Kanithi, P. Valaris, and B. Zeitlin ....................................................................... 41
Recent Results in Developing NbTi Fine Filament Superconductors
J. Teuho, T. Erkolahti, P. Haasjoki, J. Lehto,
R. Liikamaa, and V. Vanhatalo ................................................................................ 49
Recent Results in Manufacturing Niobium Titanium Superconductor
for the SSC Project
D. Frost, J. Fernandez, W. Larson, R. Randall,
G. Swenson, J. Wong, and D. Yu .............................................................................. 57
Environment, Safety, and Health

Chairman: Phil Shelley
Superconducting Super ColIider Laboratory
GISIFIS Development for the SSC

A. Oslin and M. Butalla............................................................................................ 69
Successful NEPA Compliance at the Superconducting Super Collider
Laboratory: A Case Study
B. Corning and R. Wiebe .......................................................................................... 79
Archaeology of the Superconducting Super Collider
J. Adovasio, R. Buyce, and D. Pedler ...................................................................... 85
Protection of Prehistoric and Historic Properties at the Superconducting
Super Collider
D. Lowrey and C. Marler ......................................................................................... 93
Use of Pesticides at the SSC: Approval, Reporting, and Monitoring Requirements
W. Berry and W. Pearson ......................................................................................... 99
Medical Surveillance of Employee Health at the Superconducting
Super Collider Laboratory
T. Chester ............................................................................................................... 107
Magnets I
Chairman: Phillip A. Sanger
Review of SSC Dipole Magnet Mechanics and Quench Performance

A. Devred, T. Bush, R. Coombes, J. DiMarco, C. Goodzeit,
J. Kuzminski, W. Nah, T. Ogitsu, M. Puglisi, P. Radusewicz,
P. Sanger, R. Schermer, R. Stiening, G. Spigo, J. Tompkins,
J. Turner, Y. Yu, Y. Zhao, H. Zheng, M. Anerella, J. Cottingham,
G. Ganetis, M. Garber, A. Ghosh, A. Greene, R. Gupta, A. Jain, .
S. Kahn, E. Kelly, G. Morgan, J. Muratore, A. Prodell, M. Rehak,
E. Rohrer, W. Sampson, R. Shutt, R. Thomas, P. Thompson,
P. Wanderer, E. Willen, M. Bleadon, R. Bossert, J. Carson,
S. Delchamps, S. Gourlay, R. Hanft, W. Koska, M. Kuchnir,
M. Lamm, P. Mantsch, P. Mazur,D. Orris, J. Ozelis, T. Peterson,
J. Strait, M. Wake, J. Royet, R. Scanlan, and C. Taylor ........................................ ll3
xiv
A Summary of SSC Dipole Magnet Field Quality Measurements
P. Wanderer, M. Anerella, J. Cottingham, G. Ganetis,
M. Garber, A. Ghosh, A. Greene, R. Gupta, A. Jain,
S. Kahn, E. Kelly, G. Morgan, J. Muratore, A. Prodell,
M. Rehak, E. Rohrer, W. Sampson, R. Shutt, R. Thomas,
P. Thompson, E. Willen, A. Devred, T. Bush, R. Coombes,
J. DiMarco, C. Goodzeit, J. Kuzminski, W. Nah, T. Ogitsu,
M. Puglisi, P. Radusewicz, P. Sanger, R. Schermer,
J. Tompkins, J. Turner, Y. Yu, Y. Zhao, H. Zheng,
M. Bleadon, R. Bossert, J. Carson, S. Delchamps, S. Gourlay,
R. Hanft, W. Koska, M. Kuchnir, M. Lamm, P. Mantsch,
P. Mazur, D. Orris, T. Peterson, J. Strait, M. Wake, J. Royet,
R. Scanlan, and C. Taylor ...................................................................................... 137
Detectors I
Chairman: William Chinowsky
The Engineering Development of an Actively Controlled Precise

Muon Chamber for the SDC Detector
F. Ayer, E. Berk, J. Gorman, J. Govignon, and D. Sullivan .................................... 151
Mechanical Considerations for a SDC Central Straw Tracker for the

SSCL
R. Swensrud, G. Hanson, H. Ogren, D. Rust, D. Marschik,
J. Barkell, D. Hoecker, and J. Mayhall .................................................................. 159
An Engineering Design Study of Detector Deformation Limits in the
SSC SDC-Detector
K. Leung and J. Western ........................................................................................ 167
The Prototyping and Evaluation of Remote Alignment Monitoring
Schemes for SDC Muon System Supermodules
D. Eartly and P. Johnson ....................................................................................... 175
Mechanical Design and Finite Element Analysis of the
SDC Central Calorimeter
V. Guarino, N. Hill, D. Hoecker, T. Hordubay,
J. Nasiatka, D. Scherbarth, and R. Swensrud ........................................................ 183
Cryogenics I
Chairman: Robert L. Powell
Process Systems International, Inc.
Design and Construction of the SSCL Magnet Test Laboratory

Cryogenic Systems
M. Freeman and T. Kobel ...................................................................................... 191
Commissioning and Acceptance Tests of the High Efficient 6 kW
Helium Refrigerator for the LEP 200 Project at CERN
B. Chromec, R. Clerc, W Erdt, D. Giisewell,
K. Kurtcuoglu, K. LOhlein, N. Solheim, U. Wagner,
G. Winkler, and B. Ziegler ..................................................................................... 199
xv
84K Nitrogen System for the SSC
M. McAshan, S. Abramovich, V. Ganni, A. Scheidemantle,
and M. Thirumaleshwar ......................................................................................... 207
Technical Poster Session I

Chairman: Roger W. Coombes
Materials and Processes Selection for SSC Collider Dipole Magnets

K. Agarwal, B. Croll, E. Gilchrist, and D. King .................................................... 219
Utilization of Gamma Ray Inspection System for Tomographic
Imaging and Dimensional Analysis of Complete Model Magnet Cold
Masses and Collared Coil Sections
D. Bein, J. Zbasnik, J. Bolger, B. Firey, B. Rice,
B. Crane, G. Rabaey, J. Wesick, and L. Morgan .................................................. 225
Thermal Performance of yarious Multilayer Insulation Systems
Below 80K
W. Boroski, T. Nicol, and C. Schoo ........................................................................ 235
A Proposed IR Quad for the SSC
S. Caspi, C. Taylor, and A. Wandesforde ............................................................... 243
Magnetic Field Measurements of Fermilab/General Dynamics Built
Full Scale SSC Collider Dipole Magnets
S. Delchamps, M. Bleadon, R. Bossert, J. Carson,
A .. Devred, J. DiMarco, S. Gourlay, R. Hanft, W. Koska,
M. Kuchnir, J. Kuzminski, M. Lamm, P. Mazur,
A. Mokhtarani, T. Ogitsu, D. Orris, M. Puglisi, J. Strait,
J. Tompkins, M. Wake, Y. Yu, Y. Zhao, and H. Zheng ............................................ 251
SSC Collider Quadrupole Cold Mass Design and Development
R. Farrell, F. Murray, P. Jonas, W. Mischler,
and L Blecher ........................................................................................................ 259
The Effect of Vacuum Gas Pressures and Species on Internal Heat
Leak: in the SSCL Magnet Design
D. Franks and R. Pletzer ........................................................................................ 267
Sextupole Corrector Coil Manufacture Using the Direct Wire
Fabrication Process
E. Gossler and C. Coghill III ................................................................................. 275
Support System Design Options for Accommodating CQM Cold
Mass Thermal Contraction
D. Stutzki, K. Grut, M. Hiller, and J. Waynert ....................................................... 281
Design and Construction Experience with Collet End Clamps for 50
rnm Aperture SSC Collider Dipoles
P. Heger, R. Bossert, J. Carson, S. Delchamps, K. Ewald,
H. Fulton, J. Kerby, W. Koska, J. Strait, and M. Wake .......................................... 289
A Cryogenic Support Post for SSC Quadrupole Magnets
M. Hiller and J. Waynert........................................................................................ 297
xvi
Transient Thermal and Stress Analysis of a CQM Cold Mass
X. Huang, Y. Eyssa, and 1. Waynert ....................................................................... 305
Structural Dynamic Analysis of the SSC 40 mm Collider Dipole

Magnet Under Transportation Loads
A. lalloh, E. Daly, and R. Viola ............................................................................. 313
Analytical Study of the Thermal Insulation System in the
Interconnect Region of the SSC Collider Magnets
A. lalloh and D. r "'itchi ....................................................................................... 321
High-Accuracy, Two-Dimensional Magnetic Field Multipole
Calculations for Accelerator Magnets
M. Krefta, H. Chuboy, and D. Pavlik ..................................................................... 327
Magnetic Optimization of the HEB Dipole Magnet End Section
M. Krefta and 1. Parker lr...................................................................................... 335
Transient Cooldown Stresses in Superconducting Collider Dipole
Magnet
K. Leung and G. Snitchler ...................................................................................... 343
Stress Relaxation in SSC 50mm Dipole Coils

D. Rogers and F. Markley ...................................................................................... 351
Theoretical Methods for Creep and Stress Relaxation Studies of SSC

Coil
1. McAdams and F. Markley .................................................................................. 359
Quench Performance of Fermilab/General Dynamics Built Full
Length SSC Collider Dipole Magnets
1. Strait, D. Orris, P. Mazur, M. Bleadon, R. Bossert,
1. Carson, S. Delchamps, A. Devred, 1. DiMarco,
S. Gourlay, R. Hanft, W. Koska, M. Kuchnir, 1. Kuzminski,
M. Lamm, W. Nah, T. Ogitsu, 1. Ozelis, M. Puglisi,
1. Tompkins, M. Wake, Y. Yu, Y. Zhao, and H. Zheng ............................................ 365
SSC Dipole Magnet Manufacturing Facility Implementation
K. Miller and C. Cowan ......................................................................................... 373
Electrodeposition of Copper on the Internal Walls of Colliders
in Beam Tubes
1. Puippe and W. Saxer .......................................................................................... 381
Pressure Relief Radial Passages for SSC
M. Rehak.......................................................................................................... ~ ...... 391
Overview of a Robotic System for Azimuthal Dimensions
of SSC Dipole Coils
D. Assell, 1. Cahill, 1. Carson, D. Connolly, S. Pawlak,
R. Rihel, W. Robatzek, W. Robotham, E. Schmitz, R. Sims,
T. Skweres, and R. Ullmark .................................................................................... 395
Overview and Statistical Failure Analyses of the Electrical Insulation
System for the SSC Long Dipole Magnets From an Industrialization
Point of View
1. Roach .................................................................................................................. 403
xvii
A Precise Technique for Manufacturing Correction Coil
L. Schieber .............................................................................................................. 411
A Design for a High Voltage Magnet Coil Ringer Test Set
W. Koska and R. Sims ............................................................................................. 419
Evolution of the CDM Design
H. van den Bergh .................................................................................................... 427
Mechanical Behavior of Fermilab/General Dynamics Built 15 m SSC
Collider Dipoles
M. Wake, M. Bleadon, R. Bossert, J. Carson,
S. Delchamps, S. Gourlay, R. Hanft, W. Koska,
M. Kuchnir, M. Lamm, P. Mazur, D. Orris, J. Strait,
A. Devred, J. DiMarco, J. Kuzminski, W. Nah, T. Ogitsu,
M. Puglisi, J. Tompkins, Y. Yu, Y. Zhao, and H. Zheng ......................................... 435
Quench Performance of Six Similar 17 m-Long, 40 mm-Aperture

SSC Model Dipoles
P. Wanderer and E. Willen ..................................................................................... 443
Quench Plateau, Coil Cure Temperature, and Copper-to-
Superconductor Ratio in Recent 1.8 m SSC Model Dipoles
P. Wanderer ........................................................................................................... 451
Conceptual Design of the SSC Cryogenic Transfer Lines
B. Zhang and V. Ganni ........................................................................................... 453
Automating the Strain Gauge Data Acquisition of Dipole Magnets
During Keying and Skinning
M. Winters, D. Kihlken, R. Sims, W. Koska, R. Bossert,
M. Davidson, M. Whitson, E. Gonczy, and A. Gilbertson ...................................... 461
Magnetic Measurement Instrumentation Developments at the SSC
Z. Wolf, W. Li, M. Coles, P. Pellissier, J. Dryer,
R. Fuzesy, and T. Gathright ................................................................................... 467
Design and Use of a Mechanical Gage for Measuring Alignment of
Critical Features of String Test Magnets
D. Wolman, G. Brown, D. Pena, R. Trekell, J. Woltz, and
D. Ogilvie ............................................................................................................... 475
Performance of the Magcool-Subcooler Cryogenic System
After 50 mm SSC Dipole Quenches
K. Wu ...................................................................................................................... 483
Quality Assurance I
Chairman: D. Curt Allen
A Statistical Rationale for Establishing Process Quality Control

Limits Using Fixed Sample Size, for Critical Current Verification of
SSC Superconducting Wire
D. Pollock, G. Brown, D. Capone II, D. Christopherson,
J. Seuntjens, and J. Woltz ....................................................................................... 491
xviii
Quality System Design and Development for SSC Superconductor Cable
M. Fuse, S. Saito, T. Sash ida, M. Koganeya, N. Ayai, K. Takai,
and M. Nagata ........................................................................................................ 499
Alternative Alignment Applications for the Collider Dipole Magnet:
Capability Indices
L. Neiman ............................................................................................................... 505
Application of a B&W Developed Computer Aided Pictorial Process
Planning System to CQMS for Manufacturing Process Control
D. Johanson and J. VandeBogart.. ......................................................................... 513
Magnets II
SSC Dipoles
Chairman: Eric Willen
Construction Experience with Fermilab-Built Full-Length

50 mm SSC Dipoles
M. Blessing, R. Bossert, J. Brandt, J. Carson, S. Delchamps,
K. Ewald, H. Fulton, M. Gordon, J. Haggard, W. Higinbotham,
D. Hoffman, R. Jensen, W Koska, M. Packer, R. Rihel,
W. Robotham, R. Sims, B. Smith, D. Smith, J. Strait, G. Tassotto,
D. Tinsley, M. Wake, M. Winters, and W. Zimmerman .......................................... 517
Results From the Fermilab 1.5 m Model Magnet Program
M. Lamm, R. Bossert, J. Carson, S. Delchamps, T. Jaffery,
W. Kinney, W. Koska, J. Ozelis, R. Sims, J. Strait, M. Wake,
and M. Winters ....................................................................................................... 527
Construction and Test Results From 15 m-Long, 50 mm Aperture
SSC Collider Dipole Models
M. Anerella, J. Cottingham, G. Ganetis, M. Garber, A. Ghosh,
A. Greene, R. Gupta, A. Jain, S. Kahn, E. Kelly, G. Morgan,
J. Muratore, A. Prodell, M. Rehak, E. Rohrer, W Sampson,
R. Shutt, R. Thomas, P. Thompson, P. Wanderer, E. Willen,
C. Goodzeit, and P. Radusewicz ............................................................................ 535
Results From Partial Lifetime Test of a 40 mm-Aperture, 17 m-Long
SSC Model Collider Dipole
P. Radusewicz, A.. Devred, T. Bush, R. Coombes, J. DiMarco,
C. Goodzeit, J. Kuzminski, T. Ogitsu, J. Potter, M. Puglisi,
P. Sanger, R. Schermer, J. Tompkins, Y. Yu, Y. Zhao, H. Zheng,
M. Anerella, J. Cottingham, G. Ganetis, M. Garber, A. Gosh,
A .. Greene, R. Gupta, A. Jain, S. Kahn, E. Kelly, G. Morgan,
J. Muratore, A. Prodell, M. Rehak, E. P. Roher, W. Sampson,
R. Shutt, R. Thomas, P. Thompson, P. Wanderer, and E. Willen ............................. 551
Construction and Test Results From 1.8 m-long, 50 mm Aperture
SSC Model Collider Dipoles
J. Muratore, M. Anerella, J. Cottingham, G. Ganetis,
M. Garber, A. Ghosh, A. Greene, R. Gupta, A. Jain,
S. Kahn, E. Kelly, G. Morgan, A. Prodell, M. Rehak,
E. Rohrer, W Samspon, R. Shutt, R. Thomas, P. Thompson,
P. Wanderer, E. Willen, C. Goodzeit, and P. Radusewicz ..................................... 559
xix
Detectors II
Chairman: George H. Trilling
Two-Dimensional Magnetic Analysis of the SDC Muon Toroid

M. Krefta ................................................................................................................ 575
A New Concept of Multiplexed Optical Transmission Readout
Scheme for a Silicon Strip Detector
N. Ujiie, H. Ikeda, and Y. Unno ............................................................................. 583
The Design of the Cerenkov Counter for the SDC Detector
M. Albee, G. Baranko, B. Broomer, M. Christoph, M. Daoudi,
E. Erdos, U. Nauenberg, G. Schultz, J. Smathers, C. Wente,
J. Belz, G. Daniel, C. Guss, S. Kettel, K. McFarlane, and
K. Johns .................................................................................................................. 591
Silicon-Based Tracking System: Mechanical Engineering and Design
W. Miller, M. Gamble, T. Thompson, K. Woloshun, R. Reid,
J. Hanlon, F. Michaud, G. Dransfield, H. Ziock, and
A. Palounek ............................................................................................................ 603
The Solenoidal Detector Collaboration Silicon Detector System
H. Ziock, M. Gamble, W. Miller, A. Palounek, and T. Thompson ......................... 611
Cryogenics II
Chairman: Kenneth F. Wilson
CVI Incorporated
Effects of Mounting Methods on Temperature Sensor Accuracy

Below 10K
T. Kobel, M. Kozyrczak, S. Schwenterly, and W. Bell ............................................ 619
High Efficiency, High Reliability, Improved Oil Injected
Helium Screw Compressor
M. Dreksler and T. Kishi ........................................................................................ 627
Thermal Optimization of the Helium-Cooled Power Leads for the
SSC
J. Demko, W. Schiesser, R. Carcagno, M. McAshan,
andR. McConeghy ................................................................................................. 635
Development of an Ultra Low Heat Leak Cryogenic Valve for the
SSC Tunnel Spool Pieces
E. Baker, C. Rista, and G. McIntosh ...................................................................... 643
Technology Transfer
Chairman: T. Scott Kreilick
Hudson International Conductors
Conceptual Design of a Superconducting MHD Propulsion System

H. Gural, L. Nguyen, and H. van den Bergh .......................................................... 651
xx
Superconductors II
Chairman: Wolfgang Nick
Siemens/lnteratom GmbH
Raw Materials and Early Monofilament Analysis from the Vendor

Qualification Program
J. Seuntjens, V. Bardos, D. Capone II, F. Clark, E. Coleman,
M. Erdmann, and B. Troupe ................................................................................... 661
Cable Keystone Angle Optimization for 40-mm SSC
Quadrupole Magnet Development
D. Capone, V. Bardos, D. Christopherson, F. Clark, S. Graham,
C. Hannaford, T. Reed Jr., and J. Seuntjens .......................................................... 669
Eddy Current Inspection of Superconducting Cable During Manufacturing
D. Bein, J. Zbasnik, S. Graham, and R. Scanlan .................................................... 677
Cold Weld Analysis in SSC Strand and Cable
J. Seuntjens, D. Christopherson, F. Clark, and D. Capone II ............................... 685
Results of AC Loss Measurements on SSC Conductors
G. Mallick Jr., J. Toms, W. Carr Jr., G. Snitchler,
V. Kovachev, and R. Jayakumar ............................................................................ 695
Temperature Dependence of the Critical Current Density in SSC-Type
Superconducting Wires
K. Matsumoto and Y. Tanaka ................................................................................. 703
Quality Assurance II
Chairman: Charles Owen
SSC Quality Program Overview

R. H edderick and D. Threatt .................................................................................. 711
SSCL Magnet Systems Quality Program Implementation for
Laboratory and Industry
D. Warner and D. Bever......................................................................................... 717
Total Quality-A Proven Approach for Magnet Manufacture
C. Owen and K. Malone ......................................................................................... 723
MagnetsllI
Superconducting Magnet Technology
Chairman: James B. Strait
DC Dielectric Breakdown Tests of Liquid Helium

at Temperatures 1.8 to 4.2K
J. Wu and J. Roach ................................................................................................. 731
Testing and Performance of High Temperature Superconducting
Current Leads
J. Wu, J. Dederer, O. Christianson, and S. Singh .................................................. 739
xxi
SSC 50mm Collider Dipole Cryostat Single Tube Support Post
Conceptual Design and Analysis
T. Nicol ................................................................................................................... 747
Welding Vision System for SSC 50mm Yoke-Skinning Production
J. Haggard, J. Carson, J. Jones, L. Nelson, M. Reynolds,
D. Rippinger, W. Robotham, and J. Strait... ........................................................... 757
Axial Variations in the Magnetic Field of

Superconducting Dipoles
A. Ghosh, K. Robins, and W. Sampson .................................................................. 765
Correcting Field Harmonics After Design in Superconducting Magnets
R. Gupta ................................................................................................................. 773
Detectors III
Chairman: Barry Barish
California Institute of Technology
Barium Fluoride Crystal Calorimeter for GEM

R. Zhu ..................................................................................................................... 781
Radiation Damage in Large Barium Fluoride Crystals
P. Li, Y. Xie, Y. Zhao, and Z. Yin ........................................................................... 801
Radiation Damage of Rare Earth Ions Doped Barium Fluoride
Crystals
G. Chen, H. Xiao, S. Man, J. Zhang, and S. Ren. ................................................... 809
Improved Process for Manufacture of Radiation Hard N-Channe1

JFETs for Detector Electronics
L. Rehn and D. Roberts .......................................................................................... 817
Accelerators
Chairman: Ewart Blackmore
Tri-University Meson Facility
Overview of SSC Accelerator Requirements

G. Dugan ................................................................................................................ 825
The SSC Linac: Status of Design and Procurement Activities

L. Funk ................................................................................................................... 833
An Updated Overview of the LEB RF System
J. Rogers, J. Ferrell, J. Curbow, and C. Friedrichs .............................................. 841
Engineering Design Considerations for the Energy Storage Inductors
Used in the LEB Power Supply
D. Pavlik ................................................................................................................. 849
Measured Control Characteristics of the Half-Cell 40 mm Aperture
Magnet String
A. Mclnturff, J. Weisend II, C. Dickey, R. Flora, and
D. Wallis ................................................................................................................. 859
xxii
Technical Poster Session II
Chairman: Shlomo Caspi
The Protection System for the Superconducting Electromagnet Ring

of the UNK
O. Afanasiev, A. Dubasov, A. Erochin, V. Gridasov,
B. Kazmin, O. Kurnaev, K. Myznikov, M. Prima, A. Sytin,
V. Sytchev, G. Tichin, L. Vasiliev, O. Veselov, and N. Yarygin ............................. 867
The Magnet Database System
M. Ball, N. Delagi, B. Horton, J. Ivey, R. Leedy, X. Li,
B. Marshall, S. Robinson, and J. Tompkins ........................................................... 873
The Quench Detector on Magnetic Modulator for the UNK Quench
Protection System
1. Bolotin, A. Erokhin, A. Enbaev, V. Gridasov, M. Priyma,
V. Sytchev, and L. Vasiliev ..................................................................................... 881
Requirements Control System Applied to Collider Dipole
Magnet Development
D. Brouillette .......................................................................................................... 889
A Dynamic Model for Beam Tube Vacuum Effects on the SSC

Cryogenic System
R. Carcagno, W. Schiesser, H. Shih, X. Xu, and A. Yiicel ...................................... 897
A Facility Description of the SSCL Magnet Test Laboratory
M. Coles ................................................................................................................. 905
Transportation Studies: 40 mm Collider Dipole Magnets

E. Daly .................................................................................................................... 913
Experiments to Improve Materials for SSC Magnets
E. Gregory, H. Liu, G. Ozeryansky, M. Sumption,
K. Marken Jr., and E. Collings .............................................................................. 923
Subsidence of the Pit Slab at SLC Experimental Hall
J. Inaba, Y. Himeno, and Y. Katsura ........................................... ........................... 931
Development of Fine Filament Superconducting Wires and
Cables for Accelerators
H. Ii, R. Takagi, S. Meguro, S. Shiga, K. Susai, I. Inoue,
K. Ogawa, T. Suzuki, and M. Ikeda ........................................................................ 939
Development of High Strength Aluminum-Stabilized Superconductor
for the SDC Detector
1. Inoue, T. Suzuki, M. Ikeda, A. Yamamato, and T. Kondo ................................... 943
Connectivity Among Computer-Aided Engineering Methods,
Procedures, and Tools Used in Developing the SSC Collider Magnets
N. Kallas and A. Jalloh .......................................................................................... 957
Control and Data Acquisition System for the B&W Vertical
Dewar Facility
A. Billingsly, J. Beoddy, B. Cantor, P. Hlasnicek,
J. Kelley, G. Pare, and C. Rey................................................................................ 967
xxiii
Current Oscillations in Superconducting Windings
A. Gurevich and S. Kopylov ................................................................................... 975
Evaluation of Technology Transfer on Collider Quadrupole
Manufacture at LBL
J. Boer, H. Fechteler, H. Moryson, F. Sommer,H. Gruneberg,
R. Kreutz, D. Krischel, W. Bensiek, and B. Ryan ................................................... 981
Interim Report on Technology Transfer on Collider Quadrupole
Manufacture at LBL and SSCL
R. Ryan, Jr. and W. Bensiek ................................................................................... 985
A Circuit Theory Approach to Quantifying Thermal Performance of
Cryostat Designs for SSC Dipole Magnets
T. Kupiszewski ........................................................................................................ 993
A Method of Predicting the Reliability of CDM Coil Insulation
A. Kytasty, C. Ogle, and H. Arrendale ................................................................. 1001
Software Design Philosophy for the SSCL Magnet Test Laboratory
J. Lambert, P. Craker, J. Garbarini Jr., J. [vey, W. Li,
C. Saltmarsh, and I. Syromyatnikov ..................................................................... 1007
System Design Overview for the Magnet Test Database System
J. [vey, X. Li, J. Garbarini, and P. Ball................................................................ 1015
Investigation of Azimuthal Stresses Due to Welding and Cooldown
M. LindnerandS. Mulhall ................................................................................... 1021
Design of a Sensitive Calorimeter for Use in AC Loss Measurements
G. Mallick Jr., J. Toms, W. Carr Jr., and H. Cooper........................................... 1025
Temperature Dependence of the Viscoelastic Properties of SSC Coil
Insulation (Kapton)
A. Lautenschlager, F. Markley, J. McAdams, andB. Sizemore ........................... 1031
Measurement of the Elastic Modulus of Kapton Perpendicular to the
Plane of the Film at Room and Cryogenic Temperatures
M. Davidson, S. Bastian, and F. Markley ............................................................ 1039
An Overview of Recent Progress Using Low-Cost and Cost-Effective
Composite Materials and Processes to Produce SSC Magnet Coils
and Associated Non-Metallic Parts
J. Morena ............................................................................................................. 1047
Production, Metallurgy and Utilization of High Manganese
Nonmagnetic Steel-Its Development for the SSC Magnet
K. Nohara, K. Okumura, S. Sato, and H. Sasaki .................................................. 1055
Manufacturing and Magnetic Characteristics of High Purity Ultra
Low Carbon Magnet Steel for the SSC
K. Nohara, U. Nishiike, K. Suzuki, T. Mori, K. Okumura,
H. Terami, andH. Sasaki ..................................................................................... 1065
Dipole Magnet Shuttle System
A. Zinszer, S. Pidcoe, and K. Spann ..................................................................... 1075
xxiv
Current Development of the Superconducting Wires and Cables for
SSC in Hitachi Cable, Ltd.
S. Sakai, G. /waki, Y. Suzuki, H. Nakazawa, H. Moriai, and
K. Kamata ............................................................................................................ 1081
The Cryogenic System for the MTL Magnet Test Stands
Q. Shu, D. Hatfield, P. Reddy, 1. Syromyatnikov, R. Trekell,
and A. Zolotov ...................................................................................................... 1089
A Version of the Radiation-Resistant Fast Calorimeter
Y. Akimov and B. Zalikhanov ............................................................................... 1097
High Resolution Method of the Magnetic Axis Localization for
Multipole Magnets on the Base of the Garnet Films Technology
K. Gertsev, V Gribkov, V. Liskov, and A. Chervonenkis ..................................... 1103
Parallel Technical Sessions IV

Chairman: Paul H. Gilbert
Parsons Brinckerhoff Quade & Douglas, Inc.
Concept Design of the High-Voltage Transmission System for the

Collider Tunnel
L. Norman ............................................................................................................ 1111
SDC Detector Foundation Requirements
1. Western and M. Butalla .................................................................................... 1119
Engineering the SSC Underground
1. Monsees and G. Clark ...................................................................................... 1127
Systems Engineering
Chairman: John Nonte
Systems Engineering at the Superconducting Super Collider

1. Nonte ................................................................................................................. 1137
Implementing Corrective Actions From the Failure Reporting,
Analysis and Corrective Action System (FRACAS) to Achieve
Magnet Reliability and Availability Goals
C. Arden ............................................................................................................... 1143
Risk Management as Applied to Collider Dipole Magnet
Development
R. Patel ................................................................................................................. 1149
Magnets IV
Magnet Fabrication and R&D Facilities,
Tooling, and Component Analysis
Chairman: Clyde Taylor
Summary of Production Development, Details, and Results for Model

Dipoles at SSCL
S. Dwyer, E. Vrsansky, and R. Wood ................................................................... 1159
xxv
Status of the Magnetic Design of the SSC Dipole Magnets
H. Gurol, L Nguyen, D. Bliss, J. Elliott, D. Madura,
G. Mehle, and M. Krefta ...................................................................................... 1167
Alternate Manufacturing Processes and Materials for the SSC Dipole

Magnet Coil End Parts
A. Lipski, R. Bossert, J. Brandt, J. Hoffman, G. Kobliska,
J. Zweibohmer, W. Higinbotham, R. Shields, and
R. Sims .................................................................................................................. 1177
Manufacturing Concepts for the Collider Quadrupole Collared Coils
J. Boer, H. Fechteler, R. Konschak, F. Ostwald, F. Sommer,
and D. Krischel .................................................................................................... 1187
Detectors IV
Chairman: George Yost
Neural Networks as First Level Triggers in Scintillating Fiber

Detectors
J. Orgeron ............................................................................................................ 1193
Performance of a Prototype High Pressure Gas Sampling Calorimeter
L. Demortier, D. Anderson, J. Budagov, S. Cihangir,
G. Fanourakis, N. Giokaris, K. Goulianos, D. Khazins,
M. Morgan, A. Para, A. Volkov, and J. Zimmerman ............................................ 1201
Simulated Performance of a Hybrid Tracking System for the SDC
Detector at the SSC
M. Corden, M. Mermikides, and D. Xiao ............................................................. 1209
A Novel Photomultiplier Tube for Calorimetry at the SSC
P. Cushman, M. Josue, B. Johnson, M. Madden, R. Rusack,
and M. Szawlowski. ............................................................................................... 1217
Attendees ........................................................................................................................... 1225
Author Index ..................................................................................................................... 1237
Subject Index ..................................................................................................................... 1243
xxvi
OPENING REMARKS, INTERNATIONAL INDUSTRIAL SYMPOSIUM
ON THE SUPERCONDUCTING SUPER COLLIDER
Andrew J. Jarabak, Symposium Chairman

Science & Technology Center
1310 Beulah Road
Pittsburgh, PA 15235
Welcome to the Fourth Annual International Industrial Symposium on the

Superconducting Super Collider. In choosing this year's Symposium theme: "SSC--
Discovering the Future" -- we commemorate the SOOth anniversary of the discovery of
America by Christopher Columbus. Columbus made his journey to find a better way to the
riches of Asia; instead, he opened up the treasures of a vast new world. The Super Collider
also has a well-established purpose ... but the extent of the wealth of scientific discovery it
will provide for mankind can hardly be imagined. The Super Collider is a mighty ship ... a
Starship Enterprise .,. that will take us on vast voyages of discovery ... beyond unknown
horizons in the search for new worlds. In a deeper sense, we commemorate all those who,
like Columbus, have set out to explore the unknown and expand our knowledge and
imagination.
In the spirit of this commemoration, I am pleased to report that we have made
significant progress on the SSC in the past year. Progress in our many development
programs. Progress in showing what we can accomplish when government, industry, and
scientific laboratories work together toward a common goal. Progress in manufacturing
critical components. Progress in actual construction ... this progress can be clearly seen at
all the sessions and especially at the pictorial session which is in the hall outside this room.
And progress in a deeper sense ... in a greater understanding and appreciation of the
significance of the Super Collider. I would like to comment on several of these areas this
morning.
Many of you here today ... from government, from industry, from other laboratories,
and from the academic community ... have made important contributions to the Super
Collider program. ¥our contributions will have an impact on all our lives ... provide a
legacy for our children ... advance the well-being of our country ... and, ultimately, benefit
the entire world. I am not exaggerating. In a very real sense ... in a small Texas town, here
in 1992 and in decades to come ... we are discovering the future together. We have
embarked on a breathtaking adventure ... ajoumey ... a voyage of discovery.
One of my favorite books is called The Home Planet It is a collection of photographs
from space with quotes from astronauts from many countries. They have a lot to tell us
Supercollider 4, Edited by J. Nonte

Plenum Press, New York, 1992
about the journey we are embarking on. Charlie Walker, a U.S. astronaut, put it this way ...
"My first view -- a panorama of brilliant deep blue ocean, shot with shades of green and
gray and white -- was of atolls and clouds. Close to the window I could see that this Pacific
scene in motion was rimmed by the great curved limb of the Earth. I held my breath, but
something was missing -- I felt strangely unfulfilled. Here was a tremendous visual
spectacle, but viewed in silence. There was no grand musical accompaniment, no
triumphant inspired sonata, or symphony. Each one of us must write the music of this
sphere for ourselves."
Journey with me for just a few moments ... as we look into the future. It is a truly
exciting trip ... and our adventure of a lifetime. This adventure will take us into the very
heart of science. Unlocking unimagined technologies. And adventuring into a fruitful land
of education.
Our subthemes: science ... technology ... and education. Those are really what the
Super Collider is all about. The Super Collider is the essence of science ... and of scientific
discovery. It will help us explore the origins of the universe and confirm theories of
fundamental physics. The discoveries we make will be profound. Over the last two
decades, powerful particle accelerators, at laboratories such as Fermi and elsewhere in the
world, have already helped physicists achieve startling insights into the evolution of the
universe and the laws that rule the behavior of matter. With the Super Collider, we will be
able to see closer ... and in a sense, better. By boosting energies twenty times higher than
ever before ... we will be able to resolve the subatomic world to better than a billionth of an
inch, to explore distances one thousandth the diameter of a proton ... and investigate the
laws that prevailed when the universe was less than a millionth of a billionth of a second
old. By observing phenomena at significantly higher energies, the Super Collider will help
answer questions about the ultimate structure of matter and about the forces by which
matter interacts.
Science is our first ship into the future ... and scientific discovery of the Super Collider
... is the foundation of our civilization and culture. It is in our hearts and souls ... it's the
next valley ... the far mountain ... the distant stars. Aleksei Leonov summed up this quest
for scientific discovery. He said that "I set out into the unknown and nobody on Earth
could tell me what I would encounter. There were no textbooks. This is the first time ever.
But I knew, it had to be to be done." The riddles that the Super Collider will answer are
bound up with everything else we want to know more about -- space and time, the earth and
stars, and even life and death.
Technology is our second ship into the future. Technology is Science's heir. And
technology is directly linked to our standard of living and the future standard of living of
our children by providing the means by which society prospers. The Super Collider will
have a significant impact on the discovery and implementation of new technologies. On
innovation. On quality. And on the strength of our nation's economy ... today and well
into the future. The Super Collider program offers many opportunities for extending
today's technology into our nation's tomorrows ... in broad applications of
superconductivity, in medical care, in cryogenics, in electronics, and in computers to name
just a few. Technological innovation drives economic growth. This is vitally important to
America's industrial base. It will magnify and multiply opportunities and investments in
technological progress. An investment in technology is an investment in industry. The
Super Collider will keep the United States at the forefront of scientific accomplishment.
And extend our capabilities in new frontiers of industrial achievement.
The third ship of the Super Collider's journey into the future is in education.
Education is our future. The Super Collider is an invaluable tool for learning and gives
future scientists and engineers the ability to solidify ... to actualize ... their theories, their
thoughts, their dreams. The Super Collider is already attracting some of the best and
2
brightest people from around the world. It is an invaluable teaching resource for
universities. Students will learn there by studying and working with experts. But, more
important, it is a passage to hearts ... a key to unlock -- and unleash -- minds.
The United States has found it difficult to inspire young minds to pursue careers in the
sciences. Our young people are avoiding the math and science for the lure of Wall Street ...
the glitz of the sports and entertainment industry ... and the drama of the courtroom. We
must catch their imaginations with the elegance of physics ... the symmetry of math ... and
the rhythms of the natural world. We must show them that solving the mysteries of our
universe can be as thrilling as the adventures of our astronauts ... or the explorations and
discoveries of Christopher Columbus. The Super Collider is critical to our youth for the
motivation it will provide, for the minds it will develop, and for the seeds it will plant in the
generations to come.
This educational imperative is already underway. At the workshops in the
Symposium called Adopt-a-Magnet, you can see some local children adopting magnets
while they learn about science. And at today's Discovering for Education luncheon, you
will be joining 100 local high school students listening to a talk by Sheldon Glashow, a
Nobel laureate in Physics. I have only hinted at the discoveries that await us on our
journey.
Turning once again to the subject of progress, substantial progress has been made on
this journey since our last Symposium. There has been a significant increase in the funding
for the Super Collider. Major contracts have been awarded for the production of the
superconducting dipole, quadrupole, and other magnets. Tests on the magnets have been
very successful. The industry demonstration magnets produced at Brookhaven and Fermi
have performed well. Construction has begun. People are working at the site in buildings
... and many other facilities have been started. Shafts have been completed and, perhaps
you could say, a tunnel finally started. This past year also saw the birth of a new detector
GEM, and the research phase on the SDC was completed. And also ... many magnets have
been adopted!
Progress on our journey of discovery will be evident throughout this year's
Symposium. In fact, the Symposium is a part of this journey of discovery. It provides a
showcase and an opportunity for each of us ... to discover, new realms of science,
technology, and education. In this Symposium, there are a great many plenary, parallel,
and poster sessions you can choose to attend. In the plenary sessions, speakers provide
their unique perspectives on construction progress, technical progress, the science of the
Super Collider, and the value of the research that will occur once we begin operation. The
parallel sessions and poster sessions provide a forum for the presentation and discussion of
a wide range of technical issues.
This year, new parallel sessions have been added, for instance, on educational
initiatives. There is also a panel discussion on The Superconducting Super Collider for the
Layman. The Small Disadvantaged Business session provides an opportunity for smaller
companies to discover ways that they can contribute to ... and hopefully benefit from ... the
Super Collider program. We have over 300 commitments for this session. And tomorrow,
the Government Partnership Session will give all of us invaluable insights into what needs
to be done to sustain the Super Collider program's momentum ... and continue on our
voyage. We have an excellent group of congressional and governmental speakers
participating in this session. One of the finest will be honored at tomorrow's award
luncheon.
Maybe the best tool for discovery is you and me. The Symposium gives us an
opportunity to learn from one another ... at receptions, in the exhibit hall, and during more
casual meetings ... to share, to learn, to grow ... and to discover. I would like to leave you
with these words from Yuri Gagarin, the first man in space. He said ... "And tomorrow?
3
Settlements on the moon, voyages to Mars, scientific stations on the asteroids, contact with
other civilizations ... Let us not grieve that we shall not participate in distant planetary
expeditions. We shall not envy the people of the future. Of course they are lucky, and
things about which we can only dream will be ordinary for them. But great happiness has
come our way, too, the happiness of the first steps in space. Let those who follow us envy
this our happiness.11
The Superconducting Super Collider is an instrument of vast power ... not just by
providing a closer look at the infinitesimal ... but by offering us the opportunity to travel
through time and space ... for roaming through the imagination ... and for accessing the
world on voyages of discovery.
Thank you ... And now ... let's set sail.
4
AN ASSESSMENT OF THE SSC IMPACT ON THE TRAINING AND
EMPLOYMENT NEEDS OF NORTH CENTRAL TEXAS
Charles G. Orsak) Patti McGlohen,2 and Lyssa Jenkens 3
INavarro College, Corsicana, TX

2Dallas County Community College District, Dallas, TX
3North Central Texas Council of Governments, Arlington, TX
INTRODUCTION
The Texas Higher Education Coordinating Board (THECB) has funded the Navarro
College/Dallas County Community College District SSC Technical Training Project to
determine the direct and indirect manpower needs in the eighteen-county North Central Texas
area surrounding the Superconducting Super Collider Laboratory (SSCL) and to identify
training programs to be developed by local community colleges. This needs assessment for
training has (1) revised and expanded regional labor demand estimates and forecasts from
1990 to 2000, (2) disaggregated all regional labor demand estimates and forecasts to the
county level of analysis, (3) estimated the occupational structure of the direct and indirect
employment effects of the SSC, (4) identified regional employers of the occupations
significantly affected by the SSC, (5) inventoried vocational education program completers
for community colleges and technical institutes throughout the state of Texas, and
(6) produced a custom version of the PC-based State Labor Market Information System
(LMIS) to include SSC technical training data.
The tasks of this project were specifically designed to maximize the use of existing
information resources of the various organizations and agencies concerned with labor force
development issues to provide custom databases focused upon the SSCL. The primary
resources utilized in this effort were the databases, methodologies and software developed by
INTERLINK, the regional partnership for education and training in the North Central Texas
area, the Quality Work Force Planning (QWFP) organization for North Central Texas, and the
State Occupational Information Coordinating Committee (SOICC), which is also part of the
QWFP efforts in Texas. The net result of this coordination has been the enhancement of
existing, and the addition of new, labor market information (LMI) resources in North Central
Texas.
The labor market geography used in this study is the combined 16-county North Central
Texas region plus Hill and McClennan counties, which were included because of their close
proximity to the SSCL and their community college and technical institutes' training resources
for the SSCL.l The period of study addresses the ten-yearconstruction-to-operation phase of
the SSCL, assumed to commence immediately for analytical purposes.
IThe North Central Texas region includes Collin, Dallas, Denton, Ellis, Erath, Hood, Hunt, Johnson,
Kaufman, Navarro, Palo Pinto, Parker, Rockwall, Somervell, Tarrant, and Wise Counties.
Supercollider 4, Edited by J. NonIe 5

Plenum Ptess, New York, 1992
COUNTY LEVEL LABOR DEMAND DATA
The labor demand data was disaggregated on the basis of the proportion of the total
regional employment captured by each county as reported in the 1988 County Business
Patterns publication of the Census Bureau. No industry specific or time adjustments were
developed for the individual counties.
Dallas and Tarrant counties account for the vast majority of employment in this region,
capturing 59 percent and 24 percent of the total respectively. These two core counties are
followed by McClennan, Collin and Denton, which account for another 11 percent combined.
The remaining 6 percent of regional employment is spread across 13 counties.
The nine-county SSC region of influence (ROI) captures 90 percent of the 18-county
labor market, largely because both Dallas and Tarrant counties are within the ROJ. The data
for the ROI counties are detailed below in Table 1 for selected years.
TABLE 1. ROI EMPLOYMENT

COUNTY ---TOTAL EMPLOYMENT---
1990 1995 2000
Dallas 1281143 1475274 1676053
Ellis 22291 25670 29165
Hill 5627 6482 7361
Johnson 20560 23679 26898
Kaufman 10820 12462 14162
McClennan 80300 92471 105054
Navarro 13206 15202 17273
Rockwall 5409 6233 7081
Tarrant 515140 593199 673933
TOTAL ROI: 1954496 2250672 2556980
TOTAL NeT: 2164455 2492438 2831643
The location of the SSCL in Ellis County, southward of the dominant counties in this
region, alters the magnitude of its employment effects significantly. Specifically, in the ten
year construction-to-operation period under study, the largest direct and indirect SSCL
employment impacts occur in year four, at 3,893 (Table 4) and 4,499 (Table 5) workers,
respectively. In a regional labor market exceeding 2 million, these figures account for less
than one-half of one percent of total employment. Yet, in Ellis County, with total 1990
employment at approximately 22,000, the direct SSCL jobs in year four alone promise an
increase of nearly 18 percent to total employment. And, to the extend that Ellis and other
outlying counties capture indirect employment, the ebb and flow of labor force impacts of the
SSCL will be increasingly noticeable in localized labor markets.
SSCL DIRECT EMPLOYMENT EFFECTS IN THE NORTH CENTRAL
TEXAS REGION
The direct and indirect employment effects of the SSCL in the North Central Texas
region have been estimated by Robert D. Niehaus, Inc., (RON) in the Environmental Impact
Statements (EIS) for the Department of Energy (DOE) and, most recently, in a 1990 follow-
up study for Argonne Laboratory. RON's methodologies have been effectively approved by
the DOE with their continued acceptance in the Final Supplemental Environmental Impact
Statements, and this project has not encountered any general data resources more timely than
were used in their 1990 updates. The task for this portion of the study, then, was to
disaggregate and localize the data developed by RON into more detailed occupational levels
for both the direct and indirect employment effects of the SSCL for North Central Texas.
RON's 1990 estimates of direct employment are based primarily upon SSCL cost
estimate documents, and the indirect employment figures were derived through an
input/output model. RON's figures address the nine-year construction phase, the three-year
pre-operation period as detailed in Table 2.
6
Table 2. SSC-Related Changes in Regional Economic Activity
-------------Construction year-------------- Pre-operation Year
1 2 3 4 5 6 7 8 9 10 11 12
SSC Related Jobs 3,069 6,348 8,713 9,390 8,011 6,991 6,327 7,093 5,132 4,215 5,047 5,533
Direct Jobs 1,550 2,746 3,740 3,893 3,513 3,296 3,126 3,733 2,641 2,427 2,893 3,158
Construction 824 1,675 2,279 2,262 1,867 1,417 971 653 416 0 0 0
Crafts 199 680 925 903 631 422 241 119 54 0 0 0
Technical 246 483 721 731 707 583 525 394 261 0 0 0
Management, 379 511 633 627 529 411 206 140 101 0 0 0
Clerical
Operation! 726 1,071 1,461 1,632 1,646 1,879 2,155 3,081 2,225 2,427 2,893 3,158
Pre-Operation
Professional 268 388 517 525 511 562 648 981 656 732 909 1,010
Technical 78 112 150 152 148 163 188 284 190 212 263 292
Clerical and other 360 521 694 705 686 754 869 1,316 880 983 1,220 1,356
Visiting scientists 20 50 100 250 300 400 450 500 500 500 500 500
Secondary Jobs 1,519 3,602 4,973 5,497 4,499 3,695 3,201 3,360 2,491 1,788 2,154 2,375
Manufacturing 214 615 815 847 628 468 343 307 207 150 178 196
Transportation, 117 266 358 438 404 383 397 456 363 282 342 380
Utilities
Trade 433 980 1,392 1,538 1,248 1,008 870 921 679 471 565 621
Services 569 1,327 1,836 2,042 1,693 1,392 1,197 1,255 930 655 793 874
Other 185 415 572 631 526 443 394 421 311 230 276 304
Direct employment at the SSCL was based of the 1991 staffing patterns at both SSCL
and Parsons Brinckerhoff/Morrison Knudsen (PB/MK), serving as the primary contractor, as
well as upon any published job announcements, in conjunction with the labor cost estimate
reports for the next ten years generated by the SSCL Project Management Office_
Table 3. SSCL Direct Employment Occupational Structure

I. Management! Administration
A. Managers
1. General
2. Departmental/Functional
B. Financial Specialists
C. Purchasing/Contract Specialist
D. Personnel/Human Resource Specialists
E. Other Professional Support Staff
II. Professional
A. Engineers
1. Electrical
2. Mechanical
3. Construction
4. Other
B. Scientists
1. Physicists
2. Other Physical and NaturalILife Scientists
C. Computer Specialists
1. System Analysts
2. Programmers
D. Other Professional Specialists
III. Technical
A. Engineering Technicians
1. ElectricallElectronic
2. Mechanical
3. Other
B. DesignerstDrafters
1. Electrical
2. Mechanical
3. Other
C. Other Technicians
IV. Clerical and Other Support Staff
A. Secretaries
B. Clerks
C. Other Clerical/Manual Labor
V. Construction Crafts
7
These resources present a fairly clear picture of the general occupational structure of
direct employment at the SSCL, as outlined in Table 3.
The direct employment occupational figures for the ten year construction-to-operation
period were derived by proportioning current staffing patterns across each of the functional
groups within the RDN estimates. Construction and operation data were first estimated
separately, based upon current PB/MK and SSCL employment respectively, and then
aggregated by occupation. The data were processed to the three-digit OES level of analysis
where feasible and to more summary levels of occupational groupings elsewhere. Further,
the relative proportions of each occupational grouping were held constant over time due to no
empirical basis for adjustment The direct employment estimates are detailed in Table 4.
Table 4. SSCL Direct Employment Estimates

. ..-------------------------------------Year-----------------------------------------------
OES Title 1 2 3 4 5 6 7 8 9 10
Code
Management & Administration 541 752 963 965 871 816 718 945 638 619
10000 Managers 213 293 372 371 330 296 237 291 198 177
21100 Fin. Admin. 51 72 94 95 88 88 89 125 84 88
21300 Purchasing 74 103 133 134 123 120 114 157 106 108
21500 Personnel 41 57 73 74 68 66 62 86 58 59
21999 Other Administration 163 227 290 291 263 246 217 285 192 187
Professional & Technical 612 1033 1488 1658 1666 1708 1811 2159 1607 1444
20000 Prof/Tech, NEC 27 52 77 78 76 64 59 49 33 7
22100 Engineers 293 510 731 742 719 660 663 735 490 351
22500 Technicians 103 160 222 225 219 221 241 323 216 212
24000 Scientists, Ex Visiting 117 176 240 243 237 249 278 396 265 278
24000 Visiting Scientists 20 50 100 250 300 400 450 500 500 500
25100 Computer Specialists 52 84 118 119 116 114 121 155 104 95
Clerical & Other Support 205 290 378 381 356 361 367 525 353 374
50000 Clerical/Other Support 205 290 378 381 356 361 367 525 353 374
Construction Crafts 199 680 925 903 631 422 241 119 54 0
87000 Construction, Not Cov. 38 129 176 172 120 80 46 23 10 0
87000 Construction, Covered 161 551 749 731 511 342 195 96 44 0
Total Direct Employment 1550 2745 3740 3893 3512 3295 3127 3734 2642 2427
This portion of the study revealed the two most important immediate results of the
project. First, the vast majority of employment within the construction crafts is covered by
the Davis-Bacon Act which mandates that "prevailing" wage rates be used in federal
construction projects, as determined by regularly updated Department of Labor surveys.
Typically, union wages are found to prevail in many of the crafts found at heavy construction
projects, such as the SSCL. Although this does not mandate unionized labor, there is a
perception that it may effectively result in such. This effect, combined with the significant
surpluses of experienced construction labor existing in the NCT region, indicates that there
will be few employment opportunities for inexperienced, newly trained workers in the
construction of the SSCL.
The second finding is that the SSCL direct employment effects promise to exert pressure
on occupations that are already in high demand in this region. Specifically, significant
proportions of the SSCL direct employment are accounted for by managers, secretaries,
engineers, and technicians. These occupational groups were specifically targeted in the 1990
QWFP efforts as the first, second, eighth, and ninth ranked employment fields in North
Central Texas, respectively, over the next five years, without any specific recognition of the
event of the SSCL. Additionally, the labor demand data developed for this project indicates
that these four occupational groups will remain in extremely high demand in this region
throughout the 1990s, with pressures increasing for engineers and technicians as the decade
progresses.
Each of these occupational groups requires specific training beyond high school, and the
pressures added by the SSCL indicate that even greater attention is warranted in developing
the region's educational and training program capacities. The extent to which this region is
able to produce quality labor in these occupations, especially in the engineering and technical
fields, could well affect the degree to which the SSCL will have any significant spin off
effects in the longer term.
8
EFFECTS IN THE NORTH CENTRAL TEXAS REGION
The occupational structure of regional employment generated indirectly by the SSCL was
estimated using the State of Texas Input/Output ()./O) model developed by the Comptroller's
Office. This is available for QWFP efforts through the State's automated LMI system,
known as SOCRATES.
The 32 sectors of interest represent 130 unique three digit Standard Industrial
Classification (SIC) industries and account for over 70 percent of all indirect employment
effects. The relevant proportions of total annual indirect employment, as estimated by RON,
was allocated across each industry, and the 1995 occupational structures of each were
analyzed. One hundred and eighteen (118) occupations at the five digit Occupational
Employment Statistics (OES) code level of analysis capturing at least five workers in the peak
year four were retained for analysis. This slice in the data represents less than one-tenth of
one percent of annual indirect employment at the occupation level. The remaining indirect
employment was then reallocated across the selected occupations proportionately.
The occupational data was sequentially aggregated by OES code to develop meaningful
groupings in terms of both their numerical size as well as in their skills and training require-
ments. This process yielded twenty occupational groups for the SSCL indirect employment,
at one and two digit OES codes, as displayed in Table 5.
Table 5. SSCL Indirect Employment

- -- --- -- -- ------ -- -- ------ -- Y ear- - ---- - -- ----- -- - -- -- - - --
OES Title 1 2 3 4 5 6 7 8 9 10
10000 Managers, NEC 12 45 60 68 56 45 41 41 23 14

19000 General/All Other Mgrs. 143 325 453 511 400 330 281 304 235 174
20000 Professionalrrechnical 61 137 184 202 167 139 123 130 96 69
30000 Education/Health/Related, NEC 20 47 66 76 60 49 45 46 36 23
32000 Health Occupations 61 126 174 192 158 129 117 119 96 69
40000 Sales and Related, NEC 17 45 68 72 58 45 37 24 27 21
49000 Sales/Related, Direct 324 737 988 1087 898 756 659 701 519 380
50000 Clerical/Other Support, NEC 17 77 116 126 100 81 65 72 39 34
53000 Industry Specific Clerical 25 68 95 101 81 69 60 61 43 32
55000 Secretarial/General Office 93 333 505 571 452 347 292 310 216 133
60000 Service Workers, NEC 52 109 149 161 133 111 95 102 77 55
62000 Private Household Services 74 156 209 231 190 159 140 148 116 82
65000 Food Services 251 555 733 814 673 568 497 523 394 274
67000 Cleaning/Bui1ding Services 54 120 161 177 148 120 106 106 77 62
70000 Agricultural/Farm/Related 25 51 70 76 65 54 50 46 32 25
80000 Prod/Const/Maint/Rel., NEC 42 98 137 146 123 103 84 89 68 50
85000 Mechanics/Installers/Repair 59 139 180 206 169 139 123 121 96 66
87000 Construction Trades 66 147 199 216 183 152 132 143 109 78
97000 Transportation Operators 57 135 201 221 188 141 119 128 89 71
98000 Helpers/Laborers 66 152 224 241 196 159 136 148 102 78
Total Indirect Employment 1519 36024973 5497 4499 3695 3201 3360 2491 1788
Clearly the majority of indirect employment generated by the SSCL in the ten years
under study will be directed towards the household or consumer sector of the economy. The
identified occupational groups performing consumer services directly are health, sales, food
and household workers, accounting for over 40 percent of all indirect SSCL employment.
Typically, these occupations capture approximately 25 percent of the total regional economy.
Further, many of the SSCL's indirect management, professional, clerical and other support
workers will be involved in the businesses providing personal consumer services, and most
of the construction employment will be in the residential sector. The SSCL' s indirect employ-
ment bias towards the consumer sector is further evidenced by the relatively low proportion of
general professional and production workers at less than 7 percent, compared to the regional
figures of over 18 percent.
In the near term, then, the indirect employment effects display an occupational structure
designed to service the household sector. They also suggest that few immediate spin-off
effects, in terms of high tech industries, are anticipated from the SSCL during the 10 year
construction to operation phase.
9
SSCL REGIONAL EMPLOYMENT EFFECTS
The estimated direct and indirect employment generated by the SSCL over the ten year
study period rises from 3,069 jobs in year one to a peak of 9,390 jobs in year four and settles
to 5,132 jobs in the tenth year. The RDN estimates suggest that the stabilized regional
employment attributable to the SSCL at full operation will be approximately 3,200 direct and
2,500 indirect jobs, yielding a total of 5,700.
The short term impact of the SSCL generated employment on the 18 county North
Central Texas labor market is nominal. As previously noted, even the peak year four SSCL
employment effects account for less than one-half of one percent of labor demand in this
region. Further, the general impacts of SSCL direct and indirect employment are equally
small for the nine-county ROt The figures are demonstrated in Table 6 under the assumption
that year one of the SSCL construction is currently underway.
TABLE 6. SSCL EMPLOYMENT WITHIN THE REGIONAL LABOR

MARKET
(in thousands)
YEAR 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
LOCATION
NCT(18 counties) 2221 2288 2356 2424 2492 2560 2628 2695 2763 2831
ROI(9 counties) 2005 2066 2128 2189 2250 2311 2373 2434 2495 2556
SSCL 3069 6348 8713 9390 80ll 6991 6327 7093 5132 4215
AS %OFNCT 0.14% 0.28% 0.37% 0.39% 0.32% 0.27% 0.24% 0.26% 0.19% 0.15%
Even at the county level of analysis, the general impacts of SSCL direct and indirect
employment are negligible except in Ellis. Ellis County, where the SSCL is situated, is a
small local labor market receiving the full benefits of all direct and a disproportionately large
share of all indirect SSCL employment. The county level impacts of SSCL indirect
employment were estimated by RDN, Inc., for all but McClennan within the nine-county
ROI. Their estimates, along with forecasted total county employment figures are presented in
Table 7 for the peak year four, assumed to be 1995.
TABLE 7. ROI COUNTY LEVEL EMPLOYMENT

TOTAL SSCL
EMPLOYMENT INDIRECT EMPLOYMENT
COUNTY (1995) (Year Four)
Dallas 1,475,274 2,994
Ellis 25,670 273
Hill 6,482 46
Johnson 23,679 169
Kaufman 12,462 82
Navarro 15,202 67
Rockwall 6,233 37
Tarrant 593,199 1,829
The employment effects, then, of the SSCL on the various local and regional labor
markets are significant only at the occupational levels of analysis during the initial ten years of
construction and operation. The sole exception to this is in Ellis county, the SSCL's physical
location. The occupations most significantly affected by the SSCL are managers, secretaries,
engineers and technicians. These occupations are already in high demand in this region and
the SSCL promises to exert additional pressures of some consequence.
10
NORTH CENTRAL TEXAS EMPLOYERS
Over 9,500 North Central Texas firms likely to employ workers in occupations
significantly affected by the event of the SSCL were identified for this project. The identified
firms include all regional employers of 10 or more as represented in the 1991 Zeller and Letica
databases, which are made available to local QWFP efforts through SOICC.
Table 8. Occupational Selections

OES
Code Title
22126 Electrical Engineers
22135 Mechanical Engineers
22132 Safety Engineers
22121 Civil Engineers
22199 Other Engineers
22502 Civil Engineering Technicians
22505 Electrical Engineering Technicians
22511 Mechanical Engineering Technicians
22514 Drafters
22599 Other Technicians
24102 Physicists
24111 Geologists
24199 Other Physical Scientists
25102 Systems Analysts, EDP
25104 Computer Programmers!Aides
The covered employers were identified on a very narrow range of occupations affected
by the SSCL, as detailed in Table 8, in an effort to provide the most exact targeting possible.
The OES codes for these occupations were processed through standard industrial to
occupation transformation matrices, yielding a total of 21 two-digit SIC codes upon which the
employer lists were developed. The specific SIC codes used are listed in Table 9.
Table 9. Industrial Selections

SIC
Code Title
13 Oil and Gas Extraction
15 General Building Contractors
16 Heavy Construction. Ex Building
17 Special Trade Contractors
28 Chemicals and Allied Products
29 Petroleum and Coal Products
34 Fabricated Metal Products
35 Industrial Machinery and Equipment
36 Electronic and Other Electric Equipment
37 Transportation Equipment
48 Communications
49 Electric. Gas and Sanitary Services
50 Wholesale Trade - Durable Goods
63 Insurance Carriers
73 Business Services
76 Miscellaneous Repair Services
82 Educational Services
87 Engineering and Management Services
91-93 Public Administration
The employer data contains firm name, location, phone number, employer and SIC
code information and is formatted in Dbase disk files.
VOCATIONAL EDUCATION PROGRAM COMPLETERS

Statewide community college and technical institute vocational education program
completers were inventoried to help assess the training resources and needs as the SSCL
impacts the NCT labor markets. The program completer data was provided by the Texas
Higher Education Coordinating Board for the 1989 program year, which is the most recent
11
infonnation available. The program completers are inventoried by community college district
or technical institute as well as by classification of instructional program (CIP) codes. The
inventories are provided in standard Dbase disk files in the labor supply module of the SSC
version of INTERLINK's LMIS.
There are a total of 57 institutions, including community college districts, technical
institute campuses and extensions, in this database. Only five of the districts are located
within the 16 county North Central Texas region covered by INTERLINK's existing LMIS
databases. This inventory provides the most current infonnation available of the range and
size of the many occupationally-specific vocational education programs operating throughout
the State's community college and technical institute system.
INTERLINK's LMIS FOR THE SSC TECHNICAL TRAINING PROJECT
INTERLINK's PC-based LMIS software shell was adapted to the needs of the SSC
Technical Training Project data to maximize both the access to and linkage between the labor
demand and supply data developed in this study.
The SSC Project version of INTERLINK's LMIS houses the labor demand data
developed for the I8-county combined North Central Texas and SSC regions. The county-
level data with the exceptions of Dallas and Tarrant, as well as the SSCL direct and indirect
employment files, are not processed to an adequate level of SIC or OES code detail to benefit
from incorporation into this software. Instead, they are provided in separate Dbase disk files.
All of the labor supply data developed in this project is fully incorporated into the revised
LMIS software.
One of the greatest advantages of using the LMIS software is the direct crosswalk
provided between the labor demand and supply data in the Occupational Evaluation module.
The module links the occupational level data to both directly and indirectly related vocational
education program completer figures. Further, it provides additional detail on the training,
wage and replacement characteristics of each occupation in the labor demand modules.
There are a variety of other advantages to incorporating this data into INTERLINK's
LMIS software, such as Dictionary of Occupational Title (DOT) crosswalks and inventories
of the linkages between industries and occupations.
CONCLUSIONS
The LMI resources developed in this project provide a strong foundation for examining
the impacts of the SSCL on the training and employment needs of the North Central Texas
region. The direct and indirect effects of the SSCL are analyzed to the smallest level of
occupational detail feasible. Regional labor demand estimates and forecasts were updated and
expanded, county level demand data was developed, statewide vocational education programs
were inventoried and lists of affected regional employers were extracted. All of the data
developed in this project is available in standard Dbase disk files, fonnatted to be compatible
with other LMI resources such as INTERLINK's LMIS. Further, a custom version of
INTERLINK's PC-based LMIS software was developed for the SSC Technical Training
Project to provide the greatest access to and linkage between the labor demand and supply
data developed in this study.
Future efforts to develop additional infonnation resources relevant to the event of the
SSCL must focus upon comparative analyses of similar projects, such as Fermilab and
others, to provide more detail to the products developed in this project. Further, regional LMI
data, such as the labor demand estimates and forecasts, will need regular monitoring and
updating to truly reflect the local conditions surrounding the SSCL.
12
BUILDING BLOCKS OF THE UNIVERSE
Miriam E. Bleadon!.2, W.A. Cooper!·3, T. Dejong!, P. Fales!,

R. Lindsay!, E. Malamud!·2, C O'ConneIP, C. O'Connor",
R. Pasquinelli1, M. Peshkin!·s, G. Snow!·6, M. Stanley",
S. Trombly-Freytag!, and J. Wisne4
! SciTech, Science and Technology Interactive Center
18 West Benton
Aurora, lllinois 60506
2 Fermi National Accelerator Laboratory
3 The Open University, England
4 COSI, Ohio's Center of Science & Industry
280 East Broad Street
Columbus, Ohio 43215-3773
5 Argonne National Laboratory
6 University of Michigan Physics Department
ABSTRACT
SciTech and COSI are collaborating to develop interactive exhibits to convey to a
wide audience some of the fascination of the world of the very small. In both museums,
existing exhibits already offer a rich variety of experiences of nature at the human scale. On
entering this new group of exhibits, the visitor will be led from the human scale to
progressively smaller scales of size, discovering a sequence of new ways that matter is
constituted. Some of the behavior will appear strange to most visitors, for example the
quantum jumps of a huge model atom. But on reaching the level of quarks, it will be found
that high energy physics has revealed that laws of the utmost simplicity underlie the
complications of the human world. The creation of an effective ensemble of exhibits
represents a challenge to our exhibit designers. We have to weigh the value and necessity of
classical analogs of quantum behavior against their inevitable shortcomings. We have to
engender a sense of wonder, but not of mystery. The visitor must feel that the exhibits
represent gateways that open up, not barriers to surmount; that the new worlds discovered do
not undennine common experience, but rather form an exciting extension to it. In the IISSC
symposium two years ago we presented our first ideas. Now the proposed exhibits are in the
final design and evaluation stage, and some are on the museum floor. This work is
supported by a grant from the National Science Foundation and by matching contributions
from several corporations and foundations.
Supercollider 4. Edited by J. Nonte 13

Plenum Press. New York. 1992
INTRODUCTION
The Building Blocks of the Universe Project was officially begun with the receipt of a
grant from the National Science Foundation, in June 1990. The project was entered as a
collaboration between COSI, Ohio Center of Science & Industry, and SciTech, the Science
and Technology Interactive Center, in Aurora, Illinois, in the far western suburbs of
Chicago. This promising partnership is drawing on the years of experience of the COSI
staff, which has led the center into becoming one of the leaders in interactive science exhibit
development. It also takes advantage of the wealth of talent that has amassed around the
newly formed SciTech. In its three year existence, SciTech has drawn interest and volunteer
support from nearby scientific institutions and industries, including Fermi National
Accelerator Laboratory, Argonne National Laboratory, AT&T, Bell Labs, and Amoco
Research Center, to name a few. Equally satisfying is that this support has been balanced by
a great deal of volunteer work by interested members of the public. The goal of the effort is
to assist in the quest to promote science literacy, and to produce a set of interactive
experiential science exhibits, detailing the world of high energy physics. It is focusing on the
peculiar realities of atoms, subnuclear particles, quantum mechanics, and the fundamental
principles that cooperate to tie our understanding of the cosmos, the quarks, and the coffee
table all together into a single construct.
Who needs exhibits on the basic mechanisms of the Universe? Our experiences are
naturally based on the macroscopic - that which we can see, feel and manipulate. These
experiences generate but also limit our concepts, and this applies as much to the formalized
concepts of a scientist as the intuitive concepts of others. In general, we are unfamiliar with
the workings of the microscopic, hidden behind a cloak of ignorance. However, it is
precisely this world of the microscopic which defines our own, gives it its rules and
exceptions, creates its possibilities and probabilities, and holds the secrets of the universe in
which our existences are immersed. To break through to the microscopic world we -
scientist and non-scientist alike - need new concepts, freed from the limits of our immediate
experience. The Building Blocks of the Universe exhibition seeks to bring everyone at least
to the gateway of a world which otherwise exists only in a guide book written in
mathematics.
Few other science centers or educational institutions had attempted the formidable
task of exposing this underlying world at the time the grant was issued. Present day physics,
which seeks to explore the Big Bang, the top quark, the Higgs boson, and other building
blocks seemed too complex to explain and describe in terms that an eleven year old could
understand, not to mention the non-physicist adult. However, through all the calculus,
perturbation theory, and quantum chromodynamics, we must strive to find simple analogs
and solutions, making our ideas easily communicated and understood. Recall the plight of
the astronomer Tycho Brahe. He knew the period and path of the planets across the evening
sky, but his explanation of a geocentric world with planetary epicycles and unforgiving
mathematics, was incredibly complex. He had his facts right, but he drew the wrong
conclusions. It was left for Copernicus to introduce the proper explanation, pointing towards
a heliocentric system, greatly easing the mathematics and most importantly, easing the
conceptualization. We now know who was right, and we must always remember this
lesson. No matter how she appears, Mother Nature is usually simple. It is our challenge to
unravel the perceived complexities, to fmd the hidden truth and beauty.
Most conventional science museums offer a historian's view of science and
technology. We are walked through Pascal's mathematics engine, Edison's first light bulb,
and other significant artifacts and documents, detailing our civilization's progress through the
last centuries. These museums offer a unique perspective, affording the opportunity to view
how we once lived, what theories were previously accepted, and what inventions and
innovations changed society. However, these artifacts are singular, and can exist in only one
place at a time. There are many communities that are inaccessible to the reach of these
historical stores.
Interactive science centers offer the visitor a chance to recreate these events and
manipulate the models and objects for himself. Doing exactly these things greatly
complements our learning and comprehension. These centers are not limited geographically.
They can be created in any community, aiding in the quest for science literacy. Indeed, they
have been developing around the country, and the world. A hands-on environment where
we are allowed to create, invent and discover like the scientists who preceded us, is what
14
they seek to provide. SciTech, COSI, and some 300 science centers nationwide drawing
over 50 million visitors a year, are all members of the team, providing places and tools that
take advantage of our natural curiosity, where we are allowed to ask "Why?", and where we
are allowed to answer for ourselves.
This exhibition will provide an introduction to the current understanding of matter,
energy, and interactions. It will provide a gateway from the known to the unknown.
Indeed, it will provide the proper historical perspective, detailing the ideas of Einstein, Bohr,
Curie, and others. First and foremost, however, it will provide a nurturing interactive
atmosphere, that will allow the individual to go as far as desired, limited only by his own
curiosity and imagination.
THE EXHIBITS
As originally proposed, the exhibition was to include 12 major learning stations,
where the subnuclear world would unfold. All were scheduled to be completed in a 2 year
time period. This appears to have been an overly optimistic estimate. Our resources were
severely challenged, and manpower shortages forced a reorganization of the exhibits. Some
were dropped due to financial constraints, others were added to help tie the exhibition
together. The present 12 stations are
An Investigation into the Very Small
Cutting the paper
Projection Microscope
The Atom
Rubber Ball Nuclei
Valley of the Isotopes
Quark Machines
Quarlc machine
Antiquark machine
Meson scoreboard
Particle Smasher
Half Life Cluster ( 3 stations)
Work on these exhibits has been progressing steadily, and all are in final production. It is
hoped that two of the exhibits that were left out of this list, "Matter into Energy", and
"Energy into Matter", will be spun off into an exhibition of their own, dedicated to Einstein
and his famous formula: E= mc 2.
AN INVESTIGATION INTO THE VERY SMALL

The notion of "small" generally conjures up pictures of paper clips, a strand of hair, a
sliver of cake, or a grain of salt. To those of us who are fortunate to have used a
microscope, those images might be extended down to mitochondria and proteins. This
exhibit will challenge the participant to probe even deeper into those images, exposing each
successive layer of "smallness", one by one.
The participant is offered a pair of scissors and a piece of paper, and challenged to
divide it in half as many times as possible. Most people think that dividing something in half
is trivial, and can be done numerous times if desired. This part of the exhibit gives an
appreciation of how fast factors of two multiply, as people realize that they can barely cut the
paper ten times. A magnifying glass or a simple microscope might help, but then what? The
participant, still able to hold the smallest piece of paper easily in his hands, surely feels that
there is still something smaller. At this point, he is invited to the next station, a projection
microscope, where matter can be further investigated. Pictures of the paper down to the next
few factors, are available, as are pictures of other common objects not normally viewed at
that magnification. The participant is then led through a thought experiment, imagining what
might be smaller. Is there some smallest unit that we can still call paper? Given the right
15
tools, can that smallest unit be divided into yet smaller non-paper things? Can those be
divided further? How far can we go? Is there some fundamental limit?
The goal of this exhibit is to set the stage for the entire exhibition, and to slowly draw
the visitor into the world of sub nuclear realities. It starts out with the things that he knows,
and can touch, see and manipulate. It then provides an introduction, a gateway of sorts, into
the microscopic universe, as the layers of mysticism are peeled back, one after another. It is
hoped that it will put the visitors into the proper frame of mind and mood to understand and
enjoy the exhibits which will follow.
THE ATOM
Hydrogen is the simplest atom that is possible. It consists of a single proton in the
central nucleus, around which dances a single electron satellite. Unlike the macroscopic
world, there is no energy continuum within the atom. The electron is not allowed to have
any energy it wants. Instead, the bound electron is .limited to have only certain allowed
energies. An electron with a given energy dances around and is found mostly near a
spherical shell that depends on the energy. As the energy level increases, the average radius
of the shell increases, and the electron dances further from the nucleus.
Three concentric plexiglass spheres, each covered with an array of red light emitting
diodes ( LEDs ), are used in our model of the Hydrogen atom. Each sphere represents a
single allowable energy level, and each light represents a possible position for an electron.
Our three spheres have diameters of 4, 12, and 30 inches, making our model of the atom one
billion times the size of a real Hydrogen atom. With but one electron, only one light can ever
be illuminated at a time. But, since the electron is never stationary, this point of light flickers
on and off randomly around a particular sphere. The flashing is done quickly so that the
observer can never zero in on the electron position, and sees only a cloud of light where the
electron might be found. Only the electron position is changing, not its energy.
Figure 1 • Atom
While the electron is in one of the lower energy levels, the participant is asked to
excite the atom into one of its higher energy configurations. There are three allowable
excitations: inner to middle, middle to outer, and inner to outer. Each transition is made
possible by the participant imparting energy of a characteristic frequency. The visitor is
asked to whistle. If the note matches one of the three allowable frequencies, then the electron
jumps to the indicated higher level. The three notes are roughly 1 KHz, 1.5 KHz, and
2 KHz. Musically, those are the opening notes to the theme from the popular movie, "2001:
A Space Odyssey." After some time, the electron is allowed to relax back down to a lower
energy level. Again, there are three possible transitions, each with a characteristic frequency.
A relaxation event is indicated by the automatic firing of one of three colored strobes,
positioned inside the nucleus. The blast of light represents the single photon that would be
released by the atom for such an event.
There are four major goals for this exhibit. First, it conveys the concept of quantized
energy levels. Second, it shows that the bound electron is never stationary, and that its
position at a particular energy is uncertain. Third, it demonstrates that it is the frequency
16
which detennines the energy release or absorption of a transition. Finally, it shows that
energy is radiated from the atom only when an electron drops into a lower level.
RUBBER BALL NUCLEI

It was the famous scattering experiment by Ernest Rutherford that led to the
conclusion that atoms had a central core where positive charge was concentrated. We now
know that this central nucleus itself has structure. It can be made up of a combination of
protons and neutrons. This exhibit describes the present understanding of how protons and
neutrons interact to fonn a tightly bound nucleus within each atom.
Hard rubber balls, about 2 inches in diameter represent the individual nucleons.
Those that are dyed red are protons. Those that are dyed blue are neutrons. The nucleons
are held together by a piece of clear plastic tubing, extending from the center of one ball to the
next. A plastic green bead rolls back and forth within each tube. It represents a meson, the
messenger particle that is exchanged between the two objects, delivering the attraction of the
strong nuclear force, keeping them bound together. It is important to note that the bonds are
not rigid and fixed. Instead, they are rubbery and flexible, allowing angles and distances to
change slightly, providing a more accurate representation of the atomic nucleus. This exhibit
.....
can be easily picked up, twisted and bounced. It is very appealing to the younger visitors.
NEUTRO II
DEUTERIUM PROTON II
• BlNDlNG
ENERGY
HELIUM
Figure 2. Rubber Ball uclci Figure 3· Valley of the Isotopes
VALLEY OF THE ISOTOPES

In general, when we look at a Periodic Table we see a listing of the elements, with an
indication of the number of protons and the number of neutrons found in its most stable
fonn. What we do not see is a listing of all of the other forms of the same element, which
would have differing neutron numbers. These other forms are called isotopes, and each
element has several, but the Periodic Table has no way to indicate all of these. A different
type of table is needed.
A map of the isotopes is usually three dimensional. Proton number is along the x-
axis, with neutron number along the y-axis. The third dimension represents the binding
energy of the isotope. Standard wall charts, unable to use 3 spatial dimensions, choose their
third dimension to be mapped as a color, or a shade on the grid. However, on the museum
floor, we have no such restriction, and we are free to build in x, y and z, as we please.
Therefore, we have chosen to translate the binding energy into a depth along the z-axis. As
binding energy increases, the depth in z increases. This is similar to digging a hole. The
deeper the hole is, the higher the side walls are, the greater is the amount of energy needed to
get out.
As the table is mapped, a central depression is formed, with high walls on either side.
This is called the Valley of the Isotopes. The entries which lie towards the bottom of the
valley have high binding energies and are therefore stable. The entries that are on the high
ridges, have smaller binding energies, making them unstable and prone to decay. For
instance, Iron-56 is the most stable of all isotopes, and is found at the lowest point in the
17
valley, while Sodium-22 is precipitously balanced on the ridge, a prime candidate for
radioactive breakdown.
There are thousands of known isotopes, and to have a functional and easily
maintainable exhibit for all of these would require a Herculean effort. Instead, we have
chosen to reproduce a section of the valley. Our initial effort includes all isotopes through to
Vanadium. The x-y grid is made with lights, mounted on circuit boards, and controlled by a
computer. A lucite light pipe, 5/8" square, is mounted over each light. The pipe height,
representing the binding energy, ranges from roughly 6 to 34 inches. The color of the light
represents the isotope's method of decay. Green is for stable, red is for positron emission,
yellow is for electron emission, and orange is used for the few isotopes that can decay either
way. The result is a beautiful and dynamic light sculpture.
The computer program invites the visitor to explore fusion processes, flashing the
appropriate lights as a guide through the narrative. This can be programmed to explain stellar
evolution and how our sun burns Hydrogen to make Helium and energy. It can detail
Carbon dating, and the behavior of some medical isotopes. In addition, the chemical patterns
shown in the lines of the Periodic Table are not lost is our display. In fact, they show up in
lights when requested via the computer. We anticipate that the Valley of the Isotopes will be
a tremendous teaching tool, and that it will be programmed by interested institutions in ways
that we have not yet imagined.
QUARK MACHINES AND MESON SCOREBOARD

All matter of which we are commonly aware, including all elements in the periodic
table, all nuclei, and all isotopes in the valley, can be made of three things: protons,
neutrons, and electrons. We further know that protons and neutrons are composed of up and
down quarks. The goal of this exhibit is to show that there are actually a total of six kinds of
quarks ( 5 already seen experimentally), and six kinds of antiquarks. The exhibit will also
show that there are many other particles that can be created through their various
combinations.
The basic component of these exhibits resembles a three-wheel slot machine. A color
computer screen is used to simulate the wheels and the spinning motion. An arm will be
constructed to mimic the look and feel of a true slot machine, and it will trigger the computer
program to set the wheels in motion. However, instead of lemons, cherries, and Lucky 7's
popping up, quarks will appear, and baryons will be formed. In the case of the antiquark
machine, antiquarks will appear, and antibaryons will be formed. The machines can be
played, side by side, by two competitive would-be physicists, and the resulting baryons can
be compared. Or, the two machines can be played together, in an attempt to create mesons,
which are formed from a quark and an antiquark. A scoreboard will keep track of all
interesting mesons created.
The probability of a particular quark appearing on a wheel is based on its existence in
the real world. That means, it is most likely that an up or down quark will appear, and that a
-
ORDINARY _
STRANGE _
CHARMED
- MESONS
TOP
BonoM_
t'Hoj tl,"la>j
PROTON
....
ANTIPROTON
/'
Figure 4 - Quark Machines and Meson Scoreboard
18
proton or neutron be formed. After that, there is a slim chance that either a strange or a
charm quark will appear. Finally, there is a tiny remote chance that a bottom or top quark
appears. Indeed, it would be quite an event if the top quark appeared, as it is the only one
that has not yet been seen experimentally. Perhaps the CDP and DO experiments at Fermilab
will have verified its existence by the time this exhibit is completed. The race is on.
PARTICLE SMASHER
The Particle Smasher is the next station in the quest to understand matter, and its
natural structure. By the time this exhibit is reached, the concept of order of matter has
already been seeded. An Investigation into the Very Small broke matter into molecules. The
Atom led us to the next level. The Rubber Ball Nuclei and the Valley of the Isotopes then
allowed us to peer into the atom, revealing the nucleus, protons and neutrons. The Quark
Machine further broke up these particles, and even showed what other ones might be
possible. The goal of the Particle Smasher is to bring all of these concepts under a single
umbrella, showing that there is some overall hierarchy to matter.
Built as a simple and familiar video arcade game, the screen of this exhibit presents
the participant with an array of molecules, moving and bouncing all around. Simple
molecules such as H20 and NaCI are used. These can be broken down into atoms, nuclei,
protons, neutrons, electrons, and finally quarks, with successive hits scored by the
participant, through the firing of a high energy particle beam. The player-physicist must try
to create fundamental objects such as quarks and electrons, while avoiding being struck by
the fast moving particles.
Early prototypes of this exhibit have been on the SciTech and COSI floors for over a
year. One was even at this conference last year, where we had a booth. In all cases, it was
tremendously popular among children, adults, members of Congress, and even physicists,
regardless of their age.
~
1'Z f-
r-
079500
• quark e -
Ij ~
Iii
-Tl
~
® ® ~~
q~ark
. ,
,
NaCI
quark '
©
~fEJf
I I L
Figure 5 - The Particle Smasher Figure 6 - Exponential Decay Gameboard
HALF LIFE CLUSTER

There are three stations which provide models for radioactive decay. The first station
is a simple board game. The participant is handed a container of six-sided dice. One of the
sides on each die is painted a distinctive color. In the first tum, the dice are spilled onto the
playing surface. All dice with the colored side up are removed from the group, and stacked
one upon the other on the board. The second tum is the same, with the remaining dice being
spilled, and the few with the colored sides being stacked, next to those already present. This
cycle continues until all dice have been stacked. The result is a graph, made by the individual
stacks, that represents the "decay" of the dice. With each toss, 1/6 of all dice decay. Starting
with 100 dice, the decay stacks might be as follows:
16,14,9,11,8,7,5,3,4,5,3,2,1,2,1,3,1...
19
A precocious participant might choose to start with only 50 of the original dice. He would
discover the the shape of his curve is unchanged, but its magnitude has decreased.
There is a similar game board, next to the first, but at this station, two of the six sides
are colored. With each toss, 1/3 of all dice decay. Again, it doesn't matter how many dice
are thrown on the first toss. The shape will always be the same, only the magnitude
changes. It is hoped that by having the two games juxtaposed, we reinforce the concept that
the shape of the decay curve is a fundamental characteristic of the material, and it has nothing
to do with how much of the material is actually present. It also introduces the concept of an
exponential series, if the player was not yet familiar with it.
The third station stresses these same concepts, and ties them in with radioactivity, by
introducing time and rate. The participant is allowed to change a decay rate, and to view the
results. A computer program drives graphics on a large color monitor. On the upper half of
the monitor, pixels are grouped representing atoms awaiting decay. The bottom half of the
monitor graphs the decay as it occurs. At the start of the sequence, all pixels on the upper
half are red representing non-decayed atoms. The participant is asked to enter a decay rate.
The computer begins to continuously "toss" the atoms, using the decay rate to determine
which ones "spill" with a colored face up. On the screen, the colors of these decayed atoms
are changed to green. The appropriate stacks are displayed on the graph below. This
continues until all atoms are green and have decayed. The user can also choose to enter a
total decay time. In this mode, the computer plots the decay curve for the indicated time.
The resulting screen shows how many atoms are left, and how may are decayed.
These three stations, taken together, should provide an introduction to radioactive
decay, and demonstrate the real world use of an exponential series. Accompanying text will
explain how this simple principle is used in Carbon dating and a variety of other applications.
These stations could do just as well in a mathematics exhibition, however, we use them here
to demonstrate a key concept in physics. This simple law of decay is natural for the atom and
the nucleus, but nothing in our everyday experience could prepare us for its explanation. The
behavior of an atom now is totally independent of all its previous existence. Everything we
see around us wears out from day to day, but the atom starts each half life as fresh as the day
it was born. The visitor has seen the randomness which is so characteristic of quantum
behavior simulated in the atom, and reflected in the quark machine. Here, it is epitomized in
a single exponential line.
CONCLUSIONS
People tend to learn best when all of their senses participate in the experience. Seeing
things on a blackboard, and hearing a narrative, are not enough. We need to touch. We need
to smell. We need to taste. We need to manipulate. Each of these stimuli gives us a new
piece of information, that might not have been retrieved by simple sight and sound. Our
senses, if allowed, can cooperate to form a more robust impression, which will hopefully
lead to fuller comprehension. It is also very likely that the deeper the sensory impression, the
more probable it is that we can recall it, and apply it.
Consider a Montessori-like method for teaching numbers. Not only can children be
taught to see numbers, hear numbers, and write numbers, but they can also be taught to feel
the numbers, by cutting them out of sandpaper, and stroking their fingers over the shapes.
The more senses that are involved, the deeper is the comprehension. In a similar vein,
interactive science centers attempt to involve the whole child in the learning process. There is
no abstraction allowed. There are no black boxes, no magic, no mysticism, no enigmas.
Everything is laid out in front of the child, and as many senses as practical are allowed to
participate.
In an age where technology has crept into the common workplace, into the home, into
the medical office, and into politics, it is critical that the populace be science literate.
Interactive science centers, historical science museums, industry, and the government must
all assist in the quest to educate our young. Our future might be in their hands, but at the
present time it is we - the educators, scientists, researchers, industrialists, engineers, and
politicians - who have the responsibility to make sure that the present remarkable age of
discovery is continued into the future, and does not stagnate because there is nobody left to
carry the torch. Not only must we pass our knowledge to the next generation, but we must
prepare those young minds to be able to use that knowledge, and to run with it in ways we
would never have conceived in our wildest dreams.
20
SSC EDUCATION: SCIENCE TO CAPTURE THE IMAGINATION
Tom Gadsden and Sherrie Kivlighn
Superconducting Super Collider Laboratory"

2550 Beckleymeade Avenue
Dallas, TX 75237
To the great majority of Americans, science is merely a collection of facts and theories
that should (for unknown reasons) be memorized and perhaps even understood in order for
one to function as a responsible citizen. Few see science as a way of thinking and questioning
and as an approach to learning the secrets of our world. In addition, most children and many
adults have a stereotypical view of scientists as studious men in lab coats who spend all their
time working alone in dark and smelly chemical or biological laboratories. The
Superconducting Super Collider (SSC) totally contradicts such a perception. This great
instrument is being created by thousands of scientists, engineers, business people,
technicians, administrators, and others, from dozens of nations, working together to realize a
shared vision to seek answers to shared questions. The SSCL also provides an opportunity to
change the mistaken impressions about science and scientists that have resulted in fewer
students pursuing careers in fields related to science. In addition, it will serve as a catalyst to
help people understand the roles that scientific thought and inquiry can play in bettering their
lives and the lives of their offspring.
Recognizing this problem in our society, the creators of the SSC Laboratory made a
commitment to use the SSC to improve science education as well as to seek answers to the
most fundamental questions of modern physics. Consequently, in addition to building the
world's premier high-energy physics laboratory, the SSCL has a second goal: creation of a
"major national and international educational resource." To achieve the latter goal, the
Education Office of the SSCL is charged with using the resources of the Laboratory, both
during construction and during operation, to improve education in science and mathematics at
all levels (pre-kindergarten through post-doctorate) and for all components of our society
(including the general public), in the United States and around the world. To accomplish this
mission, the Education Office has established four goals and has already begun a wide variety
of programs to address them. These goals are to:
·Operated by the Universities Research Association, Inc., for the U. S. Department of Energy under Contract
No. DE-AC35-89ER40486.
Supercollider 4, Edited by J. Nonte 21

I. stimulate interest and promote understanding in science and encourage scientific
thinking, problem solving, and effective communication;
2. encourage participation in science and technology as a career, an avocation, or a
general interest;
3. improve the classroom experience for teachers and students; and
4. provide leadership for systemic improvement locally, regionally, throughout the
state, and nationally.
These goals and the strategies identified to accomplish them are consistent with the
President's National Goals and Objectives for Mathematics and Science and with the priorities
established by the Committee on Education and Human Resources (CEHR) of the Federal
Coordinating Council for Science, Engineering and Technology (FCCSET).
As with the SSCL itself, the Education Office is currently in a phase of designing and
building. Even so, numerous programs are already in operation to meet the demands of the
education community. Examples of such programs are described below.
• Adopt-A-Magnet Program. A curriculum development project designed to

introduce to pre-kindergarten through ninth-grade students concepts of magnetism, atomic
structure, forces, superconductivity, and other areas of science related to the SSC in ways that
will increase their enthusiasm for science and their interest in the world around them. The
program uses puppets, games, songs, experiments, and interactive computer software in two-
to-three-week supplementary units of hands-on activities to accomplish its goal.
• Department of Energy (DOE) National Science Bowl. A national science

competition designed to encourage and reward academic achievements in mathematics and
science by students attending U.S. high schools. DOE national laboratories conduct regional
or state science bowls to determine the high-school team that will represent the Laboratory at a
national event in Washington, D.C., each spring.
• Super Saturday. A series of field trips to technological sites (including the SSCL)
that stimulate student interest in science and enhance awareness of technological career
opportunities by demonstrating scientific concepts and technological applications. The
program is offered to Ellis County, Texas, (site of the SSC) high-school students and is
funded by the O'Donnell Foundation.
• Student Intern Program. Provides students with summer projects at the SSC
ranging from building particle detector models to muon-ray tracking simulations. Intended to
stimulate high school and undergraduate students' interest in science and encourage them to
consider careers in science by involving them in educational and work experiences in a high-
energy physics research environment. Students also participate in special educational
activities.
• Contemporary Physics Education Project (CPEP). An interactive particle

physics project designed to promote science education by stimulating high school and
freshman level college students' interest in, and understanding of, the fundamental concepts
of particle physics. Program includes interactive curriculum materials, hands-on activities, a
computer program, and the Standard Model wall chart.
• SSC Technology Projects. Two state-funded projects designed to provide

technology transfer to faculty and students of Texas community colleges and technical
institutes for the facilitation of knowledge and expertise in SSC-related technical areas. The
programs are funded by the Texas Higher Education Coordinating Board through a Carl
Perkins Grant.
22
• Cooperative Education/Student Research Programs. Afford undergraduate
and graduate students hands-on work experiences to facilitate integration of classroom studies
with practical application in a laboratory environment. Program structure consists of periods
of academic study alternating with periods of paid employment in accordance with formal
agreements among the students, the home institution, and the Laboratory, as appropriate.
• SSC Ph.D. Thesis Program. Research opportunities that encourage doctoral

students to participate in research related to the mission of the SSCL by offering dissertation
topics to students attending U.S. institutions of higher education.
• SSC Fellowship Program. Provides fellowships to support research at

institutions of higher education and other nationallilboratories to encourage young scientists
to participate in research related to the mission of the SSCL and to enhance high-energy
physics programs at U.S. universities. Forty-four fellows have been selected to date from the
following states: California, Connecticut, lllinois, Iowa, Maryland, Massachusetts, Michigan,
Ohio, Oregon, Pennsylvania, New York, Rhode Island, Texas, and Wisconsin. The program
is funded by the Texas National Research Laboratory Commission (TNRLC).
• Minority College and University Program. SSC speakers make presentations

at historically black colleges and universities (HBCUs) and other minority campuses to
stimulate minority student interest and understanding of the SSCL and related technologies
and to motivate and encourage students to pursue scientific disciplines and subsequent careers
in science and technology. The program includes Black Executive Exchange Program (BEEP)
presentations
• Graduate Education for Minorities (GEM). A national program that provides

summer work opportunities and graduate fellowships for capable minority engineering
students to help increase the pool of minority students who receive advanced technical
degrees. The fellowships provide tuition, fees, and a $6,000 stipend per academic year.
• Student Field Trip Program. Laboratory-hosted visits to encourage and stimulate

pre-college student interest in and understanding of science and the SSe. Field trip activities
may include presentations, tours, make-and-take activities, hands-on-software activities, and
demonstrations.
• Teacher Workshop Program. In-service workshops that offer pre-college

educators an opportunity to learn about SSC technology, state-of-the-art science, available
science teaching resources, and improved instructional techniques. Each workshop is tailored
to the requesting school district's needs.
• Detector R&D Research Program. Funds detector research at U.S. national

laboratories and universities. Institutions in the following states have been awarded funding:
Alabama, Arizona, California, Colorado, Connecticut, Florida, Hawaii, Illinois, Indiana,
Iowa, Louisiana, Maryland, Massachusetts, Michigan, Mississippi, New Jersey,
New Mexico, New York, North Carolina, Oklahoma, Oregon, Pennsylvania, Rhode Island,
Tennessee, Texas, Virginia, Washington, and Wisconsin.
• Minority Opportunities in Science and Technology (MOST). Provides

work opportunities at the SSCL to motivate and encourage minority and disadvantaged high-
school students to achieve literacy in science and a college degree, and, perhaps, to pursue a
career in science. In addition to work experiences at the SSCL, the program offers
participants educational opportunities, such as special tutoring and seminars.
23
• Science Education Resource Program. A science education resource center that
will make SSCL services and resources available to educators at all levels. Through this
program, small equipment and science kits will be loaned to schools on a short-term basis to
supplement and enhance science courses. Science education resources will also include SSCL
staff (using the name "Dr. Quark") available on computer networks such as Kidsnet, TENET
(Texas Education Network), and World Classroom to disseminate information about the SSC
and to answer science questions.
• sse Student Work/Study Program. Student projects at the SSCL designed to

stimulate student interest in science and to encourage students to consider careers in science
by involving them in educational and work experiences in a high-energy physics research
environment. The program will allow students who have participated in an SSC educational
program with a fixed duration to continue working at the SSCL while enjoying special
education activities. It will also allow unique opportunities for student involvement on
projects whose timing or duration is not compatible with existing competitive programs.
• sse Technologist Program. Offers hands-on work experiences at the SSCL to

community college and technical institute students to facilitate integration of classroom studies
with practical application in a laboratory environment. The program will alternate periods of
academic study with periods of paid employment.
• sse Research Associate Program. Offers research opportunities at the SSCL to

educators to revitalize their interest in science and to enhance their understanding of new
technologies. Program participants will work with SSCL scientists and engineers on R&D
initiatives. The experience will also have a positive impact on the educators' students, who
will benefit from their teachers' renewed interest and understanding.
The Education Office will continue to develop programs throughout the SSC's
construction period and during the facility's long lifetime of operation. Through programs like
those above and through as-yet unimagined programs, the Superconducting Super Collider
Laboratory will strive to capture the imaginations of future generations of scientists.
24
SUMMARY OF THE PERFORMANCE OF SUPERCONDUCTING CABLE
PRODUCED FOR THE ACCELERATOR SYSTEM STRING TEST
PROGRAM
D. Christopherson, D. W. Capone II, I. M. Seuntjens

C. R. Hannaford, S. Graham, and D. Pollock
S SC Laboratory
2550 Becldeymeade Ave
Dallas, TX 75237
INTRODUCTION
One of the responsibilities of the Superconducting Technology Group within the SSC
is to supply the superconducting cable to support the various magnet programs, including the
Accelerator System String Test (ASST). The strand for the ASST program was purchased
from four vendors and cabled on two separate machines under the supervision of SSCL
personnel. The 30 strand SSC Inner cable was fabricated on the SSCL owned machine
located at New England Electric Wire Corporation in Lisbon, NH. This cabling machine was
built by Dour Metal S.A. in Belgium. The 36 strand SSC Outer cable was produced on the
prototype cabling machine at Lawrence Berkeley Laboratory. The cable dimensions were
measured with the in-line Cable Measuring Machine (CMM) at each of the two facilities and
the cable electrical tests were performed at Brookhaven National Laboratoryl. All cable
dimensions are reported in inches. A metric version of the CMM software is currently being
tested.
CABLE PERFORMANCE
Table 1 is a list of the average values of the strand and cable electrical measurements
for the major strand deliveries by each vendor. The Ie slope of the wire was calculated using
the ratio of the Ie measured at 5 tesla to the Ie measured at 6 tesla. This ratio is routinely
Table 1. sse cable data from measurements performed at Brookhaven National Laboratory.
The reported wire and cable Ie corresponds to fields of 7T and 5.6T for inner and outer
material respectively, at a temperature of 4.22K.
Vendor Del Sttand CableIc (A) Cabielc CV Cu/SC WireIc (A) Avg WireJe
(Kgs) (%) Deg. Slope
IGC Inner 1370 10806 1.17 1.53 343 -5.04 592
OSTInner 1360 11103 1.38 1.51 355 -4.15 577
SCN Inner 972 10393 1.46 1.5 336 -3.04 530
IGC Outer 1500 10759 1.3 1.75 302 0.83 571
OSTOuter 1340 11135 0.66 1.79 307 -0.89 613
SCNOuter 1376 10077 1.04 1.79 291 3.87 552
OTUOuter 397 10713 0.63 1.78 305 2.4 610

reported by BNL along with the other electrical test results. ~ssuming the plot of Je v~ H to
be linear, the slope was calculated for the purpose of companson between vendors. Smce Ie
is directly proportional to the magnitude of the magnetization, a larger Je slope will suggest
larger magnetization effects at the injection field of 0.6T. The data in Table 1 shows the
range of slopes between vendors to be on the order of 10% for both inner and outer
conductor. The Je slope of SSC strand is related to the processing of the material in terms of
extrusion, heat treatments, etc. Information gathered through the SSCL Vendor Qualification
Program should help us to better understand how these processing parameters affect the Je
slope of the final product.
11500 1.50
11000
ca$ 1.00 Ii
'-' I Avg lc
u -c
'=' 10500 >
8~
u 10000 0 .50 ~
'-
• CofY
9500 0.00
Figure 1. Critical current and Coefficient of Variation of SSC inner and outer cable batches.
The average critical current of batches of inner and outer cable is compared to the
Coefficient of Variation of each cable batch in Figure 1. The Coefficient of Variation (CV) is
defined as the standard deviation of the sample set divided by the average value of that
sample set. CV is a good tool to use when comparing the variations of different populations
or sample sets. By normalizing the magnitude of the standard deviation with the magnitude
of the average value, it allows direct comparisons of the variations to be made. A small CV
simply indicates a small relative variability in the process.
It is encouraging that the CY of all of the cable batches is under 1.5%. This means
that all variations related to the cable critical current, i.e. strand diameter, strand Ie, cable
dimensions and degradation, add up to less than a 1.5% variation in the final product. It is
fully expected that as the processes of strand and cable manufacturing improve, the variation
in the finished cable will decrease further.
CABLING MACHINE COMPARISONS

A plot of keystone angle versus midthickness for outer and inner cable is shown in
Figures 2a and 2b, respectively. While the outer cable data shows the majority of variation in
the angle measurements, the inner cable data displays variation in the midthickness
measurements. These observations are consistent with the turks-head design differences
between the two cabling machines.
The prototype cabling machine at LBL uses the most common type of turks-head
design. This design incorporates a two-piece roller and axle assembly with self-aligning,
tapered roller bearings. Such a design allows for keystone angle changes by adjusting the
preload on the side rollers. It is this side roller loading variation that accounts for the
keystone angle variations between cable runs.
The SSCL cabling machine utilizes a forged, one-piece roller and axle design with
non-locating cylindrical roller bearings. Because of the one-piece roller and axle and the non-
locating bearings, changes in side loading do not affect the angle of the cable. The angle is
ground into the dimensions of the roller itself and cannot be adjusted once the roller assembly
is in place. Only by removing the turks-head and re-grinding the rollers can the keystone
angle of the cable be changed.
It should be noted that some of the midthickness variation of the inner cable fabricated
from OST strand is deliberate. For reasons of mechanical stability, some of the cables
fabricated using the OST strand were intentionally produced on the lower end of the
midthickness specification. In the absence of tension, this cable becomes less stable and
26
1.06
~ 1.04 c IGCOuter
s::
< 1.02
<> OSTOuter
~ 1
~ 0.98 A SCNOuter
<>
~ 0.96 {jAc
0.94+-----------~---------~~A~--------~--------~----~
0.0453 0.0454 0.0455 0.0456 0.0457
Midthickness (in)
Figure 2a. Keystone angle versus midthickness of SSC outer cable produced on the LBL
prototype cabling machine. The axes are the approximate tolerances allowed by the
specification.
1.3
~ 1.25 c IGC Inner

<
o 1.2 <> OSTlnner
§
~
~ 1.15 A SCN Inner
1.1
0.0572 0.0573 0.0574 0.0575 0.0576
Midthickness (in)
Figure 2b. Keystone angle versus midthickness of SSC inner cable produced on the SSCL
cabling machine. The axes are the approximate tolerances allowed by the specification.
more prone to de-cabling or collapsing. By increasing the compaction, we hoped to

minimize this characteristic. This cable has been wound into several dipole magnets with no
major cases of cable failure evident.
CABLE DEGRADATION
The measured degradation is plotted against the Minor Edge Packing Factor for each
inner and outer cabling run in Figure 3. The technique used in measuring cable degradation
is explained in Reference 1. The Minor Edge Packing Factor is calculated from the equation2
M.E.P.F. = wire area/0.5 * minor edge cable thickness * wire diameter

The dotted lines show the upper and lower limits of the Minor Edge Packing Factor
for inner and outer cable as allowed by the specification tolerances of the parameters
involved.3 The tight grouping of the data along the packing factor axis versus the larger
spread along the degradation axis suggest that Minor Edge Packing Factor is not the only
variable influencing the degradation.
The degradation values of the outer and inner cable runs using each vendors' material
is shown in Figures 4a and 4b. The horizontal lines across the bars represent the weighted
average of the degradation for that batch of cable. The degradation of each individual cabling
run is weighted proportionally by its length. Thus, the longer cable runs have a larger impact
on the weighted average degradation than the shorter runs. This clearly illustrates the fact
that strand from different suppliers will produce different degradation results.
27
6.00 ·IGCOuter
4.00
I •••
•••
I
I I
t•••
.OSTOuter
..
2.00
~
I • I
• 1:
'-' • SCNOuter
§ 0.00 I I
t • ·OTUOuter
.~
-2.00
8 -4.00
•
• O~ •• [] IGC Inner
• ~
I
• o OSTInner
-6.00 • •
•
[]
-8.00
0.93
· 0.94 0.95 0.96 0.97
• ,
0.98
A SCN Inner
Minor Edge Packing Factor

Figure 3. Cable degradation versus Minor Edge Packing Factor for SSC inner and outer
cable runs. The dotted lines represent the limits of M.E.P.F. as allowed by the specification
tolerances}
-
4
~
3
'-'
1
I:: 2
1
0
SCN
Cl -1
-2
-3
Figure 4a. Cable degradation of SSC outer cable fabricated from four vendors' strand.
IGC OST SCN

O""""~T.
-1
t
§ -2
8
-31!llliilili!llillilr li"i"":I·ll i·.J·ii i ·I:/
-4 11"I:iliil~
~ -5
-6
-7~----------------------------------------~
Figure 4b. Cable degradation of SSC inner cable fabricated from three vendors' strand.
28
The copper thickness around the filament array of the SSC strand also plays a part in
the amount of degradation experienced during cabling. The ratio of rID against the average
degradation of each cable batch is shown in Figure 5, where r is the smallest distance from
the outside diameter of the strand to the filament array, and D is the strand diameter. Both r
and D were measured from micrographs of virgin strand taken at 100X.
The outer cable data shows a very strong correlation between degradation and rID.
Micrographs of cable cross sections have shown that a thicker copper jacket helps to absorb
the deformation of the cabling operation and protect the filament array. However, the rID
ratio of the strand affects cabling degradation in another respect as well. As the strand travels
around the cable edges, it experiences tensile and compressive stresses which deform its
cross section. These tensile stresses act on the filaments located at the outer radius of the
bend and reduce their cross sectional area. This reduction in filament cross section has the
effect of reducing Ie, and defines the cabling degradation. A companion paper by Capone, et
a1.4 addresses this issue further. By increasing the copper jacket, the filament pattern is
moved toward the center line of the strand cross section. This in tum reduces the tensile
stresses acting on the outermost filaments within the strand as it is deformed around the cable
edges, and lowers the magnitude of the filament area reduction. Image analysis work is
presently underway to quantify the relationship between rID and cable degradation. Results
of this work will be presented at a later date.
4
•
,.-...
~ 3
=
0
2
•
....
'"
-0
...'" 1
•
•
bIl
0
0 0
-1 •
0.05 0.07 0.09 0.11 0.13 0.15
riD ratio
Figure 5. SSC outer cable degradation versus rID ratio.
The observations from Figures 3, 4a, 4b, and 5 support the conclusion that cable
degradation is dependent upon strand vendor as well as cabling parameters. If degradation
were plotted against Minor Edge Packing Factor over a wide range of M.E.P.F.'s , the result
would be a family of vendor specific curves. The variation among these curves is dependent
upon the strand geometries, expressed here in terms of rID. Future work to define vendor
specific degradation curves using strand with different geometries is planned and will be
presented at a later date.
STRAND PERFORMANCE
The average and coefficient of variation of the critical current for batches of inner and
outer strand is shown in Figure 6. The numbers in each column represent the approximate
number of 12 inch multifilament billets in that batch. Other than approximately one-half billet
of inner material, all final strand met the SSC critical current specifications of 339 A (1.5: 1
Cu/SC) or 368 A (1.3:1 CU/SC) at 7T for inner and 286A at 5.6T for outer strand. This half
billet of inner was approximately 3% low in Ie, and a waiver was processed to accept this
material for cabling. This cable met the Ie requirement of 9990 A at 7T due to the low
cabling degradation and was wound into a dipole magnet.
29
400 .......f - - - - - - - - - - - - - ; . _ - - - , - 2.5
2
$ 350
~ 300
«I 1 () • CVIe
.g
.~ 250
u 0.5
Figure 6. Critical current and Coefficient of Variation for SSC inner and outer strand.
1.9
1.8 3
1.7
2.5
1.6
U 1.5 2 """""""""""""""""i"',j Avg Cu/SC
~
u 1.4 1.5 () • CVCu/SC
1.3
1
1.2
1.1 . 0.5
1
Inner
Figure 7. Cu/SC ratio and Coefficient of Variation of SSC inner and outer strand.
3.5
3
U
• •• •
2.5
~
u 2.
....
•
0
> 1.5 •• •
u
1 l- •
0.5 •
1 1.5 2 2.5
CVofIe
Figure 8. CV of Cu/SC ratio versus CV of Ie for SSC strand.
The important fact shown in Figure 6 is that it is possible to produce strand with an Ie
coefficient of variation of less than 2%. The average and CV of the Cu/SC ratio for the
material in question is shown in Figure 7. In most cases, the variation is slightly higher than
that of the Ie, and a comparison of the variations of the Cu/SC and Ie is shown in Figure 8.
As expected. this plot shows a trend toward higher Ie variations as the Cu/SC variations
increase. This simply suggests that tighter control over the Cu/SC ratio will help to minimize
30
the Ic variation of the final strand. This feature also emphasizes that the process delivers a Ic
in the NbTi and not an Ic in the strand Adequate process controls can produce low variation
in strand properties.
The major deficiency of the delivered strand is illustrated in Figure 9. The piece
length statistics ranged from adequate to poor. This average piece length was calculated
using only the delivered lengths. In some cases, the extremely short pieces «300 meters)
were not delivered. This would further reduce the average piece length of some of these
strand batches to even lower levels.
The yield loss due to piece-length problems of this magnitude can be summed up by
the fact that we received approximately 65,000 meters more strand from the four outer billets
with good piece length than we received from the five outer billets with poor piece length.
The dependence of yield upon piece length is shown in Figure 10. The data has been
adjusted for multifilament design differences so that a direct comparison can be made. A
paper by Seuntjens, et al. 5 discusses the piece length and yield relationship more fully.
Besides yield loss, there are the obvious problems of drawing, testing, and cabling material
of poor piece length which will not be discussed further here.
8000 350
e
'-'
7000
--/ 300
-5 6000 250 AvgPiece
~5000 ~
Inner 200 :;::I Length
..s~ 4000 ~
~ 3000
c::till 2000
150 "";;0
100 ~ • Kg/Billet
~ 1000 50
0 0
Figure 9. Average piece length and delivered Kglbillet of sse inner and outer strand.
360
....
340
c
•
.£ 320
:;::I
~
....... 300
'" 280
~
• • Outer Strand
260
c
•
c
C Inner Strand
240
0 2000 4000 6000 8000
Avg Piece Length (m)
Figure 10. Delivered Kglbillet versus average piece length for sse inner and outer strand.
Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray (EDX) analysis
were performed on selected wire breaks from the various vendors. In some cases these wire
breaks had occurred during cabling, resulting in the loss of a full coil length of cable. In
nearly all of these samples an inclusion, or evidence of an inclusion was found. 6 These
inclusions were generally stainless steel (Fe, Cr, Ni) or silicon containing particles.
We fully expect this piece length situation to improve, driven by the commitment of
each vendor to the sse project and the competitive environment generated in the Vendor
Qualification Program.
31
SUMMARY
The superconducting cable fabricated by the SSCL to supply the Accelerator System
String Test was made using strand from four different vendors. The coefficient of variation
in the cable critical current is less than 1.5%, indicating a promising consistency in cable
manufacturing. The trend toward lower degradation with thicker copper jackets is explained
by the rID ratio of the strand, and a strong relationship supporting this argument is shown in
Figure 5. Obvious differences are present in the dimensional variances of cables produced
on the two different cabling machines. Both turks-head configurations show a distinct
signature in the measurement data of the keystone angle and cable midthickness.
The critical current coefficients of variation in the batches of delivered strand are
2.5% or lower in all cases. It is expected that this variation can be lessened with a better
control of the Cu/SC variations in the batches. As was reported last year>, the piece length
statistics of the delivered strand was the major weakness in the program. There were a few
encouraging billets however, and the Vendor Qualification Program has addressed piece
length as one of the main points of focus for the vendors.
ACKNOWLEDGEMENTS
The authors thank M. Garber, Brookhaven National Laboratory, for his extensive
cable electrical testing and F.Y. Clark, SSCL, for his technical assistance.
The SSC Laboratory is operated by the Universities Research Association, Inc., for
the U.S. Department of Energy under contract number DE-AC35-89ER40486.
REFERENCES
1. M. Garber and W. B. Sampson, "Quality Control Testing of Cables for Accelerator
Magnets", IISSC, New Orleans, LA, Feb 8-10, 1989.
2. J. M. Royet and R. M. Scanlan, "Development of Scaling Rules for Rutherford Type
Superconducting Cables", ASC, Snowmass, CO, Sep 24-28, 1990.
3. SSC Specifications SSC-MAG-M-4145, SSC-MAG-M-4146, SSC-MAG-M-4147, and

SSC-MAG-M-4148.
4. D. W. Capone II, V. Bardos, D. Christopherson, F. Y. Clark, S. Graham, C. R.

Hannaford, T. Reed, and J. M. Seuntjens, "Cable Keystone Angle Optimization for 40mm
SSC Quadrupole Magnet Development", IISSC, New Orleans, LA, Mar 4-6, 1992.
5. J. M. Seuntjens, D. Christopherson, F. Y. Clark, and D. W. Capone II, "Cold Weld
Analysis in SSC Strand and Cable", IISSC, New Orleans, LA, Mar 4-6, 1992.
6. D. Christopher-son, D. W. Capone II, J. M. Seuntjens, D. Pollock, and C. R.

Hannaford, "Summary of the Performance of Strand Produced for the 1990 SSC Dipole
Program", IISSC, Atlanta, GA, Mar 13-15, 1991.
32
DEVELOPMENT OF NbTi SUPERCONDUCTOR FOR THE SSC
BY SUMITOMO ELECTRIC INDUSTRIES
S.Saito, T.Sashida, M.Ohmatsu, Y.Yamada, M.Koganeya,

A.Mikumo, M.Nagata, and M.Yokota
S SC Project Team
Sumitomo Electric Industries, Ltd.
1-1-3, Shimaya, Konohana-ku, Osaka, 554 Japan
ABSTRACf
NbTi superconductor to be used for the Inner coil of Collider Dipole Magnet has been
developing by Sumitomo Electric Industries, Ltd.(SEI). We adopt the 12 inch composite
billet with production unit weight of above 300 kg. This large billet is well designed to
improve yield of extrusion, and assembled by using an unique single stacking technique
under severe cleanliness control.
We have been investigating the dependence of some important factors, ... such as NbTi
alloy source, HIPing, hot extrusion condition, and heat treatment condition and so on ... , on
the superconductive properties, productivities and drawabilities of the conductor.
We would like to present the status of our work, including manufacturing technology
and R&D experimental results.
INTRODUcrION
SEI was nominated as one of companies to perform the R&D phase ( Phase I, II ) in
the development of NbTi superconductor for the Inner coil of the SSC Project. We recognize
that main issues of the R&D phase as follows;
In Phase I:
(1). Study on Production Variables
(2). Study on Deviation
(3). Establishment of Quality System
In Phase II:
(1). Integration of Phase I Results
(2). Establishment of an Optimized and Standard Process
(3). Application of Quality System
To respond to these requirements, SEI organized the SSC Project Team; divided into
R&D, Technical, Production, Quality System, Quality Assurance, etc, groups.
In this paper, we introduce our technical activities for the development of the
superconductor.
PRODUCfION PROCESS
Figure 1 shows the production process used at SEI to manufacture the dipole inner
conductor[1 ,2,3].
Supercollider 4, Edited by J. Nome 33

In R&D Phase of SSC Project, SEI have been examining two kinds of NbTi alloys
from two different suppliers. The NbTi alloy is assembled with Nb sheet in a large Cu can to
form a monofilament billet. This monofilament billet is then extruded and drawn to a
hexagonal size of 2.6 mm (distance across).
The multifilament billet is constructed of approximately 8,000 hexagonal pieces. A 12
inch (outer diameter) multifilament billet; with a production unit weight over 300 kg, is used
to assemble a large number of Cu/NbTi single core hexagonal pieces; enabling a savings on
assembly costs. Furthermore, SEI has been developing an unique method to reduce the
multifilament assembling hours. Figure 2 shows a cross section of a multifilament billet
which has a packing factor around 95 %. The packing factor of the billet after HIPing
becomes almost 100%.
Drawing!
Hexagonal
Segment
~J)
igurc 2. Cro ection of
multifilament billet
~U~
&
~
~
~ FIgure J. Cross section of
Figure 1. Production Process extruded rod
Figure 3 shows the cross section of an extruded rod. The arrangement of NbTi
filaments is very uniform. This is a result of suitable billet assembly, HIPing, and extrusion
conditions.
The extruded multifilament rod undergoes drawing, heat treatment, twisting, and
cabling. During these operations, the extrusion, HIPing, and heat treatment processes are
very important because they are performed above the (X- Ti aging temperature; and affect such
properties as critical current, filament breakage and wire breakage.
34
Some results for optimization of extrusion, HIPing, and heat treatment condition are
detailed in the following sections.
EXTRUSION CONDITION
Extrusion is one of the important processes for manufacturing NbTi superconductor.

Two main items are considered to decide the extrusion condition. One is extrusion force;
which is controlled by the facility capability. The second is the conductor properties; such as
critical current, filament and wire breakage. Both these items are sensitive to the affects of
extrusion temperature.
If the extrusion temperature become higher, the extrusion force can be decreased; but
the high temperature causes a risk of degrading the conductor properties because of structure
re-crystallization in NbTi alloy and Cu-Ti compound formation on interface between NbTi,
Nb and Cu. If the extrusion temperature become lower, the extrusion force must be
increased, and there is risk of not extruding the billet.
It is important to know the relationship between the extrusion force, the conductor
properties, and the extrusion temperature[1J. The extrusion force is ordinarily expressed as
the following formula (1).
Ao
F=Aokf In ( - )
Af ... (1)
F : Extrusion force
AO: Billet cross-sectional area
Af : Extruded rod cross-sectional area
k f : Deformation resistance
The deformation resistance, kf, is determined by such variables as; the mechanical strength
of the billet's component materials at high temperature, the shape of the extrusion die, the
friction, and the extrusion speed.
In R&D phase, SEI have been investigating two kinds of NbTi alloys. It is necessary
to know these material's mechanical properties at high temperature in order to estimate the
extrusion forces. Figure 4 shows the temperature dependence of tensile strength for the two
NbTi alloys and for Cu. The examined sample size corresponds to the hexagonal piece size.
35
o Alloy "A"
..-.+-1 II Alloy liB"
1-.---.--.--+-----.--.--+---.-.-
30
~~~~~~~~~~~ ~~~~~~
25 .-.-.- --.-.J .... -.-.
X Cu
20
:::::: :::::i:':::::
._._.- -_ ... - ~~.. -.
(j max
(kg/mrr7) :::::: :::::: :::::l~:::: :::::: ::::::
15 ._._.- ._ ... - .. -.- .. ._._.- _._ .. -
-~~
:::::: :::::: :::::: :::::.~::::: ::::::

.-.-.- --.-.. -.-.-.. -.. -. ·-·-ni.·-··-
10
x:::::: :::::: ::::::
.-.-.~--.-.- _... _.
5 .---.- .-_._- ._.--lX----.--+-.
x .--.-.-+--.-----4
.. -
o :::::: :::::: :::::: ::::::f::::::f:::::?
~~~~~~~~~~~~~
500 550 600 650 700 750 800

Temperature CC)
Figure 4. Dependence of tensile strength of two NbTi and Cu on temperature
35
Below 650 oC, the tensile strength of Alloy "B" is higher than the one of Alloy "A".
Above 650 oC, the tensile strength of the two alloys are the same, and it was confrrmed there
was no difference at extrusion.
Regarding the conductor properties, we are now studying the dependence of critical
current density, filament breakage and wire drawability on extrusion temperature.
HIPING CONDITION
The HIPing process is effective for improving the packing factor of the multifilament
billet. It is believed this process is necessary to make uniform metal flow during the
extrusion process. However, the HIPing cost is expensive and it will increase the production
cost; in spite of the decrease in assembly cost.
Furthermore, the HIPing process has the possibility to affect the critical current since it
is carried out above the 0.-Ti aging temperature. Thus, it is very important to examine the
effect of HIPing process for both quality and cost issues. In R&D phase, we have been
investigating the effect of HIPing and Non-HIPing.
So far, the drawability was checked using approximately 15 kg weight material, and
we could not fmd any remarkable differences between HIPing and Non-HlPing. Before a
final conclusion on the effectiveness of HIPing is made, it is necessary to pursue more
detailed examination on various other factors such as mechanical properties, mass
productivity and deviation, etc.
HEAT TREATMENT CONDmON
SEI considered several criteria to decide heat treatment. These criteria are the following:
(1).Critical current (Ie) > 368 A (7T)
(2).Lower (dJc/dH)
(3).Deviation of Ie
(4 ).Drawability
From these criteria, we describe (1).Critical current and (2). Lower (dJc/dH) in this
paper.
To ensure the Ie, we designed the Jc in terms of wire diameter and Cu ratio as
follows;.
Jc ~ _ _ _3_6_8-,(_1e..r...)- - - x(1 + 1.3+0.1)

(0.808-0.0025l x 11: /4
maximum Cu ratio
minimum diameter
=1,733 A/mm 2 ........ (2)
For the SSC CDM conductor, it is important to minimize the Jc gradient against
magnetic field (dJc/dH) so as to reduce the magnetic field distortion at low field. Figure 5
details our experimental results of the relationships between the Jc and (dJc/dH). Higher Jc
wire shows higher (dJc/dH). From the results shown in Figure 5, we have defined the in-
house standard of (dJc/dH) as 575.
SEI utilized statistical methods to investigate the correlation between the critical current
density (Jc) of Inner strand and several kinds of parameters. Based on this investigation, we
defined the Jc by following equation (3);
Jc=f (T, t, N, d) ...... (3)
In this equation, we consider that important factors to affect on Jc are as follows;
(a).Heat treatment size (d), temperature (T), time (t)

(b).Number (N) of heat treatments from the extruded size to final wire size
(c). Reduction from the extruded size to the 1st heat treatment size
(e).Reduction between each heat treatment size
(f). Reduction from the final heat treatment size to the fmal wire size.
36
The extrusion size and final wire size are ordinarily fixed. Therefore, the factors of
(c),(e), and (f) are not independent but rather are related. Other factors before multifilament
billet extrusion also have an effect. For example: NbTi alloy, extrusion temperature of mono-
and multi-filament billet et.a!. We intend to consider these other factors before extrusion in
our analysis and improve the above equation.
/
/
700 /
650 /
600
~ 550
~ / 5
/
/
500 /
/
450
/
/
/
400 /
1500 1600 1700 1800 1900

Jc (Nmm 2at 7T)
Figure 5 Relationship between (dJc/dH) and Jc
We obtain this empirical equation by multi-variable analyses[4l Figure 6 shows an

example of multi-variable analysis. There is some scattering between expected Jc values and
measured Jc values in the case of simple linear regression. However, there is good fit in the
case of multiple linear regression.
Based on the above analysis results, we selected 4 types of heat treatment conditions
for the Inner conductor. The heat treatment conditions were selected from the point of the
1200 r:::======:::::::----71 1200 r::::;::;:;:;:::;::::;:;:=~---71

Multiple Linear
Regression
1100 1100
]
>1000
~
~ 900
~
~ 800
700~~~--~--~--~--~
700 800 900 1000 1100 1200 700 800 900 1000 1100 1200
Expected Jc Value (Nmm 2at 8 T) Expected Jc Value (Nmm 2 at 8 T)
Figure 6. Multi-variable analysis on Jc
37
u- Ti precipitation by changing heat treatment time and temperature. Type 1 is a baseline
condition. Type 2 and 3 are heavy and light heat treatment respectively. Type 4 is different
temperature condition.
Figure 7 shows the dependence of Jc on the heat treatment temperature for Type 2 heat
treatment condition[2]. The expected line from multiple linear regression was plotted on
Figure 7. Measured values closely matched the expected values.
1800r-----------~------------------~
,,, ,,,
..... _. _. ---r" --_ .............. , ................... -- ~ .... _. -.. . .
~
~
o 1700 - ----- - -+- ----- -+----------- i-----_.- ----
--~,,-
1 : 1
'r;;
.@ i Meas~ed Value i i,,
1
cau
1650
. --- -_ . --_ . --:--- -_ ............ --+ .... -- --- -- .... -. --- --- --- --- i-- --- ........ -_ ..
,
I
I
I
I
I
I
I
I
•
•
•
I
•
•
•
:g ________________________________
u
1600~ ~
370 380 390 400 420

Heat treatment temperature
Figure 7. Dependence of Jc on the heat treatment temperature for Type 2
CONCLUSIONS
1. SEI have been investigating two kinds of NbTi alloys. The temperature dependence of
tensile strength for the two NbTi alloys and Cu was examined to estimate the extrusion
condition. There is a tensile strength difference between the two alloys at a temperature
below 650 0c. Above 650 oC, the tensile strength of the two alloys are the same level,
and we confirmed there was no difference at extrusion.
2. The effect of HIPing on some properties was examined. The draw ability were investigated
using 15 kg of multifilament wire. No remarkable differences were not observed between
HIPed and Non-HIPed materials.
3. SEI designed several criteria to determine the optimum heat treatment condition.
Regarding to the critical current, we designed Jc and (dJc/dH) as follows;
Jc ~ 1,733 Nmm 2 (dJc/dH) ~575
4. The empirical equation of Jc were designed using multi-variable analysis. The equation,
analyzed by multiple linear regression, showed a good fit between expected and measured
Jc values.
ACKNOWLEDGMENTS
The authors would like to express their thanks for the invaluable assistance and support
of SSCL's members and their many colleagues.
38
REFERENCES
[1].S.Saito et aI., " Recent Activities for NbTi Superconducting wire at SEI',
Swnitomo Electric Technical Review, No 29, pp51-59, January 1990
[2].S.Saito et al., " Superconducting Properties of Fine Filamentary Superconducting
Wires"
Proc. of ICFA, New York, U.S.A., May 1986
[3].S.Saito et al., " Development of NbTi Superconducting Cable for the SSC by
Sumitomo Electric Industries, Ltd."
Proc. of 2nd IISSC, Miami, U.S.A, March 1990
[4].T.Sashida et aI., "Development of Superconducting Wire and Cable for the SSC
Project in Sumitomo Electric Industries"
Proc. of 3th IISSC, Atlanta, U.S.A, March 1991
39
A NOVEL APPROACH TO MAKE FINE FILAMENT
SUPERCONDUCTORS*
H. C. Kanithi, P. Valaris, and B. A. Zeitlin
IGC Advanced Superconductors, Inc.

1875 Thomaston Ave.
Waterbury, CT 06704
ABSTRACT
Multifilamentary superconductors with 2.5 to 6 micron sized NbTi filaments are required
for most of the High Energy Physics accelerator magnets. Present manufacturing technology
employs a single stack method for conductors with up to 10,000 filaments and a double stack
method for larger numbers of filaments. A unique new approach has been developed at IGC
Advanced Superconductors, Inc. to fabricate fine filament conductors containing up to -50,000
filaments. The method offers important technological and economic advantages even for the
manufacture of conductors with less than 10,000 filaments. The feasibility of the new approach
has been successfully demonstrated on small scale pilot billets. Large scale billets are currently
being developed. Three 309 mm diameter billets with 7,251; 22,313; and 38,663 filaments
respectively, have been fabricated to strand and cable. The details of the new approach along
with its advantages will be presented in this paper. Metallurgical aspects of the NbTi filaments,
namely, diffusion barrier integrity, geometric uniformity in addition to the superconducting
properties of the strand are reported. Relevant comparisons are made with conductor fabricated
using conventional methods.
INTRODUCTION
It is well understood that superconducting A.C. devices such as generators and motors
require multifilament conductors with filaments as small as practical to reduce hysteretic energy
losses during operation. For accelerator magnet applications such as the Superconducting Super
Collider (SSC) and European Large Hadron Collider (LHC), fine filament NbTi superconducting
wire is required to minimize distortion of the magnetic fields associated with particle injection.
The magnetization of the filaments gives rise to these distortions. The magnetization is
proportional to the critical current density (Jc), the volume fraction of NbTi and the filament
diameter.
The most optimal conductor will be one that can attain the highest Ie at the smallest
practical filament size. Early designs showed that filaments 2 to 3 microns in diameter would
require no correction field windings in the SSC dipole magnets, resulting in a significant
simplification of magnet construction with a corresponding reduction in the device cost. For a
given strand size and critical current rating, the number of filaments is inversely proportional to
the square of the filament diameter. Thus, large numbers of filaments, for example, -24,000 and
-42,000 would be required in the SSC strands if the filament diameter were to be 2.5 microns.
Such large numbers would require multiple bundling and processing steps. For instance. a
40,000 filament strand can be produced conventionally by a three stage process involving
*Supported by the US Department of Energy under a Small Business Innovation Research grant.

3 extrusions - the first one for the monofilaments, the second for an intermediate composite with
200 or so monofilaments, and the third for the last 200 or so multifilament bundles yielding a
total of 200 x 200 or 40,000 filaments in the final configuration. Starting with Fermi Tevatron
type conductor with 2,000 filaments to the more recent SSC dipole magnet strands with up to
8,000 filaments can be fabricated by stacking hexagonal elements. This so-called single stack
method using small hexagonal elements has a practical limit for filament numbers in the 10,000-
15,000 range. Larger numbers beyond -10,000 - 15,000 filaments can be achieved only by
incorporating additional steps. Although this multiple extrusion approach has been employed
there are a few drawbacks:
a) each extrusion step prolongs processing time and increases manufacturing costs,
b) each extrusion step reduces the material yield by 10% to 20%,
c) the filaments in the periphery of the first bundle (intermediate stage) tend to deform
non-uniformly into the softer inter-bundle copper region,
d) multiple exposure to thermo-mechanical processing also leads to distortion and
sausaging resulting in a reduction of transport le.
IGC Advanced Superconductors, Inc. has developed a novel and patented single stack
approach I to manufacture superconductors with numbers of filaments as large as 50,000. This
paper describes the invention which circumvents all the drawbacks of the three-stage process.
Conductors with approximately 4,000; 8,000; 23,000 and 39,000 filaments have been fabricated
using this method. Results of the pilot billet will be presented here.
HEXAGONAL-CELL APPROACH
It has been shown that when a single stack approach with either hexagonal 2 or round 3
elements is used, the attainable Jc and 'n' value (superconducting-to-resisistive transition index in
the expression voltage V IX I Iel n ) are high. If large numbers, on the order of tens of thousands,
of filaments are involved, both hexagonal elements or round elements lead to extreme difficulties
in billet assembly. Small hex rods tend to twist and bend, and round elements tend to dislodge
from a perfect hexagonal close packed pattern thereby resulting in various forms of stacking
imperfections.
To enable high rate, large scale fabrication of strands for SSC or other large application,
the present invention provides for a novel hex cell billet design and an assembly technique that
uses round rod elements. As shown schematically in Figure 1, this approach consists of
assembling a number of round monofilament rods in a hexagonal close packed (HCP) pattern
inside thin hexagonal tubes which are then arranged inside an extrusion can. The number of re-
stack rods N per cell is given by the expression
N= 0.75(nd 2 -I) + I
Monofilament
Rostack Rod
- + Wlre
Monofilament
billet
Final 8illet Assembly

Hexagonal Cell with Multiplicity of Hex-Cells
Fig. 1. A schematic of the Hexagonal Cell approach to manufacture fine filament

superconductors.
42
where nd is an odd integer equal to the number of rods along the diagonal. Thus, 7, 19, 37, 61,
91, 127, 169, 217, 271, and so on, can be placed inside a cell. The exact val ue of N depends on
the total number of filaments in a given billet, billet inside diameter and number of cells. The
speed of assembly of each cell is a function of re-stack rod diameter and length and number of
rods in the cell. Cells partially filled with re-stack rods (balance with pure copper rods) can be
placed near the core and periphery of the can to approximate a circular geometry. The straight
walls of the hex tube will maintain a perfect HCP symmetry without any stacking faults. The hex
tube material should be thin enough not to upset the local area ratio of the outer filaments and it
should be strong enough to withstand handling during billet assembly. Some candidate materials
for this are pure copper, alloys of copper and nickel such as eu-30 wt%Ni or Mn doped Cu, or
dispersion strengthened copper, for example, Cu-AI203. The stronger materials will impart
further support to the outer rows of filaments and preclude local filament sausaging. Highly
resistive materials will effectively reduce eddy current losses during rapid field ramps or in A.c.
devices.
Besides maintaining a perfect HCP pattern in each Hex-Cell, this method also provides a
uniform arrangement of filaments across the entire assembly. This is not the case in the last stage
billet assembly in a triple extrusion approach discussed earlier. Not only can the hex cell
approach of the present invention be used for large numbers of fine filaments (of the order of 3
microns or less) but it can be automated for large quantity billet fabrication. The automation
approach could, for instance, take advantage of pre-packing the Hex-Cells as subassemblies. It is
contemplated that the assembly approach of the present invention may be effectively utilized to
assemble numbers of fine round monofilaments ranging from 3,000 to about 100,000.
The assembled billets will be consolidated by techniques such as hot isostatic pressing
(HIP) to eliminate all the interstitial voids. Since each group of monofilaments are well contained
in the hex tubes, the HCP structure is well preserved during the consolidation step. The entire
assembly is then subjected to extrusion and drawing down to the desired size with intermediate
heat treatments to bring out the alpha titanium precipitates in the NbTi microstructure.
The NbTi in this Hex-Cell approach can easily be substituted by other ductile NbTi-based
superconductors such as NbTiTa. The method can also be applied to other filamentary
superconductors. One example is Nb3Sn conductors which involve processing Nb filaments
along with other ductile constituents (copper and tin) made by internal 4 or tin-ringS method. A
variety of conductor configurations are possible depending on whether the tin is applied
internally or externally.
The Hex-Cell approach can be repeated to make conductors with millions of filaments.
Such conductors would have utilization in the artificial pinning centers (APC) type wire 6.7 . The
benefit of using the Hex-Cell approach is that it cuts down on the number of extrusion steps
required to produce a conductor with such large numbers of wires. The Hex-Cell method could
potentially be applied to filamentary high Tc superconductors (HTS) in the future.
PILOT CONDUCTOR FABRICATION
4,200 Filament Conductor

The first model billet which was aimed at establishing the proof of principle was designed
to produce a 0.688 mm diameter SSC Outer grade strand with 4,200 six micron filaments and a
cu/sc ratio of 1.8:1. Two hundred and seventeen re-stack rods were assembled in each of 19 cells
which were placed inside a 194 mm o.d. copper can. Nb diffusion barrier between the NbTi and
copper in the monofilament rod was only 2% (of the NbTi by volume instead of the usual 4%).
Although the billet was small when compared to the routine 2S0 mm or 309 mm diameter
production units, this composite produced a Jc(ST) of 2,793 A/mm2 and lcC8T) of 1,246 A/mm2
at 0.648 mm SSC wire size. The resistive transition index 'n' was 46 at ST. Even at a wire size
of 0.470 mm corresponding to a filament diameter of 4 microns, the superconducting properties
were virtually unaffected. Although the starting re-stack was a round rod, the filament shape in
the final conductor was hexagonal. This behavior was the result of the flow strength of inter-
filament copper relative to that of NbTi filament during all the deformation steps staring with the
HIP. A separate experiment was conducted to see if round filament shape could be retained by a
proper selection of process parameters. As Figures 2a and 2b show, the filament shape can be
controlled from a highly hexagonal shape to an essentially round configuration.
43
Fig. 2 Cross sections of a billet assembled with round re-stack rods and hot isostatically pressed
to yield a hexagonal filament shape (a); and a round filament shape (b).
Fig. 3 SSC Inner ~rrade conductor at 0.808 mm dia. containing 7,251 - six micron filaments.
There are 127 rods per cell.
7251 Filament Conductor

Upon the successful completion of the proof of principle, a 309 mm diameter billet was
designed, assembled and fabricated into an SSC Inner grade strand with a culsc ratio of 1.5: l.
The number of round monofilament rods per cell was 127 and there were a total of 55 cells in the
entire billet. There were only two wire breaks before the material reached its final size of 0.808
mm. The longest piece was over 10,000 m and weighed 43 kg.
44
Although this billet was about half the length of our production billets, the piece length
results were much better than our past SSC Inner billets using conventional billet assembly
techniques. The filament uniformity was excellent as shown in Figures 3a and 3b. The Ic(7T)
and 'n'(7T) were 1,876 A/mm2 and 40 which were at an all time high for IGC for SSC Inner
grade conductor.
This wire was cabled into a Rutherford type 30 strand Inner cable meeting all the
dimensional and mechanical properties of the SSC specification. Mechanical behavior of the
strand during cabling was judged to be superior to that of conventional strand. The inter-cell
copper was believed to impart to the strand an additional easy plastic deformation as the
composite was rolled at the edges of the cable by the turks-head. The cable Ie, as measured at the
Brookhaven National Laboratory was 10,686 A including the usual self field correction. This
value leads to the conclusion that the present conductor can be made into a cable with virtually no
critical current degradation due to cabling.
Conductors with 2.6 Micron Filaments

Since the real advantage of the hex-cell approach lies in the manufacture of conductors
with tens of thousands of filaments, we have undertaken the development of 2.5 micron filament
composites. When we embarked upon this endeavor, the 2.5 micron filament size was a
requirement for the SSC High Energy Booster dipole magnets.
Two 309 mm diameter billets - SSC Outer with 22,313 filaments and SSC Inner with
38,663 filaments were designed. The Cu/SC ratio was 1.8:1 and 1.5:1 respectively. In order to
de-couple the filaments, an inter-filament matrix material of Cu-0.6 wt% Mn was used. The thin
walled hexagonal cells were, however, made with pure copper. The number of re-stack rods per
cell was 217 for the Outer and 169 for the Inner. After the assembly, the billets were electron
beam welded, compacted by hot isostatic pressing, and extruded to rod. Samples were processed
to final size with several heat treatment cycles in order to determine an optimal cycle which
results in the best 1c and 'n' values consistent with composite drawability. Figure 4 shows an
end-view of a Hex-Cell sub-assembly and a cross section of the Outer grade composite after
extrusion. Figure 5 shows corresponding pictures for the Inner grade composite.
The individual filaments, unlike the previous 7,251 filament conductor, had deformed
into a sharp hexagonal shape. Although this observation was made in the composite rod after
extrusion, it is believed that the shape change occurred in hot isostatic compaction as a result of
the relative high strength of Cu-Mn surrounding each filament. It is also believed that the smaller
size of the re-stack adds to this behavior.
The superconducting properties of these composites were measured at final wire sizes.
The h(7T) varied between 1467 and 1664 A/mm2 depending on the particular heat treatment
cycle given. These lc values are comparable to those we had obtained and reported for 2.6
micron filament conductors made by the conventional 3 stage process 8 . 1c(7T) values achieved in
the study were 1,617 and 1,605 A/mm2 for the Outer and Inner grade strands. The
corresponding 'n' values were respectively 36 and 35. The Outer conductor had a Ic(5T) of
2,765 A/mm 2. Since only small quantities have been processed to final sizes, it is difficult to
draw any conclusions on piece lengths. However, the Outer grade produced longer pieces than
the Inner. When compared to our own recent attempts 8 to make 2.6 micron filament conductors
by the triple extrusion process, the present Hex-Cell approach yielded lengths that were about 5
times longer.
SUMMARY
At IGC Advanced Superconductors, Inc., we have successfully developed a novel single

stack approach to manufacture multifilamentary superconductors with large numbers of
filaments. It employs thin walled hexagonal cells into which round monofilament rods are packed
in a regular hexagonal close packed arrangement. This method can be used for the manufacture
of conductors with fine and ultrafine filaments.
We have demonstrated that practical conductors with 4,000 to 40,000 filaments can be
easily fabricated. The Hex-Cell approach offers several key advantages over the triple extrusion
or multi stack method:
a) Easier monofilament rod processing
b) Modular approach to billet assembly
c) Possibility of automation for large scale production
45
Fig. 4 An end-view of a Hex-Cell sub-assembly and a cross section of the Outer grade
composite after extrusion. There are 217 rods per cell and the total number of
filaments is 22,313.
Fig. 5 An end-view of a Hex-Cell sub-assembly and a cross section of the Inner grade
composite after extrusion. There are 169 rods per cell and the total number of
filaments is 38,663.
46
d) Ability to assemble large numbers of filaments
by single stack approach
e) Regularity in filament arrangements and uniform
filament geometry
f) Higher yields and reduced process time
g) Excellent superconducting properties
h) Excellent mechanical behavior during cabling
Preliminary results show very high 1c, 'n' value and piece lengths. A 1c(7T) of over
1,800 Nmm 2 with and 'n' value of 40 was achieved in an SSC Inner Strand containing 6 micron
filaments. It was shown that there was no degradation in critical current due to cabling. A 1c(7T)
of over 1,600 A/mm2 and 'n' values over 35 were achieved in 2.6 micron filament conductors.
Although Cu/NbTi composites were addressed here, this new approach can be applied to
the manufacture of superconductors such as a) conductors with Artificial Pinning Centers (APC)
in which millions of filaments are required, b) Nb3Sn conductors by internal or tin-ring methods
and potential future filamentary high T c superconductors.
REFERENCES
1. United States Patent No. 5,088,183 entitled "Process for Producing Fine and Ultrafine
Filament Superconductor Wire", Inventor: Hem C. Kanithi, filed May 1, 1990 and issued
Feb. 18, 1992.
2. H. Kanithi, M. Erdmann, P. Valaris, F. Krahula, R. Lusk, E. Gregory and B. A Zeitlin,

Supercollider-2, edited by M. McAshan, Plenum Press, New York, pp. 601-609, 1990.
3. H.c. Kanithi, C.G. King, B.A. Zeitlin, "Fine filament NbTi Conductors for the SSC", IEEE
Trans., vol 25, 2, pp. 1922-1925, 1989.
4. E. Gregory, G. M. Ozeryansky and B. A. Zeitlin, "Improvements of the Structure and

Properties of Internal Tin Nb3Sn Conductors", presented at the IntI. Cryo. MatI. Conf.,
Huntsville, Alabama, USA, June 1991, to appear in the proceedings - Adv. in Cryo.
Engineering Materials, Plenum Press, New York, 1991.
5. G. M. Ozeryansky and E. Gregory, "A New Internal Tin Nb3Sn Conductor Made by a Novel
Manufacturing Process", IEEE Trans. on Magnetics, Vol. 27, No.2, March 1991, p1755.
6. United States Patent No. 4,803,310, entitled "Superconductors Having Controlled Laminar
Pinning Centers, and Methods of Manufacturing Same", Inventors: B. A. Zeitlin, M. S.
Walker and L. R. Motowidlo, issued Feb. 7, 1989, work initiated in 1983.
7. H. C. Kanithi, P. Valaris, L. R. Motowidlo, B. A. Zeitlin and R. M. Scanlan, "Further

Improvements in NbTi Superconductors with Artificial Pinning Centers", presented at the IntI.
Cryo. MatI. Conf., Huntsville, Alabama, USA, June 1991, to appear in the proceedings -
Adv. in Cryo. Engineering Materials, Plenum Press, New York, 1991.
8. H. C. Kanithi, P. Valaris, and B. A Zeitlin, Supercollider-3, edited by J. Nonte, Plenum

Press, New York, pp. 689-693, 1991.
47
RECENT RESULTS IN DEVEWPING NbTi FINE Fll..AMENT SUPERCONDUCTORS
J.Teuho, T.Erkolahti, P.Haasjoki, J.Lehto, R.Liikamaa and V.Vanhatalo
Outokumpu Superconductors Oy
P.O. Box 60, SF-2810l Pori, Finland
ABSTRACT
During 1989-1991 several billets of fine filament wires has been manufactured and
delivered by Outokumpu Copper. These wires were made to meet SSC/Dipole, HEB,
Corrector magnet and some other specifications. For wires, which have the filament number
24,000-46,000 and filament size 2.5 JLm a double stack method has been used. A single
stack method has been used for wires, where the filament number is 3500-4200 and filament
size 5-6 JLm.
For these, in the industrial scale manufactured multifilament superconductors, test results
and various statistics focusing on piece length, critical current and copper to non-copper ratio
will be presented.
INTRODUCTION
Outokumpu Superconductors has been actively working in developing fine filament NbTi
superconducting wires for several years. In June 1989 this work was collected into a "Fine
Filament Superconductors" project. Within this program we have had outstanding results in
SSC outer and inner wires of 6 JLm and 2.5 JLm filament size, in other CuMn matrix wires as
well as in CERN LHC wires, where the filament size is specified as 5 JLm.
In connection with the first fine filament billets, a production procedure was developed
to give good springback qualities and maintain the RRR value at the highest possible level.
The procedure has already been succesfully applied to the first fine filament wires delivered
to BNL. It has been developed further for the corrector magnet wires, several batches of
which we have delivered to BNL and SSCL.
During the production of the first fine filament billets, the efforts concentrated on
increasing the unit length of wire. For that purpose, billet design, manufacturing and pickling
of the billet components, initial dimension of coil drawing, multi-die drawings as well as
drawing lubrication and cleanliness of the wire at all phases of the production cycles were
studied.

We have manufactured the first fine filament billets using diffusion barrier of 2 % and
billet size of 205 mm and current density of 2900 - 3000 A/mrrr (at 5T and 4.2K) was
achieved. We have developed a special extrusion phase for this billet size. Now we are
checking how it is possible to apply our earlier results to the bigger billet size we are using
for SSC billets. For these bigger billets the diffusion barrier of 4 % is applied.
SSC CORRECTOR MAGNET WIRE
Eight billets of SSC corrector magnet wire were manufactured using a standard
hexagonal Cu/Sc monorod billet design. These billets are identified here using numbers as
presented in Table 1. These billets were designed with the specification of SSC Corrector
Magnet Wire. In the billet design (Fig. 1) we concentrated to get springback properties as
good as possible. Nominal values for this wire are:
diameter (bare) 0,381 mm

number of filaments 450
filament size 9.8 I-'m
filament spacing 1.4 I-'m
Cu/Sc ratio 2.35:1
twist pitch lOmm
The cryogenic grade copper was produced by our foundry in Pori, and the start material
RRR value had a minimum 250. Monofilament rod was produced by a hot extrusion process
using high homogenity grade Nb-47wt%Ti alloy ingot made by TWCA. Nb barrier thickness
was 4 %. Processing of this billet serie followed conventional manufacturing operations
including hot isostatic pressing, hot extrusion and cold drawing with intermediate heat
treatments.
Fig. 1. Cross section of SSC Corrector Magnet wire manufactured by Outokumpu

Superconductors.
50
Table 1. Average results from the billets of Corrector Magnet Wire
Yield (L) Yield %L Sb Ie (A) Ie(A/mm~

Billet no. total (m) (%) > 2750m 0 Cu/Sc 4T 5T 5T
06 146515 85.1 92.2 647 2.39 120 098 2918

54 133906 78.3 66.2 731 2.37 131 108 3168
55 128066 73.7 61.5 710 2.37 128 106 3122
56 149947 89.6 76.8 687 2.40 123 101 3039
57 149308 81.5 87.5 678 2.39 125 104 3106
58 149105 81.5 88.5 747 2.39 129 107 3139
59 149775 81.6 87.4 687 2.35 128 104 3089
60 142867 81.6 65.8 721 2.38 129 106 3136
For real production period, parameters such as piece length and yield become very
important issues. Because the filaments are placed to the center of wire, this wire is very
sensitive to central bursting. However, progress has been made over the last year. The piece
length performance of each billet has been collected in table 1 and the fraction where the
piece length is longer than 2750 m is shown in figure 2. An average yield was 81.6 % and
from the total length 78.2 % was longer than 2750 m. The final wire diameter was
continuously gauged and deviation was within 4 I-'m.
ll"lr t
1.l(1
lilt bO
..,
0;
:>: ~o
:'0
0
0 5-1 55 56 57 58 59 60
BIl1e\ number
Fig. 2. Quantities over 2750 m piece lengths in different billets of Corrector Magnet
Wire.
Wire springback is dependent upon wire design and wire processing, specially in the
final phase. The results we have measured on average from different billets are shown in
Table 1. The standard deviation of all billets is 4.6 %, which is high, but a part of it is
divergency of measurements.
Six samples were cut out of each billet and tested for critical current. Table 1 lists
critical current values at 4T and 5T on average in each billet. The billet no 6 is deviating
especially with current values (see current densities at 5T), 6-7 % lower than in other billets.
In figure 3 current densities from each billet are presented as a function of position in the
billet. The deviation of Ie is relatively small as indicated by the standard deviation of 2.6 %.
If we don't take into account the billet no 6, which was made as a different batch the standard
deviation is only 1.3 %. In figures 3 and 4 we have compared the variation of Ie and Cu/Sc
ratio along the billet length collected from each billet.
51
.. • •• ..•,.... • ..• ... • .• •
3300
3200 • Spec max
•• •• .+
N 3100
E • • • •
E 3000
"~
•• •• • ••
(.)
"") 2900
• •
2800 Spec min
2700
0 20 40 60 80 100 120 140 160 180
Position [km)
Fig. 3. Critical current densities at 5T vs. position in the billets collected from the
whole production part.
2,55 1
2,50 UCL
• • •
2,45
• ••
(.)
2,40 •• •
• ••• • ••• •
(J)
...... •~ • •• • •• I •
u
:::I
2,35 • J'
• • ••• ••• • •
2,30
• • •
LCL
'"'
L,Lv '"'''
2,20
0 20 40 60 80 100 120 140 160 180
Position [km)
Fig. 4. Cu/Sc ratio as a function of position in the billets.
sse DIPOLE OUTER AND INNER WIRE

1. Dipole Outer Wire
Figure 5 shows the process applied in manufacturing SSC dipole superconducting wires
in Outokumpu Superconductors. The cleaning and packing of the components has been done
in our clean room made specially for this manufacturing phase. In our extrusion processes we
have used 2500 ton and 4500 ton "in-house" hot extrusion presses for mono and multi billets,
respectively. The multi billet size is 278 mm in outer diameter.
In this chapter we are concentrating on two billets processed together and in addition to
this, some critical current results from a billet of inner wire is presented, also. The outer wire
billets are identified by no 91 and 92 and the inner wire billet by 106. Nb-47wt%Ti and
diffusion barrier material was delivered by TWCA and cryogenic copper originated from our
own foundry.
52
NbTi -billet
Pickling. assembling
OFC Cu can
and welding
Hot extrusion
Drawing. straigthening
and cutting
CulNbTi hexagons
Pickling. assembling OFC Cu can and

and welding Cu components
Hot extrusion
Drawings and
heat treatments
Wire testing
Fig. 5. Manufacturing process of Cu/NbTi superconductor of SSC Dipole outer.
We have used hexagonal Cu/Sc elements and single stacking method, which turned out
to yield good geometrical uniformity in the filament arrangement. Tight tolerances and very
straight hexagons are achieved by using special tools. Full size billets according to SSC
specification were produced by these methods.
Fig. 6. Wire cross section of SSC Dipole outer wire taken in the final diameter.
53
The outer billets were manufactured into wire using conventional cold drawing
and heat treating methods. A wire cross section is shown in Figure 6. The yield of these
billets was 78.2% (billet 91) and 83.3% (92). Both billets produced more than 50% of
total length in over 10 kms lengths.
In Figure 7 measured results of Cu/Sc ratios and critical currents from both of the
billets are collected. In each case we can see again the variation on critical currents is
mainly dependent on variation in Cu/Sc. The minimum and maximum Ie (S.6T) values
have been 306 A and 316 A in the billet no 91 and 303 A and 311 A in the the billet no
92. The low standard deviations in Ie of 0.86% (billet 91),0.89% (billet 92) and 1.12%
(billets 91 + 92) is a demonstration of the consistency and the reproducibility.
1.90 320
...
... • 315
1.85
... .... ...• ....... ... ......
• 310
u
(h
..... 1.80 • •••
•
• •
•
J • ....
• • ••
• 305 n
u
:3
• • • • 300 ):
.....
1.75 • 295
• Ie
• Cu/Se
290
1. 70 285
0 20 40 60 80 100
Position [km]
Fig. 7. I., and Cu/Sc ratio as a function of position in the billets no 91 and 92.
2. Dipole Inner Wire
SSC dipole inner wire billet (no. 106) was assembled by single stacking method. Up to
now only critical current optimizing work has been done. The results are shown in Table 2.
Table 2. VIe values in the billet no. 106

Field I., Ie
3T 885 A 4421 A/mm2
4T 735 A 3671 A/mm2
5T 600 A 2997 Almm2
6T 475 A 2373 Almm2
7T 343 A 1713 A/mm2
8T 218 A 1089 A/mm2
54
sse HEB OUTER AND INNER WIRE
1. Manufacturing of HEB wire, ....lament size 2.5 I'm
As one of the suppliers, we have been engaged in developing superconductors for SSCL
with the filament size under 2.5 J.(m, and the filament spacing under 0.5 J.(m. When the
filament is under 2.5 J.(m, there is need for about 46,000 filaments in the inner wire and for
about 24,000 filaments in the outer wire. In our billets the diffusion barrier thickness was
under 2 %. Due to the large number of filaments, double stacking procedure was applied. The
mono and first multi stage billets were hot extruded and for the second multi stage we used
hydrostatic extrusion. The design parameters (nominal) for outer and inner wire are:
outer inner
diameter 0.648 mm 0.808 mm
number of filaments 23826 45474
filament diameter 2.46J.(m 2.40J.(m
filament spacing 0.40J.(m 0.39 J.(m
matrix to sc ratio 1.82:1 1.47:1
In comparison to earlier development efforts, a new typical problem has been how to
obtain a sufficiently thin copper layer between the filament groups. The problem had already
been investicated before the work started and a production route was created by using dummy
test pieces.
2. Some results from sse HEB wires

The optimization of critical currents have given J. values of 2700 A/mm2 at 5T for both
wire types. With this test program, unit lengths of several hundred meters in the final
dimension was reached. Later with larger quantities, unit length problems were faced. After
adjusting the cold work/heat treatment program, long piece lengths could be reached still
matching critical current goals.
In the outer/inner wires, the following critical currents (table 3) were reached, when the
length distributions have been those shown in Figures 8 and 9.
Table 3. Ie and Je values for inner and outer wires with a filament size below 2.5 J.(m.
Outer wire Inner wire
Field Ie Je(A/mm2) Field Ie J.(A/mm2)
3T 416 3597 3T 790 3807
4T 348 3009 4T 660 3181
5T 294 2542 5T 550 2651
5.6T 262 2265 6T 450 2169
6T 242 2092 7T 342 1648
7T 185 1600 8T 225 1084
8T 125 1081 9T 110 530
9T 65 536
55
As a result of improved pattern design the inner wire results were better. The yield was
82,5 % and more than 75 % of wire have piece lengths over 3 kilometers (Fig.9). An average
critical current of 338 A at 7T was achieved with this wire and the variation of:r., was narrow
as indicated by the standard deviation of only 0.7 %. Also the variation in Cu/Sc values was
very low, standard deviation was 0.7 %.
8 50
40
~ 30
c:
.
....
~
<>
20
10
0
( 0.5 0.5-1.0 1.0-1.5 1.5-2.0 ) 2.0
Cathegory [ kmJ
9 80
70
60
i 50
c
40
.....
0
:;::;
.,
<>
30
20
10
0
( 1.0 1.0-1.5 1.5-2.0 2.0-2.5 2.5-3.0 ) 3.0
Cathegory [kmJ
Fig. 8. Piece length distribution of 2.5 J-tm outer wire.
Fig. 9. Piece length distribution of 2.5 J-tm inner wire.
SUMMARY
A volume production process achieved a critical current density of more than 3000
A/mm2 at 5T and 4.2K with the SSC Corrector Magnet Wire. Further, the standard deviation
of Ie was only 1.3 % in the batch. The average values of springback and RRR were 701 and
69°, respectively.
Two dipole outer wire billets were manufactured to the final diameter. The critical
current values at 5. 6T were more than 303 A and the standard deviation for these billets was
1. 12 %. The average yield was > 80 % and both these billets produced more than 50 % of
total length in over 10 kms piece lengths.
In wires, where the filament size was under 2.5 J-tm a critical current density 2600-2700
A/mm2 at 5T, was achieved. The outer wire had more than 45 % of total wire length in piece
lengths longer than 2 kms and the inner wire more than 75 % longer than 3 kms,
respectively.
56
RECENT RESULTS IN MANUFACTURING
NIOBIUM TITANIUM SUPERCONDUCTOR FOR THE sse PROJECT
D. Frost, J. Fernandez, W. Larson, R. Randall

G. Swenson, J. Wong and D. Yu
Supercon, Inc.
830 Boston Turnpike
Shrewsbury, MA 01545
ABSTRACT
Electrical and mechanical properties, as well as production results, are

presented for 3000 kg of superconducting strand delivered to the SSC during 1991.
Piece length and yield goals of the 1991 SSC dipole magnet program have been proven
attainable for both inner and outer 50 mm dipole conductor. A high degree of
reproducibility has been maintained and deliveries of dipole strand have been made in
record time. A superconducting corrector magnet wire has been developed and
manufactured which surpasses all developmental performance specifications. It is
anticipated that further enhancement of J c for the corrector magnet wire is possible
without sacrifice of piece length and ductility.
INTRODUCTION
In 1991 Supercon delivered nearly 3000 kg of finished superconducting strand

to the SSC. This material included: 566 km of outer dipole strand * from four billets;
250 km of inner dipole strand ** from three billets; and 422 km of corrector magnet
wire *** from two billets.
The seven dipole conductor billets were supplied to fill gaps in the SSCL
scheduling requirements for the 50 mm dipole program and help meet the deadline for
the upcoming dipole magnet string test. Although awarded on a competitive fixed price
basis, delivery was an overriding priority for this materials contract. Supercon began
delivering finished strand within 140 days of receipt of NbTi ingots from SSCL. All four
outer billets were delivered complete within 187 days. The three billets of inner dipole
strand were delivered complete between 30 and 35 weeks after receipt of materials.
Most of the strand produced in this project has been incorporated into 36 strand outer
and 30 strand inner cable for the 50 mm dipole string test magnets. The performance
results of these cables are detailed in a preceding paper by SSCL. (1) Most notable is the
fact that six 50 mm outer dipole magnet cable lengths and two inner dipole cable lengths
were manufactured from this material without any cold welds.
*Outer strand manufactured per SSC-MAG-M-4146

**Inner strand manufactured per SSC-MAG-M-4145
***Corrector magnet wire manufactured per SSC-A20-000002

Corrector magnet wire from two billets was delivered by Supercon as a result of
participation in a three phase development program to optimize the superconductor wire
for use in the superconducting corrector magnets for the SSC. Phase I and II of the
development program required delivery of 380 km of wire meeting specification
requirements on a firm fixed price basis. Phase I was completed within 8 months after
receipt of materials with the delivery of 204 km of finish strand. Phase II was
completed one month thereafter with delivery of an additional 218 km of finished
strand. Phase III RFQ's have yet to be issued by SSCL.
SUPERCONDUCTING STRAND PERFORMANCE DATA
Performance data and characterization of all the superconducting dipole strand

and correction magnet wire produced by Supercon and delivered to SSCL during 1991
are presented and discussed below. All data presented here are the results of
measurements performed by the manufacturer in conformance with test procedures
specified by SSCL and are representative of QA data supplied with the material.
Supercon earned the classification as a "Certified SSCL Critical Current Test Facility" in
1991, therefore all Ic data presented is comparable to data from other certified labs.
When presenting electrical properties groups of each type of conductor strand are
identified and classified by production unit designations consistent with SSCL QA
manufacturing specifications. Mechanical properties and production results are
correlated and presented on a per billet basis.
50 MM Dipole Strand
The mechanical performance of the outer and inner dipole strand from seven
billets is summarized in Tables 1 and 2 respectively. For the outer strand 98 percent of
all delivered lengths were greater than 1680 meters, and 91 percent of all delivered
inner strand lengths were greater than 1500 meters. These are the minimum lengths
which are required to eliminate cold welded strands from within a cable length.
Overall production yield from four outer billets averaged 141,454 m per billet
or 82 percent of theoretical yield calculated on a weight basis. Per billet yields from
three inner billets averaged 83,426 m or 80 percent of theoretical yield. These yields
approach the practical limitations for the production process.
The actual distribution of strand piece lengths throughout all the outer and inner
billets are shown in Figures 1 and 2 respectively. The overall average piece length of
outer strand throughout the four billets was 7,544 meters. The inner strand produced
an average piece length of 3,247 meters throughout the three billets.
Table I. Mechanical properties/production data for SSC 50 mm dipole outer strand.

Piece Length
Billet Multifilament
Designation %>1600m AVE(m) Total(m) Yield (%)
2734-1 98 8987 143,799 82
2734-2 98 7433 141,225 82
2734-3 99 8288 140,889 83
2734-4 98 6083 139,904 82
58
Table 2. Mechanical properties/production data for sse 50 mm dipole inner strand.
Ei~Q~ !.~Dglb
Billet Multifilament
Designation %>1600m AVE(m) Tolal(m) Yield (%)
2735-1 82 2791 83,719 77
2735-2 94 3130 78,251 83
2735-3 97 3636 87,262 80.5
200000
...
til
~
150000
Q)
E
-
100000
CIS
0
~
50000
o to 3 3 to 5 5 to 10 10 to 1515 to 20 >20
Strand Length Distribution - Shipped Wire (Km)
Figure 1 Piece length distribution of sse 50 ~m
dipole outer strand from four billets.
59
80000
...en
-(1)
( 1)
E
60000
40000
-co
....
0
20000
0
o to 2 2 to 4 4 to 6 6 to 9 11 to 14
Strand Length Distribution - Shipped wire (Km)
Figure 2 Piece length distribution of SSC 50 f.Lm
dipole inner strand from three billets
Table 3. Electrical performance data for SSC 50 mm dipole outer ~trand (5.6T,4.2K)
Production Critical Qu rrenl Qurrenl Qensil3::
Unit
Average Ic(A) SDEV(%) Averaoe Jc(A/mm2) SDEV(%)
-001 293 0.78 2471 1.00

- 0 02 292 1.38 2468 0.90
-003 293 2.05 2482 1.54
- 0 04 286 0.20 2437 0.19
- 0 05 287 0.00 2398 0.90
60
Table 4. Electrical performance data for SSC 50 mm dipole inner strand (7T,4.2K)
Production Critical Qurr~ot QUrrflDt Densit:i
Unit
Average Ic(A) SPEV(%) AYerage Jc(A/mm2.) SDEV(%)
-007 349 1.51 1690 1.44
-008 348 0.83 1678 0.53
-009 333 0.58 1685 0.67
- 01 0 339 0.69 1689 0.72
- 01 2 335 0.91 1643 0.98
Electrical performance is presented in the form of critical current data for outer
dipole strand in Table 3. Over the five outer production units the average critical
current ranged from 286 A to 293 A at 5.6 T and 4.2 K. The cumulative average Ie for
the four outer billets is 291 A with a standard deviation of 1.49 percent.
Critical current data for inner dipole strand is presented in Table 4. The average
critical currents for each of the five inner production units ranged from 333 A to 349 A
at 7T and 4.2 K throughout the three billets. Throughout the five inner production units
a cumulative average Ic of 340 Amps was obtained with a standard deviation of 1.66
percent.
Corrector Magnet Wire
The piece length and yield results for the two corrector magnet wire billets are
summarized in Table 5. In Phase I billet 2695 produced an average piece length of
7,573 m with a production yield of 204,460 m or 81 percent of theoretical based on
weight. An average piece length of 8,720 m was obtained in Phase II from billet 2696
with an overall yield of 217,990 m or 83 percent. The actual distribution of piece
lengths for the two billets combined is shown in Figure 3 with an overall average piece
length of 8124 m.
Springback properties for the two production units are included in Table 5 as a
representation of ductility in the corrector magnet wire. The average values of 522 0
and 545 for billets 2695 and 2696 are based on 100 percent testing on each piece of
0
wire.
Table 5. Mechanical properties/production data for SSC corrector magnet wire.

Billet Piece Length Yield(%) Spring Back
Designation Properties
Ave (m) Total(m) (average)
2695 7573 204,460 81
2696 8720 217,990 83
61
120000 t
100000
en
-
"-
CD
CD
80000
E
60000
-
as
o
I-
40000
20000 /----A~
if///AIIIII
o 1/ if/////n 0W.
2 to 4 4 to 6 6 to 11 11 to 17
Strand Length Distribution - Shipped wire (km)
Figure 3 Piece length distribution of SSC superconducting
corrector magnet wire from two developmental billets.
Electrical properties for the two corrector magnet wire billets are summarized
in Table 6. Average current density measurements at 5 Tesla and 4.2 K for billets 2695
and 2696 are 2845 Almm 2 and 2919 Almm 2 respectively. Billet 2695 resulted in an
average 103 A critical current with 1.45 percent standard deviation. In Phase II, billet
2696 resulted in an average 107 A critical current with 1.57 percent standard
deviation.
Table 6. Electrical performance data for SSC corrector magnet wire (5T, 4.2K).
Production Critical Cumml CurreOl l:2ecsilll
Unit
Average IclA) SI:2EYl%) Average JclAlmm2.) SI:2EYl%)
-006 103 1.45 2845 2.1
- 01 1 107 1.57 2919 2.0
DISCUSSION
The low variation of electrical properties within the three billets of inner strand
and the four billets of outer strand facilitate the fabrication of cable with uniform
critical current performance without the need for "creative" cable mapping. The low Ie
standard deviations of 1.49 and 1.66 percent for the outer and inner strand respectively
approach the earlier SSCL goals of 1.0 percent variation and are evidence of a very
reproducible process.
62
The impressive piece length statistics presented above for dipole inner and outer
strand were not derived at by selective delivery of only the long pieces: overall yields
from outer and inner strand from these seven billets were 82 and 80 percent of
theoretical yields respectively on a weight basis. This is considerable when compared to
an average 72 percent yield reported over thirteen billets in the overall 1990 SSC
dipole program (2). Delivered quantities of strand from the seven billets reported here
actually exceeded ordered quantities by 6.0 percent for outer strand and 2.5 percent for
inner strand.
The superior yield attained in the seven multifilament dipole billets was
compounded by equally impressive yields in the monofilament billets. Yields in the
monofilament were so high that Supercon actually returned 20 percent of the NbTi ingot
supplied by SSCL for performance under the contract. These monofilament billets were
returned as finished monofilament.
Fe
Cr
0.480 Range= 10.230 keV 10.230 ...

Figure 4 EDS spectragraph of inclusion particles
removed from monofilament surfaces.
The piece length and yield data presented here for the dipole strand could
technically have been much better if not for the unfortunate contamination of the
monofilament billets. During the extrusion process the surface of each monofilament
extrusion was severely contaminated by particulate inclusions. EDS analysis shown in
Figure 4, of these particles revealed them to be iron particles coated with or
contaminated with chromium. The EDS analysis allowed identification of the point source
of contamination and Supercon's manufacturing process has been modified to prevent
future recurrence.
63
As shown in Figure 5 some of these inclusions were as large as 2.5 mm in
diameter. The larger inclusions were physically removed by hand from the surface of
the extruded monofilament. Many smaller inclusions were detected using Eddy-current
inspection of the in-process monofilament and were removed by cropping of the affected
section. In total more than two hundred inclusions were detected using Eddy-current
inspection and subsequently removed. This tedious process delayed final delivery of the
outer strand by approximately 3 weeks and 6 weeks for the inner strand.
Figure 5 Micrograph of inclusion Figure 6 Micrograph of internal

removed from monofilament. inclusion at point of
initiation of typical cup and
cone fracture.
Subsequent analysis of virtually all multifilament wire breaks revealed internal

inclusions as the prevailing cause. A micrograph is shown in Figure 6 detailing a typical
break with an internal inclusion at the point of the cup and cone fracture. EDS
compOSitional analYSis of the internal inclusions agree with that of the monofilament
surface inclusions. It is obvious that without removal of the inclusions from the
monofilament, piece length of strand provided under this contract could have potentially
been less than half that which was actually delivered. In any case it can be assumed that
both piece length and yield would have suffered tremendously.
The characteristics for the superconducting corrector magnet wire called out in
the development specification are summarized in Table 7. This specification restricts
the deSign parameters to a 0.381 mm diameter strand with approximately 500
filaments at 9 J.Lm diameter resulting in a copper:superconductor ratio of 2.2:1. A
micrograph detailing the filament array of Supercon's corrector magnet wire as
manufactured to this specification is shown in Figure 7.
64
Table 7. Summary of SSC corrector magnet wire specification.
Parameter Specified Value
Wire Diameter 0.381 mm
Number of NbTi filaments 480 ± 30

Mean filament size ~ 9.0 ~m
Mean filament spacing
Cu:SC Ratio 2.2:1
Critical current (Ic) @ 5 Tesla ~ 100 Amps
Minimum piece length 2750 m
Figure 7 Cross sectional micrograph of SSC

superconducting corrector magnet wire
65
Ductility of the corrector magnet wire is a major concern for magnet
manufacturability. The Supercon design assures ductility by concentrating the niobium
filaments within a minimum radius from the center of the wire. Consistent with
mechanical theory, this approach minimizes stress within the wire by minimizing the
distance of the stiffer superconductor filaments from the neutral axis of the wire. This
is achieved by minimizing both the volume percentage of the copper core and the spacing
of the filaments to the extent allowed by the specification. An average spring back of
532 over 425 km of finished corrector magnet wire attest to the soundness of this
0
approach. With 100 percent testing of the delivered strand, all pieces comfortably met
the maximum 600 0 spring back specification.
Piece lengths yielded from the two billets of corrector magnet wire were actually
much better than the production data demonstrates. It was necessary to cut many lengths
of finished strand in order to facilitate processing through a final chemical cleaning step
which was specifically requested for this conductor by SSCL. This explains the unusual
piece length distribution in Figure 3 with no pieces longer than 17 km. If it had not
been necessary to cut these strands, piece lengths for the two production units of
corrector magnet wire would have averaged 16,320 meters. Supercon is currently
procuring equipment to allow processing of the longer lengths obtained in this conductor.
Critical current results for production unit 006 produced in Phase I met the
100 Amp minimum Ic specification for corrector magnet wire.(3) SSCL requested that
Phase II material be delivered with higher current density. Therefore in Phase II of the
project the process was optimized slightly to produce an average 2919 A/mm2 current
density at 5 Tesla. Phase III of the project is expected to include a current density
speCification of 3000 A/mm2 minimum and will require further improvement in
current density.
Comparison of the results from production units 006 and 011 suggests that the
higher J c goal is attainable without significantly sacrificing other key properties such
as piece length and ductility. The minor process changes initiated in Phase II resulted in
a 2.6 percent increase in Jc with no apparent decrease in piece length. Spring back data
suggest only a slight decrease in ductility coincident with this current density
improvement. Further process modifications are apparent which are anticipated to meet
the more ambitious J c requirements without sacrifice of the mechanical properties.
SUMMARY
The overriding priorities of the 1991 dipole conductor program were to improve
piece length and yield. Consistent with previously reported results (4) Supercon has
again excelled in meeting these challenges by delivering over 2,500 kg of dipole strand
(inner and outer) of which an average 94.5 percent have piece lengths longer than the
minimum required to avoid cold welds within a cable length. These results were
coincident with delivered yields approaching practical process limitations and record
delivery times. The ultimate result was the fabrication of 6 outer dipole magnet cable
lengths and 2 inner dipole magnet cable lengths with no cold welds for use in the
upcoming 50 mm dipole magnet string test.
A superconducting corrector magnet wire has been developed for the

Superconducting Super Collider which surpassed all performance specifications.
Critical current requirements were met while maintaining the superior ductility
necessary for manufacturing corrector magnet coils. Indications are that electrical
performance can be further increased without sacrifice of piece length, ductility or
manufacturability.
66
REFERENCES
( 1) D. Christopherson, D.W. Capone II, and D. Pollack, "Summary of the Performance

of Strand and Cable Produced for the 1991 SSC Magnet Program", Fourth Annual
IISSC, paper No. 111-8-1, New Orleans, March 4·6, 1992.
(2) D. Christopherson, D.W. Capone II, J. Seuntjens, D. Pollack, and C.R. Hannaford,
"Summary of the Performance of Strand Produced for the 1990 SSG Dipole
Program", Third Annual IISSC, Atlanta, March 13-15, 1991.
(3) SSC Development Specification SSC-A20-000002, Superconducting Wire for

Superconducting Corrector Magnets.
(4) D. Frost, J. Fernandez, G. Swenson, R. Randall, and J. Wong "Production

Experience in Manufacturing Superconductor for the SSG", Third Annual IISSG,
Atlanta, March 13-15, 1991.
67
GISIFIS DEVELOPMENT FOR THE sse
Aubie Oslin
PBIMK Team Design Division-Survey/GIS

7220 S. Westmoreland Road
Dallas, TX 75237
Martin B utalla
Conventional Construction Division

Superconducting Super Collider Laboratory*
2550 Beckleymeade A venue
Dallas, TX 75237
INTRODUCTION
Throughout all phases of the Superconducting Super Collider Laboratory (SSCL)

project life decisions will be made on how to manage complex interactions of components,
systems, and people with each other and with their environment at both micro and macro
scales. The SSC has a distinct advantage compared to other large projects constructed and
operated in the past, for even in the early phases scientists and engineers can use computer
technology to provide faster computation and better modeling capability to resolve conflict
and to make design, construction, and installation decisions. Computer systems today let us
go beyond just making pictures of the components and objects under consideration. Not
only can we know what objects look like, but we can visualize what and where they are,
how they fit together to make networks, and how the networks relate to other types of
objects and networks. Two such computer-based systems that relate object position and
attributes to one another are Geographic Information Systems (GIS) and Facility
Information Systems (FIS).
The SSC GIS is a computer-based system that provides a development, management,
and analysis environment for geographically referenced (mapped) features and associated
descriptive information (attributes). GIS is typically associated with features and networks
at the macro-scale, depicting information about the local geographic surface and sub-surface
environment. The mapped information is not stored in the conventional CADD sense;
instead a GIS stores the data from which one can create the desired map or view, drawn to
suit a particular purpose.
*Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

The House Appropriations Committee Report adopted June 13, 1990, identified GIS as
a means of achieving desired coordination between various public and private entities
involved in the planning, construction, and operation of the SSC. The Committee
recognized that GIS is not simply a computer system for making maps, although it can
create maps with colorful cartographic tools at different scales and in different projections.
GIS is an analysis tool by which spatial relationships between various map features are
identified, then used to resolve conflict, evaluate environmental impact, and help manage
interactions between man and project resources.
The SSC FIS is a global, computer-based repository for both physical and attribute
information related to the facility's vital technical and conventional components. FIS
provides a development, management, and analysis environment for the components and
systems required by all divisions of the SSCL for effective laboratory management
throughout the SSC's life. Similar to GIS but at the micro-scale, the component intelligence
of FIS resides within a relational data model in both graphic and text forms to support the
development of various applications throughout the project life cycle.
GEOGRAPIDC INFORMATION SYSTEM (GIS)
GIS stores information about a particular geographic area as a series of coverages or

layers, with each coverage representing a different theme or topic of information. Coverages
consists of both graphic and descriptive information. There are two distinct data models
used for graphic representation-vector and raster. Within the vector model, data can be
stored in an unstructured "spaghetti" form or in a topologically structured "intelligent"
form. A topologically structured coverage consists of points, lines, or areas that represent
features to which descriptive information can be tied. Graphic information stored in a raster
model includes images generated by satellite and airborne remote sensing systems, thematic
grid cell models, scanned images of documents, drawings or photographs, and video frames.
Using GIS, the two basic forms of graphic data can be brought together and combined to
produce new information. Figure I illustrates the combination of GIS vector and raster data
as output on the computer screen. The graphic and descriptive information present in each
original coverage can be integrated so that users of the GIS can develop applications and
perform spatial queries.
The PB/MK Design Division-Geodetic Survey and GIS Department began
developing the SSC GIS in July 1991. The following objectives have been adopted for GIS
development and implementation: 1) Integrate both existing-conditions data through the
survey and mapping program and design data through the CADD system with other federal,
state, and local data resources; 2) Create and maintain topologically structured maps and
geo-referenced imagery suitable for planning, engineering, geodetic survey, construction
management, facility and emergency management, public relations, and environmental
applications; 3) Define attribute categories and input data into the relational database
associated with the map information; and 4) Provide an efficient database query and
cartographic system interface for end users of the GIS.
Much of the GIS discussion presented here is taken from the comprehensive GIS Plan
that summarizes work performed to date and describes the computer system, data sources,
accuracy, applications, production, maintenance, budget, and project-wide coordination
necessary to meet contractual requirements and deliverables according to three project
phases: I-Development; II-Implementation; and III-Operation and Maintenance.
70
Phase I-GIS Development
GIS and image-processing software have been installed on UNIX-based workstations

that interface through Ether networks with the Intergraph CADD system being used for sse
facility design and drawing production. Intergraph, Arc/Info, and ERDAS software are
being used to develop two forms of spatial data: the vector form that represents map
features as topologically structured points, lines, or boundaries, and the raster image form
that represents information as pixels (picture elements) or grid cells. Attribute or descriptive
data is maintained in a relational database with ties to the graphic information.
Figure 1. Combination of a 1991 Spot satellite image registered to project control and vector information from
the LIS.
Regional (1: 100,000 or 1:250,000), Planning (1 :24,000 or 1" = 2000'), Vicinity

(l :4800 or 1" = 400'), Engineering (l :480 or 1" =40'), and Land Information System (LIS)
base maps are being developed so that attribute and other map information can be managed
at various levels of accuracy. Figure 2 depicts the extremes of geographic base maps being
developed from the regional North Central Texas base to the project-specific LIS.
71
Figure 2. Layers of base map development from the regional level showing surrounding counties to the LIS
parcel base map.
Base mapping and coverage development has begun on all levels, satellite imagery is
on-line, and the digital orthophoto coverage is planned to be available by summer of 1992.
Table 1 lists coverages identified for SSC GIS implementation. Coverages are in various
stages of development. The initial engineering base map will soon be complete. LIS land
parcels have been transmitted from the Texas National Research Laboratory Commission
(TNRLC), and ownership and access agreement information is being associated with each
parcel. Several attribute categories have been identified and are being associated with
appropriate map information. Other attribute information will be identified through a User
Needs Assessment conducted by the PB/MK GIS Department with SSCL coordination. GIS
is an evolving computer-based system; as particular coverages and applications come on-
line, others will be in development. Maps, GIS analysis, and electronic data can be
transmitted as information progresses from development through QA/QC. The
implementation date listed in Table 1 is the estimated date by which the coverage will have
been edited and topologically structured. At this point coverages are ready for additional
attribute assignment and application programming.
Phase II-GIS Implementation
Implementation represents an intermediate phase in SSC-GIS evolution, when priority

GIS applications become operational at the same time new applications are in development.
Database design becomes critical during this phase, and SSCL coordination is extremely
important as the Functional User Requirements, System Specifications, and Data Dictionary
72
Table 1. GIS coverages identified for development and implementation.
Implement
Coverage Description Data Source Attributes Date
Land Information S~stem (LIS)

Geodetic Network- PBIMK, RTK& LSCS & NAD83 ST PL (x,y,z) 7/92
Monumentation USGS
SSC Lattice SSCL,PBIMK to be determined 6/92
Fee Simple & Stratified Fee Land TNRLC to be determined 1/92
Land Parcels TNRLC access, ownership, env. data 1/92
Utility Easements & R.O.W. TNRLC, PBIMK classification, ownership 7/92
Surveyed footprints PBIMK structure description 7/92
En~ineerin~ base maE (I" = 40')

Planimetric PBIMK fences, roads, structures, trees, 4/92
water
Topography PBIMK 5' index & I' intermed. contour 4/92
Digital Elevation Model (DEM) PBIMK spot elevations-TIN 4/92
Land Management PBIMK existing & planned practices 7/93
Geotech Borings PBIMK,RTK sample properties & profile 10/92
Vicinit~ base maE (I" = 400')
Digital Orthophotography PBIMK NAD83 ST PL coord. (2' pixel) 7/92

Street Network PBIMK name, class, restrictions, add. 9/92
range
Bordering Land Parcels ECAD,PBIMK parcellD 1/93
Rural Residences Ellis Co. 911, PBIMK name, address, phone number 2/93
Env. Baseline Radiological SSCL ES&H, PBIMK Monit. station !D, data type, 7/93
prop.
Hydrology-Wetlands PBIMK, JJR class, name, quality, hydraulics 7/93
Land Use-Land Cover PBIMK, JJR Anderson Classification 1/93
System*
Archaeological SSCL, SMU, JJR to be determined 7/93
73
Table 1. GIS coverages identified for development and implementation (Continued).
Implement
Coverage Description Data Source Attributes Date
Planninll base maE ~1" =2000')

Ellis Co. Street Network TNRLC,ECPW, name, class, restrictions, add. 4/92
PBIMK range
Bridges TNRLC,ECPW class, load limits, restrictions 5192
Util.(elec, gas, product, water, TNRLC class, owner 3/92
phone)
Surf. Hydro-Watersheds-Flood PI. USGS, PBIMK, class, name, quality, hydraulics 5193
FEMA
Ground Water BEG,TWC name, quality, hydraulics 10/92
Water Wells BEG,TWC ID, quality, depth, class, owner 3/92
Oil & Gas Wells TRRC ID, class, status 10/92
Surface Geology BEG name, properties 8/92
Sub-surface Geology BEG name, properties 7/93
Soils TNRLC, SCS, JJR name, properties 10/92
Political Boundaries - E911 TNRLC, Ellis Co. 911 name, characteristics 1193
Public Service & Emer. Facilities Ellis Co.NCTCOG,JJR name, phone no. 7/93
Extended Rural Residences Ellis Co. 911, PBIMK name 7193
Env. Baseline - Air TACB, EPA, FAA to be determined 12/93
Pt. Discharges Water Intake EPA, TWC, NCTCOG name, class 12/93
Extended Land Use - Land Cover PBIMK, NCTCOG, Anderson Classification System 8/93
JJR
Comprehensive Cty Plan- Zoning JJR, Ellis Co. class, restrictions, status 4/93
Rare & Endangered Sp. TP&WL,JJR name, status 4/93
Rellional ~1:100,000 & 250,000)

Regional Street Network USGS, name, class 4/92
NCTCOG,Tiger
Traffic Survey Zones NCTCOG traffic model 10/92
Political Boundaries - E911 NCTCOG,JJR name 1193
School Districts NCTCOG,JJR name 10/92
Census Tracts NCTCOG,UTD Tract ID, demographics 4/92
Zip Codes NCTCOG ID 4/92
Regional Land Use NCTCOG, UNT, JJR Anderson Classification System 10/92
Surf. Hydro. - Watersheds USGS, NCTCOG, JJR name, class 1193
Ground Water BEG, JJR name, quality 10/92
Digital Elevation Model (DEM) USGS elevation - TIN 10/92
Surface BEG,JJR name, properties 8/92
Sub-surface Geology BEG name, properties 8/92
General Soils SCS, JJR name, properties 10/92
74
documents are written. Database query, cartographic system interface, and menu options
will be developed during this phase so that efficient links to other technical systems are
provided for SSC facility management.
Priority GIS tasks include the implementation of: 1) the LIS, which defines land
parcels, SSC real property requirements, right of way, and existing utilities easements so
that access agreements, permits, and ownership information can be tied to appropriate land
parcels; 2) thematic maps such as updated topological street network for Ellis County, road
and bridge restrictions, potential construction access routes, geotech and soil information,
hydrology, land use, political boundaries, utilities, rural residences, and environmental
resource data for planning and construction applications; and 3) GIS to CADD and FIS
interface standards.
The GIS Functional User Requirements Document will be delivered upon completion
of the users' needs assessment. System Specifications and Data Dictionary documents will
be prepared in coordination with the SSCL. Phase II-GIS Implementation is planned to
begin in FY 1992 for the priority GIS applications listed above, while other GIS operations
are in development. GIS implementation is projected to continue through FY 1995, when
most of the identified coverages and applications will have undergone quality assurance and
control testing.
Phase III-GIS Operation and Maintenance
The PB/MK Geodetic Survey and GIS Department will provide effective transfer of
GIS services in a timely manner to end users of the SSe-GIS. Quality Assurance and
Quality Control programs are planned to ensure data integrity and correct operation and use
of the GIS. The PBIMK GIS staff will assist the SSCL staff in operation and maintenance of
the GIS during the transfer stage. Operation and maintenance procedures for GIS
applications, base maps, thematic maps, and attribute data begin when QAlQC programs are
in place for any particular GIS function. GIS Operation and Maintenance could begin as
early as FY 1993 for critical applications such as construction management, base map
updating, and CADD interface functions. Operation and maintenance tasks will be
performed by the PB/MK GIS group until the SSCL is prepared to take over operation. GIS
hardware, software, and database will be delivered in its entirety to the SSCL upon project
completion.
The SSC-GIS is an evolving computer-based system that will interface with other
technical application systems at PBIMK and the SSCL. However, GIS differs from many
computer development projects in that as geographic information comes on-line, users can
access the data in reference to other mapped features and can immediately perform spatial
analysis. With planned implementation and end-user coordination, GIS can be used
effectively during the early design and construction phases as well as in the installation,
operation, and decommissioning phases of the SSe.
FACILITY INFORMATION SYSTEM (FIS)
The SSC FIS is an integrated repository for the vital component intelligence of the
SSCL. The repository consists of information and related applications accumulated during
five major lab life cycle phases by each division working independently in its own arena of
responsibility. The information will be coordinated and integrated globally, supporting each
participating division's access directly to its own subset of information, and indirectly to all
other pertinent information lab-wide. Each participating division throughout all phases of
the lab life cycle are able to update, use, and maintain the information for which it has
responsibility.
75
As both conventional and technical systems components are designed, the information
regarding each component is collected and stored electronically. The evolution represents a
process of information collection and use that began during design of both conventional and
technical systems and progresses through the phases of construction, installation, operation,
and maintenance. At each phase, the "deliverable" is envisioned to be an electronic base of
data resident on Intergraph hardware with other UNIX-based hardware support, accessible
to participating groups and divisions through the use of Intergraph and other application
software, and made available to the SSCL via Ethernet network and Intergraph workstations
as well as PCs and Macintoshes.
Within the FIS database each component is guaranteed uniqueness so that pertinent
information is uniquely retrievable for each component. An SSC working group is defining
global identification of components. Every technical and conventional vital component will
be assigned a global component identifier that will remain a part of the component for its
duration of service to the Laboratory,
Figure 3 represents a sampling of applications identified as requirements for facility
management. DESIGN DATA consists of all of the design information relevant to the
selection and/or fabrication of the component. COMPONENT GEOMETRIC
DEFINITIONS comprises the geometry of each component for a 3-D graphical display of
its image on any CADD or graphical display system. OPERATIONAL
CHARACTERISTICS is a set of significant information related to the operation of each
functional component or system. This information could be used to determine the steady-
state operating condition of any component compared to an abnormality or emergency
condition. PROCUREMENT contains information and documentation relative to the
procurement trail of events for each component. VENDOR SELECTION AND
PERFORMANCE documents the selection trail for service and/or supply vendors. It is also
a log of all vendors' performance history and comparison ratings for future selection criteria
for each component. WAREHOUSING contains status information regarding the
maintenance, repair, and storage of each component that is out of commission and/or in a
storage redundancy state. When a magnet is removed and replaced by another, the removed
magnet may be shipped for repair, at which time a new location set of coordinates would be
assigned. Later, when warehoused, a ready state and new location would be assigned to the
magnet. INSTALLATION SCHEDULING reflects any necessary data pertinent to
installation procedures, problems, practices, and precautions for any component. A
HAZARDOUS MATERIAL CLASSIFICATION is issued for each pertinent component
containing hazardous material or having a potential of environmental contamination. The
classification would be a secondary key that connects the GIS base of information for
emergency response and environmental impact studies. MAINTENANCE SCHEDULING
represents information relevant to the history of a component's maintenance, its
maintenance records, and maintenance scheduling data necessary for scheduling and
budgeting future preventive maintenance activities. It also would assist in coordinating
routine preventive maintenance of different components in a congested area. OPERATING
HISTORY contains information relative to the operational track record for each component
and its expectations under various conditions. COMPONENT GEOGRAPHIC
COORDINATES is a link to the GIS base of data and uniquely identifies each component
and system with a global coordinate system location. Two coordinate systems currently
exist for objects that are located either above or below the surface. DRAWINGS,
DOCUMENTS, SPECIFICATIONS constitutes our lab-wide technical document control.
Each component appears on one or more technical drawings and/or documents, or is
referenced in specifications. The document control number for each reference to the
component becomes an attribute of the component in the document control fields of the
database.
76
Figure 3. Required applications for Facility Management.
SUMMARY
Facility management for a project of the size and complexity of the SSCL is a
challenging task. The Facility Information System/Geographic Information System
(PIS/GIS) should provide an effective tool for the demanding work ahead. Both the FIS and
GIS encompass information that many potential users across multiple disciplines will
require for effective facility management. FIS will be integrated with the GIS for
applications that involve duplicate needs of graphic and attribute data. In particular,
infrastructure networks, environmental monitoring, emergency dispatching, and hazardous
materials management have been identified as areas where the two systems will interface. In
general, the GIS will manage graphic and attribute information outside the actual structure
of the SSCL. The FIS will take over operation of components and networks within the
SSCL facility. By providing a method for informed decision-making, implementation of the
SSC FIS/GIS will facilitate the tasks involved in managing our Laboratory during all phases
of its life.
77
SUCCESSFUL NEPA COMPLIANCE AT THE SUPERCONDUCTING
SUPER COLLIDER LABORATORY: A CASE STUDY
Bruce C. Corning
EG&G W ASC, Inc.

Rockville, MD 20850
Richard G. Wiebe

Dallas, TX 75237-3997
INTRODUCTION
On January 1, 1970, the President signed the National Environmental Policy Act (NEPA) into law.
NEPA has become the basic policy-setting federal law relating to protection of the environment and
has provided the initiative for passage of other federal and state environmental statutes. Although
many of these statutes have unique requirements. there is a need to coordinate NEPA compliance with
review requirements of the other environmental statutes in order to avoid delays that can be caused by
proceeding separately under each statute. Because of its multi-purpose scope, the NEPA process is an
excellent means for accomplishing the required coordination.
The Director of the Superconducting Super Collider Laboratory (SSCL) has committed the
Laboratory to Total Environmental Compliance. Environmental Compliance involves a dynamic set of
factors--requiring system maintenance with integrated planning and control-that by design will
identify requirements, ensure implementation of mitigative actions, track follow-on efforts, and plan
for future requirements.
The Record of Decision to proceed with the building of the SSC required that several mitigation
actions be addressed. Identifying these requirements, their sources, and whether they can be
addressed within the context of existing policies and procedures is required to ensure appropriate and
timely mitigative actions. Applicable requirements may include federal, state, and local regulations,
applicable Department of Energy (DOE) Orders, best management practices, Laboratory
requirements, and the adequacy and effectiveness of DOE and contractor management programs.
Mitigative action is a principal aspect of total environmental compliance, conducted at all levels of
the Laboratory, not just as an environmental function. Identified requirements are prioritized. Goals
and objectives are set for implementing and successfully completing each mitigative action. Feedback
mechanisms required for tracking the progress of each action are developed.
*Operated by the Universities Research Association. Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

Technical assistance and environmental support is provided and/or managed by sub-contract. A
scope, schedule, and cost estimate are developed for each mitigation action. Execution priorities may
need adjustment due to fiscal, regulatory, or time constraints.
Throughout the mitigation action process, it is critical to document whatever measures have been
taken. This will (1) provide an audit trail, and (2) provide backup information in support of any
regulatory requirements. Continual interface with the regulatory community is necessary, as ultimately
concurrence on any mitigative action that requires permits or specific regulatory review will be
required.
Compliance occurs on at least two levels. As each potential impact is addressed through completion
of a mitigation action, compliance is obtained. When all mitigative actions are addressed through
completion, !program compliance is achieved. Throughout this process, regulatory approval at all
levels is essential for claiming compliance. Compliance in and of itself may not require any specific
action; however, remaining in compliance may require follow-on activities. Therefore, compliance
should not be perceived as a stopping point, but rather as a point in the cycle.
Compliance maintenance is the mechanics of tracking progress on the actions and any follow-on
activities associated with them. Feedback occurs throughout the process. Tracking may include
permitting status, monitoring results, training, reporting requirements, meetings, documentation, and
program budget execution.
Throughout the process of achieving and maintaining compliance, it is necessary to continuously
integrate requirements identification, mitigative action, and compliance maintenance. This is being
accomplished by staff of the Environmental Affairs Office in concert with the Environment, Safety and
Health function through policy development, planning and programming, oversight, and staff
coordination.
Policies are developed to incorporate legal and regulatory requirements with Laboratory objectives,
and they are promulgated by the appropriate authority. Planning provides the strategy for
implementing the policies, and programming provides the fiscal resources for implementing the plan.
Oversight is the control mechanism for determining the degree to which tbe plan, as funded, is
implemented.
The Environmental Affairs Office at the SSCL provides the framework for an integrated planning
and control function for achieving and maintaining total NEPA compliance by: (1) tracking progress
on currently identified requirements, (2) continuously examining current operations to identify new
requirements, and (3) identifying future requirements based on proposed federal, state, local, and/or
organizational rules.
COMPLIANCE ACTUALIZED
Environmental Affairs Office
The Environmental Affairs Office (EAO) of the SSCL was established in the Project Management
Office (PMO) for the primary purpose of representing and monitoring the SSC for NEPA compliance
issues and providing oversight of Mitigation Action Plan compliance during construction and start-up
operations. The initial function of the EAO was to collect and coordinate the data for and support the
preparation of the Supplemental Environmental Impact Statement (SEIS). This role has evolved into
the coordination and oversight of subsequent NEPA activities resulting from the SEIS process, such
as the development of the Mitigation Action Plan and its 38 related documents.
As one of the first organizations established as part of the SSCL, the role of the EAO has been to
provide the statute-required NEPA support during the planning, design, and construction of the SSC
and associated facilities. This support is to establish, coordinate, and manage a base of technical
experts and contacts that can collect, review, manage, and interpret environmental data.
The EAO is composed of a cadre of experts who have the historical knowledge and experience with
the SSC project from the beginning. The mission of the EAO is to provide management of
environmental issues as they arise and to provide a vast technical resource that is capable of resolving
these issues.
80
Responsibilities
Several elements of the environmental impact analysis process influence risk to the decision maker.
These elements include proposed action and alternatives, schedule, budget, affected environment,
agency consultations, permits, potential impacts, and public involvement Each element can be viewed
from both an operational perspective and an environmental perspective. The outcome of the
environmental impact analysis process may reveal that the operationally preferred alternative and the
environmentally preferred alternative are diametrically opposed; that is, the action as originally
described to accomplish the mission may significantly degrade the environment. The decision maker
has an option to choose either alternative as long as the environmental impact analysis process is
properly performed and potential impacts are accurately quantified. Most importantly, the decision
maker must have sufficient information to understand the environmental consequences of a decision.
This does not, however, give the decision maker the authority-by taking actions out of context of the
original-to violate other environmental statutes later in the project
Compliance with NEPA allows the program managers and other decision makers to maintain
schedules and to meet environmental laws. The NEPA process can be viewed as a balance between
risk of program delays and the time necessary to properly achieve environmental compliance. Prudent
decision makers can balance the risk by planning and budgeting, preparing documents, consulting
with other agencies, committing to mitigation measures, obtaining permits, and establishing an
administrative record. The NEPA process is a proactive mechanism for ensuring that programs are
prepared for a life of environmental compliance. The decision makers at the Lab have worked closely
with the EAO during all the phases of the NEPA process, which has ensured that the program remains
on an environmentally safe course from cradle to grave. Program managers and other decision makers
should use the NEPA process to minimize risk and to maximize successful program completion.
The EAO functions to assist the PMO decision makers in ensuring compliance with NEPA
throughout the construction phase of the project.
The primary responsibilities of the EAO are to:
• provide technical expertise on NEPA compliance issues

• serve as liaison between the Laboratory, the Department of Energy, and external agencies for
resolution of environmental compliance issues related to construction
• act as a single point of contact for the PMO on environmental issues
• participate in working-group design review compliance
• provide independent and objective reviews of issues that reflect environmental concerns
• monitor and update SSCL policies and procedures when changes in NEPA compliance orders
and directives occur
• enhance SSCL staff capabilities by providing short-term, expert consultation on
environmental topic areas
• support the Change Control Board with respect to possible environmental impacts of
deviations from the assessed baseline.
NEPA COMPLIANCE ISSUES

Technical Expertise
As a prescription, NEPA serves as the focus for environmental review planning and as an integral
part of "phased compliance," the term given to a comprehensive, integrated environmental planning
strategy (DOE Order 4700.1). The EAO provides guidance and technical expertise to the SSCL to
assure compliance with the NEPA process.
A NEPA Compliance Audit Protocol, developed by the Office ofNEPA Oversight (EH-25), will be
used to guide the assessment in the following areas:
81
• Overview of NEPA Issues
• Management Structure
• NEPA Compliance Planning
• NEPNCERCLA and NEPA/RCRA Integration
• Detennination of the Level of NEPA Review Required
• Procedural Aspects of NEPA Documents
• Technical Content of NEPA Documents.
The EAO serves as the lead to the SSCL in providing NEPA support. As the focus for NEPA
compliance, the EAO develops the necessary documentation, infonns the PMO of environmental
regulatory requirements, and assists in the compliance with applicable environmental regulations.
Specific technical experts are used as required. Using this cadre of technical experts, the EAO
coordinates NEPA issue resolution, manages technical input, and provides the PMO with an efficient
means to NEPA compliance.
Design Reviews
The EAO will provide guidance to SSCL working groups to ensure NEPA compliance. To meet the
environmental protection requirements, the SSCL has adopted a general "as low as reasonably
achievable" (ALARA) approach to the design, construction and operation of the SSC. Under this
approach, nonnal operations of the SSC will produce environmental impacts at or below regulatory
thresholds. This approach simplifies the compliance process and provides a margin for compliance to
future environmental protection requirements. In addition, all SSC facilities will be designed to
minimize waste through reduction and recycling, and to minimize the exposure of workers and the
public to any hazards.
One significant area ofNEPA compliance relates to the "Conventional Construction" aspects of the
project. Due to the scope and scale of the planned facility, there are a great many components related to
construction that continue to evolve and develop to a higher order of final design. It was not feasible to
have a complete fmal design available on which to base the SEIS; therefore, "Mitigation by Design"
has been a very effective method of assuring progress without fully defining the complete action.
A "worst case" scenario was developed for a number of elements and situations that define the
envelope of acceptable alternative considerations. Each time an activity advances through a stage of
design or is adjusted as a result of improved technical definition or increased knowledge of the
environment, the "original" assessment is re-visited to detennine any associated deltas which would
exceed the expected/identified impacts. When a consequence of design creates an impact of greater
negative impacts, the design is challenged to detennine if an alternative of lesser impact exists. In
many areas, design evolution has led to impact avoidance by adjustment to facility plans and location.
Of significance in this area is the relatively small infringement upon wetlands and floodplains, which
resulted in the following statement by the U. S. Anny COlps of Engineers:
Our review of the proposed SSC project, and the planning efforts of the DOE and the
TNRLC to date, reveal a strong commitment to avoiding impacts on the natural
environment, particularly waters of the United States, to the maximum extent
practicable. We commend you for those efforts, and encourage you to continue that
commitment throughout the planning, construction, and operation of the project.
The day-to-day process for compliance of Conventional Construction activities involves a number
of participants. These include participants from DOE, the State of Texas, several SSCL departments,
and the Laboratory's subcontractors. The process involves a number of iterative steps that function
through a facilities working group. The working groups are chaired by a team leader who represents
the primary "customer" or user of the facility. Design requirements are generated in the working
group, with continual reviews for compliance with regulations and commitments made in the Final
Environmental Impact Statement and the SEIS, and in their respective Records of Decision. The
commitments are summarized and tracked through the Mitigation Action Plan. Once the working
group has reached development of final technical requirements for a facility, these technical
82
requirements are developed into a Design Requirements Document (DRO) for incorporation with a
Notice to Proceed to the Architect-Engineer/Construction Manager (A-FJCM) subcontractor. The
A-E/CM then develops a draft list of pennits and applicable commitments and regulations that apply to
the specific facilities on the basis of an analysis of the DRD. TIlese are presented for review and
acceptance by the Environmental Compliance Committee. chaired by DOE. The A-FJCM develops a
draft Environmental Compliance Plan (ECP) for the facilities. The ECP is reviewed and accepted by
the SSCL once it is considered complete. The ECP is then utilized as a tool for the designers to follow
for detailed design and development of construction documents and specifications. The process is an
iterative one in which modifications are continually monitored and re-evaluated as necessary to
maintain the commitments.
Due to the large scale of this project and the nature of a one-of-a-kind facility. there is also a need to
evaluate emerging opportunities to advance the state of design. These often will include an assessment
of the compliance aspects related to NEPA. The EAO is provided technical reports on preliminary
conceptual studies that the Project Management Office has deemed worthy of evaluation. The EAO
then runs though the inventory and environmental analysis categories covered by the Impact
Statements. The various impacts are quantified, and an opinion is developed as to whether the concept
was satisfactorily covered in the Impact Assessment. This process relies on a detailed depth of
understanding of earlier analysis. The continuity of the EAO. from the early stages of the project to the
present. is a tremendous asset to the process. When thresholds are perceived to be violated. working
sessions are held with the technical groups and designer to (1) modify the concept to eliminate the
violation, or (2) develop mitigation strategies and applications. The process has been very effective in
maintaining compliance. minimizing potential impacts, and documenting the procedure.
Monitoring of Environmental Issues

As part of each working group of the SSCL, the EAO is able to assure full compliance with all
federal, state. and local environmental regulations. In fulfilling this oversight and monitoring
responsibility, the BAO will:
• assist the PMO in detennining environmental requirements and impacts to the project and in
coordinating the respective planning efforts, and will participate. to the extent requested. in
the regulatory process
• assist in the planning and implementation of PMO compliance activities
• assist in ensuring compliance with all applicable environmental, human health. and safety
statutes, standards. and regulations
• act as a liaison among federal, state, and local regulatory agencies and respond to public and
organizational interests and concerns.
83
ARCHAEOLOGY OF THE SUPER CONDUCTING SUPER COLLIDER
J. M. Adovasio, 1,2 R. Buyce,2 and David R. Pedler 1
1Archaeology Research Program

Department of Anthropology
Southern Methodist University, Dallas, Texas 75275
2Mercyhurst Archaeological Institute
Mercyhurst College, Erie, Pennsylvania 16546
INTRODUCTION
The Superconducting Super Collider (SSC) Project is one of the largest scientific endeavors
ever undertaken in the continental United States. As such, it has attracted massive amounts of
media attention and commentary. Despite its high profile, certainly one of the least publicized
aspects of this vast enterprise is the substantial archaeological and historical research associated with
the planned construction.
Since October 1989, the Archaeology Research Program (ARP) of the Department of
Anthropology, Southern Methodist University, has been continuously involved in a carefully
planned, multi-disciplinary research operation to systematically identify, evaluate, and if necessary,
conduct extensive data recovery on any archaeological or historic resources located within the direct
or indirect impact areas of the project.
As is well known, the Superconducting Super Collider Project is under construction by the
Department of Energy in conjunction with Universities Research Association, Inc., the
Superconducting Super Collider Laboratory, and the Texas Natural Resources Laboratory.
Construction of this facility will directly affect ca. 4,047 ha (10,000 acres) of Ellis County, Texas,
and may indirectly impact an additional 4,047 ha (10,000 acres).
Since the Superconducting Super Collider Project is partially funded by the federal
government, the archaeological and historic resources within the direct and, to some extent, indirect
impact zones are protected by a variety of federal laws and regulations. These include but are not
limited to the Antiquities Act of 1906 (34USC225); the Historic Sites, Buildings, and Antiquities
Act of 1935 (49USC666); the Archaeological and Historical Conservation Act of 1974
(88USC174); the National Historical Preservation Acts of 1966 and 1980 (80USC915 and
94USC2987); the National Environmental Policy Act of 1969 (83USC852); Executive Order
Number 11593 of 1971 (36CFR9821); and the Archaeological Resources Protection Acts of 1979,
1980, and 1988 (94USC2987, 102USC2778 and 102USC2983). All of these regulations direct
federal land managers to implement cultural resources inventory, assessment, and where
appropriate, mitigation (Le., full data recovery) programs in consultation with the State Historic
Preservation Office (SHPO), the Advisory Council on Historic Preservation, and other interested
parties.
Within the context of the Superconducting Super Collider Project, the lead agency responsible
for compliance with extant federal and state cultural resources protection laws and guidelines is the
Department of Energy. This responsibility is shared with the Idaho National Engineering

Laboratory Commission, EG&G Idaho, Inc., the Superconducting Super Collider Laboratory, and
the Texas Natural Resources Laboratory Commission.
REGIONAL SETTING AND IMPACT AREAS
The SSC site is located entirely within Ellis County just south of the Dallas-Fort Worth
metropolitan area, and the present landscape consists primarily of rural farmlands (Jurney, Weston,
and Chada 1990). Again, as is well known, the finished SSC will be a ring some 87 km (54 mi)
in circumference and will completely encircle the town of Waxahachie. The SSC study area is
situated within the B1ackland Prairie, which in turn is located within the northwestern part of the
West Gulf Coastal Plain physiographic province (Fenneman 1938). Broad, rolling uplands
characterize the topography with generally low elevations that range from 121.9 m (400 ft) to 228.6
m (750 ft) above mean sea level (msl). In the western part of the study area, which is underlain
by the Austin Chalk, limestone outcrops are exposed on the crests of broad hills. The eastern
portion of the study area, underlain by the soft, easily eroded shales of the Taylor Marl, is maturely
dissected and the rolling hills are cut by severe gullying.
From a geological perspective, perhaps the most prominent feature of the surficial geology of
the study area is the so-called Uvalde gravels. These gravels occur as discontinuous erosional
remnants in the residual bedrock soils of the area, and consist of thin veneers of water-rounded
cobbles and pebbles of quartzite, jasper, chert, opalized wood, and rock fragments. Aboriginally,
these gravels would have provided a ready source oflithic raw materials for stone tool manufacture
and may have been one of the primary attractions to this otherwise relatively inhospitable area.
The "footprint" of the SSC lies within the Trinity River Drainage basin. The Trinity River
has its headwaters west of the SSC area, and it flows in a southeasterly direction that would carry
it just to the south of the study area along the present course of Chambers Creek (Figure 1).
Instead of continuing its southeasterly trend, however, the Trinity River presently veers off to the
northeast before reaching the SSC impact area and bends again at Fort Worth, flowing east past
Dallas before turning south to skirt Ellis County on its eastern edge. Nearly all of the streams in
the northeastern portion of the project area flow to the southeast and empty directly into the Trinity
River. The exceptions to this pattern occur in the southwestern portion of the study area, where
several small streams flow south into Chambers Creek, which is now a southeasterly flowing
tributary of the Trinity River. The drainage divide extends from the West Campus portion of the
SSC to penetration shaft F8, north of Avalon. It is likely that in earlier times the drainage of the
Upper Trinity River flowed directly through the Chambers Creek Valley without the present detour
through Fort Worth and Dallas. This detour may well have been occasioned by stream piracy (cf.,
Jurney, Weston, and Chada 1990). The present underfit nature of Chambers Creek supports this
reconstruction.
The present streams in the SSC project area are generally of low gradient but often have steep
banks. Intermittent streams predominate with only a few perennial (base-flow) streams carrying
water through most of the year. Of the major streams in the project area, only Red Oak Creek,
Waxahachie Creek, Grove Creek, and Big Onion Creek have built substantial floodplains, all of
which occur along the lower reaches of these streams in the eastern part of the SSC footprint.
Chambers Creek, which parallels those streams to the south of the project area, has the largest
floodplain in the region. Quaternary alluvium has been mapped along each of the major streams,
and fossil floodplain deposits have been mapped as discontinuous terraces which most commonly
flank the Quaternary alluvium. Reconstruction of the ancient drainage patterns, which is a focal
part of the ongoing SSC archaeological and geoarchaeological research (see Research Issues,
below), is expected to help predict the location of fossil floodplain deposits that might contain
deeply buried archaeological sites and thereby characterize the prehistoric landscape utilized by the
study area's earlier populations.
The principal construction-related impacts to the Ellis County landscape will largely be
confined to fee-simple purchase areas for the East and West campuses, shaft penetrations, and
cooling ponds. Most of the 87 km (54 mi) ring will not produce any direct impacts to the land
surface, nor will it effect any archaeological or historical resources on that surface, since the SSC
will be constructed by boring a bedrock tunnel over 30.5 m (100 ft) below the modern ground
surface. The most-extensive surficial impacts, and hence those areas requiring the greatest
86
hlte Rod Escarpment
5 o 5 10 15 o MILES
..... - ~
5 0 5 10 15 20 KILOMETERS
~
N
Figure 1. Stream drainages in the SSC project area, Ellis County, Texas. Note the course of the Upper Trinity River (upper left) as compared to that of Chambers Creek
and the Lower Trinity River.
CD
~
archaeological and geoarchaeological scrutiny, are within the West Campus (ca. 3,035.4 ha; 7500
acres), East Campus (ca. 607.1 ha; 1,500 acres), and Ellis County Farm (ca. 182.1 ha; 450 acres)
areas. In addition, 25 shaft penetration areas that are presently planned will ultimately cover an
area of ca. 508.3 ha (1,256 acres). Some roads will also be improved and ca. 8 km (5 mi) of new
road construction is planned. Other roadways will be upgraded and improved and additional
utilities will be provided. Finally, any area under the SSC footprint which has the potential for
preserving deeply buried archaeological sites is considered to be an area of direct impact. All of
the areas specified above have received or will receive intensive reconnaissance to identify
potentially significant archaeological or historical resources (see Summary of Work Completed to
Date, below).
CULTURE mSTORY OF THE STUDY AREA
From an archaeological perspective, the sse project area has not been intensively studied prior
to this project. Nonetheless, based on research conducted around the study area (e.g., Prikryl
1987; MacGregor and Bruseth 1987; Peter and MacGregor 1988), the broad chronological outlines
of its culture history are reasonably clear. Generally speaking, archaeological and historic data
presently indicate that the sse project area has been intermittently used as a locus for human
habitation or activity for a minimum of 12,000 years. Specifically, these data indicate that all of
the cultural periods/stages currently known in this portion of Texas are represented in the general
study area (Table 1). In short, the culture history of the sse project area spans, for all intents and
purposes, the entire known chronology of human occupation in this portion of Texas.
Though time precludes a detailed treatment of what is and what is not known about each of
the sequent cultural periods represented in the sse project area, certain aspects of its cultural and
physical setting should be stressed, as they have directly influenced the research design that guides
this project. First, while the broad brushstrokes of sse study area's prehistory are visible, most
of the "finishing touches" or critical details of this 12 millennia-long portrait are either imperfectly
perceived, masked, or totally invisible.
Second, based on the research conducted to date, it is our consensus that the sse study area
is marginal, both geographically and culturally (i.e., at least until the coming of the railroad in
1871). We stress that our employment of the term "marginal" is intended in a non-pejorative,
strictly archaeological sense, and is meant to convey the notion that the study area is simply outside
of-or, more accurately, peripheral to-the mainstream of socio-cultural development witnessed
through prehistoric and historic times in adjacent portions of Texas. In the specific context of the
SSC study area, "marginality" describes a condition or set of interactive conditions in which both
the archaeological manifestations and the settings in which they occur are different in degree and
kind from their counterparts to the north, southwest, and east. On the basis of the densities and
distributions of artifacts on the modern ground surface, the archaeological sites identified to date
are ephemeral and seem to reflect transient utilization by highly mobile groups for very short
periods of time. Moreover, the array of distinct behavioral sets or activities reflected by these sites
seems to be highly circumscribed. Finally, as opaque as our picture of the prehistoric inhabitants
of the study area is, our view of its prehistoric topography and ecology is an even darker shade of
pale.
Table 1. Cultural Periods/Stages Represented in the Superconducting Super CoIlider Project Area, Ellis
County, Texas (from Prikryl 1987:104-132)
Cultural Time Range Cultural Time Range

Period/Stage Period/Stage
Paleo-Indian ?-6000 B.C. Early Ceramic 200 B.C.-A.D. 800

Early Archaic 6000 B.C.-4000 B.C. Late Ceramic A.D. 800-A.D. 1600
Middle Archaic 4000 B.C.-2000 B.C. Proto-Historic A.D. IS36-A.D. 1820
Late Archaic 2000 B.C.-200 B.C. Historic Period A.D. 1821-Present
88
RESEARCH DESIGN
To address the themes outlined above and a host of related issues, the ARP has formulated and
refined a research design for the SSC project which is directed toward defining, describing, and
explaining not only the activities of the various prehistoric "actors" in the SSC project area, but
also the history of the "stage" upon which they performed.
However, due to the marginality of the study area and the concomitant relative lack of
archaeological data from the region, especially in regard to prehistoric times, the SSC
archaeological work is not primarily designed to validate or refute prior culture-historical models,
nor are we primarily concerned with generating new processual perspectives on the study area.
In this respect our approach is a relatively simple one, in that it is focused upon the rigorous
application of sound, modern archaeological and geoarchaeological theory and practice, both in the
field and throughout the analysis and reporting phases of the project. Specifically, and taking our
lead from the highly successful, multi-year, multi-disciplinary, Meadowcroft/Cross Creek
archaeological project (Carlisle and Adovasio 1982)-which is generally considered to be the state-
of-the-art in contemporary multidisciplinary archaeological research (Fagan 1990:20)- the central
theme or goal of the SSC research project is the systematic recovery, analysis, and integration of
any and all data bearing on the archaeology, geoarchaeology, and history of the study area. If any
project is successful in reaching these fairly circumscribed goals, the resulting data sets will not
only refine existing spatial and temporal models of past human behavior, but will also allow
subsequent opportunities to generate new and more expansive models for future work.
This orientation is unapologetically empirical in philosophy and recently has been approvingly
labeled the "Mars Lander Approach" to archaeological research by Madsen (1988). According
to Madsen (1988: 14), this particular approach treats archaeological loci in much the same fashion
that an unmanned probe landing on a distant planet would treat data collection. Just as the probe
gathers all the physical information it can collect, so also do "Mars Lander archaeologists" on their
earthbound sites. Madsen (1988) further observes that though this strategy is not inexpensive and
is time consuming, it is essentially completely removed from criticism by its very character. Data
collected by this approach are amenable not only to the traditional geoarchaeological, paleo-
ecological and culture historical research issues, but also to more nomethetically oriented processual
modeling.
In other words, it is the considered opinion of the ARP that the ongoing archaeological
research in the SSC project area must first address the full range of "who, what, where, and when"
questions before any contextual "how" questions or any processual "why" or causative issues can
be addressed. To these ends, we have organized our field and laboratory research to sequentially
treat a hierarchically arranged series of problems. These problems range form basic
geoarchaeological and ecological issues, through culture historical and so-called "middle range"
questions to "Big Picture" issues that ultimately address the whys and wherefores of the central
aspect of the prehistory of the SSC project area-namely, its marginality. Some of the research
themes and issues at the core of our current research design are presented below in ascending order
of complexity.
Geological and Ecological Research Questions
l. What was the pre-contact landscape like compared to today? Specifically, what was the
presettlement drainage pattern and overall hydrology?
2. Are the current paleo-topographic reconstructions of the study area accurate? Specifically, are
the so-called Pleistocene "terraces" too old to contain cultural remains? (If they are not, deep
testing will required in areas currently considered to be older than 11 ,500 B. C.)
3. What post-depositional processes have affected sites and site integrity in the study area?
4. What role do the Uvalde gravels have in site location and areal utilization?
5. What, in fact, is the paleo-climatic history of the study area?
6. What are the "attractions," in prehistoric terms, of the study area? Are these attractions time-
transgressive? Is their onset and waning witnessed throughout the late Pleistocene and
Holocene?
89
7. What are the presumed "natural" routes of movement, both for human and faunal populations,
into and out of the study area? Moreover, how does this area interface structurally with areas
around it? Does it afford resources these other areas do not have?
Culture-Historical Research Questions
8. What is the initiation of human occupation/utilization of the study area? Where are these early
sites most likely to be located?
9. What is the pattern of Paleo-Indian settlement (if present) in the study area? If no Paleo-
Indian material is identified in the study area, were such populations absent, not detected, or
too ephemeral? What comparisons can be drawn from the data pertaining to these questions
and between the remaining culture-historical periods?
10. What, specifically, are the defined prehistoric cultures alleged to exist in the study area? Are
there any mechanisms that can explain why certain cultures assumed to be represented here
are in fact absent, and vice versa? What nearby, well-defined cultures do not exist in the
study area? Why?
11. What is/are the direction(s) of apparent cultural influence into the study area? How are these
directions signalled (e.g., by projectile point styles, ceramic types, distinctive site types, etc.)?
12. How does Euro-American contact influence or change the apparent prehistoric utilization of
the study area?
Miscellaneous, "Middle Range" Research Questions
13. What are the apparent subsistence patterns at each point in the time-culture sequence?
14. What is the distribution of site types (e.g., base camps, bivouacs, kill sites, quarries, etc.) per
culture-historical period? Do these site type distributions suggest a hierarchical arrangement,
either synchronically or diachronically, of sites in the study area? If not, why not?
15. What evidence of trade and exchange is available from the study area?
16. How do the distributions of artifacts and ecofacts at each locality explicate: a) individual site
use, and b) patterns of site utilization in the greater study area.
The "Big Issues"
Big Issue 1. Why did anybody live in this marginal area?

Big issue 2. Why is the pattern of utilization in this area different from those of surrounding
areas?
Big Issue 3. Why is this information useful for interpreting Texas prehistory, specifically, and
what implications does this information have for archaeological research in general?
SUMMARY OF WORK COMPLETED TO DATE
As of this writing, ca. 98% of all fee-simple purchase areas ofthe West campus (ca. 3,035.4
ha; 7,500 acres), East campus (ca. 607.1 ha; 1,500 acres), and service areas (ca. 3,035.4 ha; 7,500
acres) has been intensively surveyed. Approximately 99% of all fee-simple areas directly under
the SSC ring footprint and the various abort alignments has now been investigated. Access to the
remaining areas has been denied by landowners. In all, some 4,365.9 ha (10,788 acres) have been
visually examined via pedestrian reconnaissance of plowed fields or have received randomized
invasive shovel probing in areas of heavy vegetational «over. Additionally, certain other
overgrown areas and disturbed landforms have been geobored.
Prior to this investigation, only four archaeological sites were recorded for the project area.
These included one prehistoric locus, the Pecan Springs site (41ELll), and two historic loci (sites
41EL56 and 41EL57). Since the initiation of the ARP reconnaissance and testing program, ll5
new sites have been recorded. Twenty of these are purely prehistoric sites, six evidence both
prehistoric and historic components, and 89 exhibit only historic components.
In addition to the sites identified via pedestrian reconnaissance and/or subsurface
probing/coring, a geomorphological investigation (Figure 2) was conducted to determine areas
90
1
N
8 I 2 3 1Sian
. Areas of ~Ion
Figure 2. Location of areas of sediment aggradation along drainages in the SSC project area, Ellis County,
Texas. Many of these areas will be selected for archaeological deep testing.
91
where deeply buried sites might occur. The entire footprint was examined for the potential of such
sites and a minimum of 63 areas, all located in areas of past or present aggradation, have been
identified for future work.
PREVIEW AND OVERVIEW: WHERE WE HAVE BEEN

AND WHERE WE ARE GOING
As of this writing, the site identification phase of the sse archaeological project is nearly
complete. The next phase, which is currently underway, involves the intensive characterization of
selected prehistoric and historic properties to ascertain their potential eligibility for nomination to
the National Register of Historic Places (NRHP), a necessary prelude to any further intensive
excavations. Thus far, 21 historic sites have been examined for NRHP eligibility. Twelve
assessments were negative, eight assessment reviews are pending, and three sites have been
recommended for data recovery (Le., the third level of field investigation).
Over the next 12 months, the remainder of the fee-simple areas will be reconnoitred and the
eligibility testing of selected historic and prehistoric sites will continue. Assuming concurrence
from the SHPO office on our recommendations of NRHP eligibility, sites designated as eligible will
be investigated in meticulous detail as the final phase of the field project.
It is certain that the characterization studies as well as any full-scale data recovery will
generate a corpus of data which, in conjunction with the initial reconnaissance data, can then be
used to answer most, if not all of the questions raised in the research design. These answers, in
turn, will permit us not only to describe and, hopefully, explain the archaeology of the sse project
area, but also, to relate that carefully crafted archaeological record to the broader prehistory of
Texas, specifically, and to that of North America, generally.
In closing, we would like to address what is perhaps a nagging question in many minds-what
exactly is the point of all this archaeology, anyway? We believe the answer to this query is simple.
The theme of this year's sse symposium is "discovering the future," in effect, looking ahead as
a species to where we are going. The theme of sse archaeology is also discovery, but it is
discovery about the past, about where we as a species have been. If, indeed, there is any truth to
the axiom that as a species, humans cannot know where they are going unless they know where
they have been, then the sse project epitomizes both sides of this scientific coin. The sse project
is, in the end, about prediction and retrodiction, and the success of both of these scientific
endeavors will be evaluated using the same fine scale of balance-time.
REFERENCES
Carlisle, R. C., and J. M. Adovasio (editors), 1984, Meadowcroft: Collected Papers on the Archaeology of
Meadowcroft Rockshelter muf the Cross Creek Drainage, Department of Anthropology, University of
Pittsburgh, Pittsburgh.
Fenneman, N. M., 1938, Physiography of the Western United States, McGraw Hill, New York.
Jurney, D. H., G. M. Weston, and B. R. Chada, 1990, An Archaeological Survey of the Proposed Super
conducting Super Collider, Ellis County. Texas, Draft interim report prepared by the Archaeology
Research Program, Southern Methodist University, Dallas, for the United States Department of Energy
and Universities Research Association, Inc., under the terms of Contract ESC-90821.
MacGregor, D. E., and J. E. Bruseth (editors), 1987, Hunter-Gatherer Adaptations along the Prairie Margin:
Site Excavations and Synthesis of Prehistoric Archaeology, Richland Creek Technical Series, vol. 3.,
Archaeology Research Program, Southern Methodist University, Dallas.
Madsen, D. B., 1988, Context and Concepts, In The Silver Island Expedition: Anthropological Archaeology
in the Bonnevillle Basin, vol. 1., ms. on file, Utah State Historical Society, Salt Lake City.
Peter, D. A., and D. E. MacGregor, 1988, Late Holocene Prehistory of the Mountain Creek Drainage. Joe
Pool Archaeological Project, vol. 1., Archaeology Research Program, Southern Methodist University,
Dallas.
Prikryl, D. J., 1987, A Synthesis of the Prehistory of the Lower Elm Fork of the Trinity River, unpublished
master's thesis, The University of Texas at Austin.
92
PROTECTION OF PREHISTORIC AND HISTORIC PROPERTIES AT THE
SUPERCONDUCTING SUPER COLLIDER
Dino L. Lowreyl and Clayton F. Marler

EG&G Idaho, Inc.
Idaho Falls, ID 83415-3128
INTRODUCTION
When exploring the basic nature or building blocks of the universe, the matter that
physicists examine is usually subatomic, such as protons or quarks. For individuals
involved in cultural resource management and preservation, that matter is somewhat more
tangible and can consist of a dinosaur or mammoth bone, a fossil fish or gastropod, or it
can consist of the easily recognized projectile point or "arrowhead." The study of cultural
resources also includes the examination of historic structures, which happen to be the most
predominant cultural resource in the 54 mile SSC footprint.
While curiosity about the nature of matter unites archaeologists and physicists alike,
what really brings us together is law. Particularly Newton's Law which states: "For every
action there is an equal and opposite reaction." To paraphrase Newton's Law in cultural
resource terms, federal agencies, such as the Department of Energy (DOE), need to
consider what effects their actions, and actions they may assist, permit, license, or fund
may have on cultural resources. This is mandated by the National Historic Preservation Act
(NHPA) and is implemented by the Section 106 process. Again, in more general terms,
Section 106 of NHPA states that before any ground disturbing activities take place on a
federally licensed or assisted project, certain steps or processes must occur. We will
examine these steps and their resultant SSC actions later, but first, it is important to
understand how preservation laws developed and evolved in the United States.
A HISTORY OF PRESERVATION LAW, Understanding the Past

To understand the intent of cultural resource management and preservation law it is
best to turn back the pages of time to the late 1800's when the west had been won and
Americans now had the leisure time to take an interest in anthropology, archaeology, and
history in general. In fact, public interest in the continent's ancient civilizations brought
about no less than five significant developments portentous for American archaeology in
the single year of 1879. These developments mark 1879 as the beginning of the movement
that led, a quarter of a century later, to the adoption of the Antiquities Act as the Irrst
national historic preservation policy for the United States (Lee, 1970). These developments
are as follows:
• In 1879, Congress authorized establishment of the Bureau of American Ethnology.
Of interest, Major John Wesley Powell, who had lost his right arm in the Battle of Shiloh
and who in 1869 had led his remarkable boat expedition through the Grand Canyon of the
Colorado River, was appointed its first director (Hellman, 1987).

• In 1879, a superbly illustrated book devoted to the ruined pueblos of Arizona, New
Mexico and the archaeology and ethnology of the Indians of Southern California was
published.
• The American Association for the Advancement of Science for the ftrst time elected
an anthropologist as its president.
• And [mally, the Anthropological Society of Washington D.C. and,
• The Archaeological Institute of America were founded (Lee, 1970)
Unfortunately, the rising public interest in the history and art of the southwestern
Indians in the 1890's was also accompanied by a swelling demand for authentic prehistoric
objects. The desires and needs of growing numbers of collectors and dealers, exhibitors
and curators, teachers and students, added to the native curiosity of cowboys, ranchers,
and travelers, created an avid demand for original objects from the cliff dwellings and
pueblo ruins of the southwest. Most of these ruins were situated on public land or Indian
reservations. There was no system of protection and no permit was needed to dig.
Professional archaeologists were few in number, and in America their science was in its
infancy and little known to the public. The eager seeker for artifacts had one chief worry---
that someone else would reach a ruin rich in valuable artifacts before he did. The result was
a rush on prehistoric ruins of the Southwest that went on, largely unchecked, until about
1904 (Lee, 1970).
The early stages of this rush accompanied the spread of prospecting and ranching in
Colorado, New Mexico, and Arizona during the 1880's and 1890's. Cowboys pursuing
wandering cattle through the mesquite and up remote canyons began to encounter ancient
ruins never before seen by white men. In this way, one December day in 1888, ranchers
Richard Wetherill and Charles Mason discovered Cliff Palace high on a canyon wall in the
Mesa Verde area of southwestern Colorado. This silent, spectacular, many-roomed
dwelling, protected by an overhanging cave and the arid climate of the Southwest, had
survived almost undisturbed for seven centuries. On the same day, in a nearby canyon,
they discovered another large cliff dwelling they named Spruce Tree House. Neither the
walls nor the contents of these ruins were to remain intact for long (Lee, 1970).
What ensued was some of the most rampant and destructive vandalism of cultural
resources that the United States has ever witnessed. Indiscriminate looting went on for
years and continues to this day. This wanton destruction eventually caused a public
upwelling of cultural resources protection support and in 1906, the Antiquities Act was
passed.
The Antiquities Act (PL-59-209) 16-USC-431-433) was the ftrst statute to protect
antiquities on federal land and this put paleontological resources along with prehistoric and
historic resources under the preservation umbrella. The Antiquities Act also mandated that
resources were to be preserved, restored, maintained, or disturbed only under excavation
permit.
LEGAL BACKGROUND
Today, many laws direct the inventory of cultural resources on federal land, guide the
nomination of sites to the National Register, provide mechanisms to protect cultural
resources during land-use activities, and provide legal penalties for their destruction. Three
of these laws are preeminent (Miller, 1992):
The National Historic Preservation Act of 1966 (PL-89-665; 16-USC-470, et.seq;
36-CFR 60, 61, 63, 64, 65, 67, 68, 800; Guidelines: 48-FR-44716-44742; Amended
1976; 1980:PL-96-515) States the leadership role of the Federal government in
preservation of prehistoric and historic resources and promotes a policy of cooperation
between federal agencies, tribes, other nations, states, and local governments. Directs
federal agencies to assume responsibility for considering historic resources in their
activities. Provides a process (Section 106/110) by which historically important properties
must be recognized and protected. Establishes the National Register of Historic Places.
Provides for the establishment and support of State Historic Preservation Offices (SHPOs),
state historic preservation plans. Creates the independent Advisory Council for Historic
Places (ACHP) to serve as counsel on historic preservation issues to the President and the
Congress and federal and state agencies (Miller, 1992).
94
The National Environmental Policy Act of 1969 (PL-91-190; 42-USC-4321-4361,
1976). Outlines federal policy of general environmental protection by requiring
information-gathering, planning, and assessments in advance of projects or actions that
occur on federal land or are federally licensed or funded. Requires the use of natural and
social sciences in planning and decision-making with regard to project impacts on the
environment. Federal agencies must prepare detailed statements Environmental Impact
Statements (EISs), or Environmental Assessments (EAs), outlining the scope of
environmental impacts of, and alternatives to the action planned, and allow for and consider
public comments (Miller, 1992).
The Archaeological Resources Protection Act of 1979 (PL-96-95; amended PL-100-
555, 100-558, 1988) This is the primary archaeological protection law for federal and tribal
lands. The act defines archaeological resources and prohibits unauthorized excavation and
removal and the illegal buying and selling of archaeological resources; provides criminal
and civil penalties for violations of the act (Miller, 1992).
NHPA and the 106 PROCESS
Section 106 requires that federal agencies consider the potential impact of their
activities on properties on or eligible to the NRHP. Section 106 provides a process (36-
CFR-8(0) to be followed to meet this requirement and maintain compliance with federal
law. Federal agencies can comply with Section 106 by following procedures for individual
activities (36-CFR-800, Sections 4, 5, 6) or by developing a Programmatic Agreement
(PA) for large projects (36-CFR-800.3, 36). This PA is developed in consultation with, in
addition to the DOE, or primary agency, the SHPO, the ACHP, Native Americans, and
other interested parties if warranted. PA preparations for the SSC began in early 1990 and
the final version was signed on December 21, 1990. The four signatories included
representatives from the DOE, the SHPO, the ACHP, and the Texas National Research
Laboratory Commission (TNRLC). The Programmatic Agreement for the SSC was
executed by the Council. This PA outlined actions to be followed by various participating
agencies in order to complete the requirements of Sections 106 and 110 of the National
Historic Preservation Act and the Council's regulations.
Compliance with Section 106 involves the following review process (36-CFR-800):
identification and evaluation of historic properties within the project
area or jurisdiction;
assessment of the effects of the project on those properties;
consultation with the SHPO and other interested parties on ways
to avoid or minimize adverse effects;
providing an opportunity for the ACHP to comment on the plan
orMOA;and
making a final agency decision on whether or how to proceed
(implementation) with the activity.
Section 110, directs federal agencies to establish programs to locate, evaluate, and
nominate eligible historic properties under their jurisdiction to the NRHP. This requirement
is important because it stresses that federal agencies must take an active role in preservation
and management of all significant cultural resources under their jurisdiction, not only those
that happen to fall within the path of construction or modification projects.
RESULTS OF THE 106 PROCESS, Identification and Evaluation

Paleontology
Along with the Antiquities Act, the United States Department of the Interior, National
Park Service (36CFR 2.1) directs the preservation of natural, cultural, and archaeological
resources and prohibits the following: (a) Possessing, destroying, injuring, defacing,
removing, digging, or disturbing from its natural state: ... (iii) Nonfossilized and fossilized
paleontological specimens, cultural or archaeological resources, or parts thereof (Miller,
1992). While these statutes are not definitive when it comes to fossil remains, the current
NEPA emphasis on cradle to grave responsibility for both federal and private actions led to
95
the recommendation that a paleontologic study be conducted for the SSC.
Geocharacterization was already taking place and a substantial number of the geocores were
being stored at the Bureau of Economic Geology making a paleontological study both
timely and cost effective. The Paleontolo~ic Survey of the Superconductin~ Super Collider
~ by L. Edwin Garner and Jules R. Dubar, published by the Bureau of Economic
Geology, Austin, Texas, was prepared for the Texas National Research Laboratory
Commission (TNRLC) in 1989. The purpose of the survey and study was undertaken to
determine the possible existence of unique fossils in SSC substrata and to provide
background data on the distribution of fossils present within the SSC Project area.
An archival search for comparative purposes was conducted and marine fossils
generally found in the sedimentary strata of the Taylor Marl, Austin Chalk, and the Eagle
Ford Shale were described. A total of 4,390 feet of SSC core was examined during the
course of the investigation from 19 borehole locations. The most characteristic fossil of the
Taylor Marl proved to be the Exogyra ponderosa, Phylum Mollusca, Class Gastropoda,
which, to the layman, looks like an oyster shell. Ammonites were another common fossil
of the Phylum Mollusca found in these beds. The Austin Chalk, which will house the bulk
of the tunnel, was characterized by various Inoceramus and Exogyras (Garner and DuBar,
1989). No fossils were found in any part of the cores that appear in any way to be
paleontologic ally unique or even particularly unusual. Fossils in the core were generally
sparse, poorly exposed to view, fragmental, small and of low diversity. All the fossil
forms observed during the course of this investigation appear to belong to groups that are
geographically widespread and typical of the stratigraphic units in which they occur
(Garner and Dubar, 1989). This report was distributed to the various participating agencies
and no comments were returned, indicating a concurrence that no adverse effects due to
SSC construction would occur to paleontological resources. No further action was
therefore required.
Architecture
In a period of six months, from October 1989 to April 1990, the Architectural
Resource Management Consulting firm of Hardy-Heck-Moore, under contract to the DOE,
completed a windshield survey of historic resources in Ellis County. This survey covered
approximately 939 square miles or 600,000 acres, excluding the cities of Waxahachie and
Ennis which the firm had previously surveyed. This survey was conducted from public-
right-of-ways and as a result of this effort, 3,714 previously unrecorded historic resources
were identified and documented in the county (Hardy-Heck-Moore, 1990). It was
determined that construction, according to the then current 1990 Laboratory design, would
likely have a direct impact on 41 historic resources, 15 of which were potentially eligible
for listing in the National Register, and an additional 9 were considered to be possibly
eligible pending further field or archival investigations.
Because TNRLC, in the original site proposal, agreed under DOE direction to
remove, at no cost, any existing improvements on the donated land for which the DOE did
not wish to accept title, TNRLC inherited the responsibility for mitigation of any historic
properties, in the footprint, that the DOE did not wish to accept. A Draft Historic Structures
Management Plan was submitted to the TNRLC on behalf of the DOE on December 2,
1991, by DeLara Almond Architects and Southern Methodist University (SMU). This
document identified approximately 51 historic structures that were potentially eligible to the
NRHP, which would be directly impacted by construction of the sse. Of these 51, 37
have subsequently been determined to be ineligible for nomination, and 14 have been
determined to be eligible.
The question that invariably arises at this juncture is, what happens when a
disagreement about eligibility occurs between the agency and the SHPO. The agency must
obtain a formal determination of eligibility from the Keeper of the National Register, who
acts on behalf of the the Secretary of the Interior, in accordance with applicable National
Park Service regulations (ACHP, 1986). When a final report is prepared by TNRLC on all
eligible properties in the footprint and has been submitted to the DOE, the DOE will
evaluate and consider various preservation options.
96
Archaeology
In 1989, the DOE, in consultation with the SHPO, ACHP, and SSCL
representatives, initiated a pedestrian or visual reconnaissance of the SSC footprint fee-
simple lands. This action would partially fulfill the Section 106 (NHPA) identification and
evaluation requirements. To completely fulfill the Section 106 identification requirement, it
would be necessary to initiate a subsurface or shovel-probe testing phase after TNRLC
obtained title and ownership to this land. Until then, no shovel probes could take place and
the reason for this was twofold: One, private landowners did not want "holes" dug on their
property and, two, any artifacts that might have been recovered would have belonged to the
private landowner and may not have been made available for expert study and curation.
Again, due to the fact that most of this land was still in private ownership, it was necessary
for TNRLC to request and secure from private landowners, permission for SMU
representatives to gain property access.
An Archaeoloiical Survey of the Proposed Syperconductini Super Collider. Ellis
County. Texas, an interim Draft Report, was presented by Southern Methodist University
representatives to the DOE in October 1990. At that time, 90% of the fee-simple areas had
been visually surveyed. A visual survey, or pedestrian reconnaissance, was then initiated,
at the request of DOE, for both stratified and fee-simple purchase areas. The inclusion of
stratified-fee lands in the survey process was the direct result of frequent changes in the
SSC design and the recognition of the potential for inadvertent damage to cultural
resources. Currently, 98% of all fee-simple land has been visually surveyed and 99% of all
stratified-fee acreage has been visually surveyed. A few parcels are still being held by
private landowners and the condemnation process continues.
Since the land transfer process has begun, SMU archaeologists have shovel-probed,
or Tier I tested, approximately 2,000 acres, exposing 15 archaeological sites. Twelve of
these sites have received a "no further work" recommendation and three sites may require
further testing or mitigation (Tier II Phase). Tier I and Tier II testing methodologies were
developed by SMU in conjunction and concurrence with the DOE, SHPO, and ACHP
adhering to the Department of the Interior's Standards and Guidelines for Identification and
the Archaeoloiical Survey: Methods and Uses, as well as Identification of Historic
PrJwerties: A Decision Makioi GUide for Mana~. Again, site testing results and ensuing
recommendations are presented to the SHPO by the DOE for concurrence or further
negotiation. At any time, if the DOE and SHPO do not agree, the ACHP can be consulted.
FUTURE GOALS AND ACTION

• Completion of the Tier I or Shovel Probe testing phase (with the current policy
of deferrment or incremental funding, Tier I testing is expected to continue into
FY 1993.)
• Completion of the Archaeological Data Recovery Plan and the Archaeological
Resources Preservation Plan (these plans are contingent upon completion of
the Tier I testing.)
• If warranted, completion of Tier II testing and mitigation actions
• Completion of the Area of Potential Effects Management Plan, a monitoring
effort aimed at historic structures outside the SSC footprint.
• Continued consultation with the ACHP, SHPO, TNRLC, and interested
parties
CONCLUSION
Above all, our goal will be to provide the management necessary for continued
compliance with all pertinent Federal, state, and local cultural resource law and in so
doing, also contribute to the preservation of our national historic and prehistoric heritage.
Of note, the SSC itself will be eligible to the National Register of Historic Places by 2040.
97
References
Advisory Council on Historic Preservation, National Park Service, USDI. 1986, Working with Section 106
Garner, E.L. and DuBar, J.R., 1989, "Paleontologic Survey of the Superconducting Super Collider Site",
Bureau of Economic Geology, Austin, Texas
Hardy-Heck-Moore, 1990, "Historic Resources Survey ofEIlis County, Texas, An Inventory for the U. S.
Department of Energy", Hardy-Heck-Moore, Austin, Texas
Hellman, G. T., 1987, "Octopus on the Mall" in The Smithsonian, J. B. Lippincott Co., Philadelphia, Pa.
Lee, R. F., 1970, The Antiquities Act of 1970, Office of History and Historic Architecture, Eastern Center,
Washington, D.C.
Miller, S. J., 1992, Management for Cultural Resources-Draft. Prepared for the Department of Energy,
Idaho National Engineering Laboratory, Idaho Falls, Idaho
98
USE OF PESTICIDES AT THE SSC: APPROVAL, REPORTING,
AND MONITORING REQUIREMENTS
William J. Berry 1 and William W. Pearson2
EG&G Idaho, Inc.

Idaho Falls, ID 83404
EG&G WASC, Inc.

Rockville, MD 20850
INTRODUCTION
The Superconducting Super Collider (SSC) project is committed to operating with as

little impact on the environment as possible. Some people would consider the use of
pesticides to be a contradiction to this policy. Nevertheless, pesticides can be used in an
environmentally sound fashion. Effective chemical agents are available to accomplish all
of the anticipated pest control requirements at the SSC. Strategies that minimize the use
of chemical agents, such as integrated pest management, will be used wherever possible.
Furthermore, the application method, active ingredient, and formulation can be carefully
selected to minimize the impact to non-target organisms.
Definition of Pest and Pesticide
For the purpose of this discussion, pest will be defined as any insect, rodent,
nematode, fungus, weed, or any other form of terrestrial or aquatic plant or animal life,
or bacteria, that interferes with any activity carried out at the SSC and pesticide will be
defined as any substance, or mixture of substances, intended for preventing, destroying,
repelling, or mitigating any pest, and; any substance, or mixture of substances, intended
for use as a plant regulator, defoliant, or desiccant. Therefore, the term pesticide includes
materials that are commonly thought of as pesticides, such as herbicides, insecticides, and
rodenticides, but also includes materials that act against other pest organisms, such as
avicides, algicides, and bactericides.
During development of a management plan for pesticides, it became apparent that
other materials applied to land, such as fertilizers and soil fixatives, would also require
environmental review and approval. These materials will be incorporated into the review
and approval procedure for pesticides.

Anticipated Pest Control Activities
Several concerns regarding the Red Imported Fire Ant, Solenopsis invicta, were
identified during preparation of the Supplemental Environmental Impact Statement for the
Texas location of the SSC. Specifically, concerns about worker safety in fire ant infested
areas and concerns that the ants can interfere with electrical equipment led to discussion
of pest control activities that might be required at the SSC.
In addition to fire ants, other pest organisms will require control at SSC facilities.
It is anticipated some SSC land will be leased for agricultural use and these areas are likely
to require control of agricultural pests. Campus areas are likely to include areas
landscaped with ornamental plants and lawns, which will require periodic application of
fertilizers and pesticides to ensure growth and to control pest organisms, including fire
ants. Domestic pests, such as cockroaches or flies, may require periodic control in SSC
facilities, particularly cafeteria areas. An elaborate system of cooling ponds will be
required for the SSC to function and these ponds will occasionally require algae control.
Furthermore, microbial pests may need to be controlled within the cooling system. Other
pests, including rodents, birds, and mollusks may also require periodic control.
Specific pest control methodologies will be developed for control of each pest
organism in the specific environment where control is required.
REGULATIONS AND STANDARDS REGARDING PESTICIDE USE
Federal
The primary federal regulation regarding pesticides is the Federal Insecticide,

Fungicide, and Rodenticide Act (FIFRA) codified in Title 7 of the United States Code.
Regulations pursuant to the Act are found in 40 CFR, parts 150-180. Procedures for
registration of a product, labelling and packaging requirements for products are established
under FIFRA. However, this regulation also establishes safety standards for pesticides to
ensure the prevention of adverse effects upon human health and the environment. These
safety standards include worker protection standards, applicator certification requirements,
and tolerances for pesticidal materials in agricultural commodities, food and animal feeds.
FIFRA identifies pesticides as unclassified, general use, or restricted use materials.
General use pesticides are materials that, when applied in accordance with label directions,
will not generally cause unreasonable adverse effects on the environment. Restricted use
pesticide are materials that may generally cause, without additional regulatory restrictions,
adverse effects on the environment. EPA does not normally classify products for general
use and products that are not restricted remain unclassified.
FIFRA establishes requirements for availability of materials and applicator
certification based on the class of material. Under FIFRA, unless otherwise prescribed by
its labelling, a pesticide shall be applied under the direct supervision of a certified
applicator or by a competent person acting under the instructions and control of a certified
applicator even though the certified applicator may not be physically present at the time
and place the pesticide is applied. Pesticides containing the active ingredients specified in
40 CFR 152.175, have been classified by EPA as restricted and are limited to use by or
under the direct supervision of a certified applicator. This section of the CFR identifies
restricted use active ingredients by formulation and use pattern and provides the criteria
influencing restriction. Formulations of general use pesticides and some unclassified
materials are readily available to homeowners for use in routine pest control activities.
These formulations do not require applicator certification for use by homeowners.
100
State of Texas Regulations Regarding Pesticides
The Texas Department of Agriculture (TDA) controls chemicals pursuant to the Texas
Pesticide Control Act and the Texas Herbicide Law. The IDA regulates pesticides under
the Texas Pesticide Control Act (TPCA) as set forth in Chapter 76 of the Agriculture Code
and the corresponding regulations in the Texas Administrative Code. The TPCA
authorizes the IDA to regulate: (1) the collection, examination and reporting of records,
devices and samples of pesticides, (2) the safe handling, transportation, storage, display,
distribution, or disposal of pesticides and pesticide containers, and (3) labelling
requirements for pesticides and devices required to be registered.
The IDA also regulates herbicides pursuant to the Texas Herbicide Law (THL) as set
forth in Chapter 75 of the Agriculture Code and the corresponding regulations in the Texas
Administrative Code. The Texas Herbicide Law authorizes the IDA to regulate the sale,
use, and transportation of herbicides. The application of herbicides is required to be
jointly supervised by the person for whom the application of the herbicide is made and the
person who applies an herbicide to land or plants for hire (referred to as a "custom
applier"). In addition, before spraying herbicides on land or plants, other than a lawn,
persons must give notice of the intent to spray and submit a record of the spraying in
accordance with the TDA's rules, whether or not a permit is required.
The Texas Department of Agriculture operates the applicator certification program for
the State. Under the certification program applicators are certified in various categories
relating, in general, to the commodity the applicator intends to treat.
Department of Energy Standards
The Department of Energy (DOE) standard DOE 5480.4 "Environmental Protection,

Safety, and Health Protection Standards" lists FIFRA as a mandatory ES&H standard. The
provisions of this order apply to all Departmental Elements and contractors performing
work for the Department of Energy. FIFRA shall be followed whenever application of
pesticidal materials is required during facility design, construction, operation, modification,
and decommissioning of the SSC. Facilities include those owned, leased, or otherwise
controlled by DOE.
Pursuant to FIFRA (40 CFR 171 et. seq.) application of pesticides must be performed
by a certified applicator. Texas has primary enforcement responsibility for certification
and only individuals certified under 4 TAC 7 will be allowed to apply pesticides at SSC
facilities. These applicators could include, but are not limited to, land lessees, contract
applicators, and SSC personnel. Lessees will undoubtedly apply pesticidal materials to the
crops they are growing. Contract applicators could be utilized to control domestic pests
in buildings or to maintain landscaped areas. SSC personnel may also be involved in
pesticide application in some instances, such as routine algae control in cooling ponds.
Applicator certification by the Texas Department of Agriculture in the appropriate
category(ies) will be required for all applicators.
Applicator Training
Pesticidal materials are normally toxic or hazardous in nature. OSHA requires hazard
communication training for employees who use or may come in contact with hazardous
chemicals. Any DOE, SSCL, TNRLC, or subcontractor employee working for the SSC
who may be working near toxic pesticides, will be made aware of the hazards related to
the pesticides. This requirement will be fulfilled through the hazard communication
program.
101
Adequate hazard communication requires development and maintenance of a written
hazard communication program, including a list of all hazardous chemicals present;
properly labeled pesticide containers, material safety data sheets (MSDSs), and
implementation of employee training programs regarding hazards of toxic chemicals and
protective measures to be taken around them.
In order for the hazard communication program to work, all pesticide labels must be
checked for adequacy before they are received at any SSC facility. MSDSs will be
required for all pesticides used on SSC areas. Pesticide storage areas will be routinely
inspected to ensure proper, safe storage and to ensure that no pesticide labels have been
altered. If labels have been altered or are missing, new labels will be affixed to the
pesticide container, or the pesticide and container will be disposed of.
Pesticide applicators will be trained in the following areas:
• Proper application procedure for each pesticide,
• Methods for observing the presence or release of toxic pesticides,
• Physical and health hazards related to pesticidal materials,
• Environmental consequences of each pesticide,
• Safety precautions, emergency procedures, and personal protective
equipment,
• The details of the hazard communication program, including an explanation
of the labeling system and the MSDS,
• Proper methods for disposing of pesticides and containers,
• Recordkeeping requirements.
The certification process will fulfill many of these training requirements. Prior to
application of any material, the applicator is required, by law, to read the label to insure
that the product is used according to label instruction and against a labelled pest.
PESTICIDE USE AT THE SSC
Areas that may be treated during pre-operational and operational phases at the SSC
include construction sites, access sites, campus areas, cooling ponds, leased land (for
agriculture), and restored prairies. Treatment needs in these areas will be highly variable.
Access sites may require activities such as control of rodents that invade buildings at those
sites. Cooling ponds may occasionally require application of algicides or other materials
in order to maintain function. Application of pesticides to prairie restoration areas will be
avoided because it may be detrimental to development of the prairie. Land disturbance at
construction sites may be sufficient to eliminate the need for application of pesticides.
Leased land will be used for a wide range agricultural pursuits comparable to existing
agriculture in the region and pesticide use on this property will be similar to current use
in the region.
Typical agricultural activities in the region include production of cotton, com cereals,
hay, and rangeland livestock production. These agricultural activities include activities that
generally require extensive application of pesticides, such as routine production of cotton,
and activities that typically require little pesticide use, such as rangeland livestock
production.
Approval of Pesticide Use at the SSC
An inventory of pesticidal materials for use on SSC property will be developed by the
SSC Laboratory (SSCL) and submitted to DOE for approval. Selection of proposed
pesticides will be based on published guidance, label information, and consultation with
County Agricultural Agents, commercial applicators, and consultants. Intended uses for
102
a pesticide will be consistent with product labelling. Pesticides on the DOE approved
inventory may be purchased, stored or used at the SSC without special case-by case
approval. No restricted use pesticides will be included on the approved inventory. The
SSCL will maintain an up-to-date inventory of all pesticides stored in and used at SSC
facilities.
Annually, contractors and lessees will be required to consider future pest control
requirements and will communicate these requirements with the appropriate agency [Texas
National Research Laboratory Commission (TNRLC) or SSCL]. Based on the proposed
pest control needs, the SSCL will compile a master list, to be submitted to the DOE. The
list of proposed pesticides will include:
• Pesticide Name (trade name),
• Active ingredient(s) in the material,
• EPA Registration Number,
• Target Organism(s),
• Application form (granule, liquid etc),
• Application rate,
• Application frequency (including "as needed"),
• Description of the area of application (such as "lawns"),
• Indication if the material is on the approved inventory,
• Indication that an MSDS is available.
Quantities of each material that might be used will not need to be estimated, but
quantities that are actually used will need to be reported for inclusion in the annual site
environmental report. Proposed materials that are not included on the approved inventory
will be investigated and will be added to the inventory or identified as materials that
require approval for each use. All restricted use pesticides will require approval for each
use.
Occasionally, a special need to use a pesticide that was not anticipated, or to treat an
area that normally would not be treated, may arise. SSCL and DOE approval will be
required prior to application of the material. Specific contacts and alternates for approvals
at DOE and the SSCL will be designated and these individuals will have the responsibility
of expediting (same day turnaround) review of the proposed use. The contacts will be
responsible for approving the proposed use or suggesting a viable alternative. The review
must include an individual who is trained in pesticide application to the supervisory level.
Familiarity with potential pest problems, application methodologies, and pesticide
properties would enable timely review of a proposed special use. Special uses of pesticides
will be subject to all reporting requirements
Reporting Usage
Pesticide applications will be documented by the applicator and a report of each use
will be provided to the SSCL. These reports will include:
• Date of application,
• Location of application,
• Type of pesticide used and reason for application,
• Amount of pesticide used,
• Date, location, and description of any pesticide-related emergency (i.e.,
spills, fires, or any other accident which could have a detrimental effect on
the environment) and procedures carried out for remediation.
These reports of individual applications will be compiled by the SSCL and a summary
report will be included in the annual site environmental report. The report will identify
the quantity of each material used during the previous year, the reason for use, and the
crop or area to which the pesticide was applied. The summary will also include the area
of the land treated with the pesticide and the application rate of the material.
103
MONITORING
Environmental monitoring programs will be carried out during routine operations of

the SSC. These programs will be directed at insuring that environmental contamination
is not occurring as a result of SSC operations and will enable in procedural changes in the
event that environmental contamination does occur. Pesticides will be included in these
monitoring programs. Monitoring programs will be carried out by the SSCL.
Soil samples will be gathered periodically in areas that have been treated with
pesticides. These samples will be analyzed to determine if soil accumulation of pesticides
is occurring. Soil samples will also be tested for nitrate accumulation that could occur as
a result of overfertilization.
Runoff from sse property will also be collected and tested for the presence of
pesticides. Samples to be tested will include periodic stream samples and storm water
runoff. If possible, aerial application of materials will be avoided to prevent drift of
pesticides.
RESPONsmLE PARTIES
Applicators
Applicators include SSC personnel, land lessees, commercial applicators, and any
other individual applying pesticides to SSC property. These applicators will be responsible
for maintaining appropriate certifications and for being aware of current regulations
regarding use, storage, and disposal of pesticide materials.
Applicators will anticipate usage needs and provide the information to the responsible
organization (TNRLC or SSCL) for compilation and submittal to DOE. Applicators will
also be responsible for reporting actual usage of pesticides. Applicators will be responsible
for complying with all applicable procedures and regulations for use, handling, storage,
and disposal of these materials.
Texas National Research Laboratory Commission
The TNRLC will manage lands that will eventually become SSC properties and will
tum the land over to DOE as it is needed for construction. A portion of the lands under
TNRLC management will continue to be used for agricultural production by lessees.
Pesticides will likely be applied to these lands in accordance with modem agricultural
practices in the region.
The SSCL will request that the TNRLC compile information regarding pesticide use
on TNRLC managed lands. This information will be incorporated into the annual site
environmental report.
The SSCL will coordinate approval of pesticide use, report pesticide use, and conduct
environmental monitoring for pesticides. The SSCL will compile a summary of anticipated
use of pesticides and submit a list of anticipated needs to DOE. The SSCL will also
maintain a database of actual use and for preparing a summary report of pesticide use to
be included in the annual site environmental report. The database will be compared to
anticipated use needs for the following year to ensure that estimates of needs are accurate
and realistic.
104
The SSCL will be responsible for insuring that all applicators have current
certifications and will recommend corrective actions in the event that applicators do not
maintain appropriate certifications.
The SSCL will be responsible for collecting environmental samples and verifying that
releases of pesticidal materials via runoff do not occur. The SSCL will monitor for soil
accumulation of pesticides and fertilizers. Results of environmental samples will be
included in the annual site environmental report.
The Department of Energy
The DOE will be responsible for reviewing the proposed usage needs and for
approving the materials that could be used on SSC property. The agencies will also be
responsible for reviewing and approving the annual environmental reports prepared by the
SSCL.
105
MEDICAL SURVEILLANCE OF EMPLOYEE HEALTH AT THE
SUPERCONDUCTING SUPER COLLIDER LABORATORY
Thomas J. Chester
Medical Department
Dallas, TX 75237
Medical surveillance can best be defined as conducting specific, targeted medical

examinations at pre-determined intervals for the purpose of assessing whether individuals
have suffered work-related illness or injury. The objectives of the medical examinations are
to determine if there is any evidence of illness or injury and to determine whether any
illness or injury found is occupationally related. If illness or injury is found, the employee
under medical surveillance can be referred for immediate treatment. Other employees in the
same work group can be examined, and any hazardous defects in the workplace can be
corrected.
Additional objectives of these periodic examinations are to determine whether the
employee's health status and physical fitness continue to be compatible with the safe
performance of his assigned job tasks; to contribute to employee health maintenance by
providing the opportunity for early detection, treatment, and prevention of disease or
injuries; and to provide a documented record of health status that can be used in analysis of
the health of the work group as a whole.
Medical surveillance is one of several measures used in a good occupational health
and safety program to prevent occupational illness or injury. Table 1 represents a hierarchy
of preventive health and safety programs. The earlier in the list a program appears, the more
basic it is to the prevention effort and the more likely it is to prevent occupational illness
and injuries with the least risk and least expense. A good occupational safety and health
program contains all of these elements.
No. DE-AC35-89ER40486.

Table 1. Occupational Health and Safety Programs:
Prevention Hierarchy.
• System safety-design review

• Health and safety procedures
• Operational readiness review
• Management safety awareness
• Employee safety awareness
• Periodic professional inspections
- Industrial hygiene
- Health physics
- Safety
- Fire
- Medical
• Industrial hygiene/health physics monitoring
• Medical surveillance examinations
• Epidemiologic aualysis
A good program of occupational safety and health starts with a system safety design
review. When a new facility is being built or changes are being made to an existing facility,
all designs and operational procedures should be reviewed by system safety engineers to
assess the risks and hazards involved and to alter the designs to minimize these risks and
hazards. During construction, the facility should be constantly monitored for adherence to
the system safety design. Also, during the design and construction phases, operational
procedures should be developed and reviewed in order to minimize risks and hazards in the
operation of the facility. Prior to beginning operation of the facility, an operational readiness
review should be conducted to ensure that the facility meets specifications and that there are
no noticeable safety defects.
Both management and employees must be made fully aware of the safety risks and
hazards involved in the facility and of the correct operational procedures to minimize these
risks and hazards. Management and employee training should be conducted in such a
manner as to instill in both groups an understanding of the need for safe operations and an
awareness of safety and recognition of hazards. Safety is to a large extent an attitudinal or
cultural phenomenon in which the culture engenders an awareness of safe operational
procedures and of the need to rectify hazardous situations.
Safety and health professionals should conduct periodic inspections of every facility to
include industrial hygiene/health physics, safety, fire protection, and medical!
toxicology. A regular program of industrial hygiene and health physics monitoring should
be conducted to ensure that levels of hazardous agents are kept below acceptable thresholds.
As an additional check on the effectiveness of health and safety measures, employees
working in potentially hazardous areas should be placed under medical surveillance, with
periodic examinations intended to detect any occupational illness or injury that may have
resulted from hazards or risks in the workplace. In addition, the data from medical
surveillance and industrial hygiene/health physics monitoring should undergo
epidemiologic analysis on a periodic basis to ensure that there are no subtle trends caused
by workplace exposures that were not detected in the individual medical surveillance
examinations.
108
Although medical surveillance and epidemiologic analysis provide a final check on
workplace health and safety, in most situations where thorough attention is paid to health
and safety and all of the earlier steps are carried out diligently, there should be a minimum
of occupational illness or injury detected in medical surveillance or epidemiologic analysis.
Unfortunately, in the real world it is possible that untoward effects of the workplace may
occur; it is, therefore, essential that proper medical surveillance and epidemiologic analysis
be conducted. Although these methods detect problems after the occurrence of illness and
injury, it is hoped that medical surveillance will detect the problems early enough for the
illness or injury to be reversible. Also, early detection of an illness or injury in a small
number of employees should allow corrective action to be taken before the remainder of the
work group is seriously affected.
Medical surveillance and epidemiologic analysis are particularly important for
detecting untoward effects from exposures that were not previously known to be hazardous.
Examples of such previously unknown effects discovered through medical surveillance
and/or epidemiologic analysis include vinyl chloride monomer, which causes angiosarcoma
of the liver, and acrylonitrile, which causes cancer of the lungs and colon.
The Superconducting Super Collider Laboratory (SSCL) Medical Office conducts
several types of examinations as part of its medical surveillance program (see Table 2). The
Employment Entrance Examination and Employee Transfer Examination serve mainly to
ascertain the physical fitness of the employee to perform his or her job tasks safely. They
also contribute to the medical surveillance program by providing baseline data on the
employee's health status to which future medical surveillance examinations can be
compared.
Table 2. Types of Examinations Provided by the

SSCL Medical Office.
• Employment Entrance Examination

• Employee Transfer Examination
• Periodic Examination
• Hazard Examination
• Post-Injury/IlJness Examination
• Termination Examination
The post injury/illness examination is mainly intended to ensure that the employee is
physically fit to return to work after an illness or injury. It also provides an opportunity to
determine any relationship between the employee's illness or injury and the workplace.
Periodic examinations are offered on a voluntary basis annually to all employees age
45 and over and biennially to employees under age 45. While this is a voluntary program,
there is sufficient representation among program participants to allow adequate assessment
of many work groups at the SSCL.
Termination examinations represent a final opportunity to assess the employee's
health status before leaving employ at the SSCL. This examination can be considered to be
the final examination of the medical surveillance program and can be considered to be part
of both the periodic and the hazard examination programs.
Hazard examinations are mandatory and are offered in response to specific hazardous
conditions that may occur at the SSCL. Examples of the hazard examinations provided at
the SSCL are listed in Table 3.
109
Table 3. Hazard Examinations Performed by the SSCL Medical Office.
PRESENT: • Oxygen Deficiency Hazard

• Respirator Certification
• Surface Rescue Team Certification
• Drivers and Mobile Equipment
Operators
FUTURE: • Hearing Conservation Program
• Underground Work
• Underground Rescue Team
Oxygen deficiency hazard examinations are performed annually on employees who

work in an area where the risk of oxygen deficiency has been deemed to be significant.
Employees are examined to determine their physical fitness to withstand a short period of
oxygen deficiency while putting on their self-rescue apparatus and exiting the building.
Respirator certification is required annually for those employees who must wear
respirators to protect them from inhaling hazardous substances in the workplace or for those
individuals who must wear self-contained breathing apparatus in order to conduct rescue
operations in areas of oxygen deficiency. Annual examination affords the SSCL Medical
Office the opportunity to assess each employee for any physical effects that may have been
caused by the toxic substances in question.
Surface rescue team certification is conducted annually for those employees who
participate in the surface rescue team. This volunteer team is trained and equipped to
conduct rescue operations in oxygen-deficiency hazard areas.
Driver and mobile equipment operator examinations are conducted annually to ensure
that operators are physically fit to operate such equipment safely. Particular attention is paid
to vision and hearing.
Although the SSCL does not currently have any underground operations in progress,
strict medical criteria will be applied to the physical fitness necessary for underground
access, and those working underground will be placed in an annual medical surveillance
program. An underground rescue team to be developed in the future will meet even more
stringent physical criteria and medical surveillance examinations.
At present, the SSCL hearing conservation program does not have any participants
because there are no positions currently in the Laboratory that expose an employee to noises
above the eight hour time weighted average (8 hr. TWA) threshold required for participation
in such a program. In the future, individuals working in areas meeting the hearing
conservation program criteria will be placed in a medical surveillance program with
audiograms at least annually to help ensure that any effects of the noise exposure are
minimized.
The SSCL Medical Office, in conjunction with other elements of the SSCL
Environmental Safety and Health Department, is currently working to acquire a
commercially available computerized occupational safety and health program. This program
will capture all medical surveillance data, industrial hygiene/health physics monitoring data,
and employee job assignment data.
The heart of this program will be the employee work history section, which captures
data about an employee's job assignments. This is crucial over a long period of time,
because as employees move from one position to another, are promoted, or are transferred
to other departments, they will be exposed to different hazards and risks. The coding
scheme used to define these job transfers and promotions will be the same as that used for
identifying the particular jobs and positions in the industrial hygiene/health physics
monitoring database.
110
The industrial hygiene/health physics monitoring database will contain both
monitoring data and assessments of an employee's exposure to hazardous conditions in the
workplace.
The medical surveillance module of the computer system will contain the results of
medical examinations, including laboratory results. Using such a system, one can gather
information on the degree of exposure of a work group to any monitored substance or
substances in the workplace and compare these on a statistical basis to the results of the
medical surveillance examinations performed on the group of workers. Such analysis will
enable determination of trends and health data that could not be detected on the individual
physical examinations and may reveal correlations between exposures and subtle health
effects before clinical illness is apparent.
This system of workplace monitoring, medical surveillance examinations, and
epidemiologic analysis provides additional assurance of no untoward effects from
workplace exposures.
, 11
REVIEW OF SSC DIPOLE MAGNET MECHANICS
AND QUENCH PERFORMANCE*
A. Devred, T. Bush, R. Coombes, J. DiMarco, C. Goodzeit, J. Kuzminski,

W. Nah, T. Ogitsu, M. Puglisi, P. Radusewicz, P. Sanger, R. Schermer,
R. Stiening, G. Spigo, J. Tompkins, J. Turner, Y. Yu, Y. Zhao, and H. Zheng
Magnet Systems Division

2550 Becldeymeade Avenue
Dallas, TX 75237 USA
M. Anerella, J. Cottingham, G. Ganetis, M. Garber, A. Ghosh, A. Greene,

R. Gupta, A. Jain, S. Kahn, E. Kelly, G. Morgan, J. Muratore, A. Prodell,
M. Rehak, E. P. Rohrer, W. Sampson, R. Shutt, R. Thomas, P. Thompson,
P. Wanderer, and E. Willen
Brookhaven National Laboratory, Upton, NY 11973 USA
M. Bleadon, R. Bossert, J. Carson, S. Delchamps, S. Gourlay, R. Hanft,

W. Koska, M. Kuchnir, M. Lamm, P. Mantsch, P. O. Mazur, D. Orris,
J. Ozelis, T. Peterson, J. Strait, and M. Wake
Fermi National Accelerator Laboratory, Batavia, IL 60510 USA
J. Royet, R. Scanlan, and C. Taylor
Lawrence Berkeley Laboratory, Berkeley, CA 94720 USA
ABSTRACT
Eight 5-cm-aperture, 15-m-long dipole magnet prototypes have been produced and
cold-tested at Brookhaven National Laboratory (BNL) and Fermi National Accelerator
Laboratory (FNAL) under contract with the Superconducting Super Collider Laboratory
(SSCL). These magnets are the last phase of an R&D program aimed at demonstrating the
feasibility of the 5-cm-aperture designs developed by BNL and FNAL. They are also used
* This work is supported by the U.S. Department of Energy under contract No. DE-AC35-89ER40486.

as vehicles to transfer technology from the National Laboratories to the collider dipole
magnet contractors. The BNL magnets, which rely on an horizontally-split yoke, and the
FNAL magnets, which rely on a vertically-split yoke, perform according to their somewhat
different mechanical designs and have equally successful quench performance.
INTRODUCTION
In January 1990, the decision was taken to increase the aperture of the
Superconducting Super Collider (SSC) dipole magnets from 4 cm 1 to 5 cm2 in order to
improve the field quality and reduce the risk of beam 10sses.3 A large effort was then started
at Brookhaven National Laboratory (BNL) and Fermi National Accelerator Laboratory
(FNAL) to design and build 5-cm-aperture dipole magnet prototypes. The first short model
magnet of the BNL 5-cm-aperture design was produced by KEK, National Laboratory for
High Energy Physics, in the summer of 1990.4 It was soon followed by a series of 1.8-m-
long model magnets at BNL5 and a series of l.5-m-Iong model magnets at FNAL.6 In the
fall of 1991, BNL and FNAL completed the production of their first 5-cm-aperture, 15-m-
long dipole magnet prototypes. Thus far, three full-length prototypes of the BNL design
(magnets DCA207, DCA208, and DCA209) have been built and cold-tested at BNL, and
five full-length prototypes of the FNAL design (magnets DCA311, DCA312, DCA313,
DCA314, and DCA315) have been built and cold-tested at FNAL.
The main goals of the 5-cm-aperture dipole magnet R&D program are to demonstrate
the feasibility of the larger-aperture magnets and to provide to the collider dipole magnet
contractors a basis on which to start their design efforts. The production of the full-length
prototypes is also used as a vehicle to transfer technology from the National Laboratories to
the magnet contractors. Three of the FNAL-design magnets (magnets DCA313, DCA314,
and DCA315) were assembled by personnel from General Dynamics, and one of the
BNL-design magnets (magnet DCA209) was assembled by personnel from Westinghouse
Electric Corporation. General Dynamics is the leader of the collider dipole magnet contract
and is expected to assemble seven dipole magnet prototypes at FNAL. Westinghouse is the
follower of the dipole magnet contract and is expected to assemble five prototypes at BNL.
Five of these ten industrially-assembled prototypes will be used in a string test to be
performed at the SSCL by the end of 1992.
The BNL and FNAL 5-cm-aperture designs are scale-ups of the 4-cm-aperture designs
previously developed. Both designs use a common magnetic cross section.7 They rely on
similar mechanical concepts, inherited from the 4-cm-aperture dipole magnet R&D
program,8-10 but they differ in the way these concepts are implemented. The main
difference is the orientation of the yoke split. The yoke of the BNL magnets is split
horizontally, while that of the FNAL magnets is split vertically.ll In the first section of this
paper, we shall detail the key features of the two mechanical designs, and we shall explain
their rationale. In the second section, we shall review the mechanical behavior of the eight
full-length prototypes during cold-testing, and we shall discuss how the behavior conforms
to the two different designs. The last two sections will be devoted to quench performance
and ramp-rate sensitivity. Additional information on the assembly of these prototypes are
presented in References 12 and 13, while preliminary reports on their field quality can be
found in References 14 and 15.
MAGNET FEATURES
Design Concepts
The required field of 6.6 T at 6500 A is produced by a two-layer cosine-theta coil}

The inner layer contains 19 turns and 3 copper wedges, and it is wound from a
, '4
1.2 keystone-angle cable of 30 strands (strand diameter 0.808 mm). The outer layer
0
contains 26 turns and 1 copper wedge, and it is wound from a 1.05 keystone-angle cable of
0
36 strands (strand diameter 0.648 mm). The cable insulation consists of a 25.4-f.lm-thick
layer of Kapton®,* wrapped with a 50% overlap, completed by a 9-mm-wide, 102-to-
127-f.lm-thick layer of epoxy-impregnated fiberglass, wrapped with a 0.5-mm gap.
The coil is mechanically restrained by means of laminated stainless steel collars. The
1.5-mm-thick collar laminations are spot-welded by pairs and stacked into packs. The packs
are locked around the coil by means of four tapered keys,16 which are driven horizontally
into the keyways near the midplane. The collars are designed to pre-compress the coil
azimuthally. The level of pre-compression is determined in order to compensate three
effects: 10 1) stress relaxation during assembly, attributed to insulation flow or creep,
2) stress loss during cooldown, due to thermal shrinkage differentials between the various
parts, and 3) stress redistribution during excitation, mainly due to the azimuthal component
of the Lorentz force. The target pre-compressions at room temperature are 70 MPa for the
coil inner layer, and 55 MPa for the coil outer layer.
The magnet cold mass is completed by a laminated iron yoke and a 4.95-mm-thick
stainless steel outer shell. The iron yoke surrounds the collars and enhances the magnetic
field by roughly 20%. The outer shell is welded around the yoke and delimits the region of
circulation for the 4.35-K, OA-MPa forced flow of supercritical helium. Yoke and shell are
designed to tightly clamp the collared-coil assembly and to stiffen the support against the
radial and axial components of the Lorentz force. The yoke support is needed to limit the
collar deflections along the horizontal diameter, where the effect of the radial component of
the Lorentz force is largest. The yoke clamping also allows, toward the magnet ends,
distribution of the axial component of the Lorentz force between the collared-coil assembly,
the yoke, and the outer shell.
The coil is loaded axially at both extremities by means of four screws. The screws are
mounted in thick stainless steel end plates, which are anchored to the outer shell. The axial
loading allows tight compaction of the coil ends and increases their stiffness. It is also
needed to support the fraction of the axial component of the Lorentz force that is not
transmitted to the yoke and the outer shell, and to prevent stick-slip motions of the
laminated collared-coil assembly inside the laminated yoke.
BNL Design
Figure lea) displays a cross-sectional view of the BNL 5-cm-aperture dipole magnet
design. The key features of the BNL design are horizontally-split yoke and anti-ovalized,
stainless steel collars. The inner boundary of the horizontally-split yoke is a circle with a
radius of 67.82 mm. The inner and outer boundaries of the anti-ovalized collars are circular,
and their centers coincide. The collars' outer radius is the same as the yoke's inner radius.
However, the keyways at the collars' midplane are placed asymmetrically so that when the
top and bottom collars are joined and keyed (with no coil in them), the center of the top
collars is shifted downward along the vertical axis by 0.10 mm with respect to the center of
the bottom collars. As a result, the vertical outer diameter of the keyed collars (with no coil
in them) is 0.10 mm smaller than the yoke inner diameter, while the horizontal diameters
are the same.
The rationale behind the BNL design is as follows. During collaring, the coil is
squeezed into the collars with a large azimuthal pre-compression. After collaring, the coil
exerts a large pressure against the collar poles. This internal pressure causes a large outward
* Kapton® is a registered trademark ofE.1. DuPont de Nemours & Co.
115
a
Laminated
iron yoke
Stainless
steel
laminated
collars
Tapered
key
b Laminated
Iron yoke
Stainless
steel
laminated
collars
Tapered
key
Figure 1. Cross sectional view of the cold mass of 5-cm-apertme, 15-m-long dipole magnet prototype:
(a) BNL, horizontally-split yoke, and (b) FNAL, vertically-split yoke dipole magnet designs.
116
deflection of the collared-coil assembly along the vertical diameter and a slight inward
deflection along the horizontal diameter. The slight inward deflection allows assembly of
the horizontally-split yoke around the collared coil. The large outward vertical deflection is
partially compensated by the fact that the collars are anti-ovalized, but it can result in a gap
between the two yoke halves. During welding of the outer shell, the shell is put into tension,
compressing the yoke. The yoke, in tum, compresses the collared-coil assembly, and the
eventual gap between the two yoke halves closes. At the end of shell welding, the yoke gap
is closed, and the outer circumference of the collared-coil assembly fits perfectly to the
inner circumference of the yoke.
During cooldown, the coil shrinks more than the Nitronic-40 stainless steel collars,
which, in tum, shrink more than the low-carbon steel yoke. The pressure exerted by the coil
against the collar poles decreases, but it remains large enough to keep deflecting the collars
and to maintain contact with the yoke along the vertical diameter. Along the horizontal
diameter, however, the thermal shrinkage differentials result in a small gap between the
collared-coil assembly and the yoke.
During excitation, the azimuthal component of the Lorentz force causes a
redistribution of the coil stress, resulting in a decrease of the pressure exerted by the coil
against the collar poles. At the same time, the radial component of the Lorentz force causes
a bending of the collars-with a maximum displacement along the horizontal diameter-at
the location of the gap between the collared-coil assembly and the yoke. As the current
increases, the bending moment increases, and the collars keep deflecting along the
horizontal diameter until they come into contact with the yoke. At high currents, the
collared-coil assembly touches the yoke on a large perimeter on both sides of the horizontal
plane, and the yoke provides an extremely stiff support against the radial component of the
Lorentz force.
The diameter reduction of the anti-ovalized collars along the vertical axis is
determined in order to ensure that on one hand, at the end of shell welding, the gap between
the two yoke halves is closed, and that on the other hand, at the end of cooldown, there
remains a suitable interference between the collared-coil assembly and the yoke. The
amount of vertical collar-yoke interference at liquid helium temperature is determined in
order to ensure that during excitation the current at which the collared-coil assembly comes
into contact with the yoke along the horizontal diameter is well below the operating current.
Table 1 summarizes the estimated vertical collar-yoke interference at room temperature for
the three BNL magnets considered in this paper. The estimates, based on measurements of
the collared-coil assembly diameters, assume that the yoke midplane gap is closed. The
room-temperature interference appears to be of the order of 170 /lm along the vertical
diameter.
As we described, the BNL design does not have a built-in COllar-yoke interference, but
relies on deformations of the collared-coil assembly to seek out the yoke support. The
absence of a built-in interference results in a somewhat loose clamping of the collared-coil
assembly by the yoke, and toward the magnet ends only a small fraction of the axial
component of the Lorentz force can be transmitted to the yoke and the outer shell. In order
to increase the rigidity of the coil ends, the coil axial pre-load is setup to a high value of the
order of 50 kN.
Other specific features of the BNL design include internal splices between the inner
and outer layer conductors, located at the radius of the coil outer layer. As displayed in
Figure 2(a), the coil ends are supported by means of stainless steel collars, similar to the
case of the magnet body, except for the pole parts. The BNL magnets also integrate an
improved cooling scheme, known as cross-flow cooling,17 which involves the radial
circulation of helium at set intervals along the magnet length, frolll the yoke cooling
passages to the coil cooling passage.
11 7
co
Table 1. Estimated room-temperature collar-yoke interference on selected 5-cm-aperture, 15-m-long collider dipole magnet prototypes.
The estimates are based on measurements of the collared-coil assembly diameters. The quoted values refer to the vertical diameter for the
BNL magnets and to the horizontal diameter for the FNAL magnets.
OCA207 OCA208 OCA209 OCA311 OCA312 OCA313 OCA314 OCA315
At Lead-End
Strain-Gauge Pack 153 179 154 305 318 318 288 263
At Non-Lead-End
Strain-Gauge Pack nJa nJa nJa 305 318 330 300 275
In Average Over
Magnet Lengtha 170 (28) 167 (32) 162 (19) 330 (20) 340 (15) 335 (18) 290 (25) 285 (20)
a Thenumber and the number in parentheses correspond respectively to the mean and the standard deviation. For OCA2XX magnets, the
end-collars are not taken into account.
a.
Ouler coil
Beam tube
1111~1""::-ICoilarS
End force gauge
(typ. 4 pies.)
b. Lead end damp
Figure 2. Cutaway view of the end-parts of 5-cm aperture. 15-cm-aperture.

15-m-Iong dipole magnet prototypes: (a) non-lead end ofBNL, horizontally-split
yoke, and b) lead end of FNAL, vertically-split yoke dipole magnet designs.
FNAL Design
Figure l(b) displays a cross-sectional view of the FNAL 5-cm-aperture dipole magnet
design. The key features of the FNAL design are vertically-split yoke and horizontally-
ovalized, stainless steel collars. The inner boundary of the vertically-split yoke is a circle
with a radius of 67.81 mm. The outer boundary of the horizontally-ovalized collars consists
of the union of four circular arcs, extending respectively between _30· and +30· about the
midplane, and between _60· and +60· about the pole axis.1 8 The four arcs have the same
radius, which is 0.01 mm larger than the yoke's inner radius. However, the centers of the
119
arcs extending between _30· and +30· about the midplane are shifted outward along the
horizontal axis by 0.14 mm with respect to the yoke center, while the centers of the arcs
extending between _60· and +60· about the pole axis are shifted downward along the axis
by 0.46 mm with respect to the yoke center. As a result, the vertical outer diameter of the
keyed collars (with no coil in them) is 900 !lm smaller than the yoke inner radius, while
there is a 300-/lffi interference along the horizontal diameter.
The rationale behind the FNAL design is as follows. During excitation, the support of
the yoke is needed mainly to limit the collar deflections along the horizontal diameter,
where the effect of the radial component of the Lorentz force is largest. One way to ensure
that such support is provided is to design collars and yokes so that they are always in
contact along the horizontal diameter. The problem, however, is that the integrated thermal
expansion coefficient between room and liquid helium temperatures of Nitronic-40 stainless
steel is smaller than that of low-carbon steel. To maintain contact during cooldown, one
must thus start with a positive interference at room temperature. A positive collar-yoke
interference along the horizontal diameter can be produced only by a vertically-split yoke.
The vertically-split yoke and the horizontally-ovalized collars are thus designed to
ensure that there is always a suitable collar-yoke interference along the horizontal axis. The
amount of horizontal collar-yoke interference at room temperature is determined in order to
at least compensate for the cooldown shrinkage differentials. On the other hand, in order to
assemble the vertically-split yoke, one must ensure that there is enough clearance between
the collared-coil assembly and the yoke along the vertical diameter. As we have seen, the
large pre-compression of the coil during collaring results in a large vertical deflection of the
collared-coil assembly. Thus the collars' vertical diameter must be smaller than the yoke's
inner diameter. The reduction in vertical diameter of the horizontally-ovalized collars is
determined to at least compensate for the vertical deflection resulting from collaring.
Table 1 summarizes the estimated horizontal collar-yoke interference at room temperature
for the five FNAL magnets considered in this paper. As for the BNL magnets, the estimate
is based on measurements of the collared-coil assembly diameters and assumes that the
yoke midplane gap is closed. The quoted values appear very close to the design value.
The assembly process of the FNAL vertically-split yoke magnets is similar to that of
the BNL horizontally-split yoke magnets. The main difference is the use of a press during
the welding of the outer shell. The press is required in order to close the gap between the
two yoke halves, which arises from the built-in collar-yoke interference along the horizontal
diameter. As a result, the collared-coil assembly is tightly clamped inside the yoke and,
toward the magnet ends, a large fraction of the axial component of the Lorentz force can be
transmitted to the yoke and the outer shell. The axial pre-load of the FNAL magnets is set to
a lower value than that of the BNL magnets, typically between 10 and 20 kN.
Other specific features of the FNAL design include external splices between the inner
and outer layer conductors, located at a larger radius than that of the coil outer layer. As
displayed in Figure 2(b), the coil ends are surrounded by a four-piece G 10 collet,19 which at
the lead end houses the splices. Coil ends and collets are compressed radially by a tapered,
aluminium cylinder that is hydraulically slid into position. The junction between the last
collar pack of the magnet body and the collet assembly is located approximately 76 mm
before the end of the straight section of the coil inner layer. The FNAL magnets do not
incorporate cross-flow cooling.
Instrumentation
The BNL and FNAL magnet prototypes have the same basic instrumentation. It
includes voltage taps to locate the quench origins,20 beam-type strain-gauge transducers to
measure the pressure exerted by the coil against the collar poles,21 and bullet-type strain-
gauge assemblies to measure the force exerted by the coil against the end plates. 21 The
120
FNAL magnets are instrumented with two strain-gauge packs, one at the minimum and one
at the maximum azimuthal coil sizes; the BNL magnets have only one strain gauge pack, at
the minimum azimuthal coil size. Both ends of the BNL magnets are instrumented with
bullet gauges, while only the non-lead end of the FNAL magnets is instrumented. (The non-
lead end is the magnet end opposite that where the current leads are located). Strain gauges
are also mounted on the outer shell of the FNAL magnets to measure azimuthal stress. 22
The test stands' interconnects of both the BNL and the FNAL facilities incorporate
temperature sensors, and warm- and cold-pressure transducers. In the case of the BNL test
facility, the helium inlet coincides with the lead-end of the magnets, while the connections
are opposite at the FNAL test facility. Typical helium mass flow rates are 40 to 50 gls at
FNAL, and 140 to 150 gls at BNL.
MECHANICAL PERFORMANCE
Expected Behavior During Excitation
As we described earlier, the Lorentz force has three main components: 1) an azimuthal
component, which tends to compress the coil towards the midplane and to unload the collar
pole; 2) a radial component, which tends to bend the collars outwardly with a maximum
deflection at the midplane; and 3) an axial component, which tends to stretch the coil ends.
During excitation, the pressure exerted by the coil against the collar poles is thus expected
to decrease as a function of current squared, while the force exerted by the coil against the
end plates is expected to increase. Let us now qualitatively discuss the differences in
behavior that can be expected from the two magnet designs.
In the case of the FNAL magnets, collars and yoke are designed to be always in
contact along the horizontal diameter. During excitation, the yoke provides a quasi-
infinitely stiff support against the radial component of the Lorentz force, and the collars do
not bend. The unloading of the collar pole thus results only from the compression of the coil
under the azimuthal component of the Lorentz force.
In the case of the BNL magnets, the cooldown shrinkage differentials result in a gap
between the collared-coil assembly and the yoke, which extends over a certain angle on
both sides of the horizontal plane. During excitation, the collars do bend and the coil
deflects accordingly, with a maximum displacement along the horizontal diameter. The arc
length of the coil increases, resulting in a decrease of azimuthal compressive stress. The
initial unloading of the coil from the collar pole thus results from two causes: 1) the
compression of the coil under the azimuthal component of the Lorentz force, and 2) the
bending of the coil under the radial component of the Lorentz force.
From the above discussions, the BNL magnets would be expected to exhibit a faster
initial rate of unloading than do the FNAL magnets. However, as the current increases the
collars of the BNL magnets keep deflecting along the horizontal diameter, eventually
coming into contact with the yoke. As the contact occurs, the yoke provides the desired
support against the radial component of the Lorentz force, and the additional bending
moment disappears. For currents above the contact current, the unloading of the collar pole
results only from the azimuthal component of the Lorentz force, and the BNL magnets
should behave similarly to the FNAL magnets. After the fast initial unloading, we are thus
expecting the pressure plots of the BNL magnets to exhibit a change in slope and to become
parallel to those of the FNAL magnets.
As for the end force, we are also expecting the BNL magnets to exhibit a faster rate of
increase as a function of current squared than do the FNAL magnets. One reason is that the
BNL ends are pre-loaded axially to a much higher level and are thus much stiffer. Another
reason is that due to the smaller collar-yoke interference, less of the axial component of the
Lorentz force is shared by friction between the collared-coil assembly, the yoke, and the
shell, and more is directly transmitted to the end-plates.
121
Comparison of Beam-Type Strain-Gauge Measurements
Cooldown and excitation data for the coil inner and outer layers are summarized in
Tables 2 and 3. The quoted values are an average over the four quadrants of the strain gauge
packs. In the case of the FNAL magnets, which have two strain gauge packs, the data are
differentiated by means of the letters LE and NLE, which refer to the location of the strain
gauge pack with respect to the magnet ends.
Table 2. Summary of inner coil stress measurements on selected 5-cm-aperture, 15-m-Iong

collider dipole magnet prototypes. The nmnber and the number in parentheses correspond
respectively to the mean and the standard deviation over the fonr quadrants of the given strain-gauge
pack.
Befcre After Befae Initial

Magnet First First First Stress vs. fl
Nane Cooldown Cooldown W8IDlUP Slope
(MFa) (MFa) (MPa) (MPaIkA2)"
DCA207 77 (5) 49 (6) 39 (4) -1.02 (.08)

DCA208 69(9) 36 (11) 29 (to) -0.86 (.16)
DCA209 66 (2) 40(4) 35 (4) -0.92 (.02)
DCA311 (IE) 53 (8) 31 (8) 35 (9) -0.71 (.08)
DCA311 (NLE) 52(4) 32 (1) 40(3) -0.76 (.04)
DCA312 (IE) 91 (9) 59(4) 61 (5) -0.80 (.05)
DCA312 (NLE) 58 (4) 33 (5) 34(4) -0.62 (03)
DCA313 (IE) 63 (9) 44(4) 46(5) -0.64 (.04)
DCA313 (NLE) 59(4) 48 (5) 34(5) -0.65 (.04)
DCA314 (IE) 77 (3)
DCA314 (NLE) 66(9) 33 (8) 33 (8) -0.66 (.11)
DCA315 (IE) 61 (5) 36 (4) 34(4) -0.58 (.03)
DCA315 (NLE) 53 (16) 31 (11) 21 (14) -0.45 (.15)
a For a strain-gauge run performed after the quench plateau was established.
Figure 3(a) presents a summary plot of the inner layer pressure as a function of current
squared for the three BNL magnets, and Figure 3(b) presents a similar plot for the five
FNAL magnets. The data displayed in Figure 3 were taken during strain-gauge runs
performed after the quench plateau had been established. For each magnet, the pressures are
averaged over the four quadrants of a selected strain-gauge pack, and only current up-ramp
data are displayed. As expected, the initial rate of unloading of the BNL magnets appears to
be faster than that of the FNAL magnets. Also, as the current increases, the traces of the
BNL magnets flatten and become more nearly parallel to those of the FNAL magnets.
However, none of the magnets presented here exhibit a complete unloading of the collar
poles as was observed on some of the BNL 4-cm-aperture prototypes. lO
122
Table 3. Summary of outer coil stress measurements on selected 5-cm-aperture, I5-m-long collider dipole magnet
prototypes. The number and the number in parentheses correspond respectively to the meml and the standard
deviation over the four quadrants of the given strain-gauge pack.
Befoce After Befae Initial

Magnet First First First Stress vs. f2 Contact
Name Cooidown Cooldown Warmup Slope Current
(MPa) (MPa) (MPa) (MPaIkA2)& (At
DCA207 55 (11) 27 (11) 22 (7) -t>.34 (.08) 4500

DCA208 69 (10) 39 (11) 36 (9) -t>.53 (.13) 4000
DCA209 64 (10) 34 (11) 31 (12) -t>.34 (.09) 4000
DCA311 (LE) 45 (9) 26(4) 27 (4) -t>.18 (.09) nla
DCA311 (NLE) nla
DCA312(LE) 36 (14) 30 (11) 20 (11) -t>.ll (.06) nla
DCA312 (NLE) 47 (25) nla
DCA313 (LE) 73 (14) 48 (11) 57 (26) -t>.19 (.06) nla
. DCA313 (NLE) 59 (25) 34(28) 33 (19) -t>.13 (.03) nla
DCA314(LE) 56 (12) 50 (19) 50 (19) -t>.20 (.03) ilia
DCA314 (NLE) 42 (3) 36(20) 34 (21) -t>.18 (.05) nla
DCA315(LE) 43 (14) nla
DCA315 (NLE) 28 (13) 15 (8) 15 (7) -t>.ll (.02) nla
a For a strain-gauge run performed after the quench plateau was established.
123
a 70
o DCA2O'7
., 60
o
o
DCA208
DCA2C)8
!.,
.,
Q)
50
~ 40
:=
0
0
JQ)
30
20
:if
~
Jc
10
10 20 30 40 50 60
Current Squared (leA2)
.,
+ DCA311
JC DCA312
60 • DCA313
II DCA314
! 50
• DCA3115
x DCA318
I
~ 40
8
J 30
Q)
:if 20
~
Jc
10
10 20 30 40 50 60
Current Squared (leA2)
Figure 3. Summary of the changes in the azimuthal pressure exerted by the coil inner layer against the collar
pole during excitations of the 5-cm-aperture, 15-m-Iong dipole magnet prototypes: (a) BNL, horizontally-split
yoke, and (b) FNAL, vertically-split yoke dipole magnet prototypes. The pressure data are averaged over the
four quadrants of selected strain gauge packs.
124
Let us discuss in more detail the initial slope of the pressure versus current squared. As
we described, the unloading of the collar pole results from the Lorentz load on the coil. The
Lorentz load is the same magnet to magnet, but what varies are the azimuthal stiffness of
the coil and the radial stiffness of the collars. We have already explained what to expect for
the collars' radial stiffness. On the other hand, the mechanical properties of the coil are
known to be non-linear: the higher the stress, the stiffer the coil. Within each design family,
we can thus expect the initial slope of the pressure versus current squared to be an
increasing function of the coil pre-compression. Figure 4 presents a summary plot of the
initial slopes of the inner-layer pressure versus current squared as a function of the inner-
layer pressure at zero current. As expected, the magnets appear to lie on two lines with a
positive slope: one for the FNAL magnets at the bottom of the plot, and one for the BNL
magnets at the top of the plot.
1.50
'"' 1.25
01
~ IS
! 1.00 o
Q)
jl, o
....0
CIl
<>
):(
In 0.75 +
In
Q)
bCIl o DCA207
o DCA208
:;:I
0
0.50 o DCA209
0 + DCA311
)( DCA312
~ • DCA313
~ 0.25 x DCA314
• DCA3111
0.00
0 20 40 60 80
Inner Coil Stress at Zero Current (MFa)
Figure 4. Correlation between the initial slopes of the inner-layer pressure versus current squared as a function
of the pressures at zero current for the 5-cm-aperture, I5-m-Iong dipole magnet prototypes. The pressure data
are averaged over the four coil quadrants.
Figure 5 presents summary plots of the outer-layer pressure as a function of current

squared for the three BNL magnets and the five FNAL magnets. These measurements were
taken during the same strain-gauge run as that of Figure 3. For each magnet the pressures
are averaged over the four coil quadrants, and only the current up-ramp data are displayed.
As with the inner-layer pressure, the initial rate of unloading of the BNL magnets appears to
be faster than that of the FNAL magnets. As was also observed on the 4-cm-aperture
prototypes,1O the traces of the BNL magnets exhibit a clear change of slope for a current of
about 4000 A. This current is interpreted as the current at which the collared-coil assembly
comes into contact with the yoke along the horizontal diameter.
125
a 70
o
-!
DCA207
o DCA208
60 o DCA209
lIS
II') 50
II')
Q)
ben
40
==
U
0
H
Q) 30
1:i
0
Q)
:tH
20
Q)
~
10
0
0 10 20 30 40 50 60
Current Squared (kA2 )
b 70
-
+ DCA311
)( DCA312
60 • DCA313
lIS lie DCA314
!II') 50
l:I DCA315
x DCA318
II')
Q)
ben
40
==0
u
~
0
30
Q)
:t 20
~
~
10
0
0 10 20 30 40 50 60
Figure 5. Summary of the changes in the azimuthal pressure exerted by the coil outer layer against the collar
pole during excitations of the 5-cm-aperture, 15-m-long dipole magnet prototypes: (a) BNL, horizontally-split
yoke, and (b) FNAL, vertically-split yoke dipole magnet prototypes. The pressure data are averaged over the
four quadrants of selected strain gauge packs.
126
Comparison of End-Force Measurements
Figures 6 and 7 present plots of the end-force versus time during the testing of BNL
magnet prototype DCA209 and FNAL magnet prototype DCA312. Each testing included
two cycles separated by a warm-up to room temperature. Magnet DCA209 was equipped
with bullet gauges at both ends. Data in Figure 6(a) are summed over the four lead-end
bullet gauges, while data in Figure 6(b) are summed over the four non-lead-end bullet
gauges. Magnet DCA312 was equipped only with bullet gauges at the non-lead end. Each
arrow on the plots corresponds to an excitation above 1000 A, while each arrow with a
number corresponds to a quench.
100
80
80
40
Thennal cycle
20
0
0 50 100 150
TIme (BolD'll)
b 120
.' ....'
.--"
100
• • .,,'::'~_ _• "0"1
"---
80
, .•.
! 80
~ ..
E
t. 40
20
0
0 100 200 300
TIme (BoUJ'll)
Figure 6. Change in the coil axial compressive loads during the testing of BNL 5-cm-aperture, 15-m-Iong
dipole magnet prototype DCA209: (a) at the magnet lead end, (b) at the magnet non-lead end. The data are
summed over the four loading screws of the given end. The small arrows indicate magnet excitations to a
current larger than 1000 A. The large arrows surmounted by a number correspond to a quench.
The most striking difference between Figures 6 and 7 is that magnet DCA312 end-
force appears to be quite stable, while that of magnet DCA209 appears to increase
significantly over each of the testing cycles: the lead end gained 22 kN during the first
testing cycle and 15 kN during the second, while the non-lead end gained 61 kN during the
first testing cycle and 35 kN during the second. As can be seen in Figure 6, most of the
increase takes place at the time of the first excitations after cooldown. It also appears that
the warm-up to room temperature resets the end-force to a level similar to that prior to the
first cooldown, and that the overall increase during the second cycle is somewhat less than
that during the first cycle.
127
50
..
40
,; ..., ... I · : ...• ... .. ..
. .... ..
! .... .....
..
30
..-. .." ..
~.:.
:.! 20
•
~ .".
"t; " ..... .,.... "1fDi.01' 4 _ _ _ _ __ _

•• r
10
Figure 7. Change in the coil axial compressive load during the testing ofFNAL 5-cm-aperture, 15-m-long
dipole magnet prototype DCA312. The data are summed over the four loading screws at the magnet non-lead
end. The small arrows indicate magnet excitations to a current larger than 1000 A. The large arrows
surmounted by a number correspond to a quench.
The behaviors of magnets DCA209 and DCA312 that we just described are
representative of the behaviors of the other magnets of the same design. Table 4 summarizes
the end-force data for the eight magnets described in this paper. The quoted values are sums
over the four bullet gauges of each end. It appears that the end-force of the FNAL magnets
is always quite stable, while the end-force of the BNL magnets always exhibits a large
increase over each of the testing cycles. This increase also appears to be always asymmetric,
with the non-lead end gaining more than the lead end.
The origin of the end-force build-up, which was also observed on the BNL 4-cm-
aperture prototypes,10 is not yet fully understood. One possible explanation is a ratcheting
of the collared-coil assembly inside the yoke. As we described, the axial component of the
Lorentz force tends to pull the coil ends outwardly. A fraction of this axial component is
transmitted to the end plate, but, for the most part, it is shared by friction between the
collared-coil assembly, the yoke, and the shell. However, the contact surface between the
laminated collars and the laminated yoke is very rough. Thus it can happen that as the
current is ramped-up, or due to the thermal expansion following a quench, some collar
laminations slip from one yoke lamination to the other, and that as the current is ramped-
down or the magnet temperature is brought back to normal, these collar laminations stick to
their new positions. These stick-slip motions could eventually explain the incremental
increases of end-force during the first excitations or quenches.
It now remains to explain why the BNL magnets exhibit such behavior while the
FNAL magnets do not. If the model of ratcheting is correct, the explanation could reside in
the fact that the FNAL magnets have a larger collar-yoke interference than the BNL
magnets. This larger collar-yoke interference results in a tighter clamping of the collared-
coil assembly, eventually reducing the risks of stick-slip motions during excitation. As we
shall see in the next section, however, the end-force ratcheting does not seem to have any
impact on the quench performance of the BNL magnets.
Figure 8(a) presents a summary plot of the non-lead end-force as a function of current
squared for the three BNL magnets, while Figure 8(b) presents a similar plot for the five
FNAL magnets. These measurements were taken during the same strain-gauge runs as those
of Figures 3 and 5-that is, after most of the ratcheting of the BNL magnets had occurred.
For each magnet, the force is summed over the four bullet gauges at the non-lead end, and
only the current up-ramp data are displayed. As expected, the BNL magnets exhibit a much
faster rate of increase than the FNAL magnets, which is consistent with our understanding
of the collar-yoke interference. (The difference in behavior between FNAL magnet
DCA311 and the other FNAL magnets may be related to discrepancies in the yoke stacking
and end-part assembly.)
128
Table 4. Summary of end-force measurements on selected 5-cm-Aperture, 15-m-Iong collider dipole magnet prototypes. The number
and the number in parentheses correspond respectively to the sum and the standard deviation from the mean for the four bullet gauge
assemblies of the given end.
Before After Before After Before Initial

Magnet First First First Second Second Force vs. 12
Name Cooldown Cooldown Warmup Cooldown Warmup Slope
(kN) (kN) (kN) (kN) (kN) (kN/kA2)a
DCA207 (LE) 64 (1.8) 54 (1.6) 62 (1.0) 56 (1.4) 60 (1.0) 0.5 (.01)

DCA207 (NLE) 68 (1.4) 66 (1.3) 92 (1.3) 74 (1.3) 91 (1.7) 0.6 (.01)
DCA208 (LE) 45 (1.2) 51 (1.2) 59 (1.2) 59 (1.4) 66 (1.5) 0.6 (.01)
DCA208 (NLE) 54 (0.9) 47 (0.9) 68 (0.9) 61 (1.1) 70 (1.0) 0.5 (.01)
DCA209 (LE) 44 (0.9) 51 (1.1) 73 (1.2) 62 (1.0) 77 (1.0) 0.6 (.00)
DCA209 (NLE) 29 (0.9) 28 (1.0) 88 (1.8) 57 (1.6) 92 (2.4) 0.5 (.01)
DCA311 20 (0.2) 33 (0.9) 33 (1.0) 32 (0.9) 34 (0.1) 0.13 (.01)
DCA312 14 (0.2) 18 (004) 20 (004) 19 (004) 18 (0.5) 0.3 (.02)
DCA313 22 (0.7) 29 (1.0) 26 (1.0) 26 (1.0) 29 (1.0) 0.4 (.02)
DCA314 18 (2.0) 17 (104) 20 (1.5) 18 (104) 22 (1.5) 0.3 (.02)
DCA315 9 (0.3) 13 (004) 12 (0.4) 12 (0.3) 11 (0.5) 0.2 (.01)
a For a strain-gauge run performed after the quench plateau of the first testing cycle was established.
N
(.()
120
-~110
CI>
U
o'"'
rz. 100
"CI
s:I
rz:I
:3o 90
E-<
o DCA207
<> DCA20e
60 o DCA209
10 20 30 40 50 60
b 60
+ DCA311
)( DCA312
~ DCA313
50
.. DCA314
.. DCA315
-
........ x DCA318
~ 40
CI>
U
rz.'"' 30
0
"CI
s:I
rz:I
....
«I
0
20
E-<
10
0
0 10 20 30 40 50 60
Current Squared (kA2)
Figure 8. Summary of the changes in the axial force exerted by the coil against the end-plate loading screws
during strain gauge runs performed after the establishment of a quench plateau for the 5-crn-aperture, 15-m-
long dipole magnet prototypes: (a) BNL, horizontally-split yoke, and (b) FNAL, vertically-split yoke dipole
magnet prototypes. The force data are summed over the four loading screws at the non-lead end of the
magnets.
130
QUENCH PERFORMANCE
With the exception of BNL magnet DCA208, all the magnets were tested following
the same run plan. The run plan calls for two testing cycles, separated by a warm-up to
room temperature. The first cycle includes quench testing at 4.35 K and ramp-rate study.
The second cycle includes quench testing at 4.35 K, 3.85 K, and 3.5 K. Due to schedule
constraints, magnet DCA208 was not tested at low temperatures. At FNAL, the magnets are
cooled down as rapidly as the cryogenic system allows; it takes about 24 hours. At BNL, up
to and including the first cooldown of magnet DCA209, the magnets were cooled down in
such a way that the temperature difference between the magnet ends would not exceed
50 K, increasing the cooldown time to about 36 hours. This restriction was removed during
the second cooldown of DCA209 in order to match the cooldown time of the FNAL
magnets.
Figure 9(a) presents a summary plot of the quench performance of the three BNL
magnets, while Figure 9(b) presents a similar plot for the five FNAL magnets. The
quenches reported here correspond to ramp rates less than or equal to 16 Ns, with the
magnet bore tube evacuated. For all magnets, the first ramp to quench after the first
cooldown is a strain gauge run, during which the current is increased in steps. For quenches
2 through 5 of the first testing cycle of FNAL magnet DCA3l1, the current was ramped at a
rate of 16 Als, while most of the other quenches on this magnet were approached at a ramp
rate of 2 Als. For subsequent magnets, the nominal current ramp was from 0 to 6500 A at
16 Als, 10 min wait at 6500 A, and from 6500 A to quench at 4 Als. Deviations from this
nominal ramp include quenches 7 and 8 of the first testing cycle, and all the quenches of the
second testing cycle of BNL magnet DCA209, for which the ramping from 6500 A to
quench was performed at 1 Als. The design ramp rate of the SSC collider ring is 4 Ns. The
intermediate step at 6500A was introduced to remove the coil heating that results from the
ramping at 16 Als. Ramp-rate sensitivity will be discussed in the next section.
With the exception of the first quenches of FNAL magnets DCA313 and DCA3l4, the
quench performance during the first testing cycle is quite satisfactory. BNL magnet
DCA207 exhibited two training quenches, while BNL magnet DCA208 exhibited one, but
they are all at currents above 7300 A. The first two quenches of BNL magnet DCA209
occurred respectively at 7100 A and 7305 A, while subsequent quenches were all above
7400 A. Note that this magnet exhibited a slight increase in quench current when the ramp
rate was lowered from 4 Als to 1 Als. FNAL magnet DCA311 exhibited a similar trend
when the ramp rate was lowered from 16 Als to 4 Als and 2 Ns, but it did not exhibit any
noticeable training. Neither did FNAL magnets DCA312 and DCA315. The low-current
quenches of FNAL magnets DCA313 and DCA314 occurred respectively at 4935 A and
5525 A, toward the lead end of the upper-inner quarter coil pole turn, on the side opposite
the ramp splice. A missing voltage tap complicates the determination of the axial locations
of these two quenches, but they appear to be similar and close to the boundary between the
last collar pack of the magnet body and the lead-end collet. We speculate that they resulted
from discrepancies in the assembly of the magnets' ends. After these first quenches,
however, the magnets reached 7300 A, and the problem did not resurface.
The quench performance at 4.35 K after the thermal cycle is also quite satisfactory.
BNL magnets DCA207 and DCA209 did not exhibit any noticeable retraining, while BNL
magnet DCA208 exhibited one retraining quench at 7100 A. None of the FNAL magnets
exhibited a noticeable retraining. With the exceptions of BNL magnet DCA209 and FNAL
magnet DCA315, all the magnets went directly to plateau at 3.8 K. Magnets DCA209 and
DCA315 both exhibited two training quenches (at 7745 A and 7920 A for magnet DCA209,
and at 7660 A and 7815 A for magnet DCA315), and then reached plateau. All the magnets,
except FNAL magnet DCA314, also reached plateau at 3.5 K, either directly (BNL magnet
DCA207 and FNAL magnets DCA311, DCA313, and DCA315), or with a limited number
131
of training steps (BNL magnet DCA209 and FNAL magnet DCA312). The first quench of
magnet DCA314 at 3.5 K occurred at 8320 A, which corresponds to the current limit that
can be expected from the conductor used in this magnet (see below). The next two
quenches, however, occurred respectively at 7985 A and 7710 A, and originated in the
upper-outer quarter coil near the lead end. There was no further attempt to reach plateau.
The lowest plateau at 3.5 K is that of FNAL magnet DCA311, with a current of 8100 A,
corresponding to a force level 1.5 times larger than that at 6500 A. This demonstrates that
the magnets of the two designs possess a large mechanical reserve.
DCA207 DCA206 DCA209

a 9000
I I I I
I Doool I I 0000
I I I I 00: :0
8000 r-- I
DODO:
I I I :t1'l -
IJ
00001 000
'VI OOOOOOOOpOOO. pOOO o {> 0000 001000 <> <>
o I·
j 7000 r-- I
I
I
I
I
I
o I
I
-
1:l., I I I I
6000 r- I I I I -
~ I
I
I
I
I
I
I
I
.d
.,s:I<> 5000 r--
I
I
I
I
I
I
I
I
-
&. I I I I
I I I I
4000
~
I'"
I~
:'"
.'" :x
I",
I~ '"'"'" I'"
I~
x
'"
I'"
I~
:'"
.", :;x
.",,,,
-
'"
.", .", .",
~
I~ :rri :r;o; I"; ..; I~
..; I"; :Iri :";cri
I I I I
3000
Quench Sequence
No_
(1) - DCA200 _ri_ mq:peb are BNL; DCA209-on BNI/WBC
(2) - Duhed liM lnd1c..tu t.bwmal oyola
Quench 0riliDe:
o - upper inner coil
o - lcnrer innw con
(3) - Dotted Un. . .perato different operat.ina temperaturea + - upper outer coil
C4-) - After DCA2Q? ramp to quench sequence: X - lower outer coil
leA/soc to 86001., wait. 10min, fA/eeo to quench
DCA311 DCA312 DCA313 DCA314 DCA315
~o:P
8000 - I ~ I
<x9.+ . 00-
... 1;,.0
I I
'Ii' o::A:l:>dpO~ <>~
r:>o
~O~ ~
! 7000 <> I : I -
1:l.,
6000 r- -
~ o
-5
~ 5000 r- o -
&.
4000
r-:'"
..;
I'" :x
I~
1"-
.'"
.'" ..,
:'"
:00) :w
'"
~
..;
I'"
I~
I~
:'"
..,
.'"
:00)
:'"
.",
:ci
X
'"'"
..;
Ix
I~
1"-
:'"
., :'"
..",
:w :w
...,
X
~
..;
Ix
I~
1"-
'" '"'"
:g
0; 0;
:x
.,
..",
:~
.,'"'"
..;
Ix
I~
I";
:'"
::g
:t':i
...
.",
:c?
Quench Sequence
}fo_: Quen.chO~:
(1) - DCA900 _ri.. mqnet. are P'NAL; DCASt3-on FN'AL/GDSS o - upper inner coil
(2) - Duhed. Une indicatu th.nnal cycle o - lo. .r inner coil
(3) - Dotted. lin. . .perato different operat.inc temperatures ... - upper outer coil
(4) - Standard Ramp to quench ..quence: x - lower outer coil
18A/aec to 66001., wait. lOmin. WHC to quench
Figure 9. Summary of quench performance of 5-cm-aperture, 15-m-long dipole magnet prototypes: (a) BNL,
horizontally-split yoke, and (b) FNAL, vertically-split yoke dipole magnet prototypes.
132
One interesting question to answer is how the plateau currents at the various
temperatures compare to what can be expected from the known properties of the cables.
Table 5 summarizes some of the parameters of the cables wound in these magnets,
including the critical currents measured routinely at the BNL short-sample test facility.
Figure 10 presents a summary plot of the highest plateau current at the given temperature as
a function of the extrapolated short-sample current limit at the given temperature and field.
The extrapolation is done using Green's parametrization of the NbTi critical surface. 23
Although most of the data points lie slightly below the first diagonal, the agreement is
relatively good, indicating that the magnet assembly did not result in any significant
degradation of the cable current capabilities. (It also indicates that the short sample
extrapolations are quite reliable.)
Table 5. Selected parameters of inner cables wound in 5-cm-aperture, 15-m-Iong collider dipole magnet protOtypes.
Magnet Name CableID Billet Number Cu:NbTi Ic at4.22K Cable Coil

and Coil and7T(A)a RRRa RRRb
DCA207 Upper SSC-3-I-00054 819-821 1.55 10904 39 220

DCA207 Lower SSC-3-I-00055 819-822 1.54 10836 38 220
DCA208 SSC-3-I-00067 1096 1.34 11371 37 132
DCA209 SSC-3-0-00044 2594-7 1.50 11265 38
DCA311 SSC-3-S-00023 2605 1.51 10079 37 175
DCA312 Upper SSC-3-I-00047 819-822 1.53 10764 40 165
DCA312 Lower SSC-3-I-00035 642 1.53 10673 42 167
DCA313 SSC-3-I-00052 8221857 1.52 10512 42 170
DCA314 Upper SSC-3-I-00052 8221857 1.52 10512 42 174
DCA314 Lower SSC-3-I-00049 819-822 1.52 10869 39 171
DCA315 SSC-3-I-00049 819-822 1.52 10869 39 170
a Measured on cable short sample.
b Measured on magnet during cold-testing.
RAMP-RA TE SENSITIVITY
When the current in a superconducting magnet is changed, heat is generated by several

mechanisms: hysteresis in the superconductor and in the iron yoke, eddy currents flowing
within the individual strands, and eddy currents flowing from strand to strand. The resultant
temperature increase causes a decrease in the plateau current of the magnet, at a rate
approximately equal to 20% per degree.
Figure 11 shows the decrease in quench current versus ramp rate for the eight magnets
described in this paper. These magnets exhibit a far greater sensitivity to ramp rate than did
earlier SSC dipole magnet prototypes, and it is believed that this can be attributed to
unexpectedly large strand-to-strand eddy currents. This picture is supported by direct
interstrand-resistance measurements performed at BNL on cable samples, showing that
interstrand resistance is sometimes as much as two orders of magnitude smaller than
expected. 24 Further, there appears to be a correlation between the decrease in quench
current and the AC loss, as directly measured on some of the FNAL magnets. 25 In addition,
133
..-..
~ 8500
~ '~ 0
Q) x
~
t)
+
~ 8000
Q)
i11:
+>
as
..... o
p.. + DCA207
o DCA20B
+>
III o DCA209
Q)
+ DCA311
.1:1 0
bO 7500 0 )( DCA312
:E 0 >i DCA313
x III II DCA314
o!o III DCA315
+
7000~~~~~~~~~~~--~~~~~~~~~~
7000 7500 8000 8500 9000

Extrapolated Short Sample Current Limit (A)
Figure 10. Comparison between quench plateau currents and short sample current limits extrapolated using
Green's parametrization of the NbTi critical surface.
as the ramp rate is increased, quench origin shifts from the pole tum, where the field is
maximum, to the coil midplane, where the field is normal, to the wide face of the cable,
producing the largest interstrand losses; this shift in location was predicted theoretically. In
Figure 11, BNL magnets appear to be less sensitive than FNAL magnets. It is believed that
this difference can be principally attributed to a difference in heat removal from the coil, as
BNL magnets employ cross-flow cooling and are tested with a significantly higher mass-
flow rate of helium.
There is as yet no clear explanation of why today's cables appear to be lossier than
previous cables. Preliminary investigations reveal possible links to changes in processing of
the strand, cable, and coil that affect the RRR, ductility, and surface condition of the
conductor. Note that although this sensitivity is of no concern to the operations of the sse
collider ring (with its nominal ramp rate of 4 Als), it can be a problem for the high energy
booster, which requires a ramp rate of the order of 70 Als.
CONCLUSION
The BNL and FNAL magnets perform according to their somewhat different
mechanical designs, and they appear to be equally successful in terms of quench
performance. The data from these magnets provide a sound basis for the dipole contractors
to start their design effort. Aside from the ramp-rate sensitivity, the characteristics of the
magnets are well understood, and the contractors can assess the operation of the two
different designs under nominal conditions in developing a final product that is readily
manufacturable and that meets all performance requirements, including those affecting its
reliability.
134
6000
1
....
4000
llqnets Inner Cables
o DCA207
o DCA208 IGC-8XX
o DCA209
+ DCA311 IGC-1XXX
2000 )( DCA312
~ DCA313 Orlord
II DCA314
~ DCA315 Supercon
100 200 300

Ramp Rate (A/sec)
Figure 11. Summary of current at quench versus ramp rate for the 5-cm-aperture, 15-m-long dipole magnet
prototypes.
ACKNOWLEDGMENTS
We are deeply indebted to M. Tigner, who laid out the SSC magnet R&D program and
who, by heavily instrumenting the early magnet prototypes, envisioned the means of its
success. We wish to thank V. Karpenko, who promoted the 4-cm-aperture design, and
R. Palmer, who promoted the 5-cm-aperture design. We are also grateful to J. Peoples, Jr.,
who, by his charisma and determination, boosted the R&D efforts. This program, however,
owes most of its success to the hard work and dedication of the teams at Brookhaven
National Laboratory and Fermi National Accelerator Laboratory, who ensured the
continuity of the magnet program and were instrumental in designing, building, and cold-
testing the dipole magnet prototypes.
REFERENCES
1. J. D. Jackson, ed., "Conceptual Design of the Superconducting Super Collider," SSC-SR-1020, March
1986; revised, September 1988.
2. J. R. Sanford and D. M. Matthews, ed., "Site-Specific Conceptual Design of the Superconducting Super
Collider," SSCL-SR-1056, July 1990.
3. T. Garavaglia, S. K. Kauffmann, and R. Stiening, "Collider Ring Particle Loss Tracking with SSCTRK,"
Supercollider 2, M. McAshan, ed., 1990, pp. 77-84.
4. K. Hosoyama, T. Shintomi, et al., "Fabrication and Test of a 5-cm Aperture, I-m Long SSC Collider
Dipole Magnet," Supercollider 3, J. Nonte, ed., 1991, pp. 615--624.
5. J. Muratore, M. Anerella, et al., "Construction and Test Results from 1.8-m-Long, 50-mm-Aperture SSC
Model Collider Dipoles," to be published in these proceedings.
6. M. Lamm, R. Bossert, et aI., "Results from the Fermilab 1.5-m Model Magnet Program," to be published
in these proceedings.
135
7. R. Gupta, S. A. Kahn, and G. H. Morgan, "ssc 50 rom Dipole Cross Section," Supercollider 3, J. Nonte,
ed., 1991, pp. 587-599.
8. J. Strait, B. C. Brown, et at, "Tests of Full Scale SSC R&D Dipole Magnets," IEEE Trans. Magn., 25, No.
2, 1989,pp. 1455-1458.
9. J. Tompkins, M. Chapman, et at, "Performance of Full-Length SSC Model Dipoles: Results from 1988
Tests," Supercollider 1, M. McAshan, ed., 1989, pp. 33-49.
10. A. Devred, T. Bush, et at, "About the Mechanics of SSC Dipole Magnet Prototypes," to be published in
AlP Conference Proceedings Series, "The Physics of Particle Accelerators," 1992.
11. J. Strait, K. Coulter, et at., "Experimental Evaluation of the Vertically Versus Horizontally Split Yokes
for SSC Dipole Magnets," Supercollider 2, M. McAshan, ed., 1990, pp. 731-741.
12. M. Anerella, J. Cottingham, et al., "Construction and Test Results from 15-m-Long, 50-rom-Aperture
SSC Collider Dipole Models," to be published in these proceedings.
13. M. J. Blessing, D. E. Hoffman, et aZ., "Construction Experience with Fermilab-Built Full-Length 50-rom
SSC Dipoles," to be published in these proceedings.
14. P. Wanderer, M. Anerella, et al., "A Summary of SSC Dipole Magnet Field Quality Measurements," to be
published in these proceedings.
15. S. Delchamps, M. Bleadon, et al., "Magnetic Field Measurements of FermilablGeneral Dynamics Built
Full Scale SSC Collider Dipole Magnets," to be published in these proceedings.
16. C. Peters, K. Mirk, et al., "Use of Tapered Key Collars in Dipole Models for the SSC," IEEE Trans.
Magn., 24, No.2, 1988, pp. 820-822.
17. R. P. Schutt and M. L. Rehak, "Transverse Cooling in SSC Magnets," Supercollider 1, M. McAshan, ed.,
1990,pp.209-217.
18. J. Strait, J. Kerby, et al., "Mechanical Design of the 2D Cross-section of the SSC Collider Dipole
Magnet," Proceedings ofthe 1991 IEEE Particle Accelerator Conference, 1991, pp. 2176-2178.
19. P. Heger, R. Bossert, et al., "Design and Construction Experience with Collet End Clamps for 50-rom
Aperture SSC Collider Dipoles," to be published in these proceedings.
20. A. Devred, M. Chapman, et al., "Quench Start Localization in Full-Length SSC R&D Dipoles,"
Supercollider 1, M. McAshan, ed., 1989, pp. 73-83.
21. C. L. Goodzeit, M. D. Anerella, et al., "Measurement of Internal Forces in Superconducting Accelerator
Magnets with Strain Gauge Transducers," IEEE Trans. Magn., 25, No.2, 1989, pp. 1463-1468.
22. M. Wake, M. Bleadon, et al., "Mechanical Behavior of Fermilab/General Dynamics Built 15-m SSC
Collider Dipoles," to be published in these proceedings.
23. M.A. Green, "Calculating the Ie, B, T Surfaces for Niobium Titanium Using a Reduced-State Model,"
IEEE Trans. Magn., 25, No.2, 1989, pp. 2119-2122.
24. A. Gosh, private communication.
25. J. Strait, D. Orris, et al., "QuenCh Performance of Fermilab/General Dynamics Built Full Length SSC
Collider Dipole Magnets," to be published in these proceedings.
136
A SUMMARY OF sse DIPOLE MAGNET FIELD QUALITY MEASUREMENTS'
P. Wanderer, M. Anerella, J. Cottingham, G. Ganetis, M. Garber,

A. Ghosh, A. Greene, R. Gupta, A. Jain, S. Kahn, E. Kelly, G. Morgan,
J. Muratore, A. Prodell, M. Rehak, E.P. Rohrer, W. Sampson, R. Shutt,
R. Thomas, P. Thompson, and E. Willen
Accelerator Development Department

Upton, New York 11973 USA
A. Devred, T. Bush, R. Coombes, J. DiMarco, C. Goodzeit,

1. Kuzminski, W. Nah, T. Ogitsu, M. Puglisi, P. Radusewicz, P. Sanger,
R. Schermer, J. Tompkins, J. Turner, Y. Yu, Y. Zhao, and H. Zheng
Magnet Division
2550 Beckleymeade Ave.
Dallas, Texas 75237 USA
M. Bleadon, R. Bossert, J. Carson, S. Delchamps, S. Gourlay, R. Hanft,

W. Koska, M. Kuchnir, M. Lamm, P. Mantsch, P.O. Mazur, D. Orris,
T. Peterson, J. Strait, and M. Wake

Batavia, IL 60510 USA
J. Royet, R. Scanlan, and C. Taylor

Berkeley, CA 94720 USA
ABSTRACT
This paper reports results of field quality measurements of the initial 15 m-Iong, 50 mm-
aperture SSC Collider dipoles tested at Brookhaven National Laboratory and Fermi
"This work is supported by the U.S. Department of Energy

National Laboratory. These data include multipole coefficients and the dipole angle at
room temperature and 4.35 K, 4.35 K integral field measurements, and time-dependent
effects. Systematic uncertainties are also discussed.
INTRODUCTION
The initial series of full-sized SSC collider dipoles includes magnets made at Fermilab
and BNL. At this time, field quality data are available from three magnets made by BNL
and Westinghouse personnel and from five magnets made by Fermilab and General
Dynamics personnel. Partly because not all magnets have been tested and partly because
of uncertainties in the analysis of the multipole coefficient data, this report will serve
primarily as a snapshot of work in progress. Only a minimal amount of magnet
construction information is included here. More complete reports have been made in other
papers to this conference1,2,3. The field quality measurements in the Fermilab-General
Dynamics magnets have also been summarized in a separate conference paper".
A. Multipole Coefficients
A useful expression for the magnetic field is:
By +iB" =Bo L (b n + ian)(x +iyt

n=O
where x and yare the horizontal and vertical coordinates and Bo is the dipole field
strength. A pure dipole field has bo = 1 and all other multipole coefficients zero. The
skew terms are denoted by the a", the normal terms by bo • The quadrupole coefficients
have n = 1, the sextupole have n = 2, and so forth. The coefficients are evaluated at a
reference radius of 10 mm. Typical tolerances are at the level of 10-4 of the dipole field,
and this scale is informally referred to as a "unit" or a "10-4 unit" of a multipole
coefficient. The multipole data summarized in this note are for the magnet straight-section
measured with a 1. 0 m coil.
MultipoJe Coefficients at 2T
The axial variation of the normal sextupole, b2 , in a typical magnet at 2 kA and 6.5 kA
is shown in Fig. 1. (To a good approximation, 1 kA produces 1 T of field.) The value
of b2 , about 4 units, is nearly the same at both fields. The axial variation is the same at
the two fields, as expected. The size of b2 is due to the use of pole shims which were not
of the design thickness in this magnet. (This is discussed in more detail below.) The axial
variation of the normal decapole, b4 , is shown in Fig. 2. It can be seen that b4 at high
field is systematically lower than at low field. The interesting question of whether the
axial variation is the same for magnets made on the same fixturing will be addressed in
a future report.
Summaries of the multipole coefficient data for the first three BNL-Westinghouse
magnets and the first five Fermilab-General Dynamics magnets are presented in Figs. 3
and 4, respectively. For each multipole coefficient, the mean and r.m.s. variation have
been calculated and then divided by the sse systematic tolerance. In the plots, multipoles
meeting the tolerances will have ratios plotted between the dashed lines at + 1 and -1. (In
estimating the uncertainty in the mean, an allowance must be made for the small number
of magnets in the sample.)
For both sets of magnets, most of the multipole coefficients lie between the dashed lines.
The skew quadrupole, a h lies beyond the dashed lines but is particularly subject to
138
,----,,-----,----,----,----,----,
q
N
o
o
o
N
I
o
'f
o
52
I
o
ti
I
q
~
I
o
~ L-__- i____- L_ _ _ _L -__~____~__~~__~
-100 0 100 200 300 400 500
Z (inches)
Figure 1. Axial variation of b2 at 2 kA

(+) and 6.5 kA (0).
~
0
'"0
...
0
N
0
0 0
::)q o 0
::;-0
o·
N
0
I
~
0
I
~
0
I
OJ
0
I
0
., -100 0 100 200 300 400 500
Z (inches)
Figure 2. Axial variation of b4 at 2 kA

(+) and 6.5 kA (0).
139
-4
~
bl b2 b3 b4 b5 be b7 be a1 aZ a3 a485 a6 a7 a8
Normal Terms Skew Terms
Figure 3. Mean and RMS variation of

multipole coefficients
measured at 2 kA for three
BNL-Westinghouse magnets,
scaled by the sse systematic
tolerance. (The ratios for a1
and b2 exceed 5.)
I.'
:t II,
----l---;:--------------------~--i-_f-~--------
~o
~
":r 11 i
-----------------------------------------------I--
~ ::~
bl b2 b3 b4 b5 b6 b7 b8 a1 a2 0.3 84 a5 a6 87 ae
Figure 4. Mean and RMS variation of

multipole coefficients
measured at 2 kA for five
Fermilab-General Dynamics
magnets, scaled by the sse
systematic tolerance. (The
ratio for bs exceeds 5.)
additional uncertainties in the centering correction, discussed below. The allowed

multipoles (bo's with n even) are in some cases well beyond the dashed lines, but, since
they are generally in good agreement with calculations based on the actual sizes of the
magnet components5 , significant reduction can be achieved by a modest redesign of the
coil cross section. Only the decapole has a significant (0.4 unit) difference between the
design and measured values; this offset can be taken into account in the new cross section
design.
140
The correlation between room temperature and 4.35 K (2 T) multipole measurements is
shown in scatter plots of the skew quadrupole a l (Fig. 5) and the normal sextupole b 2 (Fig.
Q). Multipoles with the same value warm and cold would lie on the diagonal line. With
some scatter, the a l data follow this line. However, the b2 data lie in two groups, one for
BNL-Westinghouse magnets and one for Fermilab-General Dynamics magnets. This
grouping is due to differences in magnet construction, the most obvious of which are the
inner coil pole angles and the orientation of the split in the yoke laminations. The data
also indicate that the two designs respond differently to cooldown, although each response
is acceptable for the accelerator.
2.0
1.5
'"
.,"
~
'0
1.0
~
§. 0.5
A
~
E
-;;; C
o 0.5 1
al (Cold) (10- 4 units)
Figure 5. Scatter plot of warm and cold

(2 kA) measurements of the
skew quadrupole a l for three
BNL-Westinghouse magnets
(A-C) and five Fermilab-
General Dynamics magnets
(D-H).
A summary of the warm-cold correlation data is shown for the three BNL-Westinghouse
magnets and for the five Fermilab-General Dynamics magnets in Figs. 7 and 8. For each
multipole, the r.m.s. variation of the difference between the warm and cold measurements
is calculated and then divided by the tolerance on the r.m.s. variation of the multipole
coefficient. For all but one term, this ratio is small in comparison to 1. Hence, the use
of warm measurements to predict the cold multipoles will not significantly increase the
r.m.s. width of the cold multipole distribution, and the strategy of measuring all the
magnets warm but only a fraction cold is acceptable. (The ratio for a7 in Fig. 7 is thought
to be an artifact.)
Magnetization, Saturation, Eddy Currents
The next set of plots illustrates the sextupole and decapole variation with current.
Typical sextupole data for BNL-Westinghouse and Fermilab-General Dynamics magnets
are shown in Figs. 9 and 10 respectively. (In figures 9-13, a constant has been added to
the measured values so that a detailed comparison to the calculation can be made.) The
magnetization sextupole, the most important effect at low field, increases monotonically
141
4
,-...
rn
-'
'2
::l
...
I
0
BA
3
C
F
,-...
H E
8t-. G
(lj
~ 2
"--'
D
C\l
.0
2 3 4 5
b2 (Cold) (10- 4 units)
Figure 6. Scatter plot of normal sextupole b2 data.
Same notation as figure 5.
2.0
1.5
"0
E-
rn
S
t-.
"'- 1.0
til
S
.~
Ul
0.5
0.0
bl b2b3b4b5b6b7b8 al a2 a3 a4 a5 as a7 aB
Figure 7. RMS width of the distribution of cold-warm
differences in the multipole coefficients,
divided by the sse RMS tolerance, for three
BNL-Westinghouse magnets.
142
2.0
1.5
0.5
0.0
b 1 b2 b3 b4 b5 b6 b7 bS al a2 a3 a4 as a6 a7 as
Figure 8. Same information as in figure 7, for five
Fermilab-General Dynamics magnets.
2 - " -
~
e
u
-
T 0-
o
o
"
.
.0
-2f- •"
0
rJI.
-
"
_4~~O~~~__~_~I~ __L-~~__~I__~-L__~~__L-I-u
o 2000 4000 6000
CURRENT (Amps)
Figure 9. Normal sextupole b2 vs. current for a typical
BNL-Westinghouse magnet. In this and
subsequent figures, X denotes calculations,
o measurements. The calculations include
only magnetization effects.
143
l
b2 VS. CURRENT
41
'"' I i i i
~ 0
x
0
2 /I
.,
.,
E
(J
,
o
l
-2
x
0
x
0
-4
a 2000 4000 6000
CURRENT (Amps)
Figure 10. Normal sextupole b2 versus current for a

typical Fermilab-General Dynamics magnet.
Same symbols as figure 9.
as the current is ramped up and then down again. At these fields the hysteresis curve of
the sextupole in the two magnets is quite similar and in good agreement with the
magnetization calculation. (The calculation is adapted from the HERA program.) For
currents above 4 kA, the difference between the two magnets is qualitatively attributed to
a combination of two factors: (1) the small motion of the collared coils in the BNL-
Westinghouse magnets, which move outward until they contact the horizontally-split
yoke6 ,7; (2) the small notches in the yoke inner edge to align the collars, located at the top
and bottom in the Brookhaven-Westinghouse magnets and on the horizontal midplane of
the Fermilab-General Dynamics magnets. Although not discussed in detail here, the
variation of b2 and b4 due to yoke saturation is generally in agreement with calculations.
The variation of the decapole with current is shown in Figs. 11 and 12. At high fields,
the effects of saturation can be seen in both plots. The magnetization effect in the BNL-
Westinghouse magnet below 4 kA, a monotonic decrease of the decapole as the current
is ramped up and then down, Fig. 11, is typical, and in fair agreement with the
calculation. However, for two of the five Fermilab-General Dynamics magnets, the
decapole increases monotonically as shown in Fig. 12. It is possible that eddy currents
cause this behavior. Ramp rate effects are already known to have larger effects on the
quench currents of these 15 m magnets1,8 than on the quench currents of their smaller-
aperture, 17 m predecessors. Eddy currents may also account for the difference in the
width of the decapole hysteresis seen in Fermilab-General Dynamics magnets, Figs. 12 and
.u...(The width of the sextupole hysteresis in these two magnets is nearly the same.)
Measurements at different ramp rates will be made in future magnets.
At high field, a saturation skew quadrupole is produced because of flux leakage from the
saturated yoke, which is located above the center of the iron vacuum vessel. Uncertainties
in the centering correction (discussed below) limit the accuracy of this measurement at the
moment. A preliminary result is that a 1 decreases about 0.15 to 0.2 units at high field.
This is in general agreement with calculations, which predict a decrease of 0.1 to 0.2
units.
144
o
0.4
x
c
0.2 "
~
E
..o
u
- 0.0
I
o xxxx
o X
X
x
.D
.. -0.2 Ox
-0.4 )(
o 2000 4000 6000

CURRENT (Amps)
Figure 11. Normal decapole b4 versus current for a
typical BNL-Westinghouse magnet. Up-
ramp data are indicated by 0, down-ramp
by 0, calculation by X.
I I I
0.4 t-
)(
-
c
0.2 t-" -
.
I
~
.
E ~"WP
,.
u
I
0
0.0 t- ~~OII
o~~w,~ 1 lCiJi , . .
• .,tt •
...c.
~f(>O
.::.
/'
0
"
2-
.D
. -0.2 f-"
0
0
-
-0.4 '"'! -
I I 1 ....::
0 2000 4000 6000
CURRENT (Amps)
Figure 12. Normal decapole b4 versus current for two of
the five Fermilab-General Dynamics
magnets. Same symbols as figure 11.
145
I I I
x
0.4 -" -
000
00
-
X
0.2 T 0
X 00000 0 _
,
8u Xl 0°0°0ooo""~ _ _
-
Xx
1 0.0 D
X X x
l! X x oClov!-«> ;r : cW Udf
0 xX ,x X X
x"
!l xl!,
"rfll"<::l;1:P
.
"E
2- " "D" -
.&l
-0.2 - "CP"
0
-0.4 ..,. -
I I I
0 2000 4000 6000
CURRENT (Amps)
Figure 13. Normal decapole b4 versus current for a

magnet with measured hysteresis much larger
than the calculated values. Same symbols as
figure 11.
Sextupole Time-Dependence at Injection
On each magnet, multipole measurements are made while the current is varied as it will
be during injection and acceleration. A portion of one of these measurements is shown
in Fig. 14. After an initial quench and a cycle to high field, the magnet is brought to
injection (660A) and measurements start. The current is held constant for about an hour.
During this time, the sextupole becomes about 0.2 units more positive. When the ramp-up
is started, the sextupole "snaps back" to the value it had at the start of the constant-current
period. As the current is increased, the hysteresis curve is the same as if there had been
no constant-current period.
The magnets in this group have conductor from Oxford, IGC, and Supercon in various
combinations. The sextupole time variation of magnets selected to have cable from all
three vendors is the same to better than 0.02 units. This is interesting, because the cable
for the HERA dipoles displayed two significantly different, vendor-dependent time
constants3 •
Centering Corrections
Because the center of rotation of the coil used to measure the multipole coefficients is
not coincident with the magnet axis, the measured values of the coefficients will be
affected by feeddown from higher order terms. For these measurements, the coil lies
about 1-2 mm below the magnet axis. This offset is large enough that feeddown from
terms more than one order above the term of interest may be significant. An initial
understanding of this has been obtained by studying the effects of using different high-
order terms, bs and b lO, to make the centering correction. In these magnets, bs is twice
as large as b lO , so it is advantageous to use bs in making the centering correction if a way
can be found to remove the effects of feedown from blO • This has been achieved by taking
an initial value for the displacement from b lO , and then iterating this initial value using bg •
An interesting experiment, raising the measuring coil a known amount, has been carried
out at Fermilab. The experiment confirmed the overall method of making centering
146
3.4
3.2
.-..
.-..
N
I
."
S 3.0
-
I
0
.,
... 2.8
-a
-=-
.If
2.6
TIYE (seconds)
Figure 14. Time variation of the normal sextupole b2 at
660A (up to 2.8-3.0 ksec.) and on the ramp
up (after 2.8 ksec.) for four magnets which
contain inner coil cable from three different
vendors.
corrections. Further studies of systematic errors, using other centering methods such as
that for the HERA magnets, are underway. It is hoped that these will contribute to
understanding the current-dependence of the low-field skew quadrupole in certain magnets.
For this paper, the offset has been determined from the 20- and 22-pole terms (b1O)'
B. Dipole Angle
Thus far, the angle of the dipole field with respect to vertical has been measured with
a calibrated gravity sensor at 4.35 K in only one magnet. The important quantity in this
measurement is the difference between the warm and cold measurments, since the test
stand itself has not been constructed to replicate SSC installation. For this magnet, the
difference was less than 0.1 mrad, much less than the r.m.s. tolerance of 1 mrad. These
measurements will be made on additional magnets.
Room temperature measurements of the dipole angle before and after cold testing have
been made on all Fermilab-General Dynamics magnets. The average dipole angle showed
an increase ranging from 0.5 to 1.5 mrad. Also, there are indications of a change in the
magnet twist due to cold testing.
C. Integral Field
Precise measurements of the transfer function in the magnet straight-section, B/I, have
been made with an NMR probe in three magnets at 1.8 T. All three have B/I = 1.0453
T/kA. Integral field measurements have also been made with a combination of NMR and
Hall probes. Here, the accuracy has been limited by the absence of a precise measurement
of the axial position of the probes. Results from three magnets are about the same as the
fractional r. m. s. tolerance. 4
147
D. Short-term Reproducibility of MuItipole Measurements
Results from the first F-series "mole" rotating coil system are shown in Fig. 15. These
are of interest because this system will be initially supplied to the magnet vendors. The
plot shows the results of 50 measurements at 2 T and fixed axial position. For each
multipole, the r.m.s. width of the measurements has been divided by the SSC systematic
tolerance. All the ratios are significantly less than 1, indicating that, when the signal level
is adequate, the short-term variation in the measurement system is not a source of
measurement uncertainty.
2.0
Q)
u
Q
...
ttl
~
{3. 1.5
u
~
a:l
S
Q)
-'
~ 1.0 --------------------------.-----------------------
Ul
U
Ul
Ul
"-
.-.. 0.5
b
0.0
b 1 b2 b3 b4 b5 b6 b7 b8 a1 a2 a3 a4 a5 a6 87 a8
Figure 15. The RMS width divided by the SSC
systematic tolerance for each multipole
measured with the first F series field quality
measuring system (MOLE). The plots are
based on 50 measurements made at 2 kA and
a fixed axial position.
REFERENCES
1. M. Anerella et aI., "Construction and Test Results from 15 m-Long, 50

mm Aperture SSC Collider Dipole Models," paper to this conference
(IISSC92).
2. J. A. Carson et al., "Construction Experience with Fermilab-Built Full

Length 50 mm SSC Dipoles," paper to this conference (IISSC92).
3. P. Heger et al., "Design and Construction Experience with Collet End

Clamps for 50 mm Aperture SSC Collider Dipoles," paper to this
conference (IISSC92).
148
4. S. Delchamps et al., "Magnetic Field Measurements of Fermilab/General
Dynamics Built Full Scale Collider Dipole Magnets," paper to this
conference (IISSC92).
5. R. C. Gupta, G. H. Morgan, and P. J. Wanderer, "Comparison of the

Calculations and Measurements in the First 50 mm Magnet DSA207, " BNL
Magnet Division NOte 376-1 (SSC-MD-263), 1991, unpublished.
6. A. Devred et al., "Review of SSC Dipole Magnet Mechanics and Quench

Performance," paper submitted to this conference (IISSC92).
7. M. Wake at al., "Mechanical Behavior of Fermilab/General Dynamics Built

15 m SSC Collider Dipoles," paper to this conference.
8. J. Strait et al., "Quench Performance of Fermilab/General Dynamics Built

Full Length SSC Collider Dipole Magnets," paper to this conference.
149
THE ENGINEERING DEVELOPMENT OF AN ACTIVELY
CONTROLLED PRECISE MUON CHAMBER FOR THE
SDC DETECTOR
F. Ayer, E. Berk, J. Gonnan,

J. Govignon, and D. Sullivan
The Charles Stark Draper Laboratory

Cambridge MA 02139-3563
INTRODUCTION
As the detector configuration for the Solenoidal Detector Collaboration (SDC) evolved, a number
of concepts for MUON chambers were proposed and investigated. The Boston Group (Brandeis,
Harvard and Tufts Universities, with Draper Laboratory) has developed a concept incorporating
elliptical section drift tubes with field shaping and intennediate wire supports. This approach combines
good single and multi-track resolution with a small channel count and modularity at the tube level. Other
concepts have been developed which arise from differing interpretations of the fundamental physics and
economic requirements. These include:
• Octagonal Drift Tubes, Unsupported Wires, No Field Shaping U. Wisconsin
• Round Tubes, Unsupported Wires, w or wlo Field Shaping U. Washington
• JET Chambers (Multi-Wire, Intennediate Support, Field Shaping) KEK, Japan
This paper outlines the mechanical implementation of the Boston MUON chamber concept and
discusses the rationale for several key design decisions imposed by the stringent mechanical tolerances.
A prototype MUON chamber designed to verify design feasibility, perfonnance, and cost is also
described. A special section is devoted to the design and development of an optical alignment system
within the chamber, which provides error signals driving the intennediate supports to "true" position.
The Boston subgroup MUON chamber design is called the "wine rack" concept by virtue of its
numerous (non-structural) drift tubes laid into a supporting chamber, which provides structural and
handling capability. A number of individual e, <1>, and Stereo chambers is nonnally grouped into a
supennodule, which in tum is assembled into a multi-layer supertower. Supertowers are the basic
elements of the SDC MUON System, each comprising three supermodule layers in the barrel region and
five in the forward regions. The basic chamber concept and its relationship to the supennodules and
overall detector is shown on Figure 1.
DESIGN REQUIREMENTS
The first requirement laid on a MUON chamber is to accommodate the drift tube layout demanded
by the design specifications. The Boston MUON Chamber design specifications are described in detail
in the SDC Boston Area Group Concept Design Report or CDR (Jan. 3, 1992). These layout
specifications define the drift channel apportionment among e, <1>, and stereo orientations, the drift wire
relative positioning, and tube overlaps. Since the elliptical Boston drift tube incorporates field shaping,
the drift tubes must further be oriented so as to remain nearly projective to the interaction point. The
prototype MUON chamber layout shown on Figure 2 was fonnulated to represent a realistic geometry
some distance away from the IP, and requiring tube tilting for projectivity.
A primary requirement of the Boston subgroup MUON system is to maintain very tight location
tolerances on the drift tube, the field shaping panels, and the central filament. The current interpretation
of these tolerances is as follows:
Supercollider 4, Edited by J. Nonte 1 51

• Knowledge of wire position along its entire length to within 250 microns.
• Deviation of the wire from straight-line less than 125 microns.
• Deviation of drift tube with respect to wire less than 1 mm.
• Relative wire-to-wire position accuracy to within 250 microns.
• Positioning of wire ends to within I mm of nominal (with respect to beamline).
These requirements are conditioned both by the overall detector resolution goal «250 microns)
and the Boston drift tube configuration, which places the drift wire within 1 cm. of the high voltage field
shaping panels. The wire straightness and drift tube deflection requirements would be much less
stringent, for example, in the case of a large tube without field shaping.
Horl.onlal Eccenlric Siol
40.60-
loc:~ Screw Mounl

5.8
I
o"oleol Aag""*,,
Moun,•
....nk:al ECctnlric
5101.
_..:. j : IO~. Drift Tube eJe ranee Hole
-: I. , ...
- - 1·25" '
Figure 1. SDC MUON chamber, Figure 2. SDC Boston prototype

supertower, detector schematic. intennediate bulkhead showing tube
array and alignment provisions.
MUON CHAMBER CONCEPT

The MUON chamber concept followed logically from a series of decisions made concerning both
physics performance and mechanical implementation. The two primary tenets of this design were field
shaping and modularity to the tube level. An approach incorporating field shaping was chosen to ensure
adequate two track separation. The isolation of drift tubes from the structural support function was done
both to allow easy replacement of defective tubes and to minimize disturbances in tube performance.
The field shaping implementation was modified from that which had been used successfully on the CDF
MUON System, incorporating multiple conductor strips on G 10 boards disposed 1 cm. above and below
the (centered) sense wire. The rationale and performance predictions for this approach are described in
the SDC Boston Area Group CDR (Jan. 3, 1992). The combination of this particular field shaping
installation .and the desire to minimize a postiori corrections leads to the requirement of 125 micron
straightness in the sense wire. Using the traditional gold-plated tungsten wire, this straightness tolerance
demands wire supports at intervals no greater than 3-3.5 meters, with wire preloads at approximately
1/2 the material yield strength. This level of preload (approx. 350 grams for 50 micron diam.) has been
shown to be both reliable and stable in previous experiments.
The field shaping boards are fixed to the elliptical extrusion, and must be placed to within 1 mm
of their nominal position with' respect to the wire in order to preserve drift cell precision. This
requirement imposes a limit on extrusion deflection between supports of something less than 1 mm.
Parametric evaluation of both driftwire and tube deflections indicated that the support spans implicit in
the choice of tungsten for driftwire «3.5 meter) provided a good match with extruded aluminum drift
tubes of about 4 cm. depth in the least stiff axis. The use of advanced technology drift wires (plated
Kevlar, Graphite, or Boron) was at one time contemplated, allowing wire spans of from 4.5 to 9 meters.
In addition to their uncertain development prospects, however, such wires would have required drift
tubes of prohibitively expensive composite materials (or more tube than wire supports) to avoid
excessive tube/wire relative deflection.
The basic concept that emerged was then a 50 micron gold plated tungsten drift wire, supported
between field shaping boards disposed 2 cm apart within an elliptical-section aluminum extrusion. Both
the wire and tube are provided with intermediate supports at stations no more than 3-3.5 meters apart.
Since the maximum chamber length is less than 10 meters, the drift tubes are supported at no more than
two (and sometimes only one) intermediate points. These supports are provided by precision machined
bulkheads which possess clearance holes and quasi kinematic locating points for fiducial surfaces
extruded into the drift tube. Figure 2 illustrates one of the intermediate support bulkheads of the Boston
prototype chamber. Initially a 2x8 cm drift tube (4 cm drift) was designed and embodied in the Boston
prototype chamber. In order to reduce the overall channel count and system cost, a larger 2x 16 cm drift
152
tube (8 cm drift) was designed for use in operational supermodules, but not incorporated into the
prototype. Both drift tube cross sections are illustrated on Figure 3.
The requirement to support tubes/wires at approximately 3 meter intervals was imposed upon the
exo-structure box, which was originally presumed capable of providing the required straightness
passively. Parametric evaluation of the chamber deflections under gravity demonstrated, however, that
this was not true. Figure 4 illustrates the maximum deflections of a typical tube/box combination, simply
supported in a horizontal attitude with a 10 meter span. The abscissa indicates the box stiffness,
nondimensionalized by the stiffness (EI) of a single 2x8 em drift tube. The ordinate is the peak chamber
deflection for one of four distinct box/tube configurations. The two lower plots represent an aluminum
box structure supporting the drift tubes at one and two intermediate locations. It is evident that the
deflection asymptotes at an excessively large 2 mm, no matter how much the box stiffness is increased
(by increasing wall thickness). Another approach suggested was to glue all of the tubes into the box with
some type of epoxy foam. This result is shown on the middle curve, where it is seen that the overall
asymptote is not appreciably diminished. Finally, a very high specific modulus composite was tried for
the box structure, with the results shown on the upper curve. Although a significant reduction in the peak
.. nem Or----------------------------,
<n,t : .n~
~
Tubo Prom.
Original 8 CM X 2 CM OrIl1 Tube
wllh eln:umsc:rlbed elllpH
"Oem
..
c
el1eetlve open = lOme...
n ••", 8
~ . ,0000 IIA (BOX) : 45
..
~ 1 BUlK.lAl. Bo.n... lube.
E 2 BUlKJ Al. boxllree lube!
c 2 BUlKJ GR· EP boxllree lube.
U
AL bo unHI>.ec:llube!
· 15000 '-'----~-~-----------1
o SOOO 10000 15000 20000 25000 30000
11..., I{Box) I l(Tube)
Figure 3. SDC Boston Subgroup drift Figure 4. Parametric evaluation of drift

tube concepts. chamber structure options.
module deflections was achieved, the resulting asymptote of7 50 microns remains very much larger than
the 125 micron straightness requirement. At this point it becomes clear that an alternate solution to the
very stiff exostructure is necessary. The use of redundant box supports to the supporting truss and
MUON toroid was then considered for this purpose and discarded due to the resulting complexity and
non-kinematic support conditions.
The final solution for meeting tube/wire straightness requirements was to actuate the intermediate
support bulkheads in such a way as to compensate gravity (or other disturbance) deflections in the two
orthogonal inplane directions (transverse to the sense wires). Two directions of actuation are required
to accommodate the chamber deflections in all octants ofthe barrel toroid, where the local gravity vector
may take on four distinct orientations. This compensation is accomplished by driving the bulkheads
against the exo-structure with actuators under the control of a straightness monitor. Since the tube array
IIOVEA8L£ BUllOlEo\D
(I I, DtIIECllOfj$)
CAN IIOVE
• BUUU£AD IN 2 PEAP£HOICULAJI
• OIRECllOfj$ 11 em
Figure S. SOC Boston deflection compensating MUON chamber concept.
153
stiffness is two to three orders of magnitude less than the surrounding box structure, the adjustment
forces have negligible effect on the overall box deformations. The basic compensating box structure
concept is shown in Figure 5. An important artifact of this discrete, adjustable support concept is that
the only precision elements are the end and intermediate bulkheads, which define relative drift tube
positions. Since the intermediate bulkheads are adjustable in operation and the end bulkheads may easily
be installed with some fixed (shim) adjustment, the exo-structure box need not be constructed with any
great precision. The fabrication labor is therefore efficiently distributed between CNC machine tools
(bulkheads) and commonplace welding fabricators (box).
A prerequisite of this chamber concept is the ability to measure precisely the position and
alignment of the drift tubes in the box so as to trigger the proper corrective actions. The design and
calibration of an optical alignment system is therefore incorporated and its details are discussed later in
the paper.
PROTOTYPE CHAMBER MECHANICAL DESIGN
The Boston MUON prototype chamber is designed to verify the feasibility, cost, and performance
of the "wine rack" chamber concept, in all of its important physical attributes. Since the deflections of
wire and drift tubes are proportional to the square and fourth power of unsupported length, respectively,
it is essential to replicate a chamber length representative of the longest in the detector, about 9.15 meters.
A prototype thickness consistent with the standard unit of four phi or theta layers was also deemed
essential, leading to a minimum 0.4 meter thickness. The width of the prototype chamber is somewhat
arbitrary, but was here sized to accommodate 48 - 8 cm x 2 em drift tubes in a tilted projective array (-1
meter wide). The chamber structure specifications are described below, and were implemented in a
welded stiffened-box construction scheme.
Having decided that compensation of the tube/wire intermediate supports was the best design
strategy, the prototype chamber was configured to exercise the alignment and compensation system as
well as the basic drift tube and layout performance. The arrangement of tubes on the prototype
intermediate bulkhead is illustrated on Figure 2, along with provisions for four optical alignment
channels and the slots for position adjusting eccentrics. These intermediate bulkheads are attached in an
adjustable fashion to picture frame angle stiffeners welded to the box panels. Their transverse position
inside the box can be adjusted plus or minus 1 cm in both bulkhead principal directions, while
maintaining tight parallelism tolerances with respect to the end bulkheads. The attachment scheme must
further permit positive locking of the final adjusted position. A novel 2 degree-of-freedom flexure was
designed to perform the function of holding the adjustable bulkheads, but initial cost estimates suggested
that it would be too costly. A simple backup scheme was then formulated incorporating a locking
eccentric cam mechanism. This latter eccentric is the baseline for the prototype and is shown in Figure
6, along with the initial flexure concept.
The chamber structure supporting the tube array is a stiffened box with configuration and
dimensions shown in Figure 7. The box top and bottom panels are 3/8 inch 6061 aluminum plating, while
the side panels are 1/2 inch plate. In order to avoid panel buckling (oil canning) problems, the box is
further stiffened by rectangular bars welded to the top and bottom panels and running longitudinally, as
well as transverse "picture frame" stiffeners at approximately 1.5 meter centers. In addition, the sides
and bottom plates are welded at the comers through intermediate angle sections. The total prototype
weight is slightly over 2300 kg, with 1000 kg in the tubes and 1100 kg in the box structure (allowing an
additional 10% misc.).
Set in from either end of the box approximately 7.5 cm is a picture frame bulkhead or transverse
stiffener. To these frames the precision milled end bulkheads are bolted with some freedom to shim both
-v
fiXED
. / eUU(.HEAD
FlEXURES
OIRfCTlOH
-- J
OOF Screw-drlven lIexure Locking eccentric cam
Figure 6. Adjustable bulkhead support concepts.
154
r _. --
..-
-
.,. ,...--
~ : I
-
:
.........
•T
...
Figure 7. Prototype chamber structure layout.
the transverse position and out of plane angle. These adjustable features are intended as contingencies
against potential end bulkhead out of parallel (or square) due to box welding. On the top ofthe chamber
at either end and above the intermediate bulkheads are arranged removable "hatches", which are
intended to allow access to the optical devices and adjustment features while the box is resting on the
construction tooling. These hatches will be fastened about their perimeter at 7.5 cm centers (using 1/
4 inch screws) to the flanges formed by the box comer angles and transverse stiffeners.
The overall chamber center deflection when end-suspended in the horizontal attitude is predicted
to be 2.0 mm. The intermediate bulkhead deflection is predicted as 1.6 mm. In this situation, the box
ends tilt over (downward) approximately 0.001 radian. To cover the potential increase in deflections
due to unaccounted flexibilities, as well as fabrication deviations that may need correction, the bulkhead
adjustment range has been specified as ±1 cm using the eccentric cam adjustment mechanism.
ALIGNMENT SYSTEM DESIGN
As described above, the drift tubes and wires are supported at both ends and at two intermediate
locations by precision CNC-machined bulkheads. These bulkheads and their tube location features
ensure that the tube ends are positioned to within the CNC machining tolerance «25/J-m) with respect
to their array neighbors. The intermediate bulkheads likewise ensure the relative tube positions at the
third-points of the chamber span. The box is supported at or near the end bulkheads, and knowledge of
the support point positions will ensure knowledge of the wire end points from the accurate fabrication
of the bulkheads. Sighting fiducials will be placed on the chamber ends to provide this global reference.
The alignment is therefore split into two parts, (I) the "external" alignment of the box support points,
and (2) an "internal" alignment to maintain the straightness of the wires within the required tolerance.
It is this latter "internal" alignment we will consider below.
Assuming that the end bulkheads have been shimmed square and parallel in an assembly or
calibration fixture, the internal alignment is reduced to the proper positioning, in two orthogonal
directions (transverse to wire), of the intermediate bulkheads. Figure 8 shows a schematic cross-section
ofthe chamber under gravity load, indicating that the goal ofthe internal alignment system islo maintain
I" •• 1 BASIC 8UU(HU.o AUC/iHIII NT CONCEpT
.1 .uuotI..A.D . ' IU\.IlHUGf:::I
i'
0iQ" DODlOU' ..
0
-~ LlD C QuAOCI ....

LD••
""'"
0
......0 ,....
(i) Aller LED
OUAOCI:U
correction
Figure 8. Schematic of SOC Boston chamber Figure 9. SDC Boston MUON chamber optical
compensation mechanics. alignment characteristics.
155
points A,B, C and D in a straight line. The bulkhead deflections have been exaggerated in order to stress
the need for an alignment system which is independent of the bulkhead tilt.
The internal alignment solution adopted, shown on Figure 9, is a combination of simple optical
channels each comprising a Light Emitting Diode (LED) source, a lens, and a lateral detector sensing
the position of the LED image. Two such channels (for x and y coordinates and in plane rotation) are
needed for aligning one intermediate bulkhead with respect to the end bulkheads. A total of four channels
is therefore needed for the prototype chamber. The source and detector centers are used as references,
and the alignment consists in bringing the optical center of each lens onto the line defined by the centers
of the LED and the detector (zero output position). Component rotations due either to tilt of the
bulkheads or mounting misalignments do not produce any error in the sensed translation as long as the
rotation is about the center of the component and relatively small (a few degrees).
In order to ensure that alignment of the four monitoring optical channels will provide the required
drift wire straightness and parallelism, the optical center of each component must be accurately
referenced to the wire support points on each bulkhead. This is done by mounting and permanently
centering each component in a holding cell which has accurate external reference diameter, and
providing precise mounting holes for these cells in the bulkheads. Prototype designs of the cells for LED,
lens, and Quad Cell are shown on Figure 10, illustrating the precision (male) bore and the adjustment
features required to center components with respect to that bore. The actual component mounting and
centering is done on an optical bench fixture.
The four optical channels described above will provide an error signal for any sensed bulkhead
misalignment in the x and y translations as well as the inplane rotation. If a nonzero misalignment is
detected, mechanical means must be provided to remove it. The intermediate bulkheads of the prototype
END
8UlIIKEAD
CUIIPA1NG DETECTDR CONNECTOR
, CABLE ASS' IIOUSING
LENS
OUAD-CELl • ~~;;;;~
DETECTOR "
ADJUSllNG SCREW
(3 PlACES)
ST AKIHQ COIoIPOUNO
(3 PlACES)
Figure 10. Optical component mounting cells for MUON chamber alignment.
chamber are mounted to the box structure via eccentric cam-type devices (Figure 6) which are provided
with an integral nut to facilitate position adjustment. The operator observes a voltmeter, oscilloscope,
or other device indicating quad-cell output to obtain adjustment feedback. The fixed stem of the
eccentric is provided with a locking nut, turning against a keyed washer so as to fmnly hold a desired
cam position, without disturbing it in the process.
Based upon design trade-offs and breadboard testing, the prototype alignment system is
specified as follows:
• Quadrant Photodiodefrom UDT: Cost approximately $80/each (4x)
- Model # SPOT9DM1, operated in photovoltaic mode (no bias)
- 25 mm Cell Diameter, 10 mm Diameter active area
• Lensfrom Melles-Griot: Cost approximately $30/each (4x)
- Model #0 1LMP059
- 50 mrn Diameter, 38.1 mm Diameter clear aperture
• Light Emitting Diode (LED) from Motorola: Cost approximately $35/each (4x)
- Model #MFOEI102, AI-Ga-As
- Lamda = 820 nm, 250 micron spot size
• Bulkhead Positioning Device/Actuator: Cost approximately $50/each (8x)
- Custom Design Eccentric Cam, fabricated @ Harvard to Draper Drawings
• Optical Component Mounting Cells: Cost approximately $20/each in quantity (12x)
- Custom Design, fabricated @ Harvard & Tufts to Draper Drawings
• Detector Electronics: Cost Approximately $70/each (4x)
- Custom Design, Fabricated @ Harvard
The estimated purchase price ofthe alignment/actuation system is therefore approximately $1500,
or about $375 per channel. This figure clearly does not include the development efforts leading up to
the current design specifications, but is also quite conservative as regards unit price of systems purchased
in large lots.
'56
In selecting the above components, several design trade-offs were explored, the most important
of which was that for the detector. Three broad types of detectors can be used for sensing the
displacement of a spot of light. A mosaic of photodetectors coupled to a centroid-finding processor
provides large capture and linear ranges with great geometric stability, but is unfortunately very
expensive. A second approach is to use a planar photodetector with its sensitive surface cut into four
adjacent quadrants, creating four separate photodiodes. This type of detector combines low cost, high
precision near the center, and a very stable zero position (by construction). The third type of detector
is the lateral effect photodiode, with planar active area and output leads at the four comers. This type
of detector has a greater linear range than a quad cell of the same size active area, but is more expensive
and less sensitive in the linear range. The lateral effect detector is also less stable than the quad cell and
somewhat noisier. The quad cell was therefore chosen for prototype chamber construction.
The typical electronics necessary to process the output of a position sensor using quad-cell or
lateral-effect photodiode is composed of one low noise transimpedance (current to voltage) amplifier
for each quadrant or output of the detector, followed by two differential amplifiers, one for the X-
direction and one for the Y -direction. An improved electronics option can be constructed by modulating
the (LED) source at>200 Hz, and sensing the detector output with narrow bandpass electronics centered
on the LED modulating frequency. This approach has the advantages of eliminating the 1/f noise,
minimizing white noise, and eliminating ambient light effects. The significantly lower noise associated
with modulated sensing allows trade-offs to lower power sources (LEDs). A breadboard modulating
electronics has been built at Harvard and evaluated in both a bench alignment system at Draper and the
48 tube Boston prototype at Harvard.
The basic trade-off in light sources is betweenLEDs and Laser Diodes. This trade-off is essentially
a question of light output versus cost, with LEDs of suitable type at less than $100 and Laser Diodes at
approximately five to ten times that level with appropriate power supplies. Since the bench testing to
date indicates sufficient signal with LEDs, the trade-offs have been reduced to a selection between
competing LEDs.
PROTOTYPE CHAMBER TESTING

Several types of testing have been accomplished on the Boston prototype MUON chamber;
including bench testing of the optical alignment system, sensitivity checks ofthe system mounted in the
prototype chamber, alignment of the intermediate bulkheads from a random initial state, and alignment
stability check after lifting and dropping the chamber. The bench testing has demonstrated that the
optical components can be centered within their respective cells to a position accuracy of better than
±5 microns (2-3 microns typical). Figure 11 shows the calibration curves for the left/right portion of one
alignment channel when actually mounted in the prototype chamber. These calibration curves were
measured with the LED and Lens mounted in their prototype cells and respective bulkheads. The quad
cell was mounted on a micrometer adjustable optical bench fixture, and positioned immediately adjacent
to the mounting aperture in the prototype end bulkhead. Calibration was accomplished by moving the
quad cell in this case instead of the lens, since this arrangement was much simpler to accomplish. Note
that the capture range of the quad cell is approximately 11-12 mm, with a linear range of about 2 mm.
Ifboth the sum and difference signals are considered, an unambiguous correction direction is found, even
with the multi-valuedness ofthe difference curve. The quad cell therefore provides useful nulling signals
over its entire capture range. Figure 12 shows an expanded version of the central part of the difference
curve, illustrating the extreme sensitivity possible in the linear range. As indicated by the least squares
(Measurements taken on Prototype Chamber @ HUHEPL) (Sensitivity - 1 mV per micrometer of spot displacement)
0.6 0.5
_0.5 0.4
g 0.4 0.3
i 0.3 %0.2
; 0.2 >
-; 0.1
~ 0.1 u
" 0.0t---------.\-------=-j i 0.0
1;-0.1 ~ -0.1
§ ..(l.2 X 0111 (volt)
is
)( ·0.2
<IJ X Sum (volt)
)(·0.3 ·0.3
..(l.4 ..(l.4
·0.5 ·0.5-j,....~~~~~~~~~~~~~~....,
..(l.6-!---~~~~~~~~~~~~~~~
..(l.6 ..(l.5 -0.4 ..(l.3 ..(l.2 ..(l.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6
o 1 2 3 4 5 6 7 8 9 10 11 12 13 14
X Translation of Quad-Cell (mm) (Recentered)
X Translation of Quad-Cell (mm)
Figure 11. SDC Boston prototype alignment Figure 12_ "Linear" central region of SOC
calibration curve; x-direction (full range). Boston alignment calibration curve; x- direction.
157
fit, the detector position sensitivity is just about 1 millivolt per micron of effective cell displacement. The
sensitivity to bulkhead deflections is three times that of the calibration curve (3 millivolt per micron),
since the detector/source distance is 3x that of the lens/source.
Another series of tests accomplished on the BOSTON prototype exercised the whole alignment
and adjustment scheme. The procedure on this test was as follows:
1) Align Bulkhead A from Arbitrary Misalignment
2) Align Bulkhead B from Arbitrary Misalignment
3) Recheck Bulkhead A Alignment as Affected by Moving B
4) Lift Prototype off Supports and Suspend (reversed curvature) for 15-20 Minutes
5) Reset Prototype on Supports
6) Check Alignment of Bulkhead A
7) Check Alignment of Bulkhead B
The initial bulkhead alignments were accomplished with three of the four channels per bulkhead
(left side + right side, horizontal + vertical) coming in within < 5 microns of null, and the fourth
(horizontal) hovering around 10 microns off of center. The subsequent deviations from initial alignment
are shown in the following table. It should be noted that in resetting the prototype back on its supports,
it was inadvertently dropped for about the last 1/8-1/4 inch, causing a resounding clang.
Left Side Right Side
Change in A due to Aligning B < 1 micron (V) - 4 micron (V)

< 1 micron (H) - 2 micron (H)
Change in A due to Lift and Drop < 1 micron (V) - 5 micron (V)
< 1 micron (H) - Zero (H)
Change in B due to Lift and Drop - 11 microns (V) - 6 microns (V)
- 2 microns (H) - 1 micron (H)
These results were obtained with the eccentric cam adjustment scheme described above. It was
fairly difficult to tighten the cam locking nuts without affecting the alignment, and it appears from the
results above that bulkhead B was not secured with as much authority as bulkhead A. A self-locking
screw drive actuator mechanism would undoubtedly provide even greater stability with much superior
ease of adjustment. As shown here, however, the simple manual cam system does provide good results
with a patient hand on the adjustment wrench.
CONCLUSIONS
A self-adjustable MUON chamber concept has been formulated, designed, and executed to the
prototype stage with promising test results. This concept employs individual nonstructural drift tubes
supported at the ends and one or two intermediate locations to minimize wire/tube gravitational sag. The
tube supports are provided by precision bulkheads located within an exo-structure box of modest
stiffness. The stringent wire/tube straightness tolerances are maintained by an internal (to the box)
alignment /adjustment system which moves the intermediate bulkheads so as to maintain colinearity of
the four support stations (2x end & 2x intermediate). This approach has the important benefit of
concentrating all precision machine work in the supporting bulkheads. The exo-structure box may then
be inexpensively fabricated via common shop welding practices to indifferent precision, since the
bulkhead positions are adjustable. Both the optical straightedge system and the bulkhead actuators have
been designed, fabricated, and installed on a prototype chamber at the Harvard University High Energy
Physics Laboratory (HU-HEPL). Bench and prototype testing have demonstrated that the proposed
system meets or exceeds all SDC requirements for locating MUON chamber drift tubes.
The proposed MUON chamber concept can be implemented in several ways. The bulkheads might
be initially adjusted to true and the alignment sensors simply monitored to support a postiori analytical
corrections. This implementation would be best served by continuous lateral effect detectors rather than
quad cells, but provisions to account electronics drift and noise would be necessary. The second
approach envisions initial alignment plus continuous (or at least frequent) adjustment of bulkhead
position via servo-actuators driven by the optical nulling signals. This implementation is best served by
quad cells, and must pay the added expense of closed loop control. The final approach is a hybrid
incorporating initial alignment and infrequent (likely manual) adjustment of presumably low frequency
disturbances. Any of these approaches is feasible, with the best choice dependent upon system
optimization factors.
'58
MECHANICAL CONSIDERATIONS FOR A SDC
CENTRAL STRAW TRACKER FOR THE SSCL
R. L. Swensrud,1 2H.2
O. O. Hanson, Ogren, D. R. Rust, 2
D. Marschik, 1 J. W. Barkell, 1 D. A. Hoecker l and J. Mayhall3
lWestinghouse Science & Technology Center

Advanced Electromechanical Systems Departtnent
1310 Beulah Road
2Indiana University
Physics Department
Bloomington, IN 47405
30ak Ridge National Laboratory

P. O. Box 2008
Oak Ridge, TN 37831-6364
INTRODUCTION
The Central Tracker Detector (CfD) detector elements that do the actual particle tracking
require a support structure to support or fix them at the proper location inside the overall detector.
These detector elements or modules described in this paper are somewhat robust in and of
themselves, but they still require a relatively substantial support structure to assure alignment over
their four meter length.
Developing a sound conceptual methodology for this support structure is not a trivial matter
considering the huge size to weight ratio of the tracker. The 3.6 meter diameter and 8.0 meter
length generates a 90.0 cubic meter volume which is estimated to weigh approximately 1800
kilograms. This results in a very light average density of 20 kilograms per cubic meter. This
support system must be constructed using minimum quantities of radiation hard low radiation
length material while maintaining maximum rigidity and stability. With these goals and
restrictions in mind, a conceptual methodology has been developed for the support structure for
the straw tube modules. This support structure concept will be referred to as the "spaceframe
support system" in this paper. Figure 1 shows a view of the completely assembled central tracker
including modules, and Figure 2 a completed spaceframe support. A considerable amount of
physics evaluation and engineering analysis has been performed on this maturing concept. The
methodology dictates that to achieve our alignment and long term stability goals the following are
required:
1) Structural stability is achieved by using an absolute minimum number of mechanical

joints between graphite composite components,
2) Minimizing the tolerance buildup by using only a small number of large components
with simple geometry which are thus relatively easy to manufacture to high tolerances,
and

Plenum Press, New York:, 1992
Figure 1. Completely assembled central tracker.
SJ'f'O'· 'l·.·
Figure 2. Completed spaceframe support system.
3) Perfonning final alignment testing on the assembled support structure in its fully
simulated, in service gravity loaded support environment.
This strategy also reduces costs by requiring precision fits only at the major component and the
module-structure interfaces. Using the spaceframe support system also allows module and
structure component fabrication to occur in parallel. This is a major advantage because of its
potential to reduce fabrication cycle time and shorten the schedule. There are only two basic types
of large components used in the spaceframe support system. They are large graphite composite
cylinders and spaceframes. The space frame is shown in Figure 2, a basic support cylinder is
shown in Figure 3 and the assembly of the two can be seen in Figure 1. In the following sections,
the details of major component fabrications and designs are developed in more detail.
CTD MATERIALS REQUIREMENTS
From the structural engineering stand point, the material requirements of the central tracker
are very restrictive. The support structure must be fabricated with minimum quantities of material
160
but the final structure must have maximum rigidity or stiffness. Additionally, stability, minimum
creep, and resistance to deterioration from radiation are required.
Graphite fiber resin matrix composite has been selected as the leading candidate for the basic
structural material for a variety of reasons. This radiation hard material exhibits a high stiffness to
weight ratio with an above average nominal effective radiation length of 25 centimeters. The basic
fabrication technology exists for the proposed construction of large cylinders utilizing foam cores
and large single-unit spaceframes.
A quasi-isotropic [0/+60/-60/-60/+6010] layup has been chosen as the prime construction for
all carbon graphite composite components. This construction is a balanced and symmetrical
layup. Using the same layup for both the struts and the composite cylinders has advantages. The
cylinders need shear stiffness or a substantial shear modulus. Most of the deflection in gravity
loading without a good shear modulus will be cylinder shear deflection. For the spaceframe struts,
this layup results in a lower axial and flexural modulus than what can be achieved by placing more
fiber in the axial or zero[O] direction. This higher modulus would be achieved at the expense of a
lower strut shear modulus and an unacceptable unbalanced deferential of thermal expansion
coefficients between the cylinder and the spaceframe. During temperature changes the cylinder
would expand while the space frame would shrink and pull away from the cylinders at the
interface. The response of all assemblies to loads including gravity, thermal, and moisture should
be designed to be identical.
The resins systems being investigated included the epoxies and cyanate esters. Cyanate
esters are presently the leading candidate resin material because of low moisture absorption and
thus smaller elongation for a given humidity level.
SPACEFRAMES
The heart of the support system is the two spaceframes shown in Figure 2. The composite
spaceframes perform three basic functions:
1) Furnish four load points that support the tracker to the surrounding detector at the outside
diameter.
2) Support the silicon detector at the inside diameter.
3) Registers the five supedayer composite cylinders and thus the detector elements
themselves.
Each spaceframe is fabricated from high stiffness graphite components. These structures are
fabricated from struts, joints and rings. The struts which are the main component in the assembly
are thin walled tube shaped items made by heat curing of wrapped B-Staged graphite cloth. The
cross sectional size of these struts or tubes is 2 by 4 or 4 by 8 cm in cross section with a 0.2 cm
wall. These tube sizes have been analytically shown to be of a functional size. The joints which
connect the struts together are also fabricated thin walled hollow shaped items made by heat curing
of wrapped B-Staged graphite cloth.
The two sets of five rings which are the interface component to the cylinders are fabricated
by hand layups of autoclave heat cured B-Staged graphite cloth. The rings are manufactured with
oversized thickness for later machining either by specifying a thicker cross section or by
incorporating pads into their thickness.
Each spaceframe is assembled by adhesively joining the three components, struts, joints, and
rings into an assembly on a large fabrication tool. This tool is not required to be precision but
must be very stiff and fit snugly while rigidly holding the assembled spaceframe. Once assembled
the tool with the spaceframe is transferred to a three axis boring mill. On the mill each of the five
oversized rings is machined to a predetermined cylinder matching diameter. The diamond cutter
and high speed spindle generates only small tool forces and should not cause pan deflection. The
total operation consists of a single setup with only simple two axes moves. Diameter tolerance of
75 micron including a 50 micron concentricity should be achievable.
161
SUPPORT CYLINDERS
These fabricated, foam filled, ultra thin double wall composite cylinders are used as
structural members, with an equally important mission of furnishing a stable base for the modules.
The design features of the five cylinders can be viewed in Figure 3. Note that the construction
concept uses two composite face sheets and a foam filled core. The most restrictive design
requirement for the cylinders is the need to use minimum quantities of material to achieve the
lowest possible radiation length. The specific recommended layup construction is defined to be
matched to the space frame struts with a balanced and symmetric [0+60-60]sym layup.
The composite layup design was also selected based on the thinnest available high modulus
graphite fiber. A balanced and symmetrical layup with the needed in-plane shear modulus requires
6 plies of unidirectional fiber. Thus; each of two face sheets is 0.127 millimeter or 0.009 inch
thick foam are fabricated from 6 ply B-Staged tape lamina construction. This is considered to be
near the minimum thickness achievable. Hercules Ultra High Modulus fiber may achieve 0.006
inch thick composite layups.
The cylinders have a very large radius of curvature. A foam core is needed to give the
cylinders some out of plane bending stiffness and acts as a spacer between the composite face
sheets. Since achieving the maximum cylinder stiffness is a goal, the largest allowed core
thickness of 6 millimeter is used. This thickness limit is set by the total radiation length limit per
supedayer.
The end closeout design features of the cylinders can be viewed in Figure 3. This area forms
the mechanical connection with the spaceframe. A dense core material would be required to
withstand the forces that the mechanical connections will apply. This dense core material can
range from a denser 300 grade Rohacell foam to graphite composite layups. If a layup is used, the
coefficient of thermal expansion must be matched to the cylinder hoop expansion. Although more
engineering is needed, these issues do not appear difficult to solve.
The graphite hardware fasteners shown in the Figure 3 is commercially available.
Preliminary evaluation indicates that this hardware satisfies the requirements imposed upon it by
the interface connection. Other than for possible thermal expansion considerations, aluminum
hardware might also satisfy requirements.
The cylinders are to be formed on individual mandrels. These mandrels would be placed in a
taping machine similar to the concept shown in Figure 4 and 6 layers of the B-Staged lamina
would be applied. The foam core which would be previously thermally formed to the correct
radius would be placed on the mandrel and filament applied to hold it in place. After the end
closeouts were installed the last 6 layers of the B-Staged lamina would be applied. Vacuum bag
technology would most probably be applied to consolidate the composite. The number of cure
cycles required or whether an oven or heater strips would be used to cure the composite has not
been determined. There are advantages to keeping the cure temperature low, but this must be
weighed against other factors such as the resulting stability. Additional detailed engineering work
is needed in this area.
SHIM RING MODULE SUPPORTS
Predictable composite cylinder diameters are impossible to obtain without making several
parts of each size and iterative sizing the tooling. For this tracker program, trial and error sizing
would be prohibitively expensive. From the radiation length standpoint, machining or grinding of
a thicker outer wall composite layup is also not an attractive solution. A quality cylinder can be
produced on the first attempt by rotating the mandrel and part during the cure cycle, however the
resulting cylinder diameter and outer surface quality will be unknown.
Cylinder shim rings solve this problem of unpredictable diameter, surface quality, and other
problems. In this scheme the composite cylinders are fabricated in a normal manner. Prior to
removing the cylinder from the mandrel, the outer surface is fitted with relatively thick strips or
bands of low radiation length material bonded to the composite cylinder. A fabrication such as the
one shown in Figure 5 can be made in this manner. These shim rings can then be machined or
ground to a precision diameter.
162
7
\/
c;- -)h di'1 ~lo

1 ~ • .z ni" '50
) 21::2 Nt\,! '10
j!ii: ~(; Gl
Figure 3. The basic support cylinder for the modular tracker_
Figure 4. The mandrel used for construction of the cylinders showing the taping operation.
Figure 5. A completed cylinder with shin! rings attached.
163
The shim ring machining is done on a two axis machine equipped with a precision laser type
encoder that allows 2 to 3 arc second type indexing of the cylinder This accuracy translates
mathematically into 23 micron and 10 micron radial positioning accuracy for the largest and
smallest cylinders, respectively. Linear way systems are commercially available that will carry the
spindle and are straight to within 25 microns. It is a simple matter to optically align the mandrel
axis to the spindle way system axis to a high tolerance. A 50 micron parallel placement should be
achievable. Thus, shim rings can furnish a surface that can be machined or ground to a
predetermined diameter. They also make stereo or axial position measurement possible in that
they furnish a surface into which a stepped "hour glass" shape can be machined. The detailed
view of stereo and axial shim rings are shown in Figure 6. This same view is again shown in detail
with the modules in place in Figure 7. Axial or trigger modules require simple facets to be
machined into the shim rings. This operation requires the spindle way system and the mandrel
axis to be parallel and the mandrel to be indexed. Stereo modules require steps or shelves to be
machined into the shim rings at approximately 3 degrees. Stereo preparation of the shim rings
simply requires an adjustment in the machine tool such that the spindle way system and the
mandrel axis are parallel in the vertical plane but skewed in the horizontal plane by 3 degrees. The
accuracy of 3 degree specification is not a requirement but consistency from module to module is.
This consistency is automatically obtained because 360 degrees of modules are done in a single
setup.
MODULE ATTACHMENTS
Module attachment, in general, refers to the method by which the module is attached to the
support cylinder via the shim rings. A good view of shim rings and support cylinders without
modules is shown in Figure 6 and with modules in Figure 7. The modules will be located and
attached at approximately 80 cm intervals along their length. The shim rings will have fiducials
placed in them during the machining operation. These shim ring fiducials will mate with matching
module fiducials placed on the module shells during shell fabrication. The module fiducial
placement, since it is located by shell tooling, should be repeatable from module to module. The
shim ring machined locations should also be very good since each cylinder will be done on a
single setup and within the accuracy of the machine tool. For purposes of repair and maintenance,
the modules should be removable and replaceable from a fully assembled support structure. This
issue affects the complexity of the attachment. From the point of view of the support structure, a
wide range of options is available. The spaceframe support system can be assembled in any of the
following ways: .
1) The modules can be attached permanently to the shim ring cylinders before the cylinders
are assembled to the spaceframes.
2) The modules can be installed with removability features to the shim ring cylinders before
the cylinders are assembled to the spaceframes.
3) The modules can be installed with removability features to the shim ring cylinders after
the spaceframe support structure is assembled.
Option 3 is the preferred assembly method, since it allows for maintenance, repair and
replacement. An attachment method that has the required features is shown in Figure 8.
Provisions for shim ring module attachments are placed in the shim rings during the machining
operation that locates the shim ring fiducials. These shim ring attachments will mate with the
matching module attachment placed on the module shells during shell fabrication. The concept
utilizes an expanding tool that can reach into the inner part of the detector and release the springs
that hold that particular module so that it can be removed or replaced. This engineering effort on
module attachments is not yet complete.
164
.~
Figure 6. Detail of the stereo and axial shim rings.
Figure 7. Detail of shim rings with modules installed.
Figure 8. Module to shim ring attachment.
165
eTD EXTERNAL SUPPORT
The central tracker detector must be mounted in a stress free manner since the structure is
very stiff but somewhat fragile. From relatively small temperature changes large differential
deflections between the tracker structure and the available mounting surfaces can be calculated.
These deflections are generated by differential expansion between the carbon steel calorimeter
with its associated high thermal coefficient versus the graphite tracker with its near zero coefficient
of expansion. Shown conceptually i"n Figure 9 is a stress free system which can accommodate
radial and axial deferential expansion by allowing controlled relative motion while still anchoring
one comer of the tracker and maintaining position alignment. Radial and axial motion will be
accommodated by either the use of sliding low friction or tangent bar beam bending type mounts.
Figure 9. Central tracker external mounts.
REFERENCES
DeSalvo, G. J., and Gorman, R. W., "ANSYS Engineering Analysis System User's Manual," Swanson
Analysis Systems, Inc., Houston, PA 1089.
Outer Tracking Group of SDC, "Conceptual De,~ign Report for the Modular Straw Outer Tracking System,'
Solenoidal Detector Collaboration Note SDC-91-OO12S, Revision 1, January 21, 1992.
Trilling, G. H., et aI., "Solenoidal Detector Collaboration Letter of Intent," Solenoidal Detector
Collaboration Note SDC-90-001S1, 1990.
166
AN ENGINEERING DESIGN STUDY OF DETECTOR
DEFORMA TION LIMITS IN THE SSC SDC-DETECTOR
Kent K. Leung and Jeffrey L. Western

Dallas, TX 75237
ABSTRACT
The 18,500-metric-ton muon magnet support system (MMSS) is the major component
for the large detector proposed by the Solenoidal Detector Collaboration (SOC). The SOC
detector requires thick slabs of steel as an absorber. The purpose of this study is to
determine and understand the major impact the deflection of the toroid and stress in the
connecting bolts have on the magnet design.
INTRODUCTION
The design of the SOC detector called for limited deformation on the magnet steel
barrel toroid. 25 Finite element analysis (PEA) is utilized as a simulation tool to obtain
acceptable deflection and stress for the magnet barrel toroid. The entire assembly rests on a
10-ft-thick concrete slab which is supported by sub-soil. The steel slabs are first joined
together as blocks. The blocks are connected together as an octagonal section. Nineteen
octagonal sections are considered as significant and dispersed units for analysis. Each
section is joined by pre-tensioning bolts to eliminate gap opening. The magnet structure is
modeled in an axisymmetric configuration with the soil modulus applied as an elastic
foundation. The ANSYS finite element code is used for the present analysis.** The analysis
is not used as a failure analysis tool4 , 5 in structural design but is used early in the muon
magnet mechanical design as "a first pass" at the mechanical development cycle.
No. DE-AC35-89ER40486.
** ANSYS is a general-purpose finite-element code developed by Swanson Analysis Systems, Inc.

INTEGRA TED MODEL
The muon magnet support system (Figure 1) is composed of five units. The model is
based on the conceptual design.3 The five units are:
Steel Barrel Toroid: An 18,500-metric-ton octagonal steel barrel toroid is modeled
with ten sections. A section is made up of four long blocks and four short blocks. All the
blocks are 1600 mm thick and made up of ten steel plates. Axial symmetry is used for
conservatism and simplicity in the finite element modeling.
Connecting Bolts: The connecting bolts are employed to connect each section and join
the blocks to form the section. The stiffness of each pre-tensioning Z- Bolt system is
calculated by assuming a 45° shear cone in the clamped steel plate.
Support System: Hydraulic jacks are used for supporting a plate web girder which
carries the magnet barrel toroid. Stiffness of the hydraulic jack and the support system
vertical stiffness is obtained from a FEA model of the support. 16 , 17 The jack is modeled as a
compression member only to prevent uplift to the toroid. Coupling is employed in the toroid
and girder to account for frictional force.
Concrete Slab: A 3-m-thick slab with effective width of 17.7 m by 50 m is used. The
sub-soil properties are modeled as an elastic foundation.
Soil: The soil is recorded as Austin Chalk with an average sub grade modulus of
1000 lb per cubic in.
LOADING CONDITION
Moving load* locations are indicated in Figure 2. The loading is a concentrated

loading acting on top of the long block in the bottom toroid. Dead load is computed as 1 G
(gravity) for all the units of the integrated model.
BOUNDARY CONDITIONS
Boundary conditions are regarded as symmetrical in x = 0 plane and in axial z = 0

plane directions. These boundary conditions are rigid because only two translation degrees
of freedom are allowed. The boundary conditions in the model required symmetrical
loading and therefore moving loads of the same magnitude are acted on both ends
simultaneously.
FINITE ELEMENT MODELS FOR ANALYSIS
Model A: The Basic Model. This is the basic model of the muon magnet support
system MMSS with the number and location of all the connecting bolts considered in the
current toroid design and the current supporting system design.
Model B: Basic Model with 5 times the "c" Bolts. This is an identical model as model
A except that all the "e" bolts at the same location are increased by five times.
Model C: Basic Model with 5 times the "c" and "z" Bolts. This is identical to model
B excepted that all the "z" bolts at the same locations are increased by five times.
*The moving "forward toroid" load weighing 4000 MT moves into the barrel and induces strain, which is
considered as maximum at this particular position.
168
~~~~~i§;:;"., U
'I:::::::~~;~ii~i~~
. ,
Comer bolls (C-Bolt) are used to join the
long block to the short block together
with shear keys in the perpendicular
direction to the C-bolts. Four long blocks
and lour short blocks are united as one
toroid section.
One vertical long

block shown here ~--I....d!~t::11
Two short blocks

All Z-bolts are connecting bolts in
the Z direction with pretension to
tie the toroid sections in loming
y the SSC SOC muon barrel toroid
(SliI4).
Beam element at top of the Hydraulic Jacks

is used to add a longitudinal stiffness 01 a 3
m deep beam. This beam is ooupled at the
interface between the bottom section of the
barrel and the top 01 the beam is assumed
(non-sliding friction) (Stif 4).
'-..z Concrete slab is 3 m thick with elastic

foundation supporting at the bottom with
an elastic spring constant = 1000 poi,
considering Austin Chalk.
Figure 1. Two sections of MMSS with Gap Size Expanded to Show C-Bolts and Z-Bolts as Connecting
Elements.
r""S
------ ,
iIVertical
Direction
Reaction on nonlinear
spring (Stif 10)
elements with only
Finite element model compression stifness
Longitudinal direction is on this part.
Figure 2. Boundary and Loading Conditions of MMSS.
169
RESUL T OF ANALYSIS
The vertical deflection at the bottom Magnet Barrel Toroid (MBT) is considered as the
important analytical result for the following reasons. The moving "forward toroid" load (see
Figure 2) acts on this portion of the MBT, consequently producing maximum deflection.
The lower portion of the MBT is the interface to the bottom supporting girder. Friction
between the girder and the MBT is presumed to be large, and no relative movement between
the MBT and the girder would occur. The lower portion of the MBT needs to support the
upper portion of the MBT, accounting for more than half of the 18,500 metric-ton dead
load. The lower portion of the MBT will be in tension because of the moving load and the
elastic foundation action. Excessive separation of the sections will impair the function of the
MBT because of the magnet requirement.
This paper is focusing on the MBT deflection and connecting bolt stress (Figures 3-7).
The effect of toroid deflection by the soil spring value and the thickness of the slab were
reported in SSCL Magnet Engineering Document, Engineering Notes No.:
MD-ENG-92-A-OOl. 24
10~------~====================~--1
8
o Max. def.(mm) forward toroid load
o Min. def.(mm) forward toroid load
Differential def.(mm) of max . & min.
6
4.33 4 .351 4.309
2.789 2 .78 2 .782
1.541 1.572 1.527
Basic + C + Z Basic +5C + Z Basic + 5C + 5Z

Model
Figure 3. Maximum Vertical Deflection (mm) of the Steel Toroid.
DISCUSSIONS
The following conclusions were drawn from this study.

Connecting bolts are not effective in controlling the toroid deflection; stress (308 kips)
in the "Z" bolts is excessive (for 2.25-in-diameter bolt, the AISC allowable is 84 kips); high
bolt stress and deflection areas are found near the moving load zone; stiffness of the support
system and concrete foundation is critical in the design; deflection induced by moving loads
is highly localized and has marginal effect on the other parts of the toroid; stress in the
toroid and comer bolts is adequate by AISC standard; high stress in the "Z" bolts is induced
170
o Max axial global def (Z)
13 Max interface differential (Z)
0.169 0.178 0.17
.1
Basic + C +Z Basic + 5C + Z Basic + 5C + 5Z

Model
• The def. portion starts from the lowest part of the toroid in the area
close to the 4000 M. Ton load.
Figure 4. Maximum Longitudinal Deflection (mm) of the Steel Toroid.
10.---------------.--------------.---------------.
o Max. horiz. def. (x) @ outer vert . wall of the toroid

liD Max. vert. def. (y) @ top center portion of the toroid
8 • Max. axial (z) def. @ moving location close to edge
E
.sc 6
6.469 6 .297 6.3
.2
~
~
-0
~
Q)
4
>
a
2 1.814 1.739 1.739
Basic + C + Z Basic + 5C + Z Basic + 5C + 5Z

Model
Figure 5. Maximum Overall Deflection (mm) of the Steel Toroid.
171
Maximum forces on • Z •bolts under
4000 M . ton moving loads condition
400~------------~-------------r-------------.
o Max. Zbolt force (Kips) - for locationsee fig. 7a

• Max. shear force (Kips) - for location see fig. 7b
308
300 297
Ui'
a.
~
~
Q)
u 200
.2
(5
CD
100 93
Ol..-...J-_ _
Basic Basic + 5C Model Basic == 5C = 5Z

Figure 6. Bolt Force on Bottom Part of Steel Barrel Toroid.
Note: X· Y plane is the symmetry plane for the long~udinal axis of MMSS.
y Y·Z plane is the symmetry plane for the lateral axis of the MMSS .
Z
~~~ _ _~_ _~_ _-L~~J-~LL_ _~_ _~L--L~~ ~
Note: All Z·bolts and C·bolts are Gap size enlarged tor c larity.
2" dia. high·strength bolts with Gap sizes i-1 actual model for
initial tension as tlgh as practical execution are close 10 zero .
to create a friction connection.
Figure 7. "e" and "Z" Bolt Position on Bottom Part of Steel Barrel Toroid.
172
by the moving loads; deflection on the bottom of the toroid is produced by the moving
loads; toroid support needs consideration for distribution of the moving loads; the locations
of the toroid support are effective in carrying the body weight of the toroid but ineffective in
transferring the moving load to the supports.
RECOMMENDA TIONS
As a result of this study, the following recommendations are made for future work:
Adding a rigid support is possible directly under the moving load on the toroid to transfer
the loads into the supporting girder as much as possible; provide a very rigid loading
platform to distribute the moving loads to a larger area of the toroid and support; perform
additional finite element analysis in determining the most effective way to reduce the bolt
stress and deflection; investigate additional support options such as increasing the stiffness
of the support system and of the concrete foundation; investigate other Z connection options
such as key ways across the bottom of the toroid, continuous plates, cables and staggered
blocks.
CONCLUSION
We have succeeded in understanding the major influence the SDC muon magnet
deflection has on the magnet barrel toroid block stiffness and the moving load moment arm,
which was measured from the girder supporting point to the moving load acting point. We
also discovered a problem with the high connecting Z bolt stress and have recommended
several methods to minimize it. The movin "forward toroid" load is the most critical loading
condition for the muon barrel toroid. The results show that the critical component of the
toroid is the Z connection bolts. Our analysis indicates the Z bolt has little to do with the
overall deflection, but is overstressed. We also predict the maximum differential deflection
on the magnet barrel toroid is within the specified deflection limit at 3.0 mm.2S The
maximum differential deflection of the muon barrel toroid, as calculated from the present
analysis, is about 1.6 mm. Continuous use of FEA as a simulation tool in the advanced
design study will reduce engineering changes and achieve a "first pass final" in the final
SDC detector mechanical design.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the insightful comments and contributions of
the following people: J. Bensinger, J. Cherwinka, R. Gates, G. Snitchler, C.W. Tseng, and
T. Thurston.
REFERENCES
1. V. L. Alley et al., "Prediction and Measurement of Natural Vibrations of Multistage Launch Vehicles,"
AIAA Journal, Vol. 1, No.2, Feb. 1963.
2. C. G. Salmon et al., "Steel Structures," p. 148, Intex Ed. Pub.
3. A. Rothbart, "Mechanical Design and System Handbook," pp. 21-28, McGraw Hill.
4. K. K. Leung, "Seismic Stress of Piping System and Equipment on Heat Exchanger Supporting Structures,
"3rd International Conference on Structural Mechanics in Reactor Technology," Vol. 5,
part K, section 7-6, London, Sept. 1975.
173
5. K. K. Leung, "Stress Analysis of ALS Magnet Supports Subjected to 7.0 Earthquake at the Lawrence
Berkeley Laboratory Light Source Site," Lawrence Berkeley Laboratory Engineering Note,
LSME-201, AL 0521, M6869, July 1989.
6. K. K. Leung and T. Nicol, "A Frequency Response Study of Dipole Magnet Cold Mass End Shell,"
p. 295, Supercollider 3, Plenum Press, NY and London, 1991.
7. K. K. Leung, "Stress Analysis and DeSign of Booster Ring Girder Subjected to Seismic and Thermal
Loads," Lawrence Berkeley Laboratory Engineering Note, LSME-176A, AL 0521, M6840 A,
May 1989.
8. K. K. Leung, "Seismic Dynamic Design of Booster Ring Girder Anchor Bolt and Base Plate," Lawrence
Berkeley Laboratory Engineering Note, LSME-l00, AL 0521 M6854, July 1989, (Rev. 2.0).
9. M. S. Lin and K. K. Leung, "Final Seismic Analysis Report of Fast Flux Test FaCility," p. A4, Bechtel
Power Corp., 1975.
10. J. Jayakunar, K. K. Leung et al., "Mechanical and Electrical Analysis of 50 mm Design for The SSC
Dipole," IEEE Transaction On Magnetic, Vol. 27, No.2, March 1991.
11. K. K. Leung, "Nonlinear Stress Analysis of Superconducting Dipole Magnet Shell,"ANSYS Fifth
International Conference Proceeding, Vol. n, pp. 5-41, May, 1991.
12. S. Delchamps, K.K. Leung, et al., "ssc Collider Dipole Magnet End Mechanical Design," IEEE Particle
Accelerator Conference Records 91CH3038-7, Accelerator Science and Technology, Volume 4,
San Francisco, May 1991.
13. SSC Laboratory, "SDC Steel Toroid Specification," July 15, 1991.
14. H. A. Rothbart, Table 21.3, "Stiffness of Bolted Assembly," Mechanical Design and System Handbook,
McGraw Hill.
15. Personal communication, J. Cherwinka, Physical Sciences Laboratory, University of Wisonsin-Madison,
Dec. 1991.
16. Personal communication, J. Western, SSC Lab., Dec. 1991.
17. Jim Krebs, "Result of Modified 3-D Model of Support," presentation at SSCL Physics Research Div.,
Dec. 1991.
18. K. K. Leung, "Dynamic Motion Study of an Electron Storage Ring for The Third Generation of
Synchrotron Radiation Facility," Lawrence Berkeley Laboratory Engineering Note, LSME-169,
AL 0521, M6825, April 1989.
19. B. Seed and L. Wysmer, "Soil finite element codes LUSH, SHAK," University of California, Berkeley,
California.
20. SSC Laboratory, "SDC MBT Conceptual Design, Dwg. # R20000337," Contract No.
DE-AC35-89-ER-40486.
21. M. Hetenyi, Beams on Elastic Foundation, The University of Michigan Press.
22. PBIMK, "Preliminary Soil Data," Nov. 1991.
23. R. B. Hopkins, Design Analysis of Shafts and Beams, pp. 177-179.
24. SSCL Magnet Engineering Document, Engineering Notes No.: MD-ENG-92-A-OOl.
25. J. Bensinger, SSCL Barrel Steel Defmition and Specification Proposal, (Version 2.6, Jan. 14, 1992).
174
THE PROTOTYPING AND EVALUATION OF REMOTE ALIGNMENT
MONITORING SCHEMES FOR SDC MUON SYSTEM SUPERMODULES
David P. Eartlyl and Paul T. Johnson 2
1 Fermilab, P.O. 500, M.S.205, Batavia, lllinois 60510

2 Hutchison Career Center, Fairbanks, Alaska 99709
INTRODUCTION
The central Muon system of the Solenoidal Detector Collaboration

detector (SDC) for the Superconducting Super Collider (SSC) will be made up
of a large number of proportional drift chamber supermodules grouped in
three layer (inner layer BW1, outer BW2, BW3, or IW2, IW3 layers) projective
towers on a 28m long conventional barrel toroid magnet, with chamber
spatial resolution of 250 J..Lm. The Forward system will be made up of four or
five supermodule layers inside (FW1, FW2) and outside (FW3, FW4, FW5) the
forward toroids in each end of the barrel toroid with chamber resolution like
100 !lm. At a minimum, the inner and outer layers are independently
supported on kinematic mounts. All layers may be independently supported.
To insure the spatial measurement resolution of the systems, we must
monitor the positions of supermodules with sensor systems which have
better long term stability and position resolution, where long term
corresponds to six months to a year. Thus we began an R&D program eight
months ago to develop and test relative position measurement schemes over
such long periods to establish viable prototypes. We have tried to maximize
the use of commercial systems or parts.
SENSOR SYSTEMS UNDER TESf
The first sensor system under study is a metal target, linear analog
output, proximity sensor scheme. We have studied both inductive and
capacitive sensors. Tests on various inductive sensors in various orientations
of magnetic fringe fields up to 120 gauss indicate only small shifts of scale on
the output. Measurements of the linearity of output vs distance and the linear
range have been made on various sensors, target materials, and
configurations. For at least one brand of inductive sensor, a test group of
individual sensors had essentially identical voltage/distance slopes and only
differed by the threshold point distance for the linear response region.
Work supported by the U.S. Department of Energy under contract No. DE-
AC02-76CH03000 and contract No. DE-AC35-89ER40486.

We have set up fixed sensor-target configurations which include
temperature monitors and have subjected the setups to long term thermal
cycling. Long term monitoring of these is being done via a Keithley
multichannel scanning DMM system, IOTech488SCSI controller, and MAC
computer. Using the computer clock, typically a reading is taken every
minute for thirty minutes and averaged into a data value. Some fixed sensors
have been monitored this way for up to five months. Detailed information and
results are given in reference 1. Figure 1 illustrates the measured output of
a fixed linear analog inductive proximity sensor for a summation of one week
runs over a three month period as compared to the temperature distribution.
6.77 . , - - - - - - - - - - - - - - - - - - - - - - - - - - - - ; : ; - - - , 100
Sensor dv/dx 2.51 mY/micron

Raw Data
90
u.
6.75 Cl
(I)
'0
2! ~
'0
> ~
(I)
a.
E
6.74 (I)
~
80
~ South table Proximity sensor readings
• Temperature distribution
6.72+-------~-------~-------T-------+70
o Trials 1000 2000
Figure 1. South table reference inductive proximity sensor readings over 1000 hours
compared with the temperature distribution over the same interval.
The second sensor system under study is an infrared LED-optics-

transverse position sensing photodiode scheme. This system could be used to
measure the relative transverse positions of references in two physically
disconnected systems or segments. We have established model setups for
3.0m (BWI to BW3), 103m (BW2 to BW3), and O.3m (IW2-FW2) optical path
separations. A number of source diodes, optics configurations, quadrant and
continuous light spot position sensing diodes were evaluated to optimize the
range of model setups.! Running the systems DC, we have studied the effects
of various longpass optical filters on light background effects and photodiode
response. 1 Long term stability tests are being made on the model setups.
Another relative transverse position measurement system under study is
an AC (3Khz) wire current-differential flux pickup transformer scheme. 2
Two very different transformer prototypes have been tested and indicate a
linear voltage output of the left-right bucking pickup coils with wire
displacement form the center of the transformer as shown in Figure 2.3
176
3.0
! - - •. _ _ _ _ _ _ • ___ .C_ ) _ •. _ .. _ _ . . _ _. • . _ _ _ _ _ _ _ _ ~_
-.
f\,\ n_ ;:d r:~:;:!L~·:-.p·-'1·'3-;-~-:+~~-h-f~e w:ire~'.-d-V/-dX-=-c-0.-02c-3··2-V·IUl "'-'-I-

2.0
- --__-_-
...-_
1.0.-=\,-;--,~-7--,--',----c'---7~--;----c-+-,,---' ---c.-'-'--l:---c-'---cc---;--.-,-~--
:'\
0.0
Data taken by Z. Knaebel
...
'"
E
<8 -3.0
en
~"
i , \ \i • .--C--
i
-4.0
i i • i ' i,\". •
-5.0
,~ • i : i ! •• • : ,.• , I.. .'\.-1:,------'-----_---;----'----j
... For vertical displacement at constant x. dv/dy < 0.00003v/mii .'\ .
r--------;--_ ! . i ....:. ' • : . i ." ., I
'i • • \
-6.0 .
,
, ", J. .
-7.0
'.
-.150 -.100 -.050 0.000 .050 .100 .150 mch
-3.81 -2.54 -1.27 1.27 2.54 3.81 mm
Figure 2. HJ transformer-electronics left-right coil difference output vs the position of an

enclosed 3Khz @300ma current carrying wire.
Long term stability tests have been started with current and temperature
monitoring_ A smooth temperature dependence is observed_! We will study
other transformer configurations to optimize the response and stability and
attempt to utilize commercial LVDT circuits in some new output electronics.
To measure changes and distortions in the magnets, support systems,
and supermodules, we have been testing a sample group of precision,
thermally stabilized, liquid bubble inclinometers developed for the DO detector
at Fermilab_ 4 In a fixed setup, we are cross comparing a number of bubbles
to establish resolution and long term stability estimates.! Measuring the long
term stability is complicated by the thermal wiggling of our test support
structures. Weare going to reestablish these on a stable base to improve our
limits. On the short term, the inclinometers appear to have the 25 ).lradian
resolution to monitor the inclinations of our spatial position referencing
structures.
177
Table l.The measured properties of the various local displacement and angle sensor systems
under study; resolution, average value, deviation, extreme data values.
DBASE SENSOR M DVIDX RES <v> STDEV DT DVMAX DVMIN
I. Transverse and Longitudinal Fencepost Proximity Sensors
Sum7 MS18 d 0.0627 0.11 6. 7585v 0.0035v 77.56 +4.45 +O.OO71v -O.OllOv
Raw Data v/mi! fl (l.4fl) degF -4.27 (2.9f.l) (-4.5fl)
Sum8 MS30 Zw 0.0500 0.25 4.7648v 0.0042v +0.0135v -O.Ol58v

Raw Data v/mi! fl (2.1fl) (6.9fl) (-8.Gf.l)
Sum6 TK18 d 0.0573 0.21 7.1000v 0.0132v 78.31 +7.09 +0.0285v -0.0455v
Raw Data-Big T dep v/mil fl (5.8fl) degF -6.98 (12.4fl) (-19.8f.l)
Sum6 TK18 d 0.0567 O.0084v +O.Ol48v -0.0228v

Parametric Tcorr v/mil (3.7fl) (6.5fl) (-to.Ofl)
II. Transverse Optics ( 25 mm Continuous position sensing photodiodes)
Sum4 SC25D x 0.0111 2.3 -2.9124v 0.0496v 81.14 +12.18 +O.l43Ov -O.1099v
Raw BW3-BW2 v/mil fl (1l3fl) degF -11.59 (327fl) (-251fl)
Sum4SC25D x 0.0111 0.0478v +0.1010v -O.l063v

Parametric Tcorr (109fl) (231fl) (-243fl)
Sum3SC25D y 0.0111 2.3 -0.0709v 0.0105v +0.038Ov -0.0269v

Raw BW3-BW2 v/mil fl (24fl) (86fl) (-G1fl)
Sum3SC25D y -0.0709v O.OO63v 82.16 +11.16 +O.0338v -0.013Ov

Parametric Tcorr (14j.t) degF - 6.13 (76fl) (-30fl)
Sum4SC25D x 0.00136 13 -0.262lv 0.0073v +0. 0 l86v -O.Ol42v

Raw BW3-BW1 v/mil fl (134j.t) (34Ofl) (-26111)
Sum4SC25D y 0.00136 13 -0.OI5Ov 0.0064 +0. 0 147v -0.0169v

Raw BW3-BWl v/mil 11 (11711) (27011) (-3 1011)
III. Wire -Leakage Flux Transformers-Transverse position or distance
Fairl HJ x 0.0232 3.3 -4. 530v 0.056lv 84.34 +4.40 +0.1l4Ov -0. 115 Iv
Raw 2216612 v/mi! 11 (6011) degF -3.20 (12311) (I24j.t)
FairlHJ -4.53Ov 0.0457v +0.087Ov -0.0.914v

Parametric Tcorr (5011) (9511) (10011)
IV. Two Degree Inclinometers-Tilt angles
Sum2 INCl,B 9 1.25 26.8 -0.OI9Ov 0.1322Y 77.56 +4.92 +0.3692v -0.424Ov
Raw mY/sec flrad (512f.lrnd) degF -4.27 (real) (real)
Sum2INCl,B -0.OI9Ov 0.1195v +.3748v -0.3653v

Parametric T corr (46311) (real) (real)
Sum2 INCI,B--INCl,D +0.0003v O.OO04y +0.0013v -0.0008v

(1. If.lrnd) (I.6f.lrnd) (4.9f.lrnd) (-3.lf.lrnd)
178
Table 2. The measured properties of some distance measuring sensor systems under study.
Resolution, average value, deviation, and extreme data values are given.
DBASE SENSOR M DVIDX RES <V> STDEV DT DVMAX DVMIN
I. Long Range mtrasonic Distance
Sum7 micro d 0.250 343 5.040v 0.0054v 77.56 +4.45 +0.0191v -0.012Ov
Raw 945L my/mil ft (482.6mm) (540ft) degF -4.27 (1.91)mm (-1.20)mm
Sum7 945L 5.040v 0.0035v +0.018Ov -O.Ol23v

Parametric Tcorr (350ft) (1.80)mm (-1.23)mm
II. Laser Distance Device (Keyence LBll+LB70)
Sum7 LBll d 2.457 0.4 2.994v 0.0121v 77.56 +4.45 +0.0273v -O.028Ov
Raw Data my/mil ft (125ft) degF -4.27 (282ft) (-289ft)
Sum7 LBll 2.994v 0.0056v +0.0105v -O.0161v

Parametric TcoIT (58ft) (108ft) (-166ft)
179
For precise short range distance measurement between disconnected
elements, we have evaluated a commercial (Keyence LBll-LB70) laser diode-
optics-position sensing photodiode, linear analog output, sensor capable of
measuring a reflective target distance over a nominal range of 60-140mm.
Short term resolution of 10 !tm can be achieved when background light effects
are suppressed. Long term tests with suppressed background are in
progress. We are trying to unfold sensor thermal fluctuations from support
thermal motions to establish a long term resolution.
We have also studied two ultrasonic distance measuring sensor systems
for possible measurement of large distances. One is a simple self contained,
temperature compensated transmitter and receiver (Microswitch 945L). It
has a typical short term resolution like 500 !tm, has small thermal
fluctuations, and is capable of measuring target distances to 600mm.l A
second system is a far more complex, microprocessor controlled,
multichannel system (Migatron) where one calibration distance channel is-
used to correct all channels for thermal velocity changes and turbulence
effects. The system stabilizes electronic gain on the reflection bursts and
triggers on a given threshold-fourth pulse in the reflection burst and stack
averages readings as specified. Eight channel string readout occurs through
a serial port. Under controlled conditions, the system has a short term
resolution of 200!tm for target distances like 1200mm. Long term tests are in
progress to determine long term resolutions. Corrected sensor channels have
no evident thermal fluctuations but there are some observed random shifts
on the order of Imm. The microprocessor programming is been modified to
attempt to eliminate this effect.
RESUL1S
From the computer cycle measurements on fixed configurations we have

large sample reading databases on the long term stability of these systems.
These databases are presently being analyzed. Short term resolutions,
average data values, deviations, and extreme data values are being
determined. The measured properties of the various local displacement and
angle sensor systems described are listed in Table 1. Table 2 lists the
measured properties of some distance measuring sensors. We have allowed
more than 20 deg F fluctuations during our data taking. The Raw data
generally exhibits some thermal fluctuation effects. Some sensor readings
increase with temperature while others decrease. Resolutions are given in
Table 1. Some sensors exhibit little hysteresis in thermal fluctuations while
others exhibit a lot. For sensor systems other than the capacitive proximity
sensors, the major fluctuations in sensor readings do track the thermal
fluctuations. As such fluctuations are generally small and have some
mechanical component (motion) in them, we have tried to apply a simple
parametric linear temperature variance correction to the data for sensor
systems with little hysteresis. Some resultant temperature corrected (Tcorr)
resolutions are also given in Table 1 and Table 2. These resolutions will
improve as we learn how to make better corrections and reduce the
temperature fluctuations, as in the controlled detector environment.
CONCLUSIONS
Generally, the inductive proximity sensors are very linear in output vs
displacement. They have exhibited a very good long term resolution (few !tm)
and can be used in the SDC conventional magnet fringe fields. Based on
180
these results, we have proposed a system of boreholes through the barrel
toroid through which we project mechanical rulers or "fenceposts" from the
inside layer BWI through the outside layers (BW2, BW3) at the corners of the
supermodule towers. With x, y, z proximity sensors mounted and referenced
on the CNC machined chamber bulkheads, we can couple the positions of the
sense wires in the three layers of a projective tower and couple adjacent
towers on the same ruler. A similar scheme of rulers and sensors can be
used on the octant perimeters of the Forward system supermodule rings.
With the laser distance measuring system we can make precision (50 11m)
radial distance measurements between barrel and forward system
references. LED-optics-photodiode relative transverse position sensing can be
used between elements or systems to make precision measurements over a
range of 10mm to 100 11m or less. Small range continuous position sensing or
quad cell diode systems would provide 50 11m resolution or less on a range of 5
mm.
For the octant Intermediate region towers on the two ends of the barrel
toroid we must monitor the position and inclination of each tower wrt a
vertical datum reference plane. We propose that this vertical reference plane
can be defined by a system of wire current catenaries. Using wire position
sensing pickup transformers on the supermodules we can monitor a number
of point on each supermodule wrt the wires.
To link the relative position measurements to the detector coordinate
system, we must monitor and understand the shape changes and motions of
the detector. The precision inclinometers appear to have the sensitivity and
stability to monitor the orientations of our rulers, the magnet in sections, and
supermodules to insure a 100 11m resolution on spatial ruler position. These
will be coupled with a liquid level system to monitor the detector support
system shape.
ACKNOWLEDGEMENfS
We want to thank Hans Jostlein for his valuable comments, review, and
guidance; for the loan of a computer and scanning readout system, and for
the loan of sample inclinometers, etc. Likewise, we wish to thank A.
Lathrop, T. Mantsch, D. Shea for their help in getting us set up. We thank
the Fermilab Physics Department for their help and support, loans of
equipment, provision of facilities for the long term tests, etc. We wish to
thank R. Maly for his work in developing and building our many "widgets".
We wish to thank Mr. Donald Verstrate, Principal of Hutchison Career
Center, Fairbanks, Alaska for his support of our experiment in education: a
co-operative laboratory-school class research project on the source wire
position-transformer pickup scheme.
REFERENCES
1. David P. Eartly and Paul T. Johnson, Studies of prototype SDC muon

alignment concepts and systems, SDC Technical note SDC-92-0l95,
Document Management Office, SSCL, DeSoto, Texas (to be published)
2. Hans Jostlein, Unpublished study, Fermilab Physics Department,
Fermilab, Batavia, TIlinois
3. Measurements being taken by the students of the Electronics and
Technology classes of Hutchison Career Center, Fairbanks, Alaska
4. Hans Jostlein, The electronic alignment system at the Fermilab DO
detector, DO Technical note 999, Fermilab, Batavia, TIlinois
181
MECHANICAL DESIGN AND FINITE ELEMENT ANALYSIS
OF THE SDC CENTRAL CALORIMETER
V. Guarino,1 N. F. Hill,! D. A. Hoecker,2 T. D. Hordubay,2

J. Nasiatka,1 D. W. Scherbarth,2 and R. L. Swensrud2
1High Energy Physics Division

Argonne National Laboratory
9700 South Cass Avenue
Argonne, Illinois 60439
2Science and Technology Center

1310 Beulah Road
Pittsburgh, Pennsylvania 15235
INTRODUCTION
When designing scintillating calorimeters for the study of particle interactions result-
ing from colliding beams, a primary goal is to instrument 100% of the available solid angle.
In pursuit of this goal the challenge for mechanical designers is to minimize the amount of
structural mass and still maintain acceptable engineering standards in the design.
Argonne National Laboratory, High Energy Physics involvement in the design of a
central calorimeter for the SSC started in 1989. Our first proposal was to design a depleted
uranium scintillator calorimeter similar to the ZEUS detector presently installed at the HERA
electron-proton collider in Hamburg, Germany. Argonne was involved at the time in final
assembly of modules for ZEUS that had been designed and constructed at ANL. Due to the
cost of using depleted uranium, lead was chosen as the absorber material.
In collaboration with Westinghouse Science and Technology Center in Pittsburgh,
Pennsylvania we embarked on a program to optimize the use of lead or lead alloys in the
construction of the calorimeter. A cast lead design for the calorimeter evolved from this ef-
fort. Subsequent to this design, further pressure to reduce costs have now dictated a design
which contains lead only in the electromagnetic sections of the calorimeter. The current de-
sign is shown in cross section in Fig. 1.
The finite element analysis we will present here was done using lead for the HAD 1
section of the barrel.
CURRENT DESIGN
The current design calls for construction with the following base design features:

Plenum Press. New York, 1992
Til> Fib(' " P I'OJCC i \'C Co lo rim e l e r'
~! odc ! G
(QlIddrun Cross :ccllOn)
Figure 1. Cross section of one quadrant of the barrel calorimeter.
Barrel Calorimeter (Fig. 2):
Size: '" 9 meters in diameter x 9 meters in length
Weight: 1/2 barrel = 1376 tons

1/32 wedge = 43 tons
Segmentation: 32 wedges made up of 64 symmetric module pairs
Figure 2. Cross section of 1/2 of the barrel calorimeter.
184
End Cap Calorimeter (Fig. 3):
Size: "" 8 meters in diameter x 2.5 meters in length
Weight: each end cap = 680 tons

1/32 wedge = 21.25 tons
Segmentation: 32 wedges
Figure 3. Cross section of the end cap calorimeter.
The barrel calorimeter is constructed using 9 mm (.354 in) steel plates in HADI and
24 mm (.945 in) steel plates in HAD2 joined together by welded spacer plates that are alter-
nated to form pockets for the plastic scintillator tiles. This arrangement is shown in Fig. 4.
The readout of the scintillator tiles is achieved by embedded optical fibers that are
routed to photomultiplier tubes at the outer radius of the detector.
The construction of the end cap calorimeter is similar except that the absorber plates
are arrayed perpendicular to the detector longitudinal axis.
/
"
r.
r.
f.
--- ,
I (
f. /
f. I
=_......-
f. /
r. I
2.&Otr.m '~! L
r. _ 1
r.
Figure 4. HADI/HAD2 internal structure.
185
FINITE ELEMENT ANALYSIS
Differences between Model and Current Design
The finite element analysis was done on the configuration shown however, the
HADl sections of the barrel were constructed using cast lead as the absorber. A recent
change in design has dictated that these segments of the detector now be fabricated with steel
plates. The data for the barrel calorimeter presented here represents the lead design. The
primary changes in the data, that will be realized with this change, are a reduction in overall
weight and an increase in the stiffness of the system. The analysis is presently being re-run
using steel for these segments. The electromagnetic section of both the barrel and the end
cap are still designed for cast lead.
Model Construction
All of the analysis referred to in this paper was done using Cosmos M a product of
the Structural Research Analysis Corporation, Santa Monica, California. Due to size limita-
tions within Cosmos, it was not possible to mesh the 3D model with all of the details of ac-
tual construction. In order to deal with this limitation, it was decided to use individual
wedges that were somewhat less detailed to construct the assembly. The interface forces
and deflections were then calculated and the results will be applied to a fine meshed model
of the individual wedge. The simplified model used the following method for construction.
The structural frame which consists of the inner and outer plates, the end plates, and the
connecting bulkhead membranes were modeled in detail. The lead absorber plates, due to
the complexity of modeling laminate structures, were represented as solids with the stiffness
and density modified to represent the lead plate construction. The composite stiffness used
to model the lead structure was determined by using the ratio of load carrying area to the to-
tal area and using this ratio to modify the stiffness of solid lead. The densities are repre-
sented by using the ratio of total volume to occupied volume.
In connecting the wedge model into the assembly, the interface boundaries between
modules were separated by .030" and connected at discreet points along those boundaries.
The location of these points is shown in Fig. 5. These points were deliberately chosen since
they represent actual boundary load transfer points. The difference between modeled nodes
and actual nodes was in the quantity. The design construction will increase the number of
load transfer points by a factor of five in most cases.
The final assembled quarter barrel is represented in Fig. 6.
HAC-2 8AC~ I£ON

o DE OTES CONNECTING POINTS aOUNDAIZY
C05MOS/M
EM/HAC-I ve,-s \OT' I . 6~
BouNDA/CY p .... ob:HCI\

d .. lc : (\)5 - NOV -~1
Figure 5. Wedge module connecting points.
186
Figure 6. Quarter barrel assembly.
Analysis Results for the Barrel Calorimeter
The EM, HADI and HAD2 sections are layered structures that have been approxi-
mated by solid elements. Since the stiffness of these sections is unknown, certain assump-
tions were made. The HAD2 section is a welded structure with alternating cells, therefore it
was felt that the modulus of steel, 60 x 106 psi, would be a good approximation of the
HAD2 stiffness. The HAD 1 and EM sections however are composed of layers of lead con-
nected together with thin (.020 in.) bulkheads. The stiffness of these structures obviously
is much lower than that of solid lead. In order to establish limits for this situation, it was
reasoned that by setting upper and lower boundary conditions, the extreme limits of the
problem would be established. The upper limit of the stiffness of these sections is the
modulus of solid lead, the lower limit was found by taking the ratio of the load carrying area
to the area of a cell (the bulkhead area) to the total area of a cell and multiplying it by the
modulus of lead. Separate cases of the analysis were then run using these upper and lower
limits of the EM and HACI stiffness. This method will not allow exact values of the con-
necting forces to be calculated, however we will be able to maximize these forces and design
for the maximum condition.
Four different cases were run using a combination of stiffnesses for the EM and
HADI sections and using different size rods to connect the modules together. The cases are
summarized in Table 1. The use of different size rods allowed us to study the effect of size
on the connections. As one will see later this had little effect.
Table 1. Case numbers for Connection and stiffness variations.
Case # EM Stiffness HADI Stiffness Rod. Diameter

(psi) (psi) (inch)
I (Lead Modulus) 2 x 106 2 x 106 1/4
2 (Modified Modulus) 24 x 103 35 x 103 1/4
3 (Lead Modulus) 2 x 106 2 x 106
4 (Modified Modulus) 24 x 103 35 x 103
The connecting forces calculated for the barrel are forces which are distributed along
the length of the boundary plates. For example if a maximum normal connecting force of
400,000 lbs. is found along the EM-HAD! boundary, and this boundary as 29 bearing
points at which the modules are connected, and each point has a cross sectional area of 1
square inch, then each point carries a load of ... 14,000 Ibs. and has a stress of 14,000 psi.
The summation of the forces, both normal to the interface surfaces and the radial and axial
shear loads are represented in Figs. 7 through 9.
When the lead in the HAD 1 section is replaced with steel the forces indicated will be
reduced by the decrease in weight, however the individual connecting forces will change
due to the increased stiffness.
187
oI s.
200
-'1":
-400
~;.++:~,./i"\ .i
\
..................\ \ ·1··i
\J
-600
-800
o 10 20 30 40
MODULE NUMBER
[RON HAC 1 ~ EM
Figure 7. Forces normal to the surface as a function of module position.
-200
-250
o 10 20 30 40
MODULE NUMBER
[RON
-+-- HAC 1 ~ EM
Figure 8. Radial shear force as a function of module position.
F RC
60
40
20
-20
o 10 20 30 40
MODULE NUMBER
[RON BAC1 ~ EM
Figure 9. Beam direction shear forces as a function of module position.
188
Endcap Calorimeter Model
The endcap was modeled in a manner similar to that of the barrel. The EM front
plate, the EM-HAD 1 boundary plate, HAD1-HAD2 boundary plate and the back iron struc-
ture were modeled first using plate elements with the appropriate thicknesses to fonn the
basic frame as shown in Fig. 10. The individual modules were connected to each other at
22 points, 8 along the EM-HADI boundary, 6 along the HADl-HAD2 boundary and 8
along the back iron. The EM section once again is composed of layers of lead plates sepa-
rated by thin bulkheads. This presents the same problem of modeling as it did in the barrel,
how to model the stiffness of these structures appropriately. Instead of modeling the EM
and HADI structures using solid elements and then varying the stiffnesses of this solid to
approximate the stiffness of the structures, individual plates of lead were used. The EM
section has 12 lead plates 10.5 mm thick separated by bulkheads, these were approximated
by 4 plates which were 1.24" thick and separated by 7 bulkheads. This method approxi-
mates the stiffness of the EM structure but does not go into so much detail that the problem
becomes to large to run. Similarly the HADI section has 28 steel plates 20.5 mm thick
which were approximated by 5 plates which were 4.5" thick and separated by 9 bulkheads.
The stiffness of lead, 2 x 166 psi, was used for all of these plates. The HAD2 section was
approximated by solid elements and the stiffness of steel, 30 x 1()6 psi, was used since
HAD2 is a welded structure with very few cells therefore it was felt that it would behave like
a solid structure. Figure 11 shows the module wedge and the position of the connecting
points. This module was then copied and rotated to fonn a half endcap as shown in Fig.
12. Advantage was taken of symmetry so that only half of the endcap was modeled which
reduced the size of the problem and computer time considerably. The bottom 4 modules
were fully supported along the entire length of the outside diameter of the structural iron.
Figure 10. Wedge module frame Figure 11. Wedge connecting Figure 12. One half end
rough model. points. cap assembly.
Analysis Results for the End Cap Calorimeter
The results of this analysis produced similar results to those obtained for the barrel.
Three representative plots of the results are shown as Figs. 13, 14, 15. A summary of the
maximum forces expected at any interface boundary are represented in Table 2.
Table 2. Interface boundary maximum forces.
Boundary Max. Normal Force Max. Radial Shear Max. Z Dir. Shear
EM/HAD 1 + 0 (tension) - 3,000 (inward) 2,000
17,000 Ibs. (compr) + 7,000 (outward) + 2,000
HADI/HAD2 + 0 - 10,000 + 2,000

30,000 + 12,000 + 2,000
Outer Row + 2,000 - 15,000 + 12,000

35 1000 + 15 1000 + 2 1000
189
-20
o 5 10 15 20
Module Boundary
--t- Point 16 -*--- Point 17
-13-
Point 18
Point 15
--+-- Point 20
--t:>-::-··
Point 21
-H-
Point 22
Point 19
Figure 13. Outer iron normal forces.
10
-10
-20
-30
-40
o 5 10 15 20
Moduie Boundary
--t- Point 16 -';iE-- -13-
Point 15 Point 17 Point 18
~:r- -h-- --H-
Point 19 Point 20 Point 21 Point 22
Figure 14. Outer iron radial shear forces.
-15
o 5 10 15 20
Module Boundary
--t- -~- -13-
-<>- ~lo- --<!-
Figure 15. Outer iron Z direction forces.
ACKNOWLEDGEMENT. Work supported by the U.S. Department of Energy, Division of High

Energy Physics, Contract W-31-109-ENG-38.
190
DESIGN AND CONSTRUCfI()N OF THE SSCL
MAGNET TEST LABORATORY CRYOGENIC SYSTEMS
Michael A. Freeman and Theodore A. Kobel

Process Systems International, Inc.
20 Walkup Drive
Westborough, MA 01581-5003
INTRODUCTION
The intent of this document is to provide a brief summary of the execution, by
Process Systems International, Inc. (PSI), of the Design and Construction of the SSCL
Magnet Test Laboratory Cryogenic Systems. This $30 million project requires the
expenditure of over 200,000 manhours and the procurement of $17 million in
materials within a two year period.
Overview of the SSC MTL Cryogenic Systems
SSC magnets will be performance tested at the Ma€inet Test Laboratory
(MTL) and the Accelerator System Strin~ Test (ASST) faCIlity under conditions
simulating the environment of the SSC mam ring. The cryogenic system consists of
test stands (five for MTL, one for ASST) and the associated eqUIpment including
cryogenic storage, purification, thermal conditioning, and helium refrigeration
necessary to support the test program.
The nominal capacity requirements of the systems are given in Figure 1.
PERFORMANCE AND GUARANTEE TABLE
lOS\;
DE8IGN
100\
.Pr:CU'IC"TION
,,' '" '" ," LN,
LIO·
LN,
V/O
LN,
REFR.
W/O
LIO·
V/O
LN, LN,
=REFRIGERATION AT ... 5~ [N} 1600 700 '00 3530 1100

LN,
3300
LIQUEFACTION AT ".5)( [9/-J

" 20 16 J7 11 12.4
COMPRESSED GAS
TO PRESSURE
(9/-)
(8u]
Jl.S
11.4
'0
17 ...
"
15.1
10.5
11.1 9.0
'0
15.9
'0
17 ...
'0
17 ...
'0
16.6
'0
lB.1
..-------- ..--------
EXPECTED CUARANTEE EXPECTED GUARANTEE EXPECTED GUARANTEE
55.
..
ELlCTRIC POWER (kW) 1198 1161 1198 1045 1090 590 443 1108.2 1134.3 1077.8 1130.9 1019.9
LNl A.T 1.1 e.\R (9/-) 64.1 63.2 50.1 26.9 16.1 15.3 104
LN2 at 2 BAA G 70 '0
eOKMON fEATURES: COOLING WATER INLET 5. 305K

PRESSURE AT COLD END OF COLDBOX I 1.12 (EXCEPT PURE REfRIGERATION CASES)
eLeCTRICAL POWER IS MOTOR LEAD POWER, OOES NOT INCUIOE VACUUM, INSTRUMENT AIR AND CONTROLS
REI'V.RKS (1): J5'

or CAPACITY IS ACHIEVED WITH ONE COHPRESSOR SET ONLY.
OUTPUT COULD BE RAISED ASOVE 40\ WITH CHANCE OF TURBINE INLET tJOZZLES
FIGURE 1

PROJECT EXECUTION
On November 19, 1990, Process Systems International, Inc. (PSI), then Koch
Process Systems, Inc. (KPS), received a hmited authorization to proceed to execute
the Design and Construction of the SSC Magnet Test Laboratory Cryogenic Systems.
The scope of work included engineering, design, procurement, manufacturing,
installation, and operation of three (3) complete cryogenic facilities designated as
follows:
Accelerator System String Test (ASST) Facility
Magnet Test Laboratory (MTL) Facility .
E-1 Site (N15B) Facility
The baseline schedule requirements, and specifically the need to complete the
ASST facility as soon as pOSSIble, coupled wIth the significant manpower and
procurement activity requirements, presented a challenge not previously encountered
at PSI on cryogenic system projects. Additionally, the equipment was destined for
facilities to be supplied by SSCL under separate contract to others, but those facilities
would be designed around the PSI supplied equipment; this further magnified the
need for initial equipment design in a timely manner.
PSI's approach to this project has been to:
Develop an arrangement with certain other companies in which each
company, while providing a specific scope of supply, would complement each other's
capabilities, thus producing a truly synergistic project team.
Develop a Project Mana~ement and Execution Team within PSI consisting of
representatives from all major dIsciplines reporting directly to the Project Director.
Perform a thorough evaluation and careful selection of vendors to supply
major purchased components to meet specification, quality, and schedule
requirements.
Increase both engineering/design and manufacturing capacity through
technology, additional manpower, and extended hours of operation to meet
equipment delivery schedule requirements.
Perform a thorough evaluation and careful selection of an installation
contractor to support PSI and SSC in meeting project specification, quality, and
schedule requirements.
In conjunction with SSC, develop an operations team to ensure quick and
complete preoperative checkout, debuggmg, and testing of cryogenic facilities.
The Project Team
Early in the proposal stage of the project, PSI carefully evaluated many other
organizations, including competitors, to determine the single best combination of
talent available within the cryogenic community capable of delivering both the
technological requirements of the project as well as supporting the schedule
requirements of the overall program.
Consisting of PSI, Sulzer Brothers Limited, CCI, and CTS, the Project Team
could claim the following:
To have designed and/or supplied equipment which has liquefied over 90% of
all helium since 1908.
To have supplied equipment providing over 85% of the
refrigeration/liquefaction capacity currently owned by U.S. laboratories.
192
To have supplied over 80% of all helium systems supplied to the free world in
the last 10 years.
To have designed and/or built equipment for over 90% of the U.S.
commercial helium liquefaction capacity.
The Project Management Team
It was obvious from the outset that in order to meet the a~ressive schedule
established for delivery of the initial cryogenic systems, a well orgarnzed, established,
and dedicated project team would be essential.
Initially, it was necessary to determine the degree of autonomy the project
organization would maintain within the company. Too much autonomy might place
unreasonable requirements upon the project management team, while too little could
lessen the effectlveness and efficiency of the team. Also, it was apparent that the
project management team required a significant level of authority due to the sheer
volume of manpower and dollars to be expended.
The resulting project team or~anization as seen in Figure 2 was developed to
satisfy the above concerns. By haVlng a corporate officer as the ultimate project
manager, the authority limits required were satisfied. By having a "partially
autonomous" project oq~anization, the project management team would concentrate
on the areas and disciplmes most affectmg an efficient execution of the project, while
not needing to concern the members with support functions such as accounting,
facilities, and computer operations.
SSCL PROJECf
The PSI Project Team Organization
FIGURE 2
193
Supplier Selection
To a great de~ee, supplier evaluation and selection is a requirement of the
proposal process, dUrIng which budgetary cost is established for all major purchased
components. In this case, the evaluation process continued throughout the proposal
stage and well after contract award for a few very critical components. Of utmost
concern were the expected reliability of the components and compliance with SSCL
specifications.
The finalized list of selected major suppliers is presented herein:
SSCL Project
Msijor Supplier Selections
Component Supplier Selected
Turboexpanders Sulzer (1)

LHePumps CCI (1)
GHe Compressors Sullair
Control Systems EMX/TI
Cryogenic Valves and Filters CPC
LHeDewars CTS(l) /Cryenco
LN2 Dewars PEl
GHe Storage Trinity Industries
Vacuum Systems Leybold/Kinney
VJ Piping MVE
(1) Denotes Project Team Member
A great deal of coordination was required with all major suppliers as PSI
chose to select and provide critical valving and instrumentation tn order to maximize
reliability and interchangeability of components and minimize spare part inventory
requirements.
PSI Capacity
Although the manpower required to execute the project did not ever approach
the total manpower available within PSI, the accelerated buildup of manpower
required on the project necessitated serious consideration of how this would be
achieved without significantly impacting other ongoing projects within PSI.
The resulting plan consisted of three parts:
Technology Improvements. A significant investment was made in additional
PC's, CAD/CAM stations, software, modems, fax, and reproduction equipment, all
designated for s1?ecific use on the project within engineering and design/drafting. In
the manufactUrIng area, additional tooling was procured to increase in-house
fabrication, pipe bending, welding, and helium leak testin~ capabilities. A clean room
was built to Improve aluminum welding quality. An additIonal and larger helium leak
testing area was established and equipped. Additional mass spectrometers and other
helium leak testing equipment was procured.
Additional Manpower. While all project management, supervisory, and lead
positions were assigned to permanent PSI personnel, contract engineers, most of
whom had been utilized by PSI in the past, were employed to fill less critical
engineering and design POSItiOns. In the manufacturing area, contract employees
were utilized in the same manner. In this way, the peak manpower requirements
194
were met in the best way possible, minimizing the impact upon the balance of the PSI
organization and the work going on therein.
Extended Hours. During the most critical engineering and design period,
during which 1) major components were specified and placed on order, and 2) the site
layouts and facilities requirements established, all engineering and design personnel
were placed on mandatory 50 hour work weeks. PSI's experience with required
extended hours is that a 50 hour work week is achievable at near 100% efficiency
indefinitely, while productivity drops off with greater than 50 hours for any period of
time beyond three to four weeks. With a mandatory 50 hour work week, PSI was able
to increase capacity of the engineering and design team by 20% with no significant
loss of efficiency.
In the manufacturing area, PSI had routinely worked a 50 hour work week for
several years. Extendin~ to a 60 hour work week was not considered to be favorable
or a solution. The deciSIOn was to expand to an overlapping two-shift operation. The
requirement to overlap the two shifts by one-half hour was felt necessary due to the
complexity of the product and the manufacturing process involved. The staffing on
the second shift was to be a single crew of 10-12 individuals assi~ed only to SSC
project work. The result proved dramatically successful as eVIdenced by early
deliveries of all cold boxes and on schedule deliveries of all other PSI manufactured
subsystems and components.
Installation
The installation of the cryogenic systems presented the most significant
challenge to PSI on this project. Because of the aggressive schedule, the installation
duration was to be severely limited. Furthermore, as the planned installation start
date of October 1, 1991 approached, it became increasingly evident that the
installation of the ASST and N15B cryogenic systems would be executed in parallel
with construction of the buildings which would house the systems, and construction of
the site infrastructure.
Recognizing that this situation would further complicate an already difficult
task, the planning and execution of the installation became an immensely critical
activity. Initially, key construction management personnel were designated within
PSI and assigned to the project team. Secondly, numerous local mechanical
contractors were solicited, evaluated, and eventually a short list of contractors were
invited to submit proposals to execute the work. The evaluation process continued
throughout the proposal process, and resulted in PSI's selection of Dallas-based
Trinity Contractors, Inc. (TCI) as the prime contractor to execute installation of the
PSI supplied cryogenic systems under the direction of PSI construction and project
management personnel.
In order to continue to attempt to compress the installation schedules for the
initial ASST cryo~enic system, TCI and PSI have worked very closely with SSCL
personnel, have Increased on-site crew sizes, and have implemented extended
working hours in an effort to minimize the loss of productivity incurred as a result of
the difficult site conditions.
Additionally, under SSCL direction, daily and weekly communication has been
established between PSI, TCI, and all other site contractors in order to minimize
interferences, joint occupancy of buildings and areas, and other potential
inefficiencies during the installation process.
On future projects of this nature, the single most significant change that should
be considered to allow a more efficient installation would be to either have the
system supplier provide the building and site infrastructure in the absence of
interference by other contractors, or have that work completed well in advance of the
equipment installation.
195
Operations Team
The operations se~ment of a cryogenic systems project tends to be the most
exciting and the most difficult to predict.
PSI's approach to encountering and executing the operations portion of this
project has been to:
Designate the operations management team from within PSI and assign to the
project team.
Develop a comprehensive operations plan encompassing preoperative
checkout, initial startup, debugging, and acceptance testing.
Identify key subvendor involvement required and plan accordingly.
Develop in conjunction with SSCL the manpower requirements to satisfy the
operational and maintenance requirements.
Operations of the ASST cryogenic system are projected to begin in late April
1992 and conclude approximately three (3) months later.
EQUIPMENT STATUS
SSCL • ASST/MTLJNlSB
SYS. SUB COMPONENT DEUVERY DELIVERY DEUVERY

NO. SYS# DESCRIPTION ASST NlSB MTL
1'ower 1J..t._\westingno~, IOZ-Dec-91 IUZ-lJec-91 !Tl-1Jec-Yl
Dist. Control System (EMX/TI) Storage Storage Storage
10 16 Helium Gas Storge (Trinity) 27-Nov-91 Storage Storage
10 57 Helium Gas Storge (Trinity) 27-Nov-91 Storage Storage
10 10 Main Cold Box 25-Nov-91 12-Jan-92 20-Mar-92
10 10 Main Cold Box Turbine Pods 25-Nov-91 12-Jan-92 2O-Mar-92
10 11 First Stage Compressor-l (Sullair) 13-Jan-92 Storage Storage
10 12 First Stage Compressor-2 (Sullair) 13-Jan-92 Storage Storage
10 13 Second Stage Compressor-l(Sullair) 13-Jan-92 Storage Storage
10 14 Second Stage Compressor-2(Sullair) 13-Jan-92 Storage Storage
10 15 Compressor Oil Removal Skid 24-Jan-92 21-Jan-92 Storage
10 17 Main Cold Box Regen Skid 03-Dec-91 03-Dec-91 Storage
10 18 Syst 10 Vacuum Syst (Leybold) ll-Nov-91 11-Nov-91 Storage
30 30 Liquid Helium Pump Box (CCI) N/A N/A 04-Jul-92
30 31 Liquid Helium Pump Box (CCI) 06-Jon-92 N/A N/A
40 41-45 Test Stand Dist. System N/A N/A 22-Jun-92
40 46 Test Stand Dist Box 06-Mar-92 N/A N/A
40 47 TSO Box Vacuum Syst (Leybold) ll-Nov-91 N/A Storage
SO 50 CCWP Cold Box 09-Dec-91 28-Feb-92 2O-Mar-92
SO 51 CCWP Cold Box N/A N/A 2O-Mar-92
SO 52 CCWP Regen Skid 19-Nov-91 19-Nov-91 Storage
SO 53 CCWP Dehydration Skid 15-Nov-91 15-Nov-91 Storage
SO 54 CCWP Compressor Skid (Sullair) 13-Jan-92 Storage Storage
SO 55 CCWP Compressor Skid (Sullair) N/A N/A Storage
SO 56 CCWP Oil Removal Skid 24-Jan-92 24-Jan-92 Storage
60 60 Cold Box Assembly N/A N/A OS-Apr-92
60 61 Vac-l Vacuum Pump Assembly N/A N/A OS-Apr-92
60 62 Vac-2 Vacuum Pump Assembly N/A N/A OS-Apr-92
60 63 Oil Removal N/A N/A Storage
60 64 Cold Box Vacuum System-Portable N/A N/A Storage
70 70 Nitrogen System (Process Eng.) 20-Nov-91 04-Feb-92 Storage
70 71 LN2 Vaporizing Skid 23-Nov-91 13-Jan-92 Storage
80 81 Inst Air Compressor-l (Hope) ll-Nov-91 11-Nov-91 Storage
80 82 Inst Air Compressor-2 (Hope) ll-Nov-91 11-Nov-91 Storage
80 83 Dryer ll-Nov-91 11-Nov-91 Storage
80 84 Receiver Tank 26-Nov-91 26-Nov-91 Storage
90 Vacuum Jacketed Piping 3O-Jan-92 28-Feb-92 2O-Mar-92
90 Structural Steel 0l-Nov-91 12-Dec-91 Future
90 Interconnecting Piping In Progress In Progress Future
90 91 Oil ProcesSing System 14-Jan-92 14-Jan-92 Storage
90 92 Gas Analyzer 17-Feb-92 17-Feb-92 Storage
FIGURE 3
196
rJ:'I'Ir omr AP1r MIV" ':1I1IiI" nr -mr ~ "'1Il1" tmV" '"'1ft'" "mt
ASST.
'_T"''''~
'"""".,-
t~CePTAHcE
NI5B.
r... " .TT_
mns
..
t CcEPTANCE
MTL.
illiii'~ml
.. ..
aIllBe
.. ..
Ai:cePTA~
___ ~Irrc
c:=.::=J Acthny erIE..., DIIt.. SUPERCONDUCTING SUPER COLLIDER LAB Sheet I of I

__ critlC •• ACtlvity
F~"'''''''''
CURRENT FORECAST SCHEDULES
pr~l~et Start: ISEP90
ASST/MTL/N15B SYSTEMS
Pri ••vera Systeu, tnc. 1.... 1.. ' Pro eet Finish: 3JAN94 m· ~:~:: ~:~
CD
-.J FIGURE 4
PROJECT STATUS
A detailed summation of equipment status can be seen in Figure 3. As noted
therein, all ASST equipment has been received at site. All N15B and most MTL
equipment either has been received at site or is in storage awaiting site access and/or
space availability. Installation of the ASST and N15B systems is in progress and
ap,proximately 75% and 15% complete, respectively. Installation continues to be a
dIfficult task due to limited site access and building availability, and difficult working
conditions due to weather and joint occupancy conditions.
Control system software, a significant sco{>e of supply within this project, is
nearing completion on a first draft basis, With 15-20% debuwng complete.
Obviously, software development will continue through the operatlons phase and
beyond by both PSI and SSCL operations personnel.
PROJECT FORECAST
The presently projected schedule to completion is presented in Figure 4. As
depicted therein, the ASST cryogenic system will complete operations and testing in
early July.
The N15B system should complete operations and testing within two to three
months after the ASST cryogenic system. The MTL cryogenic system, pending
building availability on or about July 1, 1992, will complete installation in late 1992,
with operations and testing to complete in the 1st quarter of 1993.
198
COMMISSIONING AND ACCEPTANCE TESTS OF THE HIGH
EFFICIENT 6KW HELIUM REFRIGERATOR FOR THE LEP 200
PROJECT AT CERN
B. ChromecOOOO, R. ClercOOOO, W.K. ErdtOO, D. GiisewellOO,K. KurtcuogluOOOO

K. LohleinOO*, N.O. SolheimOO, U. WagnerOO", G. WinklerOO and B. ZieglerOOOO
* CERN, Div. AT, CH-1211 Geneva/ Switzerland

** Sulzer-Chemtech, Cryogenics, CH-8401 Winterthur / Switzerland
ABSTRACT
CERN has ordered helium refrigerator / liquefier plants for the LEP 200 project in 1990
with an equivalent refrigeration capacity of 6 kW at 4.5 K. The cold box of one of these
plants is installed in the LEP tunnel and is therefore limited regarding its constructional size.
The helium refrigeration plant delivered by Sulzer combines an extremely compact
construction with a remarkably good cycle efficiency. This high efficiency with a predicted
power factor of 219 W /W was achieved beside others by arranging 3 expansion turbines in
the temperature region below 20 K and one of which is expanding to subcooled liquid. This
paper describes the experience during commissioning and the results of the acceptance test of
this plant.
INTRODUCTION
The 26.7 km long underground electron positron collider LEP of the European
Laboratory for Physics of Elementary Particles (CERN) near Geneva, astride the French-
Swiss border started operation in summer 1989 1. Room temperature copper cavities are used
for the radio frequency acceleration of the particle beams up to 55 Ge V. During the coming
years the particle energy will be increased to about 90 GeV by progressively installing

superconducting cavities for acceleration. This upgrade project is known as LEP-200. To
cool a flrst string of superconducting cavities a rrrst helium plant with a nominal capacity of
6 kW at 4.5 K is presently being commissioned at LEP Interaction Point 2.
PLANT DESCRIPTION
The plant basically consists of a compressor group, a cold box and a control system. The
compressor system may be considered as the active part of the plant, providing a massflow of
helium at a preset supply and suction pressure.
The cold box itself behaves like a passive element, transfering pressure difference
available into refrigeration capacity. Its internal controls make sure, that this transfer occurs at
optimum efflciency. The compressors are located at ground level and to overcome static
pressure effects, the cold box is installed in the LEP tunnel 40 meters below the surface.
The specifled cooling capacity of the plant together with the power input expected are
given in Table 1.
Table 1. Expected cooling capacity and power input of CERN's 6 kW plant for LEP Point 2.
refrigeration capacity at 4.5 K 5000 W

liquefaction capacity at 1.3 bar 14.5 K 8 gls
refrigeration capacity at < 80 K 3000 W
total power input at terminals 1354 kW
Refrigeration process
Helium gas is compressed by a set of three screw compressors, two of them working as
boosters in parallel at a suction pressure of 1.01 bar and an outlet pressure of roughly 4 bar
and one operated as second stage machine with a discharge pressure of 20 bar (Fig. 1). Each
compression stage is followed by a bulk oil separator and a gas cooler. A flne oil separation,
consisting of a two stage coalescing fllter and a charcoal adsorber, reduces oil contamination
to 10 ppb (mass).
In the cold box refrigeration is achieved by a total of 7 turbines and 13 heat exchangers as
shown in a simplifled flowsheet in Fig. 2 and the corresponding T-s diagram in Fig. 3. A
first loop, consisting of 4 turbines arranged in series, expanding helium from 19 bar to 1.3
bar provides precooling of the IT-stream to roughly 20 K. A part of the IT-stream is extracted
at 50 K, used to provide shield cooling, returned to the plant, cooled back to 50 K and fed to
the main JT-stream again on a slightly lower pressure level. Below 20 K the
IT-stream is fed to two parallel turbines 2, operated at inlet temperatures of 16 K and 9 K,
respectively. After passing through heatexchanger 13 and turbine 7, it is available as roughly
saturated liquid at 1.3 bar.
200
~
II
--------- II
I
I
I I
L____ _________ __
, I
,
~ ~_~_
------------------~------------.-.
Figure 1. Flowsheet of the compressor system
HX 1
T1 tiX2
HXl
T2 {J HX~
T4 Hxe
HX9
T5 HXIQ
Hxn
T6 HX12
HX13
T7
Figure 2. Flowsheet of the cold box
201
t
Ii 1.2-
II.
Tl
. .....
I,. 671.1
'
"'~'4>'
7. V"tl...
.. It. "1P T2
." 75.51
.. eg . .)
....
.11.8
..
112.2
Figure 3. T-s diagram
202
Compressors
The three compressors used are of the lubricated screw compressor type, produced by
Stal, Norrkoping, Sweden. Table 2 gives an overview on the dimensions of the individual
machines.
All machines are directly coupled to the electric motors. They feature hydraulically
operated slide valve, rotor balancing piston and gliding ring shaft seal. Their rotors are axially
supported by angular contact ball bearings and radially by pressure lubricated slide bearings.
The booster compressor S93 is equipped with a separate external oil pump, whereas the S75
and the S73 machines have integrated oil pumps directly coupled to the rotor.
The two booster compressors share one common bulk oil separator, oil cooler and gas
cooler, but have individual bypass valves for startup.
The whole system was preassembled on a total of five skids.
Table 2. Dimensions of the compressors for CERN's 6kW plant for LEP Point 2.
Application booster booster 2nd stage

type of machine S93 S75 S73
rotor diameter [mm] 333 258 258
rotor length [mm] 583 657 425
swept volume at 2950 RPM fm3/hl 5750 3900 2500
Cold box
The cold box design is very much dominated by space limitations both in the access shaft
and in the LEP tunnel itself. The maximum dimensions are:
final position: maximum length 12.0 m

maximum width 3.0 m
maximum height 3.8 m
vertical shaft: length 6.0 m

width 1.85 m
max. acceptable weight 20 tons
horizontal transfer tunnel: height 2.6 m
On a very early stage it was agreed to build a cold box basically as a tube, to be lowered
to the tunnel through the shaft in vertical position and to be installed horizontally.
The very compact design required special attention during fabrication. The fabrication
sequence and the corresponding quality checks, such as radiography of welded joints and
203
Figure 4. CAD cut away vie on the cold box
leak testing had to be carefully planned, since access to the internal parts became more and
more limited during cold box assembly. Fig. 4 shows a cut away view on the cold box.
Control system
The 6 kW helium refrigerator will be run by a team of operators together with several
plants and systems built by other suppliers. For reasons of simplified operation and
maintenance all these systems will be equipped by CERN with the same microprocessor
based industrial process control system. The established CERN standard is the Master system
supplied by ABB. It provides all local control functions required and - via one of CERN's
data transmission networks - remote operation from the central cryogenics control room up to
10 km away from the individual plants.
A so called Master Piece, and independent intelligent local unit, is provided both for the
compressor system and the cold box, each with its own battery backed 24 VDC supply
system.
Separate units, called Master View, are placed in the control room for process
visualisation and operator / plant interface. The operation of the cryoplant will, in general, not
be affected by failures of the communication network or the display system.
To limit potential damages of machinery, in particular compressors and turbines are
equipped with an additional basic protection using hard wired interlock.
It is intended to operate the plant fully automatic and without intervention or even
presence of operators.
204
COMMISSIONING
Compressors
When compressors were first started in December 1991, excessive vibrations of the
piping of the oil supply system was recorded. These effects were dramatically reduced by
adjustment of the pipe routing and adequate supports. In the meantime the compressor system
has successfully completed a 200 hours test run. Massflow and efficiency measurements of
the booster compressors however were so far disappointing whereas the second stage
machine performed well. Table 3 compares design values and measured values.
Table 3. Design and operating data of the compressors for CERN's 6 kW plant for
LEPPoint2
boosters second stalZe total

desim measured desim measured desim measured
suction temperature [K] 305 292.5 300 294.2 305 292.5

suction pressure [bar] 1.01 1.01 4.45 4.1 1.01 1.01
discharge pressure [bar] 4.58 4.2 20.6 20.3 20.6 20.3
massflow [gIs] 393 377 393 377 393 377
shaft power input [kW] 620 616 706 675 1326 1291
isothennal efficiency r-1 0.549 0.530 0.532 0.546 0.566 0.533
Based on a series of measurements, Stal is presently investigating why the performance

of the booster compressors does not yet reach the design values while the second stage
machine is exceeding the predicted performance.
Cold box
The cold box logic is presently beeing tested with helium massflow applied but
without turbines being installed. Cooldwon is planned for next week.
Control system
The decision to use CERN's standard control system caused considerable problems
when defining the scope of supply and the responsibilities during erection and
commissioning. In addition the enormous work load at CERN with two 6 kW plants, built by
two different suppliers, presently being commissioned, and four 12 kW plants being under
construction forced CERN to subcontract the programming of the control system which lead
205
to considerable delays. Keeping however in mind, that the plant will be operated over a
period of say 10 years, it becomes obvious that the operator's tools have to be standardized
both on the hardware and software level even if this causes some conflicts during very limited
time. CERN and Sulzer as the supplier are both convinced that the decision taken will
positively affect the operator's work and the availability and maintainability of the plant.
REFERENCES
1 W. Erdt et aI, The refrigeration system for the LEP upgrade, "Cryogenics",
30: 9 (1990), p. 136
2 A. Klindig, Method and apparatus for liquefying a low boiling gas, US Patent
Number: 4.606.744
206
84 K NITROGEN SYSTEM FOR THE SSC
M. McAshan, M. Thirumaleshwar', S. Abramovich, V. Ganni, and A. Scheidemantle
Superconducting Super Collider Laboratoryt

ABSTRACT
The nitrogen system for the Superconducting Super Collider (SSC) is designed to
provide the 84 K (nominal) shield refrigeration for the collider rings. Liquid nitrogen is
supplied to the collider tunnel from one, two, or more locations on the surface through the
service shafts and is distributed along the 87 km of both rings by the 84 K shield lines.
Additional design requirements for the nitrogen distribution system include precooling fluid
supply to the helium plants, supplying makeup liquid nitrogen to the reservoirs located at
the entrance of the main shafts, and providing an efficient cool down means for the cold
mass from 300 K to 90 K. The operational modes and possible emergency and maintenance
conditions of the collider are taken into account for the nitrogen system design. The status
of our work, including design considerations that address thermal aspects (heat load,
recooling scheme, etc.) and hydraulic aspects (pressures, elevations, distances, etc.) of the
nitrogen system will be discussed.
INTRODUCTION
The SSC consists of two parallel magnet rings, each 87,120 m in circumference,
constructed in a tunnel 25-74 m under ground. The rings are operated in a controlled low
temperature of about 4.2 K, in order to maintain the magnet windings in the
superconducting state. The magnet cryostat! is designed with a high-quality super
insulation-i.e, a high vacuum chamber, multilayer insulation (MU), and thermal shields at
84 K and 20 K nominal temperatures. Thermal radiation and the conduction heat load
through the supports are intercepted and absorbed by the 84 K shield. Liquid nitrogen
provides the refrigeration for these loads .
• Guest collaborator from Center for Advanced Technology, Indore, India

t Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

Plenum Press. New York, 1992 207
The nitrogen system consists of a nitrogen generation plant (one, two, or more air
separation plants); ten nitrogen dewars for the collider and two for the High Energy Booster
(HEB) located on the ground at the main shaft entrances; and associated piping, heat
exchangers, controls, etc. The conceptual design for various flow schemes and the
engineering considerations are presented in this paper.
THE SYSTEM GEOMETRY

The Collider Rings
The main tunnel of the collider is 25-74 m deep. Its axis is on a plane with an
inclination of approximately 0.2°. The highest point in the tunnel is in the vicinity of NI5
(located north of the west cluster), and the lowest point is close to SIS (south of the east
cluster). The shape of the tunnel and the relative locations of the shafts are schematically
shown in Figure 1.
The circular arcs of the rings contain almost continuous strings of superconducting
magnets. The arcs are segmented into sections connected to each other by U-tubes so that
the various parts of the system,can be isolated for maintenance purposes. For practical
reasons, the length of the magnet strings between adjacent shafts varies, ranging from
8.2 km to 9.3 km, and the length of the sections varies from 360 m to 1350 m. These
sections are connected to each other through spool pieces in which the instrumentation,
controls, and U-tubes are installed.
In the east and west parts of the collider are the "clusters" that contain different pieces
of warm beam equipment. In all these regions the flows of the cryogens are bypassed
through specific transfer lines. It may be necessary in some places to route the transfer lines
through different tunnels in order to make room for the warm equipment. The detailed
design of the east and west clusters will be defined elsewhere.
TheHEBRing
The HEB is 10,800 m in circumference, built in a tilted plane parallel to the collider,
14 m above the collider plane. The main bending dipoles, the quadrupoles, and the
o =-,:::.:t~,:~"-"'--
o :::.~=,...-, ... ~
Figure 1. Geometry of the collider rings and the location of refrigeration plants.
208
correctors of the HEB are superconducting magnets with a design that is specific to the
HEB.
The Cryostat
Two nitrogen tubes, 63.5 mm 0.0. (57.15 mm 1.0.) each, are anchored to the 84 K
shield. These two lines can be used for liquid flow, two-phase flow, or nitrogen vapor flow.
In normal operation mode, one tube is used as the main distribution conduit for the liquid
nitrogen. The other tube serves as the vapor conduit for the cooling system and supplies the
vapor to the helium plants, where it is used in the helium refrigeration process and is vented
to the atmosphere at 300 K.
The cry os tats in the string are connected to each other through flexible bellows
connections. In a nominal section of 1080 m there are 84 bellows connectors and four 90°
bends per line.
The Shafts
The depths of the different shafts vary from 25 m to 74 m. The transfer lines
connecting the surface cryogenic equipment to the tunnel will be installed in the main
shafts, designated N 15, N25, SIS, etc.
THE SYSTEM REQUIREMENTS
Figure 2 shows the 84 K heat loads and the mass flow of liquid and vapor nitrogen.
The heat loads of the east and west clusters are assumed to be 50% higher than those of a
regular arc sector.
Nitrogen System Requirements
1. Provide the 84 K shield refrigeration at a nominal temperature of 84 K. The shield

temperature should not exceed 84.0 K.
2. Heat transport medium for the 84 K shields.
3. Distribution system for the nitrogen in the tunnel.
4. Make up liquid nitrogen to the reservoirs at the above-ground locations.
5. Refrigeration for the cool down process of the cold mass from 300 K to 90 K.
6. Precooling for the helium system.
Design Targets for the N2 System
1. Lowest temperature levels of the 84 K shield.

2. Lowest possible operating pressure.
3. Smallest temperature variations.
4. Maximum vapor recovery.
5. Simplest control and instrumentation requirements.
6. Low first cost and operation cost.
7. Lowest possible inventory.
8. Safe and reliable operation.
209
300g/s
300K
. . (M g/s) --..(M - 176 g/s)
Figure 2. LN210ad per sector,
OPERATIONAL REQUIREMENTS
The 84 K refrigeration to the cryostats and the supply of the precooling nitrogen to the
helium plants should be continuous and uninterrupted for all possible system
configurations-e.g" operation of one or both rings or any combination of disconnected
strings, one or more air separation plants not operating, etc. The flow rates for different
configurations are given in Figure 3,
The Maximal Pressures
The nitrogen tubes in the cryostats are made of aluminum and are designed for a
maximum pressure of 2.0 MPa. A maximum design pressure in the lines for normal
operation will be 1,0 MPa. During cooldown, fill-up, and warm-up, the pressure may rise to
1.5 MPa.
Inventory and Supply
The volume of one nitrogen tube in the cryostat is 2.565 m 3 per km. The inventory for
different parts of the collider follows:
liquid (kg) va~or (kg)
One sector (4 strings) 35,424 216
One ring 178,596 1,089
Total for the collider 357,192 2,178
Total for the HEB 38,492
12 LN2 dewars 650,400
The total inventory of nitrogen for the SSC is 1,048,200 kg, The total nitrogen
consumption for the SSC is 4200 gIs, Assuming 30% Carnot efficiency, the total power
required is 11 MW (15,000 hp),
210
Ten stand-by reservoirs of nitrogen for the collider and two for the HEB will be
located on the surface near the helium plants.
The LN 2 consumption forecast is shown in Figure 4.
The Controlled Parameters
The main parameters to be controlled in the nitrogen system are the temperature and
the mass flows.
The required range of temperatures is obtained by recooling. In a few locations
(between N20 and N25, and between S20 and S25) the tunnel is horizontal; in these places,
compact heat exchangers have to be used for recooling. In other parts of the ring, where the
tunnel is inclined, the vapor return line may be used as a continuous recooler. The pressure
in the vapor return lines and the number of recoolers are the main factors affecting the
temperature and the temperature variations in the cryostat.
If compact heat exchangers are used as recoolers, the boiling temperature depends on
the local pressure in the vapor return lines. In this case, the only parameter to be controlled
is the liquid level in the recooler.
If the vapor line serves as a continuous recooler (liquid nitrogen is injected into the
vapor line), then a control loop is required for each section of the ring to control the flow
rate of the injected liquid.
Figures 5a and 5b show the compact recoolers installed near the sector feed boxes and
end boxes. Figure 5c shows the continuous recooling system by injecting liquid into the
vapor line. Figure 5d shows a recooler with a pump that may be used if the pressure in the
lines must be reduced to a minimum .
..~15 S20 S25 S30 S35 S40 S45 S50 S55 N15 N20 N25 N30 N35 N40 N45 N50 N55
EAS. I I I I WEST I I I I EAST
CLUSTER CLUSTER I I CLUSTER
I I I I I I ' I I
~~~C~~~~~~~C~~~~~C~~
FLOW IN THE VAPOR LINE
40gfs 40gIs
I
S60 416 352 240 176
I I I
3a: ONE PLANT AT N40 FLOW IN 11IE LIQUID LINE, ONE RING.
I I I I
64 '84
38: TWO PLANTS AT NIS AND SIS FLOW IN THE LIQUID LINE, ONE RING.
I' I I
I
3C: TWO PLANTS AT NIS AND SIS. RECOOLERS EVERY 8Km, FLOW IN THE LIQUID LINES, BOTH RINGS
Figure 3. Flow in the collider rings for different configurations, in g/s.
211
M 180
I 160
L
L 140
L
I 120
I T 100
o E 80
NR 60
S S 40
0 20
F 0
1991 1992 1993 1994 1995 1996 1997 1998 1999
YEAR
Figure 4. LN2 forecast for the SSe.
Figure Sa. Recoolers near the isolation boxes. Figure 5b. Recoolers near the feed boxes.
THAOUGtI rH( u flJ& S

- -- -rr-n--'
t
a
, , ,
C~YO~l A'
CA¥O'ST A'
I I_ _ _ ~ _ _- -
I
L_
f(~OS "C 'V'·Pl R.Af l "~ ASUAl NT
.. r ...... O A .., ,, AI Of THf;!t{C I O ~
O V [5t~ )t. _,OU,j). t'II ftl~ LO ""-t.q ".a..IH Of 10fL APOA
LI...: AI 1M I,IP$P ~US UR: L OI~1A 8Ul 0 '11
Figure 5c. Continuous recooling. Figure 5d. Recooler system with vapor separator.
212
The vapor is extracted from the recooler and removed to the adjacent helium plant
through the vapor line. Based on the helium plant design, the suction pressure of nitrogen to
the plant is 0.13 MPa. The vapor flow is directed from the four strings of each sector of the
collider into the vertical conduit (transfer line) to the helium plant.
The liquid nitrogen is distributed by pumps at the air separation plants and by cold
pumps in the tunnel. The vapor generated by boiling nitrogen in the recoolers is transported
to the adjacent helium plants, where it is warmed to 300 K and vented to the atmosphere.
OPERA TION AT OFF -DESIGN CONDITIONS
Disconnected Strings
If one string is disconnected by isolating one section and the rest of that ring has to be
kept cold, then the nitrogen flow must be supplied through the parallel section (the other
ring) and/or, if possible, through the other direction of the ring. Solutions for such
conditions are shown below.
Cooldown and Warm-up of the 84 K Shield
The amount of liquid nitrogen required to cool the shield 10 a nominal section
(1080 m) is shown as:
To reduce shield temperature to 84 K 6560 kg

To fill the lines (liquid and vapor) 2220 kg
Heat load during cool down 3600 kg
TOTAL 12,380 kg
With an initial flow rate of 27 gis, increased gradually to 43 gis, the cooldown time of
the first section will be 97 h.
Cooldown Process of the Cold Mass
After cooling down the 84 K shield and maintaining the nitrogen flow required to
absorb the 84 K heat load, the next step is to cool the 20 K shield and the magnets to 90 K
by circulating cold helium. The 90 K helium is obtained by exchanging heat with liquid
nitrogen at the helium refrigeration plant. A nitrogen mass flow rate of approximately
2000 g/s is required for this process (7.2 tons/hour or 172.8 tons/day). The cool down
process takes approximately 21 days, so that more than three refills per day of the local
nitrogen reservoirs are needed to accomplish the first cooling wave of one string 4.3 km
long.
RESULTS
Scheme A: One plant located at N-40, the vapor line serving as a continuous recooler
The vapor line. For nominal sections 1080 m long and constant heat load per unit
length, the required injection of liquid in the vapor line is 16 gls per section.
Figure 6a shows the pressure drop in the vapor line of a string (four sections), for a
flow of 16 gls per section. If this liquid injection rate is maintained, the flow in the vapor
line is stratified.
213
MPA
0.3
HE
0.25 ''''' IN
P
R ../
V V CL ST RF EGI PN
0.2
E - /' - -
- -~
--
S - '4 --- / - - -f---
0.15
S
U
V
0.1
R / i'
E p ES UF EIII TH AFCS RI GS ~11r. rlOI IQI .,..,n
0.05
0 HE ~El UM PU NT
0
4.5 4 3.5 3 2.5 2 1.5 1 0.5 a 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
DISTANCE FROM THE FEED BOX IN KILOMETERS
Figure 6. Vapor pressure in the arc strings and in the east or west cluster.
The liquid line. For normal operation, the required flow rates are shown above, in
Figure 3 (see 3a for one air separation plant, 3b for two air separation plants, and 3c for two
air separation plants with N2 recoolers located 8 km apart). For pressures limited to a
maximum of 1.1 MPa in the lines, and without circulation pumps in the tunnel, the resulting
pressure levels are shown in Figure 7.
In order to pump liquid out to the LN2 dewars on the surface or to the helium plants,
the pressure at the lower leg of each vertical transfer line should be at least equal to the
static head plus the pressure in the dewar (see vertical arrows in Figure 7). If there is only
one nitrogen plant, pumps will be needed at least at S45, S55, NI5, N25, and N35. If there
are two plants, then pumps will be necessary at every main shaft.
MPa WI §ICLUS1li.R ONE PLANT AT N40
J
EAS.lCLUS~
1.2
~u~ lNG'
~
!
AP
NO
1.
;E'iJ
UMP NG
TN!
IONA
/
I~
- /
N -.....-..
~
A
N PL:t AT ! S
ITIO AL
PU> ~ING
I
P 0.8
I - r-.,. ~II~
R '-.
/
E
~
/
K
0.6 .~
/
~v
U / ,
R
E 0.4
.:..:.;.. /' / pop" "mp
'" 1.P,
0.3
/',- .. '
,
, --
/~ ....
0.2
-- -- --- --- -- - .--- -- .--- -- - - .--- -- ,~
-- - .-- --.--. .-----
V~~R I
~~~w_m ________ ~m~rnmm~
LOCATION
Figure 7. Pressure in the nitrogen system for different configurations. Vertical arrows are the required LN2
pressure in the tunnel for pumping liquid to the surface.
System operation when one critical string is disconnected. If there is only one air
separation plant, and one string between N40 and N35 (or between N40 and N45) is
disconnected, then the flow in the parallel line must be doubled in order to supply the
nitrogen to the rest of the disconnected line. The pressure drop in a 4.3-km line carrying
2000 g/s is 0.55-0.6 MPa. To prevent boiling in the first 4.3 km, the feed pressure must be
higher than 0.8 MPa, and the back pressure higher than 0.15 MPa.
System operation when two critical strings are disconnected. Assume that both
rings are disconnected between N40 and N45. The other parts of the system are cold at
normal operating temperatures. In this case, the liquid nitrogen must be supplied through the
214
remaining parts of the system-Le., through lines 87 km long. The whole nitrogen supply of
4200 gls must go through two lines (see flow rates in Figure 8a), as follows:
a. If the vapor line is used for cooling: Without circulation pumps in the tunnel, the
required supply pressure is 4.4 MPa (for disconnected string N40-N35) and
5.2 MPa (for disconnected string N40-N45). Of course, these pressures are
unacceptable, and pumps are needed to distribute the nitrogen. Figure 8b shows
the pressure distribution in the tunnel if the strings are disconnected and pumps
are used for circulation.
b. If the vapor line is usedfor liquid flow: This solution may be used only in a small
part of the system where the flow rates are higher than 1800-2000 gls. Lower
flow rates will result in an unacceptable temperature rise in the cryostat.
HEB o N1S
H60WO '60
vapor 2*135 g/! 4 6 S4 t6

IS illJ
11Qul<1 2* SO g/5 ~~i!
to H20 N40
Figure 8a. Liquid and vapor N2 flow in one ring in gls for disconnected strings east to N40. One air
separation plant (total N2 generation 4206 g/s).
MPA
4.5
P~MP~ ;P1, 2,P3. 4 1/ ! RE~ URE ORO

~~~clD
3,5
! i
I
II fuNN L \. T
P 3
! V , ,
1/
I
DISC l""NE ED,
i
R
~ I i I ~ RtNq , i
1\
1
2,5
S
U 2
!
I
I 1/1 i
,
\
R
E 1.5 i Vi ! i OPE lATta
~
l- V !
/
1
----
II
V1 V I 'L ! 'I
I--:::-::r::::I
---.
.,
.,/'
i I P?.1
0.5
o
---- Pr1.~ ··~·f!·
.-
V. V . ~, '. i V/;~ PAUlE . . ,
S40 S45 S50 S55 N05 N15 N20 N25 N30 N35 N40 N45 N50 N55 S05 S15 S20 S25 S30 S35 S40
LOCATION
Figure 8b. Pressure in the LN2 lines for disconnected strings with and without circulation pumps (one plant,
string N35-N40 disconnected).
Scheme B: Plants at N15 and SIS, the vapor line serving as a continuous recooler
For this configuration, the pressures and temperatures in the vapor line are the same as
for Scheme A (see Figure 6). The mass flow rates are as shown in Figure 3b.
Compared with the previous configuration, the mass flows in the liquid lines are only
50 percent, and the lines are shorter by 50 percent. These two factors reduce the pressure
drop in the lines. Still, in order to pump the liquid to the surface, to the helium plants, it is
required to have liquid pumps in the tunnel. If no such pumps are installed, the pressure
distribution is as shown in Figure 8. To prevent vapor in the liquid line, the pressures in the
liquid line should be above the saturation point. Two-phase flow across the U tubes may
block the liquid flow and introduce disturbances in the system.
215
Refrigeration if one string is disconnected: If a string is disconnected, then the flow
in the parallel string may be doubled, and the resulting pressures for two plants are much
lower than the resulting pressures for one plant.
Scheme C: Plants at N15 and SIS, liquid flow recooled by compact recoolers
As mentioned earlier, in the regions between NI5 and N30 and between the east
cluster and S30, the inclination of the tunnel is too small to allow the free flow of boiling
liquid in the vapor line, thereby preventing the use of this line as a distributed recooler. The
detailed design of some other parts of the ring may raise problems that prevent the use of
the vapor line as a continuous recooler-in particular in the east and west clusters. In all
these parts of the ring, the installation of compact recoolers may be necessary.
Assuming that the nitrogen recooling is based only on compact heat exchangers, the
following schemes were analyzed:
Scheme Cl only one recooler every 8 km (vapor line not in use)
Scheme C2 a recooler every 4 km + a vapor separator every 8 km
Scheme C3 a recooler every I km + a vapor separator every 8 km
Scheme C.l. In this model, all the heat load is absorbed by the liquid, which is
recooled every 8 km. The minimum flow to keep the temperature between 80 K and 89 K
for a heat load of 25,600 W/s per 8 km is 1374 gls. For the east and west cluster, the load is
assumed to be 50% higher (38,400 W/s per cluster) and the minimal flow rate should be at
least 2060 gls. Figure 3c shows the mass flows in the ring, including the flow distribution
requirements for this scheme. The resulting mass flow rates require operating pressures
above acceptable levels, even with circulation pumps installed in the tunnel every 8 km.
Scheme C.2. This scheme and some of its variations were also analyzed. It was found
that a compact heat exchanger is sufficient at the end of each sector and that a heat
exchanger, a vapor separator, and a liquid pump are needed in the feed box of the sector in
order to maintain the temperature lower than 89 K. This arrangement allows the reduction
of the pressure in the liquid lines to a level enabling two-phase flow. Figure 5d shows a
schematic diagram of the piping and recooling system for this scheme.
Figure 9 shows the pressure-temperature (P-T) diagram for a mass flow of 920 gls in
the line between SI5 and S20 and a mass flow of 838 gls in the line between S20 and S25.
The discharge pressures were reduced to a minimum of 0.1 MPa. An initial pressure of
0.4 MPa at S 15 is sufficient to maintain these flows. In other parts of the system where the
flow rates are smaller, the required operating pressures are even smaller.
Figure 9 also shows the P-T diagram for different locations in the tunnel (uphill flow
and downhill flow) and a comparison of schemes C.2 and C.3.
Scheme C.3. The results of this scheme are similar to those shown for continuous
recooling.
CONCLUSIONS
Number of air separation plants. While one air separation plant may be more
economical, two plants are preferred if availability is an important criterion for selection.
Twelve LN2 refrigerators may be preferred if the venting of large quantities of LN2 from
every helium plant is not acceptable.
216
...".,---::-:-:-------731g/s from S25to S30 1.'11 '---:54'-:-::-2g-:-/s-f::-ro-m-S:=4::5--:to-:-;;S~40~-
g WI 649g/s from S30 to 835 '" 460glS from S40 to S35 ...-
~ UPHILL ~i> _---- g lit DOWNHILL ~i>_---------

i .. c~~--------- 830
l:\! ..
~ ". ~o+.
c~~----
Wo- ... ..o~-
t-~ ffi.. ~..t-.
rfi~ '",w ~~..
<;,t- ~ '" <;,t-~
l!!
:1835",,,,,,,,,,,,,,, .. , __ .... ".... , . ..
III
u " ~ dUd M _ ~
.. IB U III .. .. ..
PRESSURE (MPs) PRESSURE (MPs)
A RECOOLER EVERY 4 Km
...
920g/s from S15 to S20
.. 164 g/s from S55 \0 N15
....
838g/s from S20 to S25
gill
W
a:
~
~
a:
W",
'"
0-
rfi'",.
~
n.
PRESSURE IMPa)
•• ••
PRESSURE (MPa)
III d
A RECOOLER EVERY 1 Km
Figure 9. LN2 temperature vs. pressure profiles.
Location of air separation plants. N40 is the preferred location from a transportation
point of view, because of its proximity to main highways 1-35E and U.S. 77. However,
N15-which is close to the HEB and has the highest elevation-is preferred from the point
of view of operation of the HEB and the LN2 distribution.
Recooling schemes. Providing a compact recooler every 4 km results in a temperature
rise of up to 89 K in the 84 K shield. If continuous recooling is implemented by injection of
liquid into the vapor line, the temperature rise in the arc strings may be limited to 84 K.
Pumps. Pumps are required in every sector to supply liquid to the surface, and also to
distribute the liquid in case of abnormal operating conditions. The flow rates and the
pumping pressures are smaller if two air separation plants are installed.
The major upset conditions. The disconnection of the two rings close to the air
separation plant will require operation of circulation pumps in order to operate the system.
The disconnection of a helium plant will cause an increase in the vapor return line
pressure and an increase in vapor temperature to 92-94 K. The pressure in the liquid line
will have to be increased to 0.4 MPa in order to operate the continuous recooling. To avoid
all these changes in temperature and pressure, cold vents of the nitrogen vapor may be used.
Supply LNZ by surface transportation (trucks) from a remote air separation
plant. After the year 1999, demand is expected to be 4200 gls (363 ton/day). This solution
will require more than 17 trucks per day. (One 8000 gallon truck may carry 22 tons LN2.)
REFERENCES
1. T.H. Nicol, "Cryostat Design for the Superconducting Super Collider," Winter Annual Meeting, ASME,
November 1990.
217
MATERIALS AND PROCESSES SELECTION FOR
SSC COLLIDER DIPOLE MAGNETS
K.L. Agarwal, B.w. Croll, E.M. Gilchrist, and D.K. King

P.O. Box 85990, San Diego, CA 92186-5990
ABSTRACT
Considerable research and development effort has gone into design and produc-
tion of prototype model dipole and quadrupole magnets for the superconducting super
collider (SSC) at the Central Design Group, Brookhaven National Laboratory (BNL),
Fermi National Accelerator Laboratory (FNAL), Lawrence Berkeley Laboratory
(LBL), and Texas Accelerator Center (TAC), in addition to work done at CERN. Experi-
ence gained by GDSS in designing and building MFTF, PMS, and LCP magnets was
merged with the work obtained by the SSC Central Design Group and other accelerator
magnet programs in making decisions on selection of materials and processes for the
collider dipole magnets. The two overwhelming issues in CDM design are high reliabil-
ity and high producibility of magnets at a low practical cost. The GDSS Materials and
Process Engineering group took a close look at the SSCL design, the selection of materi-
als, and processes, and conducted trade studies based on current CDM requirements
and past material selection, design, and fabrication experience. GDSS recommended
some changes, notably: (1) change of beam tube material from 'fYpe 21Cr-6 Ni-9Mn
CRES to Type 304LN CRES, (2) superconductor cable-to-cable insulation scheme from
one layer of polyimide film and a single barber pole wrap of fiberglass-epoxy to two
layers of polyimide film with a polyimide-type adhesive, (3) substitution of machined
G 10 CR fiberglass epoxy end pieces with less costly and mass-producible injection or
resin transfer molded end pieces, and (4) substitution of precision machined, shrink-fit
metal and composite re-entrant cold mass support posts with less expensive and mass-
producible injection or resin transfer molded posts. The last two changes are not final
and need development testing, which is under way. Several other changes are under con-
sideration, including replacement of extruded AI 6063-T6 liquid cryogenic line with
Type 304L CRES and coating the beam tube copper plating with a very thin layer of
gold for reducing photo-desorption and protecting copper plating from oxidation. The
superconductor material remains the same, SSCL optimized, high homogeneity, multi-
filamentary Nb-47% Ti wire with copper stabilizer. The cold mass skin and the end
domes shall be fabricated from 'fYpe 304LN CRES, a proven material with which GDSS
has past fabrication experience, and with an extensive design data base. This choice will

be extremely valuable for preparing the code case for ASME compliance approval of
the cold mass structure.
INTRODUCTION
The collider dipole magnets bend the path of proton beams into precise orbits dur-
ing the acceleration and beam storage phases of SSC operation. They are required to
produce 6.6 T magnetic field with a limit on allowable harmonic multipoles. There will
be 7,964 dipole magnets of 15.165 meter magnetic length and 496 magnets of 12.638
meter magnetic length. The field quality of each magnet must be within the allowed
limits of the harmonic multipole specification. The field errors could arise due to vari-
ability in the magnetization characteristics of superconductor, collar, or yoke material,
and also due to their physical placement in the magnet. The former can be minimized
by controlling chemistry and microstructure of materials and the latter by optimizing
reproducibility of manufacturing processes. The correct orbital alignment of magnets
would require congruency of magnetic and geometrical centers of each of the 8,560
magnets. This mandates dimensional stability requirements on each magnet's cold mass
support posts, which could creep, and welded vessel structures, which could relax. To
reduce creep the support posts will be lightly loaded, and to maintain stability of physical
alignment the welded structures will be stress relieved prior to installation of cold mass.
MATERIALS AND PROCESSES SELECTION CRITERION
Materials with reproducible properties, and reproducible fabrication processes are

key to producing a consistent quality product. The materials selected must meet the de-
sign and performance requirements of the dipole magnets. The heat-to-heat or batch-
to-batch variation in the key material properties should be minimized so as not to
exceed design limits. To accomplish consistent material properties, material chemistry
will have to be tightened within the limits of industry specifications, preferably without
incurring a cost penalty. This has been done in the past for Type 304LN CRES, the mate-
rial of choice for the cold mass shell, end dome, and beam tube. The Type 304LN CRES
with tailored chemistry meets and exceeds requirements of ASTM A240. General Dy-
namics has a statistically significant cryogenic data base for this material, from building
superconducting magnets for the Mirror Fusion Test Facility (MFfF-B), Preprototype
Magnet System (PMS), and Large Coil Program. To reiterate, for the collider dipole
magnet design, achieving consistency in material properties takes precedence over max-
imizing the properties to their limit. As an example, for the yoke material, it is prefer-
able to select a consistent quality low carbon electrical grade steel with reproducible
magnetic properties than to select a costlier ultra-high purity iron that may possess
slightly higher but inconsistent magnetic properties.
To summarize, the rationale for final materials and processes selection is based on
the following criteria:
1. Adequate and consistent mechanical, physical, thermal, and magnetic properties

2. Compatibility with the cryogenic, radiation, and vacuum environment
3. Large design data base and specification coverage
4. Availability in required sizes and shapes
5. Proven joining and fabrication methods
6. Service history
7. Cost.
220
MATERIALS SELECTION
Based on the above criteria, the following materials have been selected for the col-
lider dipole magnet components:
1. Cold mass materials

Conductor SSCL optimized high homogeneity Nb-47% Ti
multifilamentary cable
Cable insulation Two layers of polyimide film coated with
polyimide adhesive
Wedge material C10200 copper with polyimide film insulation
Coil end pieces Machined polyimide base glass fiber
reinforced composite
Collar Type 21Cr-6Ni-9Mn CRES steel with
15% cold work
Collar keys Phosphor bronze
Yoke Low-carbon electrical steel
Yoke keys Type 304 CRES
Beam tube Type 304LN CRES inside plated with copper
Cold mass shell Type 304LN CRES
End dome and end plate Type 304LN CRES
Cold mass cradle Type 304L cast/forged
2. Cryostat materials
Thermal shields AI 6061-T6 sheet
Cryogenic lines Type 304L CRES
Support posts Filament-wound composite/Injection molded/
RTM glass fiber-reinforced plastic
Insulation blanket MLI, PET film with vapor-deposited
aluminum and polyester spacers
Vacuum vessel Carbon steel Type SA516
Port ring support Cast carbon steel
3. Interconnect region
Bellows Type 316L CRES
Industry specifications have been utilized to the greatest extent possible. However,
materials for critical components will be procured against GDSS-developed materials
specifications. Common materials will be procured against applicable ASTM/ AMS or
military specifications.
RATIONALE FOR CHANGE OF MATERIALS
Materials changes have been proposed for several magnet components, most nota-
bly the beam tube material, cable insulation, end pieces, and cold mass support posts.
Beam Tube Material: The conceptual design called out 21Cr-6Ni-9Mn corrosion
resistant steel, also known as Armco Nitronic 40, for the beam tube material. Lower
magnetic permeability (J.Lr = 1.002) and higher tensile strength (620 MPa ultimate ten-
sile strength) were the primary drivers for selecting this material over a 300 series
corrosion-resistant steel. Welding the beam tube to a 300 series flange and the cryogenic
properties of the resultant weld joint were matters of concern. A study supported by the
Superconducting Super Collider Central Design Group and conducted at the National
Institute of Standards and Technologf indicated severe reduction in ductility at
221
cryogenic temperatures in autogenous tube-to-flange welds (21Cr-6Ni-9Mn tube to
four different flanges: Type 21Cr-6Ni-9Mn, Type 304L, Type 316L, and Type 316LN).
Obtaining code case approval per ASME Boiler and Pressure Vessel Code Section VIII,
Division 1, for 21Cr-6Ni-9Mn tube-to-Type 304L or 316L flange, would have required
a considerable effort in development testing of weld properties. GDSS Magnetics and
Stress Analysis groups assessed impact of changing the beam tube material from
21Cr-6Ni-9Mn CRES to Type 304LN CRES and concluded that either material could
be used without compromising the field quality or the structural integrity. The lower
strength of Type 304LN relative to Type 21Cr-6Ni-9Mn CRES required a slight increase
in wall thickness ofthe tube. Using Type 304LN CRES for beam tube improves chances
for ASME code case approval by allowing production welding without the need to per-
form production-level impact testing.
Cable Insulation: The conceptual design for cable-to-cable electrical insulation re-
quired one layer of 1.0 mil polyimide film (Kapton) followed by a barberpole wrap of
B-stage epoxy/fiber-glass tape. Application of temperature and pressure during curing
cycle sized the coil for subsequent collaring and yoking operations. The following reli-
ability issues were perceived with this scheme:
1. Inherent brittle nature of epoxy-type resins at cryogenic temperatures
2. Micro-cracking of epoxy resulting in quench initiation
3. Existence of only one layer of polyimide insulation at overlap locations, thus
increasing risk of electrical breakdown
4. Higher probability of initiating a crack in polyimide film by loose glass fibers
5. Loss in ductility due to radiation damage, and consequently more embrittlement
and cracking.
The BNLlDupont insulation scheme of using two separate layers of polyimide film,
the inner wrap with outside coated, and the outer wrap with both sides coated, with a
ductile, radiation-resistant polyimide adhesive, offered an attractive alternative to poly-
imide/fiberglass-epoxy scheme, and was made the baseline insulation scheme. The out-
er wrap is an alumina-filled polyimide film and has significantly higher cut-through
resistance. There were some initial problems with the higher curing temperature and
pressure (225 C and 69 MPa) of the BNLlDupont scheme. Apparently, the cable
0
strands developed some kind of electrical bond and adversely affected the fast ramping
characteristics of the coil. This problem has since been corrected by modifying the cure
cycle by applying exclusively either the relatively high temperature or the high pressure.
Model coils cured using the modified curing cycle have performed satisfactorily without
the fast ramp rate problems encountered previously.
RADIATION CONSIDERATIONS
In the collider dipole magnets, a permanent source of beam loss and associated
radiation is interaction of the circulating protons with molecules of residual gases in the
beam tube 3. In the warm sections of the beam tube, the residual gas is prominently N2.
In the cold sections, the residual gases are H2, CO, C02, and CH4• These gases come
from the beam tube wall, produced by the photodesorption process, due to synchrotron
radiation from the accelerating proton beam. The synchrotron radiation itself does not
directly contribute to the radiation dose; being of low energy, it cannot pass through
the thickness of the beam tube material.
The maximum dose occurs in the median plane toward the ring center at the inner
radius of the coil, and has been estimated at 100 Mrad. Considering the uncertainties
and the limitations of the transport codes used in calculation of radiation dose, and pos-
222
sibility of overlooking hot spots by the codes, the radiation level for design purposes
has been set to 1,000 Mrads at the inner coil location (r = 5 cm). The radiation levels
at other radial locations are scaled using an exponential relationship. Approximately
99% of the radiation dose comes from gamma radiation and the balance from neutrons.
There is a way to reduce the radiation dose by a factor of 5. The total gas desorption
drops by a factor of five and so would the radiation dose if the copper plating is supplem-
ented by a very thin plating of gold. The benefits of radiation dose reduction, reduced
gamma heating, and longer beam life are to be evaluated against additional cost of gold
plating before making an engineering judgement.
The radiation dose is too small to affect any of the physical or mechanical proper-
ties of metallic materials including superconductor. However, it can adversely affect the
structural properties of the composite materials, particularly in the cold mass region,
where the radiation dose is expected to be the highest. The following is a list of radiation-
susceptible magnet components along with the expected total radiation exposure.
Component Total Radiation Exposure

(Mrad)
1. Coil end pieces 1,000
2. Cable-to-cable insulation 1,000
3. Pultruded or molded 180
electrical bus housing
4. Multi-layer insulation 30
The highest radiation dose occurs in the coil region, and it precludes use of epoxy-
based composites for fabricating end pieces. The aromatic polymer films and aromatic
polyimide polymers, which exhibit five to ten times higher resistance to radiation-
induced changes in compressive strength, are considered for electrical insulation and
to meet structural requirements. The resin system chosen for molded or pultruded elec-
tric bus housing shall be screened for adequate radiation resistance.
COLD MASS END PIECES AND SUPPORT POSTS
Selection of materials for the coil end pieces and cold mass support posts is not
finalized pending resolution of producibility issues. A polyimide-based glass fiber rein-
forced composite material has been selected in the interim for machining the end pieces
required for the prototype magnets.
The polyimide-based composite material meets the radiation resistance require-
ments. Several injection moldable, resin transfer moldable, and compression moldable
material options are being pursued. Adequate radiation resistance is one of the require-
ments for the material. The other is a relatively high (> 230· C) glass transition tempera-
ture. The injection moldable candidate material is identified as Amoco Torlon 5030,
the resin transfer moldable candidate material as Allied Signal's phenyl triazine-based
resin "Ctyo-Rad" with glass fiber reinforcement, and compression moldable polyimide-
based glass fiber reinforced composite "Spaulding-Spaulrad."
The material for injection moldable support posts, based on requirements of low
thermal conductivity, moderate mechanical strength, and good moldability, was identi-
fied as GE Ultem 2100. Subsequent thermal and stress analysis showed that Ultem 2100
could not meet either of the requirements. The present baseline is therefore a filament-
wound post with carbon fiber reinforcement in 4.5K to 20K section, and glass fiber
reinforcement in 20K to 295K section.
223
CONCLUSION
Materials and processes have been selected for construction of collider dipole mag-
nets. The list includes structural materials like Type 304LN CRES, 'JYpe 21Cr-6Ni-9Mn
CRES, and Al6061-T6. These materials have been used for cryogenic service. They have
a long service history and a large data base.
The injection moldable or resin transfer moldable materials for the coil end pieces
and cold mass support posts are under continuing development, and their cost saving
benefits will be adopted only if test results prove they meet all the requirements for the
application.
ACKNOWLEDGEMENTS
The work described herein is being accomplished under contract to the Universi-
ties Research Association in support of the Superconducting Super Collider project for
the U.S. Department of Energy.
REFERENCES
1. J.R. Sanford, and D.M. Mathews, Editors, "Site Specific Conceptual Design of
the Superconducting Super Collider," Report No. SSCL-SR-1056, July 1990,
Superconducting Super Collider Laboratory, 2550 Beckleymead Avenue,
Dallas, TX.75237.
2. T.A Siewert, C.N. McCowan, and D.P. Vigliotti, "Cryogenic Material Properties
of Stainless Steel Thbe-to-Aange Welds," in Material Studies of Magnetic Fusion
Energy. May 1988, U.S. Department of Commerce, National Institute of Stan-
dards and Thchnology.
3. I.S. Baishev, AI. Drozhdin, and N.V. Mokhov, "Beam Loss and Radiation
Effects in the SSC Lattice Elements," Report No. SSCL-306, November 1990,
Superconducting Super Collider Laboratory, 2550 Beckleymeade Avenue,
Dallas, TX 75237.
224
UTILIZATION OF GAMMA RAY INSPECTION SYSTEM FOR TOMOGRAPIDC
IMAGING AND DIMENSIONAL ANALYSIS OF COMPLETE MODEL MAGNET
COLD MASSES AND COLLARED COIL SECTIONS
D. Bein and 1. Zbasnik

Dallas, TX 75237
1. Bolger, B. Firey, B. Rice, B. Crane, G. Rabaey, 1. Wesick, and L. Morgan
International Digital Modeling Corporation (10M)

1901 Rutland Drive
Austin, TX 78758
INTRODUCTION
Non-destructive digital photon transmission imaging can be performed with the IRIS™
system to provide internal component dimensions and locations of manufactured
superconducting magnets and sub-assemblies. With non-destructive evaluation of the
magnet components and sub-assemblies at critical stages of the manufacturing process, it is
possible to assess the quality at each stage with the goal of improving manufacturing
reproducibility and reliability. Evaluations of the feasibility of the IRIS™ system have been
performed on a NbTi billet used for the production of monofilament rods used in the
manufacturing of multifilament strand. The system has also been used to perform
inspections on 40-mm and 50-mm collared coil sections and on cold masses. The
dimensional accuracy for locating the conductor winding positions in a collared coil section
has been found to be ±80 ~m. Non-destructive analysis using the IRIS™ system offers
examination capabilities in process and post-assembly inspection previously unavailable
except through destructive physical analysis by sectioning and polishing for optical
inspection after vacuum impregnation. This novel inspection tool may give us deep insights
into the validity of our mechanical and electromagnetic models currently used to evaluate
magnet design and may lead to substantial savings per magnet analysis.
*Operated by the Universities Research Association, Inc., for the U. S. Department of Energy under Contract
No. DE-AC35-89ER40486.

DESCRIPTION OF THE CAPABILITIES OF THE IRIS™ SYSYEM
More than 12,000 superconducting magnets will be needed for the construction of the
Superconducting Super Collider (SSC).! It has been estimated that an allowable failure rate
for the Collider Dipole Magnets is 0.04 failures per million installed hours.2 Clearly, quality
control will be an important issue during magnet fabrication. The suitability of using the
IRIS™ system for non-destructive evaluation during magnet production has been
investigated. Specifically, the use of the IRIS™ system for finding flaws in NbTi billets
during manufacturing of superconducting multifilament strand and for examining collared
coil sections and cold masses will be addressed.
The devices used to acquire the images shown here contain a 10-Ci Co60 fan beam
source and 71 scintillator assemblies, spaced in 10 increments. Each detector assembly
consists of a plastic scintillator, photomultiplier tube, and discriminator card. Signals from
the discriminator cards are transferred via twisted pair cables to scaler modules located in
the CAMAC crate. Data are transferred from the CAMAC to a MicroVAX 3300 for
processing. Figure I is a schematic of one of the systems used in this study.
IDM Corporation
IRIS '' VARIScan
Inspection Unit
t.. . 5....,
........ , . . . . oJ
Figure 1. Schematic diagram of the IRIS TM system. (Reproduced with pennission of IDM Corp.)
This is the IRIS™ model 004 general-purpose inspection unit. In this system the source
and the detectors remain fixed, and the object being scanned is rotated, thereby allowing the
acquisition of a tomographic image. The object to be scanned can be placed at variable
distances from the radioactive source. This provides geometric magnification and can
increase the machine's resolution by up to 50%. With model 002, the source and the
detectors rotate around the object being scanned. Because of this, the distance between the
source and the scanned object is fixed .
226
Several types of scans can be performed on the IRIS™ instrument. The most detailed
of these is the analytic scan, which is essentially a tomographic scan. In the higher-
resolution versions of the analytic scan the detector bank is rotated about the source position
by fractions of a detector spacing, thus increasing the effective number of detectors. This is
called subpositioning. There are seven data-taking subpositions used in the highest
resolution analytic scan mode. Data are taken in 116° increments as the sample is rotated
through one complete revolution. A scan of this type typically takes about 2 hours to
perform and achieves a resolution of 1.2 mm. The lowest resolution analytic scan can be
performed in approximately 15 minutes, although the resolution suffers and is much poorer.
Scan times could be reduced by a factor of 20 in a system optimally designed for SSC
applications by using higher intensity sources and more efficient detectors. Increasing the
number of detectors or increasing their packing efficiency could further reduce the
necessary scan times.
Cracks of widths between 13 and 500 J..lm can be measured to an accuracy of 10%
using the IRIS™ system. These measurements, which are below the resolution of the
machine, can be made because of the a priori knowledge that a crack exists. This concept is
referred to as detectability, as opposed to resolution.
Another type of scan, called a gaging scan, is available for cylindrical or tubular
objects. In this type of scan fewer rotations of the object are used, and a complete image of
the object is not reconstructed. Instead, the dimensions of the object are found and flaws are
located. Because less data need to be acquired, gaging scans can be completed more quickly
than analytic scans.
A digital radiography mode is also available on the IRIS™ instruments. This gives the
same type of information one would find in a conventional X-ray. However, the data exist
in a digital form and several image-processing techniques can be applied, yielding various
flaw-detection resolutions.
EXPERIMENTAL RESULTS
NbTi Billet Scans
A NbTi billet was machined with three spherical voids of diameters of 3.18, 1.59, and
0.79 mm, for use as a sample to characterize the IRIS™ system's ability to find flaws in the
production of monofilament rods used in the production of multifilament strand used in
superconducting cable. The cylindrical billet had three circular holes electrically discharge
machined through the diameter. Cylindrical plugs were made of material cut from one end
of the billet to fit into these holes. Each plug was cut lengthwise in two, and a hemispherical
void was machined into each piece. The plugs were then cooled to dry-ice temperatures and
press-fitted into the matching hole in the billet. This gave tolerances of a few microns
between the plug and the hole in the billet.
The 3.18-mm and the 1.59-mm flaws were detected by the IRIS™ system, but the
0.79-mm defect was not. Part ofthe difficulty in detecting this flaw was the rough condition
of the exterior surface of the sample billet. This increases the noise in our flaw detection
algorithms. One expects a much smoother surface in a production environment. Currently
these billets are inspected by ultrasonic techniques. Because of the attenuation coefficient of
the NbTi material, a low-frequency transducer must be used for the inspection. The
wavelengths used in the ultrasonic inspection set a limit of 2.5 mm as the minimum flaw
size that can be reproducibly detected.
Figure 2 shows an analytic scan of the billet cross section with the 1.59-mm flaw.
227
Figure 2. Analytic scan of NbTi billet, showing 1.59-mm flaw.
The flaw is clearly visible, and the gaps between the plug and the surrounding billet
can also be seen. A subtraction between a flawed slice and an unflawed slice can be done in
order to highlight the flaw.
Taking a full tomographic scan takes too long to be practical for inspection purposes
in a production environment. Flaws can be detected in a billet using a faster data acquisition
mode in which a full tomographic reconstruction is not made. Figure 3 is a digital
radiograph of the billet taken with the IRIS™ system.
Figure 3. Digital radiograph ofNbTi billet, showing 1.59-mm flaw.
This figure shows the difference between the local density and the average density
measured over the entire billet. Each pixel is 1.9 mm in height. The outline of the plug can
be clearly seen. It is not circular due to the aspect ratio chosen. The 1.59-mm flaw can also
be seen in the center of the plug outline. Figure 4 is a triangulation display in which the
position of the flaw in the billet slice is shown.
228
Figure 4. Triangulation of the flaw position within the NbTi billet.
This is determined from data taken by seven digital radiographs similar to that shown
in Figure 3 except at different rotations. In this type of scan the number of rotations has
been reduced from a multiple of 360 to seven, giving a clear scan time. It is estimated that a
complete scan of a billet can be outlined in 2 hours using this method.
Collared Coil Section Scans
Figure 5 is an image of a 40-mm collared coil section.

The copper keys and the individual conductor windings are clearly visible. Figure 6 is
a zoom of one of the quadrants ofthe collared coil section shown in Figure 5.
Figure 5. Tomographic image of a 4O-mm collared coil section.
229
Figure 6. Tomographic image showing a damaged tum in the 40-mm collared coil section.
A damaged tum is shown in this tomographic image. Also shown are two turns that
are almost in contact with each other, perhaps indicating that the turn-to-turn insulation was
extruded from between the turns due to high prestress. International Digital Modeling
Corporation has developed software procedures that utilize our a priori knowledge of the
coil geometry to measure the conductor (turns) endpoints to within 80 !lm. This is less than
the resolution of our tomographic instrument. Figure 7 is an as-made CAD drawing of the
quadrant shown previously in Figure 6 using data obtained from this procedure.
sse Magnet Conductor Loco ions
Figure 7. CAD drawing showing winding positions.
Recently a deconvolution procedure has been developed that offers increased

resolution of our images. Figure 8 is an example of this improved resolution on a scan from
a 50-mm collared coil section. It represents our best collared coil image to date.
230
Figure 8. Tomographic image of a 50-mm collared coil section showing improved resolution using the
deconvolution technique.
Cold Mass Scans
Several scans of cold masses of both Lawrence Berkeley Laboratory (LBL) and
Brookhaven National Laboratory designs have been performed. Figure 9 shows one of the
scans of the LBL cold mass. The skin welds, alignment keys, and individual conductor
windings in the collared coil section can be seen. Figure 10 is an expanded region of the
collared coil section shown in Figure 9, to which the deconvolution procedure has been
applied. This helps resolve individual conductor positions.
Figure 9. Tomographic image of the LBL cold mass cross section (NC9 design).
231
Figure 10. Expanded image of coil section after application of the deconvolution procedure.
In order to see the conductor windings more clearly, scans can be made through the
gaps between the yoke packs. Figure 11 is a deconvolved image taken in this fashion. The
individual turns are clearly shown in this image.
Figure 11. Tomographic image shot between the yoke packs.
CONCLUSION
Several capabilities of the IRIS™ system have been demonstrated. Flaws in NbTi
billets were found that are smaller than the resolution detectable by conventional ultrasonic
methods. The positions of turns and integrity of insulation were determined in two collared
coil sections. In addition, the IRIS™ system can be used to determine conductor positions,
the symmetry and angular position of conductor blocks, and the presence and size of gaps
between the collar and yoke, and between the yoke and shell in cold mass assemblies.
232
Our future work will involve the application of the IRIS™ system to the dimensional
analysis of conductor position, yoke and shell interface, and yoke and collar interface in a
50-mm model. The intent will be to perform this analysis first at room temperature, then on
a cold magnet at temperature and fully energized. We also intend to experiment using the
IRIS™ system for alignment of the cold mass inside the cryostat and for location of the
interconnect pipes for in-process quality control.
REFERENCES
1. R. W. Baldi et al., "The ColIider Dipole Magnet Program," Proceedings of the Third International
Industrial Symposium on the Super Collider, Plenum Press, New York, 1991, p. 539.
2. Andrew Kytasty, "Proposal for an Approach to Achieve CDM Reliability," Proceedings of the Third
International Industrial Symposium on the Super Collider, Plenum Press, New York, 1991, p. 283.
233
THERMAL PERFORMANCE OF VARIOUS MULTILAYER
INSULATION SYSTEMS BELOW 80K
W.N. Boroski, T.H. Nicol, and c.J. Schoo
Fermi National Accelerator Laboratory"

P.O. Box 500
Batavia, Illinois 60510 USA
INTRODUCTION
The sse collider dipole cryostat consists of a vacuum shell operating at room
temperature, two thermal shields operating near 80K and 20K respectively, and the
superconducting magnet assembly operating near 4Kl. The cryostat design
incorporates multilayer insulation (MLI) blankets to limit radiant heat transfer into the
80K and 20K thermal shields. Also, an MLI blanket is used to impede heat transfer
through residual gas conduction into the 4K superconducting magnet assembly.
A measurement facility at Fermilab has been used to experimentally optimize the
thermal insulation system for the dipole cryostat. Previous thermal measurements have
been used to define the 80K MLI system configuration and verify system performance2•
With the 80K MLI system defined, the current effort has focused on experimentally
defining the optimum insulation scheme for the 20K thermal shield. The sse design
specification requires that radiant heat transfer be limited to 0.093 W 1m2 at an
insulating vacuum of 10-6 torr.
The radiant heat transfer budget to 20K can be met without MLI at insulating
vacuums near 10-6 torr. Research indicates that very low heat transfer rates can be
achieved between two low emissivity surfaces in the temperature regime below 80K,
provided the insulating vacuum is < < 10-6 torr. 3 However, this temperature region is
very sensitive to residual gas conduction, with small pressure increases causing
significant increases in total heat transfer. Under ideal laboratory or controlled
conditions, low insulating vacuum levels are reasonably achieved and maintained. In
contrast, insulating vacuums in large accelerators vary significantly over time. For the
Operated by Universities Research Association, Inc. under contract with the United States
Department of Energy

sse cryostat, variations in insulating vacuum will cause oscillations in refrigeration
heat loads. Thus, an important advantage of installing MLI on the 20K shield of the
sse cryostat is that the reflective layers serve as gas conduction barriers as well as
radiation shields. These barriers will buffer the effects of transient conditions on the
accelerator refrigeration system.
The performance of several MLI configurations has been evaluated under
nominal and degraded insulating vacuum. For each system configuration, the warm
boundary temperature remained fixed near SOK while the cold boundary temperature
was varied from 10K to SOK. Since the sse dipole cryostat employs aluminum thermal
shields, all experimental arrangements included aluminum tape on the warm and cold
boundaries. Heat transfer measurements were made with no MLI installed on the cold
boundary, and then with 10 and S-layer MLI blankets on the cold surface. Additionally,
the effect of aluminum coating thickness on MLI performance was studied by
measuring the performance of systems with different aluminum coating thicknesses.
EXPERIMENTAL SETUP
Thermal measurements were made in the Heat Leak Test Facility (HLTF)
illustrated in Figure 1. The HLTF as configured for 20K MLI measurements has been
previously described in detail, and thus only a brief overview is presented here.4 Test
samples are installed around the cold plate, a cylindrical copper drum with an outer
surface area of 0.36 m2• The cold plate outer surface is covered with aluminum tape
to simulate the aluminum shield of the cryostat. MLI test samples are installed as
blankets around the cylindrical portion of the cold plate and as discs on the bottom
section facing the SOK thermal shield. Blanket and disc layers are interleaved and
taped to minimize the impact of the joint configuration. The cold plate is cooled by
solid conduction through a heat flow meter. Heat flow meter accuracy is ± 1 mW near
20K, which corresponds to a heat flux of ± 2.S mW1m2 given the cold plate surface
area. The heat flow meter is attached to the anchor plate, which establishes the cold
boundary temperature. Anchor plate temperature variation is achieved by varying the
flow rate of cooling gas through tubing soldered to the cylindrical inner surface. A
temperature controller is used to balance electrical power against this flow to maintain
a constant temperature. Using the temperature controller and vent gas from SOO-liter
liquid helium dewars, constant temperatures over the range 10K to SOK are achieved.
Encompassing the test sample is a copper shield operating near SOK. The inner
surface area of the SOK shield is approximately 0.6S m2, with a radial distance of
approximately S em between the SOK shield and the cold plate. The SOK thermal
shield is cooled through a bolted connection to the bottom of an LN2 reservoir. The
inner surface of the 80K shield is covered with aluminum tape to simulate the inner
surface of the cryostat aluminum SOK shield.
System pressure in the HLTF is measured with two Bayard-Alpert ionization
gauges. Historical.z, the lowest insulating vacuums achieved in the HLTF were on the
order of 6 x 10 torr. This level of vacuum had been acceptable for past
measurements on SOK MLI systems and cryostat mechanical support structures.
However, it became clear early in the 20K MLI measurement program that insulating
vacuum in the 10-7 torr range would be necessary to reduce residual gas contribution
to negligible levels. Consequently, modifications were made to the vacuum system
during the course of the program: the pumping system was reconfigured and the
vacuum port size enlarged for increased throughput. This explains why some of the
data presented is over broader pressure ranges than others.
236
o 30
II I I II
SCALE (em)
EVACUA ION PORT
LHe / IN 2 I NER RESERVO IR

lHe , LN2 TEST VESSE L
LNZ SHIELD RESERVO IR
TEUPERA URE
I NTE~ED I A TE
fLANCE
VACUUM CAN ADAP TER
ANCHOR PLATE
~i'[I~~lff~~: HEAT METER

COLD
loll PLATE
I TEST
HEATER
CAL IBRAT ION
SAW'LE
COLD PLAT[
6DK THERIAAL SH IELO

JOOK HOT PLAT[
BAYARD- ALPERT
VAC~ CACE
Figure 1. Heat Leak Test Facility for MLI measurements below SDK
METHOD OF OPERATION
Test samples are installed with the HLTF 1800 from the orientation shown in
Figure 1. During initial pumpdown, heater tapes on the vacuum shell are powered to
heat the apparatus and accelerate system outgassing. The system is evacuated to 1
micron, rolled into operating position, and then backfilled with dry nitrogen gas. The
system is then evacuated a second time, the heater tapes turned off, and the vessels
filled with cryogens. This evacuation process occurs over 3 working days. The first
data points are taken under "good" insulating vacuum. Subsequent data points are
taken with the cold plate held near 20K and the vacuum pump throttled to create
"degraded" vacuum conditions. Finally, the vacuum is opened to the pump and the
20K data point repeated to ensure that no shift occurred during degraded operation.
MEASUREMENT RESULTS
Table 1 outlines the systems evaluated during this program. The first measurement
with aluminum tape on both boundary surfaces simulated a cryostat condition where
no MLI was used. The second system modeled the MLI system configuration currently
specified for the sse 50 mm cryostat.5 This configuration consists of 10 reflective
layers each separated by three spacer layers of spunbonded polyester mat. The
reflective layers are 0.025 mm thick Mylar film aluminized on both sides to a nominal
coating thickness of 600 angstroms. The triple spacer layers serves to maintain low
237
layer density, which is desirable at low temperatures where solid conduction is a
dominant mode of heat transfer. The remaining systems were included to study the
performance impact of blanket thickness, sewn seams, and aluminum coating thickness.
Table 1. Summary of insulation schemes evaluated near 10K

Apparent
Number of Layer Layer Heat Thermal Insulating
Test DAM Layers Height Dens~y Rux Conductivity Vacuum
No System Description (N) (em) (N/em) (W/m 2 ) (flW/em•K) (torr)
3M No. 425 aluminum tape
T1 no MLI installed -- --- - 0.057 - 3.57E-06
10 layers DAM, 600 angstroms/side

T2 triple spunbonded polyester spacers 10 0.635 15.7 0.071 0.079 2.83E-06
stepped seam / stepped butt joint
straight seam I stepped butt joint
straight seam I stepped joint
straight seam I stepped joint
Warm Boundary Cond~ions: T =8OK; Surface Rnish =3M No. 425 aluminum tape
The data in Table 1 correspond to steady-state conditions at the noted vacuum

levels. These points were selected for the table based on insulating vacuum levels that
were closest to the cryostat design specification: limit radiant heat flux into the 20K
shield to 0.093 W 1m2 at 10-6 torr. The heat flux values may appear higher than those
found elsewhere in the literature for MLI systems near 4K. 6 This is due in part to the
insulating vacuum at which this data was taken; gas conduction is a larger contributor
at 10-6 torr than at pressures of 10-7 torr and below.
Thermal performance of the three configurations (TI,T3,T4) that incorporated 600
angstrom DAM material is compared in Figures 2 and 3. At low pressures,
performance is driven by the sewn seam geometry; the stepped seam geometry offered
the lowest heat flux, and was the only MLI system below design budget. As the
pressure increases, the impact of the seam becomes less significant while the effect of
blanket thickness becomes a factor. These issues are discussed in subsequent sections.
10
C'J
<
• 10 Iyrs DAM/stepped seam
E ---0-- 10 Iyrs DAM/straight seam
...........
~ 1 ...... -:1~
::.:: • 5 Jyrs DAM/straight seam
=
C\.1
.......
0 ~
>< 0.1
r-
;::l
p;:::
.......
""
Q)
::c:
0.01
1.00E-07 1.00E-06 l.OOE-OS 1.00E-04 1.00E-03
Vacuum (torr)
Figure 2. Heat flux through MLI systems near 10K
238
10.000
• 10 Iyrs DAM/stepped seam
--0-- 10 Iyrs DAM/straight seam

~ t;::: ...
• 5 Iyrs DAM/straight seam
h V
0.010
1.00E-0? 1.00E-06 1.00E-OS 1.00E-04 1.00E-03
Vacuum (torr)
Figure 3. Apparent thermal conductivity of MLI systems near 20K
Effect of Sewn Seam on Thermal Performance
Sewn seams are used on all sse MLI blankets to hold the many layers together
during handling and cryostat installation. 2,5 As radiant heat transfer falls off rapidly
below 80K, solid conduction between MLI layers becomes a more significant
contributor to overall heat transfer. Thus, to study whether the conduction
contribution through a seam at 20K has a significant impact on overall performance,
lO-layer blankets with straight and stepped seams (T4 and T2) were evaluated and
compared. The straight seam was sewn through all 10 layers with minimal stitch
tension; the stepped seam was sewn through 5 layers, offset by 7.6 em, and then sewn
through the remaining 5 layers. Measurement data shown in Figure 4 illustrates that
the stepped seam blanket has a lower heat flux to 20K at low pressure; the heat flux
through the straight seam exceeds the design budget. As pressure increases, the
performance of both blankets is driven by gas conduction with the seam contribution
negligible by comparison. As a result of this comparison, the design of the sse dipole
cryostat 20K MLI blanket will continue to incorporate stepped sewn seams.
10 F
N( F= • stepped sewn
f=
r-
seam
8
---
f-
A Straight sewn .-.I ....
~ 1
:><::
t= seam
~
_
0
N
.....
0
;.<
0.1
... ....
~ar
E
.....ro
~
Cl.l
::;J::
0.01
1.00E-0? 1.00E-06 1.00E-OS 1.00E-04 1.00E-03
Vacuum (torr)
Figure 4. Effect of sewn seams on MLI blanket performance
239
Effect of Blanket Thickness of Thermal Performance
Since radiation falls off rapidly below 80K, the number of reflective layers required
to limit radiant heat transfer becomes less. In fact, if small heat fluxes are attainable
between bare low ·emissivity surfaces, one might argue that the number of reflective
layers is rather insignificant at these temperatures from a radiant heat transfer
standpoint. The number of layers then become significant primarily during transient
or upset conditions. To study this, a comparison was made between a 5-layer and 10-
layer MLI blanket of like geometry and layer density (T3 and T4). The singular
difference between test samples was overall thickness due to number of layers. Each
blanket was comprised of DAM sheets with a coating thickness of 600 angstroms per
side. Figure 5 shows that the performance of both blankets is comparable at low
pressures, while heat flux through the 5-layer blanket is approximately 35% higher than
through the lO-layer blanket as the pressure increases to 104 torr. This data is in
agreement with the theory that reflective sheets serve as gas conduction barriers, and
supports the continued use of lO-layer blankets for the cryostat 20K MLI system.
10.000
~
= --S-layer
N(
S == blanket
---
-
~ 1.000
---+-- 10-layer ..... ~ ....
==
~ ~ blanket
0
C-..:l
0 ~
-'
>< 10-
;:::l 0.100
~
-'
""
Cl.l
;::I::
0.010
1.00E-07 1.00E-06 1.00E-OS 1.00E-04 1.00E-03
Vacuum (torr)
Figure 5. Effect of blanket thickness on overall thermal performance
Effect of Aluminum Coating Thickness on Thermal Performance
It has been noted that the addition of MLI at temperatures near 4K actually
caused an increase in the rate of heat transfer. 3 One explanation is based on the
increased transmissivity of thin aluminum films at very low temperatures. 5 Based on
this theoretical prediction, the 20K MLI system for the sse cryostat is comprised of
reflective sheets of DAM with aluminum coating thicknesses of 600 angstroms per side.
By comparison, the 80K MLI system employs DAM sheets with aluminum coatings of
350 angstrom per side. To quantify the effect of aluminum coating thickness on MLI
performance at 20K, measurements were made on two blankets of like geometry (T4
and T5). Each blanket was comprised of 10 layers of DAM with three layers of
spunbonded polyester mat separating each reflective pair. The blanket layers were
fastened together with parallel straight sewn seams. The singular difference was that
one blanket was comprised of 600 angstrom DAM; the second was comprised of 350
angstrom DAM. Figure 6 illustrates that an increase in aluminum coating thickness
from 350 angstroms to 600 angstroms had no significant effect on overall thermal
performance. As a result of this measurement, 350 angstrom DAM can be used in all
MLI systems in the sse cryostat.
240
10
N
<
S
• 600
angstroms/side
...."
-
~
------- 1 --0--- 350
:::.:::
a == angstroms/side
C'\1
.......
0 r---'
><
;::l 0.1 ==
E;;:;
.......
co
<J.l
::t::
0.01
1.00E-07 1.00E-06 l.00E-05 1.00E-04 1.00E-03
Vacuum (torr)
Figure 6. Effect of aluminum coating thickness on MLI performance
CONCLUSION
A number of insulation schemes have been evaluated between 80K and 20K to
optimize the thermal insulation system for the sse dipole cryostat. Experimental
results verify that the 1O-layer MLI blanket currently specified for the sse dipole
cryostat 20K shield meets the design budget at an insulating vacuum of 10~ torr. From
the data, several other key points are noted:
• The region near 20K is very sensitive to residual gas conduction.

Insulating vacuum must be below 10-7 torr before the contribution of gas
conduction to insulation performance becomes negligible.
• At vacuum levels near 1O~ torr, the number of reflective layers is not as
important as how the blanket is configured. Blankets fabricated with
similar layer densities and fastening methods had the same performance
even though one blanket had double the number of reflective layers.
• The number of reflective layers becomes important when insulating

vacuum increases above 1O~ torr. The additional layers limit heat
transfer by impeding residual gas conduction.
• Sewn seams impact the overall thermal performance of MLI blankets

operating near 20K. The stepped sewn seam geometry specified for the
sse 20K MLI system allows the many layers to be handled as a single
component while still meeting the thermal design requirements.
• The use of reflective films with 600 angstrom thick aluminum coatings
offer no significant improvement in performance over similar systems
comprised of films with 350 angstrom thick coatings. This will allow the
use of reflective films aluminized with coating thicknesses of 350
angstroms in all sse cryostat blankets without compromising system
performance.
241
ACKNOWLEDGEMENTS
The authors wish to thank M. Ruschman and R Kunzelman for their contributions
in the development of the measurement facility and their assistance in the acquisition
of this data. Gratitude is also expressed to D. Franks, R Pletzer, A. Yucel, and others
for countless discussions regarding the experimental results, and to M. McAshan, R.
Coombes, and E.G. Pewitt for their support of this work.
REFERENCES
1. T.H. Nicol, Design development for the 50 mm Superconducting Super Collider

dipole cryostat, in: "Supercollider 3," Plenum Press, New York (1991), p.
1029.
2. J.D. Gonczy, W.N. Boroski, and R.c. Niemann, Thermal performance

measurements of a 100 percent polyester MLI system for the
Superconducting Super Collider: Part II, Laboratory Results (300K-80K), in:
"Advances in Cryogenic Engineering," Vol. 35, Plenum Press, New York
(1989), p.497.
3. E.M.W. Leung, et aI., Techniques for reducing radiation heat transfer between
17K and 4.2K, in: "Advances in Cryogenic Engineering," Vol. 25, Plenum
Press, New York (1980), p. 489.
4. W.N. Boroski, RJ. Kunzelman, M.K. Ruschman, and c.J. Schoo, Design and
calibration of a test facility for MLI thermal performance measurements
below 80K, presented at the Cryogenic Engineering Conference, June 11-14,
1991, Huntsville, Alabama.
5. W.N. Boroski, T.H. Nicol, and c.J. Schoo, Design of the multilayer insulation
system for the Superconducting Super Collider 50 mm dipole cryostat, in:
"Supercollider 3," Plenum Press, New York (1991), p. 849.
6. I.E. Spradley, T.e. Nast, and D.J. Frank, Experimental studies of MLI systems
at very low boundary temperatures, in: "Advances in Cryogenic Engineering,"
Vol. 35, Plenum Press, New York (1989), p. 417.
242
A PROPOSED IR QUAD FOR THE SSC *
Shlomo Caspi, Clyde Taylor, and Alan Wandesforde
1 Cyclotron Road, MS 46-161

Berkeley, CA 94720
ABSTRACT
This note outlines a detailed magnetic design of a high-gradient quadrupole for the beam
interaction region of the SSC. The 58 mm bore, 2 layer magnet uses 36 strand cable identical
to the collider dipole magnet outer cable, thin collars, a close-fitting iron yoke, and a shell for
structural support. With a 1.3: 1 Cu/Sc ratio the quadrupole short sample gradient is 274 TIm
at 1.9 K and 209.7 TIm at 4.35 K with good field quality. Assembled with 7 mm collars, the
magnet is placed inside a four-segment iron yoke and prestressed with welded outer shell.
Prestress is maintained during cooldown by aluminum spacers placed between the segmented
iron yoke blocks. This paper describes various conceptual design details including coil
geometry, load line and margin, field uniformity and saturation effects.
INTRODUCTION
Two low beta quadrupole triplets are to be installed at the SSC interaction region. The
quadrupole closest to the interaction region (QL 1) is required to have a high gradient, excellent
field uniformity, and sufficient cooling to withstand beams with 10 33 luminosity . For good field
uniformity, the bore diameter is compared to the current 40 mm collider quadrupole bore
diameter. Increasing the bore size to 58 mm has a direct effect on improving the field uniformity
as compared to the 40 mm bore quadrupole but will decrease the gradient. In order to maximize
the gradient, Jc is increased by cooling the magnet to 1.9 K and reducing the copper to
superconductor ratio. In addition, the contribution of the iron is maximized by minimizingthe
collar size. In this paper we outline a magnetic design and propose a construction and assembly
procedure based on a thin collar.
THE MAGNET
The main features of the magnet include a double layer "Cosine 29" winding, a set of four-
way collars and iron yoke, four aluminum spacers and a shell (Figure 1). The cable used in both
layers is the same size as the 36 strand cable used in the outer layer of the collider dipole
(Figure 2). The NbTi strand diameter is 0.648 mm (0.0255") with a Cu/Sc ratio of 1.8:1, we
Sltpercollider 4, Edited by J. Nonte 243

Figure 1 Low beta quad cross-section
0.Ql78 (0.0070] 1.1680 (0.4598] 0.0102 (0.0040]
0.0648 [0.0255)
0.1054 (0.0415)
Figure 2 a 36 strand cable used in layers 1 and 2
244
consider however, the possibility of using 1.3: 1 strand. The coil has a single wedge in each layer
with 13 turns and 20 turns in the inner and outer coils respectively (Figure 3). The outer layer
pole turn is intentionally left one conductor thickness away from the inner layer pole; this
simplifies the transition between layers during winding (in the end region the outer layer gains
a turn around the pole and the inner layer loses one). Special cooling channels were introduced
between layers and along the pole. These channels are thermally linked to the refrigerant
located inside 16 circular holes in the yoke introduced specifically to remove heat.
Figure 3 Quadrant showing cooling channels
COIL-COLLAR-YOKE AND SHELL ASSEMBLY
The use of thin collars and aluminum spacers to control the yoke gap assembly is a unique
feature of the design that has previously been successfully tested in a dipole magnet (magnet
D 19 - to be published). The entire coil prestress comes directly from the outer shell and both
prestress level and cooldown loss are controlled by the four aluminum spacers. The collars
provide precision conductor location andminimum prestress as an assembly aide. A 7 mm
thick stainless steel collar is sufficient to hold the coils at assembly with a prestress of 1500
psi. The key and keyway are designed to unload during cooldown. The collared coil is placed
inside an iron yoke which is split four ways. Special tabs are introduced along the 45 degree
poles of the collar and yoke to aid in alignment. Four aluminum spacers are placed in slots along
the yoke and a stainless steel shell is then pressed around the assembly and welded. The shell
is an integral part of the magnet helium container as in the collider dipoles and quadrupoles.
As the shell exerts force on the assembly the coil reaches its predetermined prestress, and a
small well controlled gap (about 5 mils) is maintained by the aluminum spacers along the
midplane between the four-way yoke. As the magnet cools and reduces in size so do the
aluminum bars; the gap closes, and the coil prestress is maintained approximately constant.
245
MAGNETIC DESIGN
The coil was designed to give negligible multipoles. Theoretically the systematic bs,b9 ,
and b 13 are 4.5x1o-", -1.7x1o-2, and -1.7x1o-" units respectively (one unit is 10-" of the
quadrupole field at r= 10 mm). The introduction ofiron saturation causes bs to vary by 0.14 units
and b9 to vary by less than 0.004 units (Figure 4). The initial transfer function of 3.03
(G/crn/A) is reduced with excitation by 7.6% at 10000 A (Figure 5). A flux plot at 1=11000 A
is shown in Figure 6.
0.14
0.12
0.10
.~ 0.08
::I
0.06
0.04
0.02
2000 4000 6000 8000 10000

I (A)
Figure 4
The effect of iron saturation on bS
The load line and short sample performance are included in Figures 7 and 8 for the inner
layer. Results for the outer layer are not included here as it has a higher margin. The short sample
curves werecomputedat4.35 K and 1.9 Kassuming Jc=2750(Almm2) at4.2K and 5 tesla [1].
Table 1 summarizes the short sample performance and Table 2 assumes a set of values for a 15%
margin. With a variation in temperature and Cu/sc ratio an operating gradient between 169
(Tim) and 232 (Tim) can be expected with a margin of 15%. The stored energy at 5755 A and
7709 A is 66 (kJ/m) and 112 (kJ/m); the respective inductance is 3.95 (mH!meter) and 3.78
(mH!meter).
246
3.00
2.95
~
~
8
.2"
V 2.90
-
"
:>
u..
~
~
I-
2.85
2.80
2.75 '--_ _--L_ _ _ ...I...-_~___' ___ _'__~_'_____'
o 2000 4000 6000 8000 10000

I(A)
Figure 5
The effect of iron saturation on the transfer function.
Figure 6
Flux plot at 11000. A.
247
9
/
Short Sample 1.9 K
8
,, 1.8: 1 cu/sc
13 1.3: I cu/sc
~
E
a:l
,,
7
6
Short Sample 4.35 K
Inner Layer Load Line
5 ~~~~~~~~~~~~~~~~~~~~
5000 6000 7000 8000 9000 10000 11000

I (A)
Figure 7 Short sample and load line
248
300
290
Shon Sample Gradient
280
270 1.9 K
260
250
240
.-. 230 Shon Sample Gradient

.§.
C 220
0
0
210
200
190
180 1.8: Icu/sc

1.3: I cu/sc
170
160
150
140
5000 6000 7000 8000 9000 10000 11000
I (A )
Figure 8 The gradient as a function of current
Table 1 Shon sample values for the IR quad
Operating
CuJSc 1.8:1 NbTi Cu/Sc 1.3:1 NbTi
Temperature
To(K) Iss (A) Bss (TJ Goss (Tim) Iss (A) Bss (T) Goss (Tim)
4.35 6835 6.915 198.4 7273 7.31 209.7
1.9 9208 9.05 259.5 9770 9.55 274
249
Table 2 Operating point with 15% margin
Operating
Cu/Sc 1.8:1 NbTi Cu/Sc 1.3:1 NbTi
Temperature
To(K) 10 (A) Go (TIm) 10 (A) Go (TIm)
4.35 5755 169 6098 178
1.9 7709 221 8182 232
He IT HEAT TRANSFER
Heat transfer to He II along the magnet was calculated for two cases, one where the entire
heat is applied at one end, and the other where the heat is uniformly distributed along the
magnet. The mechanism of heat transfer is based on the Gorter-Mellink two fluid model, and
simple relations were used here based on Reference[2]:
End Heating
f.. T )0.294
q= 8.593 ( -1-
q=(w/cm2) ; T=K ; l=cm
Uniform Heating
f.. T)0.294
q = 13.286 ( -1-
Sixteen holes of 1.27 cm radius in the iron yoke and the 2 unused bus holes will remove
over 50 watts of uniform heating via axial heat conduction over a 16 m length with negligible
f..T along the length.
REFERENCES
[1] M.A. Green. Generation of the Jc, Hc, Tc surface for commercial superconductor using
reduced-state parameters. Lawrence Berkeley Laboratory LBL-24875, UC-406, April
1988.
[2] S. Caspi. Phenomenological relations concerning heat transfer to He II. Lawrence Berkeley
Laboratory LBID-726, SU-MAG-95, May 1983.
250
MAGNETIC FIELD MEASUREMENTS OF
FERMILAB/GENERAL DYNAMICS BUILT
FULL SCALE SSC COLLIDER DIPOLE MAGNETS*
S. De1champsl, M. Bleadon l , R. Bossert l , J. Carsonl, A. Devred2,

J. DiMarc02, S. Gourlayl, R. Hanft l , W. Koska l , M. Kuchnir l ,
J. Kuzminski 2, M.J. LammI, P.O. Mazur l , A. Mokhtarani l ,
T. Ogitsu 2, D. Orrisl, M. Puglisi2, J. Strait l , J.C. Tompkins 2,
M. Wake l , Y. Yu2, Y. Zha02, and H. Zheng2
IFenni National Accelerator Laboratory
P.O. Box 500, Batavia, IL 60510 USA
2Superconducting Super Collider Laboratory
2550 Beckleymeade Avenue, Dallas, TX 75237 USA
INTRODUCTION
This paper presents preliminary results of magnetic field measurements made on a
series of 50 mm aperture 15 m long SSC collider dipole magnets designed and manufactured
at Fenni National Accelerator Laboratory (Fennilab) for use in the Superconducting Super
Collider Laboratory (SSCL) Accelerator System String Test. The magnets were assembled
by Fennilab and General Dynamics personnel, and were tested at the Magnet Test Facility
(MTF) at Fennilab. Measurements of the dipole field angle, dipole field strength, and field
shape parameters at various stages in magnet construction and testing are described.
DIPOLE FIELD ANGLE

SSC collider dipole magnets are required to have their average field angles aligned to
within ±1 mrad of the vertical. A system to measure the dipole field angle was designed and
built at Fennilab (Kuchnir and Schmidt, 1988). The field angle probe contains a small
pennanent bar magnet mounted on jeweled gimbals attached to a plastic frame. The bar
magnet is free to rotate within the frame, and the frame is free to rotate about the longitudinal
axis of the magnet whose field angle is being measured. The axis of the bar magnet oscillates
about the local dipole field direction, and the azimuthal oscillations of the frame are converted
to an electronic signal using the varying resistance of an electrolytic bubble level mounted to
the frame.
*This work was supported by the United States Department of Energy.

The magnet current used for all field angle measurements is -7 - 9 A, corresponding to
a central dipole field of -70 - 90 Gauss. Readings are taken at 76.2 mm (3") intervals along
the length of each magnet. Two complete longitudinal sweeps of the magnet are made for
each measurement, with the probe facing towards the lead end in one sweep and towards the
non-lead end in the other. Taking half of the difference of corresponding readings from the
two sweeps gives the relative dipole field angle signature of the magnet as a function of
position. Before and after each measurement, the field angle probe is calibrated in the range
±9.696 mrad.
Table 1. Preliminary Results for Field Angle Probe Measurements

on 50 mm Aperture SSC Collider Dipole Magnets
Location of Mean Value Slandani Range of
Measurement of Angle Deviation Angle about
(about the of Angle Mean Value
(see text) Vertical}
(mrad) (mrad) (rmad)
DCA311 ICB-2 0.87 2.82 13.06 ± .57
MlF-I 1.07 2.58 13.49
MlF-2 0.75 1.89 9.89
DCA312 ICB-l -0.28 2.61 9.46

ICB-2 0.00 2.25 9.75
MlF-l 0.42 1.88 9.51
MlF-2 1.37 1.12 5.42
DCA313 ICB-l 0.50 1.42 7.11

ICB-2 -0.20 1.68 8.75
MlF-l 0.76 1.68 8.38
MlF-2 1.30 1.70 7.93
DCA314 ICB-l 0.15 3.17 11.89

ICB-2 -0.11 2.04 8.38
MlF-l 0.54 2.10 8.35
MlF-2 2.00 1.62 7.52
DCA315 ICB-l -4.02 2.31 11.06

ICB-2 1.04 1.57 9.91
MFr-l 1.11 1.30 8.04
MlF-2 1.58 1.09 6.01
DCA316 ICB-l 1.44 1.92 8.04

ICB-2 1.24 1.60 7.50
MlF-l 2.14 1.68 7.77
The mean and standard deviation of the dipole field angle with respect to the vertical,
and the peak-to-peak range of field angles about the mean, are shown in Table 1. Positive
values of average field angle correspond to counter-clockwise rotation viewed from the lead
end. Measurements labeled ICB-1 are done in a set of roller cradles at the Fermilab Industrial
Center Building (ICB). ICB-2 indicates a measurement made at ICB after the cold mass has
been installed in the cryostat. M1F-1 and M1F-2 indicate measurements on the M1F test
stands before and after testing at 4.3 K.
The uncertainties in the measurement of field angle at a single position and in the mean
angle about the vertical were estimated to be ±o.4 mrad and ±o.03 mrad respectively for
measurements made on 40 mm aperture sse collider dipoles built at Brookhaven National
Laboratory and Fermilab (DiMarco, J., Kuchnir, M., and Yu, Y., 1991). The values in
Table 1 for the standard deviation of field angle about the mean are upper limits, since
contributions from random measurement errors, which can in principle be evaluated from
existing data, have not been removed. A more complete report on these measurements is
planned for a conference in the near future (Kuchnir, 1992).
252
Detailed examination of the data reveals evidence of changes in the field angle profile
between data taken before and after cryostating at ICB, and before and after cold testing at
MTF. The latter changes are implied by the changes in the standard deviation of the field
angle between MTF-l and MTF-2 measurements shown in Table 1. There is no evidence of
such changes between cryostated data at ICB and data taken at MTF before cold-testing l .
ABSOLUTE FIELD STRENGTH

Measurements of the transfer function (ratio of the central field strength to the magnet
current) are made on each magnet at MTF. A system combining a Hall Probe and and NMR
magnetic spectrometer system (Borer and Fremont, 1977; DiMarco and Radusewicz, 1991)
is used for these measurements. The Hall probe is used in the end sections of the magnet,
where the NMR system does not give a stable signal due to the large spatial field gradient.
Both NMR and Hall probe readings are taken in 304.8 mm (1 foot) overlap sections just
inside the end sections, to normalize the Hall probe readings. The NMR system is used over
the entire central section of the magnet Readings are taken every 6.35 mm (1/4") in the ends
and overlap sections, and every 25.4 mm (1 ") in the central section of the magnet. For each
NMR reading, the current and field are measured simultaneously.
Before a longitudinal sweep of the Hall ProbelNMR system, the magnet is cleared of all
persistent currents by ramping the current until a quench occurs above 5000 A. The current
is then ramped at 4 A/sec from 0 A to 6.5 kA, where it is held for 5 minutes, reduced to 115
A at 4 Nsec, where it is held for 2 minutes, and finally ramped at 4 Nsec to 1900 A, where
the sweep is performed.
The straight section transfer function average value TFave is determined by taking the
average over the data in the center (NMR-only) section, leaving out ±76.2 mm (±3") about
the strain gauge pack centers, to avoid including perturbations in the transfer function caused
by the permeability of the strain gauge block materials. Values for TFave for magnets
DCA311 through DCA316 and DCA319 are shown in Table 2.
Table 2. Transfer Function and Magnetic Length for

50 mm Aperture SSC Collider Dipoles
Magnet Transfer Magnetic
Function (TFave) Length <Lm)
(Tesla/leA) (meters)
DCA311 l.0450 14.863
DCA312 l.0454 14.867
DCA313 l.0453 14.860
DCA314 l.0453 14.862
DCA315 l.0453 14.867
DCA316 l.0452 14.866
DCA319 l.0449 14.868
Another parameter of interest, the "magnetic length" Lrn , is obtained from the formula
(DiMarco and Radusewicz, 1991)
Lrn = dTF dl)lead end + TFave*Lss + dTF dl)non-Iead end
TFave
where the integrals are obtained from the Hall Probe data taken at the lead and non-lead end
sections and Lss is the length of the straight section. Values for Lrn are given in Table 2. The
estimated uncertainty in Lrn at this time is ± 2 mm.
lThe mean angle about the vertical may change between IeB and MTF due to changes in mounting of the
cryostat
253
MAGNETIC FIELD HARMONICS
The magnetic field may be expressed as an expansion in multi-pole fields, each having
some strength relative to the fundamental dipole field strength Bo. The hannonic coefficients
bn and an may be defined by
By = Bo *[ 1+ ~ [bn cos(ne) - an sin(ne)] (r / ro)n ] n = 1, ... ,00
In this expression for the vertical component of the magnetic field in the magnet
aperture, the bn and an are "normal" and "skew" harmonic coefficients respectively. Note
that in the convention represented by this expression, bn and an are the normal and skew
(2n+2)-pole coefficients. For example, b i and al are the normal and skew quadrupole
coefficients, b2 is the normal sextupole coefficient, and so on. This differs from the
convention used more frequently in Europe. The radius ro is called the reference radius, and
has the value 10 mm for SSC collider dipoles.
To date, the device used to measure the hannonics of the 50 mm aperture SSC magnets
at Fermilab has been the B2 mole (Ganetis et al., 1987), designed and built at Brookhaven
National Laboratory. The active part of the probe is 609.6 mm (2 feet) in length and consists
of a rotating shaft with two dipole coils and a third "tangential" coil sensitive to the dipole and
higher order hannonics. The dipole coils are used to remove the dipole field component from
the tangential coil signal, so that only contributions from the higher order hannonics remain.
The hannonics of each magnet are measured after collar-keying, and again at room
temperature after the fully assembled magnet has been mounted on a test stand at MTF. The
hannonics are next measured at 4.3 K, then at room temperature after wann-up, again at 4.3
K during a second thermal cycle, and finally at room temperature following the final warm-
up of the magnet. Each measurement consists of a longitudinal sweep of the magnet aperture
with the mole center positioned at 609.6 mm (2 foot) intervals.
Collared Coil Harmonics

Table 3a shows the collared coil hannonics averaged over the straight section of the
coils for magnets DCA311 through DCA319. (The odd normal hannonics and skew
hannonics for magnet DCA316 are not available at this time due to an error during data taking
on this magnet.) The mean, standard deviation about the mean, and r.m.s. about zero for the
hannonics are given in Table 3b. The estimated uncertainty in the mean due to measurement
error is also shown. All harmonics in this paper are given in prime units at r02.
Hannonic mixing will occur if the mole device is off-center in the magnet aperture. An
attempt has been made to remove such mixing effects from the hannonics given in this paper
by calculating the horizontal and vertical centering errors ~x and ~y of the mole using an
expression for a7 and b7, both expected to be zero, in terms of as-IO, bg-IO, and ~x and ~y.
The expression is used iteratively until stable values of ~x and ~y are obtained. If the iteration
does not converge, a simpler correction based on either ag and bg (for collared coil data) or
alO and blO (for MTF data) is used to zero a7 and h7 and determine ~x and ~y.
The observed mean value of the normal sextupole 1>2 is -3.11, which is close to the
expected value -3.8. The average value of b4 0.42 differs significantly from the expected .07
and is still not well understood. The bg and blO harmonics were built into the coil cross
section to allow the removal of hannonic mixing due to imperfect centering of the mole
within the aperture. Of the remaining hannonics only b6, a2, and ~ have average values not
consistent with zero (Mean ~ - 3crNN, with N = 9 in this case.). Like b4, the non-zero b6 is
allowed for coils with perfect dipole symmetry. The small average a2 and ~ must be due to
some asymmetry in the coils, which were all wound and cured on the same tooling.
20ne prime unit is one part in 10,000 of the fundamental field, in this case the dipole. So for example,
one prime unit at ro of nonnal sextupole means that the sextupole field amplitude is .0001 times the dipole
amplitude 10 mm from the center of the aperture. (Multi-pole fields have constant amplitude on circles of
constant radius.)
254
Table 3a. Collared Coil Harmonics for DCA311 - DCA319
(Expected values for at. a2, and a3 shown in parentheses)
311 312 313 314 315 316 317 318 319

b2 -2.76 -3.15 -2.19 -3.01 -2.98 -2.52 -3.94 -3.52 -3.33
b4 0.46 0.40 0.35 0.38 0.37 0.40 0.50 0.46 0.42
b6 -0.06 -0.05 -0.05 -0.06· -0.07 -0.06 -0.07 -0.06 -0.07
b8 0.06 0.06 0.07 0.07 0.06 0.06 0.06 0.06 0.06
blO 0.02 0.02 0.02 0.02 0.03 0.02 0.02 0.02 0.02
bl 0.05 <.01 -0.39 -0.55 0.09 0.46 -0.03 -0.32

b3 <.01 -0.04 -0.04 -0.07 0.04 0.04 0.06 0.05
b5 <.01 -0.01 <.01 -0.01 <.01 <.01 <01 <.01
b7 <.01 <.01 <.01 <.01 <.01 <.01 <.01 <.01
b9 <.01 <.01 <.01 <.01 <.01 <.01 <.01 <.01
al 0.60 0.56 0.69 -0.10 -0.31 -0.15 -0.51 -0.32

(0.79) (0.67) (-0.42) (-0.62) (-0.10) (-0.28) (-0.30) (-0.35)
a2 -0.25 -0.12 -0.21 -0.08 -0.14 0.04 -0.16 -0.42
(-0.12) (-0.12) (-0.07) L-O.07) (-0.11) (-O.21} (-0.13) (-0.09)
a3 -0.07 0.10 0.06 -0.06 0.06 -0.13 0.17 -0.03
(0.03) (0.03) (-0.02) (-0.03) (0.00) (-0.01) (-0.01) (-0.02)
a4 -0.04 -0.05 -0.05 -0.06 -0.06 -0.02 -0.06 -0.08
as -0.01 -0.01 <.01 0.02 0.01 0.01 0.02 -0.01
a6 <.01 <.01 -0.01 <.01 <.01 <.01 <.01 <.01
a7 <.01 <.01 <.01 <.01 <.01 <.01 <.01 <.01
as <.01 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01
a9 <.01 <.01 <.01 <.01 <.01 <.01 <.01 <.01
alO <.01 <.01 <.10 <.01 <.01 <.01 <.01 <.01
Table 3b. Collared Coil Harmonics for

50 mm SSC Collider Dipoles
Mean cr r.m.s. ~Mean
(about mean) (about zero) (estimated
measurement
uncertainty)
b2 -3.11 0.41 3.09 0.02
b4 0.42 0.04 0.41 0.01
b6 -0.06 0.01 0.06 <.01
b8 0.06 <.01 0.06 <.01
blO 0.02 <.01 0.02 <.01
b1 -0.09 0.30 0.31 0.07

b3 0.01 0.04 0.04 0.01
b5 <.01 -- - --
b7 <.01 -- - -
b9 <.01 -- - -
a1 0.06 0.45 0.40 0.07
a2 -0.17 0.13 0.20 0.02
a3 0.01 0.10 0.07 0.02
a4 -0.05 0.02 0.05 om
as <.01 - - -
at) <.01 -- -- --
a7 <.01 - - -
as -0.01 <.01 0.01 <.01
a9
alO
<.01
<.01
--
-
--
--
--
255
An attempt has been made (Mokhtarani, 1991) to calculate the expected values of aI,
a2, and a3 based on known asymmetries in coil manufacture and pole shimming. The
predicted values for these harmonics are shown in Table 3a in parentheses below the
measured values. The correlation between predicted and measured values of al is fairly
good, with the exceptions ofDCA313 and DCA314. For a2, with the exception of DCA317
the measured harmonic has the expected sign, but the magnitude is generally greater than
predicted. For a3 there is no noticeable correlation, and the measured magnitude is always
larger than that predicted by the model. Calculations of 34 gave values much smaller than the
measured values.
Harmonics of Fully Assembled Magnets
Room Temperature Measurements. Table 4 shows the harmonics measured at
MTF before testing at 4.3 K. The mean, standard deviation, and r.m.s. about zero are
shown for each harmonic. As before, the harmonics have been averaged over the straight
section of the magnet. Only the allowed harmonics (even b's), a2, and 34 show evidence of a
non-zero central value.
Table 4. Room Temperature Harmonics for DCA311 - DCA315

(prime units at 10 mm)
311 312 313 314 315 Mean a RMS

b2 1.80 2.47 2.54 2.27 2.41 2.30 0.26 1.83
b4 0.35 0.32 0.30 0.30 0.30 0.31 0.02 0.25
b6 -0.04 -0.03 -0.03 -0.04 -0.04 -0.03 0.Q1 0.03
b8 0.05 0.05 0.06 0.05 0.05 0.05 <.01 0.04
bl0 0.02 0.01 0.01 0.02 0.02 0.02 <.01 0.01
bl -0.20 -0.06 -0.36 -0.33 0.18 -0.15 0.20 0.20

b3 0.02 -0.03 -0.03 -0.03 0.03 -0.01 0.03 0.02
b5 <.01 0.01 0.01 -0.01 <.01 <.01 -- --
b7 <.01 <.01 <.01 <.01 <.01 <.01 -- -
b9 <.01 <.01 <.01 <.01 <.01 <.01 -- -
al 0.63 0.29 1.77 -0.18 -0.16 0.47 0.72 0.68
a2 -0.20 -0.03 -0.05 -0.04 -0.05 -0.07 0.06 0.08
a3 -0.06 0.Q7 0.13 -0.04 0.03 0.03 0.Q7 0.06
a4 -0.01 <.01 <.01 -0.02 -0.01 -0.01 <.01 0.01
as -0.02 -0.01 <.01 0.01 0.01 <.01 -- --
a6 <.01 <.01 -0.01 -0.01 <.01 -0.01 <.01 0.01
a7 <.01 <.01 0.01 <.01 <.01 <.01 -- --
as <.01 0.01 0.01 0.01 <.01 <.01 -- --
fiJ <.01 <.01 <.01 <.01 <.01 <.01 -- --
alO <.01 <.01 <.01 <.01 <.01 <.01 -- --
Measurements at 4.3 K. For longitudinal sweeps of the mole at 4.3 K, the magnet
is prepared in the same manner as for Hall Probe/NMR sweeps, except that the final current
is 2000 A instead of 1900 A. Table 5 shows the values for harmonics measured at 4.3 K.
The mean, standard deviation, and R.M.S. about zero are given for each harmonic, and
where available, the SSC "systematic" and "a" values are given in the f'mal two columns3 •
Magnetization effects cause the harmonics measured at 2000 A to differ by some
amount from their geometric values4 • A correction is obtained for each harmonic from the
hysteresis loop of the harmonic during a sawtooth ramp between 115 A and 6500 A. Up and
down ramp values of the harmonic are averaged at current values between 2000 and 3000 A,
3The average value of a harmonic over the entire ensemble of magnets in the sse is required to be less
than the sse systematic value, and any given magnet must have harmonics within 3a of the mean.
4nte geometric" value of a harmonic coefficient is the value it would have if only the geometry of the
II
coils and yoke iron and the disttibution of transpon current in the coils were determining the field shape.
256
to calculate the geometric value of the harmonic. The difference between the measured
harmonic on the up ramp at 2000 A and the geometric hannonic value is then the correction
for the harmonic in question. The correction is only large (greater than .05 units) for the
normal sextupole 1>2, for which about 0.3 units should be added to the uncorrected value. 5
The measured hysteresis values for b2 are consistent with calculations based on the
critical current properties of the cable. For magnets DCA311, DCA313, and DCA316, the
predicted b4 agrees well with the observed value. For DCA312 and DCA314, the sign of the
b4 hysteresis is the opposite of that expected and the amplitude is much larger than expected.
For DCA315, the up and down ramp branches ofb4 cross, which is not expected.Two of the
magnets, DCA314 and DCA315, have shown much larger hysteresis than expected in the
skew quadrupole a 1.
Table 5. 4.3 K Harmonics for DCA311 - DCA315

(Uncorrected for Hysteresis Effects)
311 312 313 314 315 ~ 0 RMS sse sse
sys. 0
b2 1.00 1.90 1.83 1.69 1.52 1.59 0.32 1.62 0.80 1.15
b4 0.31 0.30 0.27 0.28 0.29 0.29 0.01 0.29 0.08 0.22
b6 -0.05 -0.04 -0.04 -0.04 -0.05 -0.04 0.01 0.04 0.013 0.018
b8 0.05 0.05 0.05 0.05 0.05 0.05 <.01 0.05 0.01 0.008
b10 0.02 0.02 0.Q2 0.02 0.02 0.02 <.01 0.02 na na
bl 0.06 -0.01 -0.24 -0.36 0.06 -0.10 0.17 0.20 0.04 0.50
b3 <.01 -0.04 -0.03 -0.05 0.04 -0.02 0.03 0.03 0.026 0.16
b5 <.01 <.01 <.01 -0.01 <.01 <.01 -- -- 0.005 0.017
b7 <.01 <.01 <.01 <.01 <.01 <.01 -- -- 0.005 0.01
b9 <.01 <.01 <.01 <.01 <.01 <.01 -- -- na na
al 0.58 0.55 0.57 -0.22 -0.06 0.28 0.35 0.45 0.04 1.25
a2 -0.17 -0.05 -0.08 -0.01 -0.06 -0.07 0.05 0.09 0.32 0.35
a3 -0.07 0.04 0.05 -0.01 0.02 0.01 0.05 0.05 0.26 0.32
a4 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 <.01 0.01 0.01 0.05
a5 -0.01 <.01 <.01 0.02 0.01 <.01 -- -- 0.005 0.05
a6 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 <.01 0.01 0.005 0.008
a7 <.01 <.01 <.01 <.01 <.01 <.01 -- -- 0.005 0.01
as <.01 0.01 0.01 0.01 0.01 0.01 -- -- 0.005 0.008
a9 <.01 <.01 <.01 <.01 <.01 <.01 -- -- na na
a10 <.01 <.01 <.01 <.01 <.01 <.01 -- -- na na
The allowed harmonics show systematic values inconsistent with zero. If the mean
value of b2 for subsequent magnets could be brought within the tolerance band of 0.80, then
all magnets tested so far would pass the 30 criterion. The same statement applies to b4 and
b6. Finally, b8 and bl0 are expected to be .043 and .015 for the chosen coil cross section,
and these harmonics will be tuned out in future cross section designs. Of the un-allowed
harmonics, only a2 and ~ show mean values inconsistent with zero, and both of these are
within the SSC specifications.
Warm-Cold Harmonics Correlations. Table 6 shows the mean and standard

deviation of the change in the harmonics between room temperature and 4.3 K. Also given
are the slopes and intercepts for linear fits to and correlation parameters for cold vs. warm
distributions of the harmonics. The correlations for bl and al are not good, so that it would
be difficult to correct cold values of these harmonics with changes made at room temperature.
However, as mentioned before, there is no evidence that the mean values for these harmonics
lie outside the SSC tolerances for magnets manufactured so far.
5The persistent currents affecting h2 decay slowly, so that b2 changes by -.06 units during a typical
longitudinal scan.
257
Table 6. Differences and Correlation Coefficients between
Room Temperature and 4.3 K Hannonics
<W-e> a(w - c) slope intercept correlation
l)31'3Il1eter
b2 -0.70 0.12 1.14 -1.02 0.95
b4 -0.03 0.01 0.59 0.10 0.94
b6 -0.01 0.00 0.72 -0.02 0.83
b8 -0.00 0.00 0.42 0.03 0.86
bl 0.05 0.15 0.64 -0.01 0.76
al -0.20 0.58 0.36 0.11 0.73
CONCLUSIONS
Measurements of dipole field angle, dipole field strength, and field quality have been
made on a set of full scale 50 mm aperture SSC collider dipole magnets built at Fermilab.
There is evidence of change in the field angle proIIle of magnets during assembly and cold
testing. The transfer functions and magnetic lengths of magnets measured so far appear to be
quite stable. Hannonics measurements show no evidence of systematic harmonics outside of
the SSC specifications, with the exception of several of the allowed multi-poles which may
be removed by iteration of the coil cross section design.
ACKNOWLEDGEMENTS
The authors wish to thank: the members of the Fermilab Superconducting Magnet
Construction Department and the Fermilab Magnet Test Facility. We would also like to thank:
Ms. Graciela Finstrom and Ms. Lauren Curry for their help with preparation of the paper and
accompanying poster display.
REFERENCES
Borer, K., Fremont, G., CERN 77-19 (1977).
DiMarco, J., Radusewicz, P., Summary of NMRlHall axial measurements on

4 cm aperture 17 m long dipole magnets, MD-TA-202, SSC Laboratory,
October 16, 1991.
DiMarco, J., Kuchnir, M., Yu, Y., Summary offield angle probe measurements on
4-cm aperture 17 m long dipole magnets, MD-TA-203, SSC Laboratory,
October 16, 1991.
Ganetis, G. et al., Field measuring probe for SSC magnets, BNL-39607, 1987.
Kuchnir, M., Schmidt, Ed. E., Measurements of magnetic field alignment,
IEEE Transactions on Magnets, 24, 950,1988.
Kuchnir, M. et al., SSC Collider dipole magnets field angle data, abstract submitted to
HEACC92, Hamburg, July, 1992. .
Mokhtarani, A., Effect of manufacturing errors on hannonics in 5 cm SSC magnets,

Fermilab Technical Support Section Internal Memo TS-SSC 91-038, Jan 7,1991.
Mokhtarani, A., Harmonics from left-right asymmetric midplane displacements,
Fermilab Technical Support Section Internal Memo TS-SCC 91-146, July 9,1991.
258
SSC COLLIDER QUADRUPOLE COLD MASS DESIGN
AND DEVELOPMENT
R.A. Farrell, F.S. Murray, P.A. Jonas, W.R. Mischler, and L. Blecher
Intermagnetics General Corporation (IOC)

P.O. Box 566
New Karner Road
Guilderland, NY 12084
ABSTRACT
Approximately 1,664 focussing and defocussing superconducting quadrupoles are

required for the two SSC collider rings. Collider quadrupole magnets (CQMs) must satisfy
stringent performance, reliability, life and low cost criteria. Performance requirements
include field uniformity, training, quench, tracking, thermal cycling and alignment. The
CQM cold mass design presented incorporates lessons IOC and Alsthom lntermagnetics
S.A. (AISA), our joint venture with GEC-Alsthom, learned in the design, development and
manufacture of 500 MRI, 160 high-field custom and 126 HERA quadrupole
superconducting magnets. This baseline design reflects careful quantitative assessment of
coil winding placement and collar material, evaluation of field uniformity and mechanical
performance of the magnet coil ends using 3-D modeling and analysis, and considers
tolerance and process variability.
Selected CQM cold mass design highlights and a proposed prototype development
program that allows incorporation of test feedback into the design to minimize risk are
detailed in this paper. This information may be helpful to SSCL in the design and
development of prototype CQM's.
INTRODUCTION
The IGC design combines CDM and CQM technology transfer from the National
Laboratories (LBL, Fermi, and BNL) and SSCL, European superconducting accelerator
and commercial magnet experience and original work. Highlights discussed include: (1)
Magnetic design which meets multipole and quench requirements; (2) Magnetic field
sensitivity analysis; (3) Iron yoke split 45° from the horizontal; (4) Enlarged end region;
(5) Beam tube mounting which does not contact the coils; and (6) Innovative power bus
expansion loop at the return end.

MAGNETIC DESIGN AND FIELD QUALITY
2-D Magnetic Design. The design was optimized using a combination of IGC
proprietary codes and the commercial software package TOSCA. The multipole field terms
are calculated directly from each strand of conductor using an exact analytical filament
solution developed at IGC. The contribution of the iron to the multipoles is calculated
using the finite element code TOSCA. The effect on higher-order moltipoles from the iron
are almost insignificant (Table 1). The field quality sensitivity to the iron yoke is
dominated by the effect on the quadrupole moment. The integrated quadrupole moment is
easily modified by changing the quadrupole length. Optimizing the design of the yoke for
material costs as well as dipole tracking does not have to affect the coil cross section.
To optimize coil positions a code was developed that uses 20 independent variables
to model the coils. Six of these variables describe the superconducting cables, and 14 their
placement in the quadrupole blocks. The outline dimensions of the IGC baseline design are
given in Figure 1. Y3 and Y4 were chosen as the most appropriate variables to eliminate
the b5 and b9 terms. These two variables effect the cutout in the collars, and also the shape
and placement of the copper wedges. The field quality produced by this coil (including
iron) configuration is shown in Table 2.
Table 1 Iron yoke contribution to Table 2 2-D field harmonics

moltipoIe terms.
Tenn Spec Baseline Design %ofSpec
TeslalMeter bl (TIm) - 209.6 -
Circular With Cuts With Cuts b5 (Units) 1.399 0.07 5
Multipole " Infinite " Infinite " Finite b9 (Units) 0.612 0 0
bl +13.81 +13.49 +13.36
bS +0.0002 -0.0004 .(l.OOO8
b9 0.00 0.00 0.00
2SO
Yoke Region End Region
-------.1:-------,! Effective
: Magnetic
- - Coil Only ! Length =
~ 150
- - - - Yoke Only
- - Coil + Yoke
! S.16m
fi
g'"
III
.5j
t
Integral g ell 1082 Tim
X23 --- ~ g Center Coil 196.24

C) SO 9 Center Yoke 13.36
Rl,YI R2,Y2 9 Center Total 209.60
XI4
Dimensions of the Coil Cross Section:

RI 2O.0mm Y3 2O.0Smm
Yl O.IOmm R4 2O.Omm
R2 4D.4mm Y4 10.42mm -SO ' -_ _ _---1_ _ _ _l-I-_ _ _.---J
Y2 O.IOmm XI4 O.OOmm 210 230 250 270
R3 4D.4mm X23 O.OOmm
Z Distance from Midplane, em
Figure 1 2-D baseline design coil Figure 2 Integrated quadrupole

positions moment
260
3-D End Region and Integrated Field Analyses. The end regions are
designed to minimize multipole terms while not allowing excessively large peak fields.
As the cable exits the straight section and begins to make the bend, the inner
conductor is separated and spaced out into three sections while the outer conductor
remains in two sections over the end region. The resulting peak field in this design is
approximately 5. 18T. Potential solutions to reduce this higher than desireable peak field
include spreading the end turns out, as well as modifications to the 2-D region. There are
four major contributors to the integrated performance; the straight section of conductor,
iron yoke, iron yoke end effects and conductor end region. The integrated multipole errors
are dominated by the b5 term (0.44 units, approximately 31 % of the allowed specification).
The combination of all other "designed in" multipoles is less than 0.05 units. Results of the
analysis are shown in Figure 2.
Conductor Position Sensitivity Analysis. Approximately 93% of the CQM

quadrupole moment is generated directly from the superconducting windings, mostly from
the straight sections. The major contributor to average field quality variability is the
positioning of the cable straight sections. A sensitivity analysis of field quality vs.
conductor placement was performed using an IGC developed computer code. Each of the
168 cables in a CQM cross section are modeled as a set of 15 infinitely long filaments. The
multipole moments are calculated directly from current filaments allowing computer
simulations of statistically significant quantities of magnets to be calculated. Four basic
types of displacements were investigated: (1) All conductors displaced equally, magnet
center shift; (2) Individual cables; (3) Individual coils (4 per magnet); and (4) Individual
conductor segments, consisting of a block of inner and outer conductor from the straight
section of two coils next to each other. All four cases consider radial displacement and
angular rotation, (2) - (4) use a statistical sample of 100 magnets, allow independent and
random movement and normal distributions.
The systematic error was subtracted out and the standard deviation of the multipole
errors were calculated for all of the normally distributed displacements described above
(Table 3). A linear extrapolation was used to back calculate the magnitude of the RMS
motion required to produce each individual RMS specification multipole for all types of
motions.
The errors caused by the misposition of the individual cables and larger blocks of
current created higher-order and lower order multipoles, respectively. To control
multipoles to within the RMS specification, the standard deviation for the displacement of
the conductors must be within approximately 0.015mm. This is for cables relative to each
other as well as for larger blocks of conductor. While the specification on the random
dipole field of 200 units is equivalent to a magnetic centering specification of 0.2mm, the
random skew quadrupole of 5 units is approximately equivalent to 0.25 milliradians.
The results of the 2-D calculations have been used to develop a manufacturing
tolerance specification for the 3-D situation. The specification for RMS tolerance is on the
integral field that includes the end and interconnect regions. The end regions must be
addressed separately. The straight section specification may be interpreted as an average
value to the location over the length of the quadrupole. Thus, random errors that distribute
evenly over the length of the magnet are not as important as an error that systematically
occurs over the entire length.
261
Table 3 Multipole variations for conductor positioning tolerances.
0.956 0.015
0.259 0.017
0.267 2.030 0.017
0.559 2.006 0.027
0.559 2.006 0.027
0.238 ('.480 ·0.039
0.249 0.630 0.044
45.528 0.283
12.547 91.776 0.570
12.144 0.630
25.167 93.350 0.579
25.131 1.010
11.442 27.768 2.301
10.639 19.949 1.935
Quench Propagation and Protection. There are four major parameters addressed in the
design that affect failures associated with resistive heating; copper current density, material
heat capacity and thermal conductivity, maximum allowable temperature and energy
depositions in the normal zone. Current density and energy dispositions are discussed.
The conductor operating current is 6500 amps, copper to superconductor ratio is
1.8:1 and copper current density is 1043 Nmm 2 , or 20% higher than the worst case in the
COM. This current density increase can dramatically affect the hot-spot temperature. The
adiabatic calculation indicates the CQM hot spot will reach a temperature 80% to 200%
higher than that of the COM, depending on heater delay time. The high copper current
density in the CQM acts like a series fuse with the COMs and drives requirements of the
quench protection system. The hot-spot temperature of the system could be more evenly
controlled, without affecting quantity of leads, if the protection circuit was modified to
redistribute the diodes, such that the CQM was bypassed by a separate diode.
The system quench protection unit is the half cell, although there are two sets of
current bypass diodes that also divide the cell into two parts. The current in the CQM
during a quench is related to the inductance and resistance of the series combination of two
COMs and one CQM along with the lead and diode drops. The CQM represents
approximately 1/20 of the inductance and In of the resistance of this series combination
not including diode and lead drops. The current in the CQM, during a quench, is therefore
dominated by the COMs. To protect the CQM from overheating, the heaters in at least the
two series COMs are energized once a quench is detected. With this protection scheme, the
CQM heaters do not change the rate of current decay significantly. The hot-spot
temperature in the CQM will be determined by the 50 msec quench protection circuit delay
time and the additional delay between energization of a heater and the warming of the
conductor above the critical point. IOC and SSCL experience indicate delay times near 0.1
seconds as typical. The sensitivity of hot-spot temperature to heater delay is shown in
Figure 3.
262
1,000 Heatar Delay
- - - Quadrupole Seconds
0 .10
- - - - Dipole
800
~---0 .05
200
,/ - _ - - - 0.10
0 .00
0. 1 0.2 0.3 0 .4 0.5 0.6

Time, seconds
Figure 3 CQM and CDM hotspot temperatures
HARDWARE DESIGN
Hardware Design Overview. lGC's design is based in large measure on that

developed by the National Laboratories for CQMs and CDMs, However, our design
(Figure 4) proposes several unique approaches to challenging CQM design issues.
Alignment Ba,
Stoel AST M A322
BNmTu
Aluminum
Collet
Figure 4 Cold mass design
263
Yoke Split. The yokes provide support and alignment to the collars, hence to the
coils and beam tube (Figure 5). Earlier designs had the yoke split on the horizontal center
line, an axis of maximum tangential magnetic flux. Analysis shows that this split produces
a dipole effect in field quality. Hence, we have moved the split to a 45° location, modified
the outside of the collars to have a new set of registration surfaces, and moved the skin
weld as well as other yoke features since this has significant magnetic, mechanical, and
assembly advantages.
Enlarged End Region. The end region of the CQM is rather cramped with
transition from the 28 cm CQM skin diameter to the 38 cm CDM diameter. Shortening the
skin, increasing the diameter with a dished head, and inserting a straight section of shell at
the larger diameter will provide more space for splices and magnet connections (Figure 6)
while economically containing the high pressures encountered during quench. This allows
for easier transition of bus connections to the adjacent dipole and spool piece and for
reduction of the iron diameter in the CQM to improve tracking, should that option be
exercised.
Ina eased Volume

Bener Alognmenl for Lead OoamelSl
01 Terminals & TranSlDon & E.panSlOtl
Loop
!:!2lI CoIlaIs SupponecJ aI Surtace5 "S. - 8 P a

l ces.
Clearance at al Other L.ocallOnS
Figure 5 Cold Mass cross-section Figure 6 Enlarged end region

with 45° yoke-split
Beam Tube Centering. The IGC design features a beam tube which is inserted after
collaring, yoking, and skinning. The tube, with centering devices (Figure 7), slides in from
the end of the assembly. Significant advantages include the ability to use a collaring
mandrel to assist in obtaining maximum consistency in preloading coils and a beam tube
supported from the collar pole tips, thus avoiding quench-inducing frictional interaction
with the windings. The "open" character of the beam tube support assures excellent
cooling by helium flow even at upgraded luminosity. The tips of the collar poles are
reconfigured to taper and extend inside the radius of the coils, allowing the tube to be
inserted with the heat formed kapton bumpers over the coils, then rotated into place with
the compressible bumps interfering with the pole tips. The torsion of this rotation results in
less than 0.1 mm twist at the midlength of the 5 m tube. The flats on top of the pole tips
are 5 mm wide, allowing plenty of margin for placement of the bumpers on the tube and
for insertion position.
264
Bump Unloaded for
In$8l11On; Inl8r1Gftlnal
Bum ps Uncler FIt When Sealed
Poje Tips
Bump
Unloaded
III, InsOflloo
Inside Diameler
Figure 7 Beam tube installation.
Power Bus Expansion Loop. An expansion loop must be provided in the main leads at
the return end of the CQM. This takes the form of a loop around the beam tube, with
supporting hardware (Figure 8). This loop must cover 315 0 to bring the bus from the 45 0
location in the yoke to the top center line in the interconnect region. The loop of cable is
supported by a split ring of copper, approximately Imm thick by 1.5cm wide. Mechanical,
thermal, and electrical stabilization are provided to the cable by soldering to the copper split
ring. Additional electrical stabilization in the loop is provided by additional copper split rings.
The use of thin split rings provides flexibility adequate for the possible 2cm of axial
movement at this location. Clamps of insulating material, and a thin insulating separator
between the cables complete the details of the loop.
AXIalBus to
AdJacent Element
SIC CabJe
'-'b
Copper Rings
Figure 8 Power bus expansion loop
265
PROPOSED PROTOTYPE PROGRAM
IGC's commercial experience indicates an essential part of a successful program is

the rapid feedback from prototype experience to guide the concurrent design and
development of the magnet, production tooling, and manufacturing processes.
The prototype program logic is based on a few simple assumptions: (1) It is unlikely
that the initial design quadrupoles will achieve over-field with few quenches and have field
quality fully within specification; (2) Baseline magnets will not fully satisfy all field
multipoles, and may require an iteration to correct and optimize for systematic errors; and
(3) The optimum configuration of the end region to eliminate training will require
evaluation in magnets to make final configuration selection. The result of these
assumptions is the proposed prototype program shown in Figure 9.
1
A 2 Baseline ProlOlypeS
3
4 Multiple End Configurations
B
5 To Study Quench Sensitivily
6 Cold Messes Incorporating
c 7 Design Improvements for
8 Quench & eost
9
10 Uniiormily Improved
o 11 (Incorporate Test Results, 1-3)
12
13 Group A or C with Modifications

E
:~ from Results of B Tests
Figure 9 Proposed prototype program
CONCLUSION
IGC's practical experience in the design, analysis, development and manufacture of

over 600 superconducting magnets has been applied to the CQMto create an innovative
cost effective design reflecting tolerance requirements driving the process. Examples of
this innovation are a 45° yoke split, enlarged end region, beam tube with reduced quench -
inducing features and a stabilized bus expansion loop. A five-generation prototype test
program is discussed which will allow feedback of test experience into the design prior to
pre-production.
It is recommended that industrial vendors be involved in all future SSC
superconducting magnet programs during initial design and development. This commercial
experience will shorten the overall program length and, hence, cost.
ACKNOWLEDGEMENTS
We wish to thank the technical staffs at SSCL, the National Laboratories and Martin
Marietta Astronautics Group for their comments on the work presented here.
266
THE EFFECT OF VACUUM GAS PRESSURES AND SPECIES
ON INTERNAL HEAT LEAK IN THE SSCL MAGNET DESIGN
D. E. Franks and R. K. Pletzer

Dallas, TX 75237-3946
INTRODUCTION
This work describes a procedure for calculating the heat transfer into the cryogens
inside a 50-mm Collider Dipole Magnet (CDM) for the Superconducting Super Collider
(SSC) during its operation in a Collider Ring Half Cell vacuum. Figure I is a schematic
cross-section of the CDM showing various major components and their temperatures.
Described here is the use of a species gas model for assessing the heat transfer due to
conduction through the residual gases in the cryostat insulating vacuum. Residual gas
heating is a major source of heat to the helium-cooled regimes in the magnet. It is a function
of several variables that include the gas(es) involved, and it is found to decrease linearly
with vacuum pressure and exponentially with gas temperature. Thermal radiation, in
contrast, varies with temperature and diminishes as temperatures decrease on the T4 curve.
The gas model used here is based on results from a Residual Gas Analyzer (RGA)
measurement of the vacuum in the Fermi National Accelerator Laboratory (FNAL) ER
40-mm Dipole Magnet String Test (ST). At present, no vacuum tests have been conducted
on 50-mm CDMs.
The work was done to analyze the effect on residual gas heat conduction of presumed
gas species in the vacuum total pressure. It is seen to be a significant improvement over
using a "single gas" model. Results would indicate that unless the species gas model is used.
the residual gas heating can be significantly overestimated depending on the vacuum and
gases involved.
No. DE-AC35-89ER40486.

Yocuum Yeue! Wd
Figure l. Schematic cross-section of the CDM showing components and temperatures.
GAS SPECIES MODEL
The 50-mm COM Prime Item Development Specification states that the normal
operating vacuum for the magnets in the Collider Ring Half Cell is not to exceed 1.0 E-06
Torr total pressure at the vacuum vessel wall.
Dalton's Law states that the total pressure of a gas is the sum of its component species.
It is not defined above what the gas species components involved will be.
The gas species model used here was based on the gases identified in the FNAL ER
ST RGA measurement on 9/6/91 of the vacuum for the 40-mm Dipole ST Magnets. Results
of the gases and their partial pressures are shown in Figure 2. Total pressure was slightly
over 2.0 E-08 Torr.
100
.
50
0
.
0
(At)
!COO)
0=,.
0
,
"",
0 IotWT "
-.
((lIZ)
~ 1otWT2.
~M)
~ 10
...wT"
en
en
« 0
....wwr4
~ )
- 2 O (H2)
MWT 2
10-8 10
·9
0 (><)
MWT ,
· 10
10 ."
10
GAS PRESSURE TORR

Figure 2. Species of Gases, Molecular Weight, and Partial Pressures from 9/6/91 RGA Data at FNAL ER.
268
At the time of the 9/6/91 RGA measurement the ST magnets had been under vacuum
for over a month, and it was felt that outgassing should be seen. The water vapor and the H2
in the data appear to be examples of this. Glassford 1 shows data for multi-layer insulation
(MU) such as is used in SSCL magnets. Clark 2 identifies outgassing rates for a series of
materials, including H2 from steel such as the vacuum vessel wall.
Certain of the RGA data also exist in air, including He, H2 and C02. A choice here of
the FNAL species seemed appropriate in that the 50-mm CDM design is similar to that of
the 40-mm Dipoles.
In particular, the gases selected for this analysis are shown in Table 1.
Table 1. Analysis gas species.
H2O Water Vapor

H2 Hydrogen
N2 Nitrogen
CO2 Carbon Dioxide
He Helium
The gases are a mix of those with the highest partial pressures and those that will
contribute heating to the Collider Ring helium refrigeration system.
The helium gas species above is noted to be here in a "trace" amount only. Again, it is
included because it is a component of the gas mixture for air. As such, it probably could not
be removed by N2 purging.
ANALYSIS
The analysis began by determining values of expected gas species pressure due to their
cooling as they move inward from the vacuum vessel wall. The temperature gradient
involved is extreme. It ranges from the 300-K vessel wall to the 4-K cold mass surface.
The "Thermal Transpiration" theory of Welch 3 describes this pressure calculation. It
is based on the relationship that
P
-=Constant
ft
where P = Pressure and

T = Temperature.
Starting P and T for the gas species components are the measured "room" conditions
(P by RGA). T is the room air temperature at the pressure gauge.
From this
P eold -_ P room • ~-
Teold
T
room
P eold can be calculated for any assumed Teold. This holds true until the species curve
intersects its vapor pressure curve. This intersection, if it occurs, is where the gas
cryocondenses ("freezes out"). Below this temperature value it no longer contributes to the
gas conduction.
269
Figure 3 shows a plot of the analysis gases and where they intersect their
cryocondensation curves for their FNAL RGA starting conditions. Assumed room
temperature was 300 K. The "common gases" vapor curves are found in various works. 3,4,5
For comparison, the situation of N2 as a "single gas" was included here. Room
pressure for it then is the 1.0 E-06 Torr maximum allowable Half Cell vacuum pressure.
Water vapor is projected to "freeze out" at about lS0-K, C02 at around 70 K, and N2
slightly above 20 K. H2 and He will remain gases at and below the 20-K and the 4-K
surfaces.
Also, from Figure 3 data it would appear that N2 would not exist for gas conduction to
the 4-K surface. For the Figure 2 room pressure of 1.0 E-09 Torr, it was calculated to freeze
out at about 23 K. For the higher initial pressure of 1.0 E-06 Torr it was calculated to freeze
out at around 27 K. This is due to the N2 vapor pressure curve increasing. As such, for
either case, the N2 would probably freeze out somewhere in the 20-K MLI.
II:
II:
~
W
II:
"/
:0.1
10.2
10.3
e
::::>
en
en 10.4
w
II:
c... 10.5
II:
0 10 -6
c...
:; 10.7
10.8
Figure 3. Vapor Pressure of Several Gases.
The method of analysis from here was to input the P versus T data for the five gas
species into the gas conduction subroutine of the SO-rrun CDM thermal math model.
The model used, developed by Randy Pletzer of the Magnet Systems Division
Cryostat Design Section, is of the full magnet and its Interconnect.
The computer code used for the solution of heat flows in the model was the General
Dynamics "Convair Thermal Analyzer" (CT A) Code. The routine is proprietary to General
Dynamics. Results determined here are for "steady-state" solutions.
The thermal model describes the surfaces in the vessel experiencing heat transfer and
the "paths" to be expected that connect them. The comprehensive model is solved for heat
flows resulting from and into various "fixed" temperature locations such as the ambient air
outside and the cryogens inside.
In that the CTA Code is based on the "Electrical Analogy" to heat transfer, these paths
are defined as thermal "resistances," R. They are for such effects as solid conduction,
radiation, and gas conduction between the surfaces involved.
Each pair of nodes in the thermal model that could exchange heat by the mechanism of
residual gas conduction was connected by five gas conduction resistors-one for each gas
involved. Figure 4 shows this arrangement for heat flow from the vacuum vessel wall
through the MLI and into the cold mass.
270
An equivalent situation exists for heat flow through the support posts, except it
included the solid conduction resistors.
The gas conduction equation solved for the Figure 4 resistors is that of Corruccini. 6
Barron7 presents a similar theory. In these calculations the gas P versus T was calculated by
the previously described thermal transpiration theory down to its cryocondensation value.
Below this, its pressure was input as that of the "solidified" form, which is substantially
below that of the vapor. This lower value effectively dropped its contribution out of the
calculation.
This modeling approach has the effect of letting the code and the model determine
where each species can exist in the gaseous form.
In contrast to the above, the previous procedure was to use a "single gas" model.
Figure 4. Thermal "resistance" path for heat flow from "the vacuum vessel wall through the MLI blankets and
into the cold mass.
271
RESULTS
Four analysis cases for the gas model species partial pressures were analyzed here.
These are shown in Table 2.
Table 2. Analysis cases and description.
Case Description
# 1 Species gases at the Figure 2 "room" values.

#2 Gases at lOX higher pressures than Case 1.
#3 Gases at 100X higher pressures than Case 1. Case 3 is approximately the 50-mm
vacuum total pressure operating condition. Case 1 was for a total pressure ofE-08 Torr.
#4 N2 as a "single gas" model with a room pressure of 1.0 E-06 Torr. The CTA Code logic
is such that it does not consider the freezing out of a gas. Instead, it "extends" gas
property data.
Table 3 shows the pressure versus temperature values used for H2 in Case 3. This
shows the low gas pressures to be expected in the 4-K and the 20-K regimes in the magnet.
Table 3. Pressure versus temperature data for Case 3 H2.
Temperature (K) 300 250 194 83 3

Pressure (Torr) 3.9 E-08 3.6E-08 3.14 E-08 2.06 E-08 4.1 E-09
Results from the four analysis cases were summarized and are presented here as
predicted heat loads to the three cryogens in the cryostat. They are compared to the
"budget" heat leaks for these cryogens. The budget values8 to the Collider Ring various
refrigeration systems are shown in Table 4.
Table 4. Budget heat loads for CDM and Interconnect to various Collider Ring
refrigeration systems.
4K 20K 80K
IR (W) 0.0530 2.335 18.40

Post(W) 0.160 2.400 17.34
Interconnect (W) .!l...UOO ll.32O ..2..1.00
Total (W) 0.363 5.055 37.84
Table 5 shows the results from the four analysis cases in comparison to the above.
Results for Cases 1,2, and 3 show the effect of decreasing initial ("room") pressure for the
species gas model. Theory shows that the residual gas conduction heat transfer effect
becomes less prominent as the vacuum pressure decreases. This effect is seen in the above
results for the 4-K and 20-K regions. In the 80-K region, thermal radiation is the dominant
mechanism. It largely dwarfs gas conduction.
272
Table 5. Comparison of 4-K,20-K and 80-K Refrigeration System Budget Heat Loads with
Predicted Loads for Each and % Margin.
Analysis Case
#1 #2 #3 #4
4 K Budget (W) 0.363 0.363 0.363 0.363

- 4 K Predicted Load (W) 0.119 0.120 0.136 0.1805
Margin (W) 0.244 0.243 0.227 0.1825
% 67.2% 66.9% 62.5% 50.3%
Difference Under Under Under Under
20 K Budget (W) 5.055 5.055 5.055 5.055

- 20 K Predicted Load(W) 2.937 2.891 2.989 3.062
Margin (W) 2.118 2.164 2.066 1.993
% 41.9% 42.8% 40.9% 39.4%
80 K Budget (W) 37.84 37.84 37.84 37.84

- 80 K Predicted Load (W) 36.83 36.84 36.82 35.25
Margin (W) 1.252 1.20 1.13 2.59
% 2.7% 3.22% 3.0% 6.8%
Also, it is to be noted that the predicted loads above are totals of the described budget
components of solid conduction through the support posts plus gas conduction and thermal
radiation through the MLI. These latter two effects are the "IR" term used in SSCL data.
To get the needed modeling accuracy the thermal model involved here has a total of
2,471 surfaces ("nodes") and 7,669 resistors.
Due to space limitations and formatting problems with heat load results, the individual
contribution of gas conduction could not be separately presented. As such, the effect is
inferred between the various cases above. Changes in total heat load in the above cases are
due entirely to the change in gas conduction. Essentially only the vacuum pressure changes
between the cases. Internal temperatures remain fixed for radiation and solid conduction.
Comparing the results for Case 4 for the 20-K and especially for the 4-K predicted
heat loads to those of Case 3 gives some ideas of the results of using a "single gas" model.
It is seen that heat load is overpredicted. Again, the true residual gas conduction effect is
somewhat masked by the format necessitated here.
CONCLUSIONS
A series of conclusions can be drawn regarding the preceding results. The first is that
there are no test data yet for the gas species to be experienced in the 50-mm CDM insulating
vacuum when it is in operation in the Collider Ring Half Cell. What was analyzed here is
conjecture based on FNAL ER ST RGA data.
However, if it does tum out that the Half Cell vacuum makeup is in keeping with the
Case 3 model used here, then it would be expected that heat loads to the helium
refrigeration system would be well below budget values. If the vacuum makeup were to
improve to the Case 2 or Case 1 conditions, the improvement would be even more
noticeable.
273
From discussion with various vacuum system personnel it would seem that a good
case can be made for the Case 3 gas mix.
It can be expected that the "dry" N2 used for the purge will be somewhat readily
removed during the pump down after purging because of its low "sticking" affinity for
materials in the cryostat. Consequently, this would fit the FNAL data.
Also, the species with the highest partial pressure in the vacuum total pressure will
probably be water vapor. Reference data show it will slowly outgas from the MLI. This is
due to the MLI design. As such, not all of it will be removable during pump down and
purge. Again, this fits the FNAL data.
The water vapor will be expected to freeze out, though, inside the 80-K MLI.
Consequently, it will not contribute to gas conduction below the 80-K shield.
The final conclusion drawn here is that it should be noted that the total vacuum
pressure will be a mix of species, each with its various partial pressures. Consequently, it is
not descriptive to use the "room measured" total pressure as the starting condition for
thermal transpiration calculations of gas pressure versus temperature. This would
overestimate the gas pressure actually occurring in the gas conduction regime and
consequently overpredict the effect of gas conduction there.
If possible an RGA reading is called for. This would establish the species involved and
their partial pressures at room temperature. These pressures would be the correct values to
use for the needed thermal transpiration calculations. Subsequent gas modeling would be
based on what was done here.
It will be of interest to get an RGA reading of the Collider Ring Half Cell vacuum
when possible. This would assess its makeup versus what is assumed here. If comparable,
the predicted reduction in heat loads to the helium system would be expected.
ACKNOWLEDGEMENTS
The author gratefully acknowledges the help of several people at SSCL in doing this
work. Thanks go first to Randy Pletzer, who provided the thermal model and helped to
execute it and to solve technical problems encountered in running the analysis cases. He
also reviewed the text.
Dorothy Zink and Ed Daly are duly appreciated for their help with formatting and
word-processing of this document.
Lastly, the help of Jim Lawler is appreciated for the artwork.
REFERENCES
1. A.P.M Glassford and C.K. Liu, "Outgassing Rate of Multilayer Insulation Materials at Ambient
Temperature," Journal o/Vacuum Science and Technology, (17)3, May/June (1980).
2. SSCL Internal Memo, February 28, 1986, D.M. Clark to G.J. Still. Subject: Notes, Out Gas Rates.
3. Kimo M. Welch, Capture Pumping Technology, Pergamon Press, New York, 1991.
4. CRC Handbook 0/ Chemistry and Physics, 66th ed., CRC Press, Inc., Boca Raton, Florida, 1985.
5. Marsbed H. Hablanian, High-Vacuum Technology, Marcel Dekker, New York, 1990.
6. R.J. Corruccini, "Gaseous Heat Conduction at Low Pressures and Temperatures," Vacuum, Vol 7-8,
pp 19-29.
7. Randall F. Barron, Cryogenic Systems, Oxford University Press, New York, 1985.
8. SSCL Site-Specific Conceptual Design, July 1990.
274
SEXTUPOLE CORRECTOR COIL MANUFACTURE
USING THE DIRECT WIRE FABRICATION PROCESS
E. T. Gossler and C. C. Coghill, III
Babcock & Wilcox

Accelerator & Magnet Systems
P.O. Box 785
Lynchburg, Va. 24505
ABSTRACT
The Accelerator and Magnet Systems group of Babcock & Wilcox is currently
under contract to develop the manufacturing processes necessary to produce
sextupole corrector magnets using the Direct Wire coil fabrication process. The
Direct Wire process utilizes a computer controlled wiring machine to ultrasonically
bond superconducting wire to a support tube producing corrector magnet coils. A
sextupole coil has been fabricated and the process is being refined to improve the
quality of subsequent coils and reduce manufacturing time.
INTRODUCTION
As part of the development of corrector magnets for the Superconducting Super

Collider, the SSCL has contracted with private industry to demonstrate the
manufacturability of corrector magnets using various fabrication techniques. The
Accelerator & Magnet Systems group of Babcock & Wilcox in Lynchburg, Va. is
currently in Phase I of developing Direct Wire technology to produce Sextupole
Corrector Magnets.
The Direct Wire technique is a modified version of the Multiwire 1 process
developed by the PCK Technology Division of the Kollmorgen Corporation, now
known as Advanced Interconnection ~echnology (AIT). The Multiwire process uses
an automated wiring machine consisting of an ultrasonic head, a precision
positioning system and a computerized control system to make interconnections on
high density circuit boards with small diameter wire (ie.. 1 mm). The interconnecting
wire is fed through the wiring head and under a precision-grooved stylus which is
excited by ultrasonic transducers. See figure 1. As the wire is guided along a
computer controlled path the ultrasonically excited stylus imparts energy to the wire
which locally melts an adhesive coating, bonding the wire in place.

,
ULTRASONIC
GENERATOR
_ _ ULTRASONIC
TRANSDUCER
GROOVED STYLUS
Figure 1. Multiwire Head 2
Figure 2. Wire Cross Section (115 X) 2
276
PROCESS EVOLUTION
The superconducting wire used in the fabrication of corrector coils consists of a

.015 inch diameter 2:1 Cu:SC wire spirally wrapped with .0005 inch Kapton tape and
overcoated with .0005 inch of adhesive. 3 Figure 2 shows a magnified cross section
of this wire.
In order to accommodate the larger diameter and different mechanical
properties of the superconducting wire this head went through several modifications.
The Multiwire Head was originally modified to produce a coil configuration known
as a Jellyroll Coil. This Jellyroll Coil was produced by ultrasonically bonding
superconducting wire to a flat adhesive coated substrate in a predetermined pattern
using computer generated coordinates. This wiring technique produces a series of
two layer "racetracks" at controlled spacings from a continuous length of wire. Once
the coil is completed the "racetracks" are peeled from the substrate and the excess
adhesive removed. The "racetracks" are then manually wo.und around a support
tube building the coil two layers per turn to produce the desired cross sectional
profile. Prototype magnets using this technology have been previously built and
tested. 3
The Direct Wire technology uses a technique similar to the Jellyroll however in
this process the superconducting wire is bonded directly to the support tube instead
of first wiring the coil on a flat substrate and then manually winding it around the
support tube. The support tube used in the Direct Wire process is a thin walled
stainless steel tube which is spiral wrapped with a layer of Kapton that has been
coated with a B staged thermoset adhesive. The first layer of the coil is embedded
into the adhesive coating to the Kapton wrap. Subsequent layers are adhered to
previous layers in a nesting pattern with the completed cross section resulting in the
profile shown in figure 3. 4
Figure 3. Coil Cross Section Drawing
DIRECf WIRE COIL FABRICATION
To produce a coil using the Direct Wire technique a prototype wiring machine
was modified by replacing the X-axis table used to produce Jelly roll coils with a
rotating fixture to hold and position the support tube beneath the wiring head. This
prototype machine is pictured in Figure 4.
The wiring head is mounted to a linear transport system which provides
longitudinal movement along the support tube. The wiring head also rotates about
277
Figure 4. Wiring Machine
its central axis to allow negotiation of the turns at the ends of the coil. All of the
above mentioned axes of motion are driven by computer controlled servo motors
with encoder feedback to verify position. The mounting of the ultrasonic stylus in the
wiring head provides a small amount of compliancy in the Z-axis. Since the Jellyroll
coils are never more than 2 layers high during manufacture this compliancy is
sufficient such that no adjustment of the Z-axis is necessary. The Direct Wire
technique however requires that the head height be raised as the coil height
increases. Another factor complicating manufacture is that as the coil height
increases the turns of the coil grow at a faster rate than the nested straight sections
due to crossovers. This resulting step in height eventually becomes greater than the
stylus compliance can tolerate and the wiring process has to be stopped at each
transition point to allow the head to be raised or lowered. This was initially
accomplished through manual adjustment of a fine pitch leadscrew. Test coils as well
as the first Direct Wire Coil (pictured in figure 5) have been produced using this
manual adjustment technique. Although manual intervention was used to prove the
feasibility of using the Direct Wire method to produce corrector coils, it will not be
satisfactory for fabrication of production coils.
Figure 5. Direct Wire Coil
278
DIRECf WIRE TECHNOLOGY FUTURE DIRECTION
In an effort to reduce the time required for coil fabrication and maintain coil
consistency, the wiring head Z-axis is currently being retrofitted with a prototype
automatic positioning system. This system consists of a stylus height sensor, a stepper
motor driven lead screw, and an electronic control circuit. As the stylus encounters
variations in height the control circuit detects this change through the sensor and
drives the stepper motor to maintain a constant height of the stylUS above the coil.
In order for the wiring machine to function properly and produce coils within
specifications the quality of the wire must be assured before coil wiring is begun.
Problems such as non concentric adhesive coating, too light or too heavy coating, and
damaged or missing insulation can cause malfunction of the wiring machine and I or
produce inferior coils. Babcock & Wilcox is in the process of testing various
sensoring techniques to determine their effectiveness in detecting anomalous wire
conditions. Once this testing is complete the proven sensors can be incorporated into
existing wire manufacturing lines or integrated into independent inspection systems.
As the development of Direct Wire manufacturing evolves, critical fabrication
parameters will be established allowing the design of production wiring machines
fitted with sensoring systems to monitor and maintain coil quality. Automated
fabrication with on line inspection of coil parameters insures that high quality coils
will be produced exhibiting consistent operating characteristics without variations
that can be introduced by manual fabrication techniques.
CONCLUSIONS
Although the Direct Wire technique is still in development it holds the promise
to produce high quality corrector coils, with the maximum possible current density, in
an automated fashion. Once fully developed this technique will allow quick response
to design changes such as coil cross sectional profile, length, etc. without fabrication
of new fixturing or tooling. Since the coil geometries are defined through software
coordinates, custom coils can also be produced by generating new or modifying
existing wire paths.
REFERENCES
1. Multiwire is a registered trademark of Advancd Interconnection Technology.

2. Graphic provided by Advanced Interconnection Technology.
3. J. R. Sanford and D. M. Matthews, ed., Site-Specific Conceptual Design of
the Superconducting Super Collider, July 1990.
4. Coil cross sectional drawing provided by SSCL.
ACKNOWLEDGEMENTS
Babcock & Wilcox wishes to acknowledge the contributions of BNL and SSCL
in the development of the technology for producing Direct Wire corrector magnets.
219
SUPPORT SYSTEM DESIGN OPTIONS FOR
ACCOMMODATING CQM COLD MASS THERMAL CONTRACTION
O. T. Stutzki, K. E. Grut, M. W. Hiller, and J. A. Waynert

Babcock & Wilcox
Lynchburg, VA 24505
ABSTRACT
The current design for supporting the Collider Quadrupole Magnet (CQM) cold mass
utilizes one support which rigidly constrains the cold mass and one support which restrains
the cold mass vertically and laterally, allowing axial motion during thermal contraction.
The cradle assembly concepts to date have been designed with the Collider Dipole Magnet
(COM) in mind. However, the loading and thermal contraction of the CQM are different
from the COM, therefore, alternate designs have been investigated. Sliding as well as
anchored interfaces are considered as options for support of the cold mass. The interaction
between the cradle and relevant components (i.e. thermal shields, cryogenic piping, and
support posts) have also been considered. Several cold mass support system concepts are
presented and the most feasible configuration is determined.
INTRODUCTION
The main objective of the cold mass support system is to accommodate thermal
contraction and expansion of the cold mass and support the cold mass weight. The entire
cold mass is considered to act monolithically with respect to the surrounding cryostat
components, designed to interface with each other to allow for thermal contraction and
expansion. To support the heavy cold mass with minimum heat loads and minimum sag,
two cold mass cradles attached to posts will form the cold mass support system.
The design objectives are, in decreasing order of critical attention: alignment stability,
low heat leak, high reliability, high structural strength, manufacturability and minimum
cost. The assembly is to accommodate a cold mass axial thermal contraction/ expansion of
approximately 7-8 mm between the two posts and support a cold mass weight of
approximately 1700 kgs maximum (or 850 kg per cradle). These weight values reflect a
reduced cold mass diameter of 234.188 mm.
The most difficult task will be to keep the cold mass in alignment during thermal
cycling. Currently, there is an allowable horizontal and vertical deflection of 1.0 mm and
rotational field distortion which may not differ from the average by more than +1- 10
milliradian-meters. The geometry of the cradle provides lateral and vertical alignment but
cold mass rotational alignment depends on support system design.
The cradle and sliding materials utilized will be subjected to severe temperature and
radiation environments during operation (4.35K, 1000 grey at the support posts and
5.0x10 12 neutron/cm 2 over the 25 year life in the cradle area). Other environmental
Supercollider 4, Edited by I. Nonte 281

extremes are a 1.3xlO-s Pa vacuum and 0 to 98% humidity_ The assembly must be able to
withstand large shock loads; the worst case being 3.0 g in the vertical direction. Also, the
sliding material must exhibit a low coefficient of friction as well as low wear and resistance
to radiation, creep, and outgassing. Finally, after the assembly has met these primary
requirements, the assembly must also be designed for high reliability and minimum cost.
CRADLE DESIGN
The current SSC cradle design is an assembly composed of 11 piece parts, which are
welded together to form the lower and upper cradle rings shown in Figure 1. Two snubbing
collars are used to attach the cradle to the top of the post.
UPPER
CRADLE UPPER CRADLE
ENDPI ECE RING
LOWER
CRADLE
RAD I USED UPPER
ENDPIECE CRADLE
ENDPIECE
LOWER CR./I,DLE - LOWER

RING CRADLE
ENDPIECE
BASE PL./I,TE
GUSSET
SNUBBING
COLLAR
i
j1 '
I'd91'
, I
~'
~
Figure 1. Current SSC Cradle Design
Due to the prototype nature of these cradles, it is easy to understand why the cradles
were machined from piece parts. However, for production quantities of CQM magnets,
various materials and manufacturing techniques have been investigated. It would be
feasible to make both anchor and sliding cradles identical to minimize design and
manufacturing costs. Also, it would be beneficial to reduce the number of piece parts, to
just the lower and upper cradle rings; excluding the fasteners such as studs, washers, nuts,
etc. Both of these rings can be cast from CF8 stainless steel (similar to 304 stainless steel)
282
or a less costly, lighter, permanent-mold casting aluminum alloy such as AA 355 T6. Fermi
National Accelerator Laboratory (FNAL) has investigated a cast cradle with the Design B
Casting Prototype. Once the cradle rings have been cast, only final machining of the mating
surfaces would be necessary. This includes machining the interfaces between the upper
and lower cradle rings, the cold mass shell and the cradle, and the base of the cradle and
top of the post. To simplify assembly and alleviate the recent ASME Code requirements
imposed on the CQM, the weld between the anchor cradle and cold mass will be replaced
by knurled bearing pads to provide friction and grip the cold mass. As a result of a new
support post concept, both snubbing or locking collars will be eliminated to simply bolt
the cradle to the top of the post.
Another consideration is to fabricate an injection molded cradle using a fiber
reinforced polymer such as Ultem R or Torlon R • Some advantages of such a cradle would
be weight reduction, greater thermal resistance and reduced manufacturing cost. The
creep behavior for these materials, however, is not well known and is currently being
investigated and a thorough development plan would be necessary to provide an adequate
mold configuration for the cradle.
LOWER
CRADLE
RING
Figure 2. Cast Cradle Design
SLIDE MATERIALS
As of this writing, over 10 slide materials have been tested and another 17 are in queue
for testing at FNAL's Cold Test Facility (CTF). The CTF has been designed to simulate the
conditions at the sliding interface in the COM cryostat, but only at approximately 80K.
283
Various coatings, composites, and ceramics have been tested with a graphite embedded in
a bronze/copper substrate showing the most promise to date. Candidates were selected
based on the material's ability to maintain a low coefficient of friction at the necessary
temperature and radiation environment. The CTF can simulate the vacuum, loading, and
partial thermal conditions, however, no irradiation tests have been performed on the
materials which have been tested. As the candidates become narrowed down to a few
candidates, an attempt will be made to have the finalized slide materials tested at
Brookhaven National Laboratory (BNL) during the summer of 1992.
VARIOUS SUPPORT SYSTEM DESIGNS
Although the baseline design includes an anchor cradle and a sliding cradle with a
sliding interface at the cradle ring, various other designs have been considered. While
many designs have been suggested, most can be categorized into two primary designs, that
being the anchor/slide designs and the anchor/anchor designs. The anchor/anchor
design can be considered for the CQM since the CQM's axial cold mass movement due to
the thermal cycles is approximately 1/4 that of the CDM contraction at the two outermost
sliding interfaces.
With the designs proposed, a key issue is where to accommodate movement of the
cold mass. The location of the slide interface is largely dependent on whether the slide
material can withstand the conditions at 4K. In addition, regardless of whether the sliding
interface is at 4K or 300K, the cold mass support system must interact with the cryostat
components during cool down and warm-up. This includes the piping/ cradle interface,
the shields/post interface, and the MLI/post interface. These various designs are
discussed in more detail as follows:
ANCHOR I SLIDING DESIGNS
Sliding Interface at the Cradle Ring (Modified Baseline Design)
This design is based on the FNAL design and satisfies many of the objectives and
requirements specified by SSCL, specifically alignment stability. Although this design was
devised for a CDM cold mass, some minor improvements could be implemented for a CQM
cold mass. An anchored cradle is rigidly attached to the cold mass and the post. An
alternative would be to simplify the anchor cradle such that the cold mass would be
attached to the lower cradle ring via a strap or band, thereby eliminating the machined
upper cradle ring. A slight modification to the anchor cradle would be necessary to include
the slide material at the cradle ring. In the FNAL design, the position of the bearing pads
are at 45 degrees from horizontal and large normal loads are exerted at these contact points.
A modification would be to place a single bearing pad of approximately 60 degree arc at
the bottom of the cradle to reduce the normal load exerted by the cold mass in addition to
the two bearing pads at 45 degrees on the upper cradle ring. With identical anchor and
sliding cradles, knurled pads could be placed in the recessed area of the lower and upper
cradle rings; where the bearing pads are inserted in the sliding cradle. This would provide
friction to grip the cold mass and eliminate an ASME weldment. This design is shown in
figure 3.
Sliding Interface at the Cradle Base
In this design, the sliding interface is located between the base of the cradle and the
top of the post. Both cradles would be manufactured as anchor cradles since sliding is no
longer located at the cold mass diameter. Again, the cold mass could be anchored to the
cradles using a simple band or cable. The cradles would have a generic bolt pattern at the
cradle base which would mate-up with either the top of the post at the anchor position or
284
,
.:.
. SUD_ a\AOI..E
ASSDtILY
A
COLD MASS SUPPORT SYSTEM
IX It!AJlUHG PAD
ISLlDWCI INTPFAal
. . . S><lELD
10. $MEU)
VIEW 10-1>. VIEW 10-1>.

MODI~D SUOING INTERfACE SUOING INTERfACE
AT THE CRADLE RING AT THE CRADlE BASE
Figure 3. Cold Mass Support System and Sliding Interface at the Cradle Rings and Cradle Base
a slide assembly at the sliding position. If loads allow, the lower cradle ring will no longer
require a 180 interface on both cradles. This design has been developed for other
0
superconducting magnet facilities, however, these facilities do not have the strict alignment
requirements required for the sse ring.
Sliding Interface at 300K
Another design, originally proposed by General Dynamics (GO), involves offsetting

the axial motion to the base of the post, making both cradles anchored. This design
eliminates thermal cycling and reduces the amount of radiation to which the sliding
material would be exposed. Due to the less demanding 300K environment, an acceptable
sliding material should be available. The post base and closure flange on the vacuum vessel
would need to incorporate a sliding interface and an assembly scope for this design will
need development. The eQM cryostat baseline uses a single piece vacuum vessel through
which the cold mass is inserted using a dolly-and-jack concept and the remainder of the
post is attached to a closure flange at the bottom of the vacuum vessel. A 300K sliding
assembly at the base of the post would require design changes and final assembly would
be difficult.
COlD MASS
zo It SHIELD
10 II: SMUD
a.osURf RANGE
OETAI. _
POST aASE
eN !'lATE
DETAil A-A
SLIDING INTERFACE AT 300K
Figure 4. Sliding Interface at 300K
285
ANCHOR I ANCHOR DESIGNS
Fixed Posts
Instead of a sliding interface, the fixed post design consists of two anchor cradle /
fixed post assemblies. The intent here is allow the posts to flex to accommodate the thermal
contraction and expansion of the cold mass. There are a number of ways to approach this
design. One approach is to simply assemble the magnet as described in the baseline
approach and allow the posts to flex during cold mass expansion and contraction. Each
post would then be under large loads from axial and rotational movement of the cold mass
during collider operation. Another option would be to preload the posts outward in the
axial direction during CQM assembly such that, during cool down, the posts would relax
into the vertical position or load in the opposite direction. A drawback here is the posts
would be loaded during storage and collider down-time.
Pivoting Posts
This design has not been investigated in great detail and would be considered as a last
resort, specifically if all other designs were not feasible. Basically, this is a modification of
the fixed posts design. This design has been considered to accommodate forces exerted on
the post which might be too great for the post to handle. A pivoting mechanism would
need to be designed to allow for movement of the posts and provide adequate alignment
behavior.
COlD MASS SUI'I'OIIT POST AT

_ CONOmONS
ANCHOR CllADL£
SUPl'OftT POST
AT COM OI'ERAl1NO
CONDmONS
ANCHOR f ANCHOR DESIGN
Figure 5. Anchor/Anchor Design
Comparisons of Various Designs and Cryostat Component Interfaces
With the exception of obtaining an acceptable slide material which can withstand the
conditions at 4K, a sliding interface at the cradle ring does not have any outstanding
difficulties to overcome. With the sliding interface at the cradle base, offsetting the sliding
interface does not provide any additional outstanding features other than a planar sliding
surface and a simplified cradle design. If no suitable slide material can be found, the sliding
interface must be displaced elsewhere within the cryostat. The slide interface at 300K
design eliminates many of the thermal and radiation exposures that are imposed on the
286
slide material in the 4K designs. However, the ability to integrate a slide assembly at the
base of the post as well as the assembly of such a design may prove difficult.
Both of the anchor / anchor designs were initially assumed to be practical with respect
to meeting all the objectives. However, unless further analysis shows a reduction in the
loads on the posts and/or SSC alignment requirements are relaxed, the anchor/anchor
designs are not feasible. In addition, these designs have not been proven via prototypes
or similar testing.
The sliding interface at the cradle ring is the only design which allows for cold mass
rotational movement as well as axial movement. In the remaining designs, there are
concerns with alignment stability and moment loading of the support post due to cold mass
deflection and rotation. A high torsional load is expected on the post due to the restriction
of cold mass rotation in these alternate designs. The potential key issues that arise are the
ability of the posts to withstand the large rotational forces expected and controlling
alignment with these designs.
Regardless of which cold mass support system is utilized, it must interact with the
various components which will contract and expand at different rates within the cryostat.
With the sliding interface at 300K and both anchor/anchor designs, the support system
design will need to accommodate for relative axial or pivoting motion of the posts,
respectively, at the piping/ cradle, the shields/post, and the MLI/post interfaces. With the
sliding interface at the cradle base, the posts remain stationary such that accommodating
the shields and MLI becomes more straightforward. In fact, with the sliding interface at
the cradle ring, incorporating the interfaces is minimal because the only moving part is the
cold mass.
ACTIONS
The sliding interface at the cradle ring design appears to have the most potential. This
design is being pursued most aggressively whereas the remaining designs are considered
backups and are being pursued in parallel, but to a lesser extent.
The cradle design will be similar to the FERMI design, regardless of which cold mass
support system design is considered feasible. The current configuration is to cast the
cradle. Discussions with various vendors are beginning to optimize the casting process
and reduce cost. Due to the smaller yoke diameter of the CQM, the cold mass will weigh
less such that the cradle can be reduced in both size and material to accommodate the CQM
cold mass.
A sliding material must be narrowed down to a few candidates which are able to
withstand the conditions at 4K. As FNAL completes the next round of slide material testing
in the CTF, the slide materials will be narrowed down to a few candidates which can be
irradiated during the summer of 1992 at BNL. When BNL confirms the irradiation
properties, an optimal slide material can be selected, provided all the design conditions are
met. To date, it appears a feasible slide material will be available.
The optimal design would then need to accommodate the interaction of the remaining
cryostat components such as the cryogenic piping, shields, and MLI. The segmented
shields, for example, would be rigidly attached to the post, and the cryogenic piping will
be attached to the cradle either via a polymer pipe support or aluminum bracket. Thermal
analysis is continuing to optimize these interactions. With the most emphasis being placed
on the sliding interface at the cradle ring, these cryostat component interfaces have been
analyzed in SSCL prototypes and can be implemented.
CONCLUSION
As of this writing, many of the specifications for the quadrupole magnet are yet to be
determined or are in the process of being revised such that designs and status updates
presented in this paper will need to be modified or changed.
287
Overall, the sliding interface at the cradle ring design provides the best approach. It
was designed for a CDM, can be adapted to a CQM, and has been incorporated in several
SSC prototype cryostats. This design is a practical design to develop and mass produce
and the only design which allows for cold mass rotational movement as well as axial
movement. Although the testing of SSC prototype magnets do not indicate any excessive
rotation of the cold mass, it has become apparent that this particular movement has not
been accurately determined. Due to the concerns with alignment stability and moment
loading of the support post as a result of cold mass deflection and rotation, the remaining
designs are not ruled out and are considered viable only as complications may arise with
the cradle ring design. With the objectives met, and unless some unforeseeable difficulty
arises, the sliding interface at the cradle ring will continue to be pursued to be incorporated
intheCQM.
ACKNOWLEDGEMENTS
B&W would like to thank the technical staff at Fermi and Brookhaven National
Laboratories for their cooperation and input of various support system designs.
REFERENCES
Larson, E.T., et al., "Status of Suspension Connection for sse Coil Assembly", Supercollider 1,
Plenum Press, New York, 1989, pp. 493-5Ol.
Lipski, A., et al. "Friction and Wear of Radiation Resistant Composites, Coatings and Ceramics in
Vacuum and Low Temperature Environment" Fermi National Accelerator Laboratory,
Batavia, Illinois, unpublished.
288
DESIGN AND CONSTRUCTION EXPERIENCE WITH
COlLET END CLAMPS FOR 50 MMAPERTURE SSC
COUIDERDIPOLES
Patricia Heger, Rodger Bossert, John Carson,

Stephen Delchamps, Kerry Ewald, Howard Fulton,
James Kerby, Wayne Koska, James Strait
and Masayoshi Wake
Fermi National Accelerator Laboratoryt

Box 500
Batavia, IL 60510
ABSTRACT
Collet end clamps employed on full length and model SSC dipole
magnets designed and built at Fermilab are described. Experimental data
are presented for the baseline Fermilab design. Tests using pressure-
sensitive film to register the midplane and pole coil pre-stresses in the end
clamp region are reviewed. Theoretical calculations address end clamp
mechanics at room temperature, liquid helium temperature, and under
magnetic excitation.
INTRODUCTION
Fermilab 50mm aperture SSC Collider Dipole Magnets have three

splices at the lead end. Collet end clamps [1,10] are used to achieve
continuous support of coils and splices in the end region; they prevent coil
motion which could result in a quench. This region is subjected to
gradients in prestress, magnetic field and Lorentz forces. In the Fermilab
SSC 40 mm aperture model dipole program, magnet performance was
sometimes limited by quenches occurring in the end clamps, in particular
near the inner to outer coil splices [2]. In contrast, very few quenches have
been observed in the end clamps designed for the 50 mm aperture model
and full length magnets built to date at Fermilab [3, 4, 5].
The purpose of this paper is to better understand the performance of
the end clamps by comparing models of the end clamp mechanics to
t Operated by the Universities Research Association, Inc., for the U.S.

Department of Energv.

experimental data. During construction of full length and model magnets,
the change in external diameter of the end clamps upon installation is
measured, giving an indirect indication of coil stresses in the end clamp.
In separate tests, pressure-sensitive tape was used to measure pole and
midplane stresses in the end clamp directly, and an attempt was made to
correlate the observed stresses with changes in end clamp diameter. In
this paper, these measurements are compared with mechanical models
which predict end clamp deflection as a function of coil prestress. The
analysis was taken one step further by using the models to predict changes
in coil stress under cooldown and magnetic excitation.
END CLAMP DESIGN
The Fermilab 50mm dipole end clamp design [1], at the magnet's lead
end, is shown in Figure 1. The coil straight section extends 76 mm into the
end clamp. The coils then make the turn from quadrant to quadrant. Inner
to outer coil splices are held by cutouts in the four G-10 collet quadrants
which surround the coils; the upper to lower outer coil splice is held by a G-
10 splice housing and cover. The collet quadrants are separated azimuthally
by approximately 0.81 mm.
The outer surface of each collet has a 2 degree taper; the end can has
the same taper on its inner surface. This taper allows the end can to be slid
over the collet quadrants from the collared coil assembly side. The end can
directs a radial force onto the collet quadrants which radially clamp the coils
and a set of G-10 "key" pieces. This induces a compressive azimuthal load,
generally referred to as "prestress." The splice housing, cover, and end cap
are then positioned and the latter is welded on the end can. A similar design
is used at the return end, where there are no splices.
The prestress induced during assembly is reduced during cooldown
from 300K to 4K. Lorentz forces during magnetic excitation lead to further
changes in coil stress. The magnitude of the Lorentz force is proportional to
current and to the dominant dipole field. The current is the same in the end
region as it is in the straight section. The dipole field, however, is 6.6 T in the
collared portion of the magnet and falls off to 4.8 T at a distance of 38.1 rom
into the end clamp [6]. Lorentz forces in the following model are scaled down
to 4.8/6.6 or 73% of those in the collared coil section. (All models in this paper
examine end clamp mechanics at a distance of 38.1 mm into the end clamp.
This location is far enough from the end of the collared section to eliminate
influence from the collars, but it still lies within the coil straight section.)
Figure lao End Clamp Assembly. Lead End
290
EXPERIMENTAL DATA
Measurements On Model and Full Length sse Dipoles
The clamping applied to the coils during assembly was monitored by

measuring the change in outer diameter of the end can before and after
assembly onto the coils. The change or "deflection" in diameter of the end
cans was measured on a total of twenty-six end clamps (nine full length and
four model magnets). End can diameter deflections were taken at azimuthal
locations of 0, 45, 90, and 135 degrees at a distance of 38.1 mm from the
collared section. An average diameter was also measured using a pi-tape.!
The results of the end can diameter deflection measurements are
shown in Figure 2. More detailed data for the full length magnets is given
elsewhere [7]. The average lead end vertical, horizontal, and pi-tape
deflections are 0.10, 0.04, 0.10 mm respectively. The corresponding return
end values are 0.13, 0.04, and 0.08 mm. (The uncertainty of each individual
deflection measurement is about ±0.02 mm.)
The standard deviations of lead and return end vertical, horizontal,
and pi-tape deflections for the entire set of lead and return end clamps are
0.05, 0.04, and 0.04 mm respectively. In an attempt to compensate for
variations in collet quadrant inner diameter, various amounts of radial
kapton shim, ranging from 0 to 0.25 mm, were placed between the collet
quadrants and the outer coil surfaces. Some of the large variation in end
clamp deflection may arise from the uncertainty in this procedure.
0.30
8:
-5 0.25
• &Pi·Tape
'"Po.as
~
0.20
COLLET QUADRANT r::"

0 0.15
:;l
OUTER KEY ~c::l 0.10

...
INNER KEY .s
S
as
0.05
is
0.00
324326328329 311 312 313 314 315316 317 318 318
Model Magnets Full Length Magnets

Figure lb. End Clamp Cross-Section Figure 2. Diameter Deflection Measurements
Test Installations with Pressure-Sensitive Film
The most direct measure of coil prestress inside the end clamps was
made using pressure-sensitive film. It was applied to the outer coil at the
IThe distance between the markings on a pi-tape is scaled by a factor of 1t so that when
wrapped around a circumference the number which is read is actually an average
diameter.
291
100 100
~
~
y =47.894 + 88.929>: RA2 =0.160
~ III III
.
·S + :-;::
+
..
0
0 0
+
~ 60 +
::s
~
6J
+ + +
..0'-
0
+
+
f +
~ 40 6l 40
~ f:!'"
p.,
g:,ID g:, ID
f f
.::'" 0 ~ 0
0 2 3 4 6 6 7 8 9 10 0.00 0.06 0.10 0.15 0.20
Installation Number Diameter Deflection, Pi-Tape (mm)
Figure 3a. Average Fuji Film Pressure Figure 3b. Average Fuji Film Pressure versus
Diameter Deflection Measurements
key/coil interface (pole) and to the inner and outer coils at the upper!lower
coil interface (midplane). Contact pressure and diameter deflection were
measured as the end clamp was installed eight times on the coils of model
magnet DSA322. With each installation, new pressure sensitive film was
applied. The presence of the film did not create additional prestress because
its thickness was dimensionally compensated by removing 0.13 mm (5 mils)
of azimuthal kapton. No film was applied to the inner coil pole because this
dimensional compensation was not possible at this location.
The pressure-sensitive film used was Fuji Prescale HS-H [8]. It
consists of a polyester backing coated with an emulsion containing colorant
capsules which burst under pressure, releasing a chemical which turns red
when it mixes with a developer also contained in the emulsion. The final
color of the film corresponds to the highest pressure to which the film has
been exposed. The film is scanned with a device which converts the color
intensity to a contact pressure. The manufacturer's calibration of the
scanning device (Fuji Prescale Densitometer, FPD301) was checked at
Fermilab by taking readings on tape exposed to known pressures. All end
clamp coil pressure readings were adjusted to follow this second calibration.
The measured pressures, averaged over the outer coil midplane and
pole surfaces, are plotted individually and again3t the pi-tape deflection for
all eight installations (Figures 3a and 3b). [9] The uncertainty in the pressure
readings is estimated to be ± 7 MPa (±1000psi). The average outer coil pole
and midplane stresses are 64.1 MPa (9300 psi) and 63.8 MPa (9250 psi)
respectively. A linear curve fit shows a 11.8 MPA (1700 psi) increase in
prestress for every 0.13 mm increase in end can diameter deflection. Nearly
equal average pole and midplane stresses are consistent with the simple
model of an azimuthal spring in a frictionless cavity.
Exposed films show that there are stress gradients of two kinds; large
gradients due to variation in cable insulation thickness and gradients from
the inner to outer edges of the midplane surfaces. The stress gradient across
the midplane surfaces is greater than that across the pole face; this is
predictable because the angle at the pole is less than radial and therefore the
coil has to bend less to conform a circular shape.
292
Quench Behavior
During cooldown and cold testing, no coil stress measurements or end
can defections could be made; nonetheless, quench behavior is the most
crucial measure of end clamp performance. There were only two quenches
in the end clamp regions of full length 50mm magnets. They occurred in
magnets DCA313 and DCA314. No end clamp quenches occurred in 50mm
model magnets except during high ramp rate studies[4,5].
MECHANICAL MODELS
Spring Models
Interpretation of the magnet measurements is aided by simple

mechanical models. On a small scale, the coil/key structures can be
represented as an azimuthal spring. The coils have a smaller elastic
modulus and larger length than the G-10 keys (see Table 1); therefore, they
are weaker springs which will compress more under azimuthal force.
Thin-Walled / Thick-Walled Cylinder Models

The mechanics of the entire assembly may be simplified by
representing the inner coil, outer coil, and end can as thin-walled cylinders.
For a given outer coil stress, there is a corresponding external pressure.
This pressure is scaled by the ratio of diameters and applied as an internal
pressure on the end can. For this internal pressure there is a corresponding
diameter deflection. (Calculations are simplified here by substituting the
modulus of the coils for that of the G-10 keys as well.) A more precise
relation between prestress and end clamp diameter deflection is derived by
considering the variation in azimuthal and radial stress through the width
of the coil. This effect is accounted for in equa tions which describe thick-
walled cylinders. Differences between thin- and thick-walled cylinder
predictions are shown in Figure 4. It should be noted that a diameter
deflection predicted in such a manner is an average value. As described
above, the end can actually distorts to an oval shape. However, this rough
estimate provides good correlation with measured pi-tape deflections.
Finite Element Model
Finally, a 2-D finite element model (Figure 5) further refines the

azimuthal stress contour by including the effect of the non-radial angle at the
pole and the prestress loss due to the flexibility of the end can. The relation
between outer coil prestress and end can deflection is also given in Figure 4.
The end clamp is symmetric about horizontal and vertical planes, thus
one quarter of its cross-section is modeled. Symmetry conditions are imposed
on the end can at the horizontal and vertical plane and also on the G-10 keys
at the vertical plane. Azimuthally oriented gap elements transfer force from
the coils to the key. Radially oriented gap elements transfer azimuthal
pressure from the inner coil and key to the outer coil and key, from the outer
coil and key to the collet quadrant, and from the collet quadrant to the end
can. The geometry corresponds to the cross-section at a distance of 38.1 mm
from the end of the collared coils.
The assembly process is represented by displacing the coils
azimuthally at their midplane until the desired average prestress is
293
achieved. 2 The prestress input to the outer coil is 64 MPa (9300 psi),
corresponding to the average of the outer coil pole and midplane prestress
measured with Fuji film on model magnet DSA322.
To demonstrate the validity of inducing prestress through midplane
displacements, two test cases were run: one with coils displaced dR radially
and one with coils displaced dR1t/2 at the midplane. Resulting prestress
values are within 4% of each other, which is within the precision of the Fuji
tape itself. In both cases the ratio of inner to outer coil stress is
approximately 1.6:1; this ratio is reduced to 1.5:1 when the end can and
collets are added to the model.
0.30 . , . . - - - - - - - - - - - - - - - ,
+ .DSA322 Test Installations
~ 0.25
--- Thin·Walled Cylinder.
..•.. Thick·Walled Cylinder.
Irw 0.20
- Finite Element Model.
+
~
.."~:'-;;:::::;-;;;~-:..::~:~~~:-
~ 0.15
§
0.10
~
0.00
o 40 100
Outer Coil Stress (MPa) Lx
Figure 4. Diameter Deflection versus Prestress Figure 5. 2-D Finite Element Model
Table 1. End Clamp Components - Material Properties
Material Ee Er u ex.
(N/mm2) (N/mm2) (m/mOK)
Coils
Inner 16,276 16,276 0.33 1.38.10-5
Outer 12,414 12,414 0.33 1.38.10-5
Pole Piece G-IOCR 28,000 - 0.19 1.05.10- 5
Collet G-IOCR - 10,900 0.16 3.35.10- 5
Insulation Kapton 100 100 0.30 1.38.10-5
End Can Aluminum 68,970 68,970 0.33 1.35.10-5
2 The most direct calculation of prestress could be made by measuring all

piece parts and then taking the difference between the total thickness and the
outer radius of the end can. The resulting interference could be applied to the
coils as a radial displacement. The problem with this method is that the
accuracy of the measurement is of the same order of magnitude as the total
interference.
294
Prestress Loss During Cooldown. The collet quadrants have a larger radial
coefficient of thermal contraction than the end can or coils and therefore
contract more during cooldown. This decreases the radial pressure in the
end can, and results in less clamping of the coils. In other words, stress in
the coils decreases during cooldown. The prestress loss may be estimated in
simple models by setting the change in coil radius due to prestress equal to
the change in radius of the coils during cool down «(l~TR). These equations
predict that approximately 80% of the original prestress is lost during
cooldown (49.7 and 51.7 MPa according to thin- and thick-walled cylinder
theory respectively). The finite element model shows less prestress loss, (38%
or 24.4 MPa), because it accounts for the contraction of the end can.
Prestress Loss During Magnetic Excitation. The magnetic field is three
dimensional in the end clamp region and this complicates the determination
of Lorentz forces. To simplify the problem, Lorentz forces are approximated
using the straight section coil geometry in the dominant dipole field of 4.8 T
(at 38.1 mm from the end of the straight section). Because the Lorentz forces
are non uniformly distributed in both the radial and azimuthal directions,
models based on the thin-walled and thick walled cylinder equations are less
simple to apply. The finite element model is used to predict coil stresses in a
dipole field of 4.8T. Results are plotted in Figure 6. The average outer coil
pole prestress decreases 16.5 MPa (2400 psi); the outer coil midplane stress
increases 10.8 MPa (1500 psi). The average inner coil pole prestress
decreases 25.5 MPa (3700 psi); the outer coil midplane stress increases 14.9
MPa (2200 psi).
120
----0---
Inner Coil Pole Stress
.--*--
Inner Coil Midplane Stress
100 -. --+- Outer Coil Pole Stress
CI:', - Outer Coil Midplane Stress
.,,~.....-...--.--.-.
<il 80
~
j
CI.l
60 '0.
........................
:;::: 40
0
0
20
o+-~----------------~----~
4.8T
Installation Cooldown Magnetic Excitation
Figure 6. Coil Stress at Assembly, Cooldown, Excitation
CONCLUSIONS
The end clamps on 50mm aperture sse Dipoles have performed
successfully. No quenches have occurred in the end region of magnets
whose end can diameter increased by 0.10 mm or more during installation
onto the coils. Two training quenches have occurred in magnets whose end
can diameter deflection was less than 0.10 mm and it is still not clear that
these quenches were related to lower than average prestress.
295
To understand the data from Fermilab full length and model SSC
Collider Dipole magnets, the mechanics of the end clamp assembly process
were modeled as thin-walled cylinders, thick-walled cylinders, and with a
finite element model. The results obtained, namely the end can diameter
deflection, agreed within an order of magnitude with the averaged measured
diameter deflection of nine full length and four model magnet end clamp
assemblies after installation on the coil ends. The outer coil prestress input
to all three models was 64 MPa (9300 psi), based on measurements made on
test coils instrumented with pressure-sensitive tape. For this prestress the
finite element model predicts relatively large prestress losses during
cool down and magnetic excitation. The prestress loss during cool down could
be reduced by using collet quadrants which have a smaller coefficient of
thermal contraction in the radial direction. Some effort has been made to
manufacture such collets by compression molding EM7302, and glass fiber
filled epoxy. However, part tolerances could not be maintained.
REFERENCES
1. Delchamps, S.W., et al., SSC Collider Dipole Magnet End Mechanical

Design, Proceedings of 1991 IEEE Particle Accelerator Conference,
May 6-9, p. 2185, San Francisco, CA
2. Koska, W.,et al., Tests of 40mm SSC Dipole Model Magnets with
Vertically Split Yokes, Proceedings of 1991 IEEE Particle
Accelerator Conference, May 6-9, San Francisco, CA
3. Wake, M., et al., Tests of 1.5 Meter Model 50mm SSC Collider Dipoles at
Fermilab, Proceedings of 1991 IEEE Particle Accelerator
Conference, May 6-9, San Francisco, CA
4. Lamm, M., et al., Test Results from 50mm Aperture, 15M long SSC
Dipole Magnets at Fermilab, Fourth Annual International
Industrial Symposium on the Super Collider, March 1992, New
Orleans, LA
5. Orris, D., et al., Quench Performance of Fermilab/General Dynamics

Build Full Length SSC Collider Dipole Magnets, Fourth Annual
International Industrial Symposium on the Super Collider, March
1992, New Orleans, LA
6. Orrell, D., Magnetic Analysis and Design of End-Turns, MD-TA-181,

SSCL, June 13, 1991
7. Delchamps, S., End Clamp Deflection for DCA311-DCA319,

TS-SSC 92-013, Fermilab, January 29,1992
8. Delchamps, S., Calibration of Fuji Prescale Pressure Sensitive Film,

TS-SSC 91-134, Fermilab, July 8,1991
9. Delchamps, S., Fuji Film Tests on DSA322 End Clamp Pressure,

TS-SSC 91-135, Fermilab, July 2,1991
10. Kerby, J., Finite Element Analysis of the End Collet Mechanism,
TS-SSC 90-085, Fermilab, November 6,1990
296
A CRYOGENIC SUPPORT POST FOR
SSC QUADRUPOLE MAGNETS
M.W. Hiller and J. A. Waynert
Babcock & Wilcox - Accelerator & Magnet Systems

Lynchburg, VA 24551-0785
ABSTRACT
A Cryogenic support post is being designed for the 5.4 meter long SSC Collider
Quadrupole Magnet (CQM). The dimensions of the support are being optimized using
finite element analysis (FEA) to concurrently meet the stringent mechanical, thermal, and
life cycle requirements of this application. The principal components of the post are to
be injection molded since this manufacturing method shows the most promise for being
cost effective in mass production while meeting all performance requirements. This
paper presents the details of the analysis as well as the status of the testing and
development program.
INTRODUCTION
Approximately 1700 superconducting CQM's will be produced for the SSC by

industry. The innermost portion of the CQM illustrated in Figure 1 is called the cold
mass. The cold mass contains the superconducting windings and is maintained at 4.35K
during operation. It is supported and aligned relative to the outside of the magnet
(referred to as the vacuum vessel) by two support posts. Because these posts span the
temperature range from the room temperature (3OOK) vacuum vessel to the 4.35K cold
mass, they also provide for thermal isolation of the cold mass from external heat
conduction.
Prototype SSC support posts have been developed at Fermi National Accelerator Lab
(FNAL) to meet the comparable performance requirements of the 17m Collider Dipole
Magnet (CDM)l. Even though the performance of the FNAL reentrant post appears
adequate, the high cost of its numerous tightly-toleranced piece parts and intricate
assembly procedures is unacceptable on mass produced magnets. Therefore, more cost
Supercollider 4, Edited by I. Nonte 297

MASS
...-_ _ _ _ _ .a.ppr.IL.5<t100 mom
THER~1AL
SHIELDS .pp.2400 ....
II I
d~ I
VACUUM
VESSEL
SUPPORT POST / COLD MASS
SUPPORT POST ASSEMBLY
ASSEMBLY
Side Vie'" of OLD MASS a.nd SUPPORTS

(no( {o .ac.. lfI)
Figure 1. Illustration of the Collider Quadrupole Magnet (CQM)
effective approaches for cryogenic support posts are being pursued at Brookhaven
National Lab (BNL)2, FNAV,4 , and at SSCL and its industrial associatesS,6. The
support post approach that promises the greatest cost savings utilizes an injection molded
composite (IMC) manufacturing method. This method was selected early in the B&W
CQM program to meet economic, design, and performance objectives.
MANUFACTURING
A Design-for-Manufacturing (DFM) approach is being used at B&W to design the

CQM support post. The DFM approach will reduce the part count and instill near-net
shape manufacturing technologies in order to reduce the complexity of the components
and simplify the manufacture of the magnet. The intended IMC process relies on the
softening of thermoplastics at elevated temperature so that they can be pumped into the
cavity of a mold. Final handling and finishing operations of the molded components
prior to assembly of the support post is minimal. The post was designed to be comprised
of three segments. This configuration allowed B&W to pursue a cost reducing assembly
sequence (described in reference 6) wherein the bottom segment of the post could be
installed after the cold mass assembly was slid into the vacuum vessel.
The UltemaD 2000 series of glass filled polyetherimide thermoplastics from the
General Electric Company was selected as our baseline !MC material because it exhibits
excellent material properties for this application. In addition, Ultem posts of similar
configuration have already been manufactured for BNL IS RHIC program and for
evaluation purposes by FNAL. Through a review of available literature and personal
correspondence, we compiled valuable information from their experiences with IMC
posts. In general, satisfactory support posts could be manufactured by injection molding
provided that certain design practices and process controls were adhered to. For
instance, sudden changes in wall thickness from thin to thick within the mold may result
in porosity that weakens the IMC component. Also, the locations of molded-in holes and
other features must take into account the direction from which the mold is filled,
otherwise potentially weak "knit lines" are formed in regions where the flow of !MC
material splits around a feature and then rejoins.
298
From studying the literature and the "lessons learned" at BNL and FNAL it was
concluded that good posts (those without obvious flaws) tend to fail in a similar and
systematic manner when tested to destruction in axial and transverse loading. The
mechanical failures typically occurred at the transition between the wall of the segment
and the support flange shown in Figure 2. This location was later identified as the
highest stress region through FEA and is slated for significant redesign and optimization.
segmen t
f i He t betwe n wall and flange
~---- i;y piCA fa i]1.II re 10 a ti on
Figure 2. Typical Mechanical Failure Mode of a Support Post Segment
ANALYSIS APPROACH
The ANSYSGD FEA software and hand calculations were used to "size" the segments
of the CQM support post to meet the performance requirements summarized in Table 1.
Table 1. CQM Structural and Thermal Requirements7 for CQM with Two Support Posts
CQM Structural Load Cases for Shipping and Handling (assuming 1900 kg (4200 lb) cold mass)
3.0 g vertical down results in: 2850 kg (6,300 lb) compression per post
1.0 g vertical up results in: 950 kg (2,100 lb) tension per post
1.5 g axial results in: 2850 kg (6,300 lb) side load per anchor post (no
cradle rotation) + 950 kg (2,100 lb) compressive
1.0 g lateral results in: 950 kg (2100 lb) side load per post (cradle free to
rotate) +950 kg (2,100 lb) compressive
Budgeted Thermal Requirements (conductive heatleak) per Support Post

80K intercept: 3.16 W
20K intercept: 0.48 W
4.35K intercept: 0.032 W
Radiation at SupPOrt Post

0.0003 Mgrey and TBD x 10 12 neutrons/cm2
299
We have performed structural and thermal analyses of the B&W baseline and the
BNL and FNAL IMC posts. The results have been compared to actual test data where
available. When comparing the FEA predicted performance to actual test data the
following observations were made:
1) The thermal FEA can over-predict the conductive heatleak by as much as 20% - a
plausible explanation is that the thermal contact impedance between the components
along the heatleak path is difficult to quantify in the high vacuum and cryogenic
environment so it is typically neglected (therefore, the components are assumed to be in
intimate contact).
2) The structural FEA is typically not conservative due to:
a) The dynamics of the molding process can cause variations in the orientation and
dispersion of reinforcing fibers. This results in heterogeneous and anisotropic
material properties within the component.
b) The sensitivity to manufacturing process parameters can be high - processes

such as the preheat of precursors, the temperature and pressures at injection, and
any post-treatments (curing) if necessary must be tightly controlled.
c) Erroneous FEA stress results can occur due to the intrinsic material property
mismatches within the system. For instance, a property mismatch occurs between
both the resin and the fibers of the IMC, and between the IMC and metallic
interface components (such as fasteners, flanges, and backing rings in the support
post assembly).
d) Failure criteria used during preliminary FEA analysis must sometimes rely on
published mechanical properties such as "flexure strength" which are not directly
related to the complex 3-D stress state that exists in actual components when under
load.
e) Published material properties may differ from test data obtained from coupons
that have been cut from molded components.
DESCRIPI'ION OF FEA MODEL
An ANSYS® parametric language routine named MAKEPOST.PAR was written to

automatically generate FEA support post models. The benefit of parametric modeling is
that it is easy to create many variations of a design relatively easily. This flexibility is
especially important when geometry, materials, and load cases relating to the design of
the component are subject to change. The MAKEPOST routine quickly creates 2-D or 3-
D structural and thermal support post models. The analyst sets flags prior to the
execution of MAKEPOST to define the desired mesh type and boundary conditions.
A standardized geometry input file format was developed to easily define the
complex shape of the support post models. Since a common geometry database is
utilized for both thermal and structural analyses, the integrity of the support post
configuration is insured throughout each study. The geometrical definitions include; the
300
size of the top and bottom of each post segment, the height of each segment, the wall
thicknesses, the fillet radii, the thickness of shield intercepts, and a flag that determines
the orientation of the SOK flange (radially inward or outward). A limitation of the
current MAKEPOST routine is that only segments of uniform wall thickness can be
automatically generated. In reality, this limitation may be prudent since transitions from
thin to thick walls typically result in porosity within the IMC component.
The robustness of the parametric routine was verified by generating the FEA models
shown in Figure 3 as well as variations of each.
-
FNAL TMC support Post
,.
RHiC fMC Support Post B& W Baseline Support Post
Figure 3. Examples of Parametrically Generated Support Posts
Two post-processing macros were developed to automate the interpretation of FEA

results. The first macro is utilized for structural models. It calculates a figure of merit
(FOM) which is defined as the ratio between the peak mechanical stress in the segment to
the nominal stress value within the wall. The location of the peak stress occurs at the
fillet between the wall of the post and the flange. This is the location where failure has
repeatedly occurred in support post tests at BNL and FNAL. The FOM is used to
understand and compare the mechanics of various post configurations. FOM results for
various fillet radii are illustrated in Figure 4 for variations of the BNL, FNAL, and
B&W support post designs.
It can be seen that the peak stress in the segment is at least a factor of three greater
than the nominal wall stress for all design variations. In addition, the thinner walled
segments of the BNL and B&W designs are more sensitive to the value of the fillet radius
than is the thicker FNAL segment. This is due to the development of a significant
bending moment (and bending stress) in the fillet region during typical loading. This
large bending moment is mitigated by the increased area moment of inertia of the thicker
walled FNAL segment. During the B&W development program, a more sophisticated
transition from the wall of the post to the flange will be investigated to minimize the
resulting stress.
The second macro was used to interpret thermal FEA results. It performed
mathematical operations on the nodal heat fluxes midway up each post segment and
calculated the conductive heatloads to the 4.35K, 20K, and SOK thermal stations of the
support. These values are easily compared to the requirements to evaluate the
acceptability of a support post configuration.
301
8
--------- B&W Baseline 4.32mm wall thk.

7
-0--- Lipski IFNALI 6.35mm wall thk.
6 --+--- RHIC 4.76mm wall thk.
FOM 4
o +--------+-------1--------+-------~------_+--------~----~
o 2 4 6 8 10 12 14
Radiu. of fillet let centroid. axis of ••gment wall) Imm)
Figure 4. Figure of Merit (POM) vs. Radius Fillet for BNL, FNAL, and B&W Posts
The POM for a segment of a support post is defined as the ratio between the peak
stress and the nominal wall stress.
The final dimensions of the CQM post will be optimized within ANSYSGD using the
ANSYSGD optimization module. In effect, the ANSYSGD program will be given the
freedom to vary certain design parameters (such as wall thicknesses or fillet radii) within
specified ranges in order to minimize an objective function such as the FOM or heatleak:.
In this manner the computer will automatically create numerous variations of the post
and will report back the best design relative to the specified objectives.
TESTING AND DEVELOPMENT STATUS
A prototype post has been designed based on material properties provided by the
General Electric Co. and BNL8. A Prototype mold will be built in June of this year so
that the B&W design can be verified by component testing. A thorough test and
development program is being planned to fully characterize candidate !MC materials for
this application. Mechanical and thermal testing will be performed on prototype post
segments and "coupons" removed from segment walls. The material properties in the "as
molded" configuration will be fed back to the design and analysis process. The results of
these studies will be used to verify and correlate the FEA analysis to the performance of
actual support posts.
Due to the life cycle requirements and operating environment of the support post, the
creep and radiation resistance of the IMC material must be fully characterized through
accelerated life testing. These test results and further FEA optimization will be used to
iterate the geometry of the post as necessary and to develop a comprehensive technical
specification so that a well characterized and proven support post will be available in
early 1993 for CQM prototype manufacturing.
302
SUMMARY
The B&W FEA results compare favorably with published test and analysis data from
BNL and FNAL. Design practices conducive to IMC components have been adopted and
sensitivities to process variables have been identified. A versatile ANSYSGD FEA tool has
been developed so that the geometry of the post can be further studied and optimized thru
parameterization studies. This tool readily creates meshed support post models for
thermal and structural FEA. This parametric tool for creating support post models could
readily be used for the design and analysis of support posts for other SSC cryogenic
components.
ACKNOWLEDGMENTS
Jack Sondericker, Tom Nicol, and Arie Lipski are thanked for providing valuable data
and technical discussions. Greg Lehmann is recognized for reviewing the manuscript and
providing ANSYS® expertise.
REFERENCES
1) T.H. Nicol, R.C. Niemann, and J.D. Gonczy, Design and Analysis of the SSC
Dipole Magnet Suspension System, in: IISSC-Supercollider 1, Plenum Press, NY
(1989).
2) J.H. Sondericker, Alternate Concepts for Structurally Supporting the Cold Mass of a
Superconducting Accelerator Magnet, in: IISSC-Supercollider 3, Plenum Press,
NY (1991).
3) T.H. Nicol, SSC 50mm Collider Dipole Cryostat Single Tube Support Post
Conceptual Design and Analysis, SSC Report No. SSCL-N-765, July 9, 1991.
4) A. Lipski, T.H. Nicol, and R. Richardson, Alternate Design Concept for the SSC
Dipole Magnet Cryogenic Support Post, in: IISSC-Supercollider 3, Plenum Press,
NY (1991).
5) V.L. Morris, Advanced Composite Structures For Superconducting Super Collider,
in: IISSC-Supercollider 1, Plenum Press, NY (1989).
6) M.W. Hiller et al, SSC Quadrupole Magnet Cryostat Design Alternatives, in:
Advances in Cryogenic Engineering, Vol. 37A, Plenum Press, NY (1991).
7) Collider Quadrupole Magnet System Specification, SSCL Doc. No. M80-000007,
May 17, 1991.
8) L.J. Wolf, Tensile Tests for Quality Control of Injection Molded Composite Posts,
BNL-RHIC Doc. No. AD/RHIC-78, August 1990.
303
TRANSIENT mERMAL AND STRESS ANALYSIS OF A CQM COLD MASS
X. Huang, Y.M. Eyssa, and J.A. Waynert
Babcock & Wilcox

P. O. Box 785
Lynchburg, VA 24505
INTRODUCTION
During cooldown or warm-up of the collider quadruple magnets(CQM) cold mass,

it is anticipated that large temperature gradients and consequently thermal stresses be
generated as a result of the cold mass sudden temperature change. The magnitude of
these stresses or temperature gradients are mainly function of the helium gas flow rate
and its initial temperature. The position of the CQM relative to the collider dipole
magnets(CDM) is not known to be at the beginning or the end of the cooling wave.
The cooling or warming gas inlet temperature for the CQM will depend on its position.
There is also the possibility of ramping the helium gas inlet temperature to avoid a
sudden change in cold mass temperature. To calculate the coil stresses, we need to
obtain a 2 D temperature distribution of the cold mass crosssection at different axial
positions. We have constructed an almost 3 D model that compute the temperature
distribution over the CQM crosssection at different axial positions and times. Following
the temperature calculations, the model estimates the thermal stresses in particular the
shear stresses in and between the coils.
TRANSIENT mERMAL MODEL
The analysis is for the cool-down and warm-up of the collider quadrupole magnets
(CQM). The entrance helium temperature and pressure are fixed and treated as input
in our analysis. Initially we have used a I-dimensional model which has been
developed and used to analyze the cool-down process. However, the I-dimensional
analysis does not give magnet transverse temperature distribution during cool-down,
which is very important for magnet survivability. A more detailed 3-dimensional

analysis is required. A 3-D finite element model is possible, but not necessary. Our
simple 3D approach is as follows. First, we construct 1 D (axial) model for each
separately lumped components of the cold mass. As shown in fig. 1, the cold mass is
divided into 16 lumped masses (stainless steel skin, 4 yoke parts, 9 collar sections and
2 coil regions) that are repeated at equal axial positions. These masses are connected
to each other through carefully calculated temperature dependent thermal resistances
that include all insulation and contact resistances. Each mass is assigned a temperature
dependent heat capacity. We use the helium temperature results (at different sections
and times) from the lst step to perform a simplified 3-D thermal stress analysis to
determine the survivability of the magnet.
Figure 1. CQM cold mass cross-section is divided into lumped parts.
A. Cold Mass Thermal Conduction Equations
The heat conduction equations in the axial direction for mass i (iron 0-3, ss 1-
7,coil 1 & 2, skin) is ,
(1)
where, A is the cross-sectional area, S is the volumetric heat capacity, k is the (axial)
thermal conductivity, and T is the temperature. The summation is done over all
adjacent masses j, including helium. Rij is the transverse thermal resistance ,
306
(2)
where, dij is the dimensionless thermal resistance from mass i to the ij interface, dji is
the dimensionless thermal resistance from mass j to the interface, gij and Cij are
interface gap distance and length, and ki kj and kg are thermal conductivities of
material i, material j and gap material. Between coil 1 and coil 2 and between coils
and stainless steel collar, the gap material is fiber glass epoxy. Between collar and
yoke and between iron yoke and skin, the gap material is helium gas. The resistance
between mass i and helium flow is given by
1
+-- (3)
he Wi
where hc is helium heat transfer coefficient and Wi is the wetted length of mass i.
The iron yoke and stainless steel collars are laminates. Their effective
conductivities as modified by a contact resistance.
The coils are made of Cu/NbTi wires with some voids. Their effective axial and
transverse thermal conductivities are modified, however their values are still high
compared to other thermal conductivities.
B. Helium Flow Calculations
The Helium flow rate will be divided between the central channel and two bypass
yoke channels. Over 96 % of the flow will pass through the much less hydraulic
resistant yoke channels. The helium is supercritical helium at 4.35K and 10 atm. inlet
temperature and pressure. The heat transfer co-efficient is calculated using a modified
Dittus-Boelter expression 1,2 .
(4)
Where The is the mean helium temperature and T s is the cooling surface temperature.
Temperature dependent thermal and mechanical properties are used in this
analysis 3 ,4,5.
STRESS ANALYSIS MODEL
To determine the survivability of the magnet during rapid cool-down or warm-up

process, we need to know the thermal stresses, especially the shear stresses in and
between the coils. At this time, we include a simplified thermal stress analysis in the
307
program. Our future plan is to do a detailed 3-D modeling using the ANSYS program.
Based on reasonable assumptions and simplifications, the analysis provides quick and
approximate stress solutions, which can be used to determine the survivability question.
Furthermore, the thermal stresses may be very small. If this is true, we. do not need a
very accurate analysis. The assumptions and simplifications are as follows:
1. Thermal stresses in the iron yoke and stainless steel are not of concern. Especially
since they are laminated, these stresses are small.
2. Shear stresses at the collar-coil interfaces are small and can be neglected. Note
that the collar is laminated and designed to be able to slide, it can do very little (if
any) to prevent coil from moving axially.
3. Coil deformation is axissymetric. There is no circumferential displacement.
Rradial and axial displacements are independent of circumferential position.
4. The stainless steel collar is much stiffer than the coils so that it forces the coil
outer surface to move radially as it shrinks.
RESULTS
A. Cool-down (300 - 80 K)
A cool-down analysis is performed between 300 and 80 K for 50 and 100 g/s
helium flow rates. In this analysis, we show the axial temperature gradient for the
different parts of the cold mass components (yoke, collar, and coil) and the maximum
transverse temperature difference between the inner and outer coils. Though the axial
temperature difference are high at the beginning of the cool-down process, they do not
cause any significant stresses. It is the transverse temperature difference between the
inner and outer coil that is of concern especially during warm-up as will be shown
later. Figure 2 shows the maximum axial temperature gradient for the 100 g/s flow
rate. Results for 50 g/s show slightly lower rates. Results also show a maximum
temperature difference between outer coil and collar of 3.7 and 2.6 K for the 100 and
50.g/s flow rates. A much smaller temperate difference between the coil layers exists.
Coil maximum radial and hoop stresses for both the 100 and 50 g/s flow rates were
found to be small. Initially, the SS collar is colder than the coil and it shrinks more;
therefore the radial and hoop stresses initially increase slightly from their
precompresion values(-22 MPA for radial and -54 MPA for hoop stresses) but later
these stresses decrease as the copper coils shrink more than the SS collar releasing
almost all of the precompression stresses. However it is expected that part of this loss
will be gained back during the 80 - 4 K cooling as the collar will shrink more than the
coil. Axial and shear stresses are found to be small for both flow rates.
B. Warm-up (4.2 - 300 K)
The analysis discussed above is repeated for warm-up using 100 g/s helium gas
flow rate. The temperature gradient and differences here are much higher than those in
the cool-down process due to the small values of specific heat at low temperatures.
Axial temperature gradient is shown in Figure 3. Coil to collar and coil to coil
temperature differences on the order of 50K occur at the beginning of the warm-up
process.
308
40 ,--------------------------------------.
FLOW RATE = 100 g/s

HELIUM PRESSURE = 10 atm
30
ss collar
.. "~
.. --_ .. - ..... ......
20 .'
iron yoke
.,..----
10 .-
, .- .-
,: '
:1
/
Y"noornn"rnnornnoTnnnornn"rn"rnnnTTrM
04T
o 1000 2000 3000 4000 5000
TIME, t (sec)
Figure 2. Maximum axial temperature gradient during 300-80K cool-down.
120
FLOW RATE 100 g/s =

100 HELIUM PRESSURE = 10 atm
----
S 80
...........
::.::
M 60
J N
"0
40
...."0
...........
iron
/ , ...... - ........ " "'
20 / .. ......... ~
~,'
O ~~~~~~::~~~~
o 1200 2400 3600 4800 6000 7200
TIME, t (sec)
Figure 3. Maximum axial temperature gradient during 4.2-300K warm-up with 0 sec ramp time.
309
The warm helium in beam tube annulus hits the cold inner coil at the inlet, causing
a sharp temperature rise. This very local temperature rise· causes a big radial stress
increase (negative sign indicates compression). The expansion of the warm part of the
coil is prevented by the rest of the coil and the ss collar that are still cold. Small
thermal expansion coefficient at low temperature helps but can not compensate the big
temperature difference due to primarily very small thermal· inertia. The radial
compressive stress hits a maximum value of 66.04 MPA (9.57 ksi) and then decreases
back to its pre-compression value(-22 MPA). This 9.57 ksi radial compressive stress
should not cause any concern. It pushes the inner coil against the outer coil and ss
collar which certainly can take this stress.
The cases for hoop and axial stresses are similar. The only difference is that the
axial stress, after hitting its peak value of -73.8 MPa (10.7 ksi), decreases to zero
because there is no initial axial stress in the coil. There should not be any problem for
the 10.7 ksi compressive axial stress in a very local region. The problem is the
compressive hoop stress. It has a peak value of 224.2 MPa (32.5 ksi). Another key
question is about the shear stress of the insulator between inner and outer coils which
has a maximum value of 5.888 MPA (853 psi).
C. 4.3-300 K Warm-up by 100 g/s Helium Flow with 1800 sec Ramping
A warm-up with 1800 sec ramping is calculated. Helium inlet temperature is

increased linearly from 4.3 K at t=O to 300 K at t= 1800 sec. The temperature
gradients and temperature differences are shown in Figure 4.. Maximum temperature
differences between coil sections are reduced by factor of 3 to about 14 K between the
coil layers and 13 K between the collar and the coil outer layer.
Stresses are shown in Figures. 5,6 and 7.The maximum radial stress is 42.51 MPa
(6.16 ksi), the maximum hoop stress is 128.2 Mpa (18.6 ksi), and the maximum axial
stress is 29.11 MPa (4.22 ksi). All of them are in compression. The maximum shear
stress at inner-outer coil interface is 2.364 MPa (343 psi).
80 ,----------------------------------,
FLOW RATE = 100 g/s
HEUUM PRESSURE = to alm
,....., 60
S
..........
::.:::
•6 40
~
N
'0
.......... iron yoke
!; 20
-----
O~~,,~~,,-.,,~~,,_r,,~~,,~
o 1200 2400 3600
TIME . t (sec)
Figure 4. Maximum axial temperature gradient during 4.2-300 K warm-up with a 1800 sec ramp time.
310
0
-20 FLOW RATE =

100 g/s
.-.. -40
--
~
p...
~
-60
CI.l
CI.l
f.;I;:l
~
-80
E-<
CI.l
-100
-120
-1404-rT.-,,-r"-'''-'''-'''-'''-'''-'''rT'-~
o 1200 2400 3600
TIME, t (sec)
Figure S. Maximum hoop stress during 4.2-300 K warm-up with a 1800 sec ramp time.
o~~--------------------------------~
FLOW RATE =
100 g/s
-5
~-10
p...
~
---- -15
CI.l
CI.l
~ -20
E-<
CI.l
-25
-30
-354-~'-~'-~'-rT.-~~-r~-r~.-~.-~~
o 1200 2400 3600
TIME, t (sec)
Figure 6. Maximum axial stress during 4.2-300 K warm-up with a 1800 sec ramp time.
311
3.0
FLOW RATE =
100 g/s
HELIUM PRESSURE =
10 atm
2.5
.--.
~ 2.0
:::2l
'--"'
~ 1.5
~
0::
E-o
Cf.l 1.0
0.5
0.0
0 1200 2400 3600
TIME, t (sec)
Figure 7. Inner-outer layer maximum shear stress during 4.2-300 K warm-up with a 1800 sec ramp
time.
CONCLUSION
The inclusion of central beam tube annulus helium in the analysis is important. It
does significantly change local inner coil temperature and thermal stresses.
The calculation confirms that the axial temperature gradient is not an important
factor but is the transverse temperature gradient. Furthermore, a smaller flow rate
cooling or heating has smaller stress problem than a larger flow rate case.
Cool-down seems safe. In both 100 gls and 50 gls flow rate case, the thermal
stresses are small. Warm-up may have a stress problem if we hit the cold mass with
300 K helium. Two ways to reduce the stresses are: 1) reduce the flow rate; or 2)
ramp helium temperature
REFERENCES
1. F. W. Dittus and L. M. Boelter, University of California, Berkeley, Publ. Eng.

Vol. 2, p. 443 (1930).
2. M. F. Taylor, NASA-TND-4332(1969).
3. S. W. Van Sciver, "Helium Cryogenics", Plenum Press,New York (1986).
4. H. M. Rosenberg, "The Thermal Conductivity and Thermal Expansion of Non-
metallic Composite Materials at Low Temperatures", Vol. 2 ,Plenum Press, New
York(1982).
5. R. P. Reed and A. F. Clark, "Materials at Low Temperatures", American Society
for Metals(1983).
312
STRUCTURAL DYNAMIC ANALYSIS OF mE SSC 40 MM COLLIDER
DIPOLE MAGNET UNDER TRANSPORTATION LOADS
A. R. Jalloh, E. Daly, and R. Viola

Dallas TX 75237-3946
ABSTRACT
The Superconducting Super Collider (SSC) magnets will be transported to the tunnel
site from various manufacturing locations in the United States. Dynamic loads during
transportation could damage the magnets. As part of the overall magnet development
program, the effects of transportation loads on the magnets should be investigated to ensure
that design limits are not exceeded. A validated structural finite element model of the SSC
40-mm collider dipole magnet (CDM) was excited with actual road input data measured
during transport. The test data was in the form of power spectral density (PSD). A random
vibration analysis was carried out to determine system response at selected points on the
structure. The results showed very good correlation with measured data. Dynamic stresses
were calculated to try to understand the mechanism that caused the magnets to yield during
shipment. The stresses observed were very small compared to the design limits of the
vacuum vessel, where permanent deformation was observed after shipment. It was
concluded that the yielding was not due to the transportation loads but to another
mechanism that was not included in the model.
INTRODUCTION
The SSC magnets will be transported by road to the tunnel site in Waxahachie, TX,
from various manufacturing locations in the United States. The magnets will have to be
designed to survive transportation intact, otherwise they will not be acceptable for
installation in the SSC tunnel. Thus, a thorough understanding of the impact of the
transportation loads on SSC magnets is an essential component of the overall magnet
* Operated by the Universities Research Association, Inc., for the U. S. Department of Energy under Contract
No. DE-AC35-89ER40486.

An on-going research and development effort in the the Magnet Systems Division of
the SSC Laboratory is investigating transportation loads and their effects on magnet design.
As part of this effort, the design B 40-mm collider dipole magnet was instrumented with
accelerometers, loaded on a flat bed trailer, and driven from the Fermi National Accelerator
Laboratory (FNAL) in Batavia, IL, to Brookhaven National Laboratory (BNL) in Upton,
NY, over roads and highways typical of those on which the tunnel-bound magnets will be
transported. The magnet was mounted on inflatable rubber cushions. Accelerometers were
placed at 14 points on the cold mass, vacuum vessel, and trailer bed. The accelerometer
locations on the magnet correspond to the locations where large dynamic amplitudes are
expected. Displacement time histories at the accelerometer locations were recorded during
transit) These were subsequently fed into an HP3567 A Spectrum Analyzer and converted
to PSDs. A typical PSD is shown in Figure 1. The PSD defines an upper envelope of the
distribution of vibrational energy incident at each accelerometer location.
DD2BTB - TRAILER - VERTICAL - LEAD

Power Speclrum Ch... 2 Avg-20571
Figure 1. Representative PSD Curve for Trailer Bed.
It should be noted that the range of interest for transportation loads is approximately
0-50 Hz. Dynamic displacement is inversely proportional to the square of the frequency of
interest, as shown in Eq. (1). Dynamic amplification for sinusoidal motion can be expressed
as:
d= Age (1)
(21tj)2 '
where d is dynamic displacement, A is acceleration amplitude normalized to 1 g, gc is

consistent units for acceleration due to gravity, and f is the frequency of interest.
Consequently, small amplitudes of acceleration resident at frequencies above 50 or 60 Hz
yield very small dynamic displacements. An inspection of the test data showed a drop-off
and leveling out of the PSOs after 15-20 Hz. As a result, the range of interest considered in
the present study is 0-30 Hz.
A finite-element model of the COM has been developed and validated with test data
by Jalloh et al. 2 and was used to predict beam tube displacements due to ground motion
excitation. Good correlations were obtained between natural frequencies and corresponding
mode shapes predicted by the model and those obtained experimentally. This model is a
very useful tool for all types of dynamic studies, ranging from further ground motion studies
to handling and transportation studies. Benda et al.3 investigated the stresses in the SSC
314
dipole magnet that resulted from transportation loadings. However, that work was not
validated with test data. The validated dynamic model for the 40-mm CDM developed by
lalloh et al. 2 was the basis of the present work.
For the present study, the inflatable rubber cushions were added to the model to
represent the structural dynamic system as it is mounted on the trailer bed. Natural
frequencies predicted by the augmented model were compared with those obtained
experimentally. This was done to establish that the model adequately represented the
dynamic system. The output PSDs of the trailer bed were then input into the model as
forcing functions, and a random vibration analysis was carried out. Mean square
displacements and their I-sigma deviations were obtained at points on the vacuum vessel
and cold mass corresponding to accelerometer locations. Also, response PSOs were
obtained at these points. These results were compared with test data.
After shipment, the vacuum vessel was observed to have deformed permanently. A
dynamic stress analysis was carried out to try to understand the mechanism that caused the
yield. However, because of software restrictions and limitations, the stress field could not be
computed directly in the random vibration analysis. Instead, a response spectrum analysis
was carried out with input data modified from the trailer bed PSO to determine the stress
field. The dynamic stresses predicted by the model were very low compared with the yield
strength of steel, the vacuum vessel material.
MODEL DEVELOPMENT AND ANALYSIS
The complete 40-mm COM and inflatable rubber cushions, as they were mounted on
the trailer bed, were modeled for this study. The cold mass model accounted for the
structural and dynamic characteristics of the magnet only. Thermal and magnetic effects
were neglected because they were not considered relevant to the objectives of this work.
The inflatable rubber cushions were modeled as three-dimensional helical springs. The
model also included the trailer bed as the source of dynamic excitation. However, it was
dynamically decoupled from the magnet and rubber cushion system.
The model was developed using the finite-element method and ANSYS, a commercial
finite-element software package. 4 The magnet was modeled with thin-shell and thick-plate
finite elements using ANSYS elements Stif63 and Stif43, respectively) The inflatable
rubber cushions were modeled by ANSYS element Stif14, the three-dimensional
longitudinal spring-damper element without torsion. The springs were anchored at the top to
the external feet of the magnet and at the bottom to a large mass that simulates the trailer
bed. The trailer bed was modeled with ANSYS element Stif2l, a mass element with three
translational degrees of freedom. The trailer bed was given an excessively high mass value
to ensure that its dynamic characteristics were decoupled from that of the magnet and rubber
cushion system. The mass of the trailer bed was made I x lOS times larger than the
combined mass of the magnet and rubber cushion, as recommended in the ANSYS software
users' guidelines. With this large mass anchor, the magnet was not restrained during the
dynamic analysis, as is required in ANSYS for the random vibration analysis that was
ultimately carried out. This, of course, resulted in rigid body modes. However, the non-rigid
body modes were not affected by the dynamics of the truck bed. The decoupling of the
modes was verified by carrying out two separate free-vibration analyses of the system. The
first analysis was with the large unconstrained masses attached to the springs. The second
analysis was with the large masses removed and the bottom ends of the springs constrained.
The natural frequencies and mode shapes of the non-rigid body modes from the first
analysis were the same as those from the second analysis.
After the free vibration analysis was completed, a harmonic analysis was carried out to
compute a transfer function for use in the random vibration analysis. The PSO of the trailer
315
bed in the vertical direction, shown in Figure 1, was discretized and input as the forcing
function in the random vibration analysis. This PSD is for the vertical direction only. It
should be noted that the test data showed that only the vertical direction excited
significantly during transportation. Structural damping of 2% for the magnet components
and 15% for the springs was assumed. The value of 15% for the springs was obtained from
the literature. 5
Output displacement PSDs were generated from the analysis for points on the magnet
where accelerometers were located. The results of the random vibration analysis were
consistent with test data.
In an attempt to understand the mechanism that caused the magnets to yield during
shipment, a dynamic stress analysis was carried out. The current version of ANSYS does
not lend itself to the direct computation of the stress field in random vibrations analysis.
Therefore, a response spectrum analysis was carried out using the trailer bed PSD shown in
Figure 1 as the input spectrum. However, because of software restrictions and limitations,
the input had to be modified to a constant amplitude PSD. Also, the displacement results
showed that the effects of the transportation loads were significant only in the low
frequency range. The modified input spectrum was 11 x 10-3 g 2/Hz over a frequency range
of 1-10 Hz.
RESULTS
The natural frequencies of the vertical modes predicted by the magnet and rubber
cushion system model and those obtained by test were compared to ensure that the model
adequately represented the dynamic characteristics of the structure. This comparison, shown
in Table 1, reveals very good correlation between the analytical and experimental results.
The fundamental frequency showed complete agreement. In general, there was agreement
within 8% for all modes.
Table 1. Comparison of Analytical vs. Experimental Frequencies for the 40-mm

Collider Dipole Magnet in the Vertical Direction.
Analytical Results Experimental Data

Vertical Mode Number Natural Freq (Hz) Natural Freq (Hz)
1.498 1.5
2 5.93 5.5
3 6.93 6.5
4 9.85 10.0
The mean square displacements and the I-sigma standard deviation for the lead and
return ends of the vacuum vessel and cold mass are shown in Table 2. Experimental data for
the mean square displacements for the cold mass lead and return ends were not available.
Again, there is good correlation between model results and experimental data.
The output displacement PSDs for the four locations on the structure are shown in
Figures 2-9. There is good correlation between the predicted PSD and test data for each
location on the structure. Also, the maximum dynamic displacements occur in the low
frequency ranges. The response spectrum analysis showed low dynamic stresses in the
magnet. The maximum values were about 1000 psi, in the area where the center support
316
Table 2. Comparison of Analytical vs. Experimental Mean Square Displacements
for the 40-mm Collider Dipole Magnet.
Analytical Mean Square Expt. Mean Square

Location Disp (in.2) Displacements (in. 2 )
Vac Vessel Lead End 2.58 (1.61)** 2.86

Vac Vessel Return End 2.54 (1.59) 2.31
Cold Mass Lead End 3.09 (1.76) NA
Cold Mass Return End 1.83 (1.35) NA
**The numbers in parentheses represent the I-sigma standard deviation of the
analytical mean square displacements.
post attaches to the vacuum vessel. This value is extremely low compared with the yield
strength of steel, the vacuum vessel material. Of course, it should be noted that the stress
field computation was carried out with a modified PSD because of software restrictions.
Thus, the stress field was not computed accurately. However, the stress values obtained
indicate that even if the proper analysis had been carried out, the stresses would not have
approached the yield threshold.
Diap.PSD Chan 3 Avg-21728

3.2 ,
;:;r~
°1 Funcln Lin Hz
X: 1.5Hz Y: 2.86436 in
Figure 2. Displacement PSD for Vacuum Vessel Lead EndlTest Data.
ANSYS 4.4Al
f"E8 13 1992
14:58:33
PLOT NO.
VALU POST26
2.5 AMPLITUDE
ZV -1
2.25 DIST-O.6666
XF -0.5
YF -0.5
ZF =0.5
1. 75
1.5
1.25
0.75
0.5
0.25
'----T"----,--.---r--,----.---r---""9"......'r--o FREQ
40
12 20 28 3.
vacuum vessel lead end response pad
Figure 3. Displacement PSD for Vacuum Vessel Lead End/Analytical Results.
317
Disp. PSD Chan 2 Avg=25027
2 i
:":t$F~~=~-I=r=f-t=Fj
1 Functn Lin Hz <II Marker.. 40
X: 1.75 Hz Y: 1.45592 in
Figure 4. Displacement PSD for Vacuum Vessel Return Endffest Data.
1
ANSYS 4.4Al
FEB 13 1992
14:58:35
PLOT NO. 3
VALU POST26
2.5 AMPLITUDE
ZV =1
2.25 DIST=0.6666
XF =0.5
YF =0.5
ZF =0.5
2
1. 75
1.5
1.25
0.75
0.5
0.25
FREQ
0
0 40
4 12 20 28 36
vacuum vessel return end response psd
Figure 5. Displacement PSD for Vacuum Vessel Return End/Analytical Results.
Disp. PSD Chan 3 Avg=20529

3.2 r i
P~:~~=F-=1=:~$:::3
1 Functn Lin Hz <II Marker.. 40
X: 1.5 Hz Y: 2.80309 in
Figure 6. Displacement PSD For Cold Mass Lead Endffest Data.
318
ANSYS 4. 4Al
FEB 13 1992
14: 58; 35
PLOT NO.
POST26
AMPLITUDE
ZV -=1
3.6 DIST-O.6666
XF -0.5
YF -0.5
ZF -0.5
3.2
2.8
2.4
1.6
1.2
0.8
0.4
L -_ _~=--, __-,__-,____,-__.-__,-~~~~__-, >REQ

40
12 20 28 36
cold mass lead end response pad
Figure 7. Displacement PSD for Cold Mass Lead End!Analytical Results.
2r. Disp. PSD

!
!
,
!
!
!
Chan 4
!
!
!
I
!
I
!
I
Avg..25027
!
!
!
!
OVLD
~I[=~
o
1
I ! !
Functn Lin Hz
! !
.. Marker
I ! ! !
40
X: 1.75 Hz Y: 1.56845 in
Figure 8. Displacement PSD for Cold Mass Return Endffest Data.
ANSYS 4. 4Al
FEB 13 1992
14:58:35
PLOT NO.
VALU POST26
AMPLITUDE
ZV =1
1.8 DIST-O.6666
XF -0.5
YF =0.5
1.6 zrr -0.5
1.4
1.2
0.8
0.6
0.4
0.2
12 20 28 36
cold mass return end response psd
Figure 9. Displacement PSD for Cold Mass Return End/Analytical Results.
319
CONCLUSIONS
The model accurately represented the transportation environment experienced by the

40-mm CDM, confirming that it is a useful tool for these types of studies. Computer
predictions of dynamic displacements compared well with experimental data. However,
initial computer predictions of dynamic stresses were very low compared with the yield
strength of the vacuum vessel steel. The software calculates dynamic stresses using the
dynamic displacements in combination with the stiffness matrix. This suggests that another
mechanism that was not included in the model caused the deformations seen in the vacuum
vessel. A possible candidate is stress relief of the welds that join sections of the vacuum
vessel pipe to the reinforcing rings. If these welds were already pre-stressed, lower dynamic
stresses could possibly drive the material in the weld to yield conditions.
FUTURE WORK
This work was investigative in nature, and it proved that a validated model is useful. A
validated model of a 50-mm CDM as well as other types of magnets should be developed
using this approach to complement research and development efforts on all magnet
development programs. Also, the stress field should be determined for the entire magnet
system under random transportation loading. As was pointed out, the current version of the
ANSYS software does not have this capability. It is expected that the next version will lend
itself to this type of analysis. If that proves correct, then an elaborate stress field
computation should be pursued.
REFERENCES
1. E. Daly, "Transportation Studies: 40 mm Collider Dipole Magnets," to be presented at the Fourth IISSC,
New Orleans, LA, 1992.
2. A. R. Jalloh, R. Viola, and E. Daly, "The Mechanical Response of the SSC Dipole Magnet to Ground
Motion," Proceedings of the Third lntemationallndustrial Symposium on the Super Collider,
pp. 1037-1044, 1991.
3. B. J. Benda, R. D. Campbell, D. J. Doyle, and J. J. Johnson, "ssc Dipole Magnet System: Stress Analysis
for Seismic and Transportation Loadings," EQE Engineering, January 1991.
4. G. J. Desalvo and R. W. Gorman, Ansys Engineering Users Manual, Swanson Analysis Systems, Inc.,
May 1989.
5. Industrial Engineering Manual, Goodyear Tire Company, p. 56.
320
ANALYTICAL STUDY OF THE THERMAL INSULATION SYSTEM IN THE
INTERCONNECT REGION OF THE SSC COLLIDER MAGNETS
A. R. Jalloh and D. Baritchi

2550 Becldeymeade Ave.
Dallas, TX 75237
ABSTRACT
The superconducting nature of the Superconducting Super Collider (SSC) collider

magnets requires extremely low operating temperatures. As a consequence, a heat leak:
budget with very stringent tolerances must be maintained during operation. Violating the
heat leak: budget could result in loss of superconductivity and quenching of the magnets. To
ensure that the magnets operate at the appropriate temperature levels, the interconnect
region should be maintained at similar operating temperature levels. To this end, a very
efficient thermal insulation system is required in the interconnect region. This study
presents a finite-element model of the insulation systems in the interconnect region that
serves as a tool to investigate the thermal efficiency of candidate insulation system in the
interconnect region. Temperatures predicted by the model showed good correlation with
experimental results.
INTRODUCTION
The interconnect region (ICR) links the cryostats of two adjacent magnets of the SSC
particle beam accelerator. 1,2 The insulation system in the ICR is similar to that in the
cryostat. It consists of heat shields, multilayer insulation (MU) blankets, and cryogenic
pipes. A cross-sectional view of the dipole magnet that connects to the ICR is shown in
Figure 1. The ICR cross section has to match this exactly, except for the re-entrant support
post. An elevation view of the ICR is shown in Figure 2. This shows the cryogenic pipes
and their connections to two neighboring magnets. The vacuum vessel, heat shields, and
insulation blankets have been removed from the figure for clarity.
Extremely stringent heat leak: budgets must be maintained in the SSC collider ring
magnets to ensure that the operating temperatures of the magnets do not exceed the 4.25 K
No. DE-AC35-89ER40486.
Supercollitkr 4, Edited by J. Nonte 321

threshold. If this threshold is exceeded, superconductivity is lost and quenching of the
magnets occurs. A poorly insulated ICR could become a convenient route for heat to flow
into the magnets from the outside, where the nominal temperature in the tunnel is about
300 K. Therefore, a highly efficient insulation system is required not only in the magnets
but also in the ICRs. Several candidate insulation components and systems may have to be
evaluated before a final design is adopted. This would require much effort and time if it
were done empirically, since each component change would require a complete assembly
and test of the entire system. A computer model of the system will considerably reduce the
effort and time required to evaluate the relative performance of components and systems.
In the present study, a heat transfer, finite-element computer model of the ICR
insulation system was developed and analyzed. The model accounted for conduction and
radiation heat transfer only. Residual gas conduction could not be accounted for because the
software used in the development and analysis of the model does not lend itself to handling
this phenomenon. The model also included the thermal effects of the fluid flow in the
cryogenic pipes. Only one insulation system was analyzed because of the lack of additional
MLI thermal data. Temperature distributions throughout the ICR were obtained. Average
temperatures on the exposed surfaces of the MLI and heat shields are presented as the
evaluation parameter. The temperature distribution for a different insulation system, say
different MLI blankets, would result in different average temperatures on the surfaces. The
more efficient system then would be the one that results in lower average temperatures.
~~~~~:::::-_ _- Vacuum vessel

;::: (68S.Snvn 00)
8DK shield
2DK Shield
MU
Cold mass cradle

Anchor tie bar
(1 012) 15~!;~~f:j~ Reentrant support post
Figure 1. 50-mm Collider Dipole Cryostat Cross Section.
MODEL DEVELOPMENT AND ANALYSIS
The major components of the collider magnets ICR are the heat shields, MLI
insulation blankets, vacuum vessel, the protrusions of the cold masses, and the cryogenic
pipes 1,2 as shown in Figures 1 and 2. Heat flows from the tunnel by conduction through the
vacuum vessel wall and by radiation from the inner surface of the vacuum vessel to the
outer surface of the 80-K MLI. Conduction takes place through the MLI and 80-K heat
shield. Then radiation occurs from the inner surface of the 80-K shield to the outer surface
of the 20-K MLI. Conduction occurs through the MLI and 20-K shield. Finally, radiation
322
Figure 2. Interconnect Region (elevation view).
takes place between the inner surface of the 20-K heat shield and the outer surfaces of the
cold mass protrusions and the cryogenic pipes. It should be noted that liquid and gaseous
helium and liquid nitrogen flow through the cryogenic pipes to cool the magnets. ill addition
to these modes of heat transfer, there is residual gas conduction that results from the fact
that a perfect vacuum in the cryostat cannot be achieved. This mode of heat transfer was not
represented in the model because of software limitations.
The interconnect region was modeled by the finite-element method and analyzed using
ANSYS, a commercial finite-element program) The basic conduction and radiation
equations of heat transfer4,5 are used to develop the finite elements. The flow in each pipe
was investigated to determine whether it was laminar or turbulent before an appropriate
design correlation was applied. Laminar flow was modeled by an empirical correlation, and
turbulent flow was modeled by the Petukhov-Popov design correlation for turbulent flow. 5
A three-dimensional, eight-node heat transfer element, ANSYS element stinO, was
used to model conduction through the vacuum vessel wall, heat shields, and MLI blankets.
The apparent thermal conductivity of the MLI was used in the analysis. StifS7, the ANSYS
thermal shell element with four nodes, was used to model the radiating surfaces. The
procedure for handling radiation in ANSYS consists of developing a radiation super-
element from the geometry and material properties of the elements of the radiating surfaces.
A section of the finite element mesh is shown in Figure 3. All thermal properties were input
as temperature-dependent quantities. The analysis was highly non-linear because of the
quartic nature of the temperature term in the heat transfer equation. A steady state analysis
was carried out since it represents the worst-case scenario. Model development and analysis
was carried out on a SPARCstation 2.
RESUL TS AND DISCUSSION
Temperatures predicted by the model were compared with experimental data obtained
from the Fermi National Accelerator Laboratory ER Dipole String Test. 6 There is very good
correlation between the experimental data and the analytical data obtained by the model.
The differences could be due to the fact that residual gas conduction was not included in the
model and because the thermal conductivity of the MLI used was the apparent value. A
more rigorous analysis that treats the MLI blankets as several layers of insulation instead of
a bulk material could yield a better correlation. Finally, another reason for the slight
discrepancies between the experimental and analytical results could be due to the fact that
the cooling effects of the cryostats and insulation systems of the magnets that attach to the
ICR were not considered.
323
1
ANSYS 4.4A1
FEB 13 1992
8:19:54
PLOT NO. 1
I?REP7 ELEMENTS
TYI?E NUM
xv =1
'tV =1
ZV ~1
DIST=0 .74243 5
XF -- 0.19131
ZE' -0 .6955
PRECISE HIDDEN
cryogenic mode l of intercon ne ct region
Figure 3. Finite Element Mesh of the Interconnect Region.
Table 1. Comparison of experimental and analytical temperatures on exposed surfaces of

MLI and heat shields in the interconnect region insulation system.
Component Experimental Data (K) Analytical Data (K)
80KMLI 275 262

80 K Heat Shield 136 121
20KMLI 32 29
20 K Heat Shield 18 21
This model would be very useful in carrying out comparative analysis using several
different insulation systems and studying the resulting insulation efficiency. This was not
done in the present study because thermal data for MLI, other than that used in this model,
was not available.
324
SUMMARY
A finite-element heat transfer model of the interconnect region of the sse collider ring
magnets was developed and analyzed using ANSYS. The model considered only solid
conduction and radiation between the exposed surfaces of the insulation system. Residual
gas conduction could not be represented in the model because of software limitations. The
MLI blankets were modeled with their apparent thermal conductivity. The cooling effects of
the magnets that attach to the interconnect region were not included in the analysis. A
steady state analysis was carried out since it represents the worst-case scenario. The results
showed good correlation with experimental data. This model could be used to compare the
thermal efficiencies of various insulation systems by varying the appropriate parameters in
the model and carrying out the analysis.
REFERENCES
1. Superconducting super Collider Laboratory, "Site-Specific Conceptual Design," SSCL-SR-I056,
July 1990.
2. D. Baritchi, T. Nicol, and W. Boroski, "Conceptual Design for the Thermal Bridges and Multilayer
Insulation in the Interconnect Region for the SSC," SuperCollider 3, Plenum Press, 1991.
3. F. !{reith, and M. S. Bohn, Principles of Heat Transfer, Harper & Row, 1986.
4. B. V. Karlekar, and R. M. Desmond, Heat Transfer, West Publishing Company, 1982.
5. G. J. Desalvo "ANSYS Engineering Users Manual," SASI, Houston, PA, 1989.
6. FNAL ER Dipole String Test Results, 1991.
325
HIGH-ACCURACY, TWO-DIMENSIONAL MAGNETIC FIELD MULTIPOLE
CALCULATIONS FOR ACCELERATOR MAGNETS
M. P. Krefta, H. L. Chuboy and D. Pavlik

Science and Technology Center
Pittsburgh, P A
ABSTRACT
High precision magnetic field computer programs are required to accurately

calculate the multipole content and magnetic field strength associated with a given
accelerator magnet design. In this work, a specialized version of the finite element
computer program WEMAP (Westinghouse Electric and Magnetic Analysis
Program) is applied to field harmonic calculations. The computer program relies on
a higher-order finite element formulation and high-precision computer
representation to minimize calculation error. Several calculations are made for a
SSC CDM cross section design. The effects of mesh fineness and finite element
order are investigated by examining finite element grids with increasing mesh
density and element order. A key feature of this computer program is the
capability to minimize the truncation error associated with a finite element mesh by
increasing the order of the finite element interpolating polynomial. Finite elements
of order one through six are incorporated. Inaccuracies which may occur using the
lower-order finite elements with a coarse mesh are identified in this paper. Care
must be taken in meshing the yoke iron region to accurately account for field
harmonics due to nonuniform iron permeability. Methods of extracting the spatial
harmonics from the finite element solution are discussed. The field strength and
harmonics are calculated for a full-section model of a CDM magnet cross section.
INTRODUCTION
The requirements on the SSC Collider Dipole Magnet (CDM) specify a high
degree of magnetic field uniformity and very stringent limits on magnet-to-magnet
field strength variability. An example is the requirement for the maximum
allowable systematic value of skew quadrupole (at) given as 0.04 units. This is a
very small value when compared to the dipole field level of approximately 6.7 T.
To fully appreciate the relative difference, the units of the field harmonics must be
considered. The unit of the field harmonic is expressed in parts per 10,000 of the

dipole field. Thus, the skew quadrupole component should be at least 250,000
times smaller than the dipole component. Further, these requirements must be
verified for a given magnet design through detailed magnetic field calculations.
These calculations must be capable of an accuracy of several times greater than the
maximum allowable systematic value. The calculation method used must
accurately predict a component which is about 106 times smaller than the main
field.
This paper is primarily concerned with the accuracy of numerical calculations

for the field strength and harmonic components. An estimate of the uncertainty in
the multipole calculations is necessary when comparing different design features.
In prior works, the accuracy of the calculation was judged by the differences
obtained using several numerical computer programs 1,2. Specifically, the
uncertainty in the sextupole component of field (b2) has been estimated to be about
one-quarter unit.
The approach taken in this work attributes the calculation error to the finite
element model, rather than the computer program. The models are refined so the
calculation error is minimized for a given magnet design. This process results in
uncertainties in the calculated harmonics which are an order of magnitude lower
than those previously published.
Calculations are performed on several CDM cross sections using a specialized

version of the finite element program WEMAP3. This finite element code is well
suited to problems requiring a high degree of accuracy since it utilizes higher-order
finite elements to minimize truncation or discretization error. The program is used
to analyze the CDM design cross section and determine the effects of design
changes on the magnetic field strength and uniformity.
HIGHER-ORDER FINITE ELEMENT APPROXIMATIONS AND MESH

DENSITY STUDIES
One major concern when using numerical methods to calculate magnetic

fields is the accuracy of the resulting solution. Two methods are generally used to
reduce the discretization or truncation error when using finite element techniques.
The first consists of successively refining the finite element mesh until no significant
changes occur in the solution. This approach is commonly referred to as h-type
mesh refinement. A second method is to increase the element order until the
solution converges to a single answer. This is referred to as p-type refinement. The
effects of these methods on the calculated multipoles is discussed in this section.
The accuracy of a finite element solution can be evaluated by comparing finite

element results with the spatial harmonic components obtained from a closed-form
solution. In the case considered here, the geometry consists of several current
elements internal to an infinitely permeable, smooth yoke. The currents are
distributed in rectangular regions representing the superconducting cables of CDM
design DSX-201 version 64. The magnetic field spatial harmonics can be calculated
using closed-form methods 5 . These results may be compared with harmonics
calculated using the finite element method to determine the accuracy of the finite
element model.
328
The finite element order is varied from one to six for a given mesh density.
The primary purpose is to determine if the harmonics computed using the higher-
order elements converge with those calculated using the closed-form solution. A
quarter-section model is sufficient for this study, since the geometry is symmetric in
each of the four quadrants. The finite element mesh formed for this geometry is
shown in Fig. 1. The mesh is meant to be coarse so the effects of increasing the finite
element order can be observed. The coordinates of the corner points defining the
current locations are given in Table 1. The yoke is smooth and infinitely permeable
with an inner radius of 70.0 mm.
The calculated spatial harmonics are given in Table 2. It may be observed that
the data agree to more and more decimal places as the finite element order is
increased. The case of the first-order element results in a poor harmonic calculation.
A significant gain in accuracy is achieved by using second order elements. This is
particularly true of the sextupole harmonic. An accurate calculation of some of the
smaller, higher-order spatial harmonics requires a finite element order of greater
than two for this coarse mesh.
Table 1. Location of first-quadrant conductor corner points for version 6

of CDM design DSX-201 used in finite element model study.
Cable xl(an) yl(an) x2(an) y2(an) x3(an) y3(an) x4(an) y4(an)
1 2.49889 0.04935 2.49810 0.12404 3.73203 0.13714 3.73282 0.06245

2 2.49003 0.19859 2.48765 0.27325 3.72103 0.31254 3.72341 0.23788
3 2.47221 0.34735 2.46825 0.42195 3.70051 0.48742 3.70447 0.41282
4 2.44538 0.49534 2.43984 0.56983 3.67043 0.66145 3.67598 0.58695
5 2.40948 0.64221 2.40235 0.71657 3.63072 0.83429 3.63785 0.75993
6 2.36439 0.78765 2.35569 0.86184 3.58129 1.00561 3.58999 0.93142
7 2.30582 0.97781 2.26796 1.04221 3.33175 1.66761 3.36960 1.60321
8 2.24177 1.11363 2.20256 1.17721 3.25282 1.82505 3.29204 1.76147
9 2.16993 1.24495 2.12938 1.30768 3.16565 1.97768 3.20621 1.91494
10 2.09051 1.37153 2.04863 1.43339 3.07044 2.12523 3.11232 2.06337
11 2.00368 1.49312 1.96049 1.55407 2.96739 2.26745 3.Q1058 2.20650
12 1.90961 1.60944 1.86514 1.66947 2.85667 2.40406 2.90114 2.34404
13 1.80845 1.72023 1.76272 1.77930 2.73843 2.53478 2.78416 2.47571
14 1.55791 1.95743 1.49706 2.00075 2.21284 3.00595 2.27368 2.96262
15 1.43740 2.04579 1.37564 2.08782 2.06992 3.10799 2.13167 3.06596
16 1.31185 2.12679 1.24922 2.16750 1.92168 3.20217 1.98431 3.16147
17 1.06850 2.26206 0.99935 2.29031 1.46604 3.43265 1.53519 3.40440
18 0.93096 2.32050 0.86123 2.34727 1.30356 3.49927 1.37329 3.47249
19 0.79021 2.37060 0.71992 2.39589 1.13770 3.55702 1.20798 3.53173
20 3.79236 0.04255 3.79183 0.10365 4.95978 0.11395 4.96032 0.05285
21 3.78858 0.16470 3.78696 0.22577 4.95455 0.25667 4.95617 0.19559
22 3.78086 0.28670 3.77816 0.34774 4.94503 0.39922 4.94772 0.33818
23 3.76920 0.40844 3.76543 0.46943 4.93121 0.54148 4.93498 0.48050
24 3.75361 0.52984 3.74876 0.59075 4.91309 0.68335 4.91793 0.62245
25 3.73409 0.65079 3.72817 0.71160 4.89068 0.82472 4.89659 0.76391
26 3.71063 0.77118 3.70364 0.83188 4.86398 0.96549 4.87097 0.90479
27 3.68326 0.89091 3.67520 0.95148 4.83300 1.10553 4.84105 1.04496
28 3.65196 1.00989 3.64284 1.07031 4.79774 1.24475 4.80686 1.18434
29 3.61675 1.12801 3.60656 1.18825 4.75821 1.38304 4.76840 1.32279
30 3.57763 1.24515 3.56638 1.30521 4.71441 1.52028 4.72566 1.46022
31 3.51513 1.43308 3.48599 1.48678 4.51257 2.04388 4.54172 1.99017
32 3.46653 1.54585 3.43644 1.59903 4.45304 2.17414 4.48313 2.12096
33 3.41438 1.65684 3.38336 1.70948 4.38966 2.30243 4.42067 2.24979
34 3.35873 1.76596 3.32679 1.81805 4.32247 2.42865 4.35441 2.37657
35 3.29965 1.87314 3.26679 1.92465 4.25155 2.55272 4.28441 2.50121
36 3.23718 1.97828 3.20342 2.02920 4.17695 2.67454 4.21071 2.62362
37 3.17140 2.08130 3.13674 2.13162 4.09874 2.79403 4.13339 2.74371
38 3.10234 2.18212 3.06681 2.23182 4.01697 2.91110 4.05251 2.86139
39 3.03008 2.28066 2.99368 2.32973 3.93171 3.02565 3.96812 2.97658
40 2.95467 2.37683 2.91740 2.42526 3.84302 3.13761 3.88029 3.08919
41 2.87617 2.47056 2.83806 2.51832 3.75097 3.24689 3.78908 3.19914
42 2.79464 2.56176 2.75569 2.60884 3.65561 3.35340 3.69456 3.30632
43 2.71014 2.65036 2.67037 2.69674 3.55702 3.45705 3.59679 3.41067
44 2.62274 2.73626 2.58215 2.78193 3.45525 3.55777 3.49584 3.51209
45 2.53249 2.81939 2.49110 2.86434 3.35038 3.65546 3.39177 3.61051
329
The harmonics calculated in Table 2 using higher-order finite elements do
not appear to converge precisely with those obtained using a dosed-form solution.
The reason for this is the geometry associated with the finite element method differs
slightly from that of the dosed-form solution. Specifically, the inner yoke surface is
a pure arc in the dosed-form solution. However, the finite element mesh represents
this surface as a piece-wise linear approximation to an arc. This approximation for
the yoke surface becomes more accurate as the number of elements in the finite
element model is increased.
This is illustrated by examining the harmonics resulting from the fine and
extra-fine meshes shown in Figs. 2 and 3. Examination of Table 2 shows that the
dipole component of field calculated using the coarse mesh is about 10 Gauss larger
than the dosed-form calculation. The dipole component calculated using the fine
mesh is about 3.5 Gauss too large and that calculated using the extra-fine mesh is
about 2.3 Gauss too large. This indicates that as the finite element mesh is made
finer the results converge to those obtained using the dosed-form solution.
The results for the fine and extra-fine· meshes are only given to order three.
The higher-order elements are not needed for these fine meshes, since the
harmonics have sufficiently converged using the lower-order elements. Moreover,
the number of equations solved for orders greater than three becomes excessive.
The computational costs associated with the finer meshes are high. This is seen by
examining the number of finite element nodes (or equations) required for a given
mesh and finite element order. The coarse mesh gives good harmonics for a third-
order finite element. This requires the simultaneous solution of 10,768 equations.
The extra-fine mesh using second-order elements produces a solution of
comparable accuracy. In this case 30,382 simultaneous equations are solved. In
general, the case of the coarse mesh with a third-order element is preferred due to
the savings in computer memory and solution time.
Table 2. Comparison of calculated multipoles for a quarter-section

model of magnet design DSX-201_6 with different finite
element mesh densities and element orders.
-
COARSE MESH 2367 elementS
Order eo (Tesla) b2 (uniUl) b4(uni1ll) b6 (units) b6(unill) bl0 (unill)

lal 1223 6.714758 ·0.412175 -0.029114 -0.026738 0.000465 0.018828
2nd 4812 6.715154 ·0.183782 0.001593 -0.000818 -0.000234 0.018220
3rd 10768 8.715158 ·0.183957 0.003098 -0.001310 -0.000043 0.018268
4111 19091 6.715157 ·0.183942 0.003048 -0.001293 -0.000045 0.018285
5111 29781 8.715158 ·0.183943 0.003045 ·0.001292 -0.000045 0.018284
8111 42838 6.715158 ·0.1_ 0.003045 ·0.001292 -0.000045 0.018284
Cloaed-Form 8.714194 ·0.172570 0.002587 -0.001218 -0.000022 0.018257
-
Syatematic Spec cO.8 cO.08 cO.013 cO.Ol0 cO.Ol0
FINE MESH 9388 elements
Order eo (Tesla) b2 (uni1ll) b4 (unill) b6(unila) b6(unill) bl0 (unill)

1.1 4771 6.714482 -0.225052 -0.005587 -0.007688 . -0.000154 0.018534
2nd 18927 8.714532 ·0.181339 0.002745 -0.001243 -0.000080 0.018268
3rd 42489 8.714532 ·0.181503 0.002893 -0.001291 -0.000046 0.018268
Closed·FDrm 6.714194 ·0.172570 0.002587 -0.001216 -0.000022 0.018257
-
Syalllf11alicSpec cO.8 cO.08 cO.013 cO.Ol0 c 0.010
EXTRA-FINE MESH 15071 elemenls
Order eo (Tesla) b2 (unill) b4 (uni1ll) b6 (units) b6 (unill) bl0(unill)

lal 7858 6.714243 ·0.185845 -0.004545 -0.002818 -0.000302 0.01_
2nd 30382 6.714423 ·0.181897 0.002972 -0.001311 -0.000038 0.018270
3rd 68179 6.714423 ·0.181577 0.002872 -0.001283 -0.000043 0.018266
Cloaed-Form 6.714194 ·0.172570 0.002587 ·0.001218 -0.000012 0.018257
Systematic Spec cO.8 cO.08 cO.013 cO.Ol0 cO.Ol0
(Note: R _ 2.2 ern; NPTS _ 128; Un... Case)
330
., .!
Figure 1. Coarse finite element mesh used in model accuracy study.
Figure 2. Fine finite element mesh used in model accuracy study.
Figure 3. Extra-fine finite element mesh used in model accuracy study.
331
MODELLING THE EFFECTS OF YOKE PERMEABILITY VARIATIONS
In addition to controlling the finite element size and order throughout the
coil region, the material property variation associated with the iron yoke has an
effect on the accuracy of calculated field harmonics. This effect is particularly
important at high field when magnetic saturation has a substantial influence on the
spatial harmonic content in the magnet interior. Finite element computer codes
generally represent the material variations on an element by element basis. Internal
to any given element, the material property is assumed uniform. This results in a
rather crude, discontinuous approximation for the material variation in space. A
fine mesh is required in regions where material property variations are large.
A linear variation of the material property internal to the finite element is

available in the WEMAP program. This option is used to reduce the requirements
on mesh fineness in the yoke region. However, the resulting mesh needed to
accurately represent the material variation in the yoke iron region is still relatively
fine. The higher-order elements (4,5 or 6) are not valuable in this region. Since the
mesh is sufficiently fine, the lower-order elements contribute only slightly to overall
truncation error. The added computation time and memory requirements
associated with the higher-order elements result in an insignificant gain in accuracy.
Three different quarter-section models are used to determine the required

mesh density in the yoke iron region. A coarse, fine and extra-fine mesh are used to
determine the impact of the yoke mesh density on the calculated harmonics. The
harmonics for each mesh are given in Table 3. The yoke mesh density has the
largest effect on the sextupole component (b2) of field. The higher-order harmonics
are less sensitive to yoke mesh fineness. The differences in the harmonics of the fine
and extra-fine yoke meshes are small. This indicates that a sufficient mesh density
has been attained.
Table 3. Comparison of calculated multi poles at 6500 A for a quarter-

section model of magnet design DSX-201_6 using a coarse, fine
and extra-fine mesh in the yoke iron region.
elements BO(Tesla) b2(units) b4(units) b6(units) b8(units) b10(units)
coarse yoke mesh 3484 6.5928 0.23261 -0.02473 -0.001680 -.000054 0.018605
fine yoke mesh 4545 6.5915 0.24113 -0.02430 -0.001697 -.000055 0.018608
extra-fine yoke mesh 5690 6.5916 0.24217 -0.02419 -0.001699 -.000055 0.018609
EXTRACTING SPATIAL HARMONICS FROM FINITE ELEMENT

SOLUTIONS
The spatial harmonic components can be extracted from the finite element
solutions using a variety of methods. The WEMAP program uses the magnetic
fields or the axial component of the magnetic vector potential to obtain the spatial
harmonics. The variable analyzed for harmonic content can be either the x, y,
radial, or tangential component of f~eld or the axial component of the vector
potential. Anyone of these variables is interpolated to an arc at a radius prescribed
by the user. The set of points is then Fourier analyzed to determine harmonic
content.
Theoretically the harmonic content associated with a source free region is

invariant with radial location except for a factor of rn. However, in numerical
calculations this is not necessarily true. For example, at very small radii the higher-
order harmonics rapidly approach zero. It is difficult to measure the harmonic
332
content since the size of the field harmonics is on the order of the numerical round-
off error. It is often more accurate to use a larger radius to measure the field
harmonics and then scale the results to 1 em. .
The choice of the optimal radius for evaluating the multipole content is
dependent on several factors. First, the optimal radius depends on the local
truncation error associated with the finite element mesh. In general this becomes
locally high as the interpolation radius approaches the current elements. The finite
element order also influences the choice of the optimal interpolation radius. For the
higher-order elements, the truncation error becomes very small throughout the
mesh. Here, the numerical error dominates and a large interpolation radius close to
the conductors will minimize the error.
A final factor influencing the choice of an optimal interpolation radius is the

harmonic number. For the lower-order harmonics the results are generally very
good, even for a small interpolating radius. However, for the higher-order
harmonics, a larger interpolation radius is required. For example, the fine mesh
(Fig. 2) using a third-order element to calculate the fields at an interpolation radius
of 2.2 cm gives the results in Table 2. The value of the calculated bl0 is .018266
units. If an interpolation radius of 0.7 cm is used, the calculated value of bl0 is
.017658 units. The calculation with the smaller interpolating radius is suspected to
be in error, since the closed-form solution gives an answer much closer to that
obtained using the larger interpolating radius.
CALCULATIONS FOR A MAGNET DESIGN CROSS SECTION
A finite element mesh for a full-section representation of the DSX-201_6

design is shown in Fig. 4. This mesh is being used to calculate the field strength and
harmonics in the magnet cross section. A full-section model is required since all
harmonics exist due to the presence of the off-set cryostat and the bus currents.
The skew quadrupole (al) and sextupole (b 2 ) harmonics are plotted

throughout a range of fields in Fig. 5. The cryostat has a substantial effect on the
skew quadrupole term at higher field levels. The level of al is -0.07 units at a field
of 6.7 T. This is significantly higher than the systematic specification « 0.04 units)
on this multi pole. Methods of corr~cting for this have been proposed and analyzed
using a similar finite element mesh .
The sextupole component is sensitive to iron yoke saturation. The change in

this component due to saturation is 0.47 units. This is a substantial fraction of the
systematic multipole specification (0.80 units). It has been shown that the yoke
midplru:e key ca~ b 2used to reduce the change in this multipole component due to
magnetic saturation .
CONCLUSIONS
In this work the magnetic field errors associated with two-dimensional finite
element models have been closely examined. Several models of a COM design have
been constructed and the calculated harmonics were shown to converge as the finite
element order is increased. Mesh density has an effect on the calculated multipoles
particularly in the yoke iron region where the spatial variation of permeability
influences the harmonics. The computational costs are minimized by using third-
order finite elements in combination with a relatively coarse mesh. The
uncertainties in the calculated field harmonics are within 3 % of the systematic
specification for the sextupole harmonic and below 1 % for the higher-order
333
Figure 4. Finite element mesh of a full-section model for magnet design DSX-201_6
with details of bus currents and offset cryostat included.
0.50
..
';;
.c 0.36
I
/ ~
~
Iii b2 "' .
I-
Z / 1\
/ \\
w 0.22
z
0
--
0..
::lE
0
0.06
V
0
z00 I ". .......... \,

::lE -0.06
a: ';-'.,
« a 1 -
I
J:
-0.20
2.00 2.75 3.50 ".25 5.00 5.75 6.50
'"
7.25 8.00
FIELD ST RENGTH (T )
Figure 5. Plot of skew quadrupole (a1) and sextupole (b2 ) harmonics for magnet
design DSX-201_6.
harmonics. The main dipole component of field can be calculated to within a few
Gauss; however, this requires the use of a very fine mesh to precisely represent the
geometric details and permeability variation.
REFERENCES
1. R.c. Gupta, S.A. Kahn, and G.H. Morgan, "SSC 50 MM Dipole Cross Section," Presented at
the 3rd IISse Conference, Mar. 13-15, 1991.
2. R.c. Gupta, S.A. Kahn, and G.H. Morgan, "Coil and Iron Design for SSC 50 MM Magnet,"
Presented at ASME Winter Meeting, Nov. 1990, Dallas, TX.
3. WEMAP Westinghouse Electric and Magnetic Analysis Program, Pittsburgh, PA 15235.
4. sse CDM 2D Cross-Section Mini-Review, San Diego, CA, Oct. 1991.
5. M.P. Krefta and D. Pavlik, "Two-Dimensional Magnetic Field Calculations for the SSC Dipole
Magnets," Presented at the 3rd IISSC Conference, Mar. 13-15, 1991.
6. H. Gurol et aI., "Status of the Magnetic Design of the SSC Dipole Magnets," Presented at the
4th IISse conference, Mar. 4-6, 1992, New Orleans, LA.
334
MAGNETIC OPTIMIZATION OF THE HEB DIPOLE MAGNET END SECTION
M. P. Krefta and J. H. Parker Jr.

Pittsburgh, P A
ABSTRACT
The end region of the HEB Dipole Magnet may be designed to maximize the
effective magnetic length while ensuring acceptable field uniformity and local field
margin against quench. In this work, end-turn configurations are proposed and
analyzed to determine the effective magnetic length, the integrated multipole
content and the maximum field strength in the superconductor. An end-turn
configuration with turns compacted against the ends of the magnet enclosure is
considered first to determine the maximum achievable magnetic length. The
amount of multi pole correction needed to obtain zero net integrated sextupole
component of field is determined from the analysis of this configuration. The
design is then modified to produce zero average sextupole component by adjusting
the relative axial spacing of the blocks of cables. These modifications are made so
there is a minimum decrease of the integrated dipole field. This results in a design
with both high field uniformity and maximum magnetic length. The peak fields in
the superconductor end region are calculated in order to ensure that the local fields
are within margin.
INTRODUCTION
The magnetic field strength and uniformity are central to the Supercollider
dipole magnets. Establishing a magnet design that produces a highly uniform
magnetic field with sufficient strength and maximum quench margin requires a
precise knowledge of the magnetic field distribution in the magnet cross section and
end-turn regions. In order to meet stringent mechanical and magnetic specifications
placed on the design, the end region in addition to the two-dimensional magnet
cross section must be optimized. The optimization of the mechanical end design
has been discussed previouslyl. This paper primarily addresses the optimization of
the magnetic portion of the end design.

A minimum strain energy configuration for individual or grouped blocks of
cables is a prime consideration for the mechanical design of the end1. Some of the
geometric details associated with the mechanical design such as the cable twist
angle or the exact end shape, while crucial to the mechanical design of the end, have
relatively minor effects on the magnetic field harmonics throughout the end region.
In this work, the cable twist and precise curvature of individual cables are
approximated to simplify the magnetic design and analysis of the end region.
The multipole content of an end-turn design is generally controlled by

adjusting the relative axial spacing between the blocks of conductors2. An adverse
result of this is a decrease in the effective magnetic length. In this work, axial
spacers are added to control the integrated harmonic content throughout the
magnet end region. These spacers are chosen so they have a minimal effect on the
dipole component of field. This is accomplished by first evaluating the integrated
multi pole content of a compacted end-turn design. Then, spacers are inserted
between coil blocks at locations where the undesirable harmonics can be strongly
influenced while having a minor impact on the dipole field. This results in a design
with maximum magnetic length and a high-degree of field uniformity.
MAGNETIC FIELDS ASSOCIATED WITH COMPACTED END-TURNS
The spatial harmonic content of a compacted end-turn design is calculated in

this section using a three-dimensional solution for the magnetic field distribution.
The method3 used relies on an analytic solution to Laplace's equation that satisfies
the appropriate boundary conditions. In this case, the windings are assumed to be
contained within a finite length cylindrical enclosure of infinite permeability. Axial
gaps are introduced at each end of the enclosure to represent the axial termination
of the yoke iron. Final axial closure is provided by infinite conducting plates that
reflect the mirror symmetry between successive ring magnets. This axial geometry
is illustrated in Fig. 1.
The detailed geometry to be analyzed is given in Figs. 2 and 3. The cross

section corresponds to that of magnet design DSX-201. The conductors are grouped
in blocks identified as numbers one through six. Block number one contains the
group of cables in the inner-layer winding closest to the pole while block number
six contains the cables in the outer-layer winding nearest to the midplane. The
relative axial locations of the conductors as they cross over the pole are shown in
Fig. 3. The end-turn shape in the developed plane is assumed to be elliptical with a
ratio of major (axial) to minor (azimuthal) axis of 1.5.
A plot of the dipole and sextupole components of the field are given as a
function of axial position in Fig. 4. The straight length of the winding was taken as
1.0 m, which is sufficient to realize a well developed two-dimensional region (- 0.8
m) and still yield an axial resolution of 2 mm. The sextupole component of field
integrated throughout the end region is calculated to be -6.6 unit-meters. This
design contains an excessive amount of sextupole harmonic. It is necessary to
modify this design to decrease the contribution from the end to the total integrated
sextupole harmonic.
336
~
IRON GAP
//////////
SY M ET RY
PLANE
GAP
Figure 1. Axial layout of three-dimensional geometry to be analyzed which includes

nonmagnetic gaps.
'':1 - 0 • 1•
Figure 2. Quarter-section of magnet design DSX-201 used in the end region magnetic
analysis.
ENO-TUR CROSSOVER
Figure 3. Cross section of end conductors in r-z plane for a compacted end.
337
8.00 100
Wesll~e STC
,
STRAG-t:T P,ART
TEl"
...enZ
TOT AI... LENG~
2.3-' TERS
6.00
a;
OJ
4> 0 a
!: c
4.00 ...J
W
C
..J iL:
w
ii: w
...J
W 2.00 0
...J 0..
0
-100 ::I
0.. t-
o X
W
CIl
0.00
OUTER-LAYER
If\I\ER-LAYER
-2.00 -200
OAO 0.46 0.52 0.58 0.64 0.70
AXIAL POSITION (meters)
Figure 4. Plot of the dipole and sextupole components of magnetic field as a function
of axial position for a fully compacted end-tum design.
OPTIMIZATION OF THE END-TURN SPACERS
Previous work has shown that spacers may be added to the end-turns to
reduce the net undesirable harmonic conten~. This approach is used here to reduce
the integrated harmonic content associated with the compacted end design. In
addition to reducing the higher-order multipoles, the spacers also produce a
substantial decrease in the effective magnetic length. Thus, it is desirable to size
and locate the spacers so the decrease in magnetic length is minimal while
correcting for the excessive sextupole component of field.
The addition of spacers is beneficial to the mechanical design of the end

region. This aids in minimizing the strain energy throughout the turns. It is
convenient to choose the locations for the spacers to be the same as those in the
cross section. That is, the spacers are chosen so that the cables are grouped the
same in the ends as in the two-dimensional region. The spacers are identified as 51
through 56 in Fig. 5. The spacers 51, 52, 53 are used to separate the conductor
blocks within the inner-layer winding. Spacer 55 is used to separate the two blocks
of the outer layer winding and spacers 54 and 56 are used to adjust the relative
location of the inner- and outer-layer windings.
The lengthening of a spacer may be thought of as a reduction in the straight

portion of the conductor near the center of the magnet. It is useful to consider the
problem in this way, since the fields in the straight section are well known and the
effects of the spacers on the multipoles can be quantitatively determined. For
example, if the spacer 51 (Fig. 5) is increased, its effect on the change in the
multipoles can be predicted. The change in a multipole is calculated as the product
of the decrease in the straight section length (equal to the axial increase in the spacer
length) times the value of the multipole at the center of the magnet.
From the results for the compacted end design, it has been determined that
the integrated sextupole component of field should be increased by 6.6 unit-meters
in order to reduce the end-turn contribution to zero. One method of obtaining this
338
correction is to determine which block or group of blocks in the cross section has
the largest influence on the sextupole component (b2) and then extend or decrease
the straight length of this block accordingly. Table 1 gives the contribution to the
dipole and sextupole components of field for each block in the magnet cross section.
The individual blocks have a substantial level of sextupole component. However,
the blocks are arranged in the cross section so there is a net cancellation in the
sextupole term. If block 1 is reduced in length by 1 meter, the effect on the
integrated b 2 would be an increase of 69.789 unit-meters.
The individual blocks in the cross section cannot be adjusted independently of

each other by altering the length of a single axial spacer. Rather, several conductor
blocks are affected when the length of an end spacer is altered. For instance, if
spacer S3 is increased, it is equivalent to decreasing the straight lengths of
conductor blocks 1, 2 and 3 by an equal amount. This results in an increase in the
b 2 multipole by 192.22 unit-meter/meter. The values listed in Table 2 give the
changes in the integrated sextupole and dipole field strength for a one meter
increase in any of the end-turn spacers.
The values given in the second and third column of Table 2 have been
calculated accounting for only the change in the straight length of the conductor
blocks. In an actual magnet design, the axial location of the iron yoke is altered in
addition to the conductor straight lengths. This is necessary in designing the end
region since the field level throughout the conductor end should be reduced by the
amount of the iron contribution (22 %) to minimize the likelihood of quench.
However, the change in the absolute location of the end of the iron produces an
additional change in field not accounted for in column 2 or 3 of Table 2.
Conductors whose locations are not affected by a change in a spacer length, may
still produce a significant change in the dipole field as a result of altering the yoke
iron end location. This additional component has a substantial influence on the
dipole field, but little influence on the higher-order multipoles. This occurs since
the iron contribution to the higher-order multipoles is much smaller than to the
dipole term.
Table 1. Contribution to the sextupole and dipole components of

magnetic field from six current blocks in DSX-201 design cross
section.
Block Dipole Sextupole

Number Component (T) Component (units)
1 0.206 -69.789
2 0.345 -79.051
3 1.165 -43.405
4 1.201 138.348
5 2.055 -36.816
6 1.821 92.392
Total 6.793 1.679
339
Table 2. Contribution to the sextupole and dipole components of
magnetic field due to increasing each of six axial spacers by 1
meter.
Spacer Sextupole Dipole Component Total Dipole

Number Component (unit-m) Neglecting yoke (T-m) Component (T-m)
S1 69.789 -0.206 -1.524
S2 148.839 -0.552 -1.800
S3 192.224 -1.717 -2.832
54 53.896 -2.917 -3.692
S5 36.816 -2.055 -1.644
56 -55.576 -3.877 -3.102
note: This data applies to a magnet half length. These results should be doubled to apply the
results to both ends of a magnet.
ED-TURN CROSSOVER
~
I Z
I
I
I
I,
Figure 5. Cross section of end conductors in r-z plane for an optimized end design.
The effect of changing the end region design can be mathematically expressed
as the sum of two components
6B = B2d 6L, + B (6L - 6L.) (1)
~ g w ~
where 6L i and 6Lw are the changes in length of the iron yoke and windings,
respectively, and B2 d and Bg are multipole components of the field in the two-
dimensional and axial gap regIOns. One component is associated with the change in
straight length of a conductor block due to inward or outward movement of the
end turns of that block. The change in the integrated dipole contribution is then the
amount of the end turn movement times the appropriate quantity in column 2 of
Table 1. The second component is due to the fact that the yoke end may, in general,
move relative to the end turns of any given block. If the yoke end is moved away
from the end turns, this leads to a reduction of the integrated dipole contribution
because the dipole field is reduced by .22B2d in the region that is uncovered by the
yoke moving away.
It is not difficult to obtain a good estimate of this additional change in the

integrated dipole field due to moving the yoke. An example is used to illustrate the
340
method. If spacer 53 is increased by 1 meter, the calculations for the integrated
dipole field must be further reduced by the iron contributions for coils 4, 5 and 6.
This will result in an additional reduction in the dipole field by 22 % of
(1.20+2.05+1.82) or 1.115 T-meter. A similar calculation is done for each end spacer
contained in the inner-layer turn. The spacers in the outer-layer are treated
differently. In this case, the absolute location of the iron end is considered to be
fixed. This is done because the inner-layer cable nearest to the pole generally fixes
the iron yoke end location. 5ince only the change in the winding length (6Lw) is
being altered, the component due to the change in the iron location (6Li ) should be
not be included. The net effect on the dipole field is recomputed for both inner- and
outer-layer spacers to account for yoke motion and given in column 4 of Table 2.
All of the dipole components in Table 2 possess negative signs. This indicates
the dipole field can only be decreased by increasing the size of the axial spacers.
5pacer 53 is most sensitive to producing positive b 2 to offset that in the compacted
design. This spacer could be increased by the smallest amount to generate
additional b 2 . However, there is a relatively large cost in the dipole component of
field incurred by increasing spacer 53. It is more important to increase the spacer
which will give the greatest 6b 2 per unit 6B O' This is not necessarily the spacer
which produces the largest 6b2 per meter of increased spacer.
The data in Table 2 relates the change in the spacer axial lengths to the
integrated dipole and sextupole fields. These relationships are expressed
mathematically in the following form.
6L eff = -.4486S 1 - .5306S 2 - .8346S 3 - 1.0876S 4 - .4846S 5 - .9136S 6 (2)
6b 2 = 69.86S 1 + 148.86S 2 + 192.26S 3 + 53.96S 4 + 36.86S 5 - 55.66S 6 (3)
The expression for effective length is obtained by dividing column 4 of Table 3 by

the two-dimensional dipole field (6.793 T). The coefficients are then doubled to
account for the fact that the data in Table 2 are given for half the length of the
magnet.
In the case of the compacted end design, it is desired to increase b 2 by 6.6

units while minimizing the decrease in effective length. The best way to do this is by
increasing spacer 52 by 38 mm. This assumes that the length of spacers 51' 53 and
55 is 4 mm and both inner- and outer-layers of the coil are fixed against the end of
the magnet enclosure (i.e. 54 =56 =0). The 4 mm is assumed to be the minimum size
that can be useful for the mechanical design of the end region. This will produce a
change in the magnetic length of 27.2 mm. The optimal geometry is shown in Fig. 5.
A detailed three-dimensional analysis of the optimized end-turn design given

in Fig. 5 was performed. The sextupole and dipole components of field are given in
Fig. 6. The net integral of the sextupole field has been reduced to 0.19 unit-meters
in the end region. Clearly, the cancellation of the -6.6 unit-meters for the compacted
design has been obtained. Further, the change in the effective magnetic length has
also been reasonably predicted. The three-dimensional calculation resulted in an
effective magnetic length of 522.4 mm for the compacted design and 507.3 mm for
the optimized design. The decrease in effective magnetic length for the optimized
magnet is 30.2 mm for the entire magnet. This compares with the predicted value
of27.2mm.
341
8.00 200
wesr,nfiihOuSe STC
6.00 (jj
I-
"
-;;
~
100 Z
2
!::
I
4.00 0
..J
0 W
..J
w 0 iL
LL w
..J
w 2.00 0
..J
0 n.
n. =>
e -100
l-
x
w
0.00 U)
VCt<E IRON
I OUTER-LAVER
I IN'ER-LA VER
-2.00 -200
0.40 0.46 0.52 0.58 0.64 0.70
AXIAL POSITION (metersl
Figure 6; Plot of the dipole and sextupole components of magnetic field as a function
of axial position for an optimized end-turn design.
MAGNETIC FIELDS IN THE END CONDUCTORS
The magnetic field strength throughout the superconductor in the end region
must be checked in order to verify that the local fields are within safe bounds. The
yoke iron contribution to the field is reduced in this region by terminating the yoke
prior to the turning of the ends. This is necessary to offset the local field
enhancement due to the small radius of curvature of the end conductors. The
magnitude of the magnetic field is calculated on the inner surface of the inner-layer
coil closest to the pole. This conductor generally experiences the highest fields since
it has the smallest radius of curvature and is the closest to the iron yoke. The
highest field calculated in the inside conductor surface is 5.7 T for the optimized
design.
CONCLUSIONS
A methodology has been set forth for optimizing the magnetic design of the
end region. The approach taken here is to adjust the end spacers, not only for
cancellation in the undesirable harmonics, but also to maximize the effective
magnetic length. Linear relationships are established between the end spacer
lengths and the end multipoles and effective magnetic length. An end design is
proposed with maximum magnetic length and low sextupole component of field. A
full, three-dimensional analysis of the configuration verified the predictions for
magnetic length and integrated sextupole component of field. This method can be
extended to minimize higher-order multi poles at the expense of the effective
magnetic length.
REFERENCES
1. Bossert, R.c., Brandt, J.S., Carson, J.A., Fulton, H.J., Lee, G.c., and Cook, J.M. "Analytical
Solutions to sse Coil End Design," Presented at the 1st IISse Conference, Feb. 8-10, 1989,
New Orleans, LA.
2. Hassenzahl, W.V., Caspi, S., Gilbert, W., Helm, M., Morgan, G.A. and Laslett, L.J., "Field
Quality of the End Sections of SSC Dipoles," SSCL document SSC-N-250, Sept. 1986.
3. Krefta, M.P. and Parker, J.H., "Magnetic Field Calculations in the End Sections of sse Dipole
Magnets," Presented at the 3rd IISse Conference, Mar. 13-15,1991.
342
TRANSIENT COOLDOWN STRESSES IN SUPERCONDUCTING
COLLIDER DIPOLE MAGNET
K. K. Leung and G. Snitchler
Magnet System Division

Dallas, TX 75237
ABSTRACT
The three-dimensional thermal transient stress in a collider dipole magnet (COM) is

evaluated and examined. The cool down rate is one of the factors that determines the
availability of the Superconducting Super Collider ring, but it is limited by stress in the
magnet during cooldown. The main concerns are the effects of coil stress on stable
operation and of shell stress on survival in long-term operation. Cost-effective operation can
be attained if fastest cooldown-induced magnet stress would not exceed the safe operating
stress of the magnet.
A 50-mm dipole magnet is modeled in the vertical split yoke configuration. Magnet
components include shell, yoke, end yoke, collar, coil, and kapton insulation. These
components are integrated by 3-D interface elements with preload but without friction. The
cool down condition is established by 4 K helium at 96 grams per second (gps) flowing in
four bypass tubes and 4 gps to the beam tube annular area. The cooldown time is about
2 hours in most of the center section of the magnet and about 4 hours for the end sections,
assuming that the end plate and the end yoke are also cooled down by the bypass flow. **
Purposes of this study are to understand the stress-time history and maximum stress
generated by the possible fastest cooldown condition and to demonstrate that non-linear
thermal transient stress analysis can be performed in a workstation with limited computing
resources. The results can be utilized to evaluate (1) the stress of the shell in long-term
operation, (2) the fastest safe cooldown rate in the COM, and (3) the cooldown time for the
specified cooldown rate.
No. DE-AC35-89ER40486.
**Cooldown at the end is by mix-flow, and the G 10 end assembly is not included in this simplified model.

INTRODUCTION
Three-dimension CDM thermal steady-state stress has been studied, and high stress is
found in the shell. I, 3 Two-dimension CDM thermal stress has also been studied, and the
results are reported in Reference 2. ANSYS* is used to study the 3-D thermal transient
effect on the CDM in the present analysis. The difference between a 2-D model and a 3-D
model in thermal transient stress calculation is evaluated. Five models are used for this
study. The models (A) and (B) are short models (5 cm), assembled with coil, collar, yoke,
kapton, and shell. Each component is integrated with interface elements, which in thermal
analysis represent the thermal contact resistance that can be experimentally confirmed. A
100% thermal contact area is assumed for all interfaces. This assumption is based upon the
SSC magnet design that uses a line-to-line fit for all interfaces. The helium flow paths are
represented by mass transport elements with specified flow velocities and heat transfer
coefficient. Stress analysis is performed by converting the thermal elements into structural
gap elements. The difference between models (A) and (B) is that the shell material property
is linear elastic for model (A) and bilinear kinematic hardening for model (B). Models (C)
and (D) are 30 cm long with components identical to models (A) and (B), with an end yoke
and end plate added to the shell. Models (C) and (D) are used to study the cooldown stress
on the shell in the area with differential rigidity of the end yoke to the body yoke and to
study the joint stress at the end plate and shell. Model (E) is 15 m long and is used to study
the temperature-time history of the CDM in the axial direction. This model is used for
approximating the thermal boundary condition at the beam tube area because of the short
model assumption for the present analysis.
CDM MODEL AND BOUNDARY CONDITIONS
The detail cold mass model is composed of several units. The units for the short model
are:
(1) Steel Shell. A cylindrical shell made of 304 LN stainless steel with wall thickness
of 0.5 cm. The shell is used as (a) a pressure boundary, (b) a thermal boundary, (c) a
preloading device for the CDM components in limiting micro-movement under Lorentz
force, and (d) an integrated element for CDM components in resisting extemalloads. The
functions of (a) and (b) are employed in the present model. The functions of (c) and (d) are
not used for this analysis and are expected to add stress to the CDM.
(2) Iron Yoke. Steel plates are laminated to form the yoke. The axial elastic modulus
of the yoke is small in comparison to the solid steel and was assumed to be 5% of the
modulus of the solid steel. Models (C) and (D) are used for modeling the end portions of the
CDM in which the end yoke is added to support the end shell to reduce shell-bending stress,
and the end plate is used to apply axial preload to the coil assembly.
(3) Collar. The collar is used for preloading the coils. The collared coils are the basic
assembly of the cold mass in controlling the mechanical quench performance. Ovalization
of the collared coil assembly is induced as the coils' internal stress acts on the collar. The
position of coil assembly ovalization is approximated by the following technique: The
horizontal boundary of the collared coil assembly is applied with vertical pressure
determined by test. The deformation of the coils under such a loading condition is applied to
the CDM model. The yoke, collar, and shell at the horizontal boundary have zero vertical
deflection. The coils, collar, shell, and gap elements that connect to the vertical split yoke
have zero deflection.
*ANSYS is a general-purpose, finite-element code developed by Swanson Analysis Systems, Inc.
344
POSTI -IIiP' AIi sn " .1 R I
,E6261992
17 : 26 : 16
POST I ELEMENTS
TVPE Nun
XU - -I
YU -I
ZU --I
OI ST-22 . 276
XF -6 . '196
H -6 . 5 66
2, • 20
PREC I SE H I DDEN
0; I Pr'eslr'Bssed con Coo l dn , Oullel S. 12000 'Be . T'(, Slr'es,-npa
Figure 1. Detail Cold Mass Model at End Section.
(4) Coil Assembly . Both the inner coil and outer coil are modeled with wedges
considered as part of the coils.
(5) Kapton Insulation. The Kapton is modeled in two different kinds of elements. In
thermal analysis, radial-direction, solid-conduction paths are significant, and conduction bar
elements are used for Kapton. The conduction bar elements are replaced by gap elements in
order to evaluate the thermal stress on the magnet. The application of conduction bar
elements is approximate because lateral heat transfer in the interface is not considered. But
thermal contact resistance in tests indicates insignificant lateral heat flow because the peak
and valley in the contact area break the lateral conduction paths.
(6) End Model. Models (C) and (0) are those with end yoke and end plate added to
models (A) and (B). The cold mass shell is extended and overlapped to the end yoke
without friction. The shell is modeled as an integrated part of the 3.8-cm-thick end plate.
Conduction bars are used for heat flow to sink heat from the end yoke to the shell and to the
bypass coolant. The end shell and the end plate have differential thermal capacitance, which
would generate large thermal stress at the interface. The last model, model (E), is a 15-m-
long CDM model. This model is a simplified 3-D model used to evaluate the thermal
boundary condition of the COM
The thermal boundary conditions are adiabatic at the x = 0 and y = 0 planes. These
boundary conditions are transformed into more complex structural boundary conditions. In
the thermal stress analysis, the boundaries are divided into many groups in order to match
the vertical split yoke condition and the coils pre-deflected at the horizontal boundary. At
z = 0, the axial boundary condition of the magnet is considered as follows: At the center of
the 15-m COM, a fixed translation boundary is observed because there is no thermal
345
movement in this center plane; the other end is subjected to axial compression load but the
relaxation of the coils and other components reduces the axial compression load to an
insignificant value at operation condition. The axial structural boundary condition of the
model for the COM is a fixed-free boundary condition.
LOADING CONDITIONS
Both thermal and mechanical loads are used in the dipole magnet model. The Lorentz
force generated by the coil is not included in this model because the magnets are preloading
to the point that the Lorentz force would not exceed the preload on the magnet components,
as indicated in Reference 2. And there is no Lorentz force developed in the cooldown
condition. The cooling condition at the magnet end assembly-including complex coolant
flow to the GlO end coil assembly, the end plate, vessel head, and end yoke-are to be
included in future study.
MODELING OBJECTIVES AND MODELING TECHNIQUE
Five different models are developed for the COM with different objectives. Ideally
these models can be combined into one with substantially greater computing resources. A
workstation (Sparcstation 2) can perform non-linear, finite-element analysis by dividing the
dipole model into many portions, with each portion as an independent model constructed for
calculating a particular objective of the thermal transient stress problem. This method of
modeling requires a trial procedure in order to maintain the model wave front that can be
handled by the present workstation as well as by the network system. The Sun workstation
is best for handling medium-sized thermal/structural analysis and is good for front-end
applications such as mesh generation and geometry verification for more complex problems.
The COM thermal transient stress problem needs many load steps to evaluate the thermal
stresses. The non-linear element and non-linear materials are first considered in the present
analysis. The non-linear gap element and temperature-dependent material property are the
basic requirements for the COM thermal transient analysis. Elastro-plastic material
properties in multi-linear form are needed to adequately address the relaxation of stress in
components. The Sparcstation 2 can be used to select the most critical non-linear material
properties for the particular component for large deformation effect at a particular time of
the model. Both models (B) and (0) are employed to evaluate stress at the radial and
azimuthal coil stresses and at the interfaces of the shelUend plate and shelUend yoke to
determine the long-term operation cycles of the COM. Using multi-linear kinematic or
isotropic hardening behaviors for material properties can produce the best result in stress
analysis. Limited by wave front of the Sun workstation, the classical bilinear kinematic
hardening is applied in the present analysis. This simple material behavior does account for
the Bauchinger effect by assuming that the yield surface translates only in the direction of
yielding and does not grow in size. Model (E) is a simplified version of model (A), with
axial length expanding to 15 m, but the wave front is minimized by using only significant
elements in this particular analysis.
Transient temperature gradients are directly related to the COM thermal stresses. Fast
cooldown with liquid helium may create local thermal stress in the COM because of
different thermal capacitance of the yoke, end plate, end yoke, collar, coils, and shell. Under
normal circumstances the thermal stress is greater than that obtained under steady-state
condition because the transient gradients are always steeper. The surface thermal stresses in
the shell can be expressed as 1.98 (Ll T*Lla) E.6 The stress depends on the differential
temperature and the coefficient of expansion between the yoke and shell wall and the elastic
modulus E only.
346
IP a S T I - I Ii p. RtiSVS 1 . 1AI
FEB 26 1 992
17 : 25 : 31
POSTI STRESS
STEP-I
I TER-IOOO
T l nE-12000
S I (RU G)
onx -0.053230
snN • 1 . 96 1
snx '1153
snxO-1956
xu- - I
YU -I
2U - -I
0IST-22.276
XF -0.'196
VF -6.566
2F -20
PRECISE HIOOEN
R · I B3 . 396
B - 5'16 . 266
C • 9 09 . I H
o -1272
Coil Pre~tres"ed con Cooldn,Outlel S. 12000 ,ec . ToK, Stre9s-nPa

Figure 2. Thermal Transient Stress at ShelllEnd Plate Interfaces at 12,000 Seconds.
DISCUSSIONS
The calculated temperature and stress of model A are compared to the 2-D model
results from Reference 2. The temperature-time history in the 2-D model is considered as a
good approximation of the 3-D model. However, heat flow paths for the 3-D model are
different from the 2-D model because of cavity and bypass tubes in the yoke. Particular heat
paths in the 3-D model can be fitted to the 2-D model, and the COM temperature-time
history can be studied by using a 2-D model. The comparison of thermal transient stresses
between the 2-D and 3-D models are difficult because local thermal transient gradients are
not available in the 2-D model. Stresses are also sensitive to the three-dimensional geometry
discontinuity and the boundary, which the 3-D model can describe in detail for
investigation. Stress in the third dimension produces significant effects on the combined
stress in components 5 , 6 The experimental azimuthal coil stress of the COM is based upon a
special strain gauge assembly design that uses a steel block attached to a strain gauge to
register the coil/collar interface stress. The recorded strains on this block represent the
average stress sensed by the block and provide no information on the peak stress in the coil.
The fact that the finite-element result indicates higher stress than the test may be attributed
to the averaging of stress by the strain gauge block in test.
The following observation has been made for the CDM model with 100 gps helium at
4 K as coolant. The time to steady-state condition at 4 K from room temperature is about
6000 seconds for the center section. For the end section in the present analysis, by assuming
a cooling condition for simplicity and conservatism, the time to steady state is 14,000
seconds. The cooling time would be less in mix-flow cooling condition, and different
thermal stress would be induced in the end plate, end shell, and vessel head area. This
347
cooling condition would be better studied at a workstation with greater computing
resources. The collared coils' maximum azimuthal stress is 155 MPa (23 ksi) at
12,000 seconds. The coil stress would be less if hyperelastic property of Kapton insulation
is included in the present analysis. The shell/end plate interface combined stress is
1272 MPa (184 ksi), which is less than the tensile strength of 316 LN at 77 K (1517 Mpa)
and 1697 Mpa at 4 K.7 The present model does not associate the gap to the thermal contact
resistance. The radiation between closely-spaced parallel surfaces with a diffuse reflectivity
problem is not addressed in the present analysis. The difference in 2-D and 3-D thermal
stress models is that the 3-D model produces critical information for the CDM design
specifically in the area of high stress. The 2-D model is useful in providing information for
the temperature-time history of the magnet in general. The coil stress is related to the shell
stress and material non-linearities of the Kapton. A complete stress-time history of the coil
and shell with consideration of multi-linear properties for all components under transient
cool down condition should be performed in the future, perhaps with a more powerful
workstation.
IP a 5 T I - I HP - AH5VS 1 . 1 AI
FEB 261992
I 7 :35 : 10
POSTI STRESS
STEP-I
ITER-IOOO
T lnE-12000
SV (AUG)
CSVS-I
on~ -0 . 03'1765
SnN --166,366
SnH6--196,122
sn~ - '15 , 191
Sn~6 - 61 , 259
~ U -- I
YU -0 . 2
2 U • - I
015T - 3 . 666
H =2 . 6 1 6
YF -1 . 967
2F -2 . 5
PRECISE H I DDEH
R =-155 . 001
B --1 2B . 269
C =-10 1 . 536
0 - - H , B03
- - 1 6 , 07
F --21 , 338
G -5 , )95
H -32 , 126
Co i l Pre.lre •• ed con Cooldn,O ul l et S. 12000 .ec . T-K, Stre,,·npo

Figure 3. Figure 2 Thermal Transient Stress in Coils at 12,000 Seconds.
CONCLUSION
We have succeeded in understanding the major influence of the thermal transient

effects on the CDM that are associated with differential thermal time constants of the yoke,
shell, and end plate. Based on the assumption of using 4 K helium as the coolant to obtain
348
the fastest cooldown time at about 4 hours, we found that the cooldown local stress in the
coil is about 23 ksi, and the maximum shell stress is about 184 ksi at the end shell where it
joins the end plate. These stresses will certainly reduce by a factor of 2 or more, depending
on the cooling process of the COM. We also discovered that the end yoke is working very
well to reduce the shell stress under thermal transient conditions. The shell is preloaded at
room temperature, and the longitudinal welded joints of the shell are observed to reach
yield-strength range. With additional cooldown thermal stress developed in the shell, the
horizontal seam welds on the shell will reach beyond the allowable yield strength,
producing the benefits of relaxing the residual stress in the shell and providing mechanical
training of the magnet. The shell stress as calculated has not included the residual stress,
preload, welded thermal, or external loadings in the shell.
Accelerated life tests need to be performed to identify potential fatigue-related failure
modes at the skin welds. Also, a strain gauge assembly could be used to investigate the
transient thermal stress build-up in the coil volume during cooldown.
ANSYS '1,'1A1
FEB 26 1992
16:33:08
K POST26
'100
ZU =1
360 0IST=0,6666
,
XF =0,5
320 YF =0,5
ZF =0,5
280 ~
2'10 l\.
200 ~
160
'r\\,
120
'\ \.
80 ~
'10 ~
o
'\ r-
~ 1<3 8E lfL SEC
0 1600 2'1003200 1000'1800 6006 '100 000
BOO 72008
OM coi I prestress. cooldn, Outlet Section, T=K, time @ 6000 sec
Note: Coil Temp.-Upper Curve. Shell Temp.-Lower Curve

Figure 4. Temperature-Time History of Coils at Center Section.
ACKNOWLEDGEMENT
The authors would like to acknowledge the contribution of B. Aksel of the
sse Laboratory.
349
REFERENCES
1. K. K. Leung, "An Engineering Design Study of the Dipole Magnet Cold Mass End Shell for the
Superconducting Super Collider," page 287, Supercollider 3, edited by John Nonte, Plenum Press,
New York, 1991.
2. B. Aksel and K. K. Leung, "Cooldown Stresses on the Coldmass of SSC Dipole Magnets," Conference
Record 91CH3038-7, Vol. 4, pp. 2200-2202, IEEE Particle Accelerator Conference, 1991,
San Francisco.
3. S. Delchamps, K.K. Leung, et al., "ssc Collider Dipole Magnet End Mechanical Design," IEEE Particle
Accelerator Conference Record 91 CH3038-7, Accelerator Science and Technology, Volume 4,
San Francisco, May 1991.
4. K. K. Leung, "Thermal Stress on Dipole Magnet Vessel End Shell," SSC Magnet System Division Report
MD-TA-148, SSCL-N-702, April 1990.
5. J. Jayakumar, K. K. Leung, et al., "Mechanical and Electromagnetical Analysis of 50 mm Designs for the
SSC Dipole," IEEE Transactions on Magnetics, Vol. 27, No.2, March 1991.
6. K. K. Leung, "An Engineering Design Study of Detector Deformation Limits in the SSC SOC-Detector,"
Fourth Annual Iriternational Symposium on the Super Collider, March 1992.
7. T. Ogata et aI., "Low Cycle Fatigue and Other Mechanical Properties on Aged 316 Ln Stainless Steel at
Liquid Helium Temperature," HY -05, ICMC 1989.
350
STRESS RELAXATION IN sse 50MM DIPOLE cons
Daniel Rogers and Finley Markley
Fermi National Accelerator Laboratory*

Box 500
Batavia, IL 60510
ABSTRACT
We are measuring the stress relaxation of sse 50mm outer coils with the
goal of predicting how much of the coil prestress will be lost while the coils
are warehoused between manufacture and cooldown. We manufacture 3
inch (76.2mm) long segments of coil with the same materials and techniques
that have been used for prototype coils. We are running four simultaneous
tests in an attempt to separate the contributions of the different coil
materials. Test one is a completely insulated coil section where the
insulation is the all polyimide system being tested at Brookhaven; test two is a
wire stack insulated only with the normal Kapton overwrap; test three is a
stack of bare cable; and test four is a completely insulated normal coil
section. All, except for the bare cable, include the ground insulation. The
insulated coil sections are carefully dried before loading and testing in order
to eliminate stress changes due to varying moisture content. The
temperature dependence of the stress relaxation is being studied separately.
Three companion papers presented at this conference will be:
1) "Temperature dependence of the viscoelastic properties of sse coil
insulation" 2) "Measurement of the elastic modulus of Kapton perpendicular
to the plane of the film at room and cryogenic temperatures" 3) "Theoretical
methods for creep and stress relaxation studies of sse coil."
INTRODUCTION
The purpose of these coil studies is to analyze the mechanical property of

stress relaxation with respect to time and temperature. This is an extremely
important property of superconducting magnet coils. 1 Such studies are
necessary because the magnets may fail at great cost if this critical physical
property is not fully understood, accounted for, and utilized. Magnet designs
have been improved immensely in recent years, resulting in much more
* Operated by the Universities Research Association. Inc .• for the U.S. Department of Energy.

reliable magnets. It is hoped that through additional design modifications,
the rate of magnet failure will be reduced even further. Although this is the
first stress relaxation testing on 50mm coils, this is a continuation of
previous studies conducted at this facility, with additional improvements in
testing methods and fixtures.
Stress relaxation studies involve the compressing of a magnet coil to a
predetermined size and monitoring the changing amount of force (with
respect to time) required to maintain this coil size. In other words, the strain
is held constant while the amount of stress varies. This is not the same
phenomena as creep, where the stress is constant and the strain varies.
This particular method of stress relaxation testing was developed to simulate
the conditions of an actual production magnet coil which has been
compressed and collared through the normal manufacturing process. It is
important for such coils to remain in their highly compressed state during
long periods of processing, shipping, and storage, so that on final cool down
the coil will not move causing a quench.
PROBLEMS WITH COn.. TESTING

Coil Geometry
Due to variation in coil strand size and insulation thickness, it is not

possible to accurately predict either coil size of stiffness. As a result, we
cannot very well predict coil stress when a particular coil segment is
compressed in a fixture made to represent an actual collar. These problems
cannot be eliminated, since such variations will always exist in these coils.
Environment
Other possible sources of inconsistency are temperature and humidity.

Attempts are made to control humidity by placing desiccant inside of the
acrylic boxes which contain the test fixtures. In addition, many efforts are
made to maintain a constant temperature level by thermally isolating the test
fixtures and incorporating electric heaters with PID temperature
controllers. Unfortunately, temperature controllers are not 100% accurate
and do allow a slight amount of temperature deviation. The
time/temperature effects on coil stress are not predictable, as the theoretical
time/temperature shift factor (k) for these coils has not yet been determined.
Consequently, To avoid further complications, it was decided not to introduce
any intentional temperature changes. The set point used for all of this
testing is lOO°F, (311 K).
Fabrication of Coil Samples
In order to study magnet coils, a source of these coils is necessary. It

would be extremely costly to do destructive testing of complete full size coils.
Therefore, sample coil segments were manufactured specifically for this
series of tests. This also allows certain variations in the coil samples which
would not otherwise be available. We believe that straight stacks, (ten stacks)
manufactured by flipping alternate cables, do not adequately model real coils.
It cannot be said that these samples are manufactured in every detail
absolutely identical to the real coils. They are, however, manufactured in
such a way as to simulate actual production techniques, and are believed to
be (for the purposes of these tests) reasonably representative of actual
production coils.
352
In addition to studying "regular" coils, it is also of interest to try some
coil variations for comparison. Consequently, four different types of coil
samples are fabricated. First there is the "regular" Kapton/epoxy/ fiberglass
coil. (Kapton is a "trade name" for a polyimide, which is a thermoplastic
used for wire insulation). Next, a second coil ("no fiberglass") is made,
which is identical to the first, except without the epoxy/fiberglass. Another
coil ("bare wire") is made from identical materials, except without either the
epoxy/fiberglass or the Kapton. The fourth variation is to incorporate the
experimental DuPont Kapton CI system. This fourth type of coil ("Special
Kapton" - also referred to as "Kapton only", "all Kapton", or high
temperature Kapton") requires special fabrication techniques. This "special
Kapton" coil was made as per procedures obtained from Brookhaven
National Laboratory, where tests are also being conducted on this
experimental polyimide insulation system.
TOP BLOCK
COIL LOAD
SAMPLE CELL
SPACER BLOCK
FIXTURE BASE
Figure 1. SSC 50mm Dipole Coil Stress Relaxation Test Fixture
Experimental Setup
A drawing of a coil stress relaxation test fixture is shown in figure 1.

Four of these fixtures were made for these studies. Each was designed for
one of the four coil sample types. Each fixture was intended to apply force to
the sample in order to achieve coil stress levels which fall within the range of
those of actual production coils. The fixtures for "bare wire" and "no
fiberglass" coil samples obviously need smaller cavity spaces than the
"regular" coil fixture in order to achieve adequate coil compression.
Theoretically, "regular" and "special Kapton" coils should be dimensionally
identical. However, the actual "special Kapton" coil sample is smaller than
"regular" coil samples. Therefore, shims are inserted in the "special
Kapton" fixture. In spite of designing fixtures for specific coil sizes, coil sizes
may vary and it is sometimes necessary to insert shim stock in with the coil
samples to achieve adequate loading.
Each test is conducted with the fixture contained in an acrylic box
intended for environmental control. Silica gel desiccant is placed inside the
box and the box is sealed with a gasket in an attempt to eliminate humidity.
Fixture temperature is regulated directly by a temperature controller
connected to three cartridge heaters installed in the fixture. A load cell
installed in the fixture converts force to voltage, which is monitored
353
with a voltmeter. The load cell input voltage is supplied by an HP6186c
constant current power source.
DATA ACQUISITION AND CONVERSION

Data is acquired using an IBM PC (with Labtech Notebook) to control an
HP75000 card cage/multimeter. The multimeter monitors load cell inputs
and outputs and relays data to be stored in data files in the computers hard
drive. Because mechanical properties are influenced by temperature, fixture
temperatures are also monitored. The data files generated are converted to
load and temperature data. The load cell calibration number (C) and output
voltage at zero load (Vzl) are used to convert load cell voltages to force in lbs.
Effects of possible power supply voltage fluctuations are compensated for by
incorporating the load cell input voltage (Vin), as well as the output (Vout), in
this data conversion.
Load (lbs) = VO\i.- VzI *C (1)

In
T:FSI'PROCEDURE
First, the proper positioning of the coil sample, insulating ground wrap,
and shim is verified. Then the fixture top is placed in position and bolts are
threaded into the fixture. Alignment of all fixture parts, the load cell, coil,
and shim is then verified and the entire setup is visually inspected to insure
that everything is intact. The next step is to start the data acquisition
program, and note the time, for future reference. Fixture bolts are torqued
while monitoring the load cell output to avoid overloading the load cell.
When the fixture is completely closed the coil is checked to insure that it is
properly seated. The fixture applies a force from 8000 to 16000 lbs. (35 - 71 kN)
to the coil section. Applying sufficient torque to the bolts to close the fixture is
difficult. Therefore, the entire setup is again visually inspected to see that no
damage has occurred. Finally, desiccant is placed in the acrylic box and the
box is closed and sealed.
Ideally, these procedures are performed and tests are successfully
started on the first try. However, it is sometimes necessary to adjust shim
sizes, repair damage, and repeat other procedures. It is also sometimes
necessary to delay tests to allow coils to relax after unsuccessfully attempting
to start a test.
RESULTING DATA AND ANALYSIS

Figure 2 illustrates the relationship between load and time for each test.
This data is assumed to be representative of the relationship between stress
and time, since stress is force per area, and the area is assumed to be a
constant 1.38 square inches (8.90 square cm). It is conceivable that projected
estimates of long term stress decay can be extrapolated directly from such
data. Conventional stress relaxation theory is used in analyzing this data.
One method of analysis utilized is to try fitting a curve to the data. The result
of curve fitting is an equation which theoretically represents the data, thus
allowing more accurate long term stress loss estimation. An equation of this
form can represent load or stress. Lo is the load at time t = 0, Ll is the load at
354
time t = 1, t is the time value, n is a constant, and L is the corresponding load
value.
(2)
Direct comparison of the raw data from each of the different tests is
complicated by variations in starting load (Ls) values. Simply adjusting the
scales to compare data is inappropriate. It is advantageous to modify the
data by dividing through each dataset by its L0 value, thus normalizing all
the data sets to start at the value one. The L0 values are first taken as the
highest load value in each data file (L0 = Ls). The actual initial loads are
probably higher, but are undetected because the loads change very quickly at
the start of a test. Ideally, at the start of each test, the load would be applied
instantaneously, and the L0 value would be known. However the load is not
applied instantaneously and L0 must be estimated. Therefore, estimated Ls
values are adjusted slightly (within .2%) to bring the normalized data curves
together near the beginning. The resulting curves are shown for comparison
in figure 3.
For a different perspective, the beginnings of the tests are examined by
again adjusting the initial load values, but this time normalizing the curves
to match at time (T) = 300000 seconds. This arbitrary time value was chosen
because it is relatively short term, (between 3 and 4 days), and allows for
clear comparisons of the early portion of the tests, but is still long enough to
be past uncertainties in starting times due to non-instantaneous load
application. The figure 4 data emphasizes the dissimilarity in the initial
shape of the curves and figure 3 shows overall differences in relaxation.
These differences are probably due to coil variations for the most part and
may indicate little more than that the copper dominates the total coil
relaxation.
355
50MM DIPOLE COIL
NORMALIZED STRESS RELAXATION DATA
0.98 -r-------------------------,
SK = Special Kopton
NF = No Fiberalass
BW Bore Wire
RE = Regular Coil
~ 0.97
g
...J
-c(
E:
z 0.96
o-c(
~:::===~RE SK
g 0.95 BW
NF
0.94 +-------r-----.-------,.-----..,...---~
0.0 0.5 1.0 1.5 2.0 2.5
TIME (Millions of Seconds)
Figure 3. Stress Relaxation Data normalized to match near the beginning
DATA INTERPRETATION
The expected order of initial relaxation rate, from highest to lowest, is
RE, NF, and BW; (SK is unknown). The apparent relationship between SK
and RE may indicate that the absence of the fiberglass/epoxy does not greatly
alter the initial relaxation rate, which is not unreasonable to expect.
However, the close grouping of NF and BW is surprising. Assuming that the
polyimide is the material that does the most relaxing, it is reasonable to
expect that NF would initially relax at a higher rate than BW.2
It is not yet known if this data is repeatable. It is possible that, through
repetition, it may become obvious that this individual comparison is not a
good representation of the relative properties of this coil type. On the other
hand, if this did prove to be a good coil model, it may be that the properties
were altered in an unpredictable manner. One possible explanation for a
high rate of relaxation for the bare wire coil is that the affect of the change in
contact area (between the wire strands and insulation) is overshadowing the
stress relaxation. When a coil is compressed, the stress is equal to the yield
point of copperS and the shape of the wire strands and insulation changes.
Perhaps the insulating components and wire strands are compressed in
such a manner as to increase the contact area, thusly reducing the stress.
This variation cannot be controlled because of variation in wire strand
diameter and insulation thickness. As a result, a great deal of test data
variation occurs.
The normalization implies a linear relationship between stress and
strain at any time, which is probably not true at stresses equal to the yield
stress. Therefore, a much better comparison would be to start all
measurement at the same stress value. But this is not possible, because true
internal stresses are unknown. Even starting at the same overall stress is
experimentally difficult.
356
50MM DIPOLE COIL
NORMALIZED / EARLY PORTION COMPARISON
0.98 -r----------------------"'T'" 0.98
SK = Special Kaptan
NF = Na Fiberglass
0 BW = Bare Wire
<C(
0
..J
RE = Regular Coil 0.97
0.97
..J
<C(
E
z
'-
0.96 0.96
0
<C(
0
..J
0.95 +----,----T-"----r-----,r----+ 0.95

0.0 0.5 1.0 1.5 2.0 2.5
TIME (Millions of Seconds)
Figure 4. Data normalized for early portion comparison
CONCLUSION
The actual areas of contact between cable strands in the coil (i.e. the true
area of support) is the most important element in determining stress
relaxation. Stress relaxation in the bare wire appears to dominate the total
coil stress relaxation. It is very difficult to extrapolate long term stress loss,
but we are currently preparing to make such an estimate.
REFERENCES
1. Carson, J.A. and F.W. Markley, "Mechanical Properties of

Superconducting Coils," IEEE Transactions on Magnetics, Vol.
Mag. 21 #2, March 1985, pp. 85.
2. Markley, F.W., and J.S. Kerby, "Investigation of the Mechanical

Properties of Superconducting Coils," Super Collider 2, Vol. 2, Plenum
Press, New York, 1990, pp. 753-763.
3. Markley, F.W., et. aI., "Stress Relaxation in SSC Collared Coils and their
Component Materials," Super Collider 3, Vol. 3, Plenum Press, New
York, 1991, pp. 155-163.
357
THEORETICAL METHODS FOR CREEP AND STRESS
RELAXATION STUDIES OF SSC COIL
J ames McAdams and Finley Markley
Fermilab National Accelerator Labol'atory*

Box 500
Batavia,IL 60510
ABSTRACT
Extrapolation of laboratory measurements af sse coil properties to the

actual construction of sse magnets requires mathematical models of the
experimental data. A variety of models were used to approximate the data
collected from creep and stress relaxation experiments performed on Kapton
film and sse coil samples. The coefficients for these mathematical models
were found by performing a least-squares fit via the program MINUIT.
Once the semiempirical expressions for the creep data were found,
they were converted to expressions for stress relaxation using an
approximation of the Laplace integral relating the two processes. The data
sets from creep experiments were also converted directly to stress relaxation
data by numeric integration. Both of these methods allow comparison of data
from two different methods of measuring viscoelastic properties.
Three companion papers presented at this conference will present:
1) Stress relaxation in sse 50mm dipole coil.

2) Measurement of the elastic modulus of Kapton perpendicular to
the plane of the film at room and cryogenic temperatures.
3) Temperature dependence of the viscoelastic properties of sse coil
insulation (Kapton).
INTRODUCTION
To prevent mechanical motion within the magnets during sse

operation, these magnets are manufactured so that the coils are under a
compressive stress. However, any material which is constrained in such a
manner will show a reduction in stress over time. Due to the long storage
* Operated by the Universities Research Association, Inc., for the U.S.

Department of Energy.
Supercollider 4, Edited by J. Nome 359

periods and operating lifetimes of the SSC magnets, this stress relaxation
could be considerable, and must be accounted for to avoid a reduction of
stress which would allow movement of the coil during cool-down and
operation.
In order to predict the long-term effects of this stress relaxation, a
variety of experiments were performed, as discussed in the companion
papers to this one. The emphasis of this paper will be the detailing of the
theoretical analysis used in the interpretation of data from these
experiments. The most important of these procedures is the calculation of
the Time-Temperature Superposition factor (allowing viscoelastic data to be
extrapolated to different temperatures or intervals), the conversion of data
among viscoelastic functions, and finding accurate models to describe the
behavior of viscoelastic functions.
METIIODS
In order to extrapolate data from relatively short term viscoelastic

experiments to long term effects, some method of deriving an equation
describing viscoelastic behavior from the creep data must be used. To test
various models and methods of fitting them to data sets, we selected a
representative data set from the work from Lautenschlager and compiled a
list of possible forms for the viscoelastic function from the available
literature.
To fit the models to the data set, we tried three methods: log-log plots,
selecting an appropriate number of likely data points and solving the model
as N equations in N unknowns, and by the least-squares method using
CERN's function minimization program MINUIT. We found that the best
fits to the data set were consistently given by the MINUIT program, and
performed all further work using this method of fitting models to data
excl usi vely.
Once we had a standard method, we used this method to examine a
variety of potential models. The models which we tried were:
I::: = £0 + £I t n (1)
£ = £0 + £el tn + £2t (2)
I::: = £0 + £1ln(t) (3)
£ = £0 - £0 e(-Ue 1), the Voight mod,)l of viscoelasticity. (4)
£ = £0 + £1 t + £2t2 , (5)
where £0, £1, £2, £3, and n are coefficients which are to be varied to provide the
best fit to observed data. All of these equations, except the polynomial
equation, have been reported to have been used successfully to model creep
data 1,2,4. The results of the fittings are displayed in Figure 1. For the
arbitrary data set selected, the best model of creep behavior is £ = £0 + El tn.
One complication of this work on finding useful models for visco-
elastic behavior is that a large set of values for the unknowns of our chosen
model equation exist which give similar results as shown in Figure 2.
Therefore the results of this parameterization should only be taken as
approximate solutions, and may vary considerably. More confidence can be
placed in the parameterization if anyone of the terms (such as £0, the time-
independent strain) can be accurately determined, since this limits the range
over which the other terms can be varied.
360
Comparison of Creep Models
.0022
.0021
-
~
'F-'
.0020
.... 3- ~
.0019
~~
c 4:#~
'0
...
?r
.0016
..-
Ul
.0017
.0016
.0015
o r;F0 ..,..n<:§F r;F0,.,c§)0 r;F0 000 r;F0 000 000 000

'I-,:>(j -Ju - 1,:>(j ,rfS"'- ,'1-,:>0 ,<F(j ,1,:>(j 'l-crF '}.'I-,:>(j '}.,:>O(j
Time In Seconds
Figure 1. Comparison of Creep Models.
361
Multiple fits to Creep Data
.0024
.0022
.0020
N=.OI
~~
c:: ~
~.15
-
o
....
Ul
.0018
N~
V!
.0016
° 'l-rF°rF ~~O IOrF# '0',##,'l-rF#, ,.cF#,IOOOrFO, 'OrF#'l-O#~'l-#rF

Time
Figure 2. Varying parameters of the model.
The second goal of our work was to find a method which would allow
us to obtain a model of stress relaxation behavior based on data gathered
from creep experiments.
The relationship between creep and stress relaxation is given by the
convolution integral f(E('t)D(t-'t)dT = t, or alternately by the LaPlacian
expression L{D}L{E} = 1/p2. Attempting to solve the LaPlacian expression
requires that a function for either D or E is known.
Having chosen the expression £ =£0 + q t n as the best representation of
creep behavior, we now proceed with solving the LaPlacian relationship for
an expression of stress relaxation behavior.
We start with the form of our creep model, the LaPlacian relationship
between the tensile modulus and tensile compliance, and the definition of
compliance:
~
L{E}L{D} = 2
p £ = O"D
£0 £1
assuming that 0" is a constant, D = Do + Dltn, where Do = - and Dl =-.
0" 0"
Substituting, we find: L{Do + Dltn} = ~
362
02
L{E} = --.L 0 + D r(n+1))
p2 ~ P 1 pn+l
L{E} = 1 + D 2r(n+1)
IToP IP pn+l
At this point we found no way of proceeding with the evaluation of the

anti-LaPlacian of the term on the right, and introduced the following
approximation in order to proceed:
11(1+0) - 1 - 0 for 0« 1.
Using this, we convert the expression for L{E} to:
Dl r(n+12
L{E} = 1ID op (1- DP n+l )
o P
Dl J(n+1)
L{E} = 1ID op - D o2p
n+l
Although the approximation greatly reduces the utility of this result,

we feol that the general form of the resultant equation is still useful, and the
approximate solution provides a check on the reasonableness of later results.
Based on this derivation, we chose to use the model 0(t) = 00 - 01 t n in our
efforts to model stress relaxation behavior.
The other method of obtaining a stress relaxation model from creep
data is to begin by performing the viscoelastic conversion upon the data sets
by the use of the convolution integral fE(t)D(t-T)dT = t. We performed this
integration following the method outlined by Hopkins and Hamming3 on a
representative data set. This converted data is shown in Figure 3, along with
data gather from stress relaxation experiments at the same temperature.
The converted creep data exhibited a much steeper slope than the stress
relaxation data. This suggests that while the creep data cannot be used as an
accurate predictor of stress relaxation behavior, it should provide an upper
bound for such behavior.
363
stress Relaxation Data and Converted Creep Data
Kaplon at 55 C
1.00
0.95
"0
0
0 0.90
....I
l\
0
:E
c 0.85
""'-- r--
"-
"0
~ -
---
0 0.80
0
....I
~~ Sire.. Reljxaiion Dalal
0.75 r-.
I- S're.. ReljxaflOn Dolo
:-:-<-.
Converted ICreep Dalal
I
0.70
o 100000 200000 300000 400000 500000 600000 700000 800000
Time In Seconds
Figure 3. Comparison of Stress Relaxation data and Creep Equivalent.
REFERENCES
1. Garafalo, "Fundamentals of Creep in Metals," Macmillan, New
York, NY 1965.
2. F. J. Lockett, "Nonlinear Viscoelastic Solids," Academic Press, New

York, NY 1972.
3. I. L. Hopkins and R. W. Hamming, "On Creep and Relaxation," Journal

of Applied Physics, Vol. 28, p. 906, 1957.
4. John D. Ferry, "Viscoelastic Properties of Polymers," John Wiley & Sons,

New York, NY 1980.
364
QUENCH PERFORMANCE OF FERMILAB/GENERAL DYNAMICS BUILT
FULL LENGTH SSC COLLIDER DIPOLE MAGNETS
J. Strait l , D. Orris 1, P.O. Mazur l , M. Bleadonl,

R. Bossert l , J. Carson l , S.W. Delchampsl,
A. Devred2 , J. DiMarc02 , S. Gourlayl, R.Hanft l ,
W. Koska l , M. Kuchnirl, J. Kuzminski2,
M.J. LammI, W. Nah2, T. Ogitsu 2 , J. Ozelis l ,
M. Puglisi2, LC. Tompkins 2 , M. Wake l ,
Y. Yu 2 , Y. Zha0 2 , and H. Zheng 2
1Fermi National Accelerator Laboratory

P.O. Box 500
2SSC laboratory
INTRODUCTION
In this paper we present results of quench testing offulliength SSC dipole magnets at
Fermilab. The data are from the ftrst six of a series of thirteen 15 m long, 50 mm aperture
SSC dipole magnets which are being built and tested at Fermilab. These magnets were
designed jointly by Fermilab, Brookhaven Laboratory, Lawrence Berkeley Laboratory and
the SSC Laboratory. Among the major goals for this series of magnets are to transfer magnet
production technology to the lead vendor for the Collider Dipole Magnet, the General
Dynamics Corporation, and to demonstrate industrial production by the vendor. The first
magnet in the series, DCA311, was built by Fermilab technicians to establish assembly
procedures. The second magnet, DCA312, was the "technology transfer magnet" and was
built jointly by Fermilab and General Dynamics technicians. The next seven, DCA313-319
are being built by General Dynamics personnel using Fermilab facilities and procedures.
However, Fermilab personnel still operate the major tooling, provide the welders, perform
assembly of items that would not be part of production magnets (e.g. voltage taps), and
oversee the QA program. Five of these 7 GD-built magnets will be used in the Accelerator
Systems String Test (ASST) to be carried out in Dallas later this year. The last four magnets,
DCA320-323, are being built by Fermilab alone.
The design of this magnet has been previously described 1,2,3.4. It consists of a cos e
style coil clamped by 17 mm wide stainless steel collars. A vertically-split yoke is employed
to provide mechanical support to the collars near the horiwntal mid-plane to limit deflections
* Operated by the Universities Research Association, Inc. for the U.. S. Department of Energy.

under the Lorentz force to < 0.02 mm. The coil ends have current blocks, defined by
machined G 10 spacers, that match those of the 2-dimensional cross-section. The inner-
outer coil splice is made outside the coil at the lead end. The coil ends are clamped by a collet
assembly consisting of a 4-piece G 10 insulator with a tapered outer surface and an aluminum
cylinder with a tapered inner surface. The coil end is preloaded axially against the magnet
end plate with four set screws at each end to provide axial restraint under excitation.
The magnets are instrumented with 53 voltage taps. These are concentrated in the six
inner coil turns nearest the pole and allow determination of quench locations to within several
centimeters for quenches that originate in these turns. Strain gauge load cells measure coil
azimuthal stress at the pole and axial force at the coil ends5. The azimuthal coil stress gauges
are placed at two locations, where the average coil size is largest and smallest. Strain gauges
on the magnet shell measure axial and azimuthal stress changes with excitation. The magnet
tests were carried out at the Fermilab Magnet Test Facility6.7.
In the second section of this paper the spontaneous quench performance and the
temperature dependence of the quench current are presented. Tests of the quench protection
system are presented in section three. The dependence of the quench current on the ramp rate
is compared with measured energy loss per excitation cycle in the fourth section. Other test
results from these magnets are presented in two companion papers 8•9 .
SPONT ANEOUS QUENCH PERFORMANCE

Figure 1 displays the quench histories of the six magnets tested. Unless otherwise
noted, all quenches in this plot occurred at a ramp rate ~ 4 A/sec. In most cases, the first
quench of a sequence for a magnet occurred at a nominal ramp rate of 16 A/sec during a run
to collect strain gauge data as a function of current. In such a run the magnet is ramped up in
steps of roughly 500 A and strain gauge data are collected at each step. The step-wise nature
of the ramp yields an average ramp rate of < 5 A/sec above 5000 A. Quenches 2-5 of the
first cooldown of DCA311 were taken at 16 A/sec. Because of the significant ramp rate
dependence of the quench current (see below) these quenches are at a lower current than the
others. Once the ramp rate sensitivity was recognized, the lower ramp rate was used for all
spontaneous quench tests.
On the initial test cycle at 4.35 K (Fig. la) all magnets exceeded the SSC operating
current of 6600 A (6.7 T) on the first or second quench, and by the second quench were
within about 100 A of the conductor limit. DCA313, 314 and 316 each had one quench
below the operating current. In the first two of these magnets the quench occurred, within a
measurement uncertainty of ±15 mm, at the same location: in the pole turn of the inner coil on
the side opposite the splice within ±15 mm of the collar - end clamp boundary. The cause of
the training quenches is unknown. The first quench in DCA316 occurred in the
uninstrumented portion of the inner coil, so a precise quench location cannot be determined.
Figure 1b shows the initial quench currents at 4.35 K on the second cooldown. All
magnets reached within 50-60 A of the conductor limit on the first quench. Much of the
quench-to-quench variation in the current results from temperature fluctuations. One of the
two test stands used has poorer temperature control than the other and all the even numbered
magnets were tested on this stand. Figure 2 plots the quench current versus the magnet
temperature for all quenches with dIldt ~ 4 A/sec; however, the three low training quenches
are off scale. The temperature is measured at the two ends of the magnet and the higher of
the two (normally at the helium outlet end) is used. Data from the first (Figs. 2a and 2c) and
second (Figs. 2b and 2d) are plotted separately. Magnets with inner coil conductor made by
Supercon (DCA311) and Oxford Superconducting Technology (DCA316) are in Figures 2a
and 2b and those with conductor made by Intermagnetics General Corporation (IGC) are in
Figures 2c and 2d. Shown also are the expected temperature dependences using Green's
paramaterization of the critical surface 10 •
On the second cooldown magnets are tested at 3.8 K and then 3.5 K. The low
temperature quench results are shown in Figs. lc, 2b and 2d. Only DCA315 exhibited
training at 3.8 K. (The low quench in DCA316 results from a temperature fluctuation as can
be seen in Fig. 2b.) Only DCA312 had one initial training quench at 3.5 K. However,
DCA314 "de-trained" after reaching the conductor limit on the first 3.5 K quench. It
however returned to normal behavior when the temperature was restored to 4.35 K. The
cause of this behavior is unknown and limitations on the testing time prevented further
investigation of the effect. Strain gauge data 8 were taken at currents up to 8 kA while
366
ramping to the fIrst 3.5 K quench. These data indicate that the coils remained azimuthally
loaded, that the yoke mid-plane gap remained closed, and that the axial force transfer to the
shell was similar to the other magnets. Among all magnets, four of the fIve mechanically
induced (training) quenches at low temperature occurred in the outer coil. Again, the limited
testing time has not allowed us to understand this behavior.
• ••• ••
(a)
• ••• •• ••••• ••••• • • ••• • •••
7000 -
• 20 • ~
Tev Operat ng Curr nt

- - - - - ------ - - - - - ------ - - - - - ------
•
6000 - i-
•
--
<
....c::
5000 - • ,-
II)
DCA311 DCA312 DCA313 DCA314 DCA315 DCA316
1.0
1.0
='
• •••• ••• • ••• (b)
U
..d •••• • •••• • • •••
7000 -
()
c::
II)
t-
=' 20 Tev Operat ng Curr nt

a - -- - - - - -- -- - - -- ----- -- - - - - - - -- - - -
....' ...., ...' ...,

I I ••• I··· I·
I I··· I
(c)
6000 - I.·· ,0
I
•••1 01
I
I
I
•• .,
I···
i-
I I I I d
I
0 I • I
I I I I I I
I.
I I I I I I
I
7000 - I I I I
I
I I i-
I I I
- - -1- _
20 ~e~ _9y~r_at ng L _ _ -nt- -1-
_ _ burr I I
__ - - -1- _ ---'---
Quench Sequence
Figure 1. Spontaneous quench histories at 4.35 K on the fIrst cooldown (a) and at 4.35 K (b) and 3.8 K and
3.5 K (c) on the second cooldown. Closed (open) circles are quenches that originate in the inner (outer) coil.
In (c) the data to the left (right) of the vertical dashed line are at 3.8 K (3.5 K). To the right of the second
dashed line for DCA314 the temperature is 4.35 K.
QUENCH PROTECTION SYSTEM TESTS

Quench protection is provided primarily by heater stripsll mounted between the layers
of Kapton insulation between the outer coils and the collars. Two of the four heaters, located
in opposite quadrants, are wired in parallel; the other two are spares and are not used. When
a quench is detected the power supply is turned off, causing the magnet to discharge at an
initial rate of about 1 kNs (1 TIs) into the resistance of the power system, and the protection
heaters are energized by discharging a capacitor bank of 36 to 52 mf charged to 400-450 V.
On the first magnet tested the characteristics of the heaters were established by varying the
energy deposited until the coil was observed to quench at 4.35 K with a time delay < 0.20 s
at 2 kA and < 0.12 s at 5 kA. The time integral of the current squared (MIlTs, see below)
from the time the heaters were energized is < 4 and 10 x 106 A2 s respectively. At 4.35 K a
36 mf capacitor charged to 440 V was used, resulting in a time constant of 0.24 s and an
energy deposition per strip within 0.12 s of 1 kJ. For lower temperature operation a larger
capacitance is used to give a larger energy deposition with a longer time constant.
367
I 1 I I I I I I
8500 - (a) 1', '- , (b) r-
, , '- ,
, '- ,
+,', , '-, '-
"
-
....':t:-,
..,', , , ,
8000 >-
, '-,
'- ,
'-
, ,"-,
7500 - , ""- ''-..,
..'
!-
........ +
l
~
'
...,
•+
"
DCA311 "-
It'"
d
CI)
r.. 7000 - DCA316 !-
r..
='
U
8500 - (c) (d) !-
..c::
C) ~, ,
d
CI) , .... ,
='
CY
0
,
8000 - • 4i"v , ....
r-
,,
• v ,,
7500 - DCA312 !-
~\"
0 0
x DCA313
• DCA314 ~
7000 - V DCA315 !-
I I I I I I I I
3.50 3.75 4.00 4.25 3.50 3.75 4.00 4.25 4.50
Temperature (K)
Figure 2. Quench current versus test temperature on the flrst (a and c) and second (b and d) cooldowns. The
curves in (b) and (d) indicate the calculated magnet critical currents. In (b) the dashed (dot-dashed) curve is the
calculation for DCA311 (DCA316).
In an adiabatic approximation the peak temperature of the conductor is given by the

time integral of the current squared from the time the quench begins. This integral is
conventionally expressed in units of 106 A2 s and is called "MIITs." At low current the
MIlTs are low because the current is low, and near the critical current the MIlTs tend to
decrease since the quench propagates rapidly. The maximum MIlTs, and therefore the
maximum conductor temperature, occur at some intermediate current. In these magnets the
outer coil, which has less copper in each turn, reaches a higher temperature for the same
number of MIITs. An operation limit of 15 MITTs, corresponding in the adiabatic limit to a
peak temperature of 620 K for outer coil quenches and 200 K for inner coil quenches, is
applied to these magnets. To determine the maximum MITTs as a function of magnet current
a series of quenches were induced at 4.35 K in each magnet using a spot heater located near
one end of the inner coil mid-plane. The quench protection heaters are energized when the
quench was detected. (The time delay between the first resistive voltage and quench
detection varies from about 45 msec at 4.5 kA and 20 msec at 6.5 kA to < 5 msec near the
critical current.) In magnet DCA311 the heaters were energized 35 msec after the quench was
detected; in all other magnets no delay was used. The results of this test are shown in Fig. 3.
The maximum MIITs occurs near 6 kA, or 80-85% of the critical current, in all the magnets,
and there is a roughly 20% spread in the MITTs among the magnets.
The extent to which the heaters are required for quench protection can be tested by
progressively delaying their energization. The MIITs as a function of the heater delay, for
368
spot heater induced quenches at 6 kA, is plotted in Fig. 4. For delays ~ 100 msec the MIlTs
vary linearly with the delay time. For longer delays the MIlTs increase less rapidly and for
several of the magnets the MIlTs appear to be approaching a constant value for delays around
150 msec, with some magnets above and some below the limit of 15 MIlTs. Thus, while at
least some of the magnets appear to be self-protecting for single magnet tests, the series as a
whole cannot be considered self-protecting.
• DCA311
14 + DCA312
-
0 DCA313
fIl x DCA314
N 4 DCA315
<
<0
....0 12 "
fIl
t:
SJ
10
4000 5000 6000 7000

I q (A)
Figure 3. Time integral of 12 (MIlTs) versus quench current (Iq) for spot heater induced quenches. The
protection heater fIring delay is 35 msec for DCA311 and zero for all other magnets.
"......
fIl
C'I 14
.,<
....0
t::
rn
12
• DCA311
+ DCA312
il 0 DCA313
X DCA314
4 DCA315
10+-~~~~~~~~~-.~~~~~.-~~~=r=+
o 50 100 150
" DCA316
200
Strip Heater Firing Delay (Insec)
Figure 4. Time integral of 12 (MIlTs) versus protection heater delay for spot heater induced quenches at 6 kA.
RAMP RATE DEPENDENCE AND AC LOSS MEASUREMENTS

Eddy current heating of the cable copper matrix while the magnet is being ramped can
increase the conductor temperature and reduce the quench current. The quench current as a
function of ramp rate is shown in Fig. 5. The quench current is very sensitive to ramp rate in
these magnets. In four of the magnets (DCA312-315) the quench current is roughly constant
369
up to dIldt =25 Ns and then drops rapidly to, in the most extreme case, roughly 2 kA at 200
Nsec. In the other two magnets the quench current drops more rapidly at lower ramp rates,
being noticeably reduced even at 16 Ns, but then decreases less rapidly at higher ramp rates.
(In DCA316 the fIrst quench at 200 Ns was about 600 A higher than expected from an
extrapolation of lower ramp rate data. A second 200 Ns quench was at the expected current.
The cause of this behavior is unknown.) In all cases the higher ramp rate quenches originate
in the uninstrumented turns closer to the mid-plane of the inner coil.
8000+---~~--~--L-~--~--~--~--~~--~--+
6000
-<-
0'
4000
• DCA311
+ DCA312
2000 0 DCA313
• DCA314
... DCA315
v DCA316
0
0 100 200 300
dI/dt (A/s)
Figure 5. Quench current Oq) versus ramp rate (dJ/dt).
The dramatically different behavior of these two sets of magnets correlates with the
vendor of the superconducting wire used in the inner coils: DCA312-315 use IGC conductor,
DCA311 uses Supercon conductor, and DCA316 uses Oxford conductor. It is not known
what different wire processing parameters result in the different magnet behavior. There is
also considerable variation in the high ramp rate behavior among the four magnets made with
cable from the same vendor, although the wire, cable and coil manufacturing processes are
all nominally the same for these magnets. It appears that there are processing variable which
effect the ramp rate sensitivity which are not adequately controlled. While this degree of
sensitivity is is unlikely to be a problem for the Collider, which ramps at 4 Ns, it may be a
problem during fast ramp-down if magnets with modestly more eddy current losses than
these are manufactured12• Also, similar magnets are to be used in the High Energy Booster,
which must ramp at 70 A/s, and magnets of this quality would clearly degrade its
performance.
To learn whether this ramp rate sensitivity and the magnet-to-magnet variations result
from large eddy current heating or from poor cooling, energy loss measurements were
performed on four of the magnets. Measurements were made using the system described in
Ref. 13, which simultaneously samples the magnet voltage and current using fast integrating
digital multimeters. Data were taken using a simple sawtooth ramp between 500 and 5000 A
with ramp rates varying from 30 to 150 Ns. Data were taken from two complete cycles at
each ramp rate after at least three full cycles had been completed. The measurements are
plotted in Fig. 6. Shown also is the average of measurements of three 1.5 m model
magnets 14 multiplied by 10. (The ratio of superconductor volume in long to short magnets is
10 and the ratio of magnetic lengths is 11.) A linear least squares fIt is done to the data from
each magnet. The dIldt = 0 intercept gives the superconductor and iron hysteresis loss, and
the slope characterizes the eddy current losses. The results are summarized in Table I. The
hysteresis losses are, within estimated errors, consistent among the magnets, but there is a
370
Table 1. AC Loss Measurements
Hysteresis Eddy Current

Mal!net Loss (J) Loss (]f(NS)
DCA 311 530 ± 130 16.2 ± 1.3
DCA 312 1160 ± 220 57.3 ± 3.6
DCA 314 840 ± 120 35.1 ± 1.3
DCA 315 780 + 150 50.9 + 2.2
Averal!e 780 + 130
Short Magnet Average 107 + 4 0.64 + 0.04
6000
-.........
Zl 4000
oS
~
Q)
~ 2000 -------
----sho~i
Magnets
(x10)
O+-~~~~~.-~~--~~-.~--~~~-.--+
o 50 100 150
dI/dt (A/s)
Figure 6. Energy loss measurements for a 500-5000-500 A cycle versus ramp rate (dI/dt).
-~ o
o
.....
""-
~ 40
III
III
C1.ong Magnets
oS
20
o
Short Magnets (x10)- - -
-30 -20 -10 o

dI q /di (A/(A/s»
Figure 7. Eddy current loss versus quench current ramp rate slope (dIqldI)
371
wide variation among the eddy current losses. Those magnets with the greatest ramp rate
sensitivity also show the greatest eddy current loss. This relation can be quantified by fitting
a straight line to the quench data in Fig. 5 for dIldt ~ 50 A/s. The eddy current slope is
plotted as a function of the quench current slope in Fig. 7 and an essentially linear relation is
observed. The short magnets eddy current data, when multiplied by 10, lie on the same line.
Thus we conclude that the ramp rate sensitivity and its magnet-to-magnet variation results
dominantly from eddy current heating rather than from poor and variable cooling.
CONCLUSIONS
Six full length 50 mm aperture SSC dipole magnets have been built and tested at
FermiIab. The spontaneous quench performance is good, with three of the six magnets
comfortably exceeding the operating current without quenching, and the remaining three
having one quench each below the operating current. All magnets quench at or very near
their conductor limit on the fIrst or second quench. At 3.5 K all magnets reached fIelds
above 8 T. Tests of the quench protection heaters were performed which indicate that they
are quite adequate for single magnet tests. However, the large energy deposition required
may make the use of conventional capacitor discharge fIring units impractical in a string of
magnets in which several heaters are wired in series. The quench current is quite sensitive to
ramp rate in these magnets and the ramp rate sensitivity correlates well with eddy current loss
measurements. The surprisingly large eddy current losses and the significant magnet-to-
magnet variations indicate that some parts of the manufacturing chain are not adequately
controlled. Ramp rate sensitivity and AC losses at this level are probably not a problem for
the Collider, but could be a problem for the High Energy Booster.
ACKNOWLEDGEMENTS
We would like to thank the FermiIab Superconducting Magnet Fabrication Group and
of the General Dynamics Space Systems Division staff stationed at Fermilab for producing
such splendid magnets. We also thank the staff of the FermiIab Magnet Test Facility for their
hard work that has allowed so much lovely data to be collected in such a short time.
REFERENCES
1. R.C. Gupta, et aI., SSC 50 mm Dipole Cross-Section, Sypercollider 3, p. 587,1991, J. Nonte, ed.
2. J. Sttait, et aI., Mechanical Design of the 2D Cross-Section of the SSC Collider Dipole Magnet, fmc.. Qf
~.l22!mllli ~ Accelerator Cwlf., p. 2176, 1991, L. Lizama and J. Chew, eds.
3. J.S. Brandt, et aI., Coil End Design for the SSC Collider Dipole Magnet, ibid., p. 2182.
4. S.W. Delchamps, et aI., SSC Collider Dipole Magnet End Mechanical Design, ibid, p. 2185.
5. C.L. Goodzeit et aI., Measurement of Internal Forces in Superconducting Accelerator Magnets with Strain
Gauge Transducers, ~Inms..Mag.. Vol. 25, p. 1463, 1989.
6. P.O. Mazur and T.J. Peterson, A Cryogenic Test Stand for Full-Length SSC Magnets with SuperfIuid
Capability, AdvancesinCt:yQ~nics, Vol. 35, p.785, 1990.
7. J. Sttait, et aI., Fermilab R&D Test Facility for SSC Magnets, Supercollider 1. p. 551, 1989,
M. McAshan, ed.
8. M. Wake, et aI., Mechanical Behavior of FermilablGeneral Dynamics Built 15 m SSC Collider
Dipoles, presented at the 4th International Industrial Symposium on the Supercollider New
Orleans, LA, March 4-6, 1992.
9. S. Delchamps, et aI., Magnetic Field Measurements of FermilabiGeneral Dynamics Built Full Scale SSC
Collider Dipole Magnets, presented at the 4th International Industrial Symposium on the
Supercollider, New Orleans, LA, March 4-6, 1992.
10. M.A. Green, Calculating the Jc, B, T Surface for Niobium Titanium Using a Reduced State Model,
IEEE Trans. Mag. Vol. 25, P 2119, 1989.
11. C. Haddock, et aI., SSC Dipole Quench Protection Heater Test Results, ~ Qf ~ l22! IEEE ~
Accelerator Qmf.., p. 2215, 1991, L. Lizama and J. Chew, eds.
12. G. Snitchler, private communication.
13. M.J. Lamm, et aI., Bipolar and Unipolar Tests of 1.5 m Model SSC Collider Dipole Magnets at
Fermilab,~Qf~l22llEJili~AcceleratorQmf..,p.2191.
14. MJ. Lamm et aI., Results from the Fermilab 1.5 m Model Magnet Program, presented at the 4th
International Industrial Symposium on the Supercollider, New Orleans, LA, March 4-6, 1992.
372
SSC DIPOLE MAGNET MANUFACTURING
FACILITY IMPLEMENTATION
Ken Miller and Pat Cowan
General Dynamics
Space Systems Division
Plant Operations (MZ: CI-7430)
P.O. Box 85990
San Diego, CA 92186-5990
INTRODUCTION
General Dynamics has been selected to complete the design, develop the manufacturing
systems, and begin production of the dipole magnets for the Superconducting Super Collider
(SSC) in its facility in Hammond, Louisiana. The purpose of this article is to document
recent development in the manufacturing facilities plan for SSC dipole magnets (CDM).
These achievements are the result of the concurrent engineering effort in the SSC Program on
the design and manufacturing of the magnet systems.
Hammond Facility Introduction
The General Dynamics Hammond Facility is located in southeastern Louisiana. The

selection criteria used by GD included facility compatibility and availability, transportation
access, labor availability, state and municipal participation, and quality of life characteristics.
A total of 27 different sites were evaluated. The Hammond Facility best met these
requirements and also provided features that make it a low risk and low cost facility for the
magnet program. The facility, as shown in Figure 1, has a total of 28.5 acres. The existing
building has a total of 227,000 sq ft and a 39 ft eave height, is suitable for factory space and
will support the total program space requirements through the ten-per-day rate production
requirement. The facility will support the shuttle material handling system required to
manufacture the magnets. Ample utilities support is available at the site. The facility is air
conditioned, aiding in the environmental requirement for the operation. These features
minimize the modification requirements that will be accomplished in 1992 to ready the facility
to support start of prototype production in October 1992.
Manufacturing Process
The manufacturing process to assemble the CDM is comprised of thirty-one (31)

process steps, requiring 116 production stations at a production rate of 10 magnets per day.
The movement of the parts between station will be handled by an integrated floor
transportation system, known as the "shuttle."

Figure 1. The Hammond Facility
Program Production Schedule
The production program of the CDM is broken down into four major phases:
Prototype, Preproduction, Low Rate Initial Production (LRIP), and Production. Each phase
of the program has specific goals and objectives. The prototype phase starts in October '92,
consists of 12 magnets, and will technically qualify the design of the magnet. The
preproduction phase starts in October '93, consists of 35 magnets, and will quality the
manufacturing process. Low Rate Initial Production (LRIP) starts in July of '94, consists of
251 magnets, and will qualify the ability of the tooling to achieve rate production quantities.
Upon final award of the production contract, the production phase will begin in mid '95 with
a production rate acceleration from one magnet per day to 10 per day.
Facility Activation Schedule
The major milestones in the activation of the Hammond Facility include activating the
office area by May 1992 and Annex I factory by October 1992. Initial prototype tools will
begin arriving in Hammond in August of 1992 and will be installed between August and
October. This requires all major construction requirements to be complete in August to
minimize interface of the tooling installation requirements to support an on-schedule start of
production. A schedule showing these key activities along with the Architectural and
Engineering tasks is shown in Figure 2.
374
HAMMOND FACIUTY ACTIVATION.
Apr ~ May : JUri Aug: s.p Oct NOY Dec. Jarl Flib "-r Apr May: Aug: Sep OCI NOY Dec Jan Flib Mar: Apr
Issult Oal.: 2120192
-
.u\PACOUCTSSEAVCE 00 00 oo..,."TfI..C1'l::Hcx:M'Oo
I'JlCUTYJlCl"III",T1CN
..
PlANNNg PHASE
• A & E Cmeria Speclh:atlon T

• A & E Selection
~
• 15' Archfteelural o.,lgn Review _i
· 30' Design Revl.w/Approval, ~:::;:::'\
· 60' Design Review/Approv",ls
· 90"'" DeSign Review/Approv",ls
·'00"'" Rewl ••,Approw",ls/P.rmlts
CONSTRICTION MQR Sf' EcuoN T . zl

• CM CrIt."a Selection Rew,.w
• Ct,I InterviewslAn.alysISlSe'ee:lion
• Bid & Award Ct,l Contract
CONSTRUCTION pHASE
• Order long lead M"'UslEqu~ment
• Bid & Award Const'n SuO·Conlracts

• Penorm Construction
=7
7
9
7
to
EapMTaOOUNQ 'MP' fMf1fIADON

• TOOlIng InSl.lllalion
• Cold TitS! Facliity ACtivation
FACILITY ACTIVATION <>

Figure 2. Hammond Facility Activation Schedule
TOOLING AND MATERIAL HANDLING
Stations Required
The span times for each station have been determined. This development effort
included an analysis of the task span times necessary to build the Industry Demonstration
magnets at FNAL and incorporated the differences in the GD magnet design, tooling and
material handling that would be experienced in Hammond. The actual GD technicians at
FNAL participated in this effort. Those assembling the magnets added invaluable insight
into the actual process sequences and durations. The rate tooling requirements have been
determined for each phase of the program and have been incorporated into the facility layout.
Integrated Tooling Approach
In the build-up and assembly of the CDM, the control and protection of the parts is
critical. Also, with 31 process steps required, a significant amount of the assembly cost is
involved in the movement of material throughout the facility. The GD tooling and material
handling approach to be used in Hammond integrates and addresses both concerns.
A key characteristic of GD's integrated tooling approach is that the configuration of
the part is controlled throughout the assembly sequence, by the tooling, until the part is
attached to its next assembly. The advantages are that the part is never refitted to another
tool, which decreases the chance of damage. By incorporating the material handling
requirements into the assembly tooling, one family of tooling is eliminated. Throughout the
facility, parts are moved to the next assembly station on the same tooling that they were built
up on. A good example of this is in the coil winding area where the wound coil remains
attached to the mandrel from winding through curing and extraction where it is placed on a
phenolic mandrel that controls its dimensions up through coil measuring and mate.
Integrated into the entire tooling family is the floor transportation material handling system
known as the shuttle.
375
Shuttle Concept
The Hammond facility will utilize a transfer shuttle system, shown in Figure 3, as the
major work-in-process material handling system. Comparative analysis of a shuttle vs.
conventional crane material handling methodologies indicate increased benefits with the
shuttle in the areas of product quality, increased worker safety, lower labor costs, shorter
product cycle spans and lower inventory/work-in-process costs. As a result, GD will utilize
the shuttle to handle all work-in-process movement throughout the Hammond Facility.
The shuttle will integrate with all of the work stations in the facilities. The shuttle
will be used to load and unload work from each station, utilizing controlled interfaces and
common transfer tooling. The shuttle will position itself at the appropriate position in the
aisle at the end of the tool and the part on its tooling will slide from the station tool in on to
the shuttle. After the unloading of the station is complete, the shuttle will index a short
distance and load the station with the next part. This rapid, efficient loading and unloading
of work stations reduces the number of assembly stations required and the manpower
involved.
In order to incorporate the shuttle into the facility, two large transfer aisles were
created on both sides of the building as shown in Figure 4. Assembly tooling is positioned
on both sides of the aisle creating a classic U shaped flow. To further simplify the material
handling, all assembly stations that require incoming details and material were placed to the
outside of the U facilitating access to the loading docks; test stands which require no
incoming material were placed to the inside. This creates a very efficient material handling
layout and optimizes the characteristics of the Hammond Facility.
CLEANLINESS AND ENVIRONMENTAL CONTROLS
The physical environment that is required for manufacturing the CDM magnets has
been established by identifying the characteristics that have the greatest effects on the quality
and performance requirements of the magnets. The parameters that must be controlled are
temperature, humidity, and cleanliness. The limits that must be maintained have been
established for each of these parameters and are summarized in terms of designated stations
in Table 1.
Table 1. Magnet Manufacturing Environments
Magnet Manufacturing Environments

Stations Teml!erature Humidit;r Cleanliness
1 - 22 67-75 (F) 30%-60% 80% -ASRAE - 52-76
23-26 60-80 (F) 30%-6OT 80%-ASRAE - 52-76
27-30 60-80 (F) No Control 65% - ASRAE - 52-76
These requirements are consistent with the SSCL Document #MSO-OOOOS4.
Requirements Implementation
To accomplish providing these requirements, a state of the art environmental control

system is being installed by GD in the Hammond Facility. This system will maintain the
medium levels of each parameter to minimize fluctuations in the manufacturing environment.
376
-
....
Q.
=
<j
o
U
.t:'"
'" '<J"
/
//' ...
~
377
eN II.N.RlAL. DYNAMIC:.
-.j SSC MAGNET ASSEMBLY FACILITY, HAMMOND, LOUISIANA
(XI SpSCIl Systems Division
Annex I Program 30% + Design Baseline Layout c::::::: - _ ..- .___•
B ·· · · · · 'y . '!h -
j ~, li ~=""r:==:::.=:=~ , €I
"
ti GI
L~ @
'I ; E3
Ii ii --,,- 7- '"\- ~.
I 0==1 : " : ~®
v~r--------------~
: . : ~
::J:l,;;1NNr"W!' '''''""1 : ~III '=i' ~..
~ 08
eTL-----~m-...i
:'::1
~:;=--.; ij. : . -'~fl -~-1l
i
: .
I~ 1&
I ---.- ---.. I ::'1 ~ 0
I =-= "'- -; :. .nm:nm,,=-___ ..
El - - - .- - - - •• ~i~ Ji'"='OH ; ~
I. o:; .t,.· ,: I d ~== ~ .
I. i.:11' -.... _--";;:-':
o ~ -I ,: . '--- .----t~ .:.
!~~I!.':~' i ~... j 1_
__, =--
.j[ ,
:j9'oFlt ~k
. '. '. :.t::I :. .1 ._____ ~ _, __________ ~ ~-=-~
- --- . :tLi
111
w
0 1: .::. i ~ --· · · r.," l "".d i ~Mw . :I : . . :;
: ' ''::= • ==:l; : " ~----------i . ~~ - '=1
-.---------. -_. _... I :: :j ~ ._..... - ,:
0 1; . __ _ ,..... ~ __ . ____ .. __ . _.... __ . _ __

:::::-.;:~:=:~;; I i! ~:.:.-~.
t~
- ,-;;.;,;;.;~:
-- -~i I:: -I---.· . . -f • ~
i -:...:. . ., !(.; ------J-- - ~ t:L_ -'. :

' , _ _ _ • • w : ' - ..
I I! =;;;:=--'-----:
. ~i --- , i[ , :; j0
0 ~..:.::. ~;.::.·O;,;;,'fi .. : I: Iiil_.. " :
fIE-- ,-- .. _-------
"'. -~ -.-:::-::':: ::--. -. -- - iM 11= ~ 1 ~o.w-~ ~
I!) ,0
1 .:-~l! : ~:.-- J' rio ~~-=i=~: ~ l ~ :': r~~~~: :i\ I
V. _~_'_ '_ - -.::..:;,~ iii .~, lP--,~~:&r;.- ..--T
(!)',~~ _=-:-.. . -F,. 1 ....1,.1 .
_._._~, ~ .~ •• • • • • • • • • • • • • • • -.c:>
000 0 ® 0 0 0 0 0 0
Figure 4 CDM Facility Layout

Layout Considerations
In order to further support the specialized requirements of manufacturing the CDM's,

the facility layout has been partitioned into three separate areas. These are the coil operations
area (Stat 1-9), the coldmass/cryostat assembly area (Stat. 10-26), and cold test/pack and
ship area (Stat 27-30). As depicted in Figure 3, the facility layout, this is supported by the
construction of permanent full height walls that will separate the most sensitive "clean"
operations from the stations in which welding and grinding will take place.
LA YOUT DEVELOPMENT
Electronic Pathfinder
The layout plan for the Hammond Facility was developed by considering all aspects
of the program requirement while utilizing the features of the existing facility. This approach
was further supported and validated by using a new tool called "the electronic pathfinder."
This is a computer-driven system that simulates in three dimensions the facility layout and
tooling plans, and allows real time simulation of physical movement characteristics.
Tooling Growth
The CDM Program requires that ramp-up of production rates occur while minimizing
impacts to on-going operations. This requirement represents one of the most aggressive
challenges of the facilitization requirements.
All requirements for each phase have been planned for in the facilities layouts. As
new tools come on-line, locations have been already identified that will allow for installations
to take place without impacting on-going operations. For the prototype and preproduction
phase, there will only be one installation of each tool type. This will occur in 1992 before
any operations begin. At low rate initial production, five stations will have an additional tool
installed. Provisions for these stations will be made during the installation of the initial
tooling. Prior to the production phase, the rest of the tools required to support a lO-per-day
rate will be installed. To support this requirement, GD will install and make operational the
new stations in the area that will not be currently in use and is separated by physical walls.
Once these stations are on-line the older stations will be relocated to their final locations, and
the wall will be removed.
In some cases, the cost to relocate the tools would be too great, and the final locations
must be identified as the baseline plan for beginning of the program. These stations are the
coil winding station, the coil curing station, the yoke and collaring press, and the cold test
stands. In the identification of the prototype and preproduction layout, these items were
considered to be monuments and are being placed in the locations that do not change through
all program phases.
ENERGY EFFICIENCY
A key consideration in facilitizing the CDM program is to plan for energy efficiency
in the factory and for the site as a whole.
Oven Placement Savings Analysis
By identifying the operations that require the most energy and the operations that emit
the most heat into the controlled environment, opportunities became apparent to reduce the
379
total facility energy requirements. One such example is the decision to place the curing ovens
in a new, specially constructed facility adjoining the existing structure. This approach allows
for co-utilization of energy resources while minimizing HV AC costs for the overall factory.
SUMMARY
The GD approach to facilitizing the CDM program has resulted in many key
advancements in magnet manufacturing facility design. These benefits will further support
the overall success of the SSC program and will be utilized in the superconducting magnet
industry for many years to come.
ACKNOWLEDGEMENT
The work described herein is being accomplished under contract to the Universities
Research Association in support of the Superconducting Super Collider project for the
U.S. Department of Energy.
380
ELECTRODEPOSITION OF COPPER ON THE
INTERNAL WALLS OF COLLIDERS IN BEAM TUBES
J.Cl. Puippe and W. Saxer
Werner Fliihmann AG, P.O. Box 630

8600 Diibendorf, Switzerland
INTRODUCTION
In the 150 years' history of industrial electroplating, copper has always been one of the
most used and researched elements. It is well established that the industrial copper
electroplating processes in all application fields (printed circuits, coins, decoration,
electroforming, electronics, electrotechnical applications, engineering, etc.) rely on
specially tailored plating baths containing organic additives such as thiourea, lactic acid,
f3-naphtoquinoline, 1.5- naphthalenedisulfonic acid, phenolsulfonic acid, triisopro-
panolamine, gelatine, glycine, dextrin, dextrose and many more I). The benefit of these
additives is to influence in a desired manner the mechanical and electrical properties of the
electrodeposits as well as their appearance. However, the additives are incorporated in
appreciable amounts in the deposits and lower their purity. While the contaminated deposits
may be perfectly acceptable for the above-mentioned applications, they are certainly not to
be used in applications in the field of high energy physics where high electrical
conductivity at liquid helium and room temperatures and high vacuum compatibility (low
outgassing rates) are required. Therefore applications such as copper plating of colliders'
beam tubes require additive-free baths in order to meet the morphological and surface
finish requirements. These baths, generally speaking, are more difficult to control than
regular solutions. High purity, high quality and high surface finish copper is a challenge
which has been met for the "Hadron-Electron-Ring-Anlage" HERA through the
development of a special plating process allowing highly reliable repetitivity on an
industrial level.
1) V.A. Lamb, C.E. Johnson and D.R. Valentine; J. Electrochem. Soc. 117 (9), 29IC-400c (1970)

REQUIREMENTS FOR COLLIDERS' BEAM TUBES
Specifications for the Copper Coating
HERA RHle sse
Beam tube inner diameter 55mm 72mm 32.5 mm

Length of beam tube 9'582mm 10'900 mm 17'270mm
Layer thickness of eu 12 ± 3 J.UI1 27.5 ± 5 J.UI1 75J.UI1
Electrical resistivity, p =2·1O·6 Qcm < 1.8 . 10-6 Qcm :$; 1.6 . 10-6 Qcm
of electrodeposited copper at 300' K at 300' K at 300' K
< 2.5 . 10.8 Qcm
at 4.2' K
Residual Resistivity Ratio

RRR (300' K/4.2' K) >50 >50 >250
High surface finish
(typically Ra < 1 J.UI1) x x x
Layer free of organic

additives x x x
Plating quality:
- no delamination after
thermal shocking or
after bending x x x
- no pinholes, porosity,
inclusions, nodules,
blisters, discontinuities,
discolorations x x x
Explanations for the HERA Requirements on Copper Deposits for HERA of DESY in
Hamburg
The circulating protons induce an image current in the wall of the collider's beam
tubes. The electrical resistivity of the steel would cause power losses of approx. 2 Watts per
meter. In order to minimize the losses to 0.1 Watt per meter, the electrical conductivity of
the electroplated copper layer at liquid helium temperatures must be of 4· 109 (0 mtl.
The copper coating must possess high surface finish in order to shorten the way of the
high frequency image current at the surface which would otherwise lead to high additional
power losses.
The minimum thickness of the copper layer is given by the skin depth of the high
frequency current. The thickness of the copper layer is also limited towards the upper
range: in case of superconductivity failure of magnets, because of the magnetically induced
eddy currents in the copper layer and the resulting forces, deformation of the tube's
geometry would occur. Furthermore, irregularities of the copper layer thickness would alter
the quality of the magnetic field. In view of these considerations, the optimum copper layer
thickness for HERA was 12 J.1m ± 3 J.1m.
The above-described forces on the copper layer as well as the ones resulting from the
thermical expansion between liquid helium temperature and room temperature require very
high adhesion between the steel substrate and the copper layer. Adherence failure would
lead to disastrous consequences.
382
INFLUENCE OF THE PLATING PARAMETERS ON THE COATING QUALITY
Coating Thickness Distribution
The copper plating distribution depends on different parameters, namely: the radial
electrical field distribution, the longitudinal electrical field distribution and the
hydrodynamics.
The regularity of the radial copper thickness distribution requires a symmetric
electrical field distribution given by a concentric anode disposition within the beam tube
which serves as a cathode (Figure 1).
r:x:..=.;:ft\
~~eamtu~~
a) b)
Figure 1. The radial electrical field distribution and associated thickness of copper plating depend on the
anode disposition within the beam tube. a) Concentric anode disposition leading to a regular plating thickness
distribution. b) Excentric anode disposition leading to plating thickness variations
The anode must be in a perfect concentric alignment on the full length of the tube.
This can be best achieved through vertical positioning of the beam tube during the plating
process (Figure 2).
The worse case in the anode positioning deviation is the situation where a short circuit
occurs which would seriously damage the plating quality locally.
For these reasons, the vertical positioning is preferred to the horizontal one. A
concentric positioning can be achieved with the help of distance keepers even in the
horizontal position but the contact points between the distance keepers and the beam tubes
must be continuously changed by anode or tube rotation/translation, otherwise
discontinuities in the plating occur. This motion can lead to scratches formation in the
copper layer especially in the horizontal position because the full weight of the anode is
transmitted by the distance keepers.
The regularity of the plating thickness along the beam tube depends mainly upon the
voltage distribution within the wall of the beam tube during the deposition process. The
beam tubes consisting of stainless steel are resistive enough in order to cause significant
ohmic drop along the tubes if connected only to the ends, thus decreasing the plating
velocity towards the middle of the tube. In order to avoid this problem, the tubes must be
connected along the whole length as shown on Figure 3.
The hydrodynamics play a determinant role on the plating thickness distribution as
well; higher flow velocities enhance the plating speed. The flow velocity has to be kept
constant during the process and set at an optimum value in order to
allow sufficient agitation to permit replenishment of the solution at the interface
beam tube/solution. Excessive deflection of the copper concentration at the interface
would deteriorate the plating quality in terms of surface finish.
cover the naturally occuring convective electrolyte motion due to dedensification of
the electrolyte at the interface of the beam tube.
383
to wash away potential metallic or dust particles which would otherwise be co-
deposited in the copper coating and deteriorate it. For this reason, the electrolyte
motion was made from top to bottom.
Figure 2. Installations for copper plating of the beam tubes in the vertical position
Figure 3. Electrical connection system to the beam tubes
Purity and Surface Finish of the Copper Coating
Pure electrodeposited copper can only be obtained from additive-free electrolytes

continuously filtered in order to remove dust and metallic particles. The filters used had
pore sizes of 0.25 11m and the total electrolyte volume was filtered ca. 5 times an hour. The
purity of the coating for high electrical conductivity and for avoiding degassing at low
pressure, is a determinant factor.
Producing low roughness and high purity copper is a very challenging task because a
highly lustrous finish normally depends on organic additives.
For HERA low roughness could be obtained by using a pulse reversed current.
Reversing the current periodically re-dissolves the nascent asperities and keeps the
electrodeposit smooth.
384
Adhesion and Homogeneity of the Copper Coating
Very adherent, pinhole and pore-free copper electrodeposits were obtained through the
application of special know-how and an optimized process. The main steps involve:
electrolytic cleaning
nickel strike from an acid bath
gold flash
copper electrodeposition from a pyrophosphate bath
Stainless steel for the first 3 steps and copper for the last step were used as anode
material. Since an anode change was involved before the 4th step, it was necessary to apply
a gold flash to protect the surface against oxidation during the manipulation. Oxidation at
this stage would cause adhesion problems. The know-how consists also in the right choice
of process parameters as for instance current densities, pulse reversing ratios, flow
velocities of electrolytes and rinse water, duration of each step, etc.
Repetitivity of the Electroplating Process
The repetitivity of the electroplating process and the results over the 400 dipoles of
HERA has been guaranteed by the automatization of the plating process with the help of a
process logic controller shown on Figure 4 and corresponding devices: pumps, valves,
electrical contacts, anode insertion and withdrawal, lifting the beam tubes to the vertical
position and back, Figure 5, precise positioning of tubes, electrolyte level regulation etc.
a .
~.,~
Figure 4. Board of command for the process logic controller of the plating installation
Figure 5. Beam tubes during the lifting operation towards the vertical position
385
PROCESS AND QUALITY CONTROL
Process Control
All chemicals involved as well as the mechanical devices of the equipment were
regularly analysed resp. checked on functionality.
Even though 3 beam tubes were plated simultaneously, each one was fed by a separate
power supply in order to avoid current distribution deviations from one tube to the other.
The electrolyte flow rates as well as temperatures were continuously controlled.
Besides the microprocessor control of the process, at least one person was
continuously watching the installation during operation. All deviations were documented
and communicated to the customer.
Traceability
All the tubes were identification marked. The corresponding process parameters and
ev. deviations could be specifically addressed.
Transport and Storage
Special transports with only HERA-beam tubes were made. Special containers with 6
tubes each were used.
During transportation and storage, the inside of the tubes was kept clean by closing
caps providing also protection against mechanical damage (Figure 6). The tubes were
stored indoors.
Figure 6. Beam tubes on their transport trunk (closing caps are shown beside the tubes on the trunk)
386
Quality Control
The following quality control operations were done:

visual inspection of the tubes with endoscope before plating and documentation.
visual inspection of the tubes with endoscope after plating and documentation.
measurement of the copper layer thickness with the eddy current method at 28 KHz
from the outside of the tube (see Figure 7). The measurements were documented. The
customer was using the same type of equipment for the copper thickness
measurement.
The calibrations were regularly compared with those of the customer.
The plated tubes were thermally treated 1 hour at 200 C at 10- 2 mbar in order to
0
recrystallize the copper and also to check the adhesion of the coating. The tubes were
heated up by ohmic effect, passing an electrical current of approx. 250 A through the
evacuated tubes as shown on Figure 8.
Figure 7. Measurement of the copper layer thickness with the Eddy current probe (The inner coating is
measured from the outside)
THE RESIDUAL RESISTIVITY RATIO (RRR)
The RRR values of copper electrodeposits have been measured at CERN and reported
by Dr. A.G. Mathewson. Copper layers of 100 ~m in thickness were deposited on to 1 mm
thick 3162N stainless steel substrates from sulfate/sulfuric acid baths and from
pyrophosphate baths. The measured voltage drop across the copper has been corrected for
the resistance of the stainless steel.
The results are summarized hereunder.
387
Figure 8. Heat treatment of a copper plated beam tube. The tube is evacuated with a pump and heated by
ohmic effect.
Measured RRR-values at CERN
Sample as plated after thermal after thermal after thermal

treatment Ih, 170'C treatmentlh,340'C treatment 2h, 950' C
Cu/pyrophosphate 94.2 220.6 351.6 6.6

Cu/sulfate 169.8 235.0 349.9 2.1
The increase in conductivity after thennal treatment at 170· C and 340· C is due to a
recrystallization and growth of the copper grains thus reducing the electrical resistance
resulting from grain boundaries. After thennal treatment at 950· C, the structure turned
very columnar with very large grains oriented perpendicular to the path of the electrical
current. This structure leads to dramatical losses of conductivity. Therefore brazing on the
beam tubes after the plating process might cause serious problems.
CONCLUDING REMARKS
The feasibility and the benefits of a galvanotechnical process for copper plating the
inside wall of beam tubes have been shown since the first HERA e-p collisions on Saturday
Oct. 19, 91 at 18:54. The specific requirements for colliders' beam tubes could be met
through specific know-how and adaptation of the process which was made clearly different
from the usual industrial copper plating processes.
388
Measurements of the RRR values carried out at CERN show that both electrolyte
types i.e. copper pyrophosphate as well as copper sulfate are suitable for meeting the
electrical requirements. Even though sulfate-type electrolytes show better conductivity
values in as plated conditions, after thermal treatment both types of deposit show close
RRR values.
ACKNOWLEDGEMENTS
W. Fliihmann Inc. is very grateful to the following persons and organisations for their
contribution to our work:
D. Hubert, D. Trines and R. Hensler of DESY in Hamburg for scientific and technical
support throughout the realization of the work.
Mr. Worscheck of the FH (Technical College) in Hamburg adapted the Eddy current
method for the purpose of the copper coating measurement in their HERA beam tubes.
The Management of the PSI in Wiirenlingen for putting their facilities at our disposal.
A special thank-you is addressed to G. von Holzen of PSI for support in the
construction of equipment.
RRR measurements have been performed at CERN in Geneva through A.G.
Mathewson and J.M. Dalin.
The practical work at W. Fliihmann Inc. has been performed by A. Bloch.
389
PRESSURE RELIEF RADIAL PASSAGES FOR SSe"
Margareta L. Rehak

Magnet Division
Upton, New York 11973
INTRODUCTION
In the 40 mm sse cross--section design radial helium venting passages located

every 6" were present. The purpose of these passages was to provide escape routes for
helium from the coil cooling passage to the bypasses during a quench. The passages are
introduced in the magnet by placing special yoke laminations with channels going from the
bypasses to the inner radius.
The question as to whether the venting passages are still needed in the 50 mm
design is addressed here. One might expect that they would be more needed here since
the stored magnetic field energy increases considerably faster than the coil cooling passage
cross-section, unless the width of the latter is increased.
Since the dimensions of the beam tube are not fixed at this time two cases are
considered: a coil cooling passage gap of 1.3 mm and one of 5 mm.
An existing computer code developed at Brookhaven National Laboratory by R.P.
Shutt and programmed by K. Jellet is applied to the present problem. The theoretical
contents of the code are described in [SSC--N--266 No 59]. Pressures, helium, coil, and
yoke temperatures, and mass flows are computed during a quench for one magnet with no
venting occurring in the interconnect regions. Only one magnet is considered here as
opposed to a string of magnets.
RESULTS
The relief passages are one lamination deep (d = 0.158 cm or 1116 in.). However,
due to limitations of the numerical method, the computer program will not run for a long
enough time with this value. The conservative value d = 0.06 cm will be used instead.
The case where there are no passages is simulated with d = 0.001 cm.
*work supported by the U.S. Department of Energy

1.3mm ga.p ma.ximum pre.sure around be&m t.ube with venting
.,.,
r-
~ tol=0.400
.,.,
0 sr- U340
rhr=O.0640
~
.,.,
;~Vo a4= 26.000
f
-
rh4=0.7300
:::;; ::! o
~ 0 Be= 0.0000
~ g 0
delta= 0.060
r.r; 0
! ~/o 11= 40.0
""c:::c... S
0
!
g
cj'o
0
dtq= 0.15
o Iq=1520.0
-
r.r; o
r.r; !\; o
o kp= O.l42E-03
IX) se= 0.050
;:;t: o
o
o .I..~ 0.20
...:l 0 c
o 1r=1520.0
0 ~ o
U
...:l
.,.,
I
o
0
U '"<.6 o~
c o
0 o
0 000 0
.,.; eoO 00
-0900 0 8
Ifl(Jll,oc:" 0
I.,_ _. . . . ~ e
~
(")
0.00 0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36 0.40
TIME SEC
Figure 1.
1.3mm gap ma.ximum pressure around beam tube without venting
N
(")
LEGEND lol=0.4
IX)
=P oP
0
ar:: 1.640
rhr=O.0640
'" a4= 26.600

rh4=0.7300
:::;; .,. ae= 0.0006
~ '" delta= 0.001
r.r; Ii= 40.0
W 0
dlq= 0.15
::x:: 0
0
0.. Iq=1520.0
'"
0
r.r; 0
kp= O.OOOE+OO
r.r; 0
c:
...:l
~
0 sc:= 0.050
sk: 0.20
1r=1520.0
0
U 0
0
...:l
0u S!!
0
00
..--/'./
IX)
.,.
0.00 0.02 0.04 0.06 0.06 0.10 0.12 0.14 0.16 0.16
TIME SEC
Figuloe 2.
392
mm gap maximum pressure around beam tube with venting
s:: P lol=0.400
LEGEND
0 =p or= 6.300
rhr=0.2500
'" 04= 26.800

rh4=0.7300
::::s oc= 0.0008
~ della= 0.00 I
en CO Ii= 40.0
...en
~ dtq= 0.15
0::
Iq=1520.0
§
-
kp= O.OOOE+OO
en r-
~ I se: 0.050
I
sk= 0.20
I
is
U
1r=1520.0
CO
....:l
0
U
II>
~ , ,. , ,.
0.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
TIME SEC
Figure 3.
5 mm gap maximum pressure around beam tube without venting
LEGEND lol=0.400
o =P or= 6.300
rhr=0.2500
a4= 26.800
co rh4=0.7300
::::s oc= 0.0008
~ della= O.OGO
en Ii= 40.0
g: r-
~ dlq= 0.15
Iq=1520.0
en
en
kp= 0.142E-03
c::
sc= 0.050
sk= 0.20
....:l CO Ir=1520.0
0
U
....:l
0
U
II>
~4-~~~,~~--~~~-r~~-r~r--r~~-r~r--r~r;-r~~~
0.00 0.04 O.OB 0.12 0.16 0.20 0.24 0.28 0.32 0.36
TIME SEC
Figure 4.
393
Of all the results produced by the calculations, that which is of most interest is the
helium pressure around the beam tube, where the pressure is expected to be the highest.
Figures 1 to 4 illustrate helium pressure (at the location where it is maximum) around the
beam tube as a function of time, and the table summarizes the results. Note that the final
times differ since they correspond to times where the program stops.
Table 1. Pressures around the beam tube.
Coil cooling passage gap (mm) 1.3 1.3 5 5

Relief passages yes no yes no
Maximum pressure around beam tube (atm) 12.6 29.4" 10.0" 8.5"
(* pressure still rising at end of time for which pressure is calculated)
Figure 1 shows that the maximum pressure around the beam tube is 12.6 atm when
the relief passages are present and when the coil cooling passage gap is 1.3 mm. In the
absence of these passages Figure 2 shows that the pressure rises to 29.4 atm and above
which is above the 20 atm design pressure. If the coil cooling passage gap were 5 mm
then Figures 3 and 4 show that the presence of relief passages has little effect on the
pressure around the beam tube. In Figures 2,3 and 4 pressures are still rising at the end
of the shown time periods. However, in Fig. 2 the design pressure has already been
exceeded, and at the end of the time periods in Figs. 3 and 4 vent valves in
interconnections would have opened, relieving the pressure fast. Concerning Fig. 2, the
maximum pressure in the interconnection is only 4.5 atm.
BUCKLING PRESSURE
These pressures should be compared with the 20 atm design pressure and the
critical pressure for buckling. A formula from "Roark's Formulas for stress and strain,
W.C. Young, McGraw Hill" gives the critical pressure of an externally pressurized tube:
For a beam tube of radius R = 0.8 in. (2.0 cm), thickness t = 0.06 in. (0.15 cm) and
yield strength C1 y = 30000 psi then p is equal to 1314 psi (= 91. 8 atm (9. MPa». If there
are no relief passages for the 1.3 mm gap the safety factor for the beam tube is 3. Thus
the beam tube would still be safe even while exceeding the design value of 20 atm.
On the other hand if the shell thickness is halved then p = 292 psi = 20 atm (2
Mpa) and the beam tube will fail.
CONCLUSIONS
If the coil cooling passage gap is 1.3 mm as it was in the 40 mm design then relief
passages are needed in order to remain below the design pressure of 20 atm for the beam
tube.
For a 5 mm coil cooling passage gap relief passages are not needed.
394
OVERVIEW OF A ROBOTIC SYSTEM
FOR AZIMUTHAL DIMENSIONS OF SSC DIPOLE COILS
D.L. Assell,l J.M. Cahill,l J.A. Carson,l D.F. Connolly,l S.K. Pawlak,l
R.K. Rihel,l W.R. Robatzek,l W.F. Robotham,l E. Schmitz,2 R. Sims,2
T.M. Skweres,l and R. Ullmark 2
lFermi National Accelerator Laboratory, P.O. Box 500, Batavia, TIL 60510
2Superconducting Super Collider Laboratory, Fermilab/SSC Magnet Project
P.O. Box 500, Batavia, TIL 60510
ABSTRACT
This system measures the azimuthal dimensions of SSC dipole long coils
automatically, taking over 500 measurements in less than four hours. These measurements
are then analyzed and displayed via software reports which reveal coil statistics, point by
point dimensional graphics, modulus of elasticity measurements, comparisons with
previous coils and pre-collaring shimming information. The rapid turnaround of this
system yields the ability to spot process variables and fine tune the fabrication techniques.
This process will aid in producing coils to the required precision.
INTRODUCTION
The SSC will require the production of approximately 8,000 dipole magnets. Each
dipole will require two inner and two outer coils. This will require the production of
approximately 32,000 coils, each 15 meters in length. To assess the quality of the coils
during the manufacturing process a measuring system is needed. Fermilab has developed a
coil measuring machine as part of its prototype manufacturing facilities (see Figure 1).
Figure 1. Coil Measuring Machine

MEASURING PROCESS
Coil measurements are made by confining the coil in a fixture which replicates the
collared geometry of the coil (see figure 2). The coil is then compressed by applying a
known pressure at the parting plane. Coil displacement is noted and compared to
measurements made of a precision steel block fabricated to the nominal collared coil
dimensions.
'.' Uppe r O\Jlcr COil

flllndr n n i II
/" Coll"
UpPt' t Oute r (0 11
/
'y fl u .r/r an I I
\ .''- .'
QUllcinwl III Lowe r Oul f' r ("011
\, I,.J / (}lW (/r a ll/ IV

.1'. _
1. 0'"' et Oute r COi l
, L tl ~I'r I fln r r rO ll
Figure 2. Collared Coil Assembly
The measuring process is done at 7.62xlO- 2m increments along the entire length of
the coil. The pressure applied to the coil during the measuring process, 82.74mPa, is
consistent with the prestress the coils will see when collared.
In addition, at every 20th increment or 1.52 meters, coil measurements are made at
41.37mPa, 55. 16mPa and 68.95mPa. These additional measurements are used to calculate
the modulus of elasticity for the coil. As the measurement process is being conducted the
coil resistance is also monitored for tum to tum shorts which are most likely to occur when
the coil is under pressure. This resistance monitoring affords an opportunity to find shorts
and repair before the coils are assembled.
The machine developed at FNAL to accomplish these measurements consists of the
following systems.
1) Linear motion system
2) Mandrel support system
3) Measuring head system
4) Control and data acquisition system
5) Safety system
These systems are controlled and operated by a computer, making the machine
completely automated. This automation allows the taking of over 500 data points in less
than four hours.
OPERATION
A coil is placed on the measuring machine mandrel. After assuring the coil is
properly positioned the operator starts the machine. The operation is prompted to enter
pertinent coil information. With all prompts satisfied, the measurement sequence begins.
All machine functions are now directed by the computer. The operator is signaled when the
process is complete, if a machine malfunction occurs or if a coil short is sensed.
Before loading the coil the measuring head is in a parked position at the extreme end
of the support mandrel (see Figure 3). The park position is where the calibration master
standards are located. The measurement process starts with the calibration of the
measuring head. The measuring head consists of a precision cavity which matches exactly
the nominal coil outer radius and coil pole. Pressure bars define the parting plan. The
bars, one for each side of the coil, bear against the coil parting plane. Force is applied to
the bars via hydraulic jacks. Each bar is made of 3 pieces. The hydraulic force is applied
uniformly to each of the 3 pieces. The purpose here is to eliminate end effects. Vertical
displacement of the center bar piece is measured by a linear motion probe (ee Figures 4 and
5).
396
- 1
Figure 3. Coil Measuring System
..... :.:.-:.::.:..-:
__b_ ,
.... Hf' _
..... "-
t:,:,., ,,-,
... . ..
.. -
_-... .. :
......
Figure 4. 50mm Coil Measuring Head
~'
*--'s ... "' .....
".....
~ .... I ••
l.U",
Figure 5. Three Piece Pressure Bar Assembly
Calibration is accomplished by lowering the head onto the calibration standards or

blocks. Hydraulic force is applied to achieve 41.37mPa, 55.15mPa, 68.95mPa, and
82.74mPa pressure on the blocks. At 41.37mPa the measurement probe is zeroed. All
subsequent measurements of the calibration standards and coil are relative to this zero. The
calibration will be checked again at the end of the coil measurement sequence to ascertain
that no substantial drift of the calibration has occurred.
Once the calibration is complete the measuring head is raised and moved to the first
coil measurement position. The coil is measured when compressed to 82.74mPa. The
pressure is held for a preset interval to allow for coil stress relaxation to occur. The
process is repeated until the entire coil is measured. As the measuring head moves down
the mandrel, the mandrel supports are lowered to allow the head to pass (see Figure 6).
The longest span of unsupported mandrel is 1.22 meters. The stiffness of the mandrel
provides negligible deflection.
,
.... ........
. ·(,1'_'
Figure 6. Coil and Mandrel
397
DEFINITIONS AND CALCULATIONS
Coil Width Steel Master Mean Arc LeniWt
50mm Inner 1.267xlO-2m 50mm Inner 4.050xlO-2 m
50mm Outer 1.217xlO-2m 50mm Outer 3.68OxlO-2 m
Effective Hydraulic Cylinder Area Effective Coil Area
2.477xlO- 3 m 2 1.267xlO- 2m or 1.217xlO·2m
x7.62xlO- 2m
Calculations
55. 16mPa coil probe reading minus 55. 16mPa master probe reading
82.74mPa coil probe reading minus 82.74mPa master probe reading
Pump Pressures on the Block
Pump mPa = <Block Pressure)(Coil Area)
(Hydraulic Cylinder Area)
Modulus of Elasticity between 55. 16mPa and 82.74mPa
E= Stress = PIA
Strrun AUL
PIA = 82.74mPa - 55.16mPa = 27.58mPa
L = Master length plus or minus the coil deviation with respect to the master.
& = Coil size with respect to master at 55. 16mPa minus coil size with respect to master at
82.74mPa.
The collected data is post processed from a floppy disk using EXCEL©. Figure 7
shows a sample data sheet produced by the program for one coil. The program prints
average coil size, standard deviation, range of coil size, smallest and largest reading and
modulus of elasticity for each quadrant. Figure 8 shows a sample graph of quadrants I/III
and IIIIV representing measurements taken on both sides of a coil plotted for all 190
positions. Various other graphs are plotted to characterize coil features and help to analyze
causes of variations. Figure 9 shows the whole coil average for 3 different coils. It is
apparent that the variations are systematic, probably caused by tooling variations.
CONTROL SYSTEM
The control system program, run by a 386 computer with an 80MB hard drive, is
written in ASSYST. The subprograms necessary to safely accomplish the accurate
measurement of an SSC magnet coil can be broken down into the following subsystems
(see Figure 10).
1) Linear motion system
2) Mandrel support system
3) Measuring head system
4) Sizing and data system
5) Safety system
LINEAR MOTION SYSTEM

The linear motion necessary to accurately position the head at 190 predetermined
points is accomplished with an encoder, decoder, stepper motor and indexer drive. These
are linked to the host computer via an RS-232 port. The encoder, which is always used,
knows its absolute position, even if power is lost. The indexer and motor drive are
powered by an isolation transformer to further minimize errors due to noise. This has
given the system high accuracy and repeatability. When a move command is issued by the
computer, the indexer will step the motor drive until the encoder tells the computer that the
desired position has been achieved.
398
15M-50-202l
Chart Version 2.0 10/19/91

LONG SOMM OUter Coil # 1SM-SO-2021
Master Used - Outer Master #1 Shim Size Used - .0 1a

Press Fixture Serial Number Used -
Machine Serial Number Used - AUTO-Ol Mean Arc of Master - 1.4513·
QUADRANT 11111 POSITION 1 192 QUADRANT IIIIV POSITION 1 192 WHOLE COIL
8 kpsi 12 kpsi MOE 8 kpsi 12 kpsi MOE 8 kpsi 12 kpsi MOE
-0.0018 -0.0044 2.41E+06 -0.0029 -0.0046 2.56E+06 -0.0023 -0.0045 2.48E+06
0.0009 0.0011 1.17E+05 0.0012 0.0011 1.85E+05 0.0011 0.0011 1.68E+05
0.0030 0.0067 3.22E+05 0.0042 0.00S4 6.16E+05 0.0046 0.0067 6.16E+05
-0.0006 -0.0016 2.59E+06 -0.0011 -0.0028 2.B4E+06 -0.0006 -0.0016 2.84E+06
-0.0036 -0.0083 2.27E+06 -0.00S2 -0.0082 2.22E+06 -0.0052 -0.0083 2.22E+06
Coil Master
0.9944 0.9990
0.9944 0.9990
0.9939 0.9990
0.9935 0.9990 -0.0054 -0.0009
0.9935 0.9990 -0.0054 -0.0009
0.9937 0.9990 -0.0052 -0.0007
0.9943 0.9990 -0.0047 -0.0002
0.9946 0.9990 -0.0043 0.0002
0.9946 0.9990 -0.0043 0.0002
10 0.9945 0.9990 -0.0044 0.0001
11 0.9947 0.9990 -0.0043 0.0002
12 0.9954 0.9990 -0.0035 0.0010
,.
13
IS
0.9954
0.9949
0.9951
0.9990
0.9990
-0.0036
-0.0040
0.0009
O.OOOS
0.9990 -0.0038 0.0007
,.
16 0.9954 0.9990 -0.0036 0.0009
17 0.9947 0.9990 -0.0043 0.0002
16 0.9937 0.9990 -0.0053 -0.0007
0.9939 0.9990 -0.0051 -0.0006
20 l.f' 0.9994- 1.0000 0.9976 0.9996 0.9961 0.9992 0.9947 0.9990 -0.0020 -0.0042 0.0003 2.S9E+06
20 [N 0.9969 0.9998 0.9957 0.9994 0.9950 0.9991 0.9947 0.9990
Figure 7. Sample Data Sheet
- - - - - QUADRANT 1/111 IN RESPECT TO THE MASTER AT --{}--- QUADRANT IIIIV IN RESPECT TO THE MASTER AT
12,000 COIL PSI IN INCHES 12,000 COIL PSI IN INCHES
0.0000
-0.0005
-0.0010
-0.0015
-0.0020
-0.0025
-0.0030
-0.0035
-0.0040
-0.0045
-0.0050
-0.0055
-0.0060
-0.0065
-0.0070
-0.0075
-O.OOBO
-0.00B5
-0.0090
-0.0095
-0.0100
M ~ ~ ~ M ~ ~ ~ M ~ ~ m ~ ~ ~
N N N N N M M M MM. ~ ~ ~
POSITIONS
Figure 8. Sample Graph of Quadrants I/III and IIIlV
399
0.0035 ~
VERTICAL ERROR 2001, 2002, 2003 (QUADRANT I1III + QUADRANT II/IV)/2
0.003
0.0025
0.002
0.0015
: ~,"JM~
0.001
0.0005
-0.0005 \, ~ + ,,+ +
+
-0.001
-0.0015
~~ [[]
-0.002
-0.0025 +-~~-+-~~-J!~~---t~~~+-~'='--+-~~--+-~~---t~~~.,........~- ..1--------1

0 20 40 60 80 100 120 140 160 180 200
POSITIONS
1------- 15M-50-2001 ---0-- 15M-50-2002 - + - 15M-50-20031
Figure 9. Whole Coil Average for 3 Different Coils
MANDREL SUPPORT SYSTEM
There are fourteen legs to support the mandrel and coil during the measuring
process_ The legs are attached to air driven cylinders that allow them to raise and lower
independently, clearing a path for the measuring head. The air cylinders are controlled by
solenoids activated by the computer through a 24-channel digital I/O card and a 24-channel
relay card. The host computer is informed of the measuring head location by the encoder,
then raises and lowers the mandrel support legs at predetermined positions. In the manual
mode the mandrel legs must be actuated locally.
MEASURING HEAD SYSTEM

The measuring head system can be broken down into two complementing
operations: raising and lowering the measuring head and actuating the pressure bars.
The measuring head is raised and lowered by air driven cylinders. The host
computer activates a solid state relay (SSR) with a digital I/O channel that energizes a three
way solenoid valve to deliver house air to the cylinders. The computer also detects the
head position with microswitches. Linear motion is inhibited unless the head-up switch is
engaged and sizing bar cylinders are also inhibited unless the head-down switch is
engaged.
The coil is measured at four stress values: 41.37mPa, 55. 16mPa, 68.95mPa and
82.74mPa. The pressure bars are raised by eight hydraulic cylinders. The hydraulic fluid
is supplied by a Haskel air over oil pump with a 75:1 amplification ratio. The air pressure
to the pump is controlled by a Fairchild transducer, which is programmed by the computer
through a digital-to-analog channel. The pump is equipped with a fast fill feature. This
consists of a buffer tank that contains hydraulic fluid and is topped off with an air head.
In fast fill mode an SSR activates a three way solenoid to apply house air pressure
to the top of the buffer tank. This supplies the coil sizing cylinders with a larger volume of
fluid, by forcing oil through the easy direction of the two one-way valves in the pump.
When the coil compression cylinders meet resistance by each pressing lightly on the coil
body, the pump takes over and raises the pressure to the programmed value. The coil
compression cylinders are retracted using an accumulator. The front end of the
400
Digital I/O _ I I -
A to 0 a I I for Calibration I ...... ".
from A to D n .... Un I f " " . 1
From Air "Fast FiU

Supply Valve" raises
120+ PSI 1'=(1 i I I coi I eyl inders
I I
Cab~~ ~~~~I~ r-
I I
I I
To 14
solenoid I I
14 valves
Digital Cbannel - which
I I
I/O Relay _ activate I I
Card
_ i~: maodre II I
cylInders I I
I I
I I _
Hydraulic
Probe I I -=- Pressure
Electronics Sensor
I I
0-10K PSI
Coi 1
Frame Up Detector compression
t"Head Up-Down" I-: .--~ Cy1 inders
~ (8)
Down Detector =
Raise
A to D I I Frame
and Vol tages Cyl inders L....l_t---+-! Return Side
(2) 0-10K PSI
"Return Coil
Cyl inders"E AcctIDulator
Absolute
RS232 Encoder
I_ I I HGe.rbox
Molor
Gearbox
.. I I
hos~ei llCbml t I I Rack
~ ~ Gear
o
Figure 10. Automatic Measuring System Controls
accumulator sees house air when activated by a solenoid. The back side of the accumulator
is oil filled and attached to the return side of the compression cylinders. All pressures are
regulated using feedback supplied by Sensotech pressure transducers.
SIZING AND DATA SYSTEM

All data relating to azimuthal size is taken using Sony linear probes. These are read
back by the computer at the designated stress values with an RS-232 link. Thermocouples
are used to measure the temperature of the measuring head and are read with AID channels.
The coil resistance is monitored during the sizing operation by the Valhalla Micro-ohm
meter with GPIB interface. The coil temperature is measured by an aluminum half cylinder
under the coil and if fed to the Micro-ohm meter for the correction of the reading.
SAFETY SYSTEM
The safety system includes the raise and lower head detector microswitches already
mentioned and a pressure bar up detector which also uses a microswitch. There are
bumpers installed in front and back of the measuring head. If an obstruction is detected in
the path of the head, a microswitch is activated by the bumpers and all operations are
halted. There is also an emergency stop pushbutton located on the front panel of the
control box on the measuring head. This will also halt all operations when depressed.
CONCLUSION
The automatic coil measuring system provides a constant monitoring of the
azimuthal coil size for both the inner and outer coils. This information is a valuable asset in
the production of uniform coils. It can alert one to the possibility of tooling problems
during cabling, winding or curing as well as assembly and testing. This system provides
complete data on every coil that goes into the making of the uniform sse dipoles.
402
OVERVIEW AND STATISTICAL FAILURE ANALYSES OF
THE ELECTRICAL INSULATION SYSTEM FOR THE SSC
LONG DIPOLE MAGNETS FROM AN INDUSTRIALIZATION
POINT OF VIEW
J. F. Roach

ABSTRACT
The electrical insulation system of the SSC long dipole magnets is reviewed and potential
dielectric failure modes discussed. Electrical insulation fabrication and assembly issues with respect
to rate production manufacturability are addressed. The automation required for rate assembly of
electrical insulation components will require critical online visual and dielectric screening tests to
insure production quality. Storage and assembly areas must be designed to prevent foreign particles
from becoming entrapped in the insulation during critical coil winding, molding, and collaring
operations. All hand assembly procedures involving dielectrics must be performed with rigorous
attention to their impact on insulation integrity. Individual dipole magnets must have a sufficiently
low probability of electrical insulation failure under all normal and fault mode voltage conditions
such that the series of magnets in the SSC rings have acceptable Mean Time Between Failure
(MTBF) with respect to dielectric mode failure events. Statistical models appropriate for large
electrical system breakdown failure analysis are applied to the SSC magnet rings. The MfBF of the
SSC system is related to failure data base for individual dipole magnet samples.
INTRODUCTION
Westinghouse has been following the General Dynamics design of the Collider Dipole Magnet
(CDM) main ring collider magnets since August of 1991 under the CDM Leader/Follower
agreement. 1 Through direct participation on the General Dynamics Concurrent Engineering teams,
Westinghouse has acquired a critical understanding of the approach and design clements of the
electrical insulation being developed by General Dynamics. This paper provides a brief assessment
of the present state of the CDM insulation design from the Follower's perspective.
INSULATION REQUIREMENTS
The primary electrical insulation requirements for the 15 metre CDM are found in the SSC
Magnet System Specification, Jan. 10, 1992.2 The CDM electrical insulation design must support
all voltages applied or induced during all operating, enersy relief, upset, and test conditions with
appropriate factors of safety applied to the ultimate dielectiic strength of insulation components and

system assemblies. In addition, the insulation materials must be capable of withstanding a radiation
dose of 10 MGray at the inner coil minimum radius and TBD x 10 12 neutrons/cm 2 dosage over a
25 year life. The insulation ratings between major insulation components and appropriate
reference(s) are presented in Table 1. The insulation ratings are based upon the minimum applied
voltage in air. The component insulation as well as the tum to tum insulation must be tested to the
rated voltage in Table 1. The maximum allowable leakage currents must be below 25 /JA in each
case. The component insulation fails under test if either dielectric breakdown or excessive leakage
current are occur.
The voltage ratings in Table 1 are essentially predicated upon the maximum bus voltage
expected in each arc sector of each ring which is driven by a single power supply and protection
circuit. The maximum bus voltage occurs under the energy relief cycle. For the required 20 s
energy relief time constant the ramp down rate may be as high as 300 Ns. Estimating the arc sector
total inductance as 36 H, the maximum normal voltage to ground seen by the bus in a sector during
energy dump is 1350 V. Employing a factor of safety based upon twice the maximum normal
voltage to ground plus a 1000 V, a bus insulation rating of 3700 V is estimated compared to the
5000 V bus rating requirement from Table 1.
Table 1. CDM electrical insulation requirements based on rated voltage in air.
Component Reference Minimum Rated Voltage in Air

KV
Main Coils All Other Components 5

and Ground Walls
Quench Heater at Ground 3.5
hmerCoil Outer Coil 1.5
Lower Coils Upper Coils at Midplane 3
Bus All Other Components 5
Quench Heater Ground 2
INSULA TION DESIGN
The major features of the preliminary insulation design are summarized in this section with
emphasis on departures from the SSC baseline 50 mm CDM specifications.
Cable Insulation
The all polyimide cable insulation system (Kapton CI/RCI) under development at Brookhaven
National Laboratory (BNL) is being baselined by General Dynamics as an alternate to the
polyimide/epoxy impregnated fiberglass system for prototype magnets. The essential features of the
polyimide/epoxy and the all polyimide (Kapton CI/RCI) insulation systems are shown in Figure 1.
The primary advantages of the Kapton Cl/RCI system are its excellent punch-through resistance and
improved radiation resistance compared to the polyimide/epoxy system. The high temperature
(230 C) curing cycle required for the Kapton CI/RCI system, however, must be properly controlled
0
to prevent cable deformation and interstrand resistance problems. An optimized curing cycle has
been developed by BNL where the coil curing pressure is reduced to zero during the final 230 C 0
"moldin~" of the coil and is increased for fmal coil sizing when the coil temperature has reduced to
135 C.3
0
404
Beam Tube and Support Insulation
The beam tube is wrapped with a single layer of Kapton RCI. 2 mil tape. 25.4 mm wide. with
a 48% overlap to insure a uniform taping surface for the beam tube support structure. Curing the
beam tube insulation at the RCI "molding" temperature of 230' C without damaging the inner beam
tube coating, and securing the tape at the ends of the beam tube are under investigation.
The General Dynamics beam tube support system is a unique concept which employs metallic
springs fastened around the beam tube insulation and located about every 50 cm along the axial
length.4 This beam tube spring support system concept is shown in Figure 2. (the garter springs are
304 CRES Wire which are enameled with a heavy-build of Pyre-ML polyimide before winding into
springs). This beam tube support concept provides good mechanical support. Principle concerns
are with the dielectric integrity of the system over 25 year life, since it is possible for coils to short
EPOXY PRE ·IMPREGNATED WOVEN

' .. FIBERGLASS
• 9 . ~mm wide' 0. 114mm Ihock
• SIngle hehcaJ wrap wI bun gap
FILM LAYERS
Wore • 9.5 3mm WIde x 0.0254mm thIc k
• SIngle 50"'0 overlap helICal wrap
KAPTON RCI OUTER LAYERS

~• Reonlo<ced wI ceramic filler
"$$§\$%\§) hU,,S$\$$$$\$ S5 b%'5%\\\\\§,,\\\\$\$\\§ • Thermoplastic poIylmtde adhesive on
both top and bonom 01 film
• Smgle 50% overlap
CI INNER LAYERS
• Thermoplasloc polylmide adheSIve on
lOp 01 111m only
Wore Hehum Votd • 0.0381 mm lhick film. tYPICal 10< Cl and
RCII"ms
• Smgle 50% overlap
Figure 1. Comparison of the poly imide/epoxy and Kaplon CI/RCI cable insulation systems .
.2,5 mm biro k'apton

ProfKb Coli, r"om
I . am TutM Support
1 11 , W ,r
- ~ ~G
, ,
H
Figure 2. General Dynamics beam tube support system concept.
405
either to the beam tube or the collar poles through the springs if insulation is lost simultaneous at
both coil and ground wall points. To further enhance the dielectric integrity of the system it has
been proposed to place additional polyimide sheet insulation between the springs and coil ID, as
shown in Figure 2.
Ground Wall Insulation
The ground waIl insulation is the same as the Fermi National Accelerator Laboratory (FNAL)
50 mm coil insulation system shown in Figure 3 with two modifications.
An additional 5 mil Kapton between the inner and outer coils and the addition of 5 mil Kapton
at the coil ID/garter spring interface are being considered. Brass collaring shoe (15 mil) will be
used. Pole shims will be used in model magnets and early prototype magnets to be built, but it is
anticipated that in production shims will be eliminated as experience with curing coils to aszimuthal
size is gained.
.005
Heater Strip
.005 .006
Coll aring Shoe
0.15 Brass
.005
.005
.005
.005
Figure 3. FNAL 50 mm coil and ground wall insulation design.
Quench Heater Insulation
The FNAL quench heater concept will be used. However, ongoing design and analysis tasks
may result in a new design. The Kapton insulation is applied to both sides of the heater strip with
adhesive. Kapton tape is used between the adhesive top/botton layers at the ends of the strip to
insure a uniform insulation build and prevent high mechanical point loadings at the metallic edges
which could result in insulation damage.
Bus Insulation
The bus design by MIT will be used for early prototype magnets. 5 Each bus conductor is
insulated with half lapped Kapton tape and located in a C-shaped channel cut from polyester
pultrusions. The components of the entire straight section of the bus assembly are epoxy bonded.
DIELECTRIC FAILURE MODES
The major dielectric failure mode of the CDM electrical insulation system is expected to be
short circuits between coil turns, between upper and lower coils at the midplane, and between coil
and quench heater due to insulation damage at localized high mechanical pressure points. The role
406
of foreign particles entraped in insulation during assembly may lead to unexpected insulation
damage and failure; metallic particles may also result in very high local electrical stress between
components and result in high leakage currents.
The garter spring beam tube support is vunerable to dielectric failure as the result of wear of
coil and beam tube insulation due to spring movement. The reliability of the spring wire insulation
after spring forming needs to be established. It may be possible for a spring to break; the simple
latch hooks to hold the spring around the beam tube may also fail. Any break of a spring could
result in dielectric failure. If springs are located say every half metre, then a single collider ring
would require about 120,000 spring units for beam tube support.
The present MIT bus design also appears susceptible to dielectric problems. Although a direct
breakdown failure of the bus conductors to each other or ground is unlikely, the local stress on the
lower bus conductor edge when tested in air is relatively high and could result in high leakage
currents in excess of the 25 /.IA requirement. Since the bus conductor channel wall is about 3.2 mm
thick and the test voltage required for the bus 5 KV, then the average stress from bus to grounded
yoke through the channel support is 15.6 KV/cm. The dielectric mismatch between air and the
channel pultrusion will lead to an air void stress at least twice the average stress, that is >30 KV/cm
or the breakdown strength of air. Thus, it is expected that some partial discharge will occur during
the bus verification test. For rate prodUction, therefore, the bus design is marginal from a leakage
current point of view.
RATE PRODUCTION MANUFACTURABILITY
The CDM rate production requirement is 10 magnets per day. The coil and ground wall
insulation assembly, shown in Figure 3, is probably the single most important component of the
insulation system which impacts achieving factory rate production. At the present time assembly of
the coil and ground wall insulation at both FNAL and BNL is a labor intensive operation which
requires essentially hand assembly. The interleaved coil insulation built from creased Kapton
sheets, each the length of the magnet, must be held in place as the next sheet is applied. FNAL
practice is to use ties which are later removed after assembly is complete; BNL uses small adhesive
tape to keep the insulation sheets in place during assembly as well as hot hand irons to bond
adjacent sheets together. Care must be taken to keep the coil insulation wraps smooth and free from
wrinkles. Before rate production can be realized, the assembly of the coil insulation will have to be
automated as much as possible. Adhesives applied to local areas of the creased Kapton and unique
tooling designs for handling and applying this complex insulation package are needed and are being
developed.
STATISTICAL FAILURE ANALYSIS
The insulation reliability of the CDM is determined from the requirement that only 73 magnets
of the 7964 magnets comprising the two collider rings are allowed to fail for all possible modes of
failure over the 25 year life of the magnets. If it is assumed that dielectric failures/problems will
constitute 1/3 of all possible failure modes, then the allowed insulation failure rate for insulation is
lAxlO- 8 failures/hour over 25 year life. This rate of failure is equivalent to a reliability factor for
all insulation failures of 0.997 or a failure probability of 0.003. Allocation of this overall failure
probability to individual insulation failure modes for all components of the insulation system then
allows experimental failure data on insulation components and subcomponents to be related to
magnet life predictions by appropriate statistical models.
Provided the insulation capability in subcomponent tests can be expressed as some distribution
of failure as a function of voltage stress under mechanical loading and cyclic conditions
representative of the CDM, then it is possible in principle to predict the failure probability of the
collider system. This is strictly possible for random failure sites in the collider insulation such as
the tendency for local shorts at random high pressure points; failure modes which result from non-
random assembly or processing errors would be excluded from this approach. Using an approach
similar to the statistical evaluation of dielectric breakdown in oil insulated large power apparatus,
407
the probability of random insulation failure of the entire collider can be related to the data base of
subcomponent failure probability statistics assuming that the entire system is a collection of parallel
failure sites. 6 In this way the system failure probability may be expressed as follows:
P(S,n) =1 - (1 - p(S»n (1)
where P(S,n) is the system failure probability, p(S) is the failure probability of the subcomponent
data base, S is the stress parameter of failure, and n the number of parallel failure sites in the
system. For electrical insulation failure, normal, Weibull, or extreme value statistics are generally
found when testing small sample and subcomponent insulation. For the pupose of illustrating the
approach, however, Equation 1 is applied only for a normal distribution and the result is shown in
Figure 4. In Figure 4, the cumulative failure probability P(S,n) is plotted as a function of the
normalized stress (S - M)/STD, where S is the applied stress, M the mean stress of the unit test
data probability p(S), and STD the standard deviation of the unit distribution.
As an example of the application of Figure 4 to the collider insulation failure assessment,
assume that sufficient tests are available on the insulation capability distribution from 0.1 m axial
length, full azimuthal coil insulation stack samples (such as reported by Reed of SSCL3). Then
each CDM magnet is made up of approximately 600 such subcomponents, where insulation
capability between outer and inner coils is being ignored. The collider ring consists of about 8000
CDM's or about 4.8x106 unit insulation sites. Assuming a value of P(S,n) = 0.001 for coil short
failure allocation, dielectric design margin must be >6 STD from the mean of the normal
probability distribution p(S) for n = 4.8x106 unit insulation sites. It can also be shown that
Equation 1 for a normal test distribution p(S) approaches an extreme value distribution as n
increases. 6
]! 0.999
~ 0.995
'"
Q.
.5
0.99
rJl
Q.
0.95
'"
Cl
c: 0.9
"0 0.8
~ 0.7
:c 0.6
'"
.0
e
0.5
Q. 0.4
c: 0.3
;;: 0.2
0
"0
.>< 0.1
'"
~
.0
Q)
>
til
:;
0.01
E
'"
<.) 10-3
10-4
-5 o
Normalized Stress (S-M)/STD
Figure 4_ System probability distribution for n parallel failure sites for a nonnal unit probability based upon test data.
408
CONCLUSION
The garter spring support concept for the beam tube requires special attention to dielectric
design to insure reliability.
Allocation of 1/3 of all possible failure modes of the CDM to dielectric modes requires an
allocated failure rate of 1.4xlO- 8 failures/hour over the 25 year life of the collider system.
Electrical insulation component and subcomponent failure test data can be used to predict the
cumulative failure probability of the collider system. A statistical failure analysis based upon an
approach used in large power apparatus is proposed as a method for assessment of collider
reliability for dielectric integrity. This statistical approach shows that the large volume and area of
electrically stressed regions of the collider system will demand sufficient design margin from
insulation failure test data to insure design life requirements. Based upon the analysis presented, a
design margin of >6 STD (from component test data) would be required to support an allocated
failure rate ofO.DOl for dielectric shorts.
REFERENCES
1. R. W. Balidi, et aI., "The Collider Dipole Program", Supercollider 3, Edited by John Nonte,
(Plenum Press, New York), pp. 539-547, 1991.
2. SSCL 15 Metre Collider Dipole Magnet Magnet System Specification, Doc. No: M80-000001,
Revision A, January 10, 1992.
3. Cable Insulation Working Group, SSCL MSD, Dallas, TX, February 5, 1992.
4. SSC CDM Cold Mass X-Section Mechanical Design, General Dynamics, San Diego, CA,
January 22, 1992.
5. Communications with J. Fraivillig at DuPont. DuPont preliminary spec sheet for Kapton
Poly imide film, HA for high performance amorphous structures with formability for
potential beam tube support application.
6. R.J. Thome, et aI., "Bus Design, Development and Test for the Superconducting Super
Collider," Magnet Technology-12 Conference, Leningrad, USSR, June 23-28, 1991.
7. N.G. Trinh, C. Vincent, F.A.M. Rizk, "Large Volume Breakdown of Transformer Oil," IREQ,
CEA Study 012 T 103, Progress Report #4, August 29, 1980.
409
A PRECISE TECHNIQUE FOR MANUFACTURING CORRECTION COIL*
Leonard Schieber

Magnet Division
Magnet Division - Bldg. 902A
INTRODUCTION
An automated method of manufacturing correction coils has been developed

which provides a precise embodiment of the coil design. Numerically controlled machine
have been developed to accurately position coil windings on the beam tube. Two types
of machines have been built. One machine bonds the wire to a substrate which is
wrapped around the beam tube after it is completed while the second machine bonds
the wire directly to the beam tube. Both machines use the MuItiwire® technique of
bonding the wire to the substrate utilizing an ultrasonic stylus. These machines are
being used to manufacture coils for both the SSC and RHIC.
DIRECT WINDING TECHNIQUE
An automated method of manufacturing correction coils directly on the beam

tube has been developed for the SSCL. The direct winding technique utilizes a
numerically controlled wiring machine consisting of a moveable wiring head positioned
above a mandrel which holds the beam tube. A servo positions the wiring head along
a track mounted on a bridge above the mandrel. The head has a travel of approximately
70". The head assembly contains a mechanism to control the feed rate of the wire, an
ultrasonic stylUS to bond the wire to the tube, a coil to excite the stylUS and a servo
system to rotate the stylUS about its vertical axis.
The mandrel is mounted below the wiring head on a granite table. The beam
tube is chucked into the mandrel much like the work in a turning lathe. A servo rotates
the beam tube in response to commands from the computer.
*Work supported by the U.S. Department of Energy

Plenum Press, New York, 1992 411
The desired wiring pattern is programmed on a floppy disk. The machine reads
the disk and drives the servos controlling the mandrel rotation, the head rotation and
the head translation along the mandrel; to trace out the pattern of the coil on the beam
tube. A servo controlled feed mechanism dispenses wire at the same speed that the
head is moving over the beam tube. The wire is fed under the ultrasonic stylus which
is being heated in response to signals applied to the ultrasonic coil. This melts a
thermosetting adhesive which has been placed on the beam tube; bonding the wire in
place. Multiple layers of wire may be placed on the beam tube in this fashion.
JELLYROLL TECHNIQUE
A similar technique has been used to make correction coils for the SSe. This
technique utilizes the same wiring machine by removing the Direct Wind Mandrel and
replacing it with a Material Transport Mechanism. A machine set up in this
configuration is shown in Figures 1 and 2. Instead of winding the coil directly on the
beam tube the coil pattern is laid on a flat substrate moved by the transport mechanism
to make multiple layers. The substrate is then wrapped around the beam tube like a
jellyroll. The machine shown can wind coils up to 1.2 meters wide and 8 meter long.
This results in a magnet 1.2 meters in length having up to 40 layers of wires.
Figure 3 shows the Jellyroll coil on its Kapton® substrate as wound by the
machine. The substrate is trimmed and one end secured to the beam tube. It is then
wrapped around the beam tube in the fashion of a jellyroll cake. Laminations are then
clamped around the coil winding.
RHIC MAGNET FACILITY
The techniques described are also being used to make correction coils for the
Relativistic Heavy Ion Collider. The RHIC design utilized individual substrates having
one or two layers of wire on each substrate. The substrates are then stacked in up to
six layers. Each corrector substrate is then bonded to a support tube. All support tubes
are then nested to form the complete correction coil.
Figure 4 shows four of the twelve substrates used in the RHIC corrector while
Figure 5 shows a close up of one of the Dipole layers. Dipole layers have two levels of
wire one on top of the other. The completed coil is approximately 24" in length. A
facility for manufacturing correction magnets for the RHIC project is being set up a the
Brookhaven National Laboratory. A recent photograph of the coil winding area is
shown in Figure 6. At the present time two of the machines are complete while the
remaining three are in various stages of completion.
SUMMARY
The technique described has proven to be an excellent method of manufacturing

correction coils. Wire placement accurate to .004" (lOOp.m) allows precise duplication
of the theoretical design. In addition, with proper construction techniques training
effects are virtually eliminated. Equipment is presently available to manufacture a large
variety of correction coils using this technique.
412
Figure 1. Wiring machine for jellyroll coils.
413
Figure 2. Material transport mechanism.
414
Figure 3. Jellyroll substrate for quadrupole corrector.
415
Figure 4. Four of twelve layer RHIC coil.
Figure 5. RHIC Dipole Coil.
416
Figure 6. Coil Winding Facility
417
A DESIGN FOR A HIGH VOLTAGE MAGNET COIL RINGER TEST SET
Wayne Koska 1 and Richard E. Sims2
IFermilab National Accelerator Laboratory *

Box 500
Batavia, IT.. 60510
2Superconducting Super Collider Laboratory *

M.S. 1003
Dallas, TX 75237
ABSTRACT
By discharging a bank of charged capacitors through a high power SCR switch into
an SSC dipole magnet assembly, it is possible to "ring" the coil and develop a voltage stress
of greater than 50 volts turn-to-turn, thereby verifying the insulation integrity. We will
present an overview of the test set design for a 2 kV isolated SCR ftring circuit, including
safety features, selectable capacitor banks, and digital waveform storage system. Results
from testing typical coils and magnets will be included. Possible upgrades are also
discussed.
INTRODUCTION
Superconducting accelerator magnets require good insulation integrity between coils,
and between the individual turns of the cable in a coil, since a quench which occurs in a
magnet which is a component of a string may induce in the other magnets in the string
potential differences on the order of hundreds of volts between cables, or a kilovolt or more
between coils at the mid-plane. If arcing should occur at these times the probability of
permanent damage to the magnet is high. During construction, magnets are continually
tested for electrical shorts using resistance measurements and DC hipots. However, the
potential differences developed between individual turns of a coil, (or between coils once
they have been bussed together) during these tests are less than 1 volt, and therefore cannot
determine if a situation exists where the insulation between two conductors is damaged to the
point where it cannot stand-off the voltages to which it will be exposed during operation.
The common way to test for this situation is to discharge a capacitor across the coil. This
will be referred to as a "ring test", and has been used extensively at the Brookhaven National
Laboratory 1 and at Fermilab2. Since the voltage dropped across the coil in this situation is
inductive, it will divide evenly across the individual turns. For a capacitor charged to 2000
V, and an SSC Collider Dipole magnet with 90 turns, a potential difference of 22 volts will
be developed across individual turns. In addition, the midplane insulation between the inner
coils will be stressed to the full 2000 V.
* Operated by the Universities Research Association, Inc., for the U.S. Department of Energy.

D DIGITAL
OSCILLO-
SCOPE
C
HIGH
VOLT
PROBE
D F
+ _>--'----e
HIGH VOLT
POWER
SUPPLY CAPACITOR
STORAGE 50MM
SSC
BANK Rs DIPOLE
(SWITCHABLE: COIL
60,120,180 & 240 mfD)
BASIC SCHEMATIC OF COIL RINGING TEST SET

Figure 1. The basic components of the high voltage coil ringing test set.
This paper will discuss the high voltage coil ringing test set developed at Fennilab for
production testing of SSC Collider Dipole magnets. This device uses an SCR switch to
connect the charged capacitors to the test coil instead of the more commonly used mercury
relay. Since this unit is used on the production floor, considerable effort has been made to
make it safe and easy to use for the technicians performing the tests. We will also discuss
the technique established for determining if a coil or magnet is satisfactory,
PRACTICAL CIRCUIT DESCRIPTION
Figure 1 is a simplified schematic of the coil ringing test set3. The basic components
are the high voltage power supply, the switchable capacitor bank, the SCR-diode
combination which is used to connect the capacitor bank to the test coil, the SCR firing
circuit, the digital display oscilloscope, and the safety circuits.
The power supply4 charges a bank of three, six or nine 20 Ilf bipolar capacitors5.
The capacitors are connected to the test coil via a parallel SCR-diode combination. The SCR
diode6 combination makes it possible for current to flow in both directions in the LCR
circuit. The SCR is fired using a 6 volt, 500 rnA pulse of width 30 Ils produced by a toroid
type isolated transformer? The digitizing oscilloscopeS is used to record, store and compare
the voltage waveforms produced during the ring test. The safety circuits are described in
detail below.
The major concern of a practical ringer circuit is safety. Charging capacitors of up to
240 microfarads with up to 2000 volts and discharging this much energy into a coil laying on
an insulated workbench demands that a number of conditions be monitored by "safety
circuits". Before the capacitors are charged, the following conditions are tested by the safety
circuits:
1) Is the circuit connected to an earth (building) ground via a heavy braided cable and
clamp?
420
2) Is the coil (or magnet) under test connected to the tester via heavy welding cable and
"vise grip" type clamps? The circuit must read less than about 30 ohms to pass this
test.
3) Is the door on the back of the tester closed?
4) Has the tester circuit been reset? (This clears any previous test conditions and
declares an intent to charge for a new test).
5) Is the emergency stop palm button reset?
6) Is the high voltage power supply "status OK" line true?
7) Are both "frre" buttons released?
When.all these conditions are true, the "OK to charge" green light is on, and all red
and yellow lighted buttons and indicators are off. The "charge" button may then be pressed
which switches the capacitors from a discharging resistor9 (Rd) to the high voltage pulsed
current power supply. When the capacitor voltage equals the programmed value, the power
supply is disconnected and an "at set voltage" legend and two "fIre" buttons light up green.
A Sonalert brand alarm has also switched from a pulsed sound to a continuous sound to
warn that the capacitors are charged and may be frred.
The operator now has about 20 seconds to press both "fIre" buttons. The buttons are
spaced far enough apart to require two hands, and the buttons must be depressed within 1/2
Figure 2. The ringer test set in use, testing an sse Collider Dipole coil.
421
second of each other to activate the ftring circuit. If the operator does not ftre the tester
within 20 seconds, the capacitors are automatically switched back to the discharge resistor
and the yellow "time out" and "discharged" lights come on and all green lights go out. The
tester must now be reset to charge again.
If any of the original conditions were to become false during the charged period, the
tester would revert to the discharged condition. Thus, if one of the coil connection clamps
became disconnected, or the earth ground opened while in the charged state, the tester would
dump the charge into resistor Ret.
In a normal f"lring sequence the ftring circuit would pulse the SCR gate at a 10
kilohertz rate, turning on the SCR and starting current flow through the coil under test. The
SCR continues to conduct until the current drops below a critical limit, at which time the
voltage across the capacitor bank has reached a maximum value with reversed polarity
relative to the original conditions. The current flow then reverses, now flowing through the
diode which is in parallel with the SCR. Again, when the current stops flowing in the
reverse direction, the SCR is turned back on by the gate pulses and the forward current flow
begins again. Due to resistive losses, each oscillation is lower in amplitude than the
preceding one until, after several cycles, there is not enough current to exceed the holding
current of the SCR and current flow ceases. The discharge light comes on and the operator
turns up the charge voltage to the next specifted step, presses "reset", and then "charge" to
continue the test. Figure 2 is a photograph of the ringer test set in the process of testing an
SSC Collider Dipole coil.
CHARACTERISTICS OF FAILURE MODES

When a capacitor, of capacitance C, is discharged across a coil with inductance L and
resistance R, the voltage measured across the capacitor varies as:
V = Voe-
:l!
2 cos (wt + a)
where
A change in inductance due to the development of an electrical short developing between two
adjacent cable turns will result in a change in the resonant frequency of the LCR circuit. It is
this shift in frequency which we are trying to detect. Figure 3 displays a calculation of the
shift in frequency which can be expected if the inductance of an SSC inner coil decreases by
5% due to a short when the maximum discharge voltage was 1000 volts relative to a similar
discharge at 500 volts. Of course, the Q of the circuit may also change due to the short,
which would also affect the ring frequency. A discussion of the frequency shift which
might be expected from changes in both these parameters may be found in reference 10. In
addition to a shift in frequency, erratic behavior of the voltage waveform, such as dips,
spikes or kinks, may be indications of arcing 1. If an arc occurs near the end of a magnet it is
likely that it will be heard by the technician performing the test.
TEST PROCEDURE
A standard procedure for performing ring tests on production magnets has been
developed. Prior to performing a ring test, DC hipot tests are performed where applicable.
Then the inductance and Q of the coil or magnet assembly are measured at approximately the
resonant frequency of the ringer circuit. The resonant frequency is then calculated. The
ringer capacitors are charged to a low voltage, typically 100 V, and discharged into the coil.
The waveform is recorded with a digital oscilloscope and the measured frequency compared
to the calculated value. We have found that the measured frequency is very sensitive to the
422
1000·
800
600
!l
'0
;;;.. 400
200
o
800 1000
Time (Ils)
Figure 3. The calculated response of an sse inner coil, with nominal inductance, to a ring at 500 volts, and
the response with inductance reduced by 5%, to simulate a turn to turn electrical short, at 1000 volts. The
shift in the zero volt crossing point is obvious.
calibration of the high voltage probes used, so it may differ from the calculated value by up
to 20Hz. If the difference is significantly greater than this, additional tests are performed to
determine the reason before subjecting the coils to higher voltages. Otherwise this waveform
is stored and used as the baseline. A series of rings with progressively higher maximum
discharge voltages are then performed and the resulting waveforms are stored in memory in
the display oscilloscope and overlaid on the screen. The zero voltage crossing points are
examined for any indication of a shift in the resonant frequency as the discharge voltage is
increased. The shape of the waveforms are also examined for erratic behavior.
Ring tests are performed at critical stages in the assembly of a magnet. Each coil is
tested after it is cured. These tests are performed at 100,500 and 1000 volts and produce a
maximum turn to turn potential difference of 53 V (38 V) for inner (outer) coils. Figure 4
shows the results of a ring test performed on a 50 mm aperture SSC Collider Dipole inner
I I I I , I I , I
I , I I I I I t I
_ _ _ _ L _ _ _ _ , _ _ _ _ _ 1 _ _ _ _ .J ____ .J _ _ _ _ 1 ____ L ____ L _ _ _ _ , _ _ _ _ _
I I I I I I
I I I I I
CI
o
.r;;
:~
Q
---------,----1----7----r-- I I I I
I
,
I
"
I ,
- - - -1- - - - -1- - - - .., - - - - "" - - - - ,. - - - - ... - - - -
" I I ,
' I 1 I I
100 f.lS / Division

Figure 4. The voltage traces which resulted from ringing an sse collider dipole inner coil.
423
I I I I I I I
_1I _ _ _ _ -1I _ _ _ _ ... _ _ _ _ .1I _ _ _ _ LI _ _ _ _ ,L _ _ _ _ 1I _ _ _ _ _ 1I _ _ _ _
I I I I I I I I
I I I I I I I I
j
I I I I I I I
- - - -,- - - - , - - - - T - - - - r - - - - r - - - -c - - - -,- - --
--!!
~
~
~ I I I I I I I
- - - -t- - - - -1- - - - -1- - - - '1 - - - - 1 - - - - ,. - - - - ~ - - - -,.. - - - -1- - - - -
I I I I I I I
I I I I I I I
2 ms I Division
Figure 5. The voltage traces which resulted from the final ring test performed on sse collider dipole magnet
DCA319, prior to shipment to the test facility.
coil. The convergence of the traces at the zero volt crossing point indicates that this coil has
acceptable cable to cable insulation integrity. The assembled coils are next rung after the
keying operation, which is the step at which the coils were exposed to their maximum stress.
The upper to lower splice is still accessible at this time so that the upper coils and lower coils
can be rung separately. The maximum discharge voltage is 2000 volts resulting in a tum to
turn voltage of 44 volts. The test is next performed after the shell has been welded in place;
when the cold mass is essentially completed. A final test is performed just prior to shipment
to the test facility to determine if the insulation integrity has degraded during the time the
magnet was being cryostated. These latter tests are performed with a series of discharges
beginning at 500 volts and increasing in 500 V steps to 2000 volts. The maximum turn to
tum voltage is 23 V. In addition, the potential difference across the inner coil midplane
insulation is 2000 V. Figure 5 shows the final ring test performed on the SSC collider dipole
magnet DCA319. Again the traces converge at the zero crossing points, indicating that the
insulation integrity of this magnet is satisfactory.
POSSIBLE UPGRADES
The voltage developed between individual cable turns when ringing fullr bussed
magnets is marginal for discovering faults. Much of the design for a 5 kV ringer I, which
can be implemented by firing three series connected 22W volt peak rated SCRs with a
three secondary winding transformer, has been completed I . All three SCRs must be fired
and must turn off within nanoseconds of each other to divide the 5000 volts equally.
Commutation capacitors across each SCR helps this situation. However, at high
frequencies the coils in a magnet behave like a transmission line so that the high
frequency components of the ringer pulse, to which the magnet is exposed when the SCRs
switch on, cause the early turns of the magnet to experience transient voltages ("" 10 ps
long) on the order of 10 times higher than the value obtained by dividing the whole coil
voltage by the number of turns. At very high maximum ringer voltages this would over
stress the turn to turn insulation of the coils. Initial modeling of the ringer-magnet
circuit13 indicate tha a high current, low inductance resistive choke coil inserted between
the SCR and magnet would greatly reduce these high frequency components. Further study
will need to be done on this problem before a 5 kV ringer test set can be used to routinely
test magnets.
A second improvement to the current high voltage ringer test set would be to control
the oscilloscope through a computer. This would allow the digital oscilloscope to be
automatically programmed for each test sequence and magnet situation, as well as
providing a means to collect the data, perform a routine analysis, and produce a test
report.
424
SUMMARY
A high voltage coil ringing test set has been developed at Fermilab for the production
testing of SSC Collider Dipole Magnets. Emphasis has been put on building a system that is
safe and easy to use for the technicians performing the tests. Ring tests are performed on
each individual coil prior to packaging as a magnet, then on the coil assemblies after they
have been keyed (this is after the coils have been subjected to maximum stress during the
keying process), after the magnet shell has been welded around the yokes and just prior to
shipping the magnet to test. Only one fault has been detected with this system in the 50 mm
dipole magnet program at Fermilab. This was a lead to ground arc caused by improper
insulation of the lead. It was easily repaired. Our experience during the SSC magnet
program at Fermilab is that the high voltage ringer test is useful for exposing assembly
problems early and has never damaged a coil or magnet that was not already defective.
ACKNOWLEDGEMENTS
A number of people have helped with the design of the ringer test set and the
procedures developed to test SSC magnets at Fermilab. We would like to acknowledge the
work of Donna Kubik, Richard Gaf, Dennis Gaw and Eric Schmitz. We would also like to
thank Jim Strait for his suggestions during the early development of this equipment.
REFERENCES
1 G.G. Sintchak, J.G. Couingham, and G.G. Ganetis, Electrical Insulation Requirements and Test Procedures
for SSC Dipole Magnets, Supercollider 2, 397, (1990) M. McAshan, ed.
2 P. Mazur, private communication.
3 The Fermilab drawing number for the 2000 Volt Ringer is OlD2-ME-301S04.
4 The power supply used was a CCS-200-S-P manufactured by Converter Power Inc. (part # CCS-200-S-P).
S These were 20 1J.f, 7.5 kV capacitors, manufactured by Maxwell Laboratories, Inc. (part # 3066S). The high
voltage capacitor transfer relays were manufactured by Ross Engineering Corp. (part # E1SDTlS-I-O).
6 The SCR was manufactured by Powerex, Youngwood, PA (part # T700223S04) and is rated at 2200 volts at
3S0 amps. The diode was manufactured by Powerex (part # R6002630XY A) and is rated at 2600 volts at 300
amps.
7 The toroid core used in the firing circuit was manufactured by Magnetics Co. Inc. (part # S8110-A2,
MU=60, distributed gap). It was modified by adding 200 turns of #20 wire to both the primary and secondary
windings.
8 The oscilloscope used was a Hewlett Packard model S4S01A with a model 1137A SOOO V probe.
9 High voltage, non-inductive resistors were obtained from Chaddock Electronics, Inc.
lD W. Koska, The Ringing Technique for Determining Turn to Tum Electrical Shorts in SSC Collider Dipole
Magnets-Parts 1,2 and 3. TS-SSC 92-029, TS-SSC 92-030 and TS-SSC 92-031.
11 The Fermilab drawing number for the SkV ringer is 0102-ME-290707.
12 If three SCRs are to be used in series, they must be matched in leakage and stored charge (Qrr)' It is
possible that a single high voltage SCR manufactured by GE-SPCO, Malvern, PA. (part # C716) may be used
for service up to 4800 working volts. The firing circuit may have to be modified to supply proper voltage and
current to the gate for this device.
13 E. Schmitz, Electrical Modelling of Magnets-Upper Inner Section, TS-SSC 91-231, 11(27/91, and
Electrical Modelling of Magnets-Long Time-Scale, TS-SSC 91-247,12/12/91.
425
EVOLUTION OF THE CDM DESIGN
Hugo van den Bergh

P.O. Box 85990, MjZ Cl-8442
San Diego, CA 92138
INTRODUCTION
The design goal during the prototype development phase is to take the tested and
proven Collider Dipole Magnet baseline developed by SSCL, FNAL and BNL, and adapt
it for high rate production while maintaining or enhancing performance, reliability and
uniformity between magnets. In many areas no changes are required, but a choice had to
be made between different solutions proposed by the National Laboratories. In most of
these instances the reasons to pick one solution over the other were less than compelling
since all perform about equally well.
This paper describes the major changes which are being implemented to the
superconductor insulation, the collars and yokes, the helium vessel shell, end plates,
bulkheads, the cold mass support system, the thermal shields, and the beam tube support
system. For each of these changes, a brief rationale will be given, plus an indication as to
how any uncertainty resulting from these design changes will be resolved in addition to the
testing in the model- and prototype magnets. Figure 1 shows a cross-section of the CDM
with nomenclature for the major magnet components.
COLD MASS
Superconducting Cable Insulation System
Based on extensive tests conducted at the Brookhaven National Laboratory, it was

decided to use a super conducting cable insulation system consisting of two layers of
polyimide film rather than the polyimide-fiberglass insulation system used on all
previously built FNAL-and BNL dipole magnets. This new cable insulation system
consists of one layer of polyimide, coated on one side with adhesive, and a second wrap
of alumina-loaded polyimide with adhesive on both sides. Each wrap has a 50% overlap.
Based on data generated at BNL, this insulation system appears to have considerable
advantages over the polyimide-fiberglass system:
- the higher punch-through resistance
- better radiation resistance
- improved creep performance
- more accurate and uniform conductor placement in the coils.

£,.
(WFJ 1)1""0 own ....... I 1. 1
TIIVR\IA LSf'aA"
At tt. TIURMA
Figure 1. Dipole magnet cross-section showing major design features and enhancements.
428
However, during testing at BNL of short magnets wound with the double polyimide
system, the quench current was discovered to be dependent on the current ramp rate. It
was subsequently demonstraterl that the method userl to mold and cure the coils resulted in
a significant reduction in cable interstrand resistance, which could result in the generation
of eddy currents during magnet excitation. These eddy currents probably were the cause
of quenching with high ramp rates. As a result of a subsequent series of tests at BNL, a
curing cycle for the coils was picked which showed to be superior to all other cycles with
no decrease in interstrand resistance over the initial value. This method is at this time the
design baseline. It will be used in 'ten-stack' tests, model magnets and prototype magnets
to verify its adequacy.
Collar, Yoke and Shell Design

The coil support structure requires the controlled interaction of the coils, collars,
yokes and helium vessel shell. They all play a critical role in obtaining the desired
structural performance. The ultimate goal of the design of these parts is a coil shape at
magnet operating temperature and zero field as determined by the magnetic analysis.
Figure 2 shows the cross-section of the cold mass assembly.
",ElLA IUOC'Ot..LIot5
(UMMTttt • •
Figure 2. The cold mass cross-section shows the design feature that have been modified.
A major decision concerning the design of the cold mass had to be made concerning
the use of either horizontally or vertically split yokes. Magnets constructed with either
design have been built and tested with equally good results. However, during magnet
operation, the Lorentz forces acting on the collars cause the latter to deflect horizontally. If
split vertically, this deflection in tum tends to open up a gap between yoke halves, and it is
the shell which ultimately has to react these forces and prevent opening of this gap. Since
the longitudinal welds which join the two shell halves and the heat affected zones in the
weld areas are essentially hoop stress over time and may eventually lead to opening of the
gap between the yoke halves.
429
With s horizontal split, the stiffness of the yoke resists the horizontal component of
the Lorentz force. This drastically reduces the hoop stress in the shell. Based primarily
on the above, it was decided to use the horizontally split yoke.
It was also decided to use fineblanking for the production of both the stainless steel
collars, and the iron yokes. Fineblanking allows the use of substantially thicker material
(2.5 mm vs 1.5 mm for the collars, 6 mm vs 1.5 mm for the yokes) which substantially
reduces parts count and handling costs, and results in parts with closer tolerances and a
clean sheared edge. Figures 4 and 5 show a collar pair and a yoke,respectively.
In addition it was decided to replace the two spot welds, used in previously built
magnets to join two collar laminations into a collar pair, with two semiperforated
(semiperfed) features as shown in Figure 3.
2.S I\1M TIIIC K
Figure 3. The collar laminates are joined into pairs using semiperfs.
- 6M,\1 \ ALI .
TIll ' K" F-'iS
Fig.ure 4. The yoke laminations are semiperfed to permit stacking them with adjacent parts flipped
honzontally 180 degrees,
430
LO 0 . RE\ S
TIIR li. T rLAT .
Figure 5. The end plate design allows installation after the beam tube assembly is already in place .
The press fit of the semiperfs results in a secure mechanical joint of the two laminations
and eliminates the possibility of local thermal degradation of the laminates at the spot
welds. Satisfactory joining of the collar laminates is no longer dependent on control of the
weld process.
The yoke laminations are also joined by semiperfs into yoke modules of
approximately 700 mm length. The semiperfs replace the stacking tubes and make
assembly of the yoke modules more efficient. In addition, "chevroning" of the yoke
laminates at the magnet ends is completely eliminated.
Even though the accuracy and flatness of fineblanked parts is superior to parts
manufactured by the stamping process, a slight crowning of the laminations is still
expected. To avoid having the yoke modules and thus the cold mass assume a slight
bend, adjacent yoke laminations will be horizontally rotated 180 degrees, so that the bow
in adjacent laminations is opposite. The design of the semiperf features allows this
horizontal flipping of the laminations.
Helium Vessel Shell, End Plates and Bulkheads

Figure 2 shows the horizontal split of the yokes and the helium vessel shell. It also
shows the weld backing bars which fit inside cutouts in the yokes. They prevent any weld
material to enter the helium vessel as the closeout welds are being made.
The backing bar also serves to lock the first half of the yoke modules in place after
they have been deposited in their exact position in the helium vessel half shell. To this end
the backing bar is placed in the appropriate yoke module cutouts, and tack welded onto the
helium vessel half shell, locking the yoke modules accurately in place. The collared coil
assembly and upper yoke modules can now be precisely registered with the lower yoke
modules.
At both ends of the cold mass an end plate, shown in Figure 5, is welded into the
helium vessel shell.
The cold mass bulkheads are circumferentially welded onto the end plates as shown
in Figure 6.
431
WF-LO PRFP~ _ - --,.
In I 111\1 ' F_ S L SIIEI.I.
r.. DPI. TF.
Figure 6. Welding of Bulkhead onto the End Plate.
The design of the bulkheads has to take into consideration the minimum wall thickness
required by the applicable vessel code, while minimizing the circumferential weld to
prevent inward bending of the end plate.
The design of the helium vessel has to comply with Division VIII, section 1 of the
ASME pressure vessel code. This requires reinforcements around all vessel wall
penetrations. Since the size of the end dome does not permit having individual
reinforcement rings, the whole dome was made the required 11 mm thick.
Recent experience at FNAL with welding a thick-walled dome onto the end plate
has shown that this large, multi-pass weld required a long time to complete (days), and
resulted in rather high and unpredictable compressive stresses in the coil structure. The
reason for this increased stress may be that, as the circumferential weld beads on the
outside of the end plate cool down and shrink, they introduce bending moments into the
end plate which let its center bend inward. Tapering down the dome wall thickness to 6.9
mm will greatly reduce this effect while shortening the welding time considerably.
It should be noted that both end plate and end dome have rather large central holes.
These holes enable installation of these parts by sliding them in place over the previously
installed beam tube, as shown in Figure 7.
Figure 7. Installation of the Beam Tube Assembly and Bulkhead
432
We have elected to install the beam tube as a complete assembly to avoid having to
weld flanges onto the beam tube after installation with the danger of damaging the internal
copper plating. In addition, it makes for a more risk-free design if the complete beam tube
assembly is tested prior to installation.
Cold Mass Support System, Thermal Shields
Prior to the start of our design effort, SSCL introduced the cold mass bulkheads,
plus tube and bellows connections between adjacent magnets for the cryogenic lines. As a
result of these changes, the quench forces acting on the cold mass, and hence the
supports, are greatly reduced which eliminates the need for tie bars between support posts.
Removal of the tie bars has led to a substantial simplification of the support system.
Instead of having the cold mass supported and sliding in cradles which are attached to the
support posts, which in tum are firmly attached to the vacuum vessel, saddles are welded
onto the helium vessel shell at the five post locations. The support posts are firmly
attached to these saddles (see Figure 1). All posts but the central one are allowed to slide
at their interface with the vacuum vessel as the cold mass contracts and expands. This
design change eliminates the problem of finding a slide material which is compatible with
the high radiation dose it would experience at the cold mass surface. It also replaces the
complex mechanisms of the slide cradle with only one part, the cold mass saddle.
Based on lessons learned during assembly of the COM's at FNAL, the thennal
shield assembly has been modified. Large distortions of the shields, plus the necessity to
insert slide rings around the support posts, made their installation difficult and time
consuming. The shield distortions were mostly due to the continuous longitudinal welds
used to attach the cryogenic lines to the shields. The slide rings around the support posts
were required to provide slide surfaces between the post thermal intercepts and the shields,
since all shield sections were welded onto the continuous cryogenic lines, making them
expand and contract as one unit.
The design has been changes to make the cryogenic lines structurally independent
from the thennal shields. This allows each shield segment to be firmly attached to a
support post which eliminates the need for slide rings. It also eliminates the longitudinal
welds which were the cause of shield distortion. In addition, the cryogenic lines can now
be made of stainless steel and the use of aluminum-to stainless transition tubes at the
bellows, a cost driver and design risk, is no longer required. Thennal contact between the
shields and cryogenic lines will be made by thermal straps.
The cold mass support system consists of five straight reentrant posts (see Figure
1). the design utilizes fIlament-wound, composite tubs connected to metal end fittings and
thermal intercepts by means of shrink fittings. This post is similar to the one developed
and successfully tested at FNAL.
The design was selected since the materials used are well characterized, they do not
introduce risk factors such as creep over the operational life, are relatively easy to
assemble and less expensive than the reentrant posts. A series of tests to validate the
design prior to their use in the prototype magnets is contemplated.
ACKNOWLEDGEMENTS
Research Association in support of the Superconducting Super Collider project for
the U.S. Department of Energy.
433
MECHANICAL BEHAVIOR OF FERMILAB/GENERAL DYNAMICS
BUILT 15M SSC COLLIDER DIPOLES *
M. Wake t , M. Bleadon, R. Bossert, J. Carson, S.W. Delchamps,

S. Gourlay, R. Hanft, W. Koska, M. Kuchnir, M.J. Lamm,
P.O. Mazur, D. Orris and J. Strait
Fermi National Accelerator Laboratory, Batavia, 1L
A. Devred, J. DiMarco, J. Kuzminski, W. Nah, T. Ogitsu t ,

M. Puglisi, J .C. Tompkins, Y. Yu, Y. Zhao and H. Zheng
SSC Laboratory, Dallas, TX
INTRODUCTION
A series .of full-scale demonstration SSC collider dipole magnets were built for the ASST
(Accelerator System String Test). These magnets, DCA311 through DCA319, have 50 mm
aperture and 15 m magnetic length with 6.6 Tesla uniform field. For the support structure of
the W6733B cross section[l], the Fermilab design uses a vertical split[2] in the yoke. The end
sections of the magnet have solid spacers and are supported by collet clamps[3]. The splices
between inner and outer coils are made in preforms which lie outside of the high field region.
The magnets were produced in pipeline fashion with no intentional major changes between
magnets. As a part of the technology transfer program, the last 7 magnets were built by
General Dynamics personnel using the magnet construction facilities of Fermilab, while the
first two magnets were built entirely by Fermilab personnel.
At present, the magnets up to DCA316 have been tested at Fermilab. The general charac-
teristics of the magnets have been quite satisfactory. Both of the Fermilab built magnets have
reached the conductor limited field strength with no significant training. Two of the Gen-
eral Dynamics built magnets (DCA313 and DCA314) each required a single training quench.
However, all of the magnets tested up to date meet the ASST specifications.
This report describes the mechanical properties of the ASST magnets at Fermilab based
on the currently available test results.
COIL SIZE
Primary support of the coils is provided by stainless steel collar laminations. Upper and
lower laminations are keyed together in a press to apply a pre-load to the coils. The amount
'work supported by U.S. Department of Energy
ton leave from KEK National Laboratory for High Energy Physics, Tsukuba, Japan

of pre-load achieved depends on the coil sizes. Because of this, coil sizes must be precisely
controlled.
Coil sizes were measured relative to stainless steel reference parts under 8.4 MPa compres-
sive force. Sizing measurements were made at 380 positions along the length of each coil. Fig.
1 and Fig. 2 show the histories of the average inner and outer coil sizes respectively. There
were significant variations of the size of each coil along its length. The error bars in Fig. 1
and Fig. 2 indicate the standard deviation of coil size for of each coil.
0.35
0.30
0.25
8'
.,.6
N
.0;
0.20
1-0
Q)
>
f!
0 0.15
o 10 e
r-
~
0
+
0.05
1000 1005 1010 1015 1020
Coil Number
Fig. 1 Inner Coil Size at 8.4 MPa Relative to Reference ("oversize").

Coils were paired in sequential order starting at #1003 with the
exception of #1007, paired with #1016 in DCA317.
Fig. 2 Outer Coil Size at 8.4 MPa Relative to Reference ("oversize").

Coils were paired in sequential order. The size of the last coil, #2021,
was measured using a different device, so that the value shown needs a
correction of +88 #J.m.
436
As shown in Fig. 3, the coils have quite non-linear size characteristics under pressure.
Therefore, the relationship between coil size and achieved pre-load is not straightforward.
However, the optimUTIl oversize values required to achieve initial pre-loads of 70 MPa and
55 MPa were +0.25 mm and -0.02 mm in the inner and outer coils respectively. These
values were determined empirically from 1.5 m model magnet pre-load data.
Necessary adjustments to the inner and outer coil sizes were made by adding and removing
shims applied to the curing tooling. Coil size seemed to decrease when a new curing die was
introduced. The coil size became very stable after several coils had been cured. Variations in
coil size from coil to coil could not be meaningfully correlated with cable thickness variations,
which were in the range of 5 p,m. Rather, the coil size bore the influence of the cable keystone
angle, which was systematically about 0.08 degree below the design value throughout produc-
tion.
SSC Inner Coil
6
'?
e
0..
..."
::l
4
rn
rn
"...
0..
0.4 0.5 0.6 0.7

Relative Size (mm)
Fig. 3 Relation of Pressure and Coil Size.

Coil size non-linear in applied pressure. The large hysteresis causes a
complicated relationship with pre-load.
PRE-LOADING
The entire process of collaring and yoking was monitored by pressure gauge packages placed
at the minimUTIl and maximUTIl size locations along the coil[4]. Each pressure gauge package
measured the coil pressures of all quadrants of the inner and outer coils using beam type
strain gauge transducers. The coil assembly was placed under a hydraulic press with collar
lamination packs in place. Vertical pressure was applied and maintained while tapered keys
were inserted by a horizontal press. The vertical pressure was kept at the same level during
keying to avoid deep scratches on the keys which might lead to degradation of magnetic field
uniformity. The pressure in the coils was increased by the insertion of the keys, so that the
coils experienced maximUTIl pressure during key insertion. The highest recorded value was
130 MPa, in the inner coils of DCA314. This pressure was still low enough so as not to damage
the insulation. When the horizontal and vertical pressures were removed after key insertion,
the coil pressure sprang back by about 20 MPa in the inner and 30 MPa in the outer coils.
The inner and outer coil pressures crept down another 20 MPa and 4 MPa respectively over a
few days. About 70% of the pressure loss due to creeping was recovered by yoking. The shell
was welded under pressure around the yoke, resulting in a shell tension of about 200 MPa.
Cool-down caused pressure losses of 30 MPa in the inner and 15 to 20 MPa in the outer coils,
since the collared coil contracted more than the yoke and shell.
The use of anti-ovallzed collars in conjunction with the vertically split yoke prevented the
437
thermal contraction difference between collar and yoke from causing contact loss along the
horizontal plane. A gap of 0.3 mm is expected at the vertical yoke/collar interface, but this
should have no effect on mechanical stability.
EXCITATION BEHAVIOR
The coil pressures were also measured during the excitation of the magnet. Fig. 4 and
Fig. 5 show the results. Since the vertically split design ofthe yoke provided support of the col-
lars at any current, the pressure showed a smooth monotonic dependence on the square of the
current.
40 SSC Inner Coils
-;u
! 30
Q)
"
;::!
III
III
Q)
20
"
0..
312
Q) 316
bll 313
«I
"
~ 10
m
311
Current Squared (kA**2)
Fig. 4 Inner Coil Pressure DUling Excitation
50
40 313
314
-;u
0..
30
6
~~~~315
Q)
"
;::!
III
III
~~~~;;;;;;311
r
Q)
20
"
0..
Q)
312
bll 316
«I
"
Q)
i( 10
SSC Outer Coils
0
0 10 20 30
Fig. 5 Outer Coil Pressure During Excitation
438
The yoke laminations of magnet DCA311 showed chevroning due to loose packing. The
larger than average change in coil pressure with excitation current in the DCA311 data
(especially the inner coils) is a reflection of the diminished support of the collars from the
yoke in this magnet.
The change of pressure in the outer coil is much less than in the inner coil. Even though
the magnetic field magnitude at the outer coil is less than that at the inner coil, the direction
of the field is tilted more toward the horizontal plane. Therefore, the azimuthal component
of the electromagnetic force in the outer coil should be at least 60% of the inner coil value.
The small change of pressure with excitation in the outer coil can be explained if the collar
is not making contact with the yoke in the vertical direction. The vertical spring constant of
the coil-collar system is smaller than the horizontal one because there is no backup by the
yoke. The outer coil is more influenced by the vertical motion of the collar because it occupies
azimuthal angles close to the horizontal plane. This model is also supported by the fact that
the effect of the chevroned yoke is not pronounced in the outer coils of DCA311.
The amount of pressure change in the inner coils is related to the initial pressure. In partic-
ular, magnets with large initial pressure tend to lose more pressure during magnet excitation,
as shown in Fig. 6. The deformation of the coils is determined by the hoop stress and thickness
of the collar and yoke. Therefore, the amount of deformation should not vary much between
magnets, as the azimuthal strain observations discussed later verify. IT the deformation is
fixed, coils with high effective modulus of elasticity (implying large initial pressure) lose more
pressure upon excitation. All magnets underwent excitation with a comfortable pre-load mar-
gin. The initial single training quenches of DCA313 and DCA314 might be related to unnec-
essarily large pre-load in the outer coils, which causes an imbalance of forces in the inner coils.
40
35
0311 0312
-;0 30 0311 0316
0..
6
..; 0313
..>:
l'- 25 0314
.... 314 00 ~1~6
'" 313
0"
0315
-'
" 20
....OJ
;:I o Ma.ximum Size Position
"....OJ" 15 o Minimum Size Position
0.. 0315
10
10 70
Zero Current Pressure (MPa)
Fig. 6 Excitation Pressure Loss vs. Initial Pressure.
END FORCE
The magnet is partially supported in the axial direction at each end by sets of 4 stainless
steel bullet slugs instrumented with strain gauges. Fig. 7 shows the total load experienced
by the bullets in each magnet during the excitation. The end force increases in proportion to
the current squared. However, only 10 to 15% of the total electromagnetic force (100 kN at
6.6 T) is supported by the bullets.
Most of the axial electromagnetic forces were transferred to the yoke and shell through
friction between the collar and yoke. The bullets were pre-loaded by adjusting bushing screws
439
threaded into the end plates. Due to deformation of the end plate during welding, the ini-
tialloading of the DCA311 bullets was much larger than intended. On later magnets the
initial bullet load was set low to take into account the increase during welding. Even in
the extreme case of DCA311, the end force behavior measured by the bullets was not much
different from that observed in the other magnets. Changes in bullet loads over thermal cy-
cles were observed, but these changes were not drastic nor influential to quench performance.
50
40
Z 30
C
.,
.
0
0
r..
..., 20
t::
f.il
iii
-0>
0
E-o
10
0
0 30 40 50
Fig. 7 End Force During Excitation Measured by Bullet Strain Gages.

Sum of the loads in 4 bullets is plotted as total end force.
AZIMUTHAL SHELL STRAIN

The strain in the shell was also observed with strain gauges. The change in azimuthal
shell strain during excitation is shown in Fig. 8 at different angles from the vertical plane.
All of the magnets exhibited similar behavior. No change in the rate of strain increase with
excitation was observed, as would have been the case if there were not enough pre-load in the
shell. Strain measured at angles close to the vertical remains compressive, while the high angle
gauges exhibit tensile strains. The strain profiles shown in Fig. 9 are quite different from the
uniform distribution of tensile stress predicted by a simple hoop stress model. However, since
these gauges were attached to the outer surface of the shell, they were sensitive to changes of
curvature (bending) in addition to strains in the initial plane of the shell material. The major
deformation inferred from the strain gauge data is therefore an ovalization in the horizontal
plane. Such a deformation is understandable, because the electromagnetic force on the coil is
mainly in the horizontal direction. The larger slope at angles near the parting plane could be
due to bending of the yoke facilitated by the reduced stiffness in the vertical direction caused
by the gap between yoke and collar.
The average stress, derived from the strains after Poisson correction, gives a value on
the tensile side, but not more than a few MPa at 7 kA. The hoop stress of the electromag-
netic force is apparently taken up by a decrease in the compression of the yoke, as designed.
440
50.0
, _______x:....----x
, , , , x...--x:
25.0 ---- -- -i- -- ----~,:":,,,:x~~.:-:-x~x~ -:- -- _. - - -:--- DCA315
'/' " ,
;x~x~x ~:~..r>--,<
CD
~ 0.0
x:::3-x ' - D - - - - 17°
<l
-<>-- 28°
- - f r - - 38°
-25.0
- - 0 - 61°
--x-- 90°
-50.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0
1**2 (kA**2)
Fig. 8 Azimuthal Strain Change.
AXIAL SHELL FORCE

The shell has axial strain gauges mounted at 7 longitudinal positions. Fig. 9 is the stress
profile of the axial force measured by these strain gauges. A simple spring model[5] for both
shell and coil, with coupling between these two springs provided by the deflection of the yoke
and collar laminations or the sheer stress at material boundaries, predicts an exponential
profile starting with the bullet force at each end and approaching some value at the center of
the magnet. The overall profile of the axial strain can therefore be explained by this model.
It is interesting that this transfer of the force to the shell takes place within the first 2 m of
the end. The shell tension derived from observed strain at the center of the magnet indicates
that 60% of the electromagnetic force is supported by the shell. The coils or other structures
such as collaring keys take up the remaining 40% of the force. The posts may take some of
the force as well.
The decrease in strain near the end is very large. In fact, the stress becomes compressive
at the very end. This compressive stress is due to the absence of ovalization at the end of the
magnet brought about by two causes: First, there is no radial force in this region to promote
ovalization directly, and second, the end plate suppresses any ovalization which might take
place. Close to the vertical plane, where the strain gages are mounted, the surface of the
shell has to curve toward the outside approaching the end plate. This is why we observe an
apparent compressive stress instead of the tensile stress expected from the bullets.
No significant hysteresis was observed in the axial stress. However, DCA313, DCA314, and
DCA316, which showed some training, seem to have more negative values in the stress change
at the end. The zero current strain at the very end also changed with thermal cycling. These
deformations near the end might be related to the mechanism which brings about training.
CONCLUSIONS
The mechanics of a set of 15 m ASST SSC collider dipole magnets with vertically split yokes
built at Fermilab have been explained using strain gauge data. All the observed mechanical
behavior supports the effectiveness of the design principles of the ASST magnets. Detailed
analysis might provide a clue toward solving the problem of occasional training quenches.
441
50.0 , - - - - - - - - - - - - - - - : - - - - - - : - - - - ,
25. a ------- .J - - - - - - - - '- - - - - - - - l - - - - - - -
• DCA312
- - 0 - DCA313
- - - - DCA314
-25.0
--<>--- DCA315
* DCA316
-50.0 ..l....._ _~_ _~_ _~~_ _~_ _~_ __
o 15 30 45 60 75 90
Angle from yoke parting plane (degrees)
Fig. 9 Azimuthal Strain Profile.
80
60
Oi'
3...
Ul
...0 40
(J
1
.;;;" 20 x DCA311[7052A] + DCA314[7036A]
b o DCA312[7037A] " DCA315[7034A]
rn o DCA313[7029A] oj< DCA316[7031A]
<Ii
R 0
..:
-20
0 2 4 6 8 10
Distance from End {m}
Fig. 10 Axial Strain Profile
REFERENCES
[1] R. C. Gupta et al., "ssc 50 mm Dipole Cross Section," Supercollider 3, p 587, J. Nonte
(ed), Plenum Press, New York 1991.
[2] J.Strait et al., "Mechanical Design of 2D Cross-section of the SSC Collider Dipole
Magnet," Proc. of IEEE Particle Accelerator Conference, San Francisco, p 2176, 1991.
[3] S.Delchamps et al., "ssc Collider Dipole Magnet End Mechanical Design," Proc. of
IEEE Particle Accelerator Conference, San Francisco, p 2185, 1991.
[4] M.Wake, "Pressure-Size Diagram Update with DCA Series Long Magnet Data,"
Fermilab Technical Support Section Internal Note TS-SSC-91-254, 1991.
[5] M.Wake, "DCA313-316 Axial Strain Change," Fermilab Technical Support Section
Internal Note TS-SSC-92-023, 1992.
442
QUENCH PERFORMANCE OF SIX SIMILAR 17 m-LONG, 40 mm-APERTURE
SSC MODEL DIPOLES·
P. Wanderer and E. Willen

Magnet Division
SUMMARY
The quench performance of the final six 17 m-Iong, 40 mm-aperture SSC model
dipoles made at Brookhaven National Laboratory is presented in summary form. The six
magnets are of similar construction, except for two features: (1) the copper-to-
superconductor ratio of the inner cable (either 1.3: 1 or 1.5: 1) and (2) the axial preload at
4.35K. The inner coil quench performance does not appear to be correlated with the
copper-to-superconductor ratio. There does appear to be a correlation between the outer
coil quench performance and the cold axial preload.
INTRODUCTION
The construction of the final six full-length 40 mm-aperture Collider dipoles at

BNL was intended, among other things, to test the effect of the inner coil copper-to-
superconductor ratio on quench performance. The first three magnets, DC0201 - DC0203,
had Cu:S.C. = 1.5: 1 and the last three, DC0204 - DC0206, had Cu:S.C. = 1.3: 1. The
stability of cable in short-sample measurements has been found to be dependent on the
Cu:SC ratio.l In other respects the magnets were intended to be alike, with axial and
azimuthal preload in operation up to 6.5 kA (6.6 T). Construction details pertinent to this
note are listed in Table I. The remaining construction features are described more fully
elsewhere2 •3 •
First, this note examines the quench performance of the inner coils. The quench
performance of the magnets is shown in Figs. 1 - 6. All of the magnets achieved a plateau
at 4.35 K, with the plateau quenches originating in the inner coil. Four of the magnets
quenched repeatedly in the pole turn (turn 16), indicating conductor limited quenches. For
these magnets, the ratio of the conductor limited quench current, leu to the quench current
'This work supported by the U.S. Department of Energy.

Plenum Press, New York:, 1992 443
:I : :I : I:
I I I I I
B - I
I
I
I
I
I
I
I
I
I -
: :0 0 0 0 0: : :
I 0 0 0 0 \ I 10 0 )( I
0
0.lL..2.. 0 00 0 0: -\ :0 + 0 0 0 0: :
xl
\
0 I ... I )( 1)(
I'" I x
6 - I.-
I
I
I
I
-
I I
~
I I
=
::t
I§ 1
..
-
ii p:; II':
I .
ci
:vi
IU
::t
'-' 4 r- rn -
.c I
a
.."
d
::t
.!
u
~
I
I
1
-; D Lo...r Imler Coil
2 - ..S o Upper buler CoIl

-
:f. + Upper OUter Coil
x Lower OUter Con
1 ,,"l..
4.311: 4.311: 1
1
3.811:
,-. 4.:ut 3.• I~
1-'
O~~-U~ __ II
L-ll--L~~~~~~~~~~L-~~~~L-~~
:1 I
I
I
o 10 20 30 40
Quench Nwnber
Figure 1. Quench history for magnet DC0201. This magnet was conditioned
by ramping three times to 6.8 kA at 3.5 K before the first quench.
I
,I
B
1
1
,
I
1+ 0 Q...JL....D 0
o~: 0
\
I 0
0 o -Cl.....D... 0 0 0 0 0 01
"-...
0
I.-
6
~...
.....
r:
::t
'-' 4
.c
.."
r: .
a
::t ...
U
u
o Lowa- iDDer Coil
o Upper Innor Coil
1 -; + Upper OUter CoIl
1
2
1
1
I
.~
t:!
I
I
104' 4.39: 4.3511: 3.1151t 3.SIC
"'I
"1
1
1
0
0 5 10 15 20 25
Quench Nwnber
Figure 2. Quench history for magnet DC0202.
444
I I
; ... : I
1 1
1
1
1
-
B - :
::13-9-£1 0::~x xxx::
I I
1
1
1 x
1
...
a x \ )( 0 0 0 0 0 1 \ 10
X
x 1lO X 0
6-: 1
I
I".X
X
-
1
1
I
I
I
I
4 - : -
I
I
I D lmrer ......r CoU
I
I o Upper ...... r CoU
1 x Lcnrw Outer cou
1
• Lcnrw Coillo
2 -
1
: I
-
I I
I 1
=:
,.; 1
3.l1li: :1
I ...
I
1
o~~~~--~-U~--~~-U~--~~~~~~~~--~~~
I I : I
o 10 20 30 40
QueDch Number
I
I
I
...
-'-
I
I·
B I
I
I
Ix
1
.,.- o 0 0 0 0 o x 0 0 0 xl X X X
X 0 X x 0 x x x
x
.. +
)( x
6 x
~
....
j;j
"
u 4
.
.<l
u D Lower lDDer COU
a o Upper Imler eon .!
..
;I
• Upper Outer CoU
~
CI U
x Lower Outer cou
2
I
I
I
~
~
I
I
~I
.1 4.3X 4• • :1.II1II:
"I
I
I
0
0 10 20 30
Quench Number
445
I
I
I I.
I
~
B I
I
I
1+
+ + + + +
X 0 X + o 0 0 + 0 0 1
I
0\0
X + + I
...
X + + X I
XI I
I I
6
:x I
I
I
I
~
... C cl I
.,
I
....~
;2 ;2'
110: 110:' I
C I
ci \)1 I
•. 1
4
(.)
..: "" I
.."C
;2
,
I
I
I
C Lower ......r CoD
o Upper IDDer CoD
~
t;
a I
I
+ Upper Out. CoD -;
I )( ......... Outer Coil ~
2 I
I
•
~
I
,
I
I
,
UI: 4.3511 4.351: 3.161(
I
I
0
0 10 20 30
Quench Number
Figure 5. Quech history for magnet DC0205.
I I: I ... I:
B -
I
I
I
I 0 ~Ol
/! -
.A.-4.-0 0
\ ...
:0
I
0 0 0 0 0
0
o~o
...
0
6 r- -
~
...
..c..
(.)
:3
4 :- -
..:
a
c .."
;2
u
i.
u
o Lowe .....er CoD
o Upper ......r CoD
2 '-- 1
~ ;
I
-
I
4.351[ 4.3111:
I
I
I
3.11111: 3.1i1:
I•
0 I I: I I
0 5 10 15 20 25
Quench Number
Figure 6. Quench history for magnet. DC0206. This magnet was conditioned
by ramping three times to 6.8 kA at 3.5 K before the nrst quench.
446
Table I. Magnet Construction Features
Cable
FG
Epoxy
Cable c(~)nt Yoke Weight (Ibs.)
Before After No. of
Iss(A) Cu:SC extra extra Laminations
Magnet Inner Outer Ic(7T) 4.35K I 0 Inner Lams. Lams. Added
DC0201 I-B031O S-2071 7851 6738 24 24 1.5: 1 11295.06 11295.06 0
DC0202 1-5264 I-B0359 7791 6800 20 24 1.5: 1 11306.20 11306.20 0
DC0203 1-5264 1-130359 7791 6800 20 24 1.5: 1 11306.76 11315.36 7
DC0204 S-2346 I-B0397 8368 6944 20 24 1.3: 1 11307.64 11315.24 7
DC0205 S-2346 I-B0359 8368 6944 20 24 1.3: 1 11309.29 11321.19 11
I-B0397
DC0206 S-2346 I-B0640 8368 6944 20 20 1.3: 1 11309.39 11314.79 5
I-B0641
NOTES:
1. 201: No shims; 202-206: 3 mil shims at poles between collar and yoke.
2. 202, 203, 205, 206: Increased end force before test.
3. 203: Asymmetry of one outer coil wedge reversed.
4. 201,204: Bridge, via RTV/weld of end tubes, between end plate and yoke.
5. This series of magnets used low carbon steel spacers between blocks rather than stainless steel spacers
as used in earlier magnets.
6. The yoke design specified a fixed weight of steel over the specified yoke length. When spaces
between yoke blocks developed because of more compact yoke blocks, extra laminatlOns were
mistakenly added in the later magnets, giving a higher YOKe density in these magnets.
Table II. Magnet Performance.
No. of
Training
Cold Azimuthal Prestress Total End Force (lbs.) Q
(J,Inner. a,Outer, I' for
1=0 1=0 a,Inner=O After Cold, Cold, Force at
Magnet (PSI) (PSI) (kA)' Welding Before' Q After Q I'=40(kA)' I 0 IcL(A) Icdiss
OC0201 4000 3800 27 1400 5600 7200 9000 0 0 6788 1.007
OC0202 3220 2970 28 3100 1600 4130 6800 1 0 >6899 > 1.015
OC0203 2340 4130 22 3690 1910 3060 5600 1 5 >6803 > 1.000
OC0204 3000 3800 25 1600 1800 3600 6000 3 5 6961 1.002
OC0206 2820 3900 25 2800 2000 2800 4000 0 8 6985 1.006
OC0206 5200 4400 31 10000 6000 7400 10700 1 0 6920 0.997
NOTES:
1. 202: ICk from the second quench. The other plateau quenches originated in tum 13.
2. 203: All plateau quenches originated in tum 15.
3. The outer coil quenching observed in the training of some of the magnets persists also in the lower
temperature quenching tests of these magnets.
4. The fourth column indicates the value of I' beyond which the inner coil prestress is zero.
For I = 6.5 kA, r = 42 (kA)2.
447
estimated from short-sample tests, Iss, is tallied in Table II. In two of the magnets, the
plateau quenches originated in turns near the pole turn as noted in Table II. For these
magnets, the data yield only a limit on the ratio Icdlss. The limit is probably 1 % - 2 %
below the actual value of the ratio. This uncertainly limits the accuracy of the comparison
to about 2 %, which is comparable to the uncertainty from other sources such as magnet
temperature and short-sample measurements. (The three magnets made with Cu:S.C. =
1.3: 1 all contain cable from the same spool; Icdlss varies 1 % in this set.) Within this 2 %
uncertainty, Icdlss shows no dependence on the Cu:S.C. ratio.
Another approach to testing for effects of the Cu:S.c. ratio is to count the number
of inner coil training quenches. This enumeration can be broken into three parts: initial
quenching, quenching after a thermal cycle, and quenching at low tempeerature. (Outer
coil quenches have been ignored for this exercise.) In the initial quench testing, one of
the three magnets in each group has no training quenches and the other two do. The total
number of training quenches for each group of three magnets (two for the magnets with
Cu:S.C. = 1.5: 1, four for the magnets with Cu:S.C. = 1.3: 1) is small and not
significantly different for the two groups. After the thermal cycle, the two groups have
the same performance (one magnet with no retraining, two magnets with one retraining
quench each). At low temperature, all three magnets with Cu:S.C. = 1.5:1 reach plateau
with little additional training as does the one magnet (DC0206) with Cu:S.C. = 1.3: 1
whose low temperature performance was not limited by the outer coil. In all, no
significant difference is found between the number of training quenches in the two groups
of magnets.
When the magnets were tested, quench origins in the outer coil were found in four
of the magents. A careful review of the construction features (Table I) and quench
performance (Table II) was made. The experiment with magnet DC0201, where outer coil
quenches were observed after the end preload was reduced, coupled with experience with
the 1.8 m-Iong, 50 mm-aperture magnets, which had improved quench performance as a
result of increasing the end preload,4 suggested that increasing the end preload might help.
Consequently, the room temperature end preload in the last magnet, DC0206, was
increased significantly. No quenches occurred in the outer coils of this magnet. A review
of the outer coil quenches and end loading of these magnets shows a simple correlation:
magnets with end preload at 4.35 K of 5600 lbs. or more do not quench in the outer coil;
those with cold end preload of 2000 lbs. or less do. 5
In addition to the warm end preload, two factors influenced the cold end preload.
First, the interface between the yoke and collars was changed due to the addition of 3 mil
shims after magnet DC0201. Second, the yoke packing factor was increased for the last
four magnets. The combination of these two changes clearly influenced the axial preload
during cooldown. The DC0201 end preload increased 5600 lbs. during cooldown, whereas
the largest increase in the later magnets was 200 lbs., and most lost significantly. The
large increase in the warm end preload of DC0206 compensated for the observed
cooldown loss in the magnets with yoke-collar shims and extra laminations.
REFERENCES
1. A.K. Ghosh, M. Garber, K.E. Robins and W.B. Sampson, "Training in Test Samples
of Superconducting Cables for Accelerator Magnets, " IEEE Trans. Magnetics 25: 1831
(1989).
2. J. Kuzminski et al., "Test Results of BNL Built 40 mm-Aperture, 17 m-Long SSC

Collider Dipole Magnets," IEEE Trans. Magnetics 28:311 (1992).
448
3. A. Devred et aI., About the Mechanics of SSC Dipole Magnet Prototypes, in: "The
Physics of Particle Accelerators," American Institute of Physics, NY (1992).
4. J.F. Muratore et al., "Construction and Test Results from 1.8 m-Long, 50 mm
Aperture SSC Model Collider Dipoles," paper to this conference (IISSC 92).
5. The outer coil quenches in DC0203 may not bear on the question of axial restraint,
since all but three were near the location of the incorrectly-installed outer coil wedge
(see Table I). The three quenches after the thermal cycle were far from this wedge.
But the last two of these occurred near the warm end of the magnet when the
temperature rise across the magnet was unusually high (100 mk versus 40 mk).
449
QUENCH PLATEAU, COn.. CURE TEMPERATURE, AND COPPER-TO-
SUPERCONDUCTOR RATIO IN RECENT 1.8 m SSC MODEL DIPOLES"
P. Wanderer

Magnet Division
SUMMARY
The ratio of the conductor-limited quench currents to the short-sample prediction,

IQ/Iss, is tabulated for a series of 1.8 m-Iong SSC model collider dipoles. Sixteen of these
magnets have 40 mm aperture, four have 50 mm aperture. For these magnets, the coil
cure temperature ranged from 135 C to 217 C and the ratio of copper-to-superconductor
ranged from 1.27 to 1.54. At 4.35 K, the value of ~/Iss was between 1.00 and 1.04 with
no clear dependence on cure temperature or Cu:S.C. ratio.
***************
This note collects data from recent short SSC R&D dipoles made at BNL. All the
magnets reached a stable quench plateau at a nominal temperature of 4.35 K, as
summarized in Table I. In all the magnets, these conductor limited quenches occurred in
the pole tum of the inner coil, as expected. The magnet plateau current is compared to
the expected value Iss, which has been calculated from the 7 T value of the critical current
Ie using G. Morgan's parameterization (AUTOIC90).
The Table also lists two construction features: the copper-to-superconductor ratio
of the cable used in the inner coil and the cure temperature of the coils during molding.
The cure temperatures are dictated by the choice of cable insulation. For the DSS magnets
and DSA207, 208 in the Table, the standard epoxy-fiberglass-Kapton scheme was used.
For the DS020n magnets, different combinations of all-Kapton insulation were used. The
polyimide adhesive used in DS0201-DS0204 required a higher cure temperature than the
epoxy used in DS0205, 206. DSA209, 210 also had polyimide adhesive.
-This work is supported by the U.S. Department of Energy

Generally speaking, the uncertainties in the ratio ~/Iss are estimated to be at the
level of 2 % . This includes uncertainties in both Ie (7 T) and ~ and possible vendor-to-
vendor differences (e.g., IGC SI-64 vs. OST SC14-2).1 Within this uncertainty, the
plateau current as a fraction of short sample does not appear to be affected by the different
cure cycles or copper-to-superconductor used in these magnets.
Table I. Quench Plateau and Construction Data for

1.8 m long, 40 mm and 50 mm Aperture SSC Dipoles
Nominal
40mm Coil
Aperture Cure Cable (or Cu:SC Ie (71) fa (A)
Magnets Temp. (C) Inner Coils Ratio (A) (4.35 K) Iq (A) lofIu
iDSSO\3 135 IOC SI-64 1.47 7741 6784 6950 1.024
055014 135 IOC SI-64 1.47 7741 6784 6855 1.010
OSSOl6 135 IOC SI-64 1.47 7741 6784 6950 1.024
OS5017 135 IOC SI-64 1.47 7741 6784 6925 1.020
055018 135 OST SC14-2 1.53 7020 6577 6790 1.032

055019 135 loST SC14-2 1.53 7020 6577 6770 1.029
055020 135 OST SC14-2 1.53 7020 6577 6810 1.035
055021 \35 OST 5C14-2 1.53 7020 6577 6820 1.037
050201 217 IOC SC13-3 1.53 7851 6803 6875 1.009

050202 217 IOC SC\3-3 1.53 7851 6803 6865 1.007
050203 217 IOC SC13-3 1.53 7851 6803 6895 1.012
050204 217 OST (5SC-I-O-II) 1.27 8603 6997 7060 1.009

050205 170 OST (SSC-I-O-II) 1.27 8603 6997 7040 1.006
050206 170 OST (SSC -1-0- 11 ) 1.27 8603 6997 7060 1.009
050213 217 OST (SSC-I-O-II) 1.27 8603 6997 7015 1.003
050214 217 OST (S5C-I-O-ll) 1.27 8603 6997 7036 1.006
ominal
50mm Coil
Aperture Cure Cable (or CU:5C 1e(71) fa (A)
Magnets Temp. (C) Inner Coil.s Ratio (A) (4.35 K) Iq (A) IqIIu
OSA207 135 SSC-3-I-ll 1.54 10827 7440 7500 1.008
05A208 135 55C-3-1-22 1.54 10827 7440 7600 1.022
05A209 217 55C-3-S-22 1.5 I 10079 1JIS 7390 1.010

05A210 217 55C-3-S-22 1.51 10079 1JIS 7405 1.012
IFor cable SC13-3, Iss is the average of measurements made on four different segments
covering 80% of the cable length which have an rms variation of 20 A. ~ data from one
end of the cable were about 1-1/2% lower and were not used in the average. The magnets
reported here used cable from the opposite end of the spool. For DS0203, Iss is for the
coil half in which the plateau quenches occurred.
452
CONCEPTUAL DESIGN OF THE SSC CRYOGENIC
TRANSFER LINES
B. Zha.ng and V. Ganni
Accelerator Systems Division

Superconducting Super Collider Laboratory •
Dallas, Texas 75237
Abstract: The sse cryogenic system require;; transfer lines for trans-
porting cryogens between above-ground refrigeration plants and below-
grotUld subcoolers. The transfer lines will be built in modules for con-
venience in fabrication, handling, and assembly. Each module of the
t.ransfer line consists of a vaclllUIl vessel which encloses a bundle of
seven cryogenic circuits at 4 K, 20 K, and 80 K The cryostat. includes
an 80 K shield and a multilayer insulation (MLI) system for reduction
of the heat leaks into the 20 K and 4 K circuits. Heat leaks through
various paths are est.imated and presented in conjlmction with the basic
system requirement.s. Discussions are provided on the design crit.eria,
constraints, and themw contraction handling techniques.
INTRODUCTION
Cryogenic transfer lines are required for the SSC cryogenic system to transport
cryogens between the above-ground refrigeration plants and the below-ground main
ring. Thi" work slUIl1l1arizes the issues pertinent to the design and perfonnance
analysis of the transfer lines. A transfer line consists of seven cryogenic circuits which
are enclosed in a single cryostat. An intemal suspension system supports the internal
components including the cryogenic circuits, a multilayer insulation (MLI) system,
and a thermal shield. Through this suspension system, the loading from the intemal
components is transmitted to the vaclllUIl vessel. The capacity requirement of the
internal suspension system is based on a 2g loading of all the internal components of
the cryostat. Three types of internal suspension systems are evaluated. Heat leaks
into the three cryogenic circuit groups (4 K, 20 K, and 80 K) through each suspension
system are estimated and compared.
The main ring of the collider is divided into ten sectors and the high energy
booster into two sectors. Each sector is an independent cryogenic unit with its own
• Operated by Universities Research Association, Inc., for the U.S. Department of Energy under Contract No.
DE-AC35-89ER40486

helitun refrigeration plant supplying helium for the 4 K and 20 K circuits. TIle
nitrogen supply is provided by one or two air separation plants for the entire collider.
TIlere is a nitrogen dewar for each sector which can be filled during HonnaI operation
by withdrawing liquid from the magnet nitrogen circuit. TIle dewars may also be filled
by nitrogen trailers to serve as the nitrogen supply source during col1ider maintenance
periods. One transfer line is required at each sector to transport helium and nitrogen
required for maintaining the operating temperatures of the magnets and their thermal
shields. TIle total length of the transfer lines for connecting the main ring to the
refrigeration plants is estimated at 1,250 m (4,100 ft). The transfer lines will be built
in two kinds of ba<;ic construction units, the standard module and the elbow piece.
TIle standard modules constitute over 90 percent of the transfer lines, and the elbow
pieces complete the bends at required locations. The module is approximately 12 m
long and weighs 1,800 kg (4,000 lb). As depicted in Figure 1, a transfer line starts
at the helium refrigeration plant building rmlning horizontally toward<; the utility
shaft. It continues vertically down along the shaft wall reaching the bottom of the
shaft, where it rtulS horizontally to the subcooler coldbox. TIle next section of the
transfer line cOlmects the coldbox to the feed/distribution box in the clyogenic alcove
located on the other side of the main ring. The ring tilt and the local topography
HELIUM REFRIGERATION SYSTEM
MAIN TUNNEL TRANSFER LINE
FEED BOX SUBCDDLER
Figure 1. Schematic layout of a cryogenic transfer line
dictates t.hat the length of the transfer line vary from sector to sector, particularly in
the vertical sections. It ranges approximately between 70 m (229 ft) and 120 III (394
ft). TIle cryogenic circuits are designed to meet the system requirements tabulated
in Table 1.
DESIGN AND ANALYSIS
TIle design requirements of the trmlSfer lines include high reliability; low heat
leak; easy fabrimtion at minimal cost; mKl convenience in hancUing, trmlsport, a<;-
sembly, and maintenance. In each standard module, 311 internal su<;pension system
provides mechanical mem)s to secure the clyogenic circuit buncUe, the thermal shield,
and the MLI system to the vacmun vessel. It is composed of one longitudinal mld five
raclial supports. TIle longitudinal support is located at one end of the module so that
the circuits cml move freely at the other end to acconunodate thermal contraction.
TIle radial supports distribute the radial loading, mld reduce deflection mld vibration
454
Table 1. System characteristics
Pressure Temperature Mass Flow Rate
DesIgn NOlIlll1.ai DesIgn Nominal
(Bar) (Bar) (K) (g/s) (g/s)
4 K helium supply 20 4 4.5 400 400
4 h. helium return 20 3.5 4.28 300 100
4 K gas helimn 10 .8 4.0 400 300
20 K helimn supply 20 3 16.5 200 200
20 K helium return 20 2 24.0 200 200
80 K liquid nitrogen 20 10 SO.O 400 80
80 K gas nitrogen 10 1.3 80.0 400 350
in the circuit tubes. The radial supports are evenly spaced at 2 III intervals, corre-
sponding to a natural frequency around 33 Hz for the circuit tubes. The external
support syst.em will provide measures to secure the entire t.ransfer line at designated
locations.
The vacumn vessel is made of 609.6 mm (24 in) schedule 10 carbon steel pipe.
Stainless steel tubes sizing from 50.8 llll11 (2 in) to 101.6 mm (4 in) are used for the
cryogenic circuits. The 80 K thermal shield will be rolled out of copper sheet. The
internal radial support consists of three disks of composite material, and the longitu-
dinal support is constructed using stainless steel plates and corrugated shells. In the
intercomlect region, the circuits are joined by stainless steel bellows to accommodate
thermal contractions.
Vacuum Vessel and Vacuum Barrier
The stress analysis on the vacuum vessel is based on the pressure range, concen-
trated stresses at both internal and external supports, and buckling criteria for the
vessel a'> a slender shell.
TIle t.ransfer line at each sector will be an independent vacuum lUnt. Vac-
uum barriers will be inst.alled at both ends of a transfer line to isolate the transfer
line vaCtllUll space from that of the comlecting subcooler coldbox in the tmlllel and
feedl distribution box at the surface. The baniers also serve as the local longitudinal
supports for the internal components. The cross section of a vacuum bani.er is shown
in Fig1ll'e 2. TIrree stainless steel plates are used to support the tlrree cryogenic circuit
groups. They are thermally set at nominal temperatmes of 4 K, 20 K, and 80 K,
and connected by two sets of concentric stainless steel shells. An additional set of
shells anchors the 80 K plate to the vacuum vessel. The operating vacuum level of
the cryostat will be maintained below 10-6 ton.
Internal Suspension and External Support Systems
TIle int.ernal suspension system supports the circuit tubes, thermal slneld, and
MLI blanket by transmitting the loading from these internal components to the vac-
Ul1111 vessel. As shown in Figure 3, a radial support consists of three circular composite
disks. Each disk supports one of the tlrree cryogClnc circuit groups. The disks al·e
separated and supported by stainless steel rods wlllch transmit the loading from the
4 K disk to the 20 K disk, and then to the 80 K disk. A set of rollers is attached to
455
\
./
Figure 2. Vacuum barrier for transfer lines
Figure 3. Disk-type suspension system
the 80 K disk in such a way that the disk can slide longitudinally within the vacumn
vessel following the thennal movement of the 80 K tubes. The 80 K circuit tubes are
secured to the 80 K disk, while the 4 K and 20 K tubes m·e allowed to move freely
within the 4 K mId 20 K disks in the longitudinal direction.
Other design options evaluated include a post-type slL<;pen<;ion system mId a
sleeve-type suspension system, which m·e shown in Figure 4. In the post-type sus-
pension design, three S-shaped cylindrical composite shells are u<;ed to support t.he
internal cOlpponents. TIley m·e integrated by two stainless steel shells. TIle sleeve-
type suspen<;ion design uses a single stainless st.eel plate at 80 K, which supports all
the circuit tubes. Each 20 K tube is secmed t.o the plate by a set of sleeves, and the
4 K tubes are attached to the same plate £1.<; a bWlclle.
The performances of the various internal suspension systems m·e compared mId
the systems are judged by their structural integrity mId heat leak resistmlce. Among
the systems evaluated, the sleeve-type design offers the best structural integrity mId
reliability. However, this design has the highest heat leak into the 4 K circuits. The
456
Figure 4. Post-type and sleeve-type suspension systems
post-type slL<;pension requires penetrations in the vacuum ve<;sel, which is an Illlde-

sirable feature for the transfer lines since it reduces the available cross-sectional area
for the tubes. Each internal suspension system is structurally designed to meet the
same loading requirement. The differences in building material and in construction
geometry lead to a variation in heat leaks into the three cryogenic circuit groups. TIle
estimated conductive heat leaks into the cryogenic circuits are set forth in Table 2.
Table 2. Heat leaks for different internal suspension systems
Support Heat Leak W)

Type 4K 20 K 80K
Post .0123 .259 2.038
Disk .0408 .62 5.95
Sleeve .4289 .4052 10.41
Vacuum Barrier .0225 .8472 10.73
Each transfer line module "vill be supported externally at two locations to dis-
tribute the loading. The external support system will provide means for position
adjtL<;tment in three orthogonal directions at the final assembly. Additional external
attachments are also needed for handling, transport, and assembly.
Interconnections
In the interconnect region where two adjacent transfer line uuits will be joiued,
each circuit. tube "vill be connected ,vith a bellows to accommodate the theruk"ll con-
t.raction. TIns is accomplished by compressing the bellows at. ambient temperat.nre t.o
a certain degree at the final assembly, such t.hat. when the circuit reaches its operating
temperature, t.he bellows will assllll1e a length which reduces the t.ensile st.ress in the
bellows. The thennal slneld and MLI blanket are bridged in the interconnect region.
The bridging slneld overlaps with the joining shields on both end<;. TIle layers of the
MLI system are preferably interwoven to produce a smooth transition. A sleeve t.ype
joint "vill be welded between the two jOllnng vacuum vessels to make up the gap after
all the internal components are properly joined and leak-tested.
457
Elbow Pieces
TIle transfer lines require 90° bends at several locations. Each bend causes a
change in the orientation of the cross section of the cryostat with respect to the center
line of the vacuum vessel. TIus orientation change does not have any impact to the
standard modules, as each of them can be rotated at the final assembly to match the
required cross section. However, it dictates that each elbow piece have a different
cross section with respect to the plane in wluch the elbow lies.
Another important issue concenung the elbow piece is its internal longitudinal
support. Since bellows are used to handle thernml contractions in the cryogenic
circuits, the tensile stress in each circuit tube at. the bend due t.o the static pressure
and hych·aulic hanmler has to be localized witlun t.he elbow piece. TIle longitudinal
support for an elbow piece is primarily sized for tIus loading.
Thermal Shield and MLI System
A single thermal slueld operat.ing at liquid nitrogen t.emperat.ure is adopt.ed to

intercept the heat leak into the heliwn circlut bundle. It is mechanically supported
by, and thennally anchored to, the 80 K circuit tubes. Its tIuckness is determined
by the buckling criteria for the horizontal sections, where the slueld bears the total
weight of the MLI system, and by the allowable temperature difference across the
slueld.
TIle thermal contact between the shield and the circuit tubes is established by
soldering (copper to stainless steel). TIle maxinllun length of each thernml shield
piece is limited by the thernml contraction differential between copper stainless steel
within the temperature range 300 K to 80 K Tllls problem C<"Ul also be resolved
by nUlIling a nitrogen trace line along the shield without attachment to the 80 K
lines. In this way, the conflict between the structural support ,md thernml contact, is
eliminated as they aJ:e uncoupled aJlCl treated separately.
TIle heat flux: intercepted by the thernlal slueld cau<;es a circwllferential temper-
ature variation in the shield. The maximlUn temperatme cliffercnce in a 1 nun thick
copper slueld is approximately 1 K. An MLI blanket of sixty layers is applied over the
80 K slueld and secured to the internal suspension system. Escape paths are provided
for the gas molecules trapped between insulation layers to ease the vacuwn-PlllllPing
effort. Each heliwn circuit tube is wrapped ,vith ten layers of MLI to further reduce
the heat leak.
Hydrodynamics
Two nmjor aspects of hydrodynamics in the tran<;fer line dcsib'1l are the total
pressme ch·op tIu·ough each circuit, aJlCl the static etnd dynamic loading on the int,er-
nal sl.L<;pension system. Considering the pressure drop, geometrical constraint, and
economics, the circuit tubes are sized aJld listed as follows:
• 4 K helium supply (76.2 lllill O.D. x 1.65 lIDll wall)
• 4 K helitml retmn (76.2 nun O.D. x 1.65 nun wall)
• 4 K helitllll gas return (101.6 Hllll O.D. x 1.65 lllill wall)
• 20 K helium supply (76.2 llllll O.D. x 1.65 mm wall)
458
• 20 K heliwn retum (101.6 mm O.D. x 1.65 lIllIl wall)
• 80 K liquid nitrogen (50.8 lIllIl O.D. x 1.24 lIllIl wall)
• 80 K vapor nitrogen (101.6 Illlll O.D. x 1.65 lIllIl wall)
TIle design pressure drop and heat leak per unit length for each circuit are presented in
Table 3. TIle last two colWD1lS in the table list the site-specific total pressure drop and
heat leak of each circuit for the N15 sector. At an elbow piece, the static pressure in a
Thble 3. Pressure drop and heat leak for the transfer lines
PressW'e Drop Heat Leak ~p at E1 qat E1

Pa/m W/m Pa W
4 K helium supply 9.28 .005 1114 1.37
4 K helium retum 4.86 .005 583 1.37
4 K gas heliwll 18.61 .005 2233 1.37
20 K helium supply 3.69 .014 443 22.3
20 K heliwll retum 17.13 .014 2059 22.3
80 K liquid nitrogen 16.77 1.079 2012 280
80 K gas mtrogen 62.52 1.079 7502 280
circuit tube exerts an intemal tensile stress in the wall of the circuit tube. TIns stress
will be transmitted along the circuit tube to the bellows in both directions unless the
loading is localized within the elbow. The intemallongitudinal support of an elbow
piece has to wit1lStand this loading to protect the adjacent bellows. TIns support will
also absorb the dynamic loading due to the momentum change of the Howing cryogen
at the elbow. On the other hand, the static loading is allowed to transmit along the
vacuum vessel, and the dynamic loading is localized and transmitted to the extemal
supports for the elbow piece.
SAFETY DEVICES AND INSTRUMENTATIONS
Each transfer line will be equipped with emergency pressure relief devices to
guard the vacuWll vessel against any accidental intemal pressure build-up. TIle pres-
sW'e relieving requirements for the cryogenic circuits are covered by relief devices for
both above-ground and below-ground cryogenic equipment. A vacuum-punlPing port
is provided at each module, and will also be used for leak detection and diagnosis.
On the prototypes, temperature sensors are to be installed on the intemal suspetlSion
system at designated locations to verify the estimated heat leaks.
SUMMARY
The requirements for the sse ayoganc transfer lines have been investigated.
TIle resulting conceptual design has beetl outlined in tlns work. Protot~}1)es of the
transfer line module and elbow piece will be built and tested prior to lTh'lSS produc-
tion. Testing of the prototypes will concaltrate on the perfonnance of the intemal
459
suspension system. TIle structmal integrity and thennal pertonnance of the suspen-
sion system will be examined and compared with the theoretical predictions.
REFERENCES
1. H.H. Bednar. Pressure Vessel Design Handbook, 2nd Ed., Van Nostrand Reinhold
Company Inc., New York (1986).
2. P.R. Smith and T.J. Van Laan. Piping and Pipe Support Systems, McGraw-Hill,
Inc., New York (1987).
460
AUTOMATING THE STRAIN GAUGE DATA ACQUISITION OF
DIPOLE MAGNETS DURING KEYING AND SKINNING
Michael C. Winters\ David Kihlken\ Richard Sims2,

Waynn Koska1,Rodger Bossert1,Mark Davidson1, Marty Whitson1,
Ethel Gonczy\ and Arthur Gilbertson 1
IFermi National Accelerator Laboratory

P.O. Box 500, MIS 343, Batavia, IL 60510
2Magnetic Systems Division, SSCL Magnet Project

P.O. Box 500, MIS 343, Batavia, IL 60510
ABSTRACT
Automated data acquisition and processing of strain gauge data provides

distinct advantages when compared to the manual system. Capturing of raw
strain gauge data automatically from the dipole magnet while it is being keyed or
skinned, allows us an immediate view of the relationship between the press
pressures and the internal stresses of the magnet. This data is analyzed then
converted to engineering units and displayed on the computer screen in real
time. By continuously monitoring the strain gauges and the press pressure,
movement within the magnet is detected and correlated with the applied force.
This data allows "striction" and shifting of the coils to be detected. Because the
data has been collected automatically, typographical errors in the data base are
not a concern. The data in this "soft form" is available to be analyzed on other
computers. The computer can control the press itself and automate the entire
process.
INTRODUCTION
In the past, we recorded the voltage and current signals of the strain gauge
manually using a HP3457A multimeter. Outputs were written down and the
multimeter was connected to the next strain gauge for information to be
recorded. Every strain gauge and every pressure applied to the magnet during
the keying or yoking operation was recorded in turn.
The trouble with this method is that it is very time consuming, prone to
operator error, subject to data entry error, and lacks the ability to give dynamic
data while force is being applied to the magnet, thus only showing the internal
stress of the magnet after it has been keyed or yoked. These shortcomings had
to be eliminated if we were to meet the projected schedule.
The development of the data acquisition system for the SSC dipole magnets
has overcome all of the restrictions listed above. The new design significantly
reduce inputting errors, decreases the time it takes to record the data, and
allows the observation of the internal pressures during the keying or yoking
process.
The system is made up of five main components:
1. Computer Hardware
The computer controls all functions of the system, records all
data, and displays that data in engineering units on the screen.
2. Software
All instructions are written in the Asyst@ data acquisition
language.
3. Instrumentation
Two HP3457A multimeters
4. Power Supply
HP6181C constant current source
5. Multiplexer
HP3488A switching unit
STRAIN GAUGE DATA ACQUISITION SYSTEM
The first decision that had to be addressed was which type of computer
(Macintosh, IBM Compatible, or a HP Work Station) to use. Macintosh is the
most common PC computer within our laboratory at this time and we were
leaning heavily towards it because of its ease of use and the perceived support
that would be available. In further discovery, we determined that the software
and hardware available would not be adequate for our needs.
Hewlett Packard has the VXI bus controlled by their work station which is
fast and flexible with the ability for easy expansion in the future. However, the
only language available is HP BASIC, otherwise known as Rocky Mountain
BASIC. As a BASIC this is one of the best for data acquisition"programing, but
it does not have the powerful commands that are found in the fourth generation:
languages. A more serious limitation at that time was the lack of a 6.5 digit
voltmeter card. The available 5.5 digit card did not have the resolution necessary
to take strain gauge measurements.
462
The remammg platform was MS-DOS. These computers offered us the
greatest amount of power for the money, the most variety of data acquisition
packages and languages, and a large selection of data hardware boards. This
particular combination would insure off-the-shelf solutions for our needs.
Because of the abundance of third party card manufacturers, we found support
to be readily available.
Because of the inherent noise problem of reading the strain gauges with a
voltmeter card or an A to D converter card inside the computer, it was decided
that external meters controlled through a GPIB (RS-488) bus would be used.
The current meter will read the strain gauge current at the beginning of the
measurement run and again at the end of the measurement run. The voltmeter
will be multiplexed through the GPIB bus from one gauge to the other for its
readings. All of these readings are sent from the meters to the computer over the
GPIB bus as ASCII data. In this way, internal computer noise will not have an
effect on the measurements.
The second decision, in parallel with the hardware, was which software
package to use. The software is the heart of a project. Once committed to the
software, you have to endure whatever shortcomings are discovered in the
future. Mter a lot of software research, Asyst© surfaced as the package of
choice. Asyst© is not a data acquisition program but a data acquisition language
in which you may write your own program. Its command vocabulary is verbose
enough to allow all of the programing to be done without having to import
algorithms from other language.
What the computer, H.P. instruments, and the Asyst language gave us was a
Strain Gauge Data Acquisition System using off-the-shelf equipment that would
be flexible, easy to use, and give us the reliability and resolution required to
measure the internal pressures of the dipole magnets. All that was now needed
to be done was to write the Strain Gauge Data Acquisition Program using the
Asyst Language.
THE STRAIN GAUGE DATA ACQUISITION PROGRAM
This program is a menu-orientated, stand-alone program capable of reading

all coil strain gauge signals. These signals are converted to engineering units,
displaying on the screen as coil PSI, and then immediately stored in a temporary
file on the hard drive. This prevents loss of data in the event of an error causing
the program to terminate.
When you begin the program the hard drive is rust examined for any
previous existing temporary data files as a safety precaution. Any existing files
must be deleted or copied before continuing. An existing data file cannot be
463
I II
Figure 1
464
overwritten without the operator's knowledge. Then the gauge's calibration data
is read from the disk file. GPIB addresses are examined to find out what type of
instrument is connected. A list of on-line equipment is generated and saved so
that the software can determine if there is enough hardware at the proper
addresses to perform the measurements. Missing or unpowered instruments are
listed for the operator.
The operator then selects from the menu the long or short magnet option.
The program expects to see two gauge packs for a long magnet and one gauge
pack for a short magnet. A third option allows the new data to be appended to
an existing data file. Before the strain gauge event data is read, the operator is
prompted to enter vertical and horizontal pressure being applied to the pack(s).
At this time, the graphic grid appears for each gauge pack. This grid shows
Pressure PSI on the 'Y' axis and Event numbers on the 'X' axis. Each grid will
display all eight strain gauges; the vertical press pressure, and any horizontal
keying pressure that may be applied. If the data is being appended to a previous
file, the previous data is displayed for the operator to review.
At the beginning of an event in which data is to be taken, the first check is to

ensure that there is a current loop of the proper mili-amperage through all the
gauges. This check is also taken between readings to warn the operator of
failure. The current value is recorded at the beginning of a reading and at the
end of a reading. A reading consists of not only the current values, but the
voltage value of each strain gauge. The computer reads the current and voltage
from their respective meters through the GPIB bus. Then the multiplexer is
commanded to connect the voltmeter to the next strain gauge and its signal is
read. The process continues until all strain gauges and their temperature
compensators have been read. Up to 80 events may be recorded and up to eleven
events for each pack can be graphically displayed at one time. The operator can
scroll the display(s) forward and backward to place any sequential group of
events on the screen.
The raw data is saved in the temporary file on the hard disk, converted to
engineering units, and displayed on the graphs. In addition to the individual
strain gauges displayed, an average of the inner and the outer gauges may be
displayed from a menu option. The display may be switched to text and shows
coil PSI numerically along with the R zeroes, present pressures, present
resistance, and change in resistance from R zero.
During any reading, data is checked for stability through a special software
routine. Ten readings of one power line cycle are taken. The high and low
readings are dropped and the mean of the remaining eight is returned. If the
standard deviation falls outside an allowed range, the operator is notified that
the reading was unstable. In this way, spurious anomalies, such as noise spikes,
and relay settling time dips, will be filtered out. Besides unstable readings, the
program also flags reversed polarities and gauges that become inoperative. Bad
gauges are identified on the screen between readings. If they are active gauges,
465
they are no longer shown on the graphic display. However, if they are
compensators, the computer will substitute a similar one to be used in its place.
Gauges are determined to be bad if their resistance value falls outside the
window of acceptability.
In addition to the voltage and current saved in the data file, each event will
also have the date, time, and any comments the operator cares to enter. Data
files also contain the magnet ID string, magnet size, horizontal and vertical
pressures, and the cold and warm calibration data for reference. At the end of
the operation the operator will save all accumulated data to a floppy disk under a
name of his choice. Writing this disk file also closes the temporary file, used for
safety, on the hard disk.
CONCLUSION
The three most important items of this system:
1. Total reduction of operator inputting error

2. Ability to dynamically see the internal stresses, while applying force
to key of yoke the magnet.
3. Speed at which the raw data can be formatted for later studying.
The data acquisition system will display the stresses of each strain gauge,
vertical and any horizontal hydraulic press pressures, on the CRT. The ability
to see the curves in real-time allows the immediate detection of any anomalies.
This will help to insure that all magnets are known to be acceptable before cold
testing.
With just a little additional coding, the system can be made to control the
magnet press at the same time. It is feasible that with the push of a button the
system could totally control the keying or yoke operation while continually
checking to insure that all perimeters of both the press and the magnet are
within a window of acceptability. The ability to see the curves in real-time
allows the immediate detection of any anomalies. This will help to insure that
all magnets are known to be acceptable before cold testing.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the contributions by the following

people: Clarence Penner, and a special thanks to Patti Winters for all the typing
and page layouts.
466
MAGNETIC MEASUREMENT INSTRUMENTATION
DEVELOPMENTS AT THE SSC
Z. R. Wolf, W. C. Li, M. W. Coles, P. F. Pellissier,

J. E. Dryer, R. Z. Fuzesy, and T. R. Gathright

Dallas, TX 75237
ABSTRACT
vVe review the developments in magnetic measurement instrumentation that

have taken place at the Superconducting Super Collider (SSC) Laboratory. The
primary developments have been stretched wire systems for superconducting dipole
and quadrupole alignment and field integral measurement, and our support of the
"mole" that is being developed by Brookhaven National Laboratory (BNL).
INTRODUCTION
Several developments in magnetic measurement instrumentation have taken place

at the SSC Laboratory. We have developed a stretched wire technique for precision
alignment of dipoles and quadrupoles. In addition, this technique should give the
integrated field strength of dipoles and the integrated gradient of quadrupoles. The
technique works at currents over the entire range from 10 A to 6500 A. We can use
the same instrument to do alignment and field integral measurements on warm and
cold magnets to get the warm-cold correlations.
We are also aiding the "mole" development effort at BNL. We have performed an
extensive computer simulation of the mole to establish its crucial design parameters.
The errors in multipole measurements were determined for various mechanical imper-
fections. We present construction tolerances that must be met to keep the errors on
the multipole measurements below 0.01 unit. We are also looking into several ways
to simplify the mole and to calibrate it.
• Operated by the Universities Research Association, Inc., for the U. S. Department of Energy under
contract number DE-AC35-89ER40486.

STRETCHED WIRE DEVELOPMENTS
Magnet fiducialization, the alignment of the magnetic field with respect to

external markings on the magnet cryostat, is a fundamental step in accelerator align-
ment. The measurement of the dipole-integrated field strength and the quadrupole-
integrated gradient are also important. The fiducialization and integrated field de-
termination for the SSC magnets are complicated by the fact that only a fraction of
the magnets will be tested under operating conditions. Most magnetic measurements
will be made at room temperature with low energization current. The magnets tested
cold must provide correlations between the measurements at room temperature with
low energization current and the measurements at the operating temperature with
full energization current. In developing instrumentation to do the fiducialization and
integrated field determination, we were looking for a very sensitive device capable of
working at the low field strengths of a warm magnet. In addition, we wanted a device
with a large dynamic range so the same device could be used to make all the mea-
surements necessary to determine the warm-cold correlations. We believe a stretched
wire system is ideally suited to our needs.
The technique we came up with involves oscillating the wire in the magnet. The
wire is oscillated at different positions and orientations. Null measurements give the
alignment. The magnet is energized with direct current. This allows us to avoid
problems with eddy currents, yet take advantage of the sensitivity of AC electronic
techniques. The technique was devised after R. Schermer's advocacy of AC techniques
for these measurements,] and after visits to DESY and Stanford Linear Accelerator
Center (SLAC) to observe their methods of magnet fiducialization. 2 ,3 In particular,
our technique is an adaptation and extension of Gerry Fischer's quadrupole centering
technique used at SLAC. 3 We have used our technique on dipole and quadrupole
magnets to date. We expect the technique could be used on higher-pole magnets. We
review the technique for dipoles and quadrupoles and discuss our initial results.
Dipole Fiducialization
The specification on the dipole roll angle measurement is that the average field
direction be given within 1 mrad with respect to fiducial markings on the cryostat. 4
Our stretched wire system measures the dipole roll angle in the test bench coordinate
system. A companion system finds the fiducial marks also in the test bench coordinate
system. The stretched wire system, therefore, can use only a fraction of the I-mrad
error budget. We wish to determine the dipole roll angle at the few tenths of a
milliradian leveL A specification on the integrated field value is that at full field it
must be within 0.3 Tm of the specified 4 value of 100.1 Tm, or within 0.3%. We thus
wish to measure it to a few parts in 10 4 • In this section we describe our stretched
wire system and its ability to meet these specifications. A more complete description
can be found in the rcferences. 5
The basic idea behind the stretched wire dipole alignment technique is illustrated
in Figure 1. If a wire is oscillated parallel to the dipole field, no voltage is induced
in the wire. If the plane of wire oscillation makes a small angle with respect to the
dipole field direction, a voltage given by
is induced in the wire. In this expression, v rms is the rms voltage (including the
relative phase) induced in the oscillating wire, Bo is the dipole field strength, Lm is
the magnetic length, w is 27r times the frequency of wire oscillation, A is the wire
468
oscillation amplitude, and 8 - 80 is the angle between the direction 8 of the wire
motion and the direction 8 0 of the dipole field. Sample values for the parameters in
this equation are Bo = 0.01 T (Imag = lOA), Lm = 15 m, W = 6.28 l/s U = 1 Hz),
and A = 3 mm. The I-Hz oscillation frequency was chosen to be well below the wire's
fundamental frequency of vibration, which is around 10 Hz. Inserting the parameter
v
values, the expected signal is rms = (2.2JN /mrad)(8 - ( 0 ).
Figure 1. Illustration of the measurement by which the dipole roll angle is determined.
After establishing the stretched wire measurement concept, we assembled equip-

ment to do a proof-of-principle test. \Ve stretched a 150-pm (6 mil) CuBe wire through
a full-length dipole. At each end of the dipole the wire was moved by an assembly
consisting of a synchronous motor turning a cam that pushed a slide connected to the
wire. The magnet was energized with 10 A. The voltage in the wire was measured
with a lock-in amplifier. We rotated the magnet and recorded the voltage as a func-
tion of magnet roll angle. Figure 2 shows our results. The voltage has the expected
rate of change with angle and the phase goes through a 180 0 change, as expected. A
fine scan of the voltage vs. angle demonstrated a measurement sensitivity of about
0.1 mrad. Our next step is to calibrate the system to bring the accuracy in line with
the sensitivity. A complete discussion of these preliminary results can be found in the
references. 6
8 300~--------------~
• •
6
> "iii
0
t • • :!!.
200 •••
•
I 4
""'
~ 2
• • ~
c..
100
0+-~-r~-~·~-,~-1
•••
·4 -2 0 2 4 -4 ·2 0 2 4
Theta (mrad) Theta (mrad)
Figure 2. Measured voltage, magnitude and phase, induced in the wire as a function of dipole roll
angle_
We envisage doing the integrated field strength measurement by oscillating the

wire perpendicular to the dipole field and measuring the voltage with a precision
voltmeter. The induced voltage in the wire is given by rms = (l/V2)B oL m wA. v
v
Unlike the null measurement for alignment, rms , W, and A in this expression must
J
be measured to 1 part in 10 4 if we are to determine Bdl = BoLm to a few parts
in 104 . We have ordered equipment with the required accuracy, but we have not
performed this measurement yet.
469
Quadrupole Fiducialization
The specifications on quadrupole alignment are more stringent than on dipole

alignment. The quadrupole center must be determined within 0.2 mm with respect
to fiducial markings on the cryostat. 7 The quadrupole roll angle must be determined
within 0.5 mrad. 7 As in the dipole measurements, the stretched wire system can
use only a fraction of this error budget. We thus need to make measurements with
accuracies a few tenths of these specifications. A specification7 on the integrated
gradient is that the nns variation between magnets be within 1 part in 103 of the
average value. We thus require our measurement accuracy to be a few parts in 104
for the integrated gradient measurement. We now describe our stretched wire system
for quadrupole measurements and its ability to meet these specifications. A complete
analysis of our measurement techniques can be found in the references. 8
Our procedure for doing the quadrupole alignment is illustrated in Figure 3. The
x-position of the quadrupole center is first determined by the method indicated in
Figure 3a. The wire is oscillated in the x-direction, and the center of oscillation is
moved through the magnet. The voltage induced in the wire is given approximately
by
7 rms c:::: y'2
Tw x(xeO - Xo ) ,
1 GL mW A'
where v;ms is the rms voltage (including the relative phase) at frequency w, G is the
quadrupole gradient, Lm is the magnetic length, w is the angular frequency at which
the wire is oscillated, Ax is the oscillation amplitude, XeO is the x-position of the center
of wire oscillation, and Xo is the x-position of the quadrupole center. This equation is a
simplification of the exact expression which can be found in the references. 8 Scanning
XeO gives a voltage null at the x-position of the quadrupole center.
tY f f
@~ @~ @~
%. a
*' *. b c
Figure 3. lIlustration of the measurements in our prototype alignment procedure by which we deter-
mine the a) x-position of the quadrupole center, b) y-position of the quadrupole center, and c) roll
angle of the quadrupole.
The y-position of the quadrupole center is found in a manner similar to the

x-position. The procedure is illustrated in Figure 3b. The wire is oscillated in the
y-direction, and the center of oscillation is moved through the magnet. The voltage
induced in the wire is given approximately by
1
Vwrms c:::: - y'2GL mwA y ( YeO - Yo - 3
2 8) .
The quantities in this expression have similar meaning to those used in the previous
expression for finding x. In this expression, however, the sag of the wire, 8, enters
and must be accounted for. We anticipate using the DESY technique of varying the
wire tension and extrapolating to infinite tension. The extrapolated position of the
voltage null gives the y-position of the quadrupole center.
470
The quadrupole roll angle is found in the manner illustrated in Figure 3c. The
wire is oscillated at angles near the 45° line. For this measurement, we use the voltage
induced at twice the frequency of the wire oscillation. This voltage is given by
V,Iwms ~ - ~GLmwA2(1jJ - ( 0 ).
In this expression, 1jJ is the angle of wire motion relative to the 45° line, and 60 is the
roll angle of the magnet. The other variables have been previously defined. There is
a voltage null at 1jJ = 60 •
We set up equipment to do a feasibility study of this technique. We were able
to measure the y-position of the quadrupole center with a sensitivity around 20 J1-m,
as shown in Figure 4. The roll angle could be determined only to around 1 mrad,
as shown in Figure 5. We have several improvements planned for the roll angle
measurement. The voltages for both measurements had the expected magnitudes
and dependencies. As for the dipole measurements, we still need to calibrate the
system to bring the accuracy in line with the sensitivity.
4 200
••••
•
;;;
3
•
i:!!.
!
f:!
u 2 • •
• . 100
~
• •
l1
•
D-
•
• •• • ••••••
o
°
·0.15 0.00 0.15 ·0.15 0.00 0.15
VeO (mm) VeO (mm)
Figure 4. Measured voltage, magnitude and phase, induced in the wire in a scan to determine the
sensitivity for finding the y-position of the quadrupole center.
600
I I I I
I
500 i ·50
400 I
I i
Di. I
I
~ 300
:!!.
·100
i I
~
200 .c
=
G
I i D-
·150
I
i i i
100
I I I
0 ·200
·6 ·4 ·2 0 4 ·6 ·4 ·2 2 6
Angle (m,ad) Angle (m,ad)
Figure 5. Measurement indicating our present sensitivity to the quadrupole roll angle. The plot shows
the magnitude and phase of the induced voltage as a function of roll angle.
We expectS to be able to measure the quadrupole-integrated gradient with the

stretched wire technique. As in the case of the dipoles, we have ordered equipment
with the required accuracy to make these measurements, but have not performed the
measurements yet.
MOLE SIMULATION STUDY
We have developed a simulation program to study magnetic multipole measure-

ments using a tangential coil with dipole bucking windings. In this study, our focus
was on the errors arising from imperfections in the rotating coil.
471
A typical coil is shown in Figure 6. The simulation program generates the voltage
vs. time data from a given input magnetic field and a set of configuration data about
the coil. The analysis program takes these data as input and calculates the magnetic
field. The input magnetic field 9 is shown in Table 1. The given values of an and
bn represent the limit on the allowed mean value plus 3 standard deviations. In
the simulation program, we can take the imperfections of the coil construction into
account. Examples are gravity sag, bowing, wire twist, etc. The program simulates
their effect on the measured harmonics. When we study one particular effect, only
those parameters related to this effect change, and others are kept constant with their
nominal values. The results are compared with an ideally built coil in order to give us
some design guidance. Table 1 shows the output magnetic field after the simulation
and analysis program with all parameters set at their nominal values. The small
discrepancies are due to the fact that the coil windings in any particular wire bundle
are not at the same radial and angular position, as shown in Figure 6.
tangent1al call tangent1al co11 wire bundle I
coil form tangent1al coil wire bundle 2
bucklngC0111
bucking C011 W1re bundle 2
aZ1muthai space
rad1al spac layers
(buckIng coil 2 Is not shown)
single W1re
Figure 6. Illustration of the rotating coil used for multipole measurements in dipole magnets.
Table 1. Comparison of the input and output multipole coefficients.
Multipole Input skew a Input normal b Output skew a Output normal b

(units) (units) (units) (units)
0 0.0 10000.0 0.0 10000.0

3.79 1.54 3.804 1.545
2 1.082 5.45 1.089 5.484
3 0.986 0.506 0.9958 0.511
4 0.17 0.74 0.1723 0.7496
5 0.166 0.076 0.1685 0.07716
6 0.043 0.08 0.04379 0.08144
7 0.04 0.04 0.04077 0.04076
8 0.0305 0.0425 0.0311 0.04335
The digital bucking technique was used in our analysis program. Usually the
voltage induced in a rotating coil is dominated by the fundamental component of the
magnetic field. In order to calculated the higher-order multi poles accurately, we need
to buck out the fundamental component. Figure 6 shows a typical configuration of
dipole-bucking coils in which one tangential coil and two bucking coils are used.
472
The induced voltage in the tangential coil can be written as lO
where N is number of turns of the tangential coil, C(n) is the magnitude of the
nth multipole field at the reference radius R, I is the length of the tangential coil, r c is
the radius of the tangential coil, w is the angular velocity, 13 is the initial offset angle,
and an is the angular orientation of the multipole. The induced voltages on the two
bucking coils are given by a similar expression. The skew and normal harmonics are
defined as
C(n) sin(na n ). b = C(n) cos(na n )
C(l)Rn-l ' n-l C(l)Rn-I'
For a detailed discussion of digital bucking and the error analysis, please see the
references. 10 ,11 In the following we discuss the effects of several imperfections in the
measurement coil.
Mole Construction Tolerances
In the process of constructing the mole, the coil may be bent. Since the coil is
rigid, the bend direction is fixed with the coil and rotates with it. We set the bowing
angle to 0° with respect to the coil symmetry axis. The values of bowing sagitta were
set to 3, 6, and 12 mils, respectively. Figure 7 illustrates the effect. Since the largest
change occurs linearly in b2 , the bowing should be less than 1.6 mils if we require l~b21
less than 0.01. From Figure 7, the relative changes of the sextupole are lJa2/a2 =
-0.098 and lJb2/b 2 = -0.11. vVe can also estimate these changes analyticallyll using
a first-order approximation, which results in ba2/a2 = -0.10 and lJb 2/b 2 = -0.10.
The two methods are consistent.
0.05
0
5 6 7 8 9
·0.05 .. delta a (3 mils)
·0.1 D delta b (3 mils)
-+- delta a (6 mils)

delta ·0.15
-<>- delta b (6 mils)
·0.2
-.- delta a (12 mils)
·0.25
-fr delta b (12 mils)
·0.3
·0.35
multlpales
Figure 7. Measured multipole error with coil bowing at zero degrees.
Some other sources of mechanical imperfections are: (1) the angular or radial
position of the coil winding may be offset parallel to the rotating axis; (2) the coil
winding may have a cone shape; and (3) the bearing of the coil may be offset from the
geometrical ccnt.er of the coil. The multipoles are sensitive to these effects, and the
simulation and analysis can be done similar to the bowing. If the required maximum
tolerance on any multi pole is less than 0.01 unit, then the coil has to satisfy all the
mechanical tolerance requirements in Table 2. All these simulation results are in good
agreement with the analytical calculations. l l
473
Table 2. Tolerance requirements for the coil.
effect physical parameter tolerance
bowing Ibowl 1.6 mils or 0.041 mm

tangential coil radius l~rTI 0.24 mils or 0.0061 mm
bucking coil radius l~rBI 0.53 mils or 0.0135 mm
tangential coil cone l~rTconel 0.48 mils or 0.012 mm
bucking coil cone l~rBconel 2.0 mils or 0.052 mm
bucking 1 angle IMBI 7.1 mrad
bearing offset l~xl,laYI 0.24 mils or 0.0061 mm
sag sag 40 mils or 0.10 mm
twist 1~q,1 40 mrad
tangential angle 1~(lBI 130 mrad
angular velocity of coil I~wl 1%
We also studied other imperfections, such as coil sagging, coil twist, and angu-
lar speed variations. These effects are either very small or can be corrected using
the measurement data. From Table 2, we see that harmonics are very sensitive to
the radius errors. Therefore, the most demanding requirements are on the radius
tolerance.
U sing the above fabrication tolerances as guide, we put together a machining
fixture that is now being used to fabricate coil forms.
CONCLUSION
We are in the process of developing a fast, accurate method of aligning magnets

and measuring the integrated field values. The fundamental ideas of the stretched
wire system appear sound. We have much work to do, however, to automate the
system and to prove its accuracy.
We have provided insight into the characteristics of the mole. The crucial design
parameters have been identified. Improvements in the mole construction are being
made.
REFERENCES
1. R. I. Schermer, "Quadrupole Alignment and Strength Determination from Flux Measurements,"
MD-TA-166, August 1990.
2. G. Knies, ''The DESY Stretched Wire System For Magnetic Measurements," to be published.
3. G. E. Fischer, J. K. Cobb, and D. R. Jensen, "Finding the Magnetic Center of a Quadrupole
To High Resolution," SLAC-TN-B9-01, March 19B9.
4. "15 Metre Collider Dipole Magnet, Magnet System Specification," MBO-000001, Revision A,
January 10, 1992.
5. Z. Wolf, "Moving Wire Technique for SSC Dipole Alignment," MD-TA-190, August 1991.
6. Z. Wolf and P. Pellissier, "Preliminary Stretched Wire Dipole Alignment Results," MTL-003,
January 1992.
7. "Collider Quadrupole Magnet System Specification," M80-000007, May 17, 1991.
8. Z. Wolf and P. Pellissier, "Moving Wire Technique for SSC Quadrupole Alignment," MD-TA-
209, December 1991.
9. "SSCL Site Specific Conceptual Design," SSCL-SR-1056, July 1990.
10. J. Herrera, BNL Magnet Test Group Note 225, July 29, 1982.
11. W.C. Li and Mark Coles, "Rotating Coil Simulation Study," MD-TA, to be published.
474
DESIGN AND USE OF A MECHANICAL GAGE FOR MEASURING
ALIGNMENT OF CRITICAL FEATURES OF STRING TEST MAGNETS
D. Wolman, G. Brown, D. Pena, R. Trekell, and 1. Woltz

Dallas, TX 75237
D. Ogilvie
Naval Warfare Assessment Center

Corona, CA 91720
ABSTRACT
Tight mechanical tolerances must be held for each magnet being fabricated for the
Accelerator Systems String Test (ASST) of the Superconducting Super Collider (SSC)
Particle Accelerator. The beam tube, single phase pipes, liquid and gaseous helium return
pipes, liquid nitrogen pipes, radiation shields, and vacuum vessel must be checked with a
tightly toleranced alignment gage to assure that they are installed in their intended locations
at the ends of each individual magnet. The proper alignment of these critical features within
each magnet will permit a series of several magnets to be successfully connected for the
ASST.
This paper documents the design and use of the mechanical alignment gage, verifies
its adequacy to perform the necessary measurements within tightly specified tolerances, and
presents recommendations for a second generation of gageS to be used with future
production run magnets.
INTRODUCTION
The principle of the gage (Figures 1 and 2) is straightforward: If the gage can
simultaneously fit over the beam tube flange and all the other pipe flanges, then the pipes
are accurately positioned (within their allowable tolerances) in the x and y coordinates
(Figure 3). If the gage cannot fit in this manner, then at least one of the pipe flanges is out of
tolerance from its assigned position. The gage is not designed for direct measurement of the
out-of-tolerance amounts, nor the angular orientations of the pipe flanges, nor their axial (z)
locations, although an estimate can be made by the operator. The gage can be used during
No. DE-AC35-89ER40486.

Figure 1. The gage being prepared to measure features on a magnet subassembly. The tubes of the gage will
be brought forward until the (white) plastic end pieces contact their assigned features, such as pipe flanges and
the end of the 20-K shield.
~ "Ro ll-
I
" " / . .¢-, "','
I ~ ,
I I
I ' \
Lh I - - -----.-----';-;-
, -WI
&
y' ,'
I ,I \ I,
~ - $ -',--W. . /
", I,
-$ .
\ -$, - -$' ,-,
' ... ......
'Yaw'
--.... -'
y
Figure 2. Closeup of the gage. Some of Figure 3. Coordinate system and terminology.
the sliding tubes and plastic end pieces have
been installed in this view.
476
magnet assembly to accurately position the ends of the coldmass, pipes, and shields in the
vacuum vessel. It can also be used as a go/no-go final inspection device. Just as the ASST
Magnets are prototypes, this gage is also a prototype. The design, fabrication, and early trial
uses of the gage focused valuable attention on the alignment of critical features of the
magnet interconnect region.
The gage is capable of measuring the positions of the single phase pipes, the LHe and
GHe pipes, the 20-K pipe, and both 80-K pipes at both ends of the magnet. It is also capable
of measuring the positions of the 20-K shield, 80-K shield, and vacuum vessel end rings.
This paper focuses on the piping alignment only, not shields and vessel alignment.
DESIGN
Drawings and Components
The structural core of the gage (Figure 2) is a pair of identical vertical plates welded to
a common baseplate. Stiffeners are welded between the two plates for rigidity. This welded
assembly sits atop a lower baseplate connected by three bolts with spherical washers, which
allows the gage to be oriented with respect to "pitch," "roll," and "yaw," as shown in Figure
3. Aluminum tubes with precisely machined ODs are inserted into the precisely machined
IDs of holes drilled in the two vertical plates. This allows the aluminum tubes to slide
smoothly and with a minimum of "play." Locations of these sliding tubes in the vertical
plates correspond to the nominal locations of the pipes at the ends of the magnets.
Drawings of component parts of the gage were toleranced in accordance with ANSI
Standard Y14.5-l982.1 The ANSI datum structure was intended to provide clarity for
fabrication and to produce well-defined inspection surfaces.
Materials
The plates and sliding tubes are 6061 aluminum alloy. All aluminum weldments were
solution heat-treated and aged to the T6 condition prior to final machining. Solution heat
treatment is needed to relieve internal stresses due to welding, thus stabilizing the structure.
The anodizing process was controlled to assure that dimensions specified on the drawings
were met after anodizing. Hard anodized aluminum was considered the best material choice
for local plate stiffness, overall weight, and minimum fabrication cost. Wear of sliding
aluminum tubes was not a major concern due to the limited number of uses (several hundred
at most) during the ASST program.
Each sliding tube was fitted with a plastic (acetal resin) end piece that was custom-
designed to fit over a corresponding pipe flange at the magnet's interconnect region. Plastic
end pieces were used on the gage to avoid scratching the pipe or vessel flanges, particularly
any polished demountable flanges planned in the early designs of the ASST Magnets.
Acetal resin polymer was chosen due to the stability of its structural, thermal, and moisture-
resistant characteristics. 2
Error Analysis of the Gage
Error analysis of the gage is based upon the accumulated tolerances of its component
parts. For this gage, primary considerations are the in-line accuracy of the IDs of the holes
in the vertical plates and any "play" between the tube ODs and the IDs of the holes. The
individual errors due to component tolerances may be treated as mutually independent and
totaled as the square root of the sum of the squares (SRSS), or may be treated as strictly
additive (worst possible case) and totaled as absolute values.
477
Accumulated errors built into the gage are shown in Table 1. Note that the size of each
error is small at its source (i.e., at the two vertical plates), but the resulting error (at the end
of the fully extended tube) is about 5 times larger. The predicted SRSS error of ± 0.5 mm
compares favorably with the actual errors measured on a coordinate measuring machine (see
below). The accuracy of ± 0.5 mm is sufficient for measurements of the ASST Magnets.
Table 1. Sources and Magnitudes of Errors in the Gage.
Error Source (Description) Max. Size of Error Max. Resulting Error

at its Source (mm) in Gage (mm)
Tube OD vs. hole ID at front plate 0.06 0.29

Tube OD vs. hole ID at rear plate 0.06 0.29
In-line tolerance of matched holes in plate 0.06 0.29
Total Accumulated Error (SRSS) 0.10 0.50
Total Accumulated Error (Abs. Value) 0.18 0.87
INSPECTION AND ACCEPTANCE OF THE GAGE
The gages were inspected on a coordinate measuring machine (CMM) to verify

compliance with drawing specifications and to gather actual measurements of the assembled
gage. A CMM of sufficient size was used to ensure accurate measuring without
repositioning the gage. Aluminum extension tubes and plastic end pieces were installed, and
measurements were taken with respect to the nominal beam tube centerline. Locations of
each of these tubes were measured in the fully extended position, just as they would be
during actual use. To measure any "play" between the installed tubes and the IDs of holes in
the vertical plates, the tubes' positions were measured first under their own deadweight,
then sequentially with a small force upward, right, and left. These configurations correspond
to the worst case position errors to be expected in actual use. After first checking the gage in
its lead end configuration, all tubes and end pieces were removed, the gage was rotated 1800
on the horizontal plane to the return end configuration, and measurements (under
deadweight only) were repeated.
In each configuration the diameter of the end of the tube was measured using four
approximately equally spaced positions. This approach provided both the diameter of the
end of the tube and a location for the center of the tube. The diameter should be constant for
all positions, and any variations from the known value would have provided an indication of
possible measurement errors. The measured diameter error was typically less than 0.03 mm,
and the X and Y positions for the tube centers can be expected to be accurately measured to
approximately this value.
Results of CMM measurements on the first gage were evaluated numerically. The
errors (deviations from nominal) for all tubes were consistent, with the maximum deviation
being 0.442 mm in the X direction and 0.533 mm in the Y direction. The standard deviations
were 0.182 mm in the X direction and 0.249 mm in the Y direction, with an approximately
normal distribution. These errors are comparable with the predicted values.
478
CONSIDERATIONS FOR "REFERENCE CENTERLINE" OF THE GAGE
The reference centerline for piping alignment3 at each end of the ASST Magnet was
located at the center of the beam tube flange. For straight ASST Magnets (no sagitta) this
line should ideally be the same as the geometric centerline of the cold mass and the
geometric centerline of the beam tube 10, and should be parallel to the geometric centerline
of the vacuum vessel. However, this ideal condition was not generally achieved, so various
interpretations of the reference line were used. One very detailed plan4 was developed for
possible use based upon a methodology 5 referencing the geometric centerline of the cold
mass. The matter of reference centerlines for piping alignment will receive further attention
during design of the production magnets for the SSC.
RESULTS OF GAGE MEASUREMENTS OF ASST MAGNETS
None of the ASST Magnets measured (to date) with the interface check gage has
proven to be fully in tolerance. Typically, two or more pipes have been outside their
allowable range. Estimates of the out-of-tolerance amounts range from 1 mm to 8 mm.
These errors are not surprising given the prototypal nature of the ASST Magnet program;
improvement of piping alignment performance is clearly expected for future (production)
magnets.
The inherent accuracy of the present gage was clearly shown by the CMM
measurements, but none of the magnet assembly areas had a dedicated, precisely positioned
inspection stand on which to place the gage. Therefore, the positioning and orientation of
the gage relied upon the judgement and care of the operator. Under these circumstances it is
difficult to repeatedly position and orient the gage. Interpretation of the results was further
dependent upon the choice of reference centerline (see above). The detailed plan4 for
positioning the gage rigidly on a precisely positioned, dedicated stand was judged to be a
reliable method of alleviating the repeatability problems; however, this plan was not
executed due to schedule restraints.
Based upon lessons learned, two gages will be installed on dedicated, precisely
positioned inspection stands adjacent to the cryostat assembly bench (lead and return ends)
at SSCL to facilitate in-process checking of piping, shield, and vessel alignment.
REVIEW OF THE MAGNET INTERFACES AND GAGES BY NWAC

Background
SSCL Magnet Systems Division (MSD) requested the Naval Warfare Assessment
Center (NWAC) to review specific MSD designs relating to ASST Magnet interfaces and
gaging. NW AC was consulted because the requirements and problems anticipated for
magnet assemblies parallel in many aspects the complexity and scope of Naval weapons
systems. NWAC was asked first to review and comment on drawings of the MSD-designed
ASST Magnet gage. Secondly, NW AC was asked to recommend methods of assuring
mechanical interchangeability of completed production magnets as they come together in
the field.
479
Review of the ASST Magnet Gage
The ASST Magnet gage design is appropriate only for the limited number of units it
will inspect. Sliding surfaces are not hardened and ground, so frequent checks of gage
dimensional integrity are recommended. Position and orientation of the gage to contact
datum surfaces of the beam tube are accomplished by means of adjustments to the gage
mounting bolts. This time-consuming procedure would not meet production phase
inspection rates, but is adequate for ASST units.
The ASST gage, being a "go/no-go" type, does not gather inspection data that could be
used to supply a Statistical Process Control (SPC) program. At the ASST stage of magnet
development, however, processes have not matured sufficiently to call for inspection data
beyond basic go/no-go information.
As shown in the error analysis above, the expected (SRSS) accumulation of individual
component part tolerances is ± 0.5 mm. However, the potential worst-case accumulation
(sum of absolute values) is nearly ± 1 mm, which is unacceptable. Thus, final assembly
requirements of the gage rely on the integrity of the gage manufacturer and the careful
surveillance of Magnet Division representatives. This risk can be avoided by placing
tolerance requirements at the assembly level of the gage.
Recommendations to Achieve Interchangeability

for Future (production) Magnets
NW AC recommends creation of a complete drawing package based upon the principle

of interchangeability, and suggests a thorough dimensional (worst case) analysis of the lead
and return end tolerances. Further, interface surfaces of major components and
subassemblies within the magnet assembly should be examined to assure mechanical and
functional interchangeability at worst case. This interface analysis should be independent of
and concurrent with drawing package development. It should be conducted under the
guidance of an SSCL interface control working group. The result of this effort would be to
build quality into the drawing package; that is, if built to the tolerance extremes allowed by
design, mating components and subassemblies would successfully assemble.
As the production phase approaches, control of manufacturing processes will become a
major concern. NWAC suggests that interface gages offer a cost-effective and technically
advanced means to derive feedback from many of the manufacturing processes. Such gages,
instrumented with transducers to collect data and linked to a computer to process the data,
provide the quality assurance organization with itnportant options. These include inspection
reports (electronic and hard copy) produced at the time of inspection, electronic storage and
retrieval of reports, and data analysis by means of Statistical Process Control (SPC)
software.
OTHER METHODS OF INSPECTING INTERFACE FEATURES
Several alternatives were considered prior to the decision to develop the gage in its
present form. One alternative considered was the use of several smaller and simpler gages to
check relative locations of different sets of components. However, this inspection method
was judged to be too labor-intensive and uneconomical for ASST inspections.
Development of a customized, dedicated coordinate-measuring machine was
considered. While technology was available to develop such an apparatus, and it would
have the advantages of quantified measurement for data collection and analysis, the cost
involved for equipment with such severely limited application prohibited this approach.
Use of an optical measuring system was also investigated. Such a system would also
have been capable of quantifying measurements. Not only was this method considered too
480
expensive, but there was considerable concern as to whether such an instrument could be
made sufficiently robust for the environment in which it must function. The same
observations also applied to considerations of laser measurement of ASST Magnet features.
One option not considered because of its state of development and the lack of detailed
information at the time was the Portable 3D Measuring Machine. Considerable progress has
been made recently in the development of these systems, and such a system is under
consideration as a possible alternative approach to end configuration gaging.
The present type of mechanical gage was selected for ASST over these other
inspection options because it was judged to be straightforward to build, test, and use. In
addition, as a prototype gage checking a prototype magnet, it has the benefit of providing
intuitive (visual) evaluation of assembly operations, even at intermediate stages of magnet
assembly.
SUMMARY
A mechanical interface check gage for use with the SSCL ASST Magnets has been
designed. built. and used for both in-process evaluation of subassemblies and for final
inspection. The gage has permitted assembly and inspection teams to visually assess the
condition of magnet piping alignment. Though not intended primarily for quantitative data
retrieval. it has nonetheless permitted estimates of individual piping misalignments. It has
focused valuable attention on the alignment of critical magnet features. and lessons learned
can be applied to production magnets.
REFERENCES
l. American National Standard "ANSI Y14.5-1982," American Society of Mechanical Engineers,

New York (1983).
2. Plastics, Allied Plastic Supply, Inc., Dallas, TX (1991).
3. SSCL Drawing No. M30-000806, Rev. B, "Magnet Interfaces, ASST' (1991).
4. W. Higinbotham, Fermilab Technical Note TS-SSC-91-241, "Magnet Alignment" (1991).
5. A. Lipski, J.A. Carson, and W.F. Robotham, "ssc Dipole Magnet Measurement and Alignment Using
Laser Technology," in Supercollider 2, M. McAshan, ed., Plenum Press, New York (1990).
481
PERFORMANCE OF THE MAGCOOL-SUBCOOLER CRYOGENIC SYSTEM
AFfER 50 MM SSC DIPOLE QUENCHES*
K.C. Wu
RUIC Project, Brookhaven National Laboratory

Upton, New York 11973-5000
ABSTRACT
The subcooler assembly installed in the MAGCOOL test area at Brookhaven

National Laboratory has been used for testing SSC magnets since 1989. As part of the
test program, magnets are quenched routinely in the system. The system description
and steady state capacity have been given previously!. In this paper, the thermal
behavior of the subcooler cryogenic system after quenches of the 50 mm SSC dipole
magnet DCA207 are presented.
Pressures, temperatures and flow rates in the magnet cooling loop are given as
a function of time. The heating/cooling and venting mechanisms are illustrated. The
cooling rates and the total energy removed by cooling during quench recovery have
been calculated for quench currents of 6000, 6883 and 7406 ampere at 4.35 K, and for
8413 ampere at 3.5 K. Results show that the total energy removed by cooling during
quench recovery is in good agreement with the stored energy released by the magnet
during the quench. No operating difficulties were encountered during the magnet
tests. System recovery time varies from 50 minutes for a 6000 ampere quench to 90
minutes for an 8413 ampere quench.
INTRODUCTION
The thermodynamic process following a magnet quench is complex. When a

superconducting magnet becomes normal, a large amount of stored energy is released
into the magnet and the helium stream in a fraction of a second. The fluid is
compressible, the phenomenon is transient and the geometry is complicated. The
process depends both on the magnet and on the cooling system. Not only it is difficult
to model the process, it is difficult to generate a well defined set of measurements for
the thermal process after a magnet quench.
The subcooler assembly installed in the MAGCOOL test area at Brookhaven

National Laboratory has been used to test successfully twelve 40 mm and two 50 mm
SSC dipoles. These tests included several hundred magnet quenches. The cryogenic
system recovers consistently after each quench. The assembly is operated by a CRISP
process control computer with a real time data acquisition system which offers a
unique opportunity to document the quench recovery process.
*Work supported by the U.S. Department of Energy.

SYSTEM DESCRIPTION
The cryogenic system for a single SSC magnet test consists of the
HEUB/MAGCOOL helium refrigerator, the MAGCOOL distribution headers, the
subcooler assembly, the feed can, the SSC magnet, and the return can. Fig. 1 shows
the flow diagram for the subcooler assembly.
An important feature of the subcooler assembly is the use of a circulating

compressor for closed loop circulation of single phase helium which obtains cooling
from the subcooler helium pot and delivers it to the magnet. The flow rate, pressures
and temperatures in the loop can be controlled independently of the refrigerator.
There is no need to change the operating configuration of either the refrigerator or
the subcooler assembly during a quench, and meaningful descriptions of the recovery
process can be obtained.
The cold vacuum pump is used to adjust the pressure inside the helium pot and
control magnet test temperature for steady state operation. During quench recovery,
the cold vacuum pump restores the liquid helium level in the subcooler pot by
establishing a flow from the pre-cooler pot. The wet expander, in Fig. 1, increases
system efficiency by generating cooling at liquid helium temperature. Valves 307 and
300 are two valves used to control the loop pressure. When the loop pressure
becomes higher than the set value during a quench, vent valve 307 opens to control
the loop pressure. The makeup valve 300 is used to make up the vented helium when
the magnet is cooled back to test temperature.
The subcooler assembly is well instrumented for purposes of computer control

and system evaluation. There are fifteen sets of dual diode temperature sensors, five
pressure transducers, two liquid level gages and one venturi flow meter. Real time
pressure in atms, flow rate in gls and speed in rpm are directly available. In Fig. 1,
the temperatures are shown without units and are in Kelvin.
The volume for the circulating loop piping is approximately 290 liters and there
is a warm line of 10 liter volume connected to the relief valve. Prior to a quench, the
magnet is maintained at test temperature via circulation of 150 gls of supercritical
helium. There are 40,000 grams of helium in the loop, and it takes four and half
minutes for the helium to travel through the loop. The amount of liquid helium stored
in each of the subcooler and the precooler helium pots is 110 liters and each amount
provides a cooling reserve of 240 Kilo-joules.
1.2.61.
From atm 5.41 '"B'" R.t
M~~~~~)~.~----~--~--~3-0-1--~ 4.~ 1..25atm 4.47
Cold 61rpm (16.0)
) WET EXPANDER ~~5~.5~9~~;;~1f~--~ Vae Qu.neh
4.1. Return
S.tpt - 60 rpm 77rpm 4.61. pre- 4.46 .950
W.t e::xp r- 40y' atm
07
) COLD VAC PUMP • 300
4.13
Setpt - 60 rpm
30.6 161. 5 .r-'----~~L.-r:;;.;:I~::::::::~±-_:4r.-;1;-5
'"H20 9/S
) CIRC COMPRESS 03
Cire 1. 63Z5
S.tpt/rpm 19000 Comp rpm 4. 666atm 4.356
Fig. 1 Flow diagram for the subcooler assembly
484
PRESSURE, TEMPERATURE AND FWW RATE AFfER QUENCH
The following results were obtained from quench of SSC OCA207 dipole
installed in MAGCOOL test stand A The magnet is maintained at 4.35 K and the
quench current is 7406 amperes.
The loop pressure measured at the discharge of the circulating compressor by

a strain gauge type transducer is given in Figs. 2 and 3 as a function of time. After
the quench, the loop pressure increases quickly from 5 atm to about 8 atm in one
second at a rate of 3 atm per second. Fortunately this initial pressure rise rate does
not persist as the pressure increases further to 17 atm in about 20 seconds at an
average pressure rise rate of 0.6 atm per second. It should be noted that 0.6 atm per
second pressure rise is still very fast and 20 seconds is a very short time in cryogenic
operations.
SSC DCA207 7406 AMP QUENCH 11/20/91
2ar-------------------------------~
aT------,----~~----~----~----~
D :Z:3 ..
ELAPSED T I ME FFO.1 QUENCH - MIN
Fig. 2 Loop pressure after magnet quench

When loop pressure exceeds the set pressure of 16 atm, valve 307 opens to vent
excess helium reducing the loop pressure to about 11.5 atm. The heating process
continues with the loop pressure increasing at a slower rate and the venting repeats.
Fig. 2 shows the first 5 min. of the pressure history with the vent valve opening twice
during the first minute. After the second venting, the loop pressure first increases but
soon decreases as the magnet is cooled back down. In about 15 minutes, the loop
pressure falls below 5 atm and the makeup valve 300 opens to maintain the helium in
the loop at 5 atm.
2ar-----------~~~~~~~~~~~
w' •
~
III
III
W
a:
Q. •
a~.----~~.---~,.~--~,~.----~.D~--~.~----~~
ELAPSED T I ME FFO.1 QUENCH - MIN
Fig. 3 Loop pressure after magnet quench
485
Temperatures recorded by calibrated diode sensors located at the supply and
return lines of the subcooler assembly are given in Fig. 4. As can be seen, the
temperatures respond more slowly than the pressure because it takes time for the heat
to be transferred to the helium and carried to the subcooler by the flow.
The highest temperature recorded is 22.5 K occurring approximately 3.5

minutes after the quench. After the peak temperature is recorded, the return
temperature decreases quickly to about 11 K in 3 minutes and then decreases at a
slower rate. When the loop pressure reaches 5 atm 15 minutes after the quench, the
supply and the return temperatures are 6.1 and 6.9 K respectively. Perturbations in
temperatures near the subcooler helium pot occur due to the opening of the makeup
valve.
The amount of helium vented can be calculated from the density difference
between the test condition before the quench and the condition when loop pressure
first returns to 5 atm. In the above case, approximately 92 liters of single phase
helium was vented from the loop and the same amount was needed to make up the
density difference when the loop was cooled to the operating temperature.
SSC OCA207 7406 AMP QUENCH 11/20/91

2:5 c SUPPLY
x RETURN
20 x x
15
W
0:
=>
I-
G;'0
w
0..
::::E
w
I-
5
o+-----~--~----~----~----,_--~
o 5 10 15 20 25 3[]
ELAPSED TIME FROM QUENCH - MIN
Fig. 4 Supply and return temperatures of the subcooler
In addition to pressures and temperatures, the flow rate through the loop must
be known in order to evaluate the amount of cooling transferred during the recovery
process. Using a venturi flow meter installed on the subcooler assembly, the mass flow
rate maybe evaluated from the measured differential pressure and the density
calculated from the measured pressure and temperature. During quench recovery,
helium in the loop remains in single phase. With the circulating compressor
maintained at constant speed, flow measurements are valid except during venting and
make up.
The helium flow through the magnet as a function of time is given in Fig. 5.
The mass flow rate after a quench first increases due to an increase in loop pressure
but soon decreases as the return temperature increases. After the peak temperature
is reached, the mass flow slowly increases to its original value. The momentary drop
in flow occurring 15 minutes after the quench is caused by the opening of the makeup
valve. Perturbations become smaller afterwards. The flow rate is a good indicator of
the recovery in the circulating loop and the magnet. As can be see from Fig. 5,
recovery of the magnet loop takes about 30 minutes.
486
SSC DCA207 7406 AMP QUENCH 11/20191
200
.. 0
III
~ •
110
0~a----',----~,a----~,,----~.a~--~"----~3D
ELAPSED T I ME FRCN QUENCH - MIN
Fig. 5 Mass flow rate after quench
COOLING RATES
The cooling rate of the subcooler after a magnet quench is of great interest
because the quench recovery is nothing more than a cooling process after the magnetic
stored energy is released into the helium system. However, the exact cooling rate is
difficult to calculate due to the transient nature of and the venting involved in the
process. The system which started at test conditions is eventually cooled to the
original conditions. The enthalpy flux difference between the supply and the return
lines can be considered as the apparent cooling rate applied to the magnet. The net
cooling for quench recovery equals the apparent cooling rate minus the background
heat load. The total amount of cooling for quench recovery equals the integration of
net cooling rates.
Apparent cooling rates to the magnet during quench recovery are given in Fig.
6. The apparent cooling rate peaks at 4500 watts 2 minutes after the quench and then
decreases to about 800 watts in 5 minutes. There is an artificially wide variation in
the cooling rate due to the opening of the makeup valve 15 minutes after the quench.
The average cooling rate over the first 30 minute period is approximately 900 watts.
~ ~ ~ ~
ELAPSEO T I ME FRCM QUENCH - MIN
" oa
Fig.6 Apparent cooling rate during quench recovery
The integrated totals of net cooling as functions of time for the magnet are
given in Fig. 7. As can be seen from Fig. 7, the cooling provided increases with time
and is equal to 800, 1300 and 1500 kilo-joules at 10, 20 and 30 minutes after the
quench occurred. At the end of the 60 minute quench recovery period, 1600 kilo-
487
joules of cooling have been provided with an estimated background heat load of 320
kilo-joules. Because the magnet system is fully recovered at the end of 60 minutes,
the 1600 kilo-joule represents the amount of heat released from the magnet. The
magnetic stored energy calculated from the inductance and the current, 1/2 U 2, equals
2235 kilo-joules2 • Thus the 1600 kilo-joule result as obtained from the integration of
enthalpy flux equals approximately 72 per cent of the magnetic stored energy.
2000
"'J
""
11500
!
!z 1000
~ z m ~ ~ ~
Fig. 7 Integrated totals of net cooling
RESULTS AT OTHER CURRENTS
Results for quench currents of 6000, 6883 and 8413 amperes have been
investigated. In all cases, the characteristics of pressure, temperature and flow rate
are similar to those given in Figs. 2 to 5.
Higher current quenches corresponds to larger stored energy releases. As

shown in Fig. 8, the return temperatures as a function of time for 6000, 6683 and 7406
ampere quenches clearly suggest a higher peak temperature for a higher current. The
8413 ampere quench initiated at 3.5 K gives a peak temperature of 18.7 K which is
lower than that of the 7406 ampere quench at 4.35 K.
SSC DCA207
2 5 . -________________________ NOVEMBER, ~ 1991
_____ ,
c soao AMP
)( 8BB~ AMP
v 7106 AMP
• 9113 AMP
D±-----,,----~----~~----~----~
o 3 6 !iii 12 1:5
Fig. 8 Return temperatures for four currents
The pressure rise rate is fastest for the 8413 ampere quench, followed by the
7406 ampere quench, as can be seen from Fig. 9. For the 6000 and 6683 ampere
quenches, the difference in rise rates is not as obvious. The first opening of valve 307
488
after the 6000 ampere quench occurs sooner after the quench than the first opening
after the 6683 ampere quench. The second opening of 307 after the 6000 ampere
quench occurs later after the quench than the second opening after the 6683 ampere
quench. This suggests that even through higher energy releases in the magnet
eventually translate to higher pressure rises, the process may depends on the details
of the quench occurrence in addition to the quench current.
SSC DCA207 NOVEMBER, 1991

20 -e- 6000 AMP
--
____ 6683 AMP
""""'i'- 7406 AMP
1S
~~'t-::;_illi~ 8413 AMP
w 10
IT
:::l
(f)
(f)
W
IT
0. "
O+-------~--------~--------~------~
o .5 1 1.5 2
Fig. 9 Pressure history for four currents
The cooling rate and the total cooling provided are also found to be greater for
higher quench currents. Fig. 10 shows the total cooling provided during quench
recovery for different currents.
SSC DCA207 NOVEMBER, 1991

2000 0 6000 AMP
x 6683 AMP
-:l v 7406 AMP

'>L ~ 8413 AMP
I 1500
---'
«
0:
I!)
W
f-
Z1000
~
---'
8U 500
I-
w
Z
10 20 30 40 50 60
Fig 10. Cooling provided for four currents
The results for four quench currents are summarized in Table 1. The initial
pressure rise rate used here is based on the pressure rise rate for the first five seconds
after a quench. The recovery time is obtained when both the magnet temperature and
the liquid levels in the helium pots are restored. The cooling integral and the
corresponding magnetic stored energy are also given in Table 1. The ratio of the
cooling integral to the magnetic stored energy varies from 87% at 6000 amperes to
64% at 8413 amperes.
489
Table 1. Results for four quench currents
Test Quench Init. Peak Max. Total Y2LI2 Ratio Helium Recovery
Temp current pressure return Cooling Cooling Vented Time
rise rate temp. Rate
K ampere atm/s K kw kj kj % liter min.
4.35 6000 0.95 16 4.0 1280 1467 87 92 48
4.35 6683 1.02 19 4.3 1450 1820 80 113 52
4.35 7406 1.03 22.5 4.6 1600 2235 72 128 65
3.5 8413 1.53 18.7 4.9 1850 2884 64 133 86
DISCUSSION
The results given in this paper are only valid for a 50 mm sse dipole magnet
installed on MAGCOOL test stand A. The effect of the quench origin and details of
quench phenomena are not considered. The effect due to the pressure control venting
process is also neglected in the present study. If the venting process could be
accounted for, the energy obtained from the enthalpy flux should agree better with the
magnetic stored energy. The circulating loop is connected to a 1.5 inch relief line of
10 liter volume. It is believed this volume helps to slow down the pressure rise rate,
but quantitative results are not available.
CONCLUSION
Quantitative results for the response of the MAGCOOL subcooler system after
a magnet quench have been presented. The cooling/heating and venting mechanisms
have been analyzed. The response time is of the order of seconds for pressures and
longer for temperatures. Conventional instrumentation appears to be appropriate for
data acquisition and process control. No operating difficulties were encountered with
the subcooler assembly or any part of the cryogenic system during a magnet quench.
Because the amount of helium vented outside the circulating loop is minimized by
presetting the quench vent valve 307 to a high pressure value and because cooling is
continuously provided to the magnet, operating test conditions are restored rapidly
after a magnet quench. The subcooler assembly is capable of cooling the magnet back
to test conditions in about 30 minutes and the liquid level in the helium pots are
restored in one to one and half hour.
ACKNOWLEDGEMENT
The author would like to thank D. Zantopp for developing data acquisition
software, A. Prodell for providing numerous helpful comments and suggestions, and
S. Agnetti for preparing this paper.
REFERENCES
1. K. C. Wu, D. P. Brown, J. H. Sonde ricker, Y. Farah, D. Zantopp and A.

Nicoletti,"SUBCOOLER ASSEMBLY FOR SSC SINGLE MAGNET TEST
PROGRAM", to be published in "Advances in Cryogenic Engineering", Vol. 37, 1992.
2. Private communication with A. Prodell, the inductance for DCA207 has been
measured by P. Radusewicz to be 0.0815 Henry.
490
A STATISTICAL RATIONALE FOR ESTABLISHING PROCESS
QUALITY CONTROL LIMITS USING FIXED SAMPLE SIZE,
FOR CRITICAL CURRENT VERIFICATION OF SSC
SUPERCONDUCTING WIRE
D. A. Pollock, G. Brown, D. W. Capone II, D. Christopherson,

J. M. Seuntjens, and J. Woltz
Superconducting Super Collider Laboratory *

Dallas, Texas 75237
INTRODUCTION
Quality verification of a design characteristic of a manufactured product is typically

achieved through inspection or sampling. In the case of superconducting wire critical
current, the product of one billet is presently required to be verified by testing one sample
from each piece of wire generated by the supplier's process. In some cases, the number of
tests per billet is very large, (i.e.>40). It is expected that once the Phase I part of the
Vendor Qualification Program is complete, fewer samples per billet will be necessary, both
by choice and because of anticipated improvements in manufacturing methods. As a tool
for monitoring process uniformity, current SSCL plans call for applying statistical limits to
each supplier's process, after a sufficient number of billets have been produced to establish
process expectations. This paper addresses two questions regarding the issue of sampling
frequency and process control. First, how many wire piece measurements from one billet
(or production unit 1) are sufficient for verifying billet performance to specification?
Second, how far can the measured average and range for a billet be from the established
process center and standard deviation before it is considered non-representative of the
stable process expectations?
The purpose of this paper is to demonstrate a statistical method for verifying
superconducting wire process stability as represented by Ic. The paper does not propose
changing the Ic testing frequency for wire during Phase I of the present Vendor
Qualification Program. The actual statistical limits demonstrated for one supplier's data are
not expected to be suitable for all suppliers. However, the method used to develop the
limits and the potential for improved process through their use, may be applied equally.
Implementing the demonstrated method implies that the current practice of testing all
pieces of wire from each billet, for the purpose of detecting manufacturing process errors
(i.e. missing a heat-treatment cycle for a part of the billet, etc.) can be replaced by other less
costly process control measures. As used in this paper process control limits for critical
current are quantitative indicators of the source manufacturing process uniformity. The
limits serve as alarms indicating the need for manufacturing process investigation.
* Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under
Contract No. DE-AC35-89ER40486.
Upstream sequences in the process are not part of this discussion, however they are
equally in need of direct on-line control, to the extent that they can be shown to
influence product perfonnance.
SAMPLING THEORY AND STATISTICAL SIGNIFICANCE
Two parameters are typically used to describe the center and the magnitude of
variability in the distribution of a continuous variable, the mean (Il, simple average) and the
standard deviation (cr, a measure of distribution dispersion).2 When the observations of
interest conform to a known probability distribution (i.e. Normal, Exponential, etc.), a
significant amount of infonnation may be inferred from Il and cr. Given an approximately
normal distribution, the likelihood (probability) that a given observation from the
distribution will fall within a particular range of the mean may be predicted, as a function of
Il and cr, and is typically stated as a proportion of the area under the normal curve with a
range of 0 to 1. 3
In their formal sense, 11 and cr are based on full or complete information from the source
population. When a collection of observations consists of a partial accounting, the
collection is a sample. The corresponding sample distribution parameters, mean and
standard deviation are identified XBAR (Ilx) and S (crx). 4 Any number of Ie measurements
from one billet is a sample. Practical sample size is limited to the number of pieces
generated from each billet. Economic sample size is based on: 1.) the cost of sampling; 2.)
the properties of the process distribution; and 3.) the level of security and sensitivity to
change that is necessary.
The Shewhart XBAR R Control Chart
Statistical Quality Control (SQC) methods have been developed to test the perfonnance
of samples against their expected population parameters. The methods discused in this
paper were developed by Walter Shewhart in the 1920's, from procedures for confidence
interval estimation and statistical hypothesis testing. 5 In the generalized method, one
compares a sample parameter (typically average and range, or standard deviation) for a
given number of observations (a subgroup) to the expected population value. If the
population parameters are unknown, they must first be estimated by means of an evaluation
of recent data from a stable process. Given the sample is a true representative of the
population, the parameter may be expected to fall within a certain distance of the population
center line. The allowable distance from the center is based on the level of control desired,
the subgroup size and the value of the population mean and standard deviation. If the
sample parameter falls outside the defined control limits, the method indicates that the
product represented by the sample is likely from an alternate population, (i.e. a statistically
significant assignable cause change has occurred). 6
One of the basic statistical tools for monitoring a process, called the XBAR R chart, will
be demonstrated. The application of this method involves taking samples from a process at
fixed intervals. The sample is made up of a fixed number of measurements called a
subgroup. The subgroup average (Ilx) and the range (R) are calculated and compared to the
established control limits for the sample average and the Range. For the critical current
case, the logical sampling interval is each billet (or production unit) and the subgroup is the
number of measurements performed to verify the product. Assuming the manufacturer
produces billets as single production units, the subgroup average is a measure of between
billet variation, and R is a measure df within billet variation.
492
Factors in Determining Acceptable Control Limits
The XBAR R procedure defines limits for the variability of sample averages and ranges
based on the level of natural variation in the process and choices made when designing the
chart. The key choices are the level of risk one is willing to tolerate (designated (l and ~
probabilities), and the subgroup size (n) selected for each sample. The probability that a
given sample average will be inside the selected control limits is defined by 1 - (l. As (l is
increased for a fixed n, the control limits move closer to the center line, and the tolerance
for allowed process drift is reduced, but the probability that a sample average will be within
the limits is also reduced, (i.e. more false alarms may be expected to occur). ~ is a measure
of the sensitivity of the control limits to systematic process drift, it is the probability that a
sample mean will be within the control limits when in fact it is really a representative of an
alternate process mean and should be rejected by the model. As n is increased for a fixed (l,
the control limits for sample average are narrowed while the ~ probability is reduced.
Larger subgroup size improves the overall performance of the XBAR R model, but with
the negative effect of increased sampling cost. Balancing sampling risk against cost is the
challenge for those implementing process controls.
THE REFERENCE DATASET
The reference data set for this demonstration is a sample of a supplier's process,
consisting of one critical current measurement per wire piece, as submitted by the supplier
to the SSCL for acceptance for use in SSC 50 mm CDM Inner (Type 1.5) Cable. The data
represents 5 billets of Inner wire, in 217 pieces as summarized in Table 1, Summary of
Sample Statistics by Piece.
Table 1. Summary of Sample Statistics by Piece

IeA@7T Cu:sc Ie for Cu:SC > 1.47
Number of measurements: 217 217 204
Minimum: 332 1.37 332
Maximum: 400 1.62 395
Range: 68 0.25 63
Mean: 364.9 1.54 364.6
Standard Dev: 7.35 0.038 6.91
C.V. (stdev/mean): 0.0201 0.02467 0.0190
By graphical analysis (Figure 1. Ie Distribution), Ie for the reference data set is

approximately normally distributed, and also exhibits a good degree of uniformity between
billets (Figure 2. Ie by Billet). Cu:SC data is also normal, but has a long left tail below
1.47. This is most likely due to cropping effects. The effect of the number of pieces in the
left tail has a negligible influence on the Ie distribution (see Table 1, Ie for Cu:SC > 1.47).
When applying the methods discussed in this paper, statistical theory recommends that
samples from 25 or more billets be measured to establish population expectations. For the
purpose of demonstration however, the reference data set of 5 billets is used as the
population estimate.
493
405 r----r-,---,---,--...-----,
o
60 385 --------------------.------ ------,..-----

~
<
= 0.20
~
~
0.25
50
40
E=<
r-
@) 365
11-0
0.15 ~
Z 30
....
0
~
0.10 20 345
~
0 0.05 10
11-0 o
0 325~~-~-~~~~-~
~
11-0 330 344 358 372 386 400 ABC D E
Ie FOR ALL PIECES. (A @ 7T. n=217) BILLET
Figure 1. Ie Distribution. Figure 2. Ie by Billet
A frequency histogram of the source A Box Plot ofIe variation
data Ie distribution. within and between billets.
PRELIMINARY XBAR CONTROL LIMIT CALCULATION
The following example describes the general method for calculating control limits when
~ and a are known. The text references should be consulted for a fuller treatment of the
method for calculating limits where ~ and a are estimated from process data. The example
calculates ±3a limits for XBAR. 7,8
Process Control limits = [~± (Z a./2 ) * ( a/--Jn)] ; for ~ and 0' known.
Z = [(x -~) fa], the "standard nonnal deviate". Z a/2 = the value of Z to the right or left of
~which delimits the area under the nonnal curve at CJ/2. For 3cr control limits Z CJ/2 = 3.
Given that ~ = 364.9 A @ 7T , a = 7.35 A, n = 30, and 0. = 0.0027:
~ Upper Control Limit = 368.93 A @ 7T

~ Lower Control Limit = 360.87 A @ 7T
These limits mean that if the true process mean is ~ and the standard deviation is a, a
sample of 30 randomly selected wire measurements from the process is expected to have an
average which falls between 360.87 A and 368.9 A, 99.73 times out of 100.
The Effect of 0. and n on the Control Limits
As mentioned earlier, 0. risk and subgroup size (n) have an effect on the width of the
calculated control limits. A demonstration of their effect is presented in Figures 3 and 4.
Figure 3, Subgroup Size Effect, plots XBAR upper control limits as a function of n, using
the general formula and the expected process ~ and a, for subgroups n = 1,2,3 ... 30.
Figure 4. Effect of 0. on Control Limits shows how control limits change with 0. risk, for a
< 0.40. Both Figure 3 and 4 use ~ = 364.9 and a = 7.35. 1- 0. is a measure of the
probability that a given XBAR will be inside the designated control limits. In figure 4,
sigma limits for XBAR are labeled for frequently used 0. levels. Typically 20' (0. = 0.0455)
or 3a (0. = 0.0027) control limits are selected for manufacturing process control purposes,
since they provide a good level of confidence without expecting a large number of out-of-
limit alarms (i.e. 1 to 5 per 100 samples). For the balance of this paper 3a limits (0. =
0.0027) will be used.
494
ICRAND SIMULATION
To simulate random sampling from the wire data for various subgroup sizes a simple
BASIC program ICRAND 9 was written. The program generates randomly selected
(uniformly distributed) data sets of reference spool numbers. The sample space for each
data set is the maximum number of wire pieces available for the source billet (or reference
data set). The program samples with replacement, and tests for repeating spool numbers in
each subgroup, to ensure (n) independent observations per subgroup. The output of
ICRAND is an ASCII file with four items per line [user defined subgroup size (n);
sample (i = 1 to 100); n (j = 1 to n); spool (k = (Random (1 to spool maximum
number»]. The random sample plan is able to be imported into dBASE III where a view
may be constructed with the Ie data and matched to a billet spool reference number. Several
data sets were created in this way for evaluation, each containing 100 samples of subgroup
size (n). The use of this program is not a requirement for implementing the XBAR R
procedure, it is merely a tool for demonstrating the method.
390
385 369
...:I ...:I
u
;> 380 g 368 2
~ ~
-< 375
=
><
~ 367
><
370 366
365 365 L......L........L~~~~......L..........-...J

0 5 10 15 20 25 30 0.0 0.1 0.2 0.3 0.4
SUBGROUP SIZE (n) ALPHA PROBABILITY (n=30)
Figure 3. Subgroup Size Effect Figure 4. Effect of IX on Control Limits

A plot of the XBAR Upper Control Limit A plot of the XBAR UCL for various IX
(DCL) for various subgroup sizes (n). levels with n =30.
Using SYSTAT QCSTAT® software 10 and ICRAND random sample data sets, Upper
and Lower Control limits (UCL, LCL) were calculated for the sample average and range
for subgroup sizes = 30, 10, 6, 3, 2. To improve the accuracy of the simulation, it was
repeated three times for each subgroup with different starting random seeds. As stated
above, the population values for mean and standard deviation are assumed to be Il =
364.9 A @ 7T and 0' = 7.35 A. The simulation for 100 billets shows the influence of
per billet sample size (n) on the 30' limits. As (n) is reduced the accuracy of the estimate is
also reduced, therefore the control limits expand (as a function of INn). Figure 5,
Simulation Control Limits, shows the estimated population average and calculated control
limits for the sample average, for various subgroup sizes for each run of the simulation.
Figure 6, Estimate of 0', shows the influence on subgroup size on the estimate of the
population standard deviation. As (n) is reduced the estimate becomes more uncertain.
495
...
P 390
®
~ 380
e
;:;:: 370
:4
....l
0 360
'"
!-o
~
C) 350 c MEAN
'<"
III
~ 340
v LCL
A ueL
0 5 10
SAMPLE SUBGROUP SIZE (n) SAMPLE SUBGROUP SIZE (n)
Figure S. Simulation Control Limits Figure 6. Estimate of u

Simulation results for the estimate of Il, Simulation results for the estimate of u,
and UCL/LCL for various subgroup sizes (n). * for various subgroup sizes (n).*
* Based on 100 random samples of subgroup size n, selected from the reference data set
SENSITIVITY OF THE LIMITS TO PROCESS DRIFT
A rationale for selecting the acceptable subgroup size should include a consideration of
the sensitivity of the limits to systematic process change (or drift). If the process mean
shifts up or down, what is the probability the change will be detected by the XBAR model?
The anSwer again depends On Il, cr, n, a, as well as a concept from hypothesis testing
known as the power of the test, denoted 1 - ~. 11 Table 2, Possible Decisions in
Hypothesis Testing, describes the choices that may be made when applying statistical
significance tests. 12 HO is the null hypothesis (i.e. Il = 365) and HA is the alternate
hypothesis (i.e. Il ¢ 365). When a statistical decision is made, it is subject to a and ~ error
probabilities, also only one state can be a true representative of the process. The power
concept provides a means of quantifying ~ error probabilities.
Table 2. Possible Decisions in Hypothesis Testing

STATE~ TRUE Situation TRUE Situation
(Ho is Correct) (HA is Correct)
DECISION ,J..
ACCEPT Ho Correct Decision Incorrect Decision
P(Ho)=(1-a) P( IJ ) ; Type II Error
REJECT Ho Incorrect Decision Correct Decision

P( a ) ; Type I Error P(HA)= (1-1J)
The power of the test (1 -~) against HA at the a significance level is the probability that a
normal deviate (Z) lies in one of the two regions: Z < ZI or Z > Z2. 13
1- ~ = P(HA) = [P(Z < Zl) + P( Z > Z2)] where:
ZI= (- ..In/cr)*{lJ.A -1lQ) - Za and Z2= (- ..In/cr)*{lJ.A -1lQ) + Za
For Ho: Il = 365, n = 30, cr = 7.564, a = 0.0027, the probability that Ho will be rejected
(i.e. the sample average will fall outside the 365 ± 3cr limits) if HA: the actual or True
process mean drifts 1.4% to 360 is 0.80. For the reference data set, to have a probability
better than 0.80 of detecting a change in the process mean, with a one billet sample of 6
496
measurements, the process mean would have to shift by more than ± 3.3%, (± 12 A). For
a sample subgroup size of 3 (as currently defined in the SSC wire specification 14) the
process center would have to drift by ± 4.4%, (± 16 A) for I-Ii = 0.80. An Operating
Characteristic curve is useful for visualizing Ii risk. 15 See Figure 7, OC Curve (n = 30 and
n =6, IX = 0.0027), in which Ii probabilities are shown as a function of alternate means.
1.0
f!
Si 0.8
::3
Z
;: 0.6
i-<
;
..: 0.4
<
Ill:!
><
it 0.2
I-a 1=
30
-- a = 6
~
III
Ill:!
0.0
350 355 360 365 370 375 380
TRUE PROCESS MEAN
Figure 7. OC Curve (n = 30 and n =6, IX = 0.0027)

The Operating Characteristic curve plots Ii against possible alternate means, where
J3 is the probability that an alternate mean will be accepted by the J.1O control limits.
Testing the Control Limits
To demonstrate the sensitivity of the limits to a change in Jl, a sample data set was made
by adding 3% to the Ic's from a source billet. The artificial billet data was then tested
against the population estimate. As may be expected, for a process with the VCL 2.5%
above the process center, the majority of the observations from the artificial data set were
above the upper limit, (see Figure 8, Effect of +3% Process Shift). From the OC Curve for
the process with an alternate mean of 376 and n = 6, it is expected that 25.5% of the J.1x'S
will be within the J.1O = 365 limits. The simulation had 4 (16%) sample averages with the
limits. So, using only 25 samples of subgroup size 6, the simulation confirms the model
expectations. Had this data been from a real billet (with only 6 short sample measurements
randomly selected from the finished wire), there is a relatively strong probability ( I-Ii ~
0.745) that the process shift of 3% would have been detected.
390
. . ..
I-<
380
. .
~ D XBAR
.. .
t- veL - 374
@)
370
-
<
<) CENTER ~ 365
360
LCL - 356
SAMPLE (n=6)
Figure 8. Effect of + 3% Process Shift
497
Figure 8 shows both the individual observations (filled box) and the }lx'S (open box
with connecting line), to demonstrate the sensitivity of the XBAR model to a 3% shift in
the mean. In practice the individuals would not be plotted. Note that all of the llx's are
at or above the upper control limit, indicating a significant change has occurred.
CONCLUSION
This work has demonstrated the statistical concepts behind the XBAR R method for
determining sample limits to verify billet Ie performance and process uniformity. Using a
preliminary population estimate for J.1 and (J from a stable production lot of only 5 billets,
we have shown that reasonable sensitivity to systematic process drift and random within
billet variation may be achieved, by using per billet subgroup sizes of moderate
proportions. The effects of subgroup size (n) and sampling risk (a and 13) on the calculated
control limits have been shown to be important factors that need to be carefully considered
when selecting an actual number of measurements to be used per billet, for each supplier
process. Given the present method of testing in which individual wire samples are ramped
to Ie only once, with measurement uncertainty due to repeatability and reproducibility
(typically> 1.4%) 16, large subgroups (i.e. >30 per billet) appear to be unnecessary,
except as an inspection tool to confirm wire process history for each spool. The
introduction of the XBAR R method or a similar Statistical Quality Control procedure is
recommend for use in the superconducting wire production program, particularly when the
program transitions from requiring tests for all pieces of wire to sampling each production
unit.
REFERENCES
1. SSCL document, M35-000014: NbTi Superconducting Wire For SSC Dipole

Magnets (1.3 Grade Inner), paragraph 3.3.1 and 4.4.2.
2. D. C. Montgomery, "Introduction To Statistical Quality Control", John Wiley &
Sons, New York, 1991, pp. 34 ff.
3. E. L. Grant and R. S. Leavenworth, "Statistical Quality Control", McGraw-Hill,
New York, 1988, pp. 50 ff.
4. ibid. Montgomery, p. 60.
7. ibid. Montgomery, pp. 203 ff.
8. ibid. Grant, pp. 78 ff.
9. D. Pollock and G. Brown, ICRAND.BAS program.
10. H. Stenson, "QCSTAT version 1.1: A Supplementary Module for SYSTAT",
SYSTAT, Inc., Evanston, IL, 1990, pp. 37 ff.
11. M. Ben-Horim and H. Levy, "Statistics: Decisions And Applications In Business
And Economics", Random House, New York, 1984, p. 444.
12. ibid, p. 424.
13. G. W. Snedecor and W. G. Cochran, "Statistical Methods", Iowa State University
Press, Ames, Iowa, 1980, p. 69.
14. ibid. SSCL document, M35-000014, paragraph 4.4.2.
15. J. M. Juran and F. M. Gryna. "Juran's Quality Control Handbook", McGraw-Hill,
New York, 1988, pp. 24.10 ff.
16. M.J. Erdmann, D. A. Pollock, and D.W. Capone II, "Quantification of Systematic
Error in Ie Testing", Supercollider 3, Plenum Press, New York, 1991, pp. 713 ff.
498
QUALITY SYSTEM DESIGN AND DEVELOPMENT FOR SSC
SUPERCONDUCTOR CABLE
M. Fuse, S. Saito, T. Sashida, M. Koganeya,

N. Ayai, K. Takai, and M. Nagata
SSC Project Team

Sumitomo Electric Industries, Ltd.
1-1-3, Shimaya, Konohana-ku, Osaka, 554 Japan
INTRODUCTION
In order to develop a quality system for the SSC superconductor cable, Sumitomo
Electric followed a very formal, systematic approach, comparable to those employed for such
major projects as developing a new car model.
This paper presents our approach, as well as the resulting computerized environment
for quality assurance (QA).
DEVELOPMENT PROCESS OF THE QA SYSTEM
Figure 1 illustrates the entire process of the QA system development. It starts from the
quality requirements definition, that is, what quality characteristics are required by the
customer, how they are measured, and what are the acceptable levels. A few items are added
from Sumitomo's own experiences.
QATABLE
The second step is the preparation of a QA table as shown in Figure 2. There are
numerous process factors that mayor may not affect the product quality. To achieve the
required quality, one cannot miss any important factors, while irrelevant factors must be
systematically removed to minimize control or inspection costs. The QA table has more than
300 possible process factors exhibited as rows. Its nearly 50 columns are the quality
requirements prepared at the first step. Each column is given its weight of importance, Qj,
from 1, minimum, to 10, crucial. Then each process factor's possible effect on each quality
requirement, Rij, is evaluated, and expressed by the score from 0, no effect, to 5, strong.
The evaluation was made by a team of three experienced process engineers.
The importance of process factor i is, then, obtained by aggregating the sum of Qj and
Rij over all j. The column of quality weight shows this value. A similar calculation is made
for cost factors including yield and repair costs, and the results are shown in the column of .
cost weight.

Quality Requirement
,
Definition of quality required
by the customer
QA Table
Figure 2
Evaluation of possible process

factors to be controlled
~ Figure 3
QC Process Charts
,
Determination of what factors
to be controlled and how at
each process step
Figure 4, 5
Process FMEA
Develop preventive measures
for possible failures and errors
at each process step
I Figure 6
I
Material
Standards TRACE
I Inspection Traceability &

Standards Reliability Assuring
Computerized Environment
~factUring
Standards
Figure I. Entire process of the Q.A. system development
QC PROCESS CHART
Based on this systematic enumeration and evaluation of process factors, a QC Process

Chart was prepared as shown in Figure 3. It provides the overall design of quality control
procedures to be conducted at each production process step. All the possible process factors
with the quality weight of 100 and above, those with cost weight, 250 and above, or those
required by the customer are included. The chan describes for the each process factor how
frequent it is checked, by whom, on what data sheet it is recorded, and to whom it is reponed
in case of nonconformance. This gives a grand design of the QA system to be developed.
PROCESS FMEA
Before the detailed development of the QA system, one more step is taken, that is,
process FMEA (Failure Mode and Effects Analysis) as shown in Figure 4. It sheds light on
possible failures that may occur during the process. Each failure mode is fIrst checked if it
can be perfectly detected within the process. If not, additional measures are considered to
assure 100 % detection. Then all the possible causes are enumerated for each failure mode,
and from its frequency (times/yr) and failure cost ($/time), its signifIcance is evaluated in
monetary terms. This quantitative analysis is our original extension to the standard FMEA.
The Pareto analysis of failure costs is shown in Figure 5. One can see that the top four
causes, control errors, foreign materials, operation errors, and equipment/tools worn out, are
not purely technical, but rather management problems, and, therefore, the effective QA
system once developed can also be a powerful cost saving measure. For each presumed
cause, preventive measures are taken. Among them, a computer environment to support
operators to prevent control and operation errors is one of the major approaches we adopted.
500
,--- QA ReqUIrement (Qj) Cost factors (Cj)
~
Quality Requirement
(QA Table)
5 265 250
5 5 261 500
~~ Size 333 57 0
~ .~ Nb ~C~o~rn~~~PO~!sl~·t~io~n~__-+~~3~1-r3~__r-;--+~r-t-_______ 5 249 250
C) i ~S~u~r~f~ace~________f--r~I~I-rl~__r-;--+~r-t-______ _____~S-+-5"-t I 227 500
...o S
~
Size 511
~M~ec;ha~n~ica~I~So.t~re=n=~7.h~~r-~1~~Ii--t~--t--r-t------
o
!:: ~ Cu ~c~o~rn~~~Pco~S=it=io=n=====t~~t=~=l==t:Jt:±=Jt3:±~::~~~;::;::::~~~~ill1r--,J
Il'urf,,""
I':: l.size. 1 S trangth of relations (Rij)
8__ ~ [Wash
E: Billet acid wash
\.Enreilm materials
ISllrr,,~..
5:strong 3:medium I :weak
"I eight for QA Requirement= 2: j(Qj X Rij)
IJ
'" oS 2i Nb sheet bending Foreign materials _
~ 110:. 3 [weight for Cost factors= 2:j(Cj X Rij)
~~t======j======~~~~~~~~==========~~
- I I I I I I I I I I
Figure 2. QA Table
( QC Process Chart)
Process Flow Report to,
!Standard riority Sampling pata Sheet Performed
Chart Items in case of
No No. Frequency No. by
nonconformance
Billet temperature
9.EB Welding 9 12912 1SUMace Cemperature 0 Thru each billet ~SCOOOO6 Subcon Production Engineer
Beam acceleration 0 Start of each billet Subcon trechnical
voltage 0 Start of each billet Subcon ~echnical
Output amperage 0 Each billet Subcon Production Engineer
Welding time 0 End of each billet Subcon Production Engineer
Vacuum in billet
--cvacuum in chamber 0 Each billet Subcon Production Engineer
Vacuum time Lo. Each billet Subcon Production Engineer
Each billet Subcon Production Engineer
*
Priority: Subcon Production Engineer
Production Engineer
IO.Extruding 10 12955 :Cu stomer Required ©
©:Qj;;;:;200
O:200;;;:;Qj;;;:;100 or Cj;;;:;250
Lo.:!OO;;;:;Qj and 250>Cj
*
Figure 3. QC Process Chart
501
( Process FMEA )
Detection Measures for Failure Preventive

Process Failure degree Presumed Frequency Irotal failure
100% cost measures
step mode (%) detection causes ($/time) (times/yr.) cost ($/yr.) for causes
Materials Wrong 80 Inspection of Wrong tag at 2 32,000 64,000 Barcode check

receiving materials cross section subcontract at
subcontract
Computer
--
Drawing Wire break 100 Detector for Dies worn out 0.88 2,000 1,760
wire breakage control
of die life 100-
Figure 4. Process FMEA
l00r---------------------------~-
90
80
70
60 Wrong materials (13.3%)
Failure Wrong manufacturing parameters (10.4%)
cost (%) 50 Wrong tools / dies (5.9%)
Wrong size measurement (1.8%)
40
30
1---""',.
20
10
0 - ----
Control Operation Others
errors Foreign errors Equipment
materials /Tools
worn out
Figure 5. Pareto analysis of failure costs
502
STANDARDS AND WORKER QU ALIFICA TION
Based on the deliberate, systematic analysis as stated above, detailed production and
inspection operations are designed and specified in three kinds of standards, as seen in
Figure 1. They are :
* Manufacturing standard, that specify production operations including both automatic
as well as manual operations.
* Inspection standards, for inspection and other QA procedures, and,
* Materials standards, for purchased materials.
Operators must pass a qualification examination by process engineers on whether they
have sufficient knowledge and skills required by these standards before they actually engage
in the production of SSC cables. The QA staff also regularly audits their adherence to these
standards.
COMPUTER ENVIRONMENT FOR QA
Figure 6 illustrates a computer environment for QA now being developed. Although it

covers all the aspects of manufacturing, from technical through production planning and
control, its major emphasis is on QA. From our experience of computer application to QA for
ABS (Antilock Brake System), backbone LANs, and other products requiring 100 % quality
and reliability, we developed a system called TRACE, i.e., Traceability and Reliability
Assuring Computerized Environment, which was awarded the 1990 Ishikawa award from
the Japanese Union of Scientists and Engineers, one of the most authoritative awards for
CIM applications in Japan. The most advanced, and extended version of TRACE is now
being developed for the SSC cable production.
•• STRUCTKlN •• - HISTORY "

Proceu I He.. T_I No 1890211 1 DRAWING I
T_ I .oo'tl Sot. I I 5c>Hd I
Hour 1 6OH] wet¥i I I Surlo<e 1
...... .......
..... .... .....
TAG. (I)
~
U9102 --
f-o
go
o ~o l --
goo ---
Floor center
~I workstation
J
TRACE for Superconductor Department .• Traceability & Reliability Assuring Computer Environment won the
Ishikawa Prize from JUSE in 1990.
Figure 6. Computer environment for QA
503
CONCLUSION
Quality is the product of both workers' QC circle type efforts and engineers' systematic
planning. Missing either of them leads to unsatisfactory results for the customer. At an early
stage of product and process development, however, engineering approaches must be
emphasized as shown in this paper. We hope to have an opportunity to present our workers'
efforts being extended now for the SSC cables in the future.
504
ALTERNATIVE ALIGNMENT APPLICATIONS FOR THE COLLIDER
DIPOLE MAGNET: CAPABILITY INDICES
Leonard L. Neiman
General Dynamics, Space Systems Division

PO Box 85990, MZ Cl-7430
San Diego, Ca 92138
ABSTRACT
In order to ensure magnetic field quality requirements for the Superconducting Super
Collider (SSC), highly accurate measurements will need to be performed at various points of
manufacture, test and installation. This paper will compare precision optics, coordinate
measurement analysis and laser tracking approaches as they apply to collider dipole magnet
alignment. Following a brief discussion of the dimensional tolerance allocations for
component structures, measurement system analysis will be utilized to provide a correlation
matrix indicating strengths and weaknesses of the various approaches. Special consideration
will be given to the issues surrounding beam tube alignment.
INTRODUCTION
Collider Dipole Magnet (CDM) alignment has proven to be a controversial issue. Included in
the challenge to align the magnetic field to an external reference is the opportunity to carefully
evaluate various techniques of verification. This paper will address three approaches to
collider dipole magnet alignment and will encompass objectives and techniques as they relate
to analysis and demonstration of the alignment task. Accuracy and precision will be used as
a foundational basis for comparison of these methods. Included in this discussion are
means of approaching required tolerance budgets as they pertain to manufacture. In addition,
other constraints and issues involved in the production of dipole magnets will be treated as
they pertain to impacts on the required alignment objectives.
ALIGNMENT OBJECTIVES
As of the date of this writing, the collider dipole magnet must be aligned to the
following requirements: First, the vertical angle of the dipole field must be perpendicular to
the Y-Z plane which contains the sagitta, or elliptical curvature, within ± 2.5 milliradians.
This requirement is further defined such that this tolerance must include the error of the
vertical plane created by the external fiducials (targets on the cryostat vessel) to a maximum
of 1 milliradian.
Supercollider 4. Edited by J. Nonte

Second, the coldmass design orbit (CDO) is designated as a smooth line which is
defined as the arc of a circle with a radius of 10,187 metres. The magnetic center line
(dermed by the centroid of the annulus created by the inner diameter of the collared coils) is
plotted by points to be taken no more than 1 metre apart along the Z (longitudinal) axis and
cannot deviate from the CDO by more than 3 millimetres at any measured point. This
tolerance is further restricted by a 2 millimetre RMS requirement which is to include a 1
millimetre fiducialization error.l
Dipole magnet alignment begins with the proper allocation of tolerances based on the
following considerations.
Budget and Allocation
By dermition, budgets for the coldmass design orbit are
2 mm RMS = ...j(x2+y2) (1)
By addition in quadrature where x equals the fiducialization error and all other tolerances are
in y and if x is allowed the full fiducial error of 1 millimetre then
2 mm RMS = ...j(12+y2) (2a)
Y = 1.732 mm RMS (2b)
This call be considered the true position tolerance of any point used in the measurement
of the magnetic center line. This 1.732 millimetre RMS tolerance must include all systematic
and random errors in the alignment process excluding those consumed in fiducialization.
Alignment budget allocation considerations needed to facilitate the best fit of structural
analysis and process capabilities are given in Table 1.
Table 1. Budget allocation for collider dipole magnet alignment.
COMPONENT TOLERANCE TOLERANCE RMS
Coldmass .508mm .258064mm

Cryostat 1.346mm 1.811716 mm
Alignment Equip. .200mm .04()()()() mm
"../2.109780 mm
TOTAL 1.452508 mm
SAFETY MARGIN IS 16.1 %
CONVENTIONAL TECHNIQUE
Once the distribution of tolerances is made, dipole angle alignment is the first
component of the solution to alignment requirements. It was envisioned to use the beam
tube geometric center as the equivalent of the magnetic center for the purposes of gathering
vertical plane information. Once the integrated field has been established through the use of
a field angle probe, the coldmass will be rolled to meet the ± 2.5 milliradian requirement. It
is calculated as ± .44 millimetres of arc length at the outer diameter of the helium vessel
skin. Once verticality is attained, the coldmass is tack welded at the .number 3 post position.
The next phase of the alignment task consists of adjusting the coldmass vertically and
horizontally such that all points measured along the Z axis are within the ± 1.732 millimetre
RMS coldmass design orbit tolerance. These measurements are made after the sagitta is
induced to the coldmass assembly within the cryostat.
506
LINES OF SIGHT
The frrst approach toward verification of alignment requirements to specification that
will be treated herein is the "Lines of Sight" method. This technique uses parallel lines of
sight, called bucking lines, to establish two known target paths with a known offset. This
method lends itself to collider dipole magnet alignment in that the beam tube position must
be known to a plane created by a given set of fiducials. The foundation of this method is in
the set-up of the parallel lines using conventional optics such as jig-transits or optical
telescopes. These bucking lines mayor may not be at a common height depending on the
vertical offset of the magnetic center line to the cryostat fiducials.
Once the bucking lines are set up, the magnet is then placed into a pre-determined
position so that any targeting device such as stretched wire, tooling balls, back-lit targets,
etc. will fall within the range of motion allowed by the micrometer adjustment of the optical
device. These targets will traverse longitudinally down the beam tube and sightings are
made to verify vertical and lateral displacements on a point to point basis. The assumption
in this technique is the coincidence of magnetic center and beam tube geometric center.
The next step is to take shots of the cryostat fiducials in the same manner. Lateral and
vertical readings are then used to establish the reference plane to which the position of the
coldmass is understood. An example of this technique is given in the alignment of the
DSHIP 40 millimeter dipole which was checked in two passes at about 11 metres.
Although the sample is statistically insignificant, the data does show fair repeatability on a
point to point basis. Translation accuracies for collider dipole magnet alignment are yet to
be established using this technique, but the expectation of error is .1 millimetres to .2
millimetres at 16 metres focal length based on normal sighting technique. The available data
is shown in Figure 1 and Figure 2.
M FROM TEST 1 TO 2
0.2
fIl 0
~
CII:
-0.2
~ -0.4
..
~
~
-0.6
-0.8
..
...J
...J
~
- 1
-1 .2
-1 .4
-1.6
10 10 10 10 10
v
10 10
co ,....
10 10 10
en
10 10
.,.... 10 10 10
v
10 10
co ,....
10
0
C')
~
C\I C')
,.... 10
C') en 10
DO
,.... 0
0
C\I
co
C')
C\I DO
10
V 0 co
~ ~
0 0 N C\I C') C') ~ 10 10 CD .....: ,.... DO DO ai 0 0
METRES
Figure 1. The point-to point repeatability in the lateral "X" plane of DSHIP taken in Dallas, TX.
COORDINATE MEASUREMENT SYSTEMS

The. use of coordinate measurement systems in the application of collider dipole
magnet ali~nment is a much more difficult problem than the lines of sight approach. In this
method, tn-Iateration is performed through the concurrent use of multiple optical fixtures.
The actual position of the coldmass design orbit can only be indirectly ascertained by
sighting fiducials such as tooling baIls located on the coldmass exterior. An exception to
this is the practice currently employed at Brookhaven National Laboratory wherein the actual
internal component structure is measured directly through access ports cut into the helium
vessel, an option undesirable in light of ASME requirements.
507
!J.y FROM TEST 1 TO 2
0.2
0
rI:l
~ -0.2
ell:
!- -0.4
--
~
..J
..J
:1!1
-0.6
:1!1 -0.8
- 1
-1.2
-1.4
-1.6
It) It) It) It) It) It) It) It) It) It) It) It) It) It) It) It) It) It)
0 T"" C\I ('I) "It
.....
It) <0 ..... co en
..... 0 T"" C\I ('I) "It It) <0 .....
('I) en II? ('I) en It)
~ 0 <0 C\Ico ~ 0 <0
0 0 C\I C\I ('I) ('I) "It It) It) <0 ..... ..... co co en 0 0
METRES
Figure 2. The point-to point repeatability in the vertical "Y" plane of DSHIP taken in Dallas, TX.2
COORDINATE MEASUREMENT SYSTEMS -

VERTICAL PRECISION
0.14
rI:l 0.12 ~--~~----------------------
ell: - - 0 - - ST.DEV.
~ 0.1
--
!-
~ 0.08
:1!1 0.06 ORANGE
..J
..J 0.04
o 3 SIGMA
:1!1 0.02
0
2 3 4 5
TARGET NUMBER (x3M)
Figure 3. The targeting errors associated with theodolite.
508
Multiple shots of the targets are made along with the use of three or four control
targets spread over the field of view. The coordinate measurement system then uses three
or four rixtures to lock in on the coldmass targets as well as the other points which are
used as control or pass points. The triangles which are formed by the geometries of these
instruments may now be moved to acquire new targets on the coldmass while maintaining
the original control points. This procedure is repeated along the length of the coldmass
for insertion into the cryostat.
A five target, three headed theodolite experiment was conducted in San Diego in
Fegruary of 1992. Five targets were placed in a horizontal plane at a spacing of 3.35
metres, left to right, in order to simulate over 16.75 metres of collider dipole magnet
length. Four optical sightings were made of each target and the results are given in Fig. 3.
As a result of this experiment, it is believed that the worst case of vertical alignment
in this method is approximately ±.125 millimetres at 16.75 metres. The vertical plane is
the most sensitive to error since results from experimentation3 show that lateral and
longitudinal errors equal ±.075 millimetres at 16.75 metres. Furthermore, optical
equipment manufacturers realize that errors in the X-Z plane are much more detrimental
to the overall accuracies and therefore manufacture to induce the smallest standard of
error in that plane.
LASER TRACKING
A third approach to collider dipole magnet alignment is laser tracking. A laser tracker
is a single headed optical device that uses a laser source mounted on precision gimbals that
are respondent to an optical encoder. Through the use of reflected laser light back to the
encoder, actuators micro adjust the source to form a parallel set of 5 millimetre helium-neon
laser beams. In this method, the vertical dipole field is frrst aligned to gravity as in the other
methods. This is where the similarity in alignment technique ends. The collider dipole
magnet will be re-wired at this point such that polarity in the lower coils is reversed and the
resultant quadrupole field will be used to ascertain the magnetic center of the dipole magnets
through what is termed the "skew quadrupole configuration."4 Through the use of magnetic
test equipment (mole) and a laser corner cube target attached to the coil drive end, magnetic
and geometric center data can be acquired concurrently.
The laser tracker establishes a 3 dimensional coordinate system within which the
coldmass design orbit can be ascertained quite readily. An advantage to this approach is that
the coldmass design orbit can be directly translated to the external fiducials located on the
cryostat vessel. In this method of alignment, beam tube support components need not be as
tightly toleranced as the true magnetic center is being determined as opposed to assuming the
beam tube geometric center is identical to the magnetic center. Beam tube position
accuracies are only important in as much as the allowable deviation of beam tube center fully
encompasses the good field area measured by the mole.
A sample gage repeatability and reproducibility study (Gage R&R) was performed in
San Diego on February 20 through February 25, 1992 and is shown in Figure 4. This
study was comprised of 3 operators tracking a series of 10 dynamic target measurements in
9 trials utilizing Chesapeak Laser Systems equipment. The gage repeatability and
reproducibility study shows a gage R & R of .06949 and a %R & R of 4.01 %. Typical
tolerance analysis indicates a 10% R & R for most conventional optical measurement
systems. An important point to be made is the total appraiser variation value of less than 1%
which is due to the automatic target acquisition feature of the tracker thereby virtually
eliminating operator error. In the final measurement system analysis of the laser tracker, it
was determined through statistical methodology that summation of measurement errors
would be well within prerequisite tolerance allocations.
Lateral and vertical angle coordinates are presented in Figure 5 for the purposes of
depicting point to point repeatability in the mapping of a simulated beam tube. Results from
this test indicate that the repeatability of the lateral component averages about 25 microns
and the vertical component averages 34 microns out to a length of 15.22 metres.
Very small lateral displacements were made with a micrometer translation stage at a
distance of 18.72 metres and the data is shown in Figure 6 as a test of angular resolution.
As can be seen from the graph, resolution is within 72 microns with an average dispersion
of 38 microns. This is especially important in the determination of very small cosine errors
at the far end of the beam tube.
509
Finally, angular repeatability of coordinates from 2 tracker locations was tested
through the shooting of 3 targets in 3 trials and then moving the tracker approximately 1
metre and re-shooting the same coordinates. The average vertical displacements from point
to point were .11 millimetres. as ~e worst case in the vertical plane and ~~ data r:o~ this
test is shown in figure 7. Thls Ylelds a Cpk value of 1.44 to the prereqUlSIte 1 mIllimetre
fiducialization error.
MEASUREMENT SYSTEM REPEATABILITY AND REPRODUCIBILITY REPORT

Dynamic Laser Tracker
Part No. & Name: 38.10 mm Spherical Mounted Retroflector Measuring Instrumen!: CMS-3000 Dare:2/1fJfn
Characteristic: X, Y,ZCoordinates of Targets X" Yt,Z, Meas.lnstrumen! No.: #913808 Performed By·
Larry Brown (Olesapeake)
Scott Schick (Olesapeake)
Len Neiman (GDSS)
Sp<cifICation: 1.732 mm RMS
FromDataSheet: R = .019253, XDiff= .005969
MEASIlREMENT IlNIT ANAl YSIS " WI BpANeR AN&I YSis
REPEATABR..rIY· BlUIPMENTVAlUATlON (EV.) %E.V. = 100 [(E.V.) I (TOLERANCE)]
E.V. =(R) (lId,) (6) = 100 [(.068272) I (1.732)]

= (.019253) (.591) (6) = 3.941 %
= .068272
RE1'RODUCmlUTY - AWRAlSER VARIATION (A. v.)
A.V. = --./ [(X wt) (lId,) (6)]2 - [(E.V.)' / (nr)] % A.V. = 100 [(A.V.) I (TOLERANCE)]
where n = number of partS
r =number of trials = 100[(.012831) I (1.732)]
= --./ [(.00569) (524) (6)]2 - [(.068275)'1 (30)] = .741%
= .012831
R6PEATABn.nY ANOREPRCIlUClBIUTY (R .t.R)
R & R = --./ (E.V.)2 + (A.V.)' % R & R = --./ (%E.V.)2 + (%A.V.)2
= --./(.068275)' + (.012831)2 = --./ (3.941)' + (.741)2

= .06949 =4.01 %
Figure 4. This measurement system analysis was perfonned to establish the operator, equipment and
tolerance error percentile for a laser tracker used in a simulation beam tube experiment
CONCLUSION
In summary, the lines of sight approach provides fair overall repeatability and resolution but
is subject to operator sighting errors and set-up inaccuracies. The systematic errors may, of
course, be eliminated through an understanding of their causes. Random errors, on the
other hand, may only be resolved through redundancy but the band-width for improvement
against random effects is commonly understood as
1/...Jn (3)
where n is equal to the number of observations. That is to say that the improvement
gain on r~dom errors is slow until the number of observations is steadily increased. Since
systematic and random errors are inherent to all optical work, a problem is posed for a
technique that is as labor intensive as lines of sight. It is assumed that automatic data
acquisition may be applied in all methods.
510
Coordinate measurement systems, on the other hand, are highly sensitive to errors
longitudinally and in combination with the need to transform control points at every vertex,
this technique will not support rate production requirements of 10 magnets a day.
Coordinate measurement systems also require the need for ancillary optical equipment for
beam tube position and are subject to systematic and random effects as are all methods.
The laser tracker affords collider dipole magnet alignment with exceptional accuracy,
repeatability and resolution while at the same time providing benefits in automatic targeting
and significant labor hours reduction. Perhaps the most attractive benefit of the tracker is
the establishment of a three dimensional coordinate system that will allow for dynamic
measurements which could be used for coldmass construction, sagitta inductance and
cryostat insertion. The versatility of the tracker provides solutions for other magnet
manufacturing problems such as tool installation and interconnect alignment verification as
welL
~ - ~:fl~--------------------------------------------------_.
iJ~ ~
-------------------------------
c -1:1+--r~----+-~~----~4--r-4----~~-+----~~-+~
0
---------------------------------
(Y) (Y) ,....II) co 0)
0
0
C\I
(0
C\I
C\I M (0
.,....
0) C\I
C\I
II)
C\I
C\I
C\I
~ ~ ~ ~ ~ ,....
0 C\I (Y) II) (0 a) 0) 0 .,.... N (Y) II)
~
.,.... ~
Z Length (M)
6 X1 ---O--Y1 0 X2 c Y2
Figure 5. This lateral and vertical deflection data depicts the point to point mapping and repeatability of a
simulated beam tube at equal distances of about 1 metre.
>-
-
::: 0.50
> 0.40
0.30
en 0.20
c
en E 0.00
0.10 ------------- ---------------------------
..
CI)
E - 0.10
~ -0.20
~ -0.30
CI
c -0.40
c( -0.50
0 co (0 (Y) C\I (0 co (0 ~ 0
(0
.,.... ~ 0 0) co co C\I (0
~ co ~ co 0)
~
(0
C\I M (Y) ~ II)
Micrometer Stage Deflection (mm)

Figure 6. Angular resolution at a distance of 18.72 metres.
511
TRACKER MOVEMENT ACCURACIES
-
>- 0.3
I-
.J
0.2
1-- 0.1
~e
C!)t.J
1lI:;J 0
<=
1-0 -0.1
Ill:
a.. -0.2
~
Ill:
-0.3
z x y
TARGET AXIS
--6I1!1r--- Target #1
ft -....,IJu-- Target #2 --vo-- Target #3
Figure 7. Angular repeatability with a tracker positional movement of 1 metre.
ACKNOWLEDGEMENTS
Research Association in support of the Superconducting Super Collider project for the U.S.
REFERENCES
1 SSCL Spec, "15 Metre collider Dipole Magnet, Magnet System Specification," Dallas, (1992)
2 B. Viola, Raw Data Submitted from DSHIP Alignment Tests, Dallas, (1991)
3 J. Wurdeman, Raw Data Submitted from Theodolite experiments, San Diego, (1992)
4 A. K. Jain and P. Wanderer, "Test of G. D. Centering in DS0204," Magnet Division Note, Brookhaven
National Laboratory, Upton, (1991)
512
APPLICATION OF A B&W DEVELOPED COMPUTER AIDED PICfORIAL
PROCESS PLANNING SYSTEM TO CQMS FOR MANUFACfURING PROCESS
CONTROL
Donald C. Johanson and Jeremy E. VandeBogart
Babcock & Wilcox

Accelerator and Magnet Systems
P.O. Box 785
Lynchburg, Virginia 24505
ABSTRACf
Babcock & Wilcox (B&W) will utilize its internally developed Computer Aided
Pictorial Process Planning or CAPPP (pronounced "cap cubed ") system to create a paperless
manufacturing environment for the Collider Quadrupole Magnets (CQM). The CAPPP
system consists of networked personal computer hardware and software used to: 1)
generate and maintain the documents necessary for product fabrication, 2) communicate
the information contained in these documents to the production floor, and 3) obtain quality
assurance and manufacturing feedback information from the production floor. The
purpose of this paper is to describe the various components of the CAPPP system and
explain their applicability to product fabrication, specifically quality assurance functions.
INTRODUCTION
Historically, the design and manufacturing information required to fabricate a

product has been prepared, duplicated, and supplied on paper to support office or shop
activities. Babcock & Wilcox's internally developed CAPPP system uses a network of
personal computers in the office and on the shop floor to create a totally paperless
environment. In this CAPPP system, documents such as shop travelers, inspection and
manufacturing procedures, and drawings used for product fabrication are prepared and
then supplied to the shop floor electronically. In turn, shop floor personnel use PCs at their
work centers to access these instructions, enter process data, and indicate that a particular
fabrication step has been completed. Furthermore, the document revision, review, and
approval process is performed using theCAPPP system, so that in the production of CQMs,
no official hard copy documentation is needed.

BACKGROUND
The base business of Babcock & Wilcox's Naval Nuclear Fuel Division (NNFD) is
extremely documentation intensive (the weight of the paper shipped significantly exceeds
the weight of the hardware). The material and administrative costs associated with this
high volume of paper created a desire to develop a more efficient document
generation/ distribution and records storage system. In 1989 an opportunity arose to pilot
a much discussed "paperless factory" concept. At that time, NNFD brought in-house a
major manufacturing and assembly workscope which was previously sublet. This work
would be carried out in a one and a half bay facility, or work cell, requiring minimal
interface with existing NNFD operations. These conditions allowed for a controlled
introduction of the CAPPP concept. What began as a pilot program has proven successful
over the last two years at NNFD.
BASIC SYSTEM OPERATION
The shop traveler is the base document used for defining the step-by-step operations
necessary to the fabrication of a product. With CAPPP, an engineer accesses the software
and creates the shop traveler directly on the system. Each traveler step is numbered,
contains descriptive instructions, and lists associated drawings and manufacturing or
inspection procedures. Whenever the traveler and/ or reference documents are created or
revised, they are electronically routed to affected departments for review and signoff. The
originator must then respond to and/or incorporate any comments or suggestions
received. At this time, those performing the review indicate their approval of the document
via electronic signature - accomplished when an individual scans a barcode on his/her
ID badge which is then validated against their system login name. The approved document
is then automatically released for use. Drawings are prepared and revised by the drafting
department on the CAD/CAM system in "tile" format; where each tile is a view of the
product that is specific to the particular fabrication step being performed. Drawings are
routed for approval, translated from CAD/CAM to the CAPPP system environment and
released for use.
On the shop floor, a foreman uses a computer terminal to access the work schedule
and assigns work to his/her personnel according to the indicated priorities. Shop personnel
then access the job order data which brings up the current revision of the shop traveler onto
the screen. The latest revisions of associated reference documents such as manufacturing
procedures and drawings are instantly available for on screen viewing. During the course
of the operation, process or inspection data are entered into the system as required. Once
the operation is completed, shop personnel electronically sign the traveler step worked and
"clock-off" with the bar code scanner. This action causes the system to store the collected
information on the file server.
SYSTEM CONFIGURATION
Software
The CAPPP system software is divided into two categories; creation and delivery.
Creation refers to software which supports the generation and revision of documents such
as shop travellers, work instructions, and drawings. Document review, approval, and
release are also supported by creation software. The creation portion of the system was
developed by B&W's Finance and Systems Department using Nantucket Corporation's
CLIPPER programming language. This software was developed in-house since an
acceptable creation/ delivery system was not commercially available. Delivery software
gathers and packages the various documents created for a particular task and presents or
514
displays the information to shop personnel for use in production. B&W chose to use Eyring
Computer System's Impression software for delivery. This particular software was chosen
on the basis of it's 40 to 1 image compression capability which enhances delivery speed
and reduces the amount of memory required. Other favorable aspects of Impression
include interface versatility and quality of display.
Hardware
To be compatible with the NNFD's automation strategy, CAPPP runs on a networked

microcomputer platform, using NOVELL local area networks and IBM PC clones. The
computers used by engineering and those used on the shop floor are linked to a common
file server. The current configuration of these three system elements is shown in Table 1.
Table 1. CAPPP System Hardware Configuration

Engineering PC File Server Shop Floor PC
80386 / 20 MHz proc. 80386 / 25 MHz proc. 80386 / 20 MHz proc.
4 megabytes memory 8 megabytes memory 4 megabytes memory
40 megabytes hard drive Dual 650 megabyte disk 40 megabyte hard drive
drives
VGA graphics VGA graphics
controller board controller board
Ethemet network Ethernet network
controller board controller board
Barcode reader Barcode reader
MS-DOS based Interlan Network TCP MS-DOS based
gateway
Second 80386 configured
as backup
APPLICABILITY TO QUALITY ASSURANCE
The inherent structure of and controls contained in the CAPPP system are directly
applicable to implementation of quality assurance functions and philosophies. From an
overall perspective, the CAPPP system is the result of a quality improvement effort, one
objective of which was to reduce material and administrative costs associated with a large
volume of paper documentation. A second objective was to provide people on the shop
floor with instant access to all information related to a particular operation. Input from
shop floor personnel including machinists and inspectors, as well as engineering and
management, was used in defining the functions and configuration of the system. These
people were also involved in the process of evaluating and refining the system.
The CAPPP system supports quality assurance in the area of process control. Where
a shop traveler step requires test or inspection data to be collected, an associated data
collection form complete with identification of the parameter(s) to be checked, data entry
fields, and a drawing will be displayed. The data may be keyed in or downloaded from
inspection equipment directly into the system.
Also, the software automatically compares measured values with predefined
tolerances, and graphically indicates an in or out of specification condition. Under the
CAPPP system, personnel on the shop floor have instant access to process monitoring
information as well as their work instructions. Statistical process control software is on line
and set up to automatically extract and utilize predefined pieces of data in the construction
of histograms and control charts. All data is accessible for review at any time during the
manufacturing process. The software may be programmed to check for the presence of
critical data prior to release for assembly stage upgrade (example Cold Mass released for
cryostating).
515
Implementation of the CAPPP system has made several administrative quality
assurance functions more efficient and less error prone. For example, control of document
change effectivity is automatic under the computer based system. Once a document is
approved by each appropriate department and/ or the customer, it is released for use with
an associated revision number and date. Approval of the change is electronic and therefore
implementation is instantaneous. Since no time is lost in the reproduction and physical
distribution of paper documents, there is less chance of performing an operation to
out-of-date or incorrect work instructions. Record archival is now automatic. Each
document along with its revision history, is electronically maintained, therefore the need
for clerks and large amounts of storage space is eliminated. When a Collider Quadrupole
Magnet is complete and ready for shipment, the CAPPP system will automatically
assemble all manufacturing and quality assurance history information pertinent to the
acceptability of the component, and an electronic or hard copy certification package will
be prepared.
ACKNOWLEDGEMENTS
The authors wish to acknowledge Glenn Gale and Doug Cook of B&W NNFD, whose
efforts were key to the successful implementation of CAPPP at B&W and who provided
valuable input for this paper.
516
CONSTRUCTION EXPERIENCE WITH FERMILAB·BUILT
FULL·LENGTH SOMM SSC DIPOLES
M. J. Blessing3 , R. C.Bossert 1, J. S.Brandtl. J. A. Carson 1, S. Delchampsl, K. D.Ewald 1,

H. J. Fulton 1 , M. Gordon2 , 1. E.Haggardl, W. Higinbotham 2 , D. E.Hoffman 3 , R. H.
Jensen 1 , W. Koska 1, M. D.Packer3 , R. K.Rihel I, W. F. Robotham 1 , R. Sims2 , B. E.
Smith 1, D. J. Smith 1, J. B. Strait1, G. Tassotto 1, D. A. Tinsley1, M. Wake 1, M. Winters 1 ,
and W. F. Zimmerman 1
1 Fermi National Accelerator Laboratory, P.O. Box 500, Batavia, Ill. 60510
2 Superconducting Super Collider Laboratory, Fermilab/SSC Magnet Project, P.O. Box 500
Batavia, Ill. 60510
3 General Dynamics, Space Systems Division, P.O. Box 85990, M/Z DC-9540, San Diego
CA 92186-5990
ABSTRACT
Fourteen full length SSC dipole magnets are being built and tested at Fermilab. Their purpose is to
verify the magnet design as well as transfer the construction technology to industry. Magnet design is
summarized. Construction problems and their solutions are discussed. Topics include coil winding, curing
and measuring, collaring, instrumentation, end clamp installation, yoking and electrical and mechanical
interconnection.
INTRODUCTION
A series of long 50mm SSC dipoles are being constructed at Fermilab. Magnet design is described in
detail in References 1,2,3 and 4. Special tooling is used to build the magnets as well as take measurements
during assembly. This paper describes both the production methods used to build these magnets and the
measurements taken during construction. Some analysis of the measurements is included. Organization will
be in order of construction.
The magnets being built are numbered DCA310 through DCA323. The rust three, DCA310 through
DCA312, are built by Fermilab. The next seven magnets, five of which will be used in a string test at the
SSC Laboratory, are built by General Dynamics personnel at Fermilab. The final four, DCA320 through
DCA323, will be built by Fermilab. Ten of the magnets, DCA310 through DCA319 have been completed to
date. DCA311 through DCA316 have been tested.
CABLE
The cable used for the 50mm dipoles is multistrand keystoned cable similar to that used in the 40mm
SSC dipoles5 . It is being supplied to Fermilab by the SSC Laboratory.
The keystone angle on the inner cable used in the 50mm SSC magnets, although within specification,
has been systematically at the low end of the tolerance limit. As a result, the azimuthal pressure on the coils
during both curing and collaring has been much higher on the inside surface (near thin edge) of the cable than
on the outside. Small amounts of damage tc the glass tape have been observed on these surfaces as a result.
Although this inconsistency in coil preload creates a higher risk for quenches and turn-to-turn shorts, it has
not been responsible for any known magnet problems.
CABLE INSULATION
The cable insulation system used on the long dipoles is identical to that used in the 40mm dipole
program. It consists of one layer of .02S x 9.5 mm kapton overlapped by SO% surrounded by one layer of .10
x 9.S glass tape wrapped with 90% coverage. The glass tape is impregnated with IS-20% B-stage epoxy by
weight. This is slightly higher than the amount used on 40mm dipoles. Epoxy bonds in the coils produced
with the lower epoxy content occasionally cracked in the 40mm coils. Although no known magnet
performance problems resulted from these cracks, the epoxy content was raised as a precautionary measure.
WINDING
Coil winding is a semi-manual operation. Operators form the cable around the winding keys by hand
while the table moves into the proper position. Tension is automatically kept at 310 N for both inner and
outer long coils. The winding tensioner has an accuracy of ±13 N. Tensions vary depending on cable size,
coil length and the type of winding mechanism. Specific tensions used represent a compromise between
adequate placement of end parts, straight section sag, cable mechanical stability and coil springback. Final
tensions used are determined on the production floor by trial and error.
Cable mechanical stability limits the upper level of tension on both inner and outer SOmm long coils.
If the tension exceeds 310 N, "decabling", or the unraveling of strands from their keystone shape, begins to
occur. This 310 N limit is true only for the specific winding machine used for SSC long coils at FNAL.
The point at which decabling occurs depends upon the specific geometry of the winding mechanism. The
angle at which the cable is directed toward the winding mandrel, distance from the reel of cable to the mandrel
and the number of degrees of freedom of motion allowed by the assembly all affect the permissible winding
tension. The table which was used for the SOmm dipoles has only four degrees of freedom of motion, one
less than is recommended for winding the coil ends. Use of a different winding machine may result in
different tension requirements.
The lower limit of tension on long coils is determined by the positioning of the end parts during
winding. If tension is too low, the end parts will be spaced too widely in the longitudinal direction, resulting
in inadequate compaction of the end windings. Even though hydraulic pressure is applied to the ends during
curing, it has little effect beyond the outermost current block. It is impossible to get the parts properly
positioned unless they are placed correctly during the winding process.
Coil springback after being removed from the curing mold also becomes increasingly more important
as the tension is increased. Springback is undesirable because it causes coil length to vary and because the
epoxy bonds between coil components can become broken when the coil contracts. The amount of
springback is proportional to both the magnitude of the tension and coil length. Long SOmm SSC coils
springback by approximately 12mm ±1.S and 19mm ±1.S for the inner and outer coils respectively.
WEDGES
Solid copper wedges are placed between current blocks. They are insulated in the same manner as the
cable and placed into the coils by hand during the winding operation. Since none of the wedges lie radially
with respect to the magnet bore, their ideal configuration should be nonsymmetrical. If the wedges are made
symmetrical, a portion of the coil will be left uncovered as shown in Figure I. At issue is the advantage of
the simplified assembly of the symmetrical wedge vs. the possible disadvantages of leaving a portion of the
cable unsupported. All wedges in the inner coil are symmetrical. Production began using a nonsymmetrical
outer wedge, since the unsupported area created by using a symmetrical wedge in the outer coil exceeds
.ISmm. This wedge is marked by the manufacturer to prevent inserting it backwards. Two wedge related
problems have been encountered.
1.) The inner coil wedge nearest the parting plane has a very large angle and must rotate significantly in the
mold to be pushed into position. It scrapes against the retainer, causing damage to both the retainer and
wedge insulation. One solution to this problem has been to insulate the wedges with kapton only. The
lower friction surface of the kapton does not then become damaged or cause damage to the retainer. Short
coils have been cured with wedges insulated only with kapton. No damage to retainer or insulation have
occurred. In addition coil size variations were significantly smaller (~ .07 vs. ~ .IS mm peak-to-peak) than in
the coils with glass tape insulating the wedges.
2.) The outer (non symmetrical) wedge has by design a .ISmm difference between the length of the two
surfaces. Manufacturing errors in this wedge have caused them to vary from the design to perfectly
symmetrical. In addition a mark which is supposed to identify the longer side has not been placed accurately,
causing some wedges to be placed into coils with the wrong orientation. As a result we have incorporated a
symmetrical wedge on the outer coil. It was determined that the .ISmm uncovered area would not create a
problem. 6 No performance problems have resulted.
518
Uncovered
Area
Figure 1. Symmetrical Wedge
Coil end parts are designed and manufactured according to a geometry created by computer programs at
Fermilab.3 These programs attempt to minimize the internal stresses in the cable at the coil ends. The parts
are also designed to minimize the magnetic effects of the ends. Shapes are completely defined to support
manufacturing, analysis and winding requirements.
End part placement is largely determined by proper positioning of the parts during the winding
operation. The parts are then further compressed into position during curing by applying hydraulic loads to
the ends of the coil during the curing process. Experiments were done in the short program to determine the
magnitude of the end pressure) Various end pressures were tried and a compromise was drawn between
placement of end parts, insulation damage on ends and effect of end pressure on coil azimuthal size near the
ends. 13300 N was chosen as the longitudinal end force during curing. Ends generally compress
longitudinally to within less than 3mm of their design value. The ends do not compress fully because of local
changes in the cross-sectional shape of the cable. g Two types of changes have been observed. The cable
becomes "unkeystoned" due to internal stresses caused by bending the cable. The top (thick) edge becomes
thinner while the bottom (thin) edge becomes thicker. The midplane thickness can also increase due to the
inability to apply enough end pressure to properly seat the parts without damaging the cable insulation. The
end part design has been changed to adapt to the local changes in cable shape.
CURING
Ferrnilab coils are cured in a closed cavity mold (see Figure 2).9 Four variables are manipulated during
the curing process: three independent sets of cylinders and temperature. The mandrel cylinders apply a radial
load to the coil through the mandrel. The platen cylinders apply the azimuthal load to the coil through the
sizing bars. End cylinders apply a longitudinal load to each end of the coil. Temperature is increased to
135 0 C while the cylinders apply loads which close the cavity. The sequence between the three loads and
temperature of the mold is provided by computer input to the press and is shown in Figure 3. The position of
the sizing bars determines the azimuthal coil size and is set by stops on the curing mold.
Platen forceMandrel force Platen force
Cur ing
Pre ss
S lop
SIZing
Relaine r Bar
11
Figure 2. Curing Mold
519
2800 Platten load (NI linear mm) 300
2450 260
E 2100 220 U
E
... 1750
CII
CII
£. Temperature (e)
180 ...
CII
::J
c 1400 0"'· ------------------, 140 ...
'iii
Iii
c 1050
".' \ CII
c..
""
"." [ E n d Load (not to scale) "- 100 E
0
iCII 700
,.-' " - .................
60
CII
I-
Z /:000 N " -----. ---- __!~_~E_~ ---"!. ----- --- --. ----- --- -. ---:-:'-::.= ... --.
350 ,----- -- -- --- .,__t= M~~~~~_~~e~~~_~~~~~a~_~_m) ____~-~ 20
Time (hours)
Figure 3. Press parameters
The hydraulic system applies a load of 2300 N per linear mm to the lower tooling during the curing
process (platen load minus mandrel load). This is equivalent to an azimuthal pressure of 90 MPa on the inner
coils and 95 on the outers. Only a fraction of this load is actually applied to the coils during curing. This
fraction is determined by measuring the pressure at which the mold cavity closes, i.e., the point at which the
tooling begins reacting to the hydraulic forces. Measurements using pressure sensitive tape ("Fuji film") on
the azimuthal coil surfaces indicate that the coils are sUbjected to azimuthal pressures in the curing mold of
55-70 MPa before heating and 27-35 MPa after the coil has been cured.
COIL MEASURING
Azimuthal, radial and longitudinal measurements of each coil are taken after curing. Azimuthal
measurements are used to determine the ultimate preload of the coil and to verify that the coil is dimensionally
consistent. All measurements compare the coil with a steel master of the correct size.
Most of the azimuthal measurements on 50mm SSC coils were done on a semi-automatic fixture. The
coil is placed into the fixture by hand, hydraulic pressure is applied to the coil and readings are taken with an
L VDT. Control of the hydraulic pressure as well as the measurements are done by a computer. Some coils
late in the project were measured with a fully automatic fixture developed at FNALI0. All functions,
including the linear motion of the machine along the coil, are automatically controlled.
Measurements of coil azimuthal sizes are used to predict preload. Coil mean sizes must be kept
within certain upper and lower boundaries to achieve the proper preload. Coil preload for SSC dipoles needs
to be maintained within a target window of 55-85 MPa for the inner coil and 40-70 MPa for the outer coil at
room temperature. It has become clear, however, that the coil sizes and pressures "as measured" do not yield
the absolute preloads expected in assembled magnets as read by internally placed strain gages. It is unknown
whether these discrepancies are caused by measuring fixture deflections or some other source. Nevertheless
consistent relationships exist between measured values and magnet preloads. This makes it possible to use
the measurement data to predict preload by comparing the preloads of completed magnets to the measured coil
sizes. Later coil sizes are then adjusted by appropriately changing the curing mold cavity stops at the
positions shown in Figure 2. This process led to the conclusion that the desired measured coil size is .23mm
larger than the master for inners and .03mm smaller than the master for outers. 11 Both inner and outer coils
are consistently being produced with mean sizes which yield preloads within the target window. Figures 4 and
5 show the average sizes of inner and outer long coils and the curing cavity stop size for those coils. The
cavity stop size represents the size of the mold cavity with respect to nominal.
The size variations can be divided into "horizontal" and "vertical" errors.12 The mathematical difference
between the left and right coil sizes is defined as the horizontal error. The remaining coil size error is called
the vertical error and is evident when the right and left coil sides increase or decrease together. Figure 6 shows
an individual coil "overall" azimuthal size measurement. Figure 7 displays the deviations from the coil mean
size broken down into their horizontal and vertical components. Study of many coils indicates that most
vertical errors can be directly attributed to tooling variations. Horizontal errors appear to be random but can
possibly be reduced by controlling the fit of the tooling slot. Lowering the coefficient of friction between the
tooling and coil (such as wrapping the wedges only with kapton) can also have a significant effect. Three
distinct "valleys" appear on the coils. They appear to be related to the press rather than tooling because they
appear on both inner and outer coils. Total size variations for both inner and outer coils are less than .15mm.
520
I ~ Coil Size at 85 MPa - Mold CavIty SIze
.35 .35
.30 .30
E
E .25 .25
.20 .20
.15 .15
-N~V~~~~~O-N~V~~ ~~
ggg~~~~~~§§§§§§§o§§
Coil Number
Figure 4. Inner Long Coil Sizes
1= Coll Size at 65 MPa-- Mold Cavity Size

.10 10
.05 .05
Coil Number
Figure 5. Outer Long Coil Sizes
- - Quadrant 1/111 size with respect ---a-- Quadrant II/IV Size with respect
to master at 65 MPa to master at 65MPa
E
E
o 2.5 5.0 7.5 10 12.5 15

Meters
Figure 6. Coil Size
~Horlzontal Error ---<>-- Vertical Error

(Quadrant 1/1I1-III1Vl/2 (Quadrant 11111+1111 Vl/2
E
E
-.05
0~-----2+.5----~5----4-7r.5-----1rO~~1-2~.5----~15
Meters
Figure 7. Horizontal and Vertical Errors
521
COIL ASSEMBLY AND INSULATION
The coil insulation system in Fermilab SSC 50mm dipoles consists of several layers of .13mm
kapton. "Collaring shims", traditional components of superconducting magnets, are not used. Collaring
shims typically provide two functions: to protect the kapton insulation from the serrated edges of the collar
and to provide a mechanism for preload adjustment. The 40mm SSC program has shown collaring shims to
be unnecessary. Many 40 and 50mm magnets have been built and tested without indication of coil insulation
deterioration. Fermilab is now building 50mm coils accurately and consistently. They do not require pole
shim adjustment to obtain the proper preload.
The "collaring shoe", a .4mm thick sheet of brass which lies between the outer coil and the collars,
was eliminated in the 40mm program but has been reintroduced in the 50mm models. The shoe serves two
functions. It protects the kapton ground wrap from the serrated edges of the collar. It also provides a firm
barrier to prevent the possibility of ground wrap breakdown if a large pressure differential occurs across the
outer ground wrap. This could happen if an outer coil quenches in the area of an intermittent pocket in the
collar lamination assembly.5 These intermittent pockets are approximately 3mm wide and are placed at 3mm
intervals. They result from the collar geometry and are unavoidable due to the design required for collar
assembly.
It is uncertain whether collaring shoes are either necessary or sufficient to serve these functions.
Certainly they are not needed to protect the serrated edges of the collar from the ground wrap. Over 750
Tevatron, two long and seven short 40mm SSC dipoles have been built and operated without the collaring
shoe and have not exhibited performance problems. Experiments could be performed to determine whether
they are necessary, but time did not permit them to be completed in the 50mm program.
Some performance problems may have resulted from the use of the collaring shoe. The shoe in the
initial short model (DSA321) was too short, leaving a large portion of the pole turn of the outer coil
unsupported radially. The first (and only) training quench in DSA321 was in this tum. It is possible that the
quench was precipitated by the short collaring shoe.
COLLARING
Fermilab magnets are collared using laminated collaring tooling (see Figure 8). The tooling consists of
several components. They are: a laminated structure into which the collared coil is placed, a hydraulic system
to drive in the tapered collaring keys, "key supporting bars" to support the tapered keys as they are being
inserted into the collars and a transport mecha!lism to aid in rolling the tooling in and out of the press. The
tooling is designed to operate in a "closed cavity" condition, i.e., collars are closed to a fixed dimension and
the keys are inserted.
Figure 8. Collaring Tooling
The tooling provides for two different methods of collaring a coil, called the "tapered key method" and
the "square key method". In the tapered key method a vertical load is applied until the collars are closed just
enough to allow the tapered keys to engage. The rest is done by driving in the tapered keys. In the square key
method the vertic~ load is applied with the press until the collars are completely closed. Very little side force
IS then needed to Insert the keys. The advantage of the tapered key method is that a minimum amount of
o:,ercompression of the coils is necessary to collar the magnet. The square key method requires temporary
hIgher preloads during collaring. The advantage of the square key method is that the keys are not damaged or
"grooved" during collaring. A more consistent relationship between coil size and preload can therefore be
522
achieved. The Fennilab collaring tooling can be used to collar a magnet by either method. In addition, one
can choose to use a "mixed method" by allowing the vertical and side cylinders to share the work.
Keying procedures for both short and long magnets were developed in the short magnet program) 3
Objectives of the development procedure are listed in Table 1.
Table 1. Keying Procedure Development Objectives

l. Optimize procedures to minimize peak coil stress in the collaring press. This is done by shimming the
tooling to adjust the vertical press stop and to achieve proper azimuthal alignment of the upper relative to the
lower collars. Positions of the vertical and azimuthal shims are shown in Figure 8.
a. Primary method: set the keys in (square key method). Goal is to minimize the amount of
overcompression required.
b. Backup method: Drive the keys in (tapered key method). Goal is to minimize the side force, and
consequently the defonnation of the keys, required.
c. "Mixed" method: Goal is to find a compromise which minimizes the negative effects of both
methods.
2. Detennine coil size and pole shim combination that gives prestress in the target window of 55-85 MPa for
the inner coil and 40-70 MPa for the outer coil. This ultimately detennines the coil size needed.
Short magnets were keyed and disassembled several times while changing pole shims, tooling shims and
keying method. A summary of the results is shown in Table 2. A graph of the coil preload vs. press load
during a key insertion chosen for the long magnet program is shown in Figure 9. This key insertion uses a
"mixed method" which optimizes the advantages of both square and tapered key insertion.
Table 2. Keying Results

I. Full range of key insertion types has been covered by varying the press stop by .35mm.
2. Springback preload loss is approximately 28 MPa for the square key method and approximately 7 MPa
using the full tapered key method (525 N/linear mm horizontal force used to insert keys).
3. Using full tapered key insertion the collar and key defonnations result in vertical collared coil changes of
approximately .04mm. This is not expected to have a large effect on field quality.
4. Final choice of press stop for long magnets is a mixed method requiring 175-350 N/linear mm horizontal
force for key insertion. Springback is approximately 21 MPa.
5. Pole shim adjustment indicates that measured coil size to achieve the proper preload should be .23mm
larger than the master for the inner coil and .03mm smaller than the master for the outer coil.
VertIcal press load (N/mm) l, 160

4000 140
E 120
3000 Inner Coil Preload (MPa) ~
.....
E 100
.
,.=
QI 80
2000 Co.
Outer Co11 Preload (MPa)J 60 z:
Z
1000 40
Horizontal Press Load (N/mm) 20
0 0
1 2 3 4 5 6 7 8 9 10 111213141516
Sequence Number
Figure 9. Collaring Procedure
Ten long and several short 50mm magnets have been collared and keyed to date using this method. No
significant problems have occurred during the collaring process.
END CLAMPS
The coil end and splices are enclosed in a collet style clamp assembly4 shown in Figure 10. The lead
end configuration is shown. The return end clamp is identical except that there are no splices. The coil is
surrounded by a four piece G-IO collet. The collet is closed by driving on an aluminum tapered sleeve,
thereby compressing the end sections of the coil. The assembly fixture has a capacity of 266000 N. Both
inner-to-outer and upper40-10wer coil splices are made during this assembly step. All splices are compressed
within the end clamp.
523
Experiments were done to develop a method of measuring the preload inside the end clamp.1 4 Pressure
sensitive "Fuji film" was used to measure the preload in a short magnet (DSA322) end clamp. The pressure
applied to the film is determined by the degree of color apparent on the surface of the film . The film was first
calibrated to ensure the accuracy of the pressures read on the film.1 5 Four different measurements were made
during assembly as an attempt to read externally the preloads inside the end clamp. They are: pressure
required to close clamp, diameter of clamp at various angles and circumference of clamp (pi tape).
Measurement positions are shown in Figure 10. All four of these values were plotted against the pressures
read by the Fuji film. The pi tape vs. Fuji film is the only relationship which shows a clear correlation.
Based on these measurements, end clamp preloads appear to be adequate, although slightly lower than in the
magnet body.
Vertlce'
Horlzon!.1 t+---F9f-=~-+I
Mee.urltment Pos ition

(mm 'rom end)
Figure 10. End Clamp Measurements
YOKE
The collared coil in SSC 50mm magnets is surrounded by a 330mm diameter iron yoke. The yoke
consists of 1.6mm thick low carbon steel laminations stacked into packs. The yoke pack layout is shown in
Figure 11. There are three different types of packs. Standard packs are 3.5m long and are held together by
13mm diameter pins. The pins are flared on one end and welded to the final steel lamination on the other.
"Monolithic" packs are placed at various positions throughout the pack assembly as shown. The laminations
in these packs are coated with epoxy and cured while being compressed to create monolithic blocks which are
highly resistant to shear. They are placed between standard packs to increase the column stability of the
assembly. Other special yoke packs called "end filler packs" are placed around the end clamp area to prevent
a large stress discontinuity in the shell at the ends of the yoked portion of the cold mass. They have no
magnetic function. They are made of high manganese stainless steel which has a similar thermal contraction
to that of low carbon steel but is non-magnetic.
The shell is made of full length 5mm thick stainless steel. The yoke and shell are placed around the
collared coil, compressed and welded at the alignment keys. 2 kN/mm is applied to the shell exterior to close
the vertical gap between the yoke packs.
.26 .25 3.44 .12 3.44 .12 3 . 44 .12 3.44 .25 . 26

(Pack langth In meters)
Figure 11. Yoke Pack Stacking Assembly
A vertically split yoke is used. The iron laminations are drawn onto the collars from the horizontal
direction allowing the collared coil to be inserted easily into the yoke while still maintaining the appropriate
horizontal interference necessary at room temperature.
When the completed magnet is cooled to helium temperatures, the yoke assembly contracts
longitudinally as well as radially. The stainless steel shell contracts more than the iron yoke putting the shell
into tension and the yoke into compression. Care must be taken that the "air gaps" between the yoke
laminations are sufficiently large to allow the yoke to compress at least as much as the differential thermal
contraction between the yoke and the shell. The "packing factor", or ratio between the steel volume and total
524
space within the yoke was chosen conservatively to be 98% for the SOmm dipoles.!6 This was estimated to
provide 20 times the space needed for thermal contraction.
Experience with the early 50mm dipoles indicated that the unwelded yoke assembly did not have
sufficient column stability to withstand the press loads without buckling.!? The initial design, in addition to
using a 98% packing factor, included monolithic packs only at each end of the cold mass. Yoke laminations
in magnet DCA311, the first magnet yoked with this design, buckled severely near the center of the magnet.
Shell diameters decreased by as much as Amm in this area. The solution was to increase the packing factor to
99% and to add monolithic packs between each regular pack as shown in Figure 31.18 Yoke buckling did not
occur on subsequent magnets.
Strain gages are applied to the shell during magnet assembly for the purpose of measuring shell stresses
during welding and testing. They are read in the yoke press before welding and after each pass. Shell tension
after welding was estimated both from the shell gages 19 ,20 and from coil preloads. 21 Although the data is
not in total agreement and is difficult to interpret, indications are that shell stresses are in excess of 180 MPa
in the area near the alignment keys. This is sufficient to keep the vertical gap between the yoke halves closed
during operation.
FIELD ANGLE MEASUREMENTS

Measurements of the vertical plane of the magnetic field are taken of the completed cold mass. 22 These
measurements indicate the degree of twist inherent in the assembly. Measurements for a "typical" cold mass
(DCA313) are shown in Figure 12.
Coil Twist Magnet DCA3 13
10
8
6
'0 4
'"
L
E
2
0
'" -2
0,
c:
« -4
-6
-8
-10
-7.5 -5.0 -2.5 2.5 5.0 7.5
Return End Position (meters) Lead End
Figure 12. Field Angle Probe Measurements
INTERCONNECTION
The cold mass interconnection area begins with a 38mm thick end plate. Four preload screws are
mounted into the end plate which apply a longitudinal load to the coil through the end clamp. Total load
applied to each end is 9000 N at room temperature. The area exterior to the end plate includes electrical
connections for the power leads, strain gages, voltage taps, strip heaters, spot heaters and thermal sensors.
All is enclosed by a stainless steel end dome which is welded to the shell. Primary mechanical components of
the interconnection are shown in Figure 13.
Beam Tube
(instnJrnented on
...,tum end onlyl
Figure 13. Interconnection Components
525
Problems developed while welding the end dome to the end plate. 23 The weld configuration is shown
in Figure 14. The weld between the end plate and the dome extension tube was made very large to meet an
ASTM pressure vessel specification. Heat from the welding process caused the end plate to deflect inward,
increasing the end force exerted by the preload screws. The screws cannot be turned after welding. The
extension tube was chamfered as shown, reducing the end plate deflections.
Redesigned
Weld- 6 mm
deep
Figure 14. End Dome Weld
CONCLUSION
Full length SSC 50mm dipoles have been produced successfully at Fermilab. The magnets are being
tested and are operating within the criteria set for the ASST string test. Construction technology has been
transferred to industry. The basic design has been determined and requires little but a few refinements. The
dipole program is ready to enter the industrialization phase of development.
REFERENCES
1. Gupta, R. C., et aI., "ssc SOmm Dipole Cross Section, in Supercollider 3, p. 587, 1991, J Nonte, ed.
2. Strait, J., et al, "Mechanical Design of the 2D Cross Section of the SSC Collider Dipole Magnet, Proc.
of the 1991 IEEE Particle Accelerator Conf., p. 2176 1991, L. Lizama and J Chew, eds.
3. Brandt, 1. S., et. aI., "Coil End Design for the SSC Collider Dipole Magnet", ibid, p. 2182 and Fermilab
Technical Memo TM-1735.
4. Delchamps, S., et al., "SSC Collider Dipole Magnet End Mechanical Design", ibid, p. 2185.
5. Pewitt, E, ed., "50mm Collider Dipole Magnet Requirements and Specifications" (Yellow Book).
6. Strait, J., "Outer Coil Wedge - Symmetric or Asymmetric", Fermilab Technical Support Section internal
note TS-SSC 91-144.
7. Brandt, J., "ssc Dipole End Pressure Experiments", TS-SSC 92-036.
8. Bossert, R., "ssc End Part Modifications", TS-SSC 91-042.
9. Carson, J., et ai, "ssc Dipole Coil Production Tooling", Supercollider 1, p. 51, M. McAshan, ed.
10. Rihel, R. W., "Overview of a Robotic System for Azimuthal Dimensions of SSC Dipole Coils", these
proceedings.
11. Wake, M., "Relationship between Coil size and Preloading in 50mm Magnets", TS-SSC 91-070.
12. Sims, R., "Tooling and Process Parameter Improvements Leading to Improved Dimensional Control of
the Cured SSC Dipole Coils", Supercollider 3, p 57., J. Nonte, ed.
13. Strait, J., "Development of Collar Keying Procedures", TS-SSC 91-060.
14. Delchamps, S., "DCA311 - DCA319 End Clamp Deflections", TS-SSC 92-013.
15. Delchamps, S., "Calibration of Fuji Prescale Pressure Sensitive Film", TS-SSC 91-134.
16. Gordon, M., "Axial Compliance and Packing Factor for the Yoke of the 50mm SSC Dipole", TS-SSC
91-143.
17. Gordon, M., "Yoke Chevron Effects", TS-SSC 91-192.
18. Gordon, M., "50mm Dipole Revised Yoke Assembly", TS-SSC 91-169.
19. Strait, J., "DCA312 Shell Gage Data During Welding", TS-SSC 91-179
20. Strait, J., "DCA312 Shell Gage Data During Welding", TS-SSC 91-184.
21. Strait, J., "Estimate of Shell Tension from Coil Stress Data", TS-SSC 91-182.
22. Delchamps, S., et al., "Magnetic Field Measurements of Fermilab/General Dynamics built SSC Collider
Dipole Magnets", these proceedings.
23. Higinbotham, B., "Welding End Bell Assembly to End Plate", TS-SSC 91-208.
526
RESUL TS FROM THE FERMILAB 1.5 M MODEL MAGNET PROGRAM
M.J. LammI, R. Bossertl , J. Carsonl,S. Delchampsl, T.S. Jafferyl ,

W. Kinneyl, W. Koska l , J. Ozelis 1, R. Sims2, J. Strait 1, M.Wake l
and M. Winters2
1Fenni National Accelerator Laboratory

P.O. Box 500
Batavia, II 60510
2SSC Laboratory, Fennilab/SSCL Magnet Project

Fenni National Accelerator Laboratory
P.O. Box 500
Batavia, II 60510
ABSTRACT
A series of model SSC collider dipole magnets (50 mm aperture, 1.5 m magnetic length) are
being built and tested at Fennilab. The eady magnets have been built according to the Fennilab
baseline design. The magnets are precursors to the full length 15 m models to be used in the
Accelerator Systems String Test at the SSCL. All magnets are instrumented with voltage taps for
quench localization and with strain gages to measure stresses and deflections in the coil and support
structure. Measurements are made during assembly, cooldown, excitation, and wannup.
Comparisons are made between data and design expectations, and magnet behavior is correlated with
changes in the magnet design and assembly methods.
INTRODUCTION
Tests have been perfonned at Fennilab on 50 mm aperture 1.5 m long model magnets built at
Fennilab for the SSC project. Because of the reduced longitudinal scale of the magnets (1.5 m vs 15.
m) and because these magnets are tested in a vertical dewar (and hence need not be placed in their
own cryostat), it is possible to build and test magnets approximately 3 times faster than the longer
magnets and at a much reduced cost. This allows the program to validate many design features in a
timely fashion which in tum may be incOJporated into the long magnet program.
Table 1 is a summary of the eight 50 mm model magnets that have been constructed to date.
These magnets incorporate the features of the Fennilab full length baseline design l ,2,3,4. The
inner(outer) coils are made of 30 (23) strand NbTi cable with 6 micron filament diameter and Cu:SC
ratio of 1.49. The copper field shaping wedges are insulated either with glass-epoxy tape and
Kapton or Kapton only. The 1.52 mm thick stainless steel collar laminations are locked in place
using tapered brass keys. The outer coil is insulated from the collars by 5 layers of 0.13 mm Kapton
insulation. On some magnets, quench protection heater strips are placed between these layers. The
results of heater tests have been reported elsewhere5 . The inner-outer coil splice is located external
to the high field body region, and is supported along with the magnet end by G-lO collets and an
Work Supported by the Department of Energy

Table 1. Summary of Fermilab Constructed 1.5 m Model Magnets.
Magnet Variation of Collar Post-Yoke Pre-Stress

Name Fermilab Baseline Shimming Stress Loss with
(mm) (MPa) Cooldown
Inner/Outer Inner/Outer (MPa)
DSA321 Stainless Steel 0./0. 62./86. -19./-14.

End Can
DSA322 Assembly
Experiments Only
DSA323 Stainless Steel 0./0. 52.n2. -30./-10.
End can
DSA324 Aluminum . 13/-.13 76./60 . -33./-8.
End Can
DSA326 Aluminum 0./0. 69./56. -37./-23.
End Can
DSA327 Assembly
Experiments only
DSA328 Kapton Only .09/-.13 70./45 . -36./-9.
Wedge Insulation
Inner Coil Onlv
DSA329 KaptonOnly .09/0. 69./39 . Presently
Wedge Insulation Under Test
aluminum or stainless steel cylindrical can. The yoke consists of vertically split low carbon high
quality steel laminations. A stainless steel shell surrounds the yoke and is seam welded
longitudinally.
The magnet is instrumented with voltage taps at each coil splice. Additionally, the first two
current blocks nearest the the inner coil pole are instrumented for quench localization. Strain gage
transducers 6 are located in the collars to measure azimuthal stress between the collars and the coils.
Gages are also located on the surface of the shell to measure the stress due to welding, cooldown,
and the force transferred by friction from the coil during excitation. End force gages measure the
force of the coil on the end plate due to differential thermal contraction during cooldown and the
Lorentz force due to magnet excitation.
EXPERIMENTAL DETAILS
Collar strain gages were monitored during the collaring and yoking process. The design post-
yoke stress is 70 ± 14 MPa for the inner coils and 56±14 for the outer coils. This coil stress is
adjusted by adding or subtracting Kapton insulation at the pole as well as modifying the molding
fixture. As shown in Table 1, the desired inner stress is obtained except in magnet DSA323. For
DSA321 and DSA323, the outer stress was systematically higher than design but was adjusted lower
on subsequent magnets. Magnet measurements were performed after collaring and then again after
yoking using the BNL B2 Mole measuring system7 . The system consists of 0.6 m tangential and
dipole bucking coils rotating at 0.31 Hz. The voltages were read by fast sampling DVMs. The
resultant voltages are analyzed using an FFT algorithm.
Cold measurements are performed in a 3.1 m vertical dewar of boiling helium. Magnet
temperatures of 3.0 K to 4.35 K are achieved by varying the boiling helium bath pressure. Strain
gages and voltage taps, as well as dewar temperature, pressure and helium liquid level are monitored
through a DEC MicroVax II computer. Magnetic measurements are performed by inserting warm
probes into a stainless steel anti-cryostat, which is in tum inserted into the bore of the magnet. The
dipole field strength is measured using a Rawson-Lush field meter and has an accuracy of
approximately 1 part in 5000. Field harmonics are measured using Morgan and tangential coils
rotated at 6 Hz. The 46 cm active length Morgan coil was used on the earlier tests8, and was later
528
replaced with a 25 cm active length tangential coil. The resultant voltages are processed using a VIF
based data acquisition system. The Morgan coil data has windings up to the decapole only, thus no
transverse centering through 16 pole or 20 pole elimination is possible. For these tests only the
allowed b2 and b4 are reported. Magnetic measurements are performed as a function of excitation
current, and as a function of longitudinal position at 5 kA. The strain gage data are recorded in
increments of magnet excitation squared. The magnet is repeatedly ramped to quench at 16 A/s to
establish a quench plateau at 4.35K, 4.2K and 3.8 K. Additional quenches are performed at 4.35 K
at ramp rates up to 300 A/s. AC loss measurements were performed on DSA323, DSA324 and
DSA328. Each magnet was tested through at least two thermal cycles in order to study retraining.
Tests of five of these magnets(DSA321, DSA323, DSA324, DSA326, DSA328) have been
completed. DSA329 has not yet been cold tested. Two magnets (DSA322, DSA327) have been used
for collaring experiments, and will not be cold tested.
RESULTS
Quench Performance
The quench performance at 4.35 K (first thermal cycle) and 3.8 K (performed on either first or
second thermal cycle) is shown in Figure 1. For all quenches the magnet is ramped at 16 A/s from
zero excitation current. The quench velocity for such plateau quenches is typically 80 m/s9. With the
exception of the first quench of DSA321, these magnets exhibit no training at 4.35K. The quenches
occur in the straight section of the inner pole tum which is the high field region of the coil. The
training quench in DSA321 occurred in the outer coil and is probably attributable to a manufacturing
error 10. No training quenches are observed during subsequent thermal cycles. At 3.8 K less than
one training quench was observed on average. The exception to this was DSA324 which exhibited
three quenches in the outer coil.
During the initial test of DSA323, the magnet was found to quench unpredictably at
approximately 3 kA on the down ramp only. The quenches occurred in the pole tum of one of the
inner coils at the return end. The magnet was partially disassembled and the end cylinder was
removed. It was discovered that the coil end saddles were not all the same length. Thus, the shorter
coil might not be properly supported axially through the cylindrical end can. The end saddle lengths
were adjusted, the magnet was reassembled and retested. Unfortunately, down ramp quenches were
also found during this second test. Further studies are needed to understand this problem.
Figure 2 shows the ramp rate dependence of the quench current. At ramp rates less than 25-
50 A/s the quench current is insensitive to the ramp rate. The quenches occur in the high field region
of the conductor. For two magnets, DSA321 and DSA323, this small ramp dependence extends to
200 A/s and beyond. DSA323 quenches above 200 A/s occur in the ramp splice near the pole tum.
For magnets with more severe ramp rate dependence, the high ramp rate quenches occur in a less
instrumented portion of the magnet, towards the midplane. The high ramp rate quench location near
the midplane is expected because there the magnetic field direction is most perpendicular to the wide
cable face, thus facilitating inter-strand eddy currents.
~~~ iii:::: :~:::::~::::::)Ii

8250
x····· ·X
323 x····.:)t····.':~..e·:-::;:~:::::~··.
8000
326 07f..: .... of
32.
7750
326
7500
32.
32.
7250
321
7000
0 2 4 6 8 10
Quench Number
Figure 1. Quench history for first thermal cycle.
529
7500
:$
<:
...;:l~ 7000
u
.<:
o
.,
~
326
;:l
0' 6500 326
X loner pole turn

X Inner non pole lurn
324
o Ramp Splice
Ramp Rate (A/S)
Figure 2. Quench current as a function of ramp rate at 4.35 K
Magnet DSA324
300
S
.,
<l
»
u
""-
<II 200 Magnet DSA328
,..,0
<II
»
.,""...
~
w Magnet DSA323
100
Ramp Rate (A/S)
Figure 3 AC loss as a function of ramp rate. The magnet is cycled between 0.5 kA and 5 kA Solid
lines are least square fit to data. Slope of lines are 0.703, 0.624, and 0.594 J/A/S for DSA324, DSA328,
and, DSA323 respectively.
AC Loss Measurements
For three magnets (DSA323, DSA324, DSA328) the AC loss as a function of ramp rate was
measured using fast digitizing DVM's to measure the magnet voltage and current 11,12. Figure 3
shows the AC loss as a function of ramp rate using a 0.5 kA to 5.0 kA ramp cycle. The data from
each magnet are fit to a straight line. The extrapolated value at dl/dt =0 is the AC loss due to iron and
superconductor magnetization. All three lines extrapolate to the same value to within measurement
errors at dI/dt =0. This is expected since these magnets have the same iron contribution and are
made with similar conductor. The difference in slope among the magnets is due to the different
levels of eddy current loss. A comparison of Figure 2 with Figure 3 shows that there is a correlation
between eddy current loss and quench current degradation at high ramp rate. The variation of AC
loss as well as ramp rate dependence among these similar magnets (DSA323 and DSA324 are made
with conductor from the same reel) is an interesting phenomenon that is under study.
530
Inner coil. + DSA321
X DS.\323
Outer coils
<> 0SA.324
o DSA326
-+ DSA328
60
.~
p..
e
.,'"'" 40 '::6
.......
rn
. '. : :"5'.
. ... l!>..
'0
u o~
. :~.
. :'~."'+
.., ... -0
.+~ "¢"
20 "'x "rr:"+ " . 'I- "0
" ·x .. ,. X '. '0, ::+. "+ .' .. +.
. . . X .. . x. .,.0 ... +, .. 'I- '+
" 'x. ". ·0.·:·;.,.".+
'. 'X'" .x'
.. X
o ~-L~~~-L~~~~~~~~~-L~~~-L~~~~
o 10 20 30 40 50 60
(Magnet Current)2 (kA2)
Figure 4 Average coil stress as a function of excitation current squared.
Mechanical Measurements
Figure 4 shows the average prestress for inner and outer coils as a function of excitation
current squared (12). As shown in Table 1, each magnet loses 30 (15) MPa of coil stress during cool
down to LHe temperatures for inner(outer) coils. With the exception of DSA323, both inner and
outer coils show adequate prestress; no unloading is observed even at the highest 12. The slope of
stress vs 12 is approximately linear. A linear relationship between increasing end force and 12 is also
observed (not shown). The average end force at 7.5kA is 15 kN, which implies that approximately
90 percent of the the coil Lorentz force at this current is transferred to the collar/yoke/shell by
friction.
Magnetic Measurements
Table 2 summarizes the magnetic measurement at room .temperature and in atmospheric
pressure boiling helium. The cold measurements are an average of the readings taken at 2kA from
the up and down portion of a ramp cycle with maximum excitation current greater than 7.0 kA.
Only b2 and b4 are reported for DSA321 and DSA324 because only data from the Morgan coil is
available. For all magnets except DSA32 1, the probe active length is positioned longitudinally at least
20 cm away from the strain gage transducers. This is because the strain gage block becomes
magnetic upon excitation at cyrogenic temperatures and contributes significantly to b2 and b4 at 2 kA
excitation current. For DSA321 the gage pack was located at the magnet's longitudinal center. Thus
it was not possible to take data without the influence of the gage pack. For this magnet the effect of
the gage pack is estimated from studies on a later magnet13.
For the room temperature measurements, the data are an average of ± lOA measurements in
order to eliminate the effect of any external magnetic field such as the earth's magnetic field. The data
are recorded at several longitudinal positions. However, we present only the data taken at the same
longitudinal location as the corresponding to cold measurement.
With the exception of pole shims and differences in copper wedge insulation, the variations in
hannonics are due to measuring and manufacturing errors. To estimate the wann and cold random
error, several data points were taken at the same longitudinal location for one magnet (DSA326).
The random errors are small compared to the magnet to magnet variation. The effect of Kapton pole
shimming which should affect only the allowed hannonics is estimated using I:field l calculations 14 •
On the whole, the RMS deviation of the hannonics are all within the required sse random error
tolerances. For many of the hannonics the systematic values are larger than the sse systematic
tolerance, even after correcting for the effects of pole shimming as shown. For b8, the present
design has approximately +0.05 units built in to aid in the centering corrections. The large variation
in b2 may be due in part to the distribution in cable thickness. The average shim corrected b2 value
for DSA321-6 (glass-epoxy tape insulated wedges) is 2.1±.6 compared to 0.5±.3 for DSA328-9
531
Table 2 Measured warm and cold harmonics. Room temperature measurements are averages of ± 10 amps.
Cold harmonics are average of up ramp-down ramp value at 2 kA for Imax >7 kA cycle. Cold Transfer
function measured with the Rawson Lush Field meter at 3 kA. For b2, b4 ,b6, b8 warm harmonics corrected
for pole shims are also presented.
HARMONICS (UNITS @ 10 mm)
Pole 321 323 324 326 328 329 326 SSC SSC
Ran. Sys. Ran.
b2Warm 2.952 1.378 1.762 2.190 0.069 1.796 0.082 0.80 1.15
Cold 3.2 1.497 1.935 2.243 0.427 0.025
Warm corrected 2.95 1.38 1.67 2.19 0.21 0.87
b4 Warm 0.327 0.215 0.186 0.350 0.255 0.079 0.015 0.08 0.22
Cold 0.22 0.134 0.040 0.284 0.099 0.005
warm corrected 0.33 0.22 0.34 0 .. 35 0.39 0.20
b6Warm -0.075 -0.082 -0.036 -0.069 -0.027 -0.037 0.007 0.013 0.02
Cold -0.046 -0.030 -0.026 0.002
Warm corrected -0.075 -0.082 -0.051 -0.069 -0.041 -0.055
b8Warm 0.047 0.056 0.052 0.056 0.058 0.046 0.005 0.01 0.01
Cold 0.045 0.049 0.038 0.011
Warm corrected 0.047 0.056 0.056 0.056 0.061 0.049
b10Warm 0.023 0.015 0.018 0,015 0.018 0.011 0.003
Cold 0.016 0.012 0.014 0.000
b1 Warm -0.071 0.095 -0.090 0.065 -0.272 -0.467 0.119 0.04 0.5
Cold 0.192 0.374 -0.132 0.042
b3Warm 0.027 -0.011 -0.090 0.101 0.010 0.045 0.017 0.026 0.16
Cold 0.058 0.080 -0.041 0.018
b5Warm 0.022 -0.021 -0.032 -0.023 0.025 0.012 0.012 0.05 0.02
Cold -0.022 -0.004 0.005 0.004
b7Warm -0.004 -0.005 0.000 0.004 -0.002 0.001 0.002 0.05 0,01
Cold 0.000 0.032 0.000 0.005
b9Warm 0.002 0,015 -0.002 0.006 0.010 0.005 0.005
Cold 0.000 0.041 0.000 0.003
a1 Warm -2.151 0.407 -0.003 1.025 -0.966 -0.666 0.133 0.04 1.25
Cold -0.332 0.964 0.041 0.033
a2Warm 0.104 0.027 0.427 0.689 0.099 0.287 0.082 0.32 0.35
Cold 0.588 0.575 0.094 0.006
a3Warm -0.024 0.021 0.190 0.086 0.092 -0.039 0.056 0.26 0.32
Cold -0.151 -0.002 0.156 0.017
a4 Warm 0.097 -0.067 0.044 0.055 0.023 0.040 0.022 0.01 0.05
Cold 0.046 0.023 -0.019 0.013
as Warm 0.043 -0.020 0.049 0,015 0.021 -0.027 0.014 0.005 0.05
Cold -0.009 0.021 0.027 0.007
a6Warm 0.004 0.003 -0.008 -0.016 -0.023 0.010 0.006 0.005 0.008
Cold 0.006 -0.017 -0.001 0.011
a7Warm -0.008 0.035 0.000 0.004 -0.010 -0.001 0.002 0.005 0,01
Cold 0.004 0.012 0.000 0.018
a8Warm 0.019 0.010 0.003 0.006 0.001 0.003 0.008 0.005 0.008
Cold 0.004 -0.007 0.003 0.005
a9Warm 0.001 0.009 -0.002 0.016 0.000 0.002 0.004
Cold 0.000 0.014 0.000 0.001
a10Warm 0.004 -0.014 -0.002 0.008 0.005 0.002 0.005
Cold 0.002 -0.005 0.000 0.001
Predicted T.F.
TF collar 0.795 0.794 0.794 0.793 0.794 0.796 0.0002 0.794
Warm 1.042 1.043 1.043 1.042 1.041 1.041 0.0002 1.045
Cold 1.042 1.043 1.044 1.042 0.002 1.045
532
=:..
.qo
,.I:l 0
~--------------------- {!II}
~
-1 {IV}
CURRENT (A)
Figure S. Measured and predicted transfer function vs excitation current.
1.10
+
~ 1.05
~ ~
l'i
o
:;:l
§ 1.00
r..
...
~., + Tangential Coil Data
~
t:. 0.95 - Field Calculation
0.90
o 2000 4000 6000 8000
Current (A)
Figure (i. Measured and predicted harmonics vs excitation current. (I) b2 from DSA326 (II) b4 from
DSA326, (III) Predicted Iron Saturation for b4, (IV) predicted Iron Saturation for b2 .
(Kapton only insulated wedges). It is estimated that a .005 mm displacement of the middle tum of the
inner coil towards the midplane (pole angles kept constant and all tum thickness adjusted linearly)
would change b2 by + 1.1 units 15 .
There is generally good correlation between warm and cold harmonics. For b2 and b4 (where
we have the most statistics) the RMS of the warm-cold distribution is less than 20 percent of the sse
random error tolerance. Thus the warm measurements should be a good indication of the magnet's
cold geometric harmonics at fields where the iron saturation effects are negligible.
Also included in Table 2 is a comparison of the measured transfer function for the collared coils
and warm and cold assembled magnets with prediction4. The cold measurements are taken at 3 kA
excitation current with the Rawson-Lush field meter. The effect of iron saturation at higher fields is
shown in Figures 5 and 6. Figure 5 shows the transfer function for DSA326, measured with one of
the dipole windings on the tangential probe. Above 4 kA, the iron saturation becomes significant.
533
For all magnets, both warm and cold, the measured value is systematically lower than predicted.
Figure 6 shows b2 and b4 as a function of excitation current, for currents above 2 kA. The dip in
both distributions at 4 kA is due to iron saturation in the poles, the subsequent rise is due to iron
saturation at the midplane. Cutouts are included in the iron yoke midplane to balance these two
saturation effects. The predicted b2 and b4 are also shown. b2 is expected to vary by approximately
0.6 units. The measured variation is smaller. It is interesting to note that the shape ofb2 is different
for Fermilab and BNL designed model dipoles. This is likely due to a collar alignment notch which is
located in the midplane of the vertically split Fermilab magnets and in the pole of the horizontally split
BNL magnets.
CONCLUSION
Tests have been performed on several Fermilab-built 50 mm aperture 1.5 m long model
magnets. The quench performance has been very good, with few training quenches. All quenches at
the nominal 16 Ns ramp rate exceed the SSC design operating current by at least 600 amps. There is
a correlation between AC eddy current energy loss and ramp rate dependence of the quench current.
All magnets exhibit good mechanical behavior. With the exception of the DSA323 inner coil,
there is adequate stress to prevent unloading during magnet excitation.
The warm and cold harmonics have been measured. Though the statistics are limited, the RMS
variation appears to be well within the SSC specifications. There is a good correlation between warm
and cold harmonics. There is good agreement in shape of the transfer function, though the warm and
cold transfer functions are systematically lower than predicted. The measured variation b2 due to iron
saturation is smaller than predicted.
ACKNOWLEDGMENTS
We would like to thank the staffs of the Fermilab Superconducting Magnet Production Group
and the Superconducting R &D Test Facility for their assistance in performing these studies.
REFERENCES
l. J. Strait et al., "Mechanical Design of the 2D Cross-section of the SSC Collider Dipole
Magnet", Proc. of the 19911EEE Par. Acc. Coni, San Francisco,Vo14, pp. 2176-8,1991.
2. S. Delchamps et al., "ssc Collider Dipole Magnet End Mechanical Design", Proc. of the 1991
IEEE Par. Acc. Con/. San Francisco, Vol 4, pp. 2285-8, 1991.
3. R. C. Gupta et al., "ssc 50 mm Dipole Cross Section", Supercollider 3, pp.587-599 1991.
4. E.G.Pewitt (ed), "50 mm Collider Dipole Magnet Requirements and Specifications", Fermilab
1991.
5. C. Haddock et aI., "ssc Dipole Quench Protection Heater Development", Proc. of the 12th
Int. Coni on Mag. Tech., June 1991, Leningrad, USSR.
6. C. L. Goodzeit et al., "Measurement of Internal Forces in Superconducting Accelerator
Magnets with Strain Gage Transducers",IEEE Trans. Magn., Vol. 25, pp. 1463-1468, 1989.
7. G. Ganetis et al., "Field measuring probe for SSC magnets" ,Proc. 1987 IEEE Particle
Accel. Coni, Washington D.C., 1987, p.1393.
8. M.J. Lamm et al., "Magnetic Field Measurements of 1.5 Meter Model SSC Collider Dipole
Magnets at Fermilab", Proc. of the 12th Int. Coni on Mag. Tech., June 1991, Leningrad,
USSR.
9. T. Jaffery, "Quench Summary of DSA324", Fermilab Technical Support Internal Note
TS-SSC 91-175, Sept. 1991.
10. M. Wake et al., "Quench behavior of 1.5 m model SSC collider dipole magnets at
Fermilab", Proc. of the 12th Int. Coni on Mag. Tech., June 1991, Leningrad, USSR.
11. M.J. Lamm et al. "Bipolar and Unipolar Tests of 1.5 m Model SSC Collider Dipole Magnets
at Fermilab", Proc. of the 1991 IEEE Par. Acc. Coni,San Francisco,Vol4, pp. 2191-4,1991.
12. J. Ozelis, "Status of AC Loss Measurements of 1.5 m Model Dipole Magnets at Fermilab"
Fermilab Technical Support Internal Note TS-SSC 91-205.
13. S. Delchamps "Effect of strain gage blocks on normal sextupole moment in DSA321 and
DSA323, Fermilab Technical Support Internal Note TS-SSC 91-104 ,1991.
14. R. Gupta, private communication.
15. R. Gupta, "Effect of the Azimuthal Variation in the Cable Thickness on the Field Harmonics in
SSC 50 mm Dipole Cross Section", BNL Magnet Division Note 379-1(SSC-MD-264), 1991.
534
CONSTRUCTION AND TEST RESULTS FROM 15 m-LONG, 50 mm APERTURE SSC
COLLIDER DIPOLE MODELS"

M. Rehak, E. P. Rohrer, W. Sampson, R. Shutt, R. Thomas, P. Thompson,
P. Wanderer; and E. Willen

C. Goodzeit and P. Radusewicz

Dallas, Texas 75237 USA
ABSTRACT
Construction details, assembly data, coil stress and end force measurements are
reported with quench data for the initial full-length SSC model dipoles with 50 mm
aperture being built at BNL.
INTRODUCTION
A series of six Collider Dipole model magnets are being assembled and tested at
BNL. These magnets embody the design principles that have been employed in the
previously tested series of 40 mm aperture magnets that have been built at BNL over the
past few years in support of the SSCL program. [I] The major change in these magnets (in
addition to the large aperture) is the use of wider cables for the inner and outer coils to
provide an increased margin, greater than 10%, at operating field. Other basic design
features that were used in the 40 mm magnets such as horizontally split yokes, internal
ramp splices between the inner and outer coils and individually determined ends clamped
in collars were retained. These magnets are instrumented with strain gauge collar packs
to measure coil azimuthal stress and end force transducers to measure the force of the coil
"Work supported by the U.S.Department of Energy.

ends against their support. Voltage taps applied to the inner coil turns and at coil splices
are installed in these magnets for quench origin determination. This paper discusses the
design details and some test results for the first three magnets tested in this series. This
includes mechanical assembly coil stress and end force histories, quench performance
results and coil stress and end force measurements made during testing. Field quality and
multipole measurements are discussed in a separate paper.!2J
DESIGN
The magnets utilize a two-layer cosine e coil design with 50 mm aperture and 100
mm outer diameter. The details of the magnetic design of this magnet have been
previously published. [3] The characteristics of the superconductor used in the three
magnets discussed here are given in Table I. Table II lists the construction details for the
three magnets. The two dimensional cross section of the cold mass is shown in Figure 1.
The collars used in this cross-section have 4 mils of compensation. This means that the
locations of the collar keyways are 6 mils closer to the midplane than those which would
make the collar outside diameter perfectly circular. When tapered keys are inserted into
the keyways and the clearance between the keys and keyway is 2 mils the vertical diameter
of the keyed collar is 4 mils less than circular. This effect has been referred to as "anti-
ovalization" and is now called compensation. This is to correct in part for the typical 10
mil vertical ovality created by deformation of the collars due to coil stress and insertion
of the tapered keys.
Table I.
Cable Parameters Mechanical Inner Coil Outer Coil

Filament diameter, jJ. 6.0 6.0
Strand diameter mm 0.808 0.648
Number of strands 30 36
Bare cable width mm 12.34 11.68
Bare cable mid-thickness, mm 1.458 1.156
Keystone (max/min) thickness, mm 0.262 0.206
Strand Properties Electrical
Ie @ 5.6 T Ie @ 7.0 T
Magnet Coil Mfg. Cu:SC and 4.2 K and 4.2 K
DCA207 Inner-upper IGC 1.55 2716 1811
Inner-lower IGC 1.54 2661 1787
Outer-upper IGC 1.75 2397 1645
Outer-lower IGC 1.75 2397 1645
DCA208 Inner-upper IGC 1.34 2571 1732
Inner-lower IGC 1.34 2571 1732
Outer-upper Oxford 1.79 2638 1756
Outer-lower Oxford 1.76 2614 1740
DCA209 Inner-upper Oxford 1.5 2700 1835
Inner-lower Oxford 1.5 2700 1835
Outer-upper Oxford 1.83 2665 1771
Outer-lower Oxford 1.79 2629 1750
536
Table II. Construction features of 17 m, 50 mm magnets DCA207-209.
1. W6733 coil design.
2. Cable insulation: 48% overlap wrap of 25 J.'m Kapton type H film covered
with a butt wrap of Hexcel F 185 epoxy impregnated
adhesive.
3. Solder filled internal ramp splice epoxy bonded to adjacent tum.
4. Collar design: 4 mil vertical OD compensation (reduction).
5. Collars spot welded, tapered (3° per side), alternating LlR pairs.
Material: 21-6-9 stainless steel, strain hardened to 90,000 psi yield strength.
6. Collar packs: 6-inches long assembled on brass tubes and adjusted to length
with variable brass shims in center of pack.
7. Outer coil scuff guard: brass, 15 mil.
8. Coil shims: brass, mechanically seated at collar pack ends.
9. Monolithic type (emulating a one-piece half-yoke) horizontally split.
10. Cross flow cooling.
11. Epoxy bonded stainless steel yoke modules at lead and return ends.
12. One piece, 1.5 inch (37 mm) stainless steel end plates welded to helium
containment shell.
13. Helium containment shell extension with access ports for instrumentation
mounted at lead and return ends, closure with dished heads.
14. Preloaded bullet type end force strain gauge transducers at both ends.
The collared coils are installed into the yoke with a horizontal split line and
nominal line to line fit between collar outside diameter and yoke inner diameter. The yoke
is made up of individually assembled blocks that are stacked without gaps between them.
Spacer laminations between the yoke blocks have slots which together with flow directional
plugs on the collar and in the end plate bypass holes, provide the directed cross flow of
helium for increased cooling in the annulus between the beam tube and inner coil. This
is referred to as "cross flow cooling" .[4] The yoke halves are prevented from lateral
displacement by means of alignment keys at the horizontal split line.
Two .197 in. thick type 304 stainless steel shell halves are welded around the yoke
at each side of the horizontal split line, separated from one another by stainless steel
alignment rails, so that a total of four longitudinal welds are made for each magnet.
During welding, closure of the yoke split line is verified by feeler gauge measurements
through access ports in the alignment rails, which are later welded closed. At each end
of the magnet, a 1.5 in. thick circular end plate is welded directly to the shell ends.
537
EL
STAI LESS s r EEL

lJIMlNATED COlLAR
Figure 1. Cold mass cross section.
Figure 2 is a diagram of the construction of the cold mass at the lead end. The
construction at the return end is similar. Coil end axial restraint is provided at each end
by four instrumented set screws installed into threaded holes in the end plates. They are
loaded against stainless steel pressure plates mounted to the coil ends.
END PLATE
BLOt: TUBE
END COLlARS
COIL END LOADI G SCREW
\ (WITH TRANSDUCER)
t
YOKE MODULE ASS'Y TUBE
Figure 2. Return end.
538
The magnet cold mass is installed into a cryostat with multi-layer insulation and
both a 4 K and 20 K heat shield. The magnet is supported at five locations with "folded
post" type supports which minimize heat leaks. DCA207 utilized a BNL design cable bus;
DCA208 and subsequent magnets feature an MIT design cable bus.
COIL FABRICATION EXPERIENCE
All coils were cured at pressures higher than 7 kpsi due to the consistent use of
oversized cable. Coils were molded to target sizes which would result in molding
pressures of 7 kpsi when nominal size cable is used. The measured sizes of the molded
coils, referenced to a standard of the design size, are given in Table III.
Table III. 15 m SSC Coil Size Data.
Std. Dev.!Max.
Magnet Coil Azimuthal Size1 Dev.
(+mils) (mils)
INNER COILS
DCA207 DCAI2001 -0.5 1.4/4/5
DCAI2002 -0.4 1.3/3.75
DCA20S DCAI2003 -0.5 1.0/3.75
DCAI2004 -2.4 1.3/4.75
DCA209 DCAI2005 -O.S 0.9/3.5
DCAI2006 -1.3 1.0/4.25
OUTER COILS
DCA207 DCA02001 S.3 1.5/5.25
DCA02002 S.2 1.3/4.75
DCA20S DCA02003 9.0 1.3/3.25
DCA02004 8.1 0.S/3.0
DCA209 DCA02005 9.1 1.0/4.0
DCA02006 8.3 1.0/3.25
1 As compared to the magnetic design coil size, with

coil measured while under the design compressive stress.
MECHANICAL BEHAVIOR
Collaring and Assembly History
Assembly shim thicknesses were chosen in each case to provide the desired
azimuthal coil stress after collar assembly. The inner coils used assembly shims larger
than design by 15 mils; the outer coils used shims which matched the design value within
2 mils. (The relationship between measured coil size and required assembly shim
thickness is not fully understood and shall be studied in the near future.)
The coils were mounted in collars and placed in the collaring press which applies
both vertical and horizontal pressure to the collars to make use of the "tapered key"
539
method to collar these coils. The full-hard bronze keys have a three degree taper which
when inserted into the keyways applies tension to the collars. This occurs at the end of
the keying process and results in significantly less loss in coil stress when the hydraulic
pressure is removed than if the tapered keys were not used. The inner coil stresses show
an initial loss amounting to about 10% in the first hour. The rate of relaxation decreases
significantly with time.
After collaring, (and when the coil stress relaxation has stabilized) the fit of the
collar into the yoke was verified by measuring the deflection of the collars containing the
compressed coils. Measurements were taken at 6-inch intervals along the length of the
straight section of the coils and at one inch intervals at both the lead and return ends.
These measurements record the vertical and horizontal dimensions. The fit of the collared
coil in the yoke is inferred from these measurements using the measured inside diameter
of the yoke laminations. The interference or gap for the three magnets as determined by
this method is shown in Figure 3. The average coil stress for each magnet is indicated
along the X-axis. It is seen that although there is some variation of the stress in each
magnet the vertical yoke diametral interference is close to 7 mils in each case and that
there is virtually a line to line fit of the collared coil into the yoke along the horizontal
midplane.
.01 1
.01
.009
.008
DCA207 DCA209
.007 DC'?nA
0
.006
JX1S
.00'1
.D03
.cxJ2
.001
DCA209
o[ DCA207
DC~ fl
-.001
-.002
-.003
7609 9233 9045
Average pre-stress. psi
Figure 3. BNL ASST Magnets. Average collar-yoke interference in the magnet

straight section (5.340" yoke I.D.). The vertical interference is
indicated by 0, the horizontal by D.
Coil Stress and End Forces
The coil stress histories and end force variation will be examined as follows:
• During the assembly steps up to test.

• Cool down and warm up effects.
• Coil Stress and End Force Effects in Testing.
In the discussion of the above effects, data from the assembly and testing of DCA209 will
be used. DCA207 and DCA208 had similar behavior.
540
a. Coil Stress and End Force History
The stress history of the ambient temperature polar stress for the average inner and
outer coil stresses is shown in Figure 4. Starting with the data that was taken a day after
collaring, there is a steady relaxation of coil stress up to the time that the shell was
welded. At the time the shell is welded there is an increase in stress as the coils are
compressed by the tensile force induced in the shell by the welding process. The yoke
horizontal midplane gap was measured to close after the root pass of shell welding for all
three magnets. The rate of stress relaxation increases again at the higher stress levels after
shell welding. The relaxation continues at a low rate and the stresses measured
approximately 80 days after the shell welding and the magnet has completed testing show
that the inner coil stress has dropped from 11,000 psi to 9800 psi and the outer coil stress
has decreased from 9800 psi to 8800 psi.
16C00
14000
12000 -,.---- ~She~eld

Start Test
0- End Test
°fii
lDOOJ WQ- -{]
~ ?-¢.
A-
'V- -0
ui
8000 <>--
/
fa
e::
-4"
....
fI)
6000
4000
2000
o
o 10 ZO 30 40 50 60 70 80 90 100 110 lZO 130
Time, days
Figure 4. Magnet DCA209, Coil Stress at Ambient Temperature. The inner
layer is indicated by 0, the outer by O.
The end force history during the same period is somewhat more complicated and
is shown in Figure 5 . The initial setting of the load screws that support the ends of the
inner and outer coils is about 4000 lbs. per end. In order to complete the assembly of the
cold mass a 7/16 inch wall cylinder is welded onto the end plate as shown in Figure 6.
The purpose of this extension is to provide adequate space for the instrumentation boards
and connections that are used in these model magnets. During this welding process the
end force increased by about 10,000 lbs. on the lead end of the coil and by 6000 lbs. on
the return end of the coil. This was believed to be caused by warpage of the end plate
during the welding process. Referring to Figure 5 again, it is seen that the end force stays
about the same until installation in the test stand where there is about a 1500 - 2000 lb.
decrease. However, after the magnet had completed testing the end force at both the lead
and return ends had increased to the 15000-20000 lb. range. This behavior was caused by
a tendency of the end force to increase during testing. Figure 7 shows the end force for
magnet DCA209 as a function of time throughout the first cold test cycle. Short arrows
on the plot indicate excitations greater than 1000 A and long arrows represent excitation
541
whereby the magnet was quenched. From the plot one observes that with each magnet
excitation there is a corresponding increase in the end force. A possible explanation for
this is due to a retention of a portion of the axial Lorentz force produced during magnet
excitation at the coil ends. In this case, the Lorentz forces expand the coil axially.
Following excitation, the friction between the collars and the yoke laminations may prevent
the collared coil assembly from returning to its original position and therefore increase the
end load.
.~
-5OCO
~
iii
0.
I
~a:
\0_
- I!XXXI D
Iii
-Ign)
0,\
-2Oll)
-2SOO)
-~ o 100 l~
Time. dny"
Ao Initial Sellmg B: Exlension Tube Weld
C: Slarl Tesl D:End Tesl
Figure 5. Magnet DCA209, End Force at Ambient Temperature. A: Initial
Setting; B: Extension Tube Weld; C: Start Test; D: End Test. The
lead end is indicated by D, the return end by o.
Figure 6. Lead end extension.
542
150
125
Z
C- lOD
(I)
(,)
1-0
0
r:..
"0
J::
75
r.l
'iii
.,.>
0
E- 50
I I 4. I • 7 • • 10111111
25
0
0 20 40 60 BO
Figure 7. Magnet DCA209 1st Test Cycle Return End Force
vs. time (hours).
b. Cool down and warm up effects
Table IV presents a summary of the changes in coils stress and end forces for the
three magnets tested from the start to the completion of testing. There is some consistency
in the amount of stress loss in the inner and outer coils when cooled from ambient to
operating temperature. The coils typically loose about 4000 psi. The end forces are not
as consistent from ambient to operating temperature. In some cases there is a small
increase, in others a small decease. Characteristically it appears that they remain
substantially the same through cool down. In most cases the end forces increase during
testing as mentioned above; however, the return end force for DCA207 did not show this
effect.
c. Coil stress and end force during magnet excitation
Data from the testing of DCA209 will be used to illustrate the effect of magnet
excitation on the behavior of the coil stress and end forces. The variation of inner coil
stress with current is shown in Figure 8. The polar stress indicated by the gauges in each
quadrant is plotted as a function of magnet current squared. The serial number of the
gauges is shown in the legend. Note that the stress deceases somewhat quadratically with
current to about the operating point at 6600 A. At that time the poles of the coils are still
under compression. This run was made under sub-cooled conditions so that the magnet
could be powered above 8000 A. It is seen that as the current rises above 7000 A, the
slope of the curves start to flatten out indicating that the coils are becoming unloaded at
the poles. This gives a good method of verifying the cold calibration of the gauges since
the flat part of the curves indicate zero polar stress. The offset shown of several hundred
psi is the error in the compensating gauges resistance tracking with the active gauges.
Thus one can correct these curves to indicate zero stress at above - 7500 A. Note that
these magnets are capable of running well above the point where the poles of the inner
coils have become unloaded without quenching.
543
Table IV. Summary of Stress and End Forces from Magnet Tests.
DCA207 DCA208 DCA209
Inner coil stress (psi):

l. Before cooldown 11292 9370 10295
2. After cooldown 6617 5317 6428
3. Change 4675 4053 3867
4. Percent loss 41.4% 43.3% 37.6%
5. Warm, after test 10307 8919 9742
6. Delta for test -985 -451 -553
7. Percent change for test -8.7% -4.8% -5.4%
Outer coil stress (psi):
1. Before cooldown 7399 8117 8994
2. After cooldown 3988 3596 4778
3. Change 3411 4521 4216
4. Percent loss 46.1% 55.7% 46.9%
5. Warm, after test 7045 7851 8898
6. Delta for test -354 -266 -96
7. Percent change for test -4.8% -3.3% -1.1%
Lead end force (pounds):
2. After cooldown 10685 11457 11722
3. Change 3808 -1482 -1821
4. Percent increase -26.3% 14.9% 18.4%
5. Warm, after test 16652 14410 15868
6. Delta for test 2159 4435 5967
7. Percent change for test 14.9% 44.5% 60.3%
Return end force (pounds):
2. After cooldown 14262 11415 7090
3. Change 1333 -40 -743
4. Percent increase -8.5% .4% 11.7%
5. Warm, after test 13370 17214 19356
6. Change for test -2225 5839 13009
7. Percent change for test -14.3% 51.3% 205.0%
The behavior of the outer coils is somewhat different as shown in Figure 9 in

which the outer coil quadrant stresses are plotted as a function of current squared. The
Lorentz forces have relatively little effect on these coils and the stress only decreases a
small amount with current. However, there is an interesting effect here. The curves of
coil stress vs. current squared seem to show a dual slope. Initially the average slope for
the gauges is -7.07 x 10.5 psi/A2. However at -4500 A, the slope becomes -2.86 x 10-5
psi/ A2. The interpretation is that there is a slight gap between the collars and the yoke
near the mid plane. When the magnet is energized a horizontal component of the Lorentz
force is produced which tends to increase the horizontal deflection of the collars.
However, as soon as the collars contact the yoke, the bending stiffness of the yoke
decreases the amount of horizontal deflection with force (and resulting stress decrease).
544
Thus, the slope of the curves become less. In this case, we assume that the collared coil
is firmly supported by the yoke at 4500 A (2 x 107 A2), and above.
The change in end force with excitation is shown in Figure 10 in which the total
end force at the lead and return end are plotted as a function of amperes squared. It is
seen that these curves are quite linear and that the total end force change up to 8400 A is
about 10,000 lbs. at the lead end and 6000 lbs. at the return end.
8000
,
7000
II,..
6000
~
---- ."'"
5000
OJ
0.. 4000
~ ..... ~
, ~ ~~
UJ
UJ 3000 K
~
~i
W
0:
I- 2000
"-
~
~~
~
UJ
1000
Ida,
~
~ .~
~
0 ---- -- -- --- - - -- - - - - - - - - - - - - - - - - - - - - -
- - --
-1000
o 1000 2000 3000 "'000 5000 6000 7000 8000 9000
CURRENT, AMPS
Figure 8. Magnet DCA209, inner coil stress (3.5 K) vs. I.
8000
7000 ~
""-
6000 ~
...........
~
--.,
r---- r--.
5000
OJ 'I
0.. 4000
, ~
UJ
3000
~ 't ~'"
UJ
~ I!.--s-.:
W ~
0: If-
I- 2000
UJ
L--2
1000
0 ----- - - - - - - - - - - - - - -- --- ---- - -- - - - - - -----

-1000
o ~7 ~7 ~7 ~7 ~7 ~7 h7 ~l
CURRENT, AMPS**2
Figure 9. Magnet DCA209, Outer coil stress (3.5 K) versus 12.
545
o ---- - - - - - - -- - - - ---- - -- --- - - - -- - ---- --- --
-4000
-8000
ui -12000
.0
----
~ -16000
:su.. -20000 ~ ~
w
u
~
o
z -24000
-s-...,
~ t--s. -----r---- r----f<_
W
~
-28000 I-BBss
-32000
-36000
o 1e7 2e7 3e7 4e7 5e7 6e7 7e7 8e7
CURRENT, AMPS**2
Figure 10. Magnet DCA209 lead end force (O) and return end force (X) vs. J2
(3.5 K).
QUENCH ANALYSIS
The quench test procedure for all three magnets consisted of ramping up the magnet
current at a specified rate until a quench was generated, and performing this until the
quench currents achieved had reached a four-quench plateau, within an allowable range
of 30 A, and the quench origins were at locations that implied that they were conductor-
limited. For the first quench at each testing temperature, a strain gauge run to quench,
where the ramp was stopped at specific current steps to take strain gauge measurements
on the way to quench, was used instead of a continuous ramp. In addition, at the end of
each set of plateau quenches, strain gauge measurements were made at current steps almost
up to quench plateau current and then on the way down again. Then, after a warm-up to
room temperature and re-cooling (thermal cycle), a quench plateau was again established
at 4.35 K. In the case of DSA207 and DSA209, quenching was also performed at the
lower temperatures of 3.85 K and 3.5 K, where the central field is about 8.1 T and 8.5
T, respectively. The quenching at lower temperatures is done to test the limits of a
magnet's mechanical performance under the stress of the higher magnetic fields possible
at the lower temperatures. DCA208 was not tested under these conditions.
Figure 11 shows the quench history for the magnets tested. For brevity, the plot
shows only the quenching at 4.35 K and does not exhibit quenches done at lower
temperatures or those done at various ramp rates to study ramp-dependent effects. These
results will be discussed later. All three magnets exhibited only minor training as they
went quickly to plateaus which were close to values predicted from measurements of short
sample cable. Small variations about the mean plateau quench current are attributed to
cryogenic temperature fluctuations. All the plateau quenches were located in a straight
section of a pole turn (turn 19) of an inner coil; this is expected for conductor-limited
quenches, since these are the regions of highest magnetic field where the critical field can
be expected to be reached first. It also should be noted that none of the magnets
experienced quenches in the outer coils under any of the conditions imposed during the
different testing regimes. As can be seen from the plot, DSA207 had two training
546
quenches at 7358 A and 7307 A, both more than 10% above the SSC 20 TeV operating
current of 6618 A, at the start of testing before achieving a plateau of conductor-limited
quenches. Subsequent plateaus at two thermal cycles were re-established without any more
training quenches. Mean plateau current was 7407 A, about 0.5 % lower than the value
of 7442 A predicted from short sample cable measurements. The three plateaus were at
slightly different mean values due to slight differences in the test temperature between
thermal cycles. All plateau quenches were conductor-limited and occurred in the straight
sections of the pole turns, as is expected. The conductor-limited quench plateau was
achieved in DSA207 with the standard test ramp rate of 16 A/s.
,, I:, I ,,, , ,,, I

B -
,,,
,,,
,
,, ,,, -
, coo ,"" 010 c
,,
[JI 000
o a 0
, ,,
0 0 0:0 0 0 0:0 0 0 0 o ,0
0
D COD 0:0
I I
I
0 0
,, ,
-------,-----,------- -------t--------
I
, ,, -------l---!---~
,,
I
6 r-
,
I
I
I
I
,20 TeV :
~
I
I I
I I ! !
.
-=t: i ! ! i
- "
1 1
-
a 1
::l 01
t.> 4 s
..
.c
g tl
::l
01 i i
I
i
I I
2 - ,
I
I ,
I
I
I
,
I
I
-
! I !
DCA20? DCA208 DCA209
i
I ,i I
,
i
I
i
I
I
o
o 10 20 30 40
Quench Sequence at 4.35K (nom)
Figure 11. BNL-Built 50 mm, 15 III SSC Dipole Quench Tests at 4.35K
During the second cycle of tests for DCA207, quench studies were also done at the
lower temperatures of 3.85 K and 3.5 K, after the second plateau at 4.35 K had been
established, as mentioned above. In both temperature regimes, the magnet achieved a
plateau without any training quenches. At 3.85 K, the plateau mean current was 8073 A,
0.3 % above the short sample prediction of 8047 A; at 3.5 K, the plateau mean current was
8422 A, 0.1 % below prediction. All quenches were in the lower inner coil ramp-splice
section, a region of high field.
From Figure 11, it is also seen that DSA208 was just as well-behaved. There was one
training quench at 7383 A before achieving a conductor-limited plateau; after a thermal
cycle warmup and re-cooling, the magnet once again established a plateau with one
training quench, this time at 7407 A. Both training quenches were about 12% above the
SSC 20 TeV operating point. The plateau mean was 7571 A, 0.6% above the short
sample prediction of 7524 A, and all plateau quenches were located in the upper inner coil
pole turn left straight section. The ramp rate used to achieve the plateau in this magnet
was 4 A/s. At the 16 Als ramp rate, the quench plateau was uniform but about 100 A
lower in current and in a turn further in toward the midplane. Again, due to cryogenic
test temperature differences between thermal cycles, the two plateaus were at slightly
different mean currents.
At 4 Als, after two training quenches, DCA209 reached a plateau of 7405 A, 1.4%
below the short sample prediction of 7508 A and not conductor-limited, as evidenced by
547
its non-pole turn location. It was found that at 1 A/s, (point A), the magnet exhibited a
conductor-limited plateau at 7488 A, 0.3% below the short sample value and located in
the same cable region as in DSA208. The lowest training quench was 7102 A, 7% above
the 20 TeV operating point of 6618 A. Due to differences in the conductor characteristics
among the three magnets, the maximum ramp rate at which a conductor-limited quench
plateau could be established was different for all three magnets. In general, at higher
ramp rates, heating due to eddy currents generated in loops between the wires in the cable
result in quenches located in cable turns further in toward the midplane rather than in a
pole turn and also, because of lower field strength, in lower quench currents. It is
suspected that a lowering of inter-strand resistance during the coil curing process may be
the cause of this effect, and it is presently under investigation. The ramp-dependent
behavior of all three magnets was studied by performing a series of quench tests at
increasing ramp rates up to 300 A/s for each magnet. The results are shown in Figure 12.
Each magnet shows a significantly different behavior, both by the way quench current
varies with the ramp rate and by the maximum value at which a conductor-limited quench
plateau is achievable. DCA209 showed the unexpected behavior at a noticeable ramp rate
dependence in going from 4 A/s to 1 A/s as shown in Figure 12, point A. It is interesting
to note that for this cable, the ramp rate dependence is less severe at high ramp rates than
at the low one. Studies of this effect are continuing.
6
zo TeV
~
:!.
;:
.
I!
u" 4
0_
.
.e.
,,-
u
c:
"
Q1
• DCA207
2
100 200 300 400

Ramp RaUl (A/_)
Figure 12. 15 m, 50 mm SSC Dipole Magnet Ramp Rate Studies at 4.35 K.
REFERENCES
[1] J. Kuzminski, et al., "Test results of BNL built 40-mm aperture, 17 m-Iong SSC
Collider dipole magnets", IEEE Trans. Magnetics 28, p.311 (1992).
[2] P. Wanderer, "A Summary of SSC dipole magnet field quality measurements",
paper to this conference (Fourth Annual International Industrial Symposium
on the Super Collider, New Orleans, LA, March, 1992).
548
[3] R.C. Gupta, et aI., "SSC 50 mm Dipole Cross Section", Supercollider 3, p.387,
Plenum Press, NY 1991.
[4] R.P. Shutt and M.L. Rehak, "Transverse Cooling in SSC Magnets",
Supercollider 2, p.209, Plenum Press, NY 1990.
549
RESULTS FROM A PARTIAL LIFETIME TEST OF A
40-MM-APERTURE, 17-M-LONG SSC MODEL DIPOLE
P. Radusewicz, A. Devred, T. Bush, R. Coombes, J. DiMarco, C. Goodzeit,

J. Kuzminski, T. Ogitsu, J. Potter, M. Puglisi, P. Sanger, R. Schermer,
J. Tompkins, Y. Yu, Y. Zhao, and H. Zheng


M. Rehak, E.P. Roher, W. Sampson, R. Shutt, R. Thomas, P. Thompson,
P. Wanderer, and E. Willen

Upton, NY 11973 USA
M. Bleadon, R. Hanft, M. Kuchnir, P. Mantsch, P.O. Mazur, D. Orris,

T. Peterson, and J. Strait

J. Royett, R. Scanlan, and C. Taylor

Berkeley, CA 94720 USA
ABSTRACT
A 40-mm-aperture, 17-m-long Superconducting Super Collider (SSC) model dipole

was assembled at Brookhaven National Laboratory (BNL) and tested initially at Fermi
National Accelerator Lab (FNAL) and later at BNL. At BNL an extended cycle test was
devised to examine the magnet's performance through numerous cold tests and thermal
No. DE-AC35-89ER40486.

cycles. This paper discusses the magnet's mechanical and quench performance and magnet
field measurements during the tests.
INTRODUCTION
One area of interest in an SSC dipole magnet is in its performance during a test cycle
approximating that which will be seen by an actual dipole within the collider. However,
planning and carrying out such long tests on a magnet would require a great amount of time;
therefore, test plans have been derived to examine a "Partial Lifetime" test for a collider
dipole magnet. Such a test has been developed and carried out at BNL, and it has become
known officially as the "Extended Cycle Test" (ECT).
In this paper we will discuss magnet DD0028, which completed the first ECT
performed on a SSC dipole. Construction and magnet features will be reviewed first.
Discussion will then tum to the test plan carried out at FNAL and BNL. Emphasis will be
placed on number of quenches, power cycles, thermal cycles, etc., with comparisons to
actual "Life Time" expectations of a collider dipole magnet. Finally, we will summarize
quench performance, mechanical stability, and field quality along with any changes
observed due to the extended period of testing on the magnet.
MAGNET FEATURES
DD0028 was constructed 3 with the conceptual cross section design designated
C358D.1,2 It is a 40-mm-aperture dipole magnet with a magnetic length of 16.6 m. The
inner and outer conductors have copper-to-superconductor ratios (Cu:SC) of 1.3:1 and
1.8:1, respectively. The cables are wrapped in two layers of insulation, with the first layer
being double wrap, 50% overlap, I-mil Kapton. The second layer is a single wrap of 4-mil
pre-pregnated fiberglass with an epoxy content of 24%.
The cables are then wound to form the inner and outer, upper and lower half of the
coil. The coil ground insulation consists of Teflon (l mil), Kapton (5 mil), strip heater
(15 mil), and two more layers of Kapton (5 mil each). Mechanically seated shims are
located at the poles, with the inner and outer coil shims being 25 and 32 mil, respectively.
The coils are collared using alternating left/right (spot-welded) Armco
Nitronic 40 collars using the tapered key method. The collared coils are then placed in the
yoke assembly, which consists of horizontally split modules. The long, straight section of
the magnet consists of modules made of iron laminations, whereas the modules making up
the end sections are stainless steel. The yoke assembly is then enclosed in a horizontally
split stainless steel outer shell that is manually clamped together, then welded along the
magnet's length.
The magnet is also fully instrumented with voltage taps and various strain gauge
assemblies. Voltage taps, used for quench detection and analysis, are located on the quarter
coil leads, on the inner-outer coil ("ramp") splice region, and on the first three turns at the
pole of the inner coils. Special strain gauge collar packs are included to measure both the
inner and outer coil stresses. Two packs are included, one at the location of the azimuthal
high point in the coil size and the other at the corresponding low point. Four strain gauge
assemblies (dubbed "bullets") are also located at the magnet's return end to measure the
force that the coil exerts against the end plate.4
TEST PLAN
After assembly of the DD0028 cold mass at BNL, the magnet was transported to
FNAL, where it was placed in a cryostat and run through the first series of tests. The test
552
plan at FNAL consisted of cold testing followed by thermal cycling and retesting. During
the cold tests, the magnet was brought to liquid helium temperatures, followed by initial
cold electrical test prior to any quench testing. After all checks were complete, a series of
quenches was performed at a rate of 16 A1sec to examine the magnet's quench performance.
Quench testing continued until four conductor-limited quenches were established.
Following quench testing, magnetic measurements were made. These typically included
2000-A axial scans, NMR/HaIl measurements, time-dependence measurements, and
magnetization loops. Once the cold test program was complete, the magnet was brought
back up to room temperature, then recooled for more cold testing. After four cold test cycles
at FNAL, the magnet was returned to BNL for further cold testing.
The test plan used at BNL-the Extended Cycle Test--consisted of weekly test cycles
whereby quenches and quench performance would be observed, followed by a series of
power cycles, strain gauge measurements. and magnetic measurements. Warm-ups and
cooldowns were planned for the weekends for more efficient use of time.
The effect of "conditioning" on the magnet's training performance was also examined.
Training occurs when a magnet quenches below its expected short sample limit and
continues until a conductor-limited quench is established. These training quenches are
usually energy-deposited quenches that occur when there is movement of the conductor
within the collared coil assembly. These movements, coupled with frictional forces, tend to
cause frictional heating that may lead to a quench. Conditioning is the procedure in which
the magnet is cooled to a temperature lower than test temperature. and a series of excitations
is done to a current of 6.8 kA, just above the operating current of 6.5 kA. The expectation is
that if any coil movement is to occur within the magnet, it will occur at these lower
temperatures where the added thermal margin may allow for movement of the collared coil
assembly. The collared coils can then position themselves in an equilibrium position
without quenching. Therefore, during 4.35-K quench testing. any large conductor motion
has already occurred, and the chance of training during quench testing is reduced. To
examine the effects of conditioning, alternate thermal cycles included a conditioning period.
TEST SUMMARIES
FNAL testing consisted of four thermal cycles, with cooldown schemes shown in
Table 1.
Table 1. Cooldown Schemes in FNAL Testing.
Cooldown Restricted Unrestricted Conditioning
x X
2 X
3 X X
4 X
Restricted cooldowns consisted of lowering the magnet's temperature while

maintaining a given temperature rise across the magnet. The ~T restriction at FNAL was
125 K. During unrestricted cooldowns the ~ T restriction is removed, and the magnet is
553
cooled to 4.3 K as quickly as possible. Cooldown times for restricted cooldowns are
generally around 24-32 hours, while unrestricted cooldowns are approximately
12-14 hours. A summary ofthe four test cycles is shown below:
Quench Testing (47 total quenches):
I) initial quench during conditioning.
2) 27 quenches at 4.35 K.
BNL's ECT consisted of eight full test cycles, with conditioning beginning with the
first thermal cycle, then alternating between thermal cycles. Test summaries are shown
below:
Quench Testing (46 total quenches):
1) 31 quenches at 4.35 K with Cold Bore Tube.
2) 15 quenches at 4.35 K with Warm Bore Tube.
Power Cycles (100 A --7 6500 A --7100 A):
1) 997 total at a rate of 100 Alsec (4.35 K).
2) 1332 total at a rate of 200 Alsec (3.85 K).
Prior to the last thermal cycle of cold testing at 4.3 K, the magnet was cycled eight
times between a temperature of 300 K and 77 K, where 90% of the thermal contraction
takes place. This was done to examine the effect of multiple thermal contractions and
expansions on the mechanical and quench performance.
The totals from the partial lifetime test results include: 2329 magnetic cycles (25%),
20 thermal cycles (67%), and 93 total quenches (62%). Percentage of lifetime durability
requirements are indicated in parentheses.
QUENCH TEST RESULTS
To examine the magnet quench behavior, the magnet's temperature is lowered to 4.3 K
and energized at a rate of 16 A/sec until the magnet quenches. This is repeated until a
plateau of four quenches is obtained. Plateau quenches are typically conductor-limited
quenches, with quench currents remaining stable from one quench run to the next. Figure 1
shows a summary of the quenches conducted on 000028 throughout its test history.
Influence of Conditioning
During the first conditioning cycle at FNAL, the magnet quenched on one of the
excitations to 6800 A. Following a warm-up to 4.35 K, the magnet exhibited one further
training quench, then attained plateau for the remaining quenches. After completion of the
first thermal cycle, training was observed in each test cycle where conditioning was left out.
In no thermal cycle where conditioning was done was there any training. It therefore
appears that an initial conditioning cycle prior to quench testing improves the training
performance of the magnet following a thermal cycle.
Plateau Quench Currents and Origins
The quench currents and origins showed good stability throughout the testing at both
labs. All FNAL plateau quench origins were located in the pole turn, right straight section,
upper inner coil, near the middle of the magnet, while BNL quench origins were located
similarly in the lower inner coil. These plateau quenches occurred at the region in which the
field is at a peak, as expected for conductor-limited quenches. Differences in quench
554
locations between FNAL and BNL data could be attributed to differences in cryogenic
conditions during testing. Helium flow rates at FNAL were approximately 45-50 g/sec,
whereas BNL flow rates were 160-165 glsec.
Plateau quench currents from one thermal cycle to the next also showed good stability.
There was no indication of decreased quench performance and, therefore, no signs of
degradation of the conductors from thermal cycle to thermal cycle.
8000 '-~--'I---'~---'---'I---'--~~-'I--------------'
I I I Quencb Locations
I I I
I I I o Upper Inner
I I I o Upper Outer
I I I
I I I o Lower outer
I I I l( Lower Inner
I I I
I I I
I I I
I I J I
S 7000 I-- :
en I I I I
: : I I I: I : -
~ ~ <>o<>~ 00 <>0:0<><> i 00' (0):0:( i )l;):():():( 1:tt:):():( i ):001: i )::():( i ):():():i)):(o <>):{):( 1):{):();I:):(
..., io I I 10 I I I I I 0 I I
.:: : 10 I I 0 I I I I I I I
'........" --~---~---t--f---
I I I
---io---t __ ~ ___ L_l----i----1-~T~--
I I I ID I I
;:J I I I I I I I I I
u I I I I I I I I I
I I I I I I I I I
.r:: I I I I I I I I I
g
Q)
6000 r-- I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
-
I I I I I I I I I
;:J I I I I I I I f I
01 I I I I I I I r I
I I I I I I I I I
I I I I I I I I I
I I I I I I I I I
uc u 877K
FNAL TestinB BNL Test.iD8

5000 L-______________ ~ ____________________________ ~
Quench Sequence
Figure 1. Quench history of DD0028. Conditioning cycles indicated by "C," unrestricted cooldown by "U."
MECHANICAL TEST RESULTS
Inner and Outer Coil Stress Histories
Figures 2 and 3 show the inner and outer coil stress as measured during different
stages of testing. For each thermal cycle of testing, a set of 10 strain gauge reads are made
with zero current on the magnet prior to cooldown, after cooldown, and again prior to
warm-up. The figures show the average of these 10 reads for each initial warm, initial cold,
and final cold reads for a test cycle. Indicated in the plot is also the point at which the
temperature was cycled eight times between 300 K and 77 K.
The initial reads prior to any testing indicate that the magnet was assembled with
similar levels of inner and outer coil stress and that the change in prestress throughout
testing decreases slightly from start to finish. The inner coil stress shows a loss of
approximately 8 MPa in prestress. while the outer shows a loss of 2 MPa in prestress by the
time the last warm reads are made.
During cooldown, both the inner and outer coils exhibit a loss in prestress of 40% and
39%, respectively. This loss in prestress remains fairly stable throughout testing, leaving
consistent and acceptable levels of coil stress at 4.3 K. Coil stress values showed only slight
decreases during excitation, with 4 MPa observed for the inner coils and the outer coils
remaining stable. It is also observed that the coil stresses show further decreases with
testing, which is evident in the final cold reads of each test cycle. The inner coils show a
decrease in each cycle (except for three cycles), while a decrease is seen in each cycle for
the outer. These further decreases in coil stresses during testing at 4.3 K typically occur
during the first few excitations of the magnet to high fields; thereafter they remain stable.
555
70
Each thermal cycle contains:
--;; 60
Co Initial 1farm Read
+ Initial Cold Read
e
0..
l:( Final Cold Read
~ 50
OJ
b
[fl
40
0
u
....
OJ
<=i
30
.8
OJ
20
""....'"
OJ
>
-0:
10 B Thermal
1"'1 Cycles to 77K
0::
Tested at FNAL Tested at BNL
0
History Sequence
Figure 2. Inner coil stress history of return end gauge pack. Each thermal cycle contains an initial warm, cold,
and final cold read.
70
--;; o Initial Warm Read

60 + lnitie.l Cold Read
e
0..
)( Final Cold Read
'"'"
OJ 50
b
[fl
'0 40
u
....
OJ
;:; 30
0
OJ
t
20
""....'"
OJ
>
-0: 10 • Thermal
1"'1 Cycles to 77K
0::
0
History Sequence
Figure 3. Outer coil stress history ofreturn end gauge pack. Each thermal cycle contains an initial warm, cold,
and final cold read.
End Force Histories
Figure 4 shows the total end force plotted in the same way as the inner and outer coil
stress. For each thermal cycle of testing. the total end force prior to cooldown. following
cool down. and prior to warm-up are plotted for both FNAL and BNL data. Differences
between FNAL and BNL data can be attributed to the horizontal test setup at BNL. During
setup of the test stand. the bore tube is welded to the extension bellows at both the lead and
return end. Following this welding, it is observed that the end force increases. This has also
been observed in previous magnets tested. where reads taken prior to setup show lower
values of end force than those after welding. Therefore. it is not believed that these
differences in end force observed between the two labs are due to either transportation or to
an error in the calibration.
The initial end force levels prior to cooldown for each cycle show a considerable
increase in the first three test cycles at FNAL. whereas the last six test cycles at BNL show
end force levels that remain rather stable. This initial increase in end force is due to a
"ratcheting" effect that takes place during cold testing. This is seen by comparing the initial
556
cold reads with the final cold reads of each test cycle. During testing, it is observed that for
each excitation of the magnet to high fields, there is a corresponding increase in end force.
This increase typically occurs following the first few excitations, after which the magnet
remains stable through the test cycle. Ratcheting occurs when the magnet is excited to high
fields, where the Lorentz forces tend to expand the coils radially and axially outward. When
the current is brought down to zero, the collar-yoke interference tends to "hold" the collared
coil assembly at the position it reached during excitation rather than allowing it to return to
its original configuration prior to excitation. Therefore, an increase in end force is observed.
As mentioned above, this increase is observed only after the first few excitations during a
test cycle.
50,-------------,----------------------,
<) Initial Warm Read
+ Initial Cold Read

:It Final Cold Read
40
Z
C
...<>'" 30
....0
.
~ ~
"0 ~ ~ ~ 9
.:., ..):{
0:
r:.:l •" ~~r + "
'itl 20
+'
0
~ r
"
Data lost B Thermal
'+' during 3rd TC Cycles to 77K
10

0
History Sequence
Figure 4. Total end force history of the return end. Each thennal cycle contains an initial wann, cold, and final
cold read.
MAGNETIC MEASUREMENT TEST RESULTS
Throughout the testing of DD0028, magnetic measurements were carried out in order
to examine the effects of quench testing, power cycles, and thermal cycling on the multipole
components of the magnetic field. An axial scan of the magnetic field was made with a
rotating coil assembly known as a "MOLE" at a current of 2000 A following a pre-cycle to
6500 A. The coil is 1 m in length, allowing 17 positions along the magnet's length to be
measured. The data obtained from the MOLE is used to calculate the normal and skew
components of the magnetic field. Therefore, measurements taken from each thermal cycle
can be examined for any changes that may occur throughout testing.
Generally, one focuses on the lower-order normal and skew terms of the magnetic
field. These lowest-order terms, quadrupole and sextupole, typically show the largest effects
due to construction errors and any changes that may occur after the original construction.
Any deformations of the collared coil assembly that may occur throughout testing will
therefore show up as a corresponding change in these terms.
Axial scans were done at FNAL before and after the initial quenching of the magnet.
Axial scans were then repeated the following thermal cycle and in each thermal cycle during
the ECT at BNL except for the sixth cycle. Axial scans that were done at BNL were
followed up with a second scan in order to check the repeatability of the measurement.
Figure 5 shows the sextupole component of the magnetic field for each axial scan. Each
point represents the average along the magnet's length where the end regions and strain
gauge collar pack locations have been excluded from the average. Differences betweep
FNAL and BNL data can be attributed to the fact that different measuring devices were
used. At FNAL, measurements were made with MOLE Bl, while at BNL measurements
557
were made with MOLE Dl. Calibration of MOLEs Bl and Dl is necessary to resolve the
question of whether it's really the MOLE or the magnet.
During the seventh thermal cycle, there is also an apparent shift in b2 as compared to
axial scans measured in the 5th and 8th thermal cycles. These scans were made using the
same MOLE, and it is believed that the prior excitation loop caused a different initial b2 for
these runs.
1.0
FNAL BNL o 1st Z-acan in a
Data Data thermal cycle
00 2nd Z-8can done
for repeatability
0.8 - -
-'"""'
-'"""'
I No Z-BCIlD. duri.ns
S
C)
Bm. 8th TC
... 0.6 - -
l
0
I
....
0
0
)(
0 <> 0 0
0 <>
0 <> 0 0 <>
...,rn <>
'8 0.4 - -
2-
,.Cl'"
5
;
"~
~
01
r
8 Thermal
Cycleo to 77K
.
":!! ]~
0.2 t-- iI -
" ,g
! .'!
0.0
History Sequence
Figure 5. Sextupole history of DD0028. Each point represents an average taken along the magnet's length.
End regions and strain gauge collar pack locations are omitted from the average.
REFERENCES
1. G. Morgan, "C358A-A New SSC Dipole Coil and End Design," SSC-MD-155, March 1987.
2. G. Morgan, "C358D: A Revision of the SSC Coil Design C358A," SSC-MD-183, March 1988.
3. A. Devred et aI., "About the Mechanics of SSC Dipole Magnet Prototypes," AlP Conference Proceedings
Series, The Physics of Panicle Accelerators, January 1992.
4. C. Goodzeit et aI., "Measurement of Internal Forces in Superconducting Accelerator Magnets with Strain
Gauge Transducers," SSC-N-547, September 1988.
558
CONSTRUCTION AND TEST RESULTS FROM 1.8 m-LONG, 50 mm
APERTURE SSC MODEL COLLIDER DIPOLES t
J.F.Muratore, M.Anerella, J.Cottingham, G.Ganetis, M.Garber, A.Ghosh,

A.Greene, R.Gupta, A.Jain, S.Kahn, E.Kelly, G.Morgan, A.Prodell, M.Rehak,
E.P.Rohrer, W.Sampson, R.Shutt, R.Thomas, P.Thompson, P.Wanderer,
and E. Willen

e.Goodzeit and P. Radusewicz

Dallas, Texas 75237
ABSTRACT
A series of four l.8 m-long model sse dipole magnets has been constructed at
BNL in order to test the new, wider 50 mm aperture sse design prior to measurements
of the full length 15 m-long dipoles. Except for the length, these magnets have the same
baseline design and features as the full length models. All four short magnets were tested
successfully, have exhibited reliable and uniform behavior, and have exceeded the 10%
quench margin above the sse operating point. In addition, some interesting results have
led to further ongoing investigations into the effects of ramp rate on quench performance
and magnetic multipoles, and into the effect of end force preload on training. This report
will describe the basic features of the 50 mm design of these magnets and the experimental
methods used in the testing of the magnets. Results from quench tests, ramp rate studies,
strain gauge measurements, and magnetic field harmonic measurements will be presented.
INTRODUCTION
In order to raise the quench margin to 10% of the sse design operating field of
6.7 T and to improve magnetic field quality, it was decided at the beginning of 1990 to
!Work supported by the U.S.Department of Energy.

increase the coil inner diameter from a 40 mm to a 50 mm aperture. An extensive
program of short (1.8 m) and full length (17 m) 40 mm aperture dipole magnets had
already been well underway and had proven a successful design with the C358D cross
section, which makes use of a two-layer cos (J coil geometry to generate the magnetic
field. The basic design and concepts of those magnets have therefore been retained in the
new 50 mm aperture W6733 design while incorporating a number of improvements, in
addition to the wider aperture, in order to extend the SSC operating point margin, improve
quench training performance, and provide good field qUality. The W6733 cross section,
shown in Fig. 1, employs wider cables with larger numbers of strands: 30 in the inner
cable and 36 in the outer. There are now 19 turns of cable in each inner coil and 26 turns
in each outer coil. Because of the larger coils, the collars and yoke also had to be
redesigned. Details of the new design and the changes made can be found elsewhere. [1]
Magnets built by BNL and FNAL are the same except for small differences in the yoke
geometry, since those built at FNAL have a vertically-split yoke and the BNL dipoles
employ a horizontally-split yoke. Four short 1.8 m-Iong model dipoles and seven full
length 15 m-Iong dipoles have been constructed as part of the BNL 50 mm program. The
last five of the long magnets have been built by Westinghouse, with BNL assistance, to
initiate their contribution to the production prototype program in industry. The four 1.8
m-Iong models have been assembled and tested at BNL in order to try out the new 50 mm
aperture design prior to measurements of the full length magnets. Except for the length,
these short magnets have the same baseline design and features as the long magnets.
STAJNL SS STEEL
SUPPORT SHELL
8 ' A... lUBE
SIAl SS S" lL
LAM' "'lO COL R
Figure 1. W6733 Cold Mass Cross Section.
560
MAGNET DESIGN AND CONSTRUCTION
General Features
All four magnets share the same basic design features, with variations among them
due only to differences in conductor and cable wrap characteristics. Common features
include the W6733 coil cross section design, already described and shown in Fig. 1, and
all four magnets have identical coil, collar, and yoke geometries, and incorporate all
improvements made in the 50 mm design. In order to increase mechanical stability in the
ramp-splice between the inner and outer coils, that section of cable has been solder-filled,
the ramp section of cable was lengthened, and the adjoining short straight section of cable
in the pole turn was epoxied under pressure to the adjacent cable turn. The entire coil is
clamped firmly in place by a collar constructed of 621 MPa yield strength Nitronic 40
stainless steel laminations. These are spot welded in pairs and then assembled into 15.24
cm (6 in.) long collar packs by stacking the pairs in alternating left/right symmetry in
order to reduce axial twisting of the laminations during collaring. The laminations are 1.5
mm thick and are configured for phosphor-bronze keys of tapered cross section to reduce
the pressure needed to close on the coil. Mechanically-seated, collar pack length brass
shims are placed at the pole-most turns of each quarter coil to bring the overall coil size
to design specifications. One collar pack is designed with beam-type strain gauge
transducers to measure coil azimuthal prestress during all phases of magnet assembly and
testing, and it is located at the axial position of minimum coil radius.
Enhancement of the magnetic field and further mechanical stability is provided by
a monolithic, horizontally-split yoke, made of 16 gauge, extra-low carbon steel
laminations. The top and bottom halves of the yoke are aligned at the midplane with
stainless steel keys, whose size and location are determined by magnetic field
considerations. The yoke modules at the ends are made of stainless steel laminations and
extend over the coil ends and the last 36 mm of the coil straight sections to lower the
magnetic field there, and are bonded to the coil to provide axial rigidity. The collared coil
and yoke are contained by a 4.95 mm thick stainless steel support shell with 38.1 mm
thick, one-piece, stainless steel end plates welded directly to the shell ends. This scheme
replaces the bonnet design of the 40 mm aperture dipole magnets. Axial end force preload
is applied to the coil ends by four set screws in each end plate. One or both end plates
are equipped with "bullet"-type strain gauges, mounted inside hollowed out set screws, to
measure the end force at assembly and throughout testing. The magnitude of this force,
and how it relates to thermal cycles and axial Lorentz forces during coil excitation, has
been shown to be a factor in quench performance. During the testing of the four magnets,
end force preload was increased from 9 kN per end to 36 kN per end.
In addition to the collar pack and bullet strain gauges, magnet instrumentation also
included two diode temperature sensors at the lead end plate, and also, in DSA209 and
DSA210, carbon-glass resistor sensors to monitor temperatures at the gas-cooled leads.
Each of the inner coils was equipped with 22 voltage taps to help localize quench origin.
Taps were placed in the pole turn and the next three turns, two of which are adjacent to
the pole-most wedge. The ramp-splice section is more heavily tapped in order to obtain
better resolution in locating quench origins more accurately in that sensitive region. There
is also a tap at the turnaround between the inner and outer coils after the ramp-splice. In
addition, DSA209 and DSA210 had an added tap at the midplane turn to supplement the
taps monitoring the superconducting leads.
Design Variations
The magnets in this series differed by design only in the conductor and the cable
insulation. Table 1 lists the conductor parameters and type of cable wrap used for the
561
inner coils of all four magnets. As can be seen, they form two pairs with respect to
conductor characteristics and cable insulation type: DSA207 and DSA208 inner coils are
wound from the same spool and use the standard fiberglass-epoxy adhesive cable wrap
with 20% epoxy content; DSA209 and DSA210 inner coils are also wound from the same
spool and employ a double-layer all-Kapton cable wrap. The outer coils, not included in
the table, use the same insulation schemes as their respective inner coils. The slight
variation in conductor properties between the two magnet pairs results in a small difference
in their quench plateaus. The difference in the parameters used to cure the cable wrap,
however, is believed to have more significant effects on magnet performance. As seen in
Table 1, during the coil molding process, the coils with all-Kapton wrap were cured at 217
C, 82 C higher than that necessary for the epoxy-impregnated fiberglass wrap. Also, due
to the thickness of the double Kapton layers, molding pressure to the coils had to be much
higher, by as much as 71 MPa in the inner coils, as shown in Table 1. Outer coil
pressures were increased by as much as 46 MPa. Coil cure pressure and temperature are
being investigated as a possible cause of lowered cable interstrand resistance, leading to
the ramp rate-dependent effects seen in the two all-Kapton magnets and discussed in a later
section of this report.
Table 1. Inner coil cable and cure parameters related to quench performance
and ramp-dependent effects in the 1.8 m 50 mm SSC dipoles.
Coil Cure
Cable Ip.O T) Cu:SC Cable Coil Cure Pressure
Magnet ID [AJ Ratio Wrap Temp. [C) [MPa]
DSA207 SSC-3-1-22 10827 1.54 Kapton/Epoxy- 135 27/34
Fiberglass
DSA208 SSC-3-1-22 10827 1.54 Kapton/Epoxy- 135 23/23

Fiberglass
DSA209 SSC-3-S-22 10079 1.51 Kapton/Kapton 217 90194
DSA210 SSC-3-S-22 10079 1.51 KaptonlKapton 217 72/72
Another consequence of the thicker all-Kapton wrap was the smaller coil assembly
shim sizes used in the two all-Kapton magnets. Shim sizes are listed in Table 2, and the
difference between the two pairs of magnets is evident. While DSA207 and DSA208 have
shims close to or at the design shim size, DSA209 and DSA210 required shims
significantly smaller. Also, it can be seen that the two all-Kapton magnets differed from
each other in shim size. Such apparent disparity in coil size among the magnets will show
up as variations in the values of the allowed harmonics during magnetic field
measurements. These results will be discussed in a later section.
Table 2. Coil assembly shim sizes for the 1.8 m 50 mm SSC dipole magnets.
Magnet Inner Coil Shim Size (mm) Outer Coil Shim Size (mm)
Design 0.762 1.016
DSA207 0.762 0.965
DSA208 0.762 0.965
DSA209 0.635 0.381
DSA210 0.508 0.457
562
EXPERIMENTAL TECHNIQUES
Cryogenic testing of the short dipoles was performed in vertical dewars and cooling
was by liquid helium bath and control of dewar pressure. Measurements were taken at
three test temperatures: the SSC operating temperature of 4.35 K and the subcooled
temperatures of 3.85 K and 3.35 K (0.5 K steps).
The typical test procedure for all four magnets consisted of training by ramping up
the magnet current at a specified rate until a quench was generated and performing this
until the quench currents achieved had reached a four quench plateau, within an allowable
range of 30 A to allow for temperature fluctuations. In addition, quench origins were
verified to be at locations reasonable for conductor-limited quenches. For the first quench
at each test temperature, an increasing sawtooth current ramp was utilized as a type of pre-
conditioning procedure to allow the magnet to gradually experience mechanical movement
due to the Lorentz forces in small steps to possibly avoid enough heat generation to
produce an energy-deposited quench. In addition, at the end of each set of plateau
quenches, a strain gauge run-to-quench was performed, in which strain gauge reads were
taken at 500 A steps up to about 500 A below the plateau current, then at 100 A steps to
quench. This series of measurements was done to study the dependence of azimuthal coil
stress and end force on magnet current and field. Strain gauge loops, in which strain
gauge reads are taken at current steps on both an up ramp and back down without
quenching, were also done to obtain data on mechanical hysteresis.
After quench testing at the SSC operating temperature of 4.35 K, where the central
field can reach 7.4 T at quench, training was also performed at the sub cooled temperatures
of 3.85 K and 3.35 K, where the central field is about 8 T and 8.5 T, respectively. The
training at subcooled temperatures is done to test the limits of a magnet's mechanical
performance under the stress generated by the higher magnetic fields possible at the lower
temperatures. Then, after warm-up to room temperature and re-cooling (thermal cycle),
quench plateaus were again established at the three test temperatures.
In addition to the standard quench tests to establish conductor-limited quench
plateaus at the three temperatures, all four magnets were also tested, while at 4.35 K, at
varying ramp rates, to study the effect of ramp rate-induced eddy current heating on the
maximum quench current. Tests were done at varying ramp rates, starting with the rate
at which the current predicted by short sample cable measurements was reached, to 500
A/s in some cases. Some interesting differences were found between the two pairs of
magnets with different types of cable insulation wrap. These results will be shown later.
Quench origin location and velocity of propagation were determined by the start
times and slopes of the resistive voltage signals between pairs of voltage taps surrounding
the quench hot spot and the cable segments adjacent to each side. In most cases, a time-
of-flight calculation was adequate and accurate to within 4-10 mm, depending on quench
propagation velocity. The slope and shape of the voltage signal was also useful in
determining the general location and can be used to calculate the quench velocity.
Quench Tests
Table 3 summarizes the quench plateau results for all four magnets. As can be
seen, all reached a plateau at each of the three test temperatures and exceeded the quench
margin of 10% above the collider operating current of 6618 A at 4.35 K. The quench
history plots referred to in this section do not include tests done at ramp rates other than
those needed to achieve stable conductor-limited quench plateaus. Except for DSA209,
such quenches did not present a possible interference with standard training to plateau.
The special case of DSA209 will be discussed later in this section. Results of the ramp
563
Table 3. Performance of conductor with respect to Qredicted short sample
limit and sse operating current for 1.8 m 30 mm sse dipole magnets.
4.35 K 3.85 K 3.35 K Plateau

Ramp
Is.[A] I QlIss la/lop Is.[A] I alIss Is.[A] I II/Iss Rate
Magnet INtl
DSA207 7422 1.007 1.130 8048 1.008 8578 1.014 16
DSA208 7422 1.022 1.146 8048 1.024 8578 1.027 16
DSA209 7315 1.011 1.117 7910 1.016 8437 1.025 2
DSA210 7315 1.013 1.120 7910 1.021 8437 1.029 2
rate studies, with quenches done at varying rates, will be presented separately.
The quench history for DSA207 is shown in Fig. 2. During the standard quench
tests, DSA207 reached mean plateaus of 7476 A, 8109 A, and 8702 A at the test
temperatures of 4.35 K, 3.85 K, and 3.35 K, respectively; these are 0.7%, 0.8%, and
1.4 % above values predicted from measurements of short sample cable. A difference of
54 A between the quench plateaus at 4.35 K before and after the thermal cycle was due
to a difference in the test temperatures for those two runs. There were no training
quenches at 4.35 K and 3.85 K. At 3.35 K, there were three training quenches, two
before and one after the thermal cycle, and these were only 0.6% below the mean plateau
quench current. One of these originated at the turnaround between the upper inner and
upper outer coils. The other two training quenches and all the plateau quenches but one
were located in the pole tum short straight section adjacent to the ramp-splice section, and
outside the G 10 box which houses the ramp-splice. One plateau quench occurred in the
tum 18 lead end section, but was probably in the pole tum right straight section part of
that tap segment and not in the curved end itself. All the ramp rate study quenches
originated in the pole tum short straight section, but those whose ramp rate was 200 A/s
or greater were slightly inside the GIO box.
l I :0 ~\o 00 I
:0 0
I
<> 0 0 ()
B I 10 Q....<W> 01 10 0 0 0 0:
a <>~o:o
I
0°: I \ I
Ia 000001
I I
I
'\: : I.. :
'0°1I I
I I
I
6 I II :I
I I
I I
I I
I I
I I
I I
I I
4 I:
I !
2
I..
I
I
I
I
4.35X I
4.3fSK 3.85K I ::S.35X
8
f. I
I
II
0
0 10 20 30 40
Quench Number
Figure 2. Quench performance of DSA207.
564
For DSA208, a total of 115 quenches were performed in two separate tests. Fifty-
two were done during the first test and 63 in a later test in which the magnet was rebuilt
with a non-monolithic yoke (the only magnet in the series so constructed) and increased
end force preload of 36 kN/end (8000 lb./end). Of those 63 quenches, 13 were at varying
ramp rates to study ramp dependent effects and 50 were at the standard 16 A/s ramp rate.
The first test was done at the standard three temperature regimes but the later retest was
performed at 4.35 K only. During the first test there were nine quenches which took place
in the superconducting leads and outside the magnet. They occurred in either lead and
resulted in unusually low quench currents. All but one were at 3.35 K and were attributed
to inadequate lead flow to the gas-cooled leads during current ramps to quench. After this
test, improvements to the leads resulted in the elimination of such quenches. Since these
quenches were external to the magnet and were subsequently eliminated as a problem, they
are not included in the quench plot, which appears in Fig. 3, for the first test of DSA208.
I 1<>1 01 o
I
I I I I
1<> <> <> <> <>1 I I
8 1- II I
I I
<> <> <> <> 0 0 <> 0 01I \ II I 0 <> <> 0
10 <> <> <> <> <> <> <> <>
I
\
Iss
I .. i
:
I
I
I
I
I I
I
6 I
I
I
I
I
I
I
I
I
4 I
I
I
I
I
I
I
i
I
2 r(>-u-p-pe-r-lnnL-er-C-Oil~ I
I
D Lower Inner Coil I I
I
I I
I", "'I
4.35K 3.85K I till 4.35K ~ I 4.35K 4.35K 3.85K 3.35K
ltOl "I
I
I I I
O~-L~ __~-L-U__~~I-W__~-LL-U-~-LL-L-~~ __L--U~__~~
o 10 20 30 40
Quench Number
DSA208 reached mean quench plateau currents of 7586 A at 4.35 K and 8238 A
at 3.85 K, which are about 1.5% higher than those for DSA207. This can be attributed
to lower actual test temperatures, possibly because of offsets between the temperature
sensors mounted in the two different dewars that were used. Like DSA207, there were
no training quenches at these temperatures. All quenches at 4.35 K were located in the
pole turn short straight section near the edge of the G 10 box, either inside or outside,
except one low current quench in the lead. At 3.85 K, all quenches were also in the pole
turn short straight section, but all except two were farther away from the G 10 box than
the other quenches in that section. At 3.35 K, the quench behavior was erratic, and a
plateau was not established at this temperature before the thermal cycle; this was mostly
due to the lead quenches already mentioned. After the thermal cycle, at 3.35 K, there
were two lead quenches and five plateau quenches, which occurred in the pole turn short
straight section and formed a mean plateau of 8810 A, about 1 % higher than that for
DSA207. The quench locations were like those at 3.85 K and were far from the GlO box.
565
In the DSA208 retest, shown in Fig. 4, with the non-monolithic yoke and the
increased end force preload, 50 quenches established a mean plateau of 7630 A with a
standard deviation of 10 A. This plateau was 0.6% higher than for the earlier test due to
slightly lowered test temperature. Quench origin locations were in the lower inner coil
pole tum right straight section or in the upper inner coil pole tum short straight section,
not near the GlO box. As in the first test at 4.35 K, there were no training quenches.
I I ' I i I j
8-
oo\oooooooooooooooooooooooOOOOOOOOOOOOOOOOOOOOOOOD
-1 ~
~ ~
~
6f-
~
""t: -
" 4f-
I .
t:)
.c
"
c
I~
"
a"
Lower lDDer Coil
o Upper Inner Coil ~
'i!
2f--
t:
5 -
0 I 1 I
60 70 80 90 100
Quench Number
Figure 4. Quench performance of rebuilt version of DSA208.
The test results for DSA209 can be divided into two parts. In the first part, all
standard quench tests were done at the standard ramp of 16 A/s, as they were for the
previous two magnets. It was found, however, that after three training quenches, located
in the turnaround cable between the lower inner and lower outer coils, a uniform plateau
at the short sample-predicted quench current could not be established, and quench currents
were somewhat erratic. Most of the quench origins were not in the pole tum but in the
15 tum, untapped cable segment between the midplane tum and tum 15; this verified that
these were not conductor-limited quenches. Because of this behavior, the magnet was
warmed up to room temperature and removed from the test dewar in order to increase the
end preload from 9 kN/end to 36 kN/end. After a second cooldown, testing resumed and
a very uniform plateau was established at 7255 A, 0.8% below the short sample limit.
The quench origins were still somewhere in turns 1-15 and not in the pole tum. Because
of this, and a plateau below the conductor limit, it was decided to perform a ramp rate
study by quenching the magnet at various ramp rates. It was discovered that, in order
to establish a plateau at the conductor limit, the maximum ramp rate that would work was
2 A/s. At this point, all further quench plateau tests were done with the 2 A/s ramp rate,
and the standard test schedule was then carried out at the three test temperatures before
and aftera thermal cycle. The quench tests just described, prior to and leading up to this
point, are not included in the DSA209 quench plot in Fig. 5 for the purpose of brevity.
As can be seen from the plot, uniform plateaus were established at all test
temperatures with little or no training. DSA209 reached overall mean plateau currents of
7395 A at 4.35 K, 8037 A at 3.85 K, and 8648 A at 3.35 K; these were 1.1 %, 1.6%, and
2.5 % above their respective short sample predictions. All plateau quenches occurred in
566
the pole turn straight sections of either inner coil, on the left or right side. There were
three training quenches: before the thermal cycle at 3.85 K, 0.5% below plateau, and two
after the thermal cycle, one at 4.35 K and one at 3.35 K, which was 2.2% below plateau.
The lowest was the one at 4.35 K and was 7165 A, 3.1% below plateau but still 8.3%
above the sse operating point. All three training quenches originated in the turnaround
section between the lower inner and lower outer coils, like the three training quenches at
the start of testing with the 16 A/s ramp rate. It is not possible to know what training
would have been like if the initial testing had started with the 2 A/s ramp rate.
I I
1<> <> 0
I
<> 0
10
\ 01
I
8 ~:JOOI <> <> 0 <>
\
I I
I I
t..
"--"-" 0 I
0 I
\
I I
1.. I I
I I
I I
1" I I
I
6 I
I
I
I
<-
~
I
I
I
I
I
I
"t
I
I I
I I
I I
I
" 4 I
.
'-' I I
I I
-§ I I
""
I
.
~
I
"
0' I
I
I
u
2
I
I ~
I
I
t:
I
I
4.36X 3.65K I 3.351 4.35K 3.35K
I
I
I
0
30 40 50 60
Quench Number
Because of the ramp rate-dependent effect seen in DSA209 and because the
DSA210 inner coils were wound from the same spool, it was decided to continue with 2
A/s as the standard ramp rate for DSA21O, whose quench plot is shown in Fig. 6. As can
be seen, DSA210 reached stable plateaus at all three temperatures, both before and after
the thermal cycle. The overall mean plateau currents were 7411 A at 4.35 K, 8076 A at
3.85 K, and 8680 A at 3.35 K; these were 1.3%,2.1 %, and 2.9% above their respective
short sample limit predictions. All plateau quenches originated in the upper inner coil pole
turn straight sections, left or right side. There were three training quenches before the
thermal cycle and no retraining quenches after. The first, at 4.35 K, was at 7345 A, 0.8%
below plateau and 11 % above the sse operating point; it occurred in the ramp section
inside the GlO box. The other two were at 3.35 K: one was 1.9% below plateau and in
the ramp-splice and the other was 5.8% under plateau and in the upper outer coil, close
to or in the pole turn. This was noteworthy as being the only outer coil quench seen in
any of the four magnets.
Ramp Rate Studies
Because of differences in the conductor characteristics between the two pairs of

magnets, the maximum ramp rate at which a conductor-limited quench plateau could be
established was different for the two pairs. As discussed previously with the quench
training results, the standard ramp rate of 16 A/s was adequate for DSA207 and DSA208.
567
For DSA209 and DSA210, however, it was necessary to lower the rate to 2 A/s to avoid
ramp-dependent effects. These effects, also seen in the full length 15 m dipoles built at
both BNL and FNAL, are of functional importance for the sse collider, since the collider
design ramp rate is 4 A/s.
I 10 I ~
+-\
0 0 <> 0 <> : 0 <> 0 0
B- 10 <> <> <>

1-
0: I
IQQOOOI
I
I
I -
I I
<> C ~ <> 01 \ I", 0000 0 : :
\ !
I I
I", I I
I I
I I
I",
6-
I
I
J
I -
I I
I I
~
I I
I J
I I
d
~
I
I
I
I
~
I I
~
-
"
t.>
.c
4- I
I
I
I
I
I
u I I
C I I
~
I I
a" 0 Lower Inner COU I I J
<> Upper lnner Coil I I
1+ Upper Outer CoU I I
2- I
I
I
I -
J I
I I
I I
".3MC '.BOX 9.36JC ".:iSle I 5.a5K I S.S6K
I I
I I
oL-~~ __ ~~~ I__ ~~ __ L-~-U I__ ~~~L-~~ I II __ ~~
o 10 20 30
Quench Number
Figure 6. Quench performance of DSA21O.
Because of this, the ramp rate-dependent behavior of all four magnets was studied
by performing a series of quench tests at varying ramp rates for each magnet. Ramp rates
used in a particular magnet depended on the magnitude of the effect in that magnet. When
the effect was large, lower ramp rates were needed to improve resolution. The ramp rate
study for DSA208 was done during its retest with non-monolithic yoke and increased end
preload. The results for the four magnets are combined in Fig. 7, and it is immediately
apparent that the two pairs show significantly different behavior, both by the way the
quench current varies with the ramp rate and, as already mentioned, by the maximum
values at which a conductor-limited quench plateau is achievable. The two magnets in
each pair, however, showed similar behavior.
As the ramp rate increased, the locations of the quench origins changed from being
in a pole turn straight section, where conductor-limited quenches are e:<pected to occur.
This behavior also depended on the magnet and the severity of the ramp effects. In
DSA207, which exhibited only a small effect at 100 A/s, the quenches at 100 A/s and
higher still originated in the pole turn short straight section, but moved inside the GIO
box. In DSA208, which had a slightly greater effect at 100 A/s, the quenches at 100 A/s
and higher were in the ramp section, farther inside the GIO box. For DSA209 and
DSA2IO, which showed large ramp rate dependence, quenches at 16 A/s and higher, in
addition to being below plateau, were located away from the pole turn toward the
midplane.
It is believed that these effects are caused by heating due to eddy currents generated
in loops between the strands in the cable as the current is ramped up. This phenomenon
becomes more pronounced at lower ramp rates when the interstrand resistance is lower
than a critical value. It is suspected that a lowering of this resistance may occur during
the coil curing process when high pressures and temperatures are used. From Table I,
568
I ' ,
l
-------------------------r------
20 TeV
<
=-c
.
u
t:
:s 4 L
.c
.:sc
u ~
J
Cl DSA207
Of
o DSA20B
x DSA20g
+ DSA210
O~-L~~~~~~_L~~ __ ~~~~~~_L~~~~
o 100 200 300 400 500 60C

Re.mp Rate (A/_)
Figure 7. Quench tests at varying ramp rates for the 1.8 m 50 mm sse dipole
magnets.
it was seen that, because of the all-Kapton insulation scheme used with the DSA209 and
DSA210 coils, they had to be cured at significantly higher pressure and temperature than
the coils for the other pair, and they subsequently demonstrated a much steeper IQ vs. dIldt
curve. Furthermore, the molding pressure for DSA209 was about 20 MPa greater than
for DSA210 and its curve was slightly steeper. Additionally, from the ramp rate studies,
it was seen that a conductor-limited plateau was possible for DSA210 at 6 A/s, 4 A/s
faster than that used for DSA209 and 2 A/s faster than the collider design ramp rate.
The 15 m-Iong sse dipoles built at BNL and FNAL have shown varying ramp-
dependent effects, even though the magnets tested so far have standard epoxy-fiberglass
wrap and did not experience the higher molding temperature and pressures at curing.
Another mechanism for lowered interstrand resistance has been suggested: removal of
oxidation layers in the cables during cable preparation and consequent decrease in
resistance between strands.
Mechanical Behavior
The mechanical behavior of the magnets was studied by extensive measurements

of coil azimuthal prestress with collar pack beam strain gauges and of coil end force with
bullet strain gauges mounted in end plate set screws. The strain gauges were routinely
monitored during all phases of assembly and testing, and during ramps of current to high
field. Variation of the stress and end force with current was measured during strain gauge
runs-to-quench, as described earlier, and this data is summarized for all four magnets in
the plots of Figs. 8-10, which show stress or force plotted against the square of the
current, since the Lorentz forces are expected to vary as 12. These plots represent data
taken during the strain gauge runs to the highest field for each magnet, while testing at
3.35 K. The maximum current reached was 8600 A.
Figure 8 shows the average inner coil stress for each magnet. On average, the
inner coils lost about 47% of their initial prestress by azimuthal thermal contraction during
cool down , then 22 % of that during quench tests and magnetic excitation cycles. On
warmups, most of the initial prestress was restored, to within 11 %, on average. The plot
in Fig. 8 demonstrates how the inner coil prestress decreases linearly with 12 , then
569
becomes nonlinear at a current which varies with magnet before completely unloading to
zero stress. The current at which the prestress unloads varies with magnet pair: 48 x
106 A2 (6900 A) for the standard wrap magnets DSA207 and DSA208; 36 x 106A2 (6000
A) for the all-Kapton magnets DSA209 and DSA21O.
o 20 40 60 80
Current 2 (10 6 amps2)
Figure 8. Average inner coil stress vs. 12.
The average outer coil prestress is shown in Fig. 9. On average, the outer coils
lost 49% at cooldown, another 18% during testing, and recovered to within 7% of the
initial prestress during warmup. The plot shows that the outer coil prestress decreases
linearly with 12 in two steps with a change of slope at about 20 x I()6A 2 (4500 A). The 15
m-Iong sse magnets built at BNL also exhibit this behavior.[2] The slope changes where
calculations indicate that the collar, deformed by Lorentz forces as the current increases,
makes contact with the horizontally split yoke. Before that point of contact, both
azimuthal and radial components of the force contribute to the slope; after contact is made,
only the azimuthal component remains. Also, from the plot, it is seen that the outer coil
prestress decreases linearly at different rates (slopes) for different magnets.
Total end force is shown for the four magnets in Fig. 10. Each curve in the plot
represents the total end force for a magnet end where there were bullet strain gauges
mounted in the end plate set screws. It must be noted that all four magnets had gauges
in the return end but only DSA209 and DSA210 had them in the lead end as well;
r
therefore, six curves appear in the plot. All show linear increases of end force with at
various slopes, implying differences in coil stiffness. The typical increase is about 33 kN
over the 8600 A current range. Offsets in the plots are attributable partly to differences
in the initial end force preload. DSA207 and DSA208 started with 9 kN per end, while
the other pair had been initially torqued to 36 kN per end. Also, DSA208 had one
malfunctioning gauge, so its total is for three gauges only. The linear increase in end
force is typical of both the short and full length dipoles. All four magnets lost 50% or
more of their preload at cooldown, but preload increased significantly after quench testing.
This ratcheting effect is also seen in the long magnets. Preload returns close to initial
values at warmup.
570
40
)( DSA207
-;a 30
0...
6
'"
......w'"
Ul
20
W
till
...ctlw
>
-< 10
o L-~~~~ _ _L-L-~~~~~L-~~~-L~_ _L-~~~
o 20 40 60 80
Figure 9. Average outer coil stress vs. 12.
60 ,
t
I
~r
50 r
z
~
C 40
w
...0
C)
r.x..
-:l
30
c:
r::!
(ii
-' 20
;:)
:- x RE DSA207
o RE DSAZ06
10 ORE DSA209
• HE DSA210
)( I.E DSA209
+ IE DSA210
0
0 20 40 60 80
Figure 10. Total end force vs. 12.
571
MAGNETIC FIELD MEASUREMENTS
Table 4 gives a summary of the cold geometric multipoles for all four magnets.
With geometric multipoles, persistent current contributions are removed by averaging the
harmonics of both the up and down ramps of a measurement loop to high field; therefore,
the harmonics in the table provide sensitive indications of small variations in construction
among the four magnets. For example, the allowed harmonics (even-n b's) point out coil
size variations from the standard. In particular, the large difference in the normal
sextupole (b 2) between the standard wrap and all-Kapton wrap magnets can be understood
by recalling Table 2, which showed the disparity in coil assembly shim sizes between the
magnet pairs. Also, the difference in b2 between DSA207 and DSA20S can be attributed
to 76 p.m collar-yoke shims placed at the collar top and bottom in DSA207 only, to
compensate for collars that did not have the correct dimensions. Distortion of the coil in
DSA207 due to these shims would produce a geometric difference not seen in the other
magnets, which did not have shims between the collar and yoke. The large skew
quadrupole terms for all four magnets are due to size mismatches between upper and lower
coils. DSA209, which had the largest mismatch, also had the largest skew quadrupole.
Taking these geometric effects into consideration, the harmonic data are understandable
and agree with expectations. [3]
Table 4. Geometric multipoles for the 1.S m-Iong 50 mm SSC dipole magnets
while cold. SSC allowed systematic and random tolerances are
included for comparison. Units are 104 of 130 at 1 cm radius.
Multipole DSA207 DSA20S DSA209 DSA210 Systematic Random

b2 1.67 1.96 -3.20 -3.11 O.SO 1.15
b4 0.36 0.42 O.4S 0.70 O.OS 0.22
b6 -0.02 -0.01 -0.01 -0.04 0.01 0.02
bg 0.04 0.04 0.05 0.05 0.02 0.01
blO 0.01 0.01 0.01 0.01 0.00 0.00
bl2 0.00 0.00 0.00 0.00 0.00 0.00
bl 0.04 0.09 0.03 -0.37 0.04 0.50
b3 -0.01 -0.03 -0.03 0.06 0.03 0.16
b5 0.01 0.00 0.007 0.01 0.02 0.02
b7 0.00 0.00 0.00 0.00 0.01 0.01
al -1.34 -1.22 -1.46 -1.31 0.04 1.25
a2 -0.13 -0.17 -0.16 0.23 0.03 0.35
a3 -0.03 -O.OS 0.00 0.15 0.03 0.32
~ 0.01 -0.01 0.03 -0.01 0.02 0.05
a5 0.02 0.00 0.00 0.00 0.02 0.05
CONCLUSION
The results of the tests on the l.S m-Iong 50 mm SSC dipoles at BNL have
demonstrated that the new 50 mm aperture design meets the standards for successful
collider operation, and they have paved the way for subsequent tests of the full length 15
m-Iong dipole magnets of that design. All four magnets exhibited uniform quench
572
performance that exceeded the 10% quench margin, with little or no training. Magnetic
field quality met expectations within the range of well-monitored construction variations.
In addition, ramp rate-dependent effects on the quench performance in these magnets have
led to ongoing investigations of importance to the SSC project into the causes of such
phenomena. Also, in this series of magnets, experimentation with end force preload and
curing pressures for coils with all-Kapton insulation have added much to our knowledge
and experience for future SSC magnet construction.
REFERENCES
[1] R.C. Gupta, S.A. Kahn, and G.H. Morgan, "SSC 50 mm Dipole Cross Section",
presented at the 3rd International Symposium on the Super Collider (IISSC) ,
Atlanta, GA, March 13-15, 1991.
[2] M. Anerella et al., "Construction and Test Results from 15 m-Long, 50 mm

Aperture SSC Collider Dipole Models", presented at the 4th International
Symposium on the Super Collider (IISSC), New Orleans, LA, March 4-6, 1992.
[3] R.C. Gupta et al., "A Comparison of Calculations and Measurements of the Field
Harmonics as a Function of Current in the SSC Dipole Magnets", presented at the
1991 Particle Accelerator Conference, San Francisco, CA, May 6-9, 1991.
573
TWO-DIMENSIONAL MAGNETIC ANALYSIS OF THE SOC MUON TOROID
M. P. Krefta

Pittsburgh, P A
ABSTRACT
In this work, the magnetic field strength and distribution throughout the cross
section of the SOC intermediate and barrel toroids are calculated using two-
dimensional, finite element models. The variation of the magnetic field throughout
the muon steel and the integrated magnetic field strength, JBxdL, at different
angular locations are of prime concern. The ANSYS computer program is used to
determine the magnetic field effects of small gaps in the muon steel which
potentially may occur in manufacturing the toroids. A method of increasing the
field uniformity of the intermediate toroid by introducing a nonuniform air gap into
the magnetic path is shown to be effective. The fields are also calculated using
measured magnetic material characteristics for actual samples of the muon steel,
rather than using standard data available for low carbon steels.
INTRODUCTION
The Solenoidal Detector Collaboration (SOC) has proposed several toroidal

magnets for the muon detector system 1. This includes the barrel toroid which
encloses a 24.6 m axial length of the detector, an intermediate toroid which forms an
endcap for the barrel toroid and a forward toroid which is located outside the
intermediate toroid. The magnets generate a cylindrical magnetic field of 1.8 Tesla
that surrounds the interaction region through which muons pass and are deflected.
Each of the toroids will be constructed from straight lengths or blocks of

ferromagnetic steel forming an octagonal cylinder2. The blocks are large to avoid
assembling the toroid with an excessive number of pieces. There has been some
concern about the magnetic effect of small gaps introduced into the magnetic path
during the toroid construction3. Gaps or separations may potentially occur as a
result of tolerance buildup in the each of the eight corners of the toroid. Gaps may
also exist throughout the axial length of the iron stackup. Further, the large inertial
loads on the muon toroid structure can result in small gaps in the magnetic path.
This paper discusses calculations that were made to determine the effect of small
gaps on the magnetic field and describes a method of obtaining a uniform field
throughout the toroid cross section by purposely introducing nonuniform air gaps.
ANALYSIS OF THE BARREL TOROID
A two-dimensional magnetic analysis of the barrel toroid is performed to

determine the amount of ampere-turns needed to produce a 1.8 T field and obtain
the field distribution throughout the magnet cross section. The analysis is
performed on an octant of the geometry, since the fields are assumed to be
symmetric in the other seven octants. The dimensions associated with an octant of
the barrel toroid is given in Fig. 1. The muon steel is taken to be 1010 carbon steel.
The magnetic characteristics for this material are given in Table 1.
A finite element model was created using the ANSYS STIF13 finite element4 .
The mesh consisted of 4641 first-order triangular and quadrilateral elements (Fig. 2).
A thin region was introduced in the comers of the octagon so a small air gap could
be represented by simply changing the material properties of this region to those of
air. The finite element mesh shown in Fig. 2 illustrates a higher density of elements
near the corner of the toroid. This is necessary due to the large relative difference
between the size of the air gap (.050 in.) and the characteristic dimensions of the
toroid (a few meters). In order to obtain accurate results, finite elements with
reasonable aspect ratios (less than 20:1) are required. This requires a very fine grid
in the vicinity of the gap. This fine of a mesh is not needed external to the gap
region where larger elements can be used to obtain acceptable accuracy.
An analysis is first performed with the elements in the gap region assigned
magnetic characteristics of 1010 carbon steel. A total of 20,000 Ampere-turns is
applied to each conductor block to obtain a flux density of 1.74 T throughout the
central portion of the top iron block. This result is slightly lower than previous field
calculations performed on the barrel toroid5 in which a 1.8 T was obtained with the
same amount of Ampere-turns. Since two blocks of turns in the octant are
analyzed, a total of 40,000 Ampere-turns per octant are needed. Additional
Ampere-turns must be supplied to produce a 1.8 T field.
The effective magnetic path length for the barrel toroid can be determined by
dividing the total Ampere-turns by the magnetizing strength required to generate
1.74 T in 1010 steel. A magnetizing strength of 5,900 Aim is obtained by
interpolating the data in Table 1 to a field of 1.74 T. This gives an effective magnetic
path length of 7.78 meters per octant of the toroid.
The magnitude of the quantity JBxdL is of interest for several different radial
paths throughout the cross section of the toroid steel. This quantity has been
calculated and plotted in Fig. 3 for several different angles throughout the cross
section. Here, an angle of 0 degrees corresponds to a radial path through the top of
the toroid. An angle of 22.5 degrees corresponds to a radial path through the
comer of the octagon. The vertical scale has been expanded to show the variation in
JBxdL with angular location. In general, the variation is very small « 2%) except
for a slight peak near the angular location of the coils. Also superimposed on this
plot, is the field integral for the case in which a .050 in. air gap has been introduced
576
Table 1. Magnetic characteristics for low carbon 1010 steel used in magnetic
analysis of barrel toroid.
Point Magnetizing Strength Flux Density

H(A/m) 8m
1 160 0.25
2 208 0.50
3 344 0.90
4 720 1.30
5 1120 1.45
6 1760 1.55
7 2560 1.60
8 3600 1.66
9 4400 1.69
10 5400 1.72
11 6800 1.77
12 10400 1.83
13 16800 1.92
14 36000 2.05
15 72000 2.14
16 160000 2.28
,..
·r t- 1
r
2m300 460 B60
300620
1
1m500
2m999
2m300 ·1
LOW CARBON
STEEL
Figure 1. Dimensions of muon barrel toroid used in magnetic analysis.
577
AKSTS 4 . 4.\ 1
ru 2't l'U
.1:J. :JI
,un 1
'UP, tL-tlCDrTS
Tn!. Ilt.
tV -1
DIn · ... :H
ICf .~ lOJ
'fr .1"
Figure 2. Finite element mesh for muon barrel toroid magnetic analysis using 4,641
elements.
277
27
,
~ 2.75
::
III
273
..
2.72
ou·e< CO,I _
2.7
0 5 '0 25
ANGLE !DEGREES)
Figure 3. Comparison of the magnitude of JBxdL throughout a range of angles for

cases of no air gap and .050 in. air gap in the eight comers of the barrel
toroid.
at the corner of the toroid. The effect of the air gap is evident, but very small in
magnitude. It has the effect of decreasing the field integral uniformly throughout
the toroid steel by less than one percent.
The case in which the magnetic field is tangent to an iron-air gap interface has
a more substantial effect on the magnetic field. This situation occurs as a result of
stacking the steel blocks throughout the axial length of the toroid. In this case a
large discontinuity in the magnetic flux density occurs because the tangential
component of field is discontinuous by the relative permeability of the iron. Since
all the field is tangential to the iron-air interface, a large field discontinuity results
regardless of the size of the gap. The field drops from 1.8 T in iron to .0064 T in the
air gap. These gaps should be minimized near the center of the toroid where rays
can travel a large distance through the air gap and greatly influence the integral
JBxdL. A larger block thickness can be used to avoid gaps in this region.
578
ANALYSIS OF THE INTERMEDIATE TOROID
An octant of the intermediate toroid is shown in Fig. 4. Unlike the barrel

toroid, there is a substantial difference between the inner and outer diameters in
this case. An analysis is performed for an octant of the geometry to determine the
amount of field variation throughout the cross section. Here, a current of 22,500 AT
is applied to each of the conductor blocks. The iron material is assumed to be 1010
carbon steel described by the magnetic characteristics given in Table 1.
Contours of the constant flux density throughout the iron are shown in Fig. 5.
The field varies from 1 Tesla near the outside corners to greater than 2.1 Tesla near
the inside corners of the toroid. The variation is considerably larger than in the
barrel toroid, due to the large inner to outer diameter ratio. This is undesirable
since it is difficult to calculate the field integral JBxdL for a ray passing through the
endcap iron where the flux density is varying substantially throughout the path.
A method for obtaining a highly uniform distribution of field in the

intermediate toroid is to introduce a nonuniform air gap into the magnetic path.
This can be used to compensate for the variation of magnetic reluctance with radius.
The air gap can be chosen to be large at the inside radius of the toroid where the
iron path length is relatively short. Correspondingly, the air gap is chosen to be
small at the outside radius where the iron path length is relatively long. The air gap
geometry is described by the parameters gl and g2 as shown in Fig. 6.
500 I. ·1· 2111499 ---.~I
Figure 4. Dimensions of intermediate muon toroid used in magnetic analysis.
579
" t tA l
n.1 n \ttl
11 H I li
PWf., I
POSTl J'TUU
lTCP-I
Ina- 10
iliUM'"''''''
ptJI- Qo . 'IU'J
iKII - ' . )Ol
LY - I
. -,
ClUT- ) , h l
., - 1 . U
rr -t . IU
lI>«
I -1.0"
C " 1 Ul
o -I n.
- l .U t
" 1 JU
- I t' l
- I ~~
" 1 Ut
1"
.. t lQol
"1
- I U t
.. l . ' t )
" lO~1
-2 .•
Figure 5. Contours of constant magnetic field regions for intennediate muon toroid
with no compensating air gap.
Figure 6. Octant of intennediate muon toroid with nonuniform air gap illustrating
shape of air gap required to produce a unifonn magnetic field.
The values of the half gaps, gl and g2' are iteratively refined to obtain a more
uniform distribution of magnetic field in the iron. In this case, the optimal values of
gl and g2 were determined to be 14 mm and 1.4 mm, respectively. This produces a
highly uniform magnetic field of 1.8 T throughout the muon steel as shown in Fig.
7. The field variation throughout the steel has been reduced to less than 2 percent.
The added field uniformity is not obtained without cost. The cost in this case
is the additional Ampere-turns needed to overcome the air gap reluctance. The
amount of this can be estimated by taking the mean air gap length and multiplying
by the field strength in the gap. In this case, the mean gap length is 14.2 mm and
the field strength is 18 KOe (1.4 x 10 6 Aim) resulting in an additional 20,000
Ampere-turns per octant of toroid.
580
ANSYS 4 ",,1
MAR 2 .992
23·2&d l
I LOT ... 0 2
POST STRESS
$ TEP -2
I TER-20
BSU ... ,AVG)
- .6" 1
SKl( -).8' "
tV -1
DIST-l. )48
xr ·2.~6
y,. - 4 . 19
EDGE
A - 1 16
8 · 1 165
C
o
-1. ""
· 1 .17~
C -1."'8
r -1.'8~
G -1. 7~
-I . 19~
-1.8
-I. eO!)
-1.81
Figure 7. Magnetic field contours for intermediate toroid with air gap shaped to
produce uniform magnetic field distribution.
FIELD CALCULATIONS USING MEASURED MATERIAL

CHARACTERISTICS
The magnetic characteristics have been measured for a sample of the actual
muon steel to be used in constructing the toroid. The measured data are plotted in
Fig. 8 encompassing a range of fields from 20 to 2000 Oe. This data is compared
with standard data available for low carbon steels. At low fields, the induction is
less for the measured data. However, at higher fields, the measured data for
induction becomes greater than that of the standard data.
It is necessary to determine the impact of different magnetic characteristics on

the magnetic field uniformity of the optimized toroid shown in Fig. 7. This
geometry is analyzed using the measured material characteristics shown in Fig. 8.
The resulting flux density distribution remains very uniform for this case. Further,
the absolute field strength is nearly the same. This is not surprising due to the
similarity of the two material characteristics especially at the 1.8 T level of field
where the toroid operates.
CONCLUSIONS
The magnetic field distribution in the muon barrel and intermediate toroids
was calculated using a two-dimensional magnetostatic analysis. It was determined
that the effects of gaps on the order of .050 in. (1.27 mm) located at the corner of the
barrel toroid would decrease the field throughout the muon steel by less than 1
581
2.50
2.30
E sample
>-
I-
(j)
Z
2.10
, .. - , standard low carbon steel
W
0
1.90
X
:::l
-1
LL
1.70
1.50
t
~~~~~~~~~~~~~ww~~~~~~ww~~
o 200 400 600 800 1000 1200 1400 1600 1800 2000
MAGNETIZING FIELD (Oel
Figure 8. Comparison of measured magnetization data and a standard magnetization

curve for low carbon steel.
percent. The case of gaps between the laminations throughout the axial length of
the barrel toroid has a much larger local effect on the magnetic field. These gaps
should be minimized near the center of the barrel toroid (z=O) to avoid radial gaps
in the magnetic field.
The magnetic field can be made more uniform in the intermediate toroid by
introducing a nonuniform gap in the magnetic path. This was accomplished at a
field of 1.8 T using an inner half gap distance, gl' of 14 mm and an outer half gap
distance, g2' of 1.4 mm. The cost of incorporating the nonuniform gap is additional
ampere-turns which must be supplied to overcome the air gap reluctance.
REFERENCES
1. G. H. Trilling, et aI., "Solenoidal Detector Collaboration Letter of Intent," Solenoidal Detector
Collaboration Note SDC-90-00151, 1990.
2. W. Edwards, et aI., "SDC Muon Barrel Toroid Conceptual Design Review Final Report of the
Review Committee," Solenoidal Detector Collaboration Note SDC-91-OOO15, Mar. 15, 1991.
3. S. J. Hong, "Effects on the Magnetic Induction due to the Gaps within the SOC Muon Toroid,"
Solenoidal Detector Collaboration Note SDC-91-00005, 1991.
4. G. J. DeSalvo and R. W. Gorman, "ANSYS Engineering Analysis System User's Manual,"

Swanson Analysis Systems, Inc., Houston, PA, 1989.
5. J. Bensinger, "Field Map for SOC Muon Toroids," Presented at the SOC Muon Steel meeting,
Jan. 8, 1991.
582
A NEW CONCEPT OF MULTIPLEXED OPTICAL
TRANSMISSION READOUT SCHEME FOR A SILICON
STRIP DETECTOR
Norihiko Ujiie, Hirokazu Ikeda, and Yoshinobu Unno
National Laboratory for High Energy Physics, KEK

1-1, Oho,Tsukuba-city, Ibaraki-ken, 305 Japan
Abstract
We have been proposing a new concept of multiplexed optical-transmission readout
of a factor. of 1000 reduction in the number of cables, yet individually readout.
A wavelengh division multiplexing (WDM) transmission system using optical devices
is described which has possible application to a detector with an enormous number of
readout channels (larger than 5 million) e.g. the Silicon(Si) micro(p; )-strip detector
of SDC, or in high energy physics experiments in the near-future.
1. Introduction
One of the major detector in the experiments in high energy physics has been the
tracking detector to measure the momentum of a charged particle with the magnetic
field. In the coming collider detectors in the SSC, this will be still true. The measure-
ment of momentum is done by measuring the positions of the passage of the charged
particle. With increasing the available collision energy, the momentum of interest will
increase, thus, require either larger tracking length, higher magnetic field, or better
position resolution.
One of the cost-effective way, to meet the requirement is to have better position
resolution, e.g. by using a Silicon-strip detector born from the recent technology of
Silicon semiconductor device. An average resolution of 15 p;m can be achieved by YES-
NO readout of 50 p;m pitch spacing of the strips. Present drift chamber have a typical
resolution of 150 p;m with a channel spacing of about 5 mm.

Thus, the radius of the Silicon-strip detector can be reduced to square-root of 1/10
to maintain the same momentum resolution; if better momentum resolution, larger
than square-root of 1/10. However, there is a price we have to pay: about 100 times
of readout channels, and if we divide the length of the drift chamber to 1/10, 1,000
times. A typical number of readout channels of present drift chamber is about 10,000,
and the resulting readout channel of the Silicon-strip detector will be 10,000,000.
So far, we have kept one-to-one correspondence between the readout channels and
the data processing electronics residing out of the detector. To cope with this huge
number of readout channels, the present concept of sending informations is to send only
hit channel addresses through a manageable number of digital-bus cables by using in-
telligent electronics at the detector realized by LSI circuits. One-to-one correspondence
will be no-longer maintained.
Disadvantages of losing one-to-one correspondence with integrated circuit at the
detector are: gross damage of information if a LSI chip has malfunction; more chances
of damage, e.g. by radiation, for more integrated circuits, no direct link to a channel
for debugging, etc.
Here, we propose a new concept of multiplexed optical transmission readout without
electrical data compressing circuits mentioned above to regain one-to-one correspon-
dence. The multiplexing can be done in the domain of wavelength of light (wave-length
division multiplexing: WDM) transmitted in the optical fiber, and in addition, in the
domain of time axis (time division multiplexing: TDM). The usage of optical fiber can
dramatically reduce the volume and mass of the cables compared with the electrical
cables.
In the following sections, we will describe the WDM and TDM system, impres-
sive progresses of optical devices such as laser-diode arrays, multi/demultiplexer, and
typical receiver circuit, etc. which can be applicable to the sse experiments.
2. Optical wavelength division multiplexing (WDM)
The concept of an optical WDM transmission system with a number of "n" readout
channels is shown in Figure 1. A WDM transmission system is basically composed of
three parts: Transmitter, Optical fiber, and Receiver. Singals in individual channel
modulate it's own laser diode (LD) with one wavelength, a carrier. Several carriers
whose wavelengths are spaced out within the bandwidth of the optical fiber are mixed
with an optical multiplexer. A light with many wavelength is transmitted through a
fiber, and then demultiplexed at the receiver. A demultiplexer may work as optical
filter to pass a specific wavelength, or an extra optical filtering may be employed.
Thus, information-carrying capacity can be increased dramatically for a fiber and, if
the multiplicity is large enough, it becomes possible to transmit every channels with
reasonable number of optical fibers.
584
The WDM system has excellent features such as:
1) no need for intelligent processing circuits for data compression at the detector-
end, and
2) better reliability because any damage at the detector-end is channel by channel
except the multiplexer,
in addition to the common features of the optical transmission system such as:
3) electrical noise immunity, and
4) dramatically less mass than electrical cables.
Transmitter (E/O) Receiver (O/E)
Double l i ded ~1
S1 jI- .trip
dete c t.o r
........
.... "
: Optical:
: filter : PHD Pre-ar.'lp
-
Demu! t 1p le xe·~····-·-·~ array
Expenmenlal Hall .....••...... EleclroOlcs hUI

- ................................................. .................................................
Figure 1. Principle of optical wavelength division multiplexing (WDM) system.
Furthermore, multiplicity can raised easily by introducing an idea of TDM which

will be described in the next section.
To develop an adequate WDM system, three factors have to be kept in mind for
device parameters:
a) laser diode should have low power consumption with enough power margin and
stable oscillation frequency,
b) wide enough bandwidth for the system, and
c) good Signal-to-Noise ratio.
The power margin of a light source should be designed such as (P r P r P r) > 10 dB
where P t is emitted power from light source, P J=Lxa, L is the Length of the fiber,
a is unit loss of the fiber, and P r is detectable power at receiver. The bandwidth of
individual devices should be wide enough, or in other words, the rise time of individual
device should be fast enough so that the quadratic sum of each rise time should be less
than the system rise time.
585
3. Combined multiplexing system of WDM and TDM (WTDM)
Multiplicity into a single optical fiber can be increased by dividing the transmitting
time (Time Division Multiplexing: TDM). When combined with a WDM system, the
system will be WTDM (Wavelength and Time Division Multiplexing). SSC has a very
fast beam crossing interval of 16 ns( 62.5 MHz), and individual data must be carried
out from the detector within the time interval. Since 10 GHz is achievable for a optical
device, this 16 ns interval can be divided into at most 160 intervals. If a WDM has
a multiplicity of 100 and a TDM of 16, we can transmit a multiple of data as high as
1600 with the WTDM system.
Receiver (OlE)
: t ~
/1liLlL=L
: Tunc:
: - N %
: :t:t :t
.I. ~ ~ ...... ~
~ ........................... "."" ..... Electronics Hut

............................ ~
Figure 2. Optical WTDM(WDMxTDM) combined transfer system.
A WTDM system is shown in Figure 2 in which the Ins delay is made optically.
The 1 ns delay can also be realized by CMOS switching delay used in the time memory
cell (TMC) concept 2 • To realize the 1 GHz TDM, a new memory system will be
required which can be operated with 1 GHz at the receiver-end, in addition to the time
resolution of peripheral electronics.
586
4. State of art of the optical devices
4-a). Light source(LD)
The main element of the transmitter is the light source. We think an array of
laser-diode (LD) is the prime candidate. Each LD in an array is processed to emit
with slightly different wavelength which is driven by a driver circuit corresponding
to a detector channel. Shown in Table 1 are the fabricated laser diode arrays in the
past few years 3 - 8 • To be applied to a Silicon strip detector, an LD-array must have
following features: small size within the strip pitch, e.g. 50 J.Lm , high density including
driver circuitry and multiplexer, low power-consumption (low-threshold drive current
v.s. the output light: L-I curve), high speed, and radiation hardness. Surface Emitting
Laser-diodes (SEL) are being developed explosively; Vertical-Cavity type (VCSEL) has
excellent features of very ·small size (~10 J.Lm ) and low threshold current with a short
resonator. SEL generally has shorter resonator length than other LD types such as
Distributed Feedback (DFB), excellent mono-coherent wavelength, and small radiative
angle (5°) for a good coupling efficiecy to an optical fiber. Development of LD-arrays
are being watched with keen interest regarding WDM realization.
Table-I. Fabricated laser diode(LD) in the past a few years.
Wavelength Channel Authors &

LD-Type Features Fabrication
(~m) number published
............................. ............................ ...................................................... ............................ .............................
liA<O.S nm (=SA) SR lithography (NTT)
1.55 20 DFB Ith=IS·22 mAo
M.Nak.o et.a! [3]
Ind-drive S=400XSO I'm x2/ch accuracv<O I nm Elect Lett. VoI2S/'89
Fabry-perot <U.=4.s jim. (SONY)

0.86 102 MQW. MOCVD H.Narui cui [4]
Dig.12Inll.Scmicon.
Buried-heterostructure Ith=I.8mA. Ti=42% Ridge-sh.ped G.As
Laser Conr.Davos. '90
> 1 Million .,,1-7j1m.

MBE (AT&T)
0.96 (p>2 MilJion/cmx2, VCSEL Ith=I.S mA(SxSjlmx
J.L. Jewell, et.al
2-Dim Matrix) ISQW),.Ti>28% InGaAs:Resonator CLEO'90. CFFI [S]
Bellcore+AT &1
64 6).;=3S4j1m,
MBE
0.93-0.95 VeSEL 1U.=26+/-3A [6]
C.J.Ch.ng-H.nsn.,"
(8x8 2-Dim M.trix) hh=14.7mA InGaAs Elect Lett Juner91
6W4.62Sjlm
64 xl MBE (AT&T) (7)
0.85 VeSEL 1U.-849.4 +/-0.8 nm
R.A.Morg.n et ..•1
(I.Dimensional) Ind-drive ~~~:i;t!i2v\ mA G.As/AIG.As Elect Lett. Augf91
Planer (Sandia NT!)
0.85 64 xl VCSEL 1U.=20jlm,
self-aligned
liA=O.3A , IOGHz G.r. Olbright,
(I-Dimensional) Ind-drive AIAs/AIGaAs CLEO, JThA3 '91 (8)
[ 6l.: LD spacing, Ill.: Wavelength space, Ith: Threshold cunent, Vth:Thn:shold voltage, MQW: Multi Quantum Well, "A": Angstrom,
Ti: quantum efliciency,.: Cavity diameter of SEL-LD, SR: Synchrotron Radiation, MOE: Molecular beam epitaxy,
DFO: Distributed feedback, VCSEL: Venieal cavity surface emitting L. MOCVD: Metal-organic chemical vapor deposition.
587
4-b ). Multiplexer/Demultiplexer
The driver includes a multiplexer and the receiver contains a demultiplexer. If a

diffraction-type grating or interference filters is used, it will work as the mutiplexer
to mix carriers with different wavelengths into a light and the demultiplexer to split
a light into individual carrier with filtering function. Table 2 shows fabricated multi-
plexing/ demultiplexing devices in the last a few years 9 - 12 •
Table-2. Fabricated Multi/Demultiplexer in the past a few years.
Wavelength Diffraction Authors &

Type Characteristics Fabrication published
UJ.m) order
............................ ........................... ........................... ............................ ............................ ..............................
/U.o=lnm spacing RIE+PHOTO·LI (NTT)
1.3 Waveguide Resolution..o.63 nm ·THOGRAPHY
42 (28) grating Crsdk<· UldB H.Takahashi, cui
Si_IOx2O mmx2 Si02/117059/SiOl Elccl Lell, Jani '90 (9)
Coplaner 4A.-3 nm spacing RIE (Osaka Univ)

0.78-0.802 9 grating lens Crsllk<·2OdB(·7 dB) Shogo Ura, el. al
Si.....23xIO mmx2 Si·base Applied optics,
Vol 29. Marl '90 [10
? 20 Reflecting .u..I nm spacing ? . CREMER,C.,et.al

ECOC'90
4A.-3 nm spacing OMCVD (Bellcore)

1.5 >50 Spectrometer Rcsolulion=O.3nm +CAIBE
Crsllk<·19 dB J.B.O. Soole, el••1
Si_12x2 mmx2 InP/lnGaAsP ElccILeU, Jani 91[ II J
Grating 4A.-3.7nm spacing, Low pressure (Siemens)

1.2-1.6 >30 spectrograph Crsllk<·25dB epitaxy c. Cremer, cui
Loss-7dB
( Capable of lOOch) ( Rowland IypC ) SIIP2x3mmx2 InGaAsPlinP ~~~,.~~YSics.urY2)
C Crsdk: Cross Iallc, RIE: Rcacdve Ion elChi,!g, OMCVO: Org!",ic "';Clal chemical vapor deposi.ion, )
CAIBE: Chcmically usisred Ion beam elChing. Elecl Leu: ElcclrOn.cs leuen.
4-c), Transmitter and receiver circuit
A wide bandwidth of an optical fiber (single-mode) has stimulated reserchers to

consider increasing bit-rate to increase the transmission capacity. Recently, the trans-
mission bit-rate has been about (or above) 10 Gbit/s. This high transmission speed re-
quires not only a fast LD (or LED) light source and driver circuitry but also fast receiver
circuitry: photo detector, electrical decision and memory circuitry. These progresses
will enable to combine a TDM system with a WDM system. 1. AMAMIYA et al. of
Fujitsu have demonstrated a high speed receiver circuitry with HBT(AlGaAs)/(GaAs)
process 13 • A circuit and it's typical performances are shown in Figure 3.
588
a b c
r---.---.--<> GN 0
~80'----------~
c::
CD - Measured
IN ~ ..... Simulated
S 10'"
2l 60 ~
OUT
c:
III
E 10-1
"C iii
CIl ffi 10"
~40 10"
AI ·iii
c:
~
I- 20 '--.L......----'---'_'---'-.......L--'_...J
'----+---<> VEE 0.1 10
Frequency (GHz)
Figure 3. Typical high speed receiver circuits 13 :

a) transimpedance amplifier, b) frequency performancse, c) bit-error rate.
5. Radiation damage of optical devices
The radiation dose rate at the sse is a few mega(M) rads for charged particles and
I-rays and a few times 1012 /
for neutrons over 10 years of a luminosity of
cm2
1033 cm- 2 s- 1 at a radius of 10 cm from the beam axis. Radiation damage tests must
be carried out to confirm long-term reliability of the devices. The radiation damage on
a Ti-diffused LiNb0 3 waveguide device was investigated for several waveguide devices
containing straight, bent, and directional couplers with I-rays and high-energy elec-
trons. It was reported that such waveguide devices were radiation-hard to I-rays at a
dose of up to 3.8 Mrads l 4, with a reproducibility within a measurement error of ±5%.
Unfortunately, we can't find measured damage data for LD's or photo detector devices,
except for GaAs materiaP5. The radiation hardness of GaAs LD's must be measured,
especially SEL devices with a low injection current at the end of transmitter, which
will become a key device for a WDM transmission system.
6. Conclusions
We have proposed a wavelength division multiplexing (WDM) transmission system

using optical devices for the readout system of a future Silicon microstrip detector with
an enormous number of readout channels( 2:5 million) for high-energy physics experi-
ments at colliders. As a readout system, aWDM transmission system can elminate an
inteligent data-compressing LSI circuitry at the detector-end where the environment
is hostile in space and in radiation, and as a result can enhance reliability. A WDM
system can be combined with a TDM system to enhance the multiplicity. We have
assumed a multiplicity of a 100 for WDM and 16 for TDM: total multiplicity of 1600
to suffice to have reasonable number of optical fibers.
589
It is necessary to make an array of LD's and photo detectors together with high-
speed transmitter and receiver circuitry of, say, 10 GHz. These high density circuitry
is called Opto-Electronic Integrated Circuit (OEIC), which is a next target of micro-
electronics technology. Our requirement for the OEICs is high-density integration of
at least 2-order of magnitude larger than a few channels at present.
Surface Emitting Laser-diode (SEL) looks like the most promissing device for the
light source, which has a very small size (a few /-Lm) and also has possibility to lower the
injection current for low power-consumption. At present, 2-dimensional (8x8 matrix)
and I-dimensional (1 x64) SEL's have been fabricated. Improving the sensitivity of the
photo detector at the receiver-end is also important to decrease the LD driving power
at the transmitter-end.
Radiation damage tests of the optical devices must be carried out to ensure that
the system is viable in the SSC experiment.
7. Acknowledgements
We would like to acknowledge support and suggestions by Prof. T. Kondo,

T. Ohsugi, and Y. Watase of the JSD(Japanease Solenoidal Detector) group, and
Prof. S. Iwata, the director of physics department of KEK.
References
1. Letter of Intent by the solenoidal detector collabollation, 30 Nov(1990).

2. Y.Arai, et al., IEEE Journal of SSC, Vol-27, No3 Mar(1992).
3. M.Nakao, et al., IEEE.J. Lightwave technology, 8, Jan(1990).
4. H.Hirata, et al., Appl.Phys.Lett.58 (4),28 Jan(1991), pp.319.
5. J.L.Jewell, et al., CLEO (1990), CFF1.
6. C.J.Chang-Hasnain, et al., Electron. Lett., Vol 26, No 13(1991), pp.940-942.
7. R.A.Morgan, et al., Electron. Lett., Vol 27, No 16(1991), pp.l400.
8. G.R.Olbright, et al., CLEO JThA3(1991).
9. H.Takahashi, et al., Electron. Lett., Vol 26, No 2(1990), pp.87.
10. S.Ura, et al., Applied Optics, Vol 29, No 9, Mar(1990), pp.1369.
11. J.B.D.Soole, et al., Electron. Lett., Vol 27, No 2(1991), pp.132-134.
12. C.Cremer, et al., Appl.Phys.Lett 59(6), 5 Aug(1991), pp.627-629.
13. I.AMAMIYA, et al., (1990) Autumn convention record, part4 Japan, ppA-llO.
14. C.A.JACK. SPIE Vol.1l77(1989), pp.274-279.
15. Proceedings, SPIE,Vol.541, (1985).
590
THE DESIGN OF THE CERENKOV COUNTER
FOR THE SDC DETECTOR·
M. Albee,! G. Baranko,! B. Broomer,! M. Christoph,! M. Daoudi,l

E. Erdos,! U. Nauenberg,! G. Schultz,! J. Smathers,! C. Wente,!
J. Belz,2 G. Daniel,2 C. Guss,2 S. Kettel,2 K. McFarlane,2 and K. Johns 3
! University of Colorado, Boulder, Colorado

2 Temple University
3 University of Arizona
INTRODUCTION
One of the characteristics of hadron colliders is the large collision rates and the associated large
flux of particles at small angles relative to the beam direction (0 < 25°). In the case of the SSC,
preliminary conservative calculations indicate that this flux of particles can be expected to be upwards
of 10 7 per sec at a luminosity of 10 33 cm- 2 sec-!.! These rates will be discussed further below. These
calculations do not include the effect due to the background associated with the large particle fluxes
being produced inside the beam pipe and near the low radius cells of the Cerenkov counter nor the
background expected from the low energy neutrons behind the calorimeter in the present detector
design. This last effect is poorly known and is being measured. 2 This problem makes it difficult to
design an efficient trigger system for high energy muons with the use of standard scintillator counters.
In addition, these calculations show that most of this background comes from muons resulting from
decays of particles produced in the jets and, as a result, their momenta in the muon detector locations
are, on the average, very low (P < 10 GeV Ic). As a result of these observations we propose to use a gas
Cerenkov counter 3 which is very efficient in the detection of high momentum muons (P > 20 Ge V I c)
and completely insensitive to low momentum particles. In the SDC design we propose to use a multi-cell
nitrogen gas Cerenkov counter. We now proceed to describe this element of the SDC detector.
PARAMETERS OF THE SDC CERENKOV COUNTER
The behavior of a gas Cerenkov counter is essentially understood by the equations below which
describe the angle of photon emission relative to the direction of the incident particle and the number
of photoelectrons detected by a phototube on which the Cerenkov photons are incident.
* Research supported by the U.S. Department of Energy Contract DE-AC02-86ER40253 and the Texas
National Research Laboratory Commission Grant RFY9102.

where,
f3 = velocity of the particle in units of the velocity of light
n = index of refraction of the Cerenkov medium = 1.000309 for N2 gas
A = Cerenkov photon wavelength
f(A) = detection efficiency of the phototube
In Figure 1 we presen~ the dependence of the Cerenkov angle on the particle's momentum. This
angle has effectively a constant value of 24.8 mrad. In Figure 2 we show the number of photoelectrons
we expect to observe using a 5" BURLE 8854 phototube assuming a 100% mirror reflectivity. These
results are in excellent agreement with the measured values described later on. In Table 1 we present
the parameters of the design being proposed for the SDC detector.
CERENKOV ANGLE
2.
20
j .0
~
~ .0
°o·~~~~~~~~~.o~~~~~~~~~~~~~~,oo
MOMENTUM (GeV Ie)
Figure 1. The angle of emission of the Cerenkov light by a charged particle of a given momentum.
PHOTOELECTRONS PER METER
°0~~~~~~~~'0~~~~~~~~~80~~~~'OO
MOMENTUM (GeV/e)
Figure 2. The number of photoelectrons expected to be observed by a BURLE 8855 phototube

assuming 100% mirror reflectivity.
The operation of the counter as a trigger for high momentum muons is described in Figure 3,
where we show the alignment of a particular mirror-phototube axis relative to the beams collision
point. As a muon is bent by the magnetic field in the forward toroids, the angle of incidence of the
muon relative to this axis changes in proportion to the momentum. This angle would then be reflected
in the number of photoelectrons produced in the phototube since the Cerenkov light ring would also
move off the phototube face as indicated in Figure 3.
592
Table 1. Design Parameters of the Gas Cerenkov Counter.
Parameter Value
Gas Medium N2
Gas Pressure 1 Atmosphere
Index of Refraction (>.=3500 A 0) 1.000309
Width of Counter 2.00 meters
Inner Radius 2.15 meters
Outer Radius 6.55 meters
Width of Cerenkov Medium 1.85 meters
Mirror Reflectivity 85%
Phototube 5/1 BURLE 8854
Expected # of Photoelectrons 17.5
# of Cells per Counter 200
# of Mirrors per Cell 4
Size of Each Mirror ~ 40x45 cm 2
Focal Length of Each Mirror 1.40 meters
Cerenkov Light Ring Radius at Phototube ~3.73 cm
Momentum Threshold for muons 4.25 GeV Ic
MIRRDR~
PHDTDTUBE ~ "
TDROID-
~ INTERACTION POINT
NOT TO SCALE
Figure 3. A schematic description describing the path of a charged particle through the forward
element of the SDC detector. It shows how the Cerenkov light is directed away from the
detecting phototube for a given particle momentum. This describes the process by which
the Cerenkov counter operates as a muon trigger.
TEST BEAM RESULTS
In order to confirm the concepts presented here, we constructed a Cerenkov cell unit reflecting
the design. This unit was then installed in Fermilab's high energy muon beam laboratory behind
E-665. We carried out a series of measurements to determine the number of Cerenkov photon-electrons
observed for various angles of incidence of the muons relative to the mirror-phototube optic axis. This
change in angle was accomplished by placing the Cerenkov unit cell on a rotating platform. The
rotation of this platform was measured to better than 0.10 by means of a sprocketed wheel attached to
the rotating lead screw. A counter would detect the number of times the sprockets would pass by as
the wheel rotated in unison with the platform attached to the lead screw. The mirror-phototube optic
axis is so aligned that the nominal 0 0 occurs when the expected beam direction is normal to the cell
and the Cerenkov ring is focused on the center of the phototube. Under these conditions the number
of observed photoelectrons is a maximum.
593
In Figure 4 we show our measurement of the linearity of the ADC relating the charge associated
with the pulse and the ADC channel count. A small break in the linearity is seen at an ADC count of 50;
this break is taken into account in our results stating the number of photoelectrons being detected. The
performance of the Cerenkov counter was determined for two values of the Cerenkov medium length
by placing a cover at a specific location inside box. This was done to check for consistency in the
observed number of photoelectrons. The two lengths were 148 cm and 215 cm. In Figure 5a we show,
for a Cerenkov length of 148 cm, the single photoelectron peak measured with the counter rotated by
3.5° and 3.75°. This measurement gives us the value of the number of ADC counts and, therefore, the
charge generated by a single photoelectron peak. In Figure 5b we present the collected charge, in terms
of ADC counts, for various angle of rotations. We also state the equivalent momentum that the muons
would have in the SDC design to approach the Cerenkov counter at the same angle. In Figure 5c and
Figure 5d we show the same results when the Cerenkov length was increased to 215 cm. These results
clearly show the effect of the number of photoelectrons detected decreasing with angle and essentially
dropping to zero for angles larger than 4°. All the results presented are for clockwise rotations. We
carried out a complementary series of measurements for counterclockwise rotations and obtained the
same results. This reflects the similarity of the performance of the system on particles of opposite
charge. These measurements lead us to the conclusion that we produce 10.2 ± 1.0 photoelectrons per
meter, in excellent agreement with the calculated value. The details of this determination are shown
in Table 2. These results are then used in the simulation of the counter performance as the trigger
signal in the SDC detector. This is discussed in the next section.
LINEAlUTY OF THE ADC
300
Q=1.402 x ADC + 24.41 -+

U
~
200
"'"
0
It:
ilu
100
... Q=1.511 x ADC + 17.63
0
0 60 100 150 200 250
ADC CHANNEL
Figure 4. The linearity of the ADC used in the Fermilab beam test.
In addition, we have measured the effect of oxygen and the effect of extending the magnetic
shielding from 2" to 4" past the face of the phototube on the number of observed photoelectrons.
The results indicate that the presence of oxygen at the level of a few percent or less does not affect
the number of observed photoelectrons, even though we are coating the surface of the phototube with
Pt-Phenol which enhances the detection of Cerenkov photons in the far ultraviolet. On the other hand,
extending the magnetic shielding 4" past the face of the phototube does increase slightly the number
of observed photons. Hence this arrangement is incorporated in our design.
594
H.V.~ 2896 VOLTS
C LENGTH ~ 148 eM.
.
30000 '" ROT.~ 3.75 DEG. CW, SPE PEAK
""
... ROT.~ 3.50 DEG. CW, SPE PEAK
o
u
_ 20000 ~ PEDESTAL X3
o
...
"
.D
§ SIGNAL
Z lUUOU
o 0~·"""~1~0,,"K:~~~~~~~~~~""4~0~""~50
ADC CHANNEL
40000 r..,,----r~~-~- .. ~-----r~~-~~b~
'" PULSE HEIGHT AT 4.0 DEG. CW (17 GeV/e)

" PULSE HEIGHT AT 3.0 DEG. CW (23 GeV/e)
" PULSE lU,;IGHT AT 2.0 DEG. CW (35 GeV/e)
30000
.. PULSE HEIGHT AT 1.0 DEG. CW (69 GeV/e)
!l x PULSE HEIGHT AT 0.0 DEG. CW
§
o
u
...... 20000 H.V. ~ 2696 VOLTS
o
... C LENGTH ~ 146 CM .
"
~
Z 10000
It" 150
ADC CHANNEL
H.V.~ 2096 VOLTS

C LENGTH ~ 215 CM.
'" ROT.~ 4.00 DEG. CW, SPE PEAK
..
30000
... ROT.~ 3.50 DEG. CWo SPE PEAK

§o
U
_20000 ~ PEDESTAL X3
o
...
"E
.D
"
Z 10000
.. ~ SINGLE P .E. SIGNAL
10 20 ~ 40 .0
ADC CHANNEL
40000 r"~-"~~~-~'--~~--"~~--~d'"
" PULSE HEIGHT AT 4.0 DEG. CW (17 GeV/e)

x PULSE HEIGI·IT AT 0.0 D(';G. CW
.
30000
""
U
o
...... 20000
H.V. ~ 2696 VOLTS
C LENGTH ~ 215 CM .
o
...
"
~
Z 10UOU
100 200
ADC CHANNEL
Figure 5. a) The single photoelectron peak in the Fermilab beam test for a 148 cm. Cerenkov medium.
b) The observed pulse height distribution for various angles of incidence of the muon beam
relative to the mirror optic axis. c) The single photoelectron peak for 215 cm. Cerenkov
medium. d) The associated observed pulse height at 0 deg. muon beam incidence.
595
Table 2. Determination of Number of Observed Photoelectrons.
ITEM ADC Channel Charge (pc) Charge/p.e. # of p.e.
L c =148 cm
Pedestal (0=3.50°) 18.5 45.58
Pedestal (00=3.75°) 18.6 45.73
Pedestal (0=4.00°) 18.6 45.73
Single P.E.( 0=3.50°) 21.9 50.72 5.14

Single P.E.( B=3. 75°) 22.1 51.02 5.29
Single P.E.(0=4.000) 22.5 51.63 5.90
Ave. Charge/p.e. 5.44
# of P.E.(O=O.OO) 70.0 122.55 14.1±1.4
L c =215 cm
Pedestal (0=3.50°) 18.3 45.28
Pedestal (0=4.00°) 18.1 44.98
Single P.E.(0=3.500) 21.6 50.27 4.99

Single P.E.(0=4.000) 22.3 49.81 4.83
Ave. Charge/p.e. 4.91
# of P.E.(O=O.OO) 97.5 161.l0 23.6±0.9
AVERAGE # OF P.E./METER= 10.2±1.0
DETECTOR SIMULATION AND TRIGGER EFFICIENCY
Making use of the pulse height distribution as a function of muon incident angle measured in the
Fermilab test we develop the simulation of the detector performance as a trigger counter. The Cerenkov
medium in the design is about 1.8 meters. We assumed that the phototube observes 15 photoelectron
for the case of a very high momentum muon instead of the expected 18 in order to err on the side
of conservatism. We propagate muons through the various elements of the SDC detector including
multiple scattering and energy loss. The muon is bent by the forward toroids and is then propagated
into the Cerenkov counter. We use the design magnetic field of 1.8 Tesla. It is then allowed to emit
the Cerenkov photons such that the observed pulse height simulates the pulse height from the number
of photoelectrons generated by the phototube for such an incident muon direction. In the propagation
of the photons we include the effect of spherical and astigmatic aberrations produced by the rather
large spherical mirrors. We then determine, how many times the phototube observes the number
of photoelectrons above a specified threshold. This method then leads to our trigger efficiency as a
function of muon momentum. To describe the effect of this simulation effort 4 we show in Figure 6a the
typical Cerenkov light ring on the face of the phototube for a very high momentum, P = 300 GeV /c,
a 50 GeV /c, and a 20 Ge V /c muon. This case is for the center mirror in the smallest radius layer (see
Figure 8) while in Figure 6b we show the case for a corner mirror in the largest radius layer. These
2 cases cover the extreme cases of how the Cerenkov light ring appears at the phototube face. The
effect of multiple scattering is minimal compared to the toroid bending power; the multiple scattering is
about 1/8 of the bend angle and this ratio is independent of momentum. The multiple scattering effect
is clearly seen in the increasing spread of the 50 GeV /c and 20 GeV /c light ring when compared with
that of the 300 GeV /e ring. The scatter in the 300 GeV /c ring is due mainly to spherical aberrations.
From these two diagrams we note that the detection efficiency versus momentum is very similar to the
point that the trigger efficiency versus momentum is the same for all mirrors.
596
In Figure 7a we display the simulated trigger efficiency versus momentum. The two curves are for
extreme threshold requirements. This gives one a sense of the dynamic range in trigger rates available
for this design. In Figure 7b we show the same trigger efficiency versus the transverse momentum for
the five radial layers of mirror cells (see Figure 8). These results are then used to determine the trigger
rate. We use the results of the work of Green and Hedin 1 which present the cross section for a muon to
appear beyond a certain amount of material versus transverse momentum and versus angle. In Table 3
below we present the trigger rates expected under the present design. This design can be altered by
the addition of Winston cones which allows one to reach smaller values of transverse momentum in
the outer layers if one desires to have a more universal trigger efficiency versus transverse momentum
than that shown in Figure 7b. Since the outer layers have a low trigger rate this addition would not
alter significantly the results presented in Table 3. Such a possibility is discussed in a latter section.
These trigger rates are quite low. They are not affected by the low energy charge particle debris
that is associated with the back of calorimeters or materials that are placed in the path of charged
tracks. This is because of the properties ofCerenkov counters; namely, light is only emitted by relatively
high momentum tracks (P > 4.3 Gev / c for muons and P > 21 Me V / c for electrons) and the light must
be within a 5° cone of a direct line from the target. In addition, it is totally insensitive to a flux of
low energy neutrons which might be quite intense in the forward direction.
a D
" -
Ce Y/c:.
-. J"
Pho l o l ube Foce -0
/'
Pho l otu bc Fo ce
EDCE MllUlon UPPERMOST LEVEL

CE TEn MJMOR LOWEST LEVEL - 11
-. .. .... (: m
I' -0
"
Figure 6. a) The simulation of the Cerenkov light ring due to a muon of a given momentum traversing
the various materials of the detector and emitting Cerenkov light in the detector. This
particular distribution is for a particle incident on the center cell, lowest row of the detector.
See Figure 8 for the description of the cells. The effect of multiple scattering is clearly visible
in the light rings of the low momentum particles. The limiting effects of optical aberrations
is seen in the high momentum ring. b) Similar simulation for a mirror in the right hand
corner, uppermost row of the detector. See Figure 8.
MECHANICAL STRUCTURE OF THE CERENKOV COUNTERS
In this section we describe the structure and dimensions of the counter. The counter in each end
is divided into eight sections (octants) located radially and perpendicular to the beam line. In Figure 8
we show a frontal schematic view of such a unit where we show the mirror structure and location of
the 25 phototubes, each phototube observing four mirror cells. The location of the phototube relative
to its 4-mirror cell is located so as to minimize spherical and astigmatic aberrations. The distance
597
100 100
00 uo
g 60 E 00
...c.; ...
c.;
<oJ
'"
'"'-'
t..l 40 '"'-'
t..l 40
'" FIRST. THRESHOLD ~ 7 P.E.
'-' '" THRESHOLD ~ 1 P.E. '-'
[g x SECOND. THRESHOLD = 5 P.K
~ x THRESHOLD = 0 P.E. .,. THIRD. THRESHOLD ~ 4 P.E
20 20 - x FOURTH. THRESHOLD ~ 3 P •
+ I'WfH. THRESHOLD ~ 1 P.E.
20 40 60 66 100 10 15- 20 2. 30
P (GeV/c) P, (GeV/c)
Figure 7. a) The trigger efficiency versus momentum for two extreme pulse height discriminator levels.
b) A possible trigger efficiency versus Pt for the five rows of the detector. See Figure 8.
1 .o-j
,...-----·--------~54S.5_S--------------.,
38.1-r44
4S .5
4 4.0
- 38.77
@ =PM TUBE LOCA TlON

x =4 MIRROR GEO.CENTER
Figure 8. The cell arrangement of an octant section of the Cerenkov element in the SDC detector.
Four mirrors, each of approximate size 45.6 x 38.4 cm 2 are associated with one phototube,
described in the figure by a double circle. There are 5 rows and a total of 25 cells in each
octant.
Table 3. Trigger Rates Summed Over Both Ends.

Trigger Rate per Octant = 1/16
Luminosity = 1033 cm- 2 sec- 1
Layer Radius (m) e (deg.) 1') Trigger Rate
2.64 9.4 2.50 2.3x 10 3

3.55 12.5 2.21 1.1x10 3
4.46 15.6 1.99 4.1x 10 2
5.37 18.6 1.81 1.6 x 10 2
6.28 21.5 1.66 9.2x 10 1
598
from the interaction point to the mirror plane is ~ 16 meters. The depth of each octant is ~ 2 meters
and the volume is 3.9 m 3 .
All eight octants attach to a "spider", a welded structural frame made of Aluminum I-beams.
This frame is shown in Figure 9. The estimated weight of this structure is 27001bs. We plan to have
dowells on the spider for precision mounting of the octants. These dowells will have bearings to allow
for some motion of the octant for alignment purposes. This system will assure precision positioning
of the octants. The spider and the octants are joined on the downstream side. The octants will have
removable panels as described below. The assembly of the octant and spider will still allow for the
removal of the panels in situ. The octants on the upstream side will be bracketed together to add to
the rigidity of the structure. The bracketing will occur on the inner and outer radius.
5.Plutll . (Il.tft; .~S.t"11. y

U or Co..lJilAOO
Figure 9. The spider structure to which each octant of Cerenkov cells is attached.
The octants are constructed as a frame made of structural angle and welded panels (2 sides,
top, and bottom). Both frame and panels make up the integral, rigid part of the structure. Each
side has flanged opening ports . These ports are necessary to service the photomultipliers and mirrors.
The panels consist of a I" honeycomb-aluminum sheet laminates. They are ~ 4 m2 and weigh 1251bs.
Each panel has a flange around its periphery which mates with the flange making up the ports. A
flat rubb er gasket between the panel flange and the port flange will seal the opening when the two are
mated. The estimated weight of each octant, with the mirror weight included, is 4300 pounds.
The panels covering the outer element of the octants will have a gastight recessed utility trough.
This trough is located in the upstream inactive region of each octant so as not to project a shadow on
the mirrors and affect the Cerenkov light collection. The dimension of the trough is 300 cm. in length,
50 cm. in width, and 20 cm. deep. It will accommodate the gas supply and return connections, provide
high voltage connection to the phototube and the output signal connections. In addition, it will have
ample space for the location of the front end electronics. Each one of these troughs will service the
25 photomultipliers in the octant.
In the interior of each octant, parallel to the upstream and downstream sides, separated by about
190 cm., are tubular weldments. They consist of 5 horizontal and 3 vertical pipes. These are shown in
Figure 10. The horizontal pipes will hold the phototube assemblies and cables on the upstream side,
and the mirror assemblies on the downstream side. The vertical pipes serve as stiffeners , specially on
the downstream side, where the mirrors are located. There are 25 photomultipliers and 100 mirrors per
octant as shown in Figure 8. The diameter of the pipes are 2.0" o.d . by 0.25" wall, designed at present
to hold only 3 mm. thick glass. The total weight of the glass is 2801bs. In addition, the hardware to
hold the mirrors in place adds about 5001bs. If thicker glass is needed then thicker support pipes will
be required on the downstream side only. The holders for the phototubes and mirrors will allow us to
adjust orientations in x, y, and z and will incorporate rotational capabilities. This will permit maximal
adjustment of the photomultipliers and the mirrors to obtain good focus location . The spacing between
the mirror plane and the photomultiplier plane is about 140 cm. A 3-d view of the structure is shown
in Figure 11.
The counter will run at atmospheric pressure. Assuming an operating pressure of 0.5" of water
above atmospheric pressure the net average force on a side is ~ 5001bs.
599
,.-----001:'----,
CH(R(N OV UKLOSUR(
U. 0 COLO~ADD
Figure 10. Various views of the mechanical structure of each octant.
.1 ~ (. 1\0'1(1 .... 'l""ar

tNCJ.CWII't 'W'II lot """'-L l.-c. ¥t II
CH R(NKDV-MODULE<TYP)
U or COLORADO
Figure 11. Three dimensional and edge view of the Cerenkov octants.
ALIGNMENT
We need to maintain an active alignment mechanism that will allow us to detect, for every octant,
angular deviations of a fraction of a degree. Our present design is shown in Figure 12. It consists of
arrays of 4 light detecting diode cells on each outermost corner of an octant structure that observes
a small light beam emitted by a light source with a focusing lens attached to elements of the beam
pipe. This light-lens structure is carefully surveyed to be perpendicular to the beam line. A deviation
of 0.25° over a distance of 4 meters leads to a light spot displacement of 1.7 cm, easily observable by
the 4 diode structure. By observing the size of the pulse signal from each of the 4 diodes we can easily
maintain the alignment of each octant relative to the beam line.
POSSIBLE IMPROVEMENTS
One simple future addition to the present design is to add Winston cones to the two outermost
or more layers. One such cone is shown in Figure 13. The max. angle of incidence of the Cerenkov
light on the surface of the Winston cone for the present design is 27° so that a 35° Winston cone
allows for a high collection efficiency. This Winston cone increases the detection radius from 6.3 cm.
to 11 cm. or less depending on the length of the cone. This allows us to reduce the threshold for the
outermost layers shown in Figure 7b from 8 GeV/c to ~ 4.6GeV/c or higher. Similarly for the next
to the outermost layer, by using a slightly shorter Winston cone. Since the trigger rates are already
low, adding these does not affect the basic low trigger rate behavior of the counter.
600
CH[REN 0 OCTANT
Figure 12. The alignment system of the Cerenkov counter. It consists of a pair of light sources
for each octant attached to the beam pipe and a set of four detecting photo-diodes to
detect the light beam. This system would define the plane of each octant and maintain
its alignment.
:35 DEC WI '!:iTO CO. E
E
~
'" 10
__ __ ____- L_ _ _ _- L - L_ _
- II> -10 -$ 0 10 15
X (em)
Figure 13. A diagram of a Winston Cone that would allow us to extend the trigger efficiency to lower
values of P t if deemed desirable.
The present design trigger using the Cerenkov counter requires that each cell be placed in coinci-
dence with a scintillator cell that corresponds in dimension to the Cerenkov cell or with the beam clock
in case the scintillator cell has too much background. The singles rate of the Cerenkov phototube due
to its dark current has been measured to be 100 Hz. Such a rate when placed in coincidence with the
beam clock leads to an accidental trigger rate of ~ 100 Khz which is far higher that the actual trigger
rate due to muons as presented in Table 3. One solution we proposed is shown in Figure 14. It consists
of using 4 2" phototubes that cover the Cerenkov light ring. The Cerenkov signal for the trigger is the
coincidence between any two or more of the phototubes. This then reduces the singles rate from the
Cerenkov unit to a negligible amount, ~ 10- 4 instead of 100 Hz. This reduces the trigger rate, using
the beam counter as the other signal, to a negligible level. The efficiency of this trigger as a function
of the muon momentum is shown in Figure 15 and we note that we hardly loose any efficiency when
compared to the trigger efficiency using 1 phototube which is shown in Figure 7a. For completeness
we also present the trigger efficiency if we demand a coincidence of three or more phototubes. It is
quite clear that this phototube arrangement is effective in reducing the accidental trigger rates while
maintaining the high efficiency at moderate values of Pt.
601
Cl>lll>Nl(OV LlGllT ~lMULATIO
"
/'
-0 Phololube Foce
- 1/ Ct:;NTEfi MlnIlOR LOWEST LEVt:;L
- I" -. "
Figure 14. A four phototube detecting system superimposed on the simulated Cerenkov light ring
of Figure 6a. It shows how the tubes could be used in coincidence with one another to
reduce the singles rate from the Cerenkov cells to a negligible level.
Rao
r.:
"-
~
0::
~ 40
t}
t} !;I 2 OR MORE COUNTERS 0
...ii: x 3 OR WORE COUNTERS 0
o 00 7~ 100 I~
P (GeVjc)
Figure 15. The calculated trigger efficiency associated with the phototube arrangement in Figure 14.
REFERENCES
1. D. Green and D. Hedin, Muon Rates at the SSC, Nuc. lnst. and Meth. A297:11l-120 (1990).
D. Green, Rates for Triggers, SDC-90-00075.
2. K. Johns, University of Arizona. Measurements being carried out at Ferrnilab.
3. We would like to thank Michael Marx for his comments to us that lead us towards the present
design.
4. These rings are the overlay of many tracks that strike the reflecting mirrors so as to cover the
area of the 4 associated with one cell. Hence the number of photoelectrons seen in the fig-
ure is much larger than the nominal 17 per muon.
602
SILICON-BASED TRACKING SYSTEM:
MECHANICAL ENGINEERING AND DESIGN
W. O. Miller, M. T. Gamble, T. C. Thompson, K. A. Woloshun, R. S. Reid,

J. A. Hanlon, F. D. Michaud, G. D. Dransfield, H.-J. Ziock, and A. P. T. Palounek
Los Alamos National Laboratory

Los Alamos, NM 87545
SILICON TRACKING SYSTEM MECHANICAL DESIGN
The Silicon Tracking System (STS) is composed of silicon strip detectors arranged both in a cylindrical
array and in an array of flat panels about the interaction region. The cylindrical array denotes the central region
and the flat panel arrays, which are normal to the beam axis, denote the forward regions. The overall length
of the silicon array is 5.16 m, and the maximum diameteris 0.93 m. The Silicon Tracking System Conceptual
Design Report l should be consulted for the body of analysis performed to quantify the present design concept.
For the STS to achieve its physics goals, the mechanical structures and services must
• support 17 m 2 of silicon detectors and stabilize their positions to within 5 J..U11,
• unifonnly cool the detector system to O°C, while potentially removing up to 13 kW of waste heat generated
by the detector electronics,
• provide up to 3400 A of current to supply the 6.5 million electronics channels, and
• supply all control and data transmission lines for those channels.
These objectives must be achieved in a high-ionizing radiation environment, using virtually no
structural mass and only low-Z materials. The system must be maintainable during its 1O-year operating life.
Tables 1 and 2 enumerate the silicon detector modules and STS performance specifications, while Figure I,
a cutaway view of the STS, identifies its major components.
The method for supporting the individual central region silicon layers combines adjacent layers into
superlayers. The four superlayers in the central region are supported pairwise by two composite support
cylinders composed of graphite/cyanate-ester resin (G/C-E) to minimize structural materials. The maximum
silicon detector module length in the central region is 24 cm and, correspondingly, the maximum shell length
is 24 cm. Three silicon shell assemblies, two 24-cm and one 12-cm long, comprise a 6O-cm-long central
region superlayer. The three shell combination is staggered in z as a function of r, to disperse the structural
support mass, while the edge-to-edge overlap of each silicon layer provides circumferential hermeticity.
Electronics for powering the strips and data readouts are mounted at each end of the module assembly.
Ten electronic read-out chips (128 channels each), five per side of the doubIe-sided strip detector, are required
for each detector module. The silicon module is stiffened by edge-bonding two ultrathin G/C-E strips along
the length of the detector edges. The silicon modules are bonded to the structural rings made from G/C-E to
form the silicon shell assembly. Besides providing structural support, these rings, hereafter denoted cooling
rings, contain an internal heat pipe wick artery structure. Because the detectors in the central region are canted
at an angle of 7.4 0 (Table 1), the top and bottom surfaces of the cooling ring assume a serrated appearance.
Butane working fluid is wicked to this area. Heat from the electronics flows through the thin-walled
composite cooling ring surface, onto which the module is mounted, and evaporates the cooling fluid medium.

Table 1. Silicon detector module specifications. Table 2. STS perfonnance specifications.
Central Region Forward Region

Positional stability: r, r<p , z-80, 5, 250 I'm
Shape: Rectangular Trapezoidal Mounting precision: r, r<p, z-80, 25, 250 I'm
Active detector width: 3.2 cm ~6.4 cm
Number of channels: 1280 2560 Radiation exposure: 10 Mrad over 10 year service
Heat dissipation: 2 mW/channel 2 mW/channel life. No measurable stability
Module heat load: 2.56 W 5.12 W effects >5 I'm.
Thennal conductive area: .8 cmx3.3 cm ~.8 cmx6.5 cm Material radiation Structure and related service
(@ cooling ring surface) length budget: connections < 3% (excepting
Module heat flux: 1.0 W/cm 2 1.0 W/cm2 transition from central to for-
(@ cooling ring surface)
ward and high rapidity regions).
Cant angle: 7.4° 0°
The two forward regions are each composed of thirteen silicon annular disks. The detector modules in
the forward region are not canted, which significantly simplifies the cooling ring design. Large separations
between individual silicon layers make the prospect of combining adjacent layers into superlayers
unattractive. Indeed, only the silicon layers near the central region are in sufficient proximity for this
arrangement to be feasible. Individual modules are arranged circumferentially around the beam axis and are
approximately twice as wide as those in the central region. The forward modules are trapezoidal in shape to
fit as individual arc sectors. Circumferential hermeticity is preserved by overlapping adjacent trapezoidal
modules. Further simplification of the cooling and structural ring construction is achieved by having adjacent
modules in one disk on alternating sides of the cooling ring.
The space frame provides structural support for the cylindrical arrays of the central region and the flat,
planar arrays of the forward regions. The design of this structure is intended to satisfy the STS performance
specifications set forth in Table 2. The dimensional stability requirements motivated the use of metal matrix
composites (MMC) construction materials. Their ultrahigh elastic modulus is supplemented by their
immunity to moisture-induced distortions and by the ability to tailor the longitudinal coefficient of thermal
expansion (CTE) to approach zero. The selection of these materials also facilitates the use of adhesives in
the demountable joints as a design option. The frame features an inner hexagonal geometry rotated about its
,t,£'AL. 41,. ,. rr/f \ -

SP"" CE FRI\III£
"'/'0 r(RI "'LS

C~l II'C€R
Figure 1. STS cutaway view.
604
longitudinal axis to avoid aligning with the outer hexagonally arranged members. TIlls design diminishes the
radiation length experienced by any particle. Finally, the MMC space frame readily satisfies the requirement
for butane compatibility in a high-radiation environment.
In addition to containing the evaporated butane coolant, the enclosure also provides the primary conduit
for transferring the mechanical loads on the silicon tracker space frame to the outer tracker space frame. The
outer structure and the kinematic mounts prevent extraneous loads from being directly imposed on the
delicate silicon structures while maintaining precision alignment. Hence, the outer shell structure is a key
member in maintaining STS stability. An ultralightweight G/C-E sandwich construc'ion has been chosen for
the outer shell and two end cover plates. A O.5-mm thin beryllium shell will be used for the inner liner. TIlls
combination will provide a stiff low-mass structure.
Materials Investigations
The materials selection for the STS is motivated by three primary factors: (1) 5-~m stability in r 11>, (2)
3% radiation length restriction, and (3) radiation resistance in the presence of hydrocarbon fluids. The large
scale of the detector dictates that linear expansion effects must be minimized, whether from thermal or
moisture effects. During the conceptual design effort, materials choices were narrowed to MMCs, whose
CTEs can be tailored to be approximately zero, and to hydrophobic cyanate-ester resin systems for the 0/
C-E composite candidate. Although the cyanate-ester resins have low moisture saturation levels, typically
1% (producing an overall laminate coefficient of moisture expansion (CME) that is 1/50 that of a OlE
laminate), uncertainties still exist in overall stability in a radiation and butane environment. The CTEs of G/
C-E composites are tailorable and have long radiation lengths. Experimental verification of the CTE and
CME of the G/C-E system is in progress; Table 3 lists the experimental results of the mechanical properties
verification of this materials system.
Radiation Length Considerations
Throughout the design process, the need for ultralightweight, low-Z, radiation resistant materials will
be stressed. Structures are configured to have minimum material thicknesses along the particle trajectories.
The design requirement is expressed in terms of the number of equivalent radiation lengths. The design is
restricted to using no more than 3% of a radiation length at normal incidence for all materials exclusive of
the silicon detectors. Figure 2 illustrates the radiation length associated with silicon and the silicon detector
support material, detector electronics, cabling, cooling, etc., expressed as a function of rapidity. It is clear that
the baseline design nominally meets the requirement at zero rapidity «3% ), but as expected, its contribution
increases slightly with increasing rapidity. This is an area of concern, and the preliminary design effort must
address how further reductions in support materials may be attained.
COOLING RING DEVELOPMENT

Mold Development
A 30° cooling ring arc segment mold was fabricated to produce partial cooling ring prototype
components for compression molding evaluation. Small components present similar technology issues as
their larger counterparts. When an epoxy-based molding compound proved unsuccessful, MX4925, a
Table 3. Mechanical test results of G/C-E materials system P75/954-3.
Test Neat Resin Longitudinal Transverse Quasi -isotropic

Tensile Modulus 3.8 Gpa 323.8 GPa 6.4 GPa 96.5 GPa
Tensile Strength 58.6 MPa 1 GPa 23.4 MPa 293.8 MPa
Poisson Ratio 0.39 0.32 0.0063 0.33
Compo Modulus 3.4 GPa 268.7 GPa 6.2GPa 77.2GPa
Compo Strength 6.9MPa 337 MPa 41.3 MPa 131 MPa
Shear Modulus 0.96GPa 4GPa 1.6 GPa
Shear Strength 27.6 MPa 40MPa 19.3 MPa
605
phenolic-based compound, was used because of its ease of fabrication. Both compounds were composed of
approximately 63% graphite fibers and 37% resin. The cooling ring wall thickness of 0.4 mm is substantially
thinner than typical compression molded components and proved troublesome. Poor material flow, quick
curing, and difficult uniform charge deposition complicate fabrication. Ultimately, molding success was
achieved using MX 4925 molding compound, and the desire to use an epoxy resin was replaced by plans to
use an advanced hydrophobic resin system.
Cooling rings composed of a very high elastic modulus graphite fiber and hydrophobic cyanate-ester
resin are the ultimate objective of this program. Hence, a molding compound composed was produced ofP7 5
graphite fibers that were ball-milled to approximately l00-llm lengths and mixed with 954-3 resin. These
were significantly shorter than the 2.5-mm-Iong fibers used with the phenolic resin. The mold was again used
to produce 30° arc segments from this compound. This resin system exhibited much more flow than the
phenolic but failed to evenly distribute the fibers. The parts also tended to adhere tenaciously to the mold.
These problems were overcome by carefully shaping the partially cured, or advanced, charge in the mold that
had been Teflon coated. Uniform arc segments with well-filled walls were produced.
A larger mold, spanning 120° of arc, was fabricated to further this compression molding evaluation.
Problems have been encountered because the larger mold possesses subtle geometric complications and it
requires more time for filling, hence allowing resin hardening during charge deposition. Resin flow and fiber
uniformity issues are difficult to control using this new mold. High quality components are being produced
with the mold but the rejection rate must be minimized. Once this is accomplished, material qualification and
thermal conductivity enhancement will be undertaken using this mold. Figure 3 is a photograph of high-
quality 30° and 120° arcs produced. Ultimately, a 360° cooling ring mold will be fabricated, and rings will
be produced for prototype testing, manufacturing optimization, and deployment into the STS.
Thermal Conductivity Enhancement
Chopped fiber/resin compression molded products are theoretically quasi-isotropic. The fibers,
however, tend to align with the mold wall, yielding much higher in-plane physical properties than those
transverse to the wall. More specifically, both mechanical and thermal properties are higher in the wall plane.
Figure 4 depicts an ultrasonic image of the distribution of fibers in the plane of a compression molded arc
segment. Material physical properties transverse to the wall are fixed, largely by the matrix physical
properties. Because the resin and the transverse fiber thermal conductivity are low, the transverse wall
thermal conductivity is typically quite low. This situation is of significant concern for the STS.ltis important
to minimize the thermal gradient from the electronic readout chips to the cooling system's evaporative wick
structure. If it is not minimized, the silicon strip detectors will operate above the O°C base temperature chosen
to maximize performance amid a radiation environment. Enhancement of the transverse thermal conductiv-
ity is in progress, while emphasis is concurrently placed on obtaining uniform cooling ring properties.
The thermal gradient through the cooling ring was investigated for conditions simulating the
polystyrene wick. Finite element (FE) solutions were obtained as a function of transverse thermal
conductivity (k,) for a constant heat flux. This solution isolated the fm effect of the cooling ring sidewall. For
0.10
0.08
Figure 2. Average fraction of a radiation length

0.06 in the silicon detector expressed as a function of
pseudorapidity (11 = -In (tan 9/2), where 9 is the
angle with respect to the beam axis). a) Silicon
0.04 detectors alone (central and forward regions
~,.. " !I'\"'" ~'i ':'I,.\jl!o'.~~ "'!"~j'''tr ~ shown separately below, and both combined),
r'~'~i:t;. b) including the gas inside enclosure vessel, c)
0.02 . '. ":~t"''''''''__ ' ' '\' including the beam pipe, d) including the outer
enclosure vessel wall, cooling rings and wicks,
support cylinders, and the space frame averaged
over<jl, and e) including the electronics packages,
cabling and inner enclosure vessel wall.
606
Figure 3. 30° and 120° partial cooling ring arc segments.
low k" e.g., 0.5 W /moC, characteristic of the MX 4952 carbon/phenolic system, heat flows down the sidewall,
as well as through the top surface of the ring. The peak: temperature gradient is ameliorated by this effect. For
higher assumed kt values, the gradient directly across the ring top surface diminishes significantly. Because
the temperature gradient is reduced noticeably, the sidewall fm effect diminishes similarly. In short, as the
transverse wall thermal conductivity of the molded ring is improved, a significant reduction in the
temperature gradient through the wall is achieved. The temperature gradient in the cooling ring sidewall
becomes insignificant as the fm effect diminishes to zero. The results of this study clearly identified the benefit
of achieving a k t of roughly 4 W /moC. For this case, the temperature gradient becomes the temperature drop
across the wick, 2.5°C for the effective wick convective film coefficient chosen.
Numerous ring segments and material specimens have been fabricated and their thermal conductivities
measured during the initial phase of this investigation. We studied the effects of high-conductivity additives
in the resin and the potential benefit of using shorter fiber lengths. A seven-fold increase in kt, up to
3.6 W/moC, has been achieved. Although the results of this program are quite encouraging, they are
preliminary. Noticeable improvements to this particular aspect of the ring design have been achieved, and
a factor of 10 increase over the earlier development ring models is ultimately expected. The fmal results of
this investigation will be reported in the coming STS research and development report.
Thermal Stability Studies
An analytical assessment of the ability of the silicon substructures to resist thermally induced
distortions is in progress. Precise measurement of the charged-particle momentum requires an ultrastable
platform. It is important to note that the STS cooling system under development is intended to maintain a high
degree of thermal stability. The evaporative cooling system should preclude any significant thermal gradient
Figure 4.0ne-mm-wide ultrasonic image of graphite fiber distribution in a compression molded component.
607
within the support structures. An understanding of the structure's sensitivity to thennally induced distortions,
will alleviate concerns regarding structural stability. An FE model of the central region silicon shell module
has been developed and will aid in the design and construction of a prototype to be used in stability tests. As
the STS design progresses, this analytical effort will be expanded.
An FE model of a typical central region structure was developed that represents a structure with outer
and inner diameters of 420 mm and 360 mm, respectively, and a length of 240 mrn. Thirty-six silicon strip
detector modules are attached to each diameter. The cooling ring cross-section is 8 mrn wide by 30 mrn high,
with a wall thickness of OAO mIll. The cooling ring will ultimately be a compression molded G/C-E resin
composite component, possessing approximately 60% fiber by volume. The analysis was perfonned
assuming mechanical properties characteristic of IM7 graphite fibers embedded in an epoxy matrix. This
combination yields an elastic modulus of approximately 80 GPa. Verification of the fmal compression
molded cooling ring properties will be undertaken. A larger modulus of elasticity is anticipated for the fmal
product than is used in this analysis. Boundary conditions applied to the model simulate kinematic restraints
at three equally spaced locations on each end. Figure 5 depicts the distortion that occurs when the temperature
of the outer layer of silicon is allowed to rise unifonnly. For small temperature changes, the distortion of the
structure is linear with temperature.
For the size module studied, a 6°C to 8°C temperature difference between the upper and lower silicon
layers composing a silicon superlayer could be tolerated because of the small, 2.5-f.U!1 radial and 0.94-f.UI1
axial, displacements that are induced.. The larger silicon shell modules will be more flexible and are likely
to exhibit more distortion under this temperature variation.
EVAPORATIVE COOLING SYSTEM WICK DEVELOPMENT
The dimensional stability requirements of the SSC silicon detector array dictate precise thennal control
of the entire array to a nominal operating temperature with minimal thennal gradient within the array. The
cooling system developed to achieve this thennal stability is by the free-surface evaporation of a cooling
liquid in near-equilibrium with its vapor. I- 2 Thus, the entire detector containment vessel is filled with the
coolant vapor and liquid is distributed to the cooling rings by a combination of hydrostatic and capillary
forces. Based on the desire to maintain a nominal pressure of I atrn while operating near DoC, butane was
chosen as the coolant.
The critical research and development component of this system is the distribution wick. The wick must
have sufficient capillary pumping capability to overcome its flow impedance. The gravity contribution
dictates a maximum pore size of 5A f.U!1. The flow impedance is minimized by providing a low-impedance
artery circurnferentially around the cooling ring. However, even a short flow path through the wick can
contribute a substantial pressure drop if the wick penneability is low. The actual pressure drop contribution
caused by flow impedance is a function of wick thickness, which must be minimized; proximity of the heated
surfaces to the circumferential artery; and wick penneability, which is coupled to the pore size.
Based on these requirements, the wick development approach has been to examine candidate wick
structures that have a maximum pore size of 4 f.U!1 or less, with a high enough penneability to keep the wick
thickness to an acceptable value, that can contain a distribution artery and, that can be integrated into the
complex geometry of the cooling rings in the central region.
The baseline wick material is a polystyrene foam that can be molded to the desired shape and can
incorporate the distribution artery in the molded fonn. The foam is produced by mixing water with a monomer
with a ratio of 90/10. An emulsifying liquid and an initiator is added to the mixture. The emulsion is
polymerized at 60°C and then dried. The resulting foam has densities of less than 0.1 g/cm3 and a porosity
of 95%. A scanning electron micrograph of a typical foam wick cross section is shown in Figure 6. The
structure consists of a honeycomb pattern oflarge pores connected by smaller pores that govern the effective
capillary pore size. This structure has the potential of offering a relatively high penneability for a given
capillary pore size.
The desired pore size has not yet been achieved, but it has been bounded. A 30-f.U!1 effective pore radius
sample has been machined from bulk material. Micrographs of this material show pores sizes less than 10
11m, which suggests that the effective large pore size is a result of damage to that structure due to machining.
Geometrically simple molded wick samples are currently being produced that have a capillary pore size of
less than 111m. The penneability of these molded samples is unacceptably low. Development efforts are
focusing on tailoring the pore size and increasing the penneability by altering the foam mixing process.
608
Figure 5. FE model of a typical central region structure, Figure 6. Electron micrograph of the polystyrene wick
deformed configuration. structure demonstrating pores of 6-10 J.1m diam.
Temperature drop was measured as a function of power density at the cooling ring surface, across a
cooling ring/wick assembly. The test apparatus was a 30° arc segment on a 4O-cm radius, with electronic
resistive heaters on each of six flats. The measured temperature drop for I W/cm2 is approximately 4.5°C.
Further experiments were conducted to estimate the effective thermal conductance across the wick
only. A resistance heater is mounted on a thin aluminum plate. The wick is attached to the opposite side of
that aluminum plate. The measured temperature drop is from the center of the aluminum plate to the wick-
vapor interface. Wick thicknesses o£O.25, 0.50, and 1.25 mm were tested. An effective conductance of about
0.3 W/cm 2 °e was calculated from the temperature measurements. This conductance was independent of
wick thickness and power density in the range of 0.4 W/cm2 to 1.3 W/cm2 •
These results, together with the assumption that wick pore size can be tailored to the desired value,
indicate that the polystyrene wicks will meet the system design requirements. These wicks will be molded
to the cooling ring shape in machined delrin molds thereafter bonded in place. Delrin has good machinability
for cost -effectivemold production, and the good wettability, which assures foam pore configurations at the
foam-mold interface are essentially the same as in the bulk material.
ALIGNMENT AND ASSEMBLY CONSIDERATIONS
The overall alignment and assembly concept must be integrated with the mechanical structure in many
important ways. The concept must provide a method for positioning silicon ladders to micron accuracies
while they are being attached to the cooling rings and must provide a method for integrating the structural
shells and forward region planar arrays onto the detector space frame. The concept must also include
alignment considerations for the completed detector, including alignment shifts in shipping, reassembling,
and positioning the detector onto the sse beam line; in situ alignment; verification; calibration; and
knowledge of alignment changes over the detector operating life. These alignment and assembly tasks are
necessary in addition to component and subassembly testing, structural verification, and correlation of test
data with FE models, which are also important components of the overall alignment and assembly process.
One of the most important and challenging alignment tasks is to develop a time-saving alignment
maintenance system for the calibrated detector. With a recently developed fiber optic system, it will be
possible to continuously monitor the relative three-dimensional position of one subassembly relative to
another and the position of the detector to the outside world using a compact, nonintrusive fiber-optic motion
sensor that functions by observing variations in light intensity. For this to be possible, the integrity and
stability of the subassemblies must be assured so that the subassemblies can be treated as rigid bodies. This
is another important reason for careful subassembly testing.
609
Detector Stability
The stability of silicon ladders, barrel, and planar subsystems will be tested using TV holography which
provides quantitative and qUalitative infonnation about object shape changes in the presence of heat or
mechanical forces. 3 The results will be used to optimize subassembly designs, aid in the choice of
appropriate adhesives, benchmark FE models, and provide experimentally tested analytical models for
predicting many other detector mechanical perfonnance features or structural perfonnance characteristics
under unusual operating conditions. TV holography offers submicron shape-change resolution and
hundreds of microns of dynamic range over a l-m field of view using 30 mw He-Ne lasers. The system
operates like an optical interferometer for objects, which are not of interferometric quality, by comparing
the shape of the object under load with its unloaded shape. Fringe patterns provide the infonnation about
shape change. Recent TV holography advances provide a way to interpret the fringe patterns and obtain
quantitative infonnation about the test object. As an example of the technique, a silicon ladder glue joint
was tested using TV holography. Two silicon pieces, each 30 mm x 60 mm x 0.3 mm, were glued together
along the 30 mm x 0.3 mm sides with Epoxy Technology H70S adhesive. The assembly was secured and
loaded by deflecting its free end. The arrow in Figure 7 marks the position of the butt joint, while continuity
of the fringes across the joint indicate that the bonding is mechanically equivalent to a single piece of silicon.
CONCLUSIONS
The STS mechanical engineering and design is dedicated to producing a system capable ofaccomodating
the detectormodu1es specified in Table 1 while achieving the perfonnance requirements enumerated in Table
2. This will be accomplished by using high-elastic-modulus graphite fibers embedded in hydrophobic resin,
and MMCs for the cooling rings and space frame, respectively. The long-radiation-Iength cooling rings have
CTEs nearly equal to that of silicon, and include maximized out of plane thermal conductivity to avoid
thermal distortions, while the space frame longitudinal CTE approaches zero. Thermal gradients are
minimized by the evaporative cooling system. Alignment and assembly tasks, as well as long-term stability
interrogation, are accomplished optically, using real-time quantitative holographic means.
Figure 7. Continuous fringe pattern across silicon detector butt joint.
REFERENCES
1. W. O. Miller, M. T. Gamble, T. C. Thompson, et al., Superconducting Super Col/ider Silicon Tracking Subsystem
Research and Development Report, LANL Report LA-12029 (1990).
2. W. O. Miller, T. C. Thompson, M.T. Gamble, et aI., Mechanical engineering and design of silicon tracking devices,
in: "Proc. of the Symposium on Detector Research and Development for the Superconducting Super Collider,"
T. Dombeck, V. Kelly, and G. Yost, eds., World Scientific, Singapore (1991).
3. R. J. Pryputniewicz and K. A. Stetson, Measurement of vibration patterns using electro-optical holography, in:
"Laser Interferometry: Quantitative Analysis of Interferograms," Proc. SPIE 1162:456 (1989).
610
THE SOLENOIDAL DETECTOR COLLABORATION SILICON
DETECTOR SYSTEM
Hans-J. Ziock, Michael T. Gamble, William O. Miller, Andrea P. T. Palounek,

and Timothy C. Thompson

Representing the Solenoidal Detector Collaboration
INTRODUCTION
Silicon tracking systems will be fundamental components of the tracking systems for
both planned major SSC experiments. Fig. 1 shows the central tracking system and the
calorimeter of the detector being proposed by the Solenoidal Detector Collaboration (SDC).
The small shaded central area is the Silicon Tracking System (STS). Despite its seemingly
small size, it occupies a volume of more than 5 meters in length and 1 meter in diameter and
is an order of magnitude larger than any silicon detector system previously built.
'1: ....
UlL..!--:;---- - 'j: '.f':
Fig. I. The central tracking system and the calorimeter of the detector being proposed by the SOc. The
small shaded region in the center is the silicon tracking system. (lP) designates the interaction point where
the beams intersect and the events originate.

GEOMETRY
An isometric view of the STS is given in Fig. 2 and a cross sectional view of the system
including the various support hardware is shown in Fig. 3. The STS is symmetric about the
interaction point, which is located at the center of the detector. It is where the two counter-
rotating proton beams collide. The STS consists of a central region in which there are 8
cylindrical detector layers of 30 cm half-length. In addition there are two forward regions
each of which contains 13 annular planar arrays of detectors. The system is configured to
cover a large solid angle in order to have reasonable acceptance for all the relevant particles in
interesting events. The detector also provides a minimum of 6 position measurements over
as large a radial extent as possible for any particle that is within its acceptance. The transition
from a cylindrical to disk geometry provides a more effective coverage with a minimal
amount of silicon, and also reduces the pathlength of the particles through material (i.e. the
particles never strike the detectors at grazing incidence). The system consists of
approximately 17 square meters of silicon micros trip detectors, which are subdivided into
approximately 6.5 million independent channels. Considerable effort has gone into
optimizing the STS design to ensure that it can perform the required physics measurements
while meeting budgetary constraints. .
Fig. 2. An isometric view of me STS showing thc 8-laycr central region and the two 13·layer forward
regions.
CUTER ENCL OSl-RE SEAL, GRAPH I TE /EPOX WI Th

CARBON FOAM INTE RI OR
,
~=;:::=:!::;;t;:~======!;:~=====:;==' =~t:± .. , :
-, I
I
I
MATR IX COMPOSI TE
FRAME
,
I
FLEXIB_ E II -EPFACE....J
SUPPOR T CY LINDERS
Fig. 3. A cross sectional vicw of the STS showing mc various support hardwarc.
612
PURPOSE.
The STS is one of three components which form the central tracking system of the
detector being built by the SDC. The STS will consist of silicon micros trip detectors and
possibly silicon pixel detectors. The other two components are an outer barrel tracker, which
will consist of straw tubes or scintillating fibers; and an outer intermediate angle tracker,
which will consist of gas microstrips. The components are designed to work as an integrated
system. Each component has specific strengths, but is individually incapable of providing
the overall performance required by the physics goals of the sse. The STS contributes
substantially to all of the functions described below.
Pattern Recognition
End and side views of a typical interesting event are shown in Fig. 4. Each solid line
represents a charged particle's track as it traverses the detector. One of the goals of the
tracking system is to find all the charged particle tracks shown. Unfortunately, the tracking
system is incapable of measuring an entire track, but only measures discrete isolated points
on each track. Those points must then be connected with very high reliability and very high
efficiency to the other points generated by the same particle. By performing a "connect-the-
dot" operation one is able to reconstruct that track. From a first glance at Fig. 4, that appears
to be impossible. However, Fig. 5. shows that even the most dense region of tracks is not
intimidating when magnified. The trick is to have sufficient magnification, and then to look
for a set of points that lie on a constant trajectory through multiple layers. Sufficient
magnification is achieved by having a large number of very small detector elements so that
two different particle tracks are only rarely intercepted by the same detector element. The
verification of a constant trajectory requires a sufficient number of separate layers so that the
trajectory is measured many times. These requirements necessitate the large number of
channels and the multiple layers of detectors found in the STS.
Fig.4a. Side view of the tracks in the central tracker for a typical event of interesl Charged particles are
represented by solid lines, neutral particles by dashed lines.
613
I~
Fig. 4b. End view of the tracks in the central tracker for a typical event of interest. Charged particles are
represented by solid lines, neutral particles by dashed lines.
E
u
<rI
Fig. 5. A magnified view of the event depicted in Fig. 4b showing a region of very high track density. This
illustrates the low track density at the scale of the individual sensitive clements of the STS.
614
Charge determination
The charged particles produced in the collision of the proton beams follow a curved path
due to their interaction with the 2 Tesla magnetic field in which the tracker is located In three
dimensions, the particles follow helical paths, which if viewed from the end of the detector,
will appear to be a circle, or at least a small part of one. Depending on the direction of the
magnetic field and from which end of the detector the event is viewed, positively charged
particles will follow a clockwise or counterclockwise arc, while negatively charged particles
will curve in the opposite sense. The sense of the arc, hence indicates the charge (±) of the
particle. In contrast, neutral particles are not affected by the magnetic field and furthermore
are invisible to the STS. However, these particles are detected by the calorimeter. They are
identified as being neutral by a lack of hits in the silicon system along the expected track of
those particles from the interaction point to the calorimeter.
~omentum measurement
As mentioned above, charged particles follow a curved path while traversing the tracking
volume. By measuring the radius of curvature of a track it is possible to determine the
component of the particle's momentum transverse to the magnetic field. The magnetic field is
parallel to the beam axis, and, if in addition one measures the angle of a track with respect to
the beam axis, one can determine its total momentum. The problem is that high momentum
particles follow an arc which has a very large radius of curvature and over the radial
pathlength measured by the silicon detector deviates from a straight line by only a small
amount (e.g. a particle with a 55 GeV/c transverse momentum has an -90m radius of
curvature and deviates from a straight line by only about 100 J.l.m over a 27 cm pathlength).
That again illustrates the need for a very high resolution, finely divided, position measuring
device. It can be shown that the deviation of an arc from a straight line increases as the
square of the radial extent over which one follows the path of the particle. That is one of the
primary reasons for having the other two components of the overall central tracking system.
They provide measurements of the position of the particles at large radii.
Impact parameter measurement
Although the primary interaction occurs in an extremely short time (-10-24 sec), some of
the exotic particles created in the reaction live for relatively long periods of time (_10- 12 sec ),
and in that time can travel hundreds of microns. (Special Relativity helps in that the particles
have such a high velocity that from our point of view they exist on the order of ten times as
long as if they were at rest.) The STS has sufficient position resolution to show that some of
the particles originated from a particle that decayed a finite distance away from the primary
interaction point. An example of this is shown in Fig. 6. By measuring the distance the
parent particle travels from the interaction point to its decay point, and the parameters of the
particles into which it decayed, the initial particle can typically be identified, and its lifetime
measured. The ability to determine the origin of the particles also helps in separating events
from one another. Often when the two proton beams interact, there is more than one primary
interaction. The multiple interactions are typically separated by relatively large distances,
centimeters, from each other. By identifying where the particle tracks originated, one can
cleanly assign them to one interaction or the other.
Time stamping
The two counter rotating proton beams at the SSC will cross and collide with each other
at the heart of the detectors once every 16 nsec and produce an average of 1.6 interactions or
events. The vast majority of these are uninteresting interactions and are called minimum bias
events. The preceding values translate into 100 million events every second. In addition,
each minimum bias event has on the average 40 charged tracks each of which produces on
the average 4 hits per layer in the silicon detector system. The interesting events that we keep
for further study occur at rates on the order of 1000 per second, while the truly interesting
Higgs events are expected to be produced at rates of only -100 per year. In order to unravel
this mess and have a chance of assigning tracks to a unique event one would like to be able to
assign every event to a particular beam crossing. That requires the silicon system to measure
615
E
u
Fig. 6. A reconstructed detached decay vertex. The point (P) is the interaction vertex, and the point (S) is
where one of the primary particles decayed as indicated by the intersection of the projected tracks from its
decay products.
the time of a hit with a total accuracy which is better than the time between beam crossings,
namely 16 nsec. From the above, it is evident that the silicon tracker generates a tremendous
amount of information, namely (4 hits!layer/track· 8 layers· 40 tracks/event· 100,000,000
events/second), or about 0.1 trillion hits every second.
THE DETECTORS
The basic component of the STS is a silicon microstrip detector. A detector for the
central region is shown in Fig. 7. It consists of a 300 11m thick piece of a single crystal of
high resistivity (-5000 ncm) silicon approximately 3.4 cm wide and 6 cm long. A bias
voltage of approximately 100 V is applied between the top and bottom surfaces of the
detector. When a charged particle passes through the 300 11m thickness of the detector it
ionizes a number of the silicon atoms, creating about 24,000 electron-hole pairs. The
positively charged holes are attracted to the negatively biased top surface of the detector,
while the electrons are attracted to the positively biased bottom surface. The collected
electrons and holes are the detected signals. Position information is obtained by using
photolithography to subdivide both the top and bottom surfaces into a series of fine strips.
The spacing, or pitch, of the 10 11m wide strips is 50 11m. The strips span essentially the full
6-cm-Iength of the detector. There are 640 strips that cover the 3.2 cm active width of the
detector on both the top and bottom surfaces. The bias voltage is actually applied only to the
strips which therefore alone collect the generated electrons and holes. The electrons and
holes move respectively to the closest strip or pair of strips on the bottom and top surfaces
and thus generate a signal only on those strips. By reading out each strip individually one
can isolate where the charged particle passed through the detector. The signals collected on a
single strip are amplified and compared to a threshold value and only a hit (1) or miss (0)
value is stored. There is a unique amplifier, comparator, and data storage array assigned to
each strip. This is done using very large scale integration (VLSI) electronics which are
mounted on the detectors and have the same 50 11m pitch as the strips. If a strip is hit, the
position of the center of that strip is assigned to that hit. Thus the maximum position error is
half the strip pitch or 25 11m, and on the average the error is even less. The resolution of the
616
Axial Strip Side Structure
a
Ibondlnc
pod. forblu
I
-~
'"
-~- bias
"'~~
&
guard
-~ ring
"""'.,.,
--( I
-~
-~
.
2
~::"~
~ J
-~
~
• -~
..7.,
• -~
~ 1
~
bonding pad poly~Si

rcsisl<>r
bonding pad bias rcsisl<>r

t
strip . / n~substratc
p.~
t-O
•
t>O
t
h
C
300).lm v I<.
n n
n e
n
s
__ wIdth _
s
t
Fig. 7. a) A top view of a silicon microstrip detector. b) A side view of the detector illustrating the
collection of the generated electrons and holes.
position measurement from a single side of the detector is about 14 ~m in one dimension.
The coordinate in the second dimension is determined by having the strips on one side at a
small "stereo" angle with respect to those on the other side. The value of the coordinate in
the second dimension is given by the the region over which the hit strips on the top and
bottom sides of the detector overlap. That, and the concept of ghost hits are illustrated in
Fig. 8. The maximum resolution in the second dimension would clearly be best if the stereo
angle were 90°. This, however, also maximizes the problem of ghost hits.
For example, when 3 particles strike a detector, 3 strips on both the top and bottom
surface of the detector are hit. If the stereo angle is 90°, the 3 hit strips on the top surface
cross the all 3 strips on the bottom surface and one has a total of 9 places where the strips
overlap. There are thus 3 real top-bottom combinations, together with 6 ghost combinations.
Since there is no way to correlate a specific hit on the top surface with a given hit on the
bottom surface, there exists a very large ambiguity. By reducing the stereo angle, a strip on
the bottom surface crosses only a few strips on the top surface, greatly reducing the chance
of producing ghost hits. The undesirable consequence of reducing the stereo angle is the
reduction of the resolution in the second dimension, as now a given strip on the top surface
overlaps a given strip on the bottom surface over a longer distance. However, since the
required resolution is much poorer in the second dimension and it is very important to
617
90° stereo angle real hit ghost hit "small" stereo angle real hit
Fig. 8. An illustration of a double sided detector and the concept of ghost hits. Note that the small angle
stereo detector has no ghost hits, while the 90° stereo angle detector has 6 ghost hits.
minimize the number of ghost hits, a small stereo angle is chosen. The SDC STS stereo
angle is 10 milliradians, which corresponds to a resolution of approximately I mm for the
second coordinate. The position in the third dimension is determined by the accurate
placement of the detector. To obtain sufficient accuracy in the different dimensions, it is
required that the detectors be positioned to an accuracy of 25 Jlm and that they maintain that
position to an accuracy of 5 Jlm over long periods of time.
CHALLENGES
The large particle fluxes, the short times between beam crossing, the high channel count,
and the required very high position measurement accuracy pose challenging problems that
must be solved. Furthermore, to avoid degrading the measurements, the solutions must be
achieved using only a minimal amount of material. An additional constraint is that only low-
Z materials are allowed. If that were not difficult enough, the solutions must also be
affordable. The requirements and problems facing the STS group include:
• The large particle fluxes expose the detectors and electronics to radiation doses of several
Megarads over the lifetime of the experiment.
• The 6.5 million channels of amplifiers, comparators, data storage, and readout are
required to fit on the ends of the detectors. As the signals are very small, the electronics
must produce very little noise and be immune to noise. Furthermore, they must operate at
speeds in excess of 60 MHz and consume very little power; the design goal is 1 mW/chan.
• The 6.5 million channels of electronics, even at I mW / channel, require a total of -2000 A
of low voltage current, and generate 6.5 kW of waste heat which must be removed.
• In addition to removing the waste heat, the STS must operate at approximately O°C to
reduce some of the effects of radiation damage and improve the detector performance.
• The detectors must be placed with an accuracy of approximately 25 Jlm and must remain
positionally stable at - 5 !lm for periods of many months. The stability must be achieved
over the entire 5 m tracker and must include freedom from vibrations.
• The detector must be maintainable and repairable.
CONCLUSION
Because of the formidable problems and large list of secondary issues associated with
each one, an incredible amount of work remains. Nevertheless, several years of research and
development work, summarized in references I and 2, has significantly assisted in
developing solutions to these problems. The next several years of research and development
will go to implementing and testing the solutions on a wide variety of prototypes.
REFERENCES
I) A. Weinstein el ai., "Silicon Tracking Conceptual Design Report", Santa Cruz Institute for Particle
Physics report no. SCIPP 92/04 (January 1992).
2. W. O. Miller el ai., "Mechanical Engineering and Design of Silicon Tracking Devices", these procecdings.
618
EFFECTS OF MOUNTING METHODS ON
TEMPERATURE SENSOR ACCURACY BELOW 10K
Theodore A. Kobel, Maksymilian A. Kozyrczak 1,

S.W. Schwenterly2, and William M. BeU3
IProcess Systems International, Inc.

20 Walkup Drive
Westborough, MA 01581-5003
20ak Ridge National Laboratory
P.O. Box 2009
Oak Ridge,TN 37831-8070
Yropsfield Engineering Service
57 Water Street
Hingham, MA 02043
ABSTRACT
Silicon diode cryogenic thermometry has been designed and tested for use on
the Cryogenic Systems of the SSCL Magnet Test Facilities (MTL, ASST). This paper
describes the test setup, results of testing, final design configuration, and performance
specification.
INTRODUCTION
The design goal was to deliver silicon diode thermometry with combined loop
accuracy of not less than ± 150 mK in the range 2 to 6K and with thermal response
time comparable with in-stream mounted sensors. Silicon diode mounting techniques
were tested in a cold box. The test results indicated the following: 1) a sensor
attached to the outside of the pipe measures the temperature with comparable
accuracy to a sensor mounted directly in the process stream for slow changing process
conditions, 2) there is an alternative mounting method which approaches the accuracy
of an in-stream sensor, 3) radiation effects are negligible on sensor operating below
80K, and 4) low mass flow of process gas has impact on the sensor reading. Based on
the test results, two thermometers were developed: one for elevated accuracy with
quick thermal response time, and one for less critical, nominal accuracy
requirements. All silicon diode tests were performed using the Lake Shore
Cryotronics, Inc. diode DT-470.

THERMOMETER TEST SETUP AND RESULTS
Samples and Apparatus
Figure 1 shows outlines and descriptions of the various diode packages tested.
The diodes were specified to conform to Lake Shore's Standard Curve #10 within the
following accuracy: ± 100 mK for 4K to 6K and ± 300 mK for 6K to SOK. Figure 2
shows the test cold box with sensors mounted on a I-inch Schedule 10 stainless steel
pipe in various ways. Two type BO diodes were mounted directly in-stream and
located 2 inches above the bottom end of the pipe. The diode leads were brought out
through a liB-inch capillary to a room temperature feedthrough in the top plate.
Two diodes (one type BO and one type CU) were mounted on the outside surface of
the pipe with 4-40 screws at a position roughly even with the in-stream diodes.
Copper pads were soft soldered to the side of the pipe to provide a diode mounting
surface which could be drilled and tapped for the diode mounting screws. Either
O.OOS-inch indium foil or vacuum grease was used on contact surfaces. A third type of
installation, initially referred to as "off-stream," was conceived to provide the
advantages of surface mounting while improving the heat sinking to the gas stream
and reducing the effect of radiation. Four such arrangements were tested: one made
completely of copper (Cu), one with stainless steel (SS) sidewalls and a Cu bottom,
one made completely of SS, and a fourth with a thin walled SS well. The fourth
arrangement was ultimately selected as the Quick Response Thermometer (QRT).
The electrical leads were externally heat sunk on the process pipe by wrapping them
around copper posts screwed to the pipe. A copper radiation shield of 71A-inch
diameter and lIlA-inch high was supported by a Micarta collar around the upper part
of the sample pipe. A 30 W heater was also mounted on the pipe, upstream of the
diodes and radiation shield. The whole apparatus was installed in a vacuum tank
lined with 24 layers of MU superinsulation. Uquid or gaseous helium was brought in
through a bayonet in the top flange and entered the top end of the sample pipe. The
helium exited the bottom of the pipe and was returned to the recovery system. All
the diodes were connected with a four-lead scheme to a CAMAC system which also
provided the 10 pA exciting current. The digitized CAMAC data was stored and
displayed with a PC-based data acquisition system with resolution ± 10 mK.
Summary of Test Results
Reproducibility for diode test runs was limited to 20 mK. Results were
evaluated in terms of comparisons between the various diodes on the sample pipe
with respect to responses to such factors as shield temperature, applied heat, and
diode mounting scheme.
The performance of the two BO diodes mounted directly in the stream of the
process flow was used as the base reference for test results.
Estimate of Measurement Errors
Even when corrections for the offsets at liquid helium temperature were
applied to the indicated diode temperatures, there was still some residual scatter with
gaseous helium as coolant. With the shield at 77K, the difference between the
highest and lowest indicated temperatures was as much as 150 mK. The scatter in
the values can be attributed to several sources:
620
L~
I
BC? 5: t--
SE'J SOP I
HEA Sl~ ~
- L 294 DIA 79
:;E SOP SE SOP SENSOR

Dl-~ 70- CU 01- 4 70- ED OT-~7 0 - C D
Figure 1. Silicon Diode Sensors used for test. Source: Lake Shore
Cryotronics, Inc., Westerville, Ohio, USA, all dimensions in millimeters.
N,
'N
ALUMINUM VAC.UUfV.
~'2%·O .D.
< 1..0' HIGH
'f,." T HK
_ __ ~r
Figure 2. Test Equipment Configuration.
621
1) Actual changes in the gas or wall temperature over the length of the sample
pipe.
2) Deviations from Lake Shore's Standard Curve #10 which are not compensated
by the LHe temperature corrections.
3) Errors due to imperfect heat transfer between the diode and the gas stream.
Heat loads come from resistive heat generated by the diode, external
radiation, or lead heat loads.
4) Instrumentation errors in the CAMAC system used to read the diode voltages.
Gas and Wall Temperature Variations.<t) The gas or pipe wall temperature
can vary either because of radiation heat input to the pipe or due to transient
conditions during cooldown. The gas flows in at the top and out the bottom of the
pipe. Both of these processes would make the lowest sensors (the in-streams) the
warmest, since the gas at the bottom would pick up all the integrated radiation or
pipe wall heat. Analysis indicates that a period of a few minutes should easily suffice
for reaching thermal equilibrium once the coolant gas stream temperature is
stabilized.
Deviations from the Calibration Curve. All sensors were ordered with a
maximum deviation of O.IK from Lake Shore's Standard Curve #10 in the
temperature range between 4K and 6K. The additive corrections determined near
4.3K help reduce this scatter, but the corrections are strictly valid only at this
temperature. When gaseous helium at about 5K is used as coolant, slight differences
in the calibration curve slopes evidently increase the variation range to about 70 mK.
Heat Transfer Limitations. Heat generated in the diode or conducted in by its

leads is transferred to the helium through the thermal resistances of the diode's own
electrical insulation, the interface between the diode and sample pipe wall, the wall
itself, and the gas film heat transfer coefficient at the wall or diode surface. Lead
heat loads should be small because external heat sinks on the sample pipe were
installed for all diodes. The DT-470 diode generates about 20 IlW of self-heating
power at 4.2K. The Lake Shore source(2) allows us to use 0.15 mK/IlWas heat sink
capacity for DT-470. For 20 IlW, this would add 3 mK, considering that the diodes
are installed with a thin layer of grease or indium. If the diode heat load is spread
over an area of 0.25 cm2, conduction through the pipe wall (about 2 mm thickness)
gives about a 10 mK temperature rise. Finally', the gas film heat transfer coefficient
gives another 10 mK. Since the gas film and electrical insulator thermal resistances
apply equally to an in-stream sensor, we conclude that the self-heating effect would
cause a diode mounted in vacuum on the pipe surface to read roughly 10 mK higher
than an in-stream one.
Instrumentation Errors. The total CAMAC accuracy at 4K is ± 25 mK. The

smallest temperature step which could be displayed by the digitizer was 10 mK.
Test Conclusions
The effects of diode mounting scheme, lead heat sinking, gas flow rate, and
radiation shield temperature have been assessed. When the sample pipe was filled
with liquid helium, an independent measurement of the temperature could be
622
12
TEMPERATURE SENSOR TE-4621
11 10 GIS @ 4.31 BAR - 1.682" ID PIPE
Tr ~ 80 K - 10 LAYERS MLI
Tss = 10 K· AT 2 sec, STEP TO 3.6 K in .005 sec
10
THERM MODEL OSTSEN6
9
:,,::
w 8
0:
:::J
I- 7
«0: @ t ~ 18 sec
w
!L 6 T-ORT = 3.603 K
:2 HT SINK
T-ST = 3.614 K
w
I- ST T-HS ~ 3.977 K
5
4
He GAS
3
2
0 2 4 6 8 10 12 14 16 18 20
TIME - seconds
Figure 3. PSI Temperature Sensor Response.
12
TEMPERATURE SENSOR TE-4621
11 100 GIS @4.31 BAR - 1.682" ID PIPE
Tr = 80 K - 10 LAYERS MLI
Tss = 10 K - AT 2 sec, STEP TO 3.6 K in .005 sec
10
HOLD FOR 10 sec - STEP TO 10K in .005 sec
THERM MODEL OSTSEN5
9
:,,::
w 8
0:
:::J
I- 7
« @ t = 12 sec
0:
w
!L 6 T·ORT = 3.603 K
:2 HT SINK
T·ST = 3.608 K
w
I- T-HS = 3.911 K
5
4
He GAS
3
2
0 2 4 6 8 10 12 14 16 18 20
TIME - seconds
Figure 4. PSI Temperature Sensor Response.
623
derived from the vapor pressure. The offsets of the diodes from this temperature
varied from 10 mK to 100 mK. showing that the diodes were within the specified
deviations from Lake Shore's Standard Curve #10. With gaseous helium cooling near
5K, and relatively low flow, the uncorrected indications were spread over a band
180 mK wide. Applying the offsets determined from the LHe readings decreased this
to less than 80 mK. An improved thermowell, the QRT concept, has been developed.
Keeping the parts as thin and light as possible gave thermal response better to that of
diodes directly immersed in the process stream. Since the diode is on the vacuum
side of the well, it can be replaced without cutting into the pipe, and no cold (4K)
helium tight electrical feedthroughs are necessary. For temperatures in the
neighborhood of 5K, the estimated error due to self-heating for this arrangement is
about + 10 mK compared to a diode immersed in the process stream. Surrounding
the sample pipe with superinsulation gave a pronounced decrease in the response of
the surface diodes to changes in the radiation shield temperature. There did not
appear to be any difference in performance between the type BO and type CU
diodes.
ST SURFACE THERMOMETER
The surface mounted silicon diode is a good solution for measuring

temperature in most locations of the helium plant cryogenic systems. Typically,
process temperatures are nearly constant and design helium flow velocity is at
average 4 m/s. The PSI type ST thermometer will provide inaccuracy due to
mounting not greater than 20 mK and thermal time constant of 2.2 seconds. The
following design details are implemented on this type of thermometer:
The four lead phosphor bronze DT-470 diode follows the Lake Shore
Standard Curve # 10 with accuracy ± 100 mK in 2 to 6K range. A copper pad
(aluminium pad for aluminium surfaces) is soft soldered to the measured surface for
diode mounting. The contact area of the pad is moistened with APIEZON N grease.
A special lead heat sink design was implemented to achieve 1 mK temperature rise
for 200 jlW leads heat leak. Aluminized mylar superinsulation (MLI) is used on the
diode assemblies; 10 layers for average radiation source below 80K and 30 layers
when above 80K.
QRT, QUICK RESPONSE THERMOMETER
The PSI type QRT thermometer will be used at locations where shorter
thermal response time is important for control purposes and for refrigerator
performance calculations. The QRT sensor provides inaccuracy due to mounting not
greater than 10 mK and thermal time constant of 1.2 seconds. Computer model(3,4)
response time for type ST and QRT are given on Figures 3 and 4. Figure 5 shows real
data taken during the diode test. The QRT diode is mounted in a light-weight well.
The bottom of the well serves as the mounting pad for the diode sensor. The top has
a thicker collar to allow for welding to the process pipe. The diode, lead heat sink
and MLI insulation are identical to that used on the ST thermometer.
624
"LOT 30A, COOLANT: GI-'E, LN2 SI-'T~77,
260 --,,-------~----------------~---,
240
220
200
180
~60
~ 140
D-
2' 120
w
f-
iOO
80
60
40 ~
l
20
0
2 :3 4 5 6 7 8 10
EXPERIMENT ELAPSED "ME, MINUTES

-- IN~STREA~ DIODE A SURFACE MOUNT DIODE
£:, SS WELL W.CU BOTTOM x SS LIGHT WELL
Figure 5. Diode Test #30 for SSCL, May 10, 1991.
ACKNOWLEDGEMENTS
Special thanks for contributions to the diode test:
Rao Ganni, SSCL Cryogenic Division, for his inspirational comments;

Willem Keyer, Process Systems International, Inc., for engineering assistance;
Hank Valcour (the diode test engineer), Process Systems International, Inc., for his
time and patience; and John Larkin, Highland Technology, Inc., 320 Judah Street,
San Francisco, CA 94122, for supplying us with CAMAC diode test instrumentation.
REFERENCES
1. S.W. Schwenterly, "Summary and Analysis of PSI Cryogenic Thermometer

Diode Test," Oak Ridge National Laboratory, Fusion Energy Division,
Memorandum, December 19, 1991.
2. B.C. Dodrill, J.K. Krause, P.R. Swinehart, V. Wang, "Performance

Characteristics of Silicon Diode Cryogenic Temperature Sensors," Lake Shore
Cryotronics, Inc., Westerville, OH.
625
3. Gordon N. Ellison, ''Thermal Computations for Electronic Equipment," Van
Nostrand Reinhold Co., 1st Ed., New York, 1984.
4. "Thermal Radiation Analysis System, TRASYS II," Martin Marietta, NASA

Contract NAS9-15832 MCR-73-105 (Revision 5), June 1983.
626
HIGH EFFICIENCY, HIGH RELIABILITY, IMPROVED
OIL INJECTED HELIUM SCREW COMPRESSOR
Moshe Y. Dreksler (1) and Takayuki Kishi (2)

(1) MYCOM Corporation
19475 Gramercf Place
Torrance, Cal1fornia 90501
(2) MYCOM Advanced Technology Lab
Mayekawa Mfg. Co. Ltd.
Moriya, Ibaraki, Japan
INTRODUCTION
Fundamental investigation of the factors affecting oil
injected Helium screw compressors performance was initiated
about 5 years ago after expiration of the basic screw pa-
tents issued late 1960's, resultin~ in developments which
provide significant improved effic1ency and reliability.
These theoretical and empirical efforts were demonstrated
on two-stage Helium Compression test-system showing shaft-
power reduction with improved Isothermal and Volumetric
efficiencies, as function of modified fixed discharge ports
with corresponding reduced hydraulic effects. Includin~
change of lubricant type and flow-rate also increased t1P-
speed with modified rotors profile, and optimized volume-
ratio of first to second stage compressors. Further inves-
tigation of these governin~ variable factors is expected to
enhance present understand1ng of their relative affects on
performance, also facilitate refinement of the preliminary
formulae developed for predictability of each variable,
accordingly provide a significant advancement in state of
the art.
During the 1970's it has been demonstrated {1},{2}, that
the oil injected helium screw compressors are more reliable
than reciprocating compressors and are suitable for high
pressure ratio in single-stage mode, and are best choice in
two-stage or compound configuration for large Helium refri-
geration/liquefication systems, with expected oil purity
concentrations of less than 1.0 ppm by weight based on 1970
existing technology. During the 1980's performance confir-
mations of the oil injected Helium compound screw compres-
sors at Fermi National Accelerator Laboratory (FNAL) have
shown {3},{4}, that the 2016-LL/MS-C (400 HP) compound model
provides 56.8g/s to 59.7 gls at 17.PSIA to 18PSIA suction,
310.PSIA discharge with 40% to 42% Isothermal efficiency.
Also that oil carry-over is less than 0.1 ~pm by weight be-
fore entering the charcoal and molecular S1eve vessels. It

was shown {5}, that these (40) screw compressors exceeded 1.4
million total combined hours of sustained operation without
unpredictable shut-down of the accelerator. Of course numer-
ous Helium screw compressors have been operating interna-
tionally on super-conductive systems as well as process
applications with similar reliability and in some cases
higher Isothermal efficiency, as function of size and bal-
anced displacements of the two stages for optimized power
requirement.
The technolog¥ developed for helium compression s¥s-
terns has been applled successfully on low molecular welght
gas compression duty, including low temperature conventional
refrigeration {6}. Of course, recent experiences with Helium
two-stage system improved performance test results {7} are ex-
pected to introduce new innovations adaptable to various ap-
plications with related improved performance and reliabili-
ty. These innovations include means to reduce hydraulic ef-
fects which is accomplished by discharge ports modification,
and reduced internal gas leakages which is accomplished by
use of modified rotors profiles, aided b¥ improved oil man-
agement techniques and electronic operatlng controls.
FEATURES AND CONFIGURATIONS

Screw units configurations
There are basic schemes used in staging Helium screw
compressors based on performance, reliability and economic
considerations. The typical flow diagram shown in Figure-1
represents separate units, each with the required oil manage-
ment components and controls. Providing for parallel hook-up
of mUlti-units with common interconnection of the two-staged
operation, including intermediate side-load connection, and
option of intermediate heat-exchanger. The oil separator,
whether horizontal or vertical includes first stage coales-
cer, and the common discharge of the second stage compressors
is connected to aftercooler and mUlti-stage vessels for oil
and moisture clean-up {4}.
<>- SUCTION
-::> SIDE
~""'-.l lOAD
12ND ST:G
CONT'l ,..........-~
#2
O il COOLER
Fig. 1. Typical Two Stage Sep- Fig. 2. Typical Two Stage Unit
arate Screw Units System Flow With Single Oil Management
Diagram. System Flow Diagram.
628
The system shown in Figure-2 utilizes sin9le oil cir-
cuit without intercooler option, because the flrst stage
compressor(s) discharge the helium gas-oil mixture into suc-
tion of high stage, while all other features are similar to
the separate units packaged system. This series staged arr-
angement uses a smaller floor space but requires proper mon-
itor of the oil circuit in accordance with recently esta-
blished improvements.
SUCTION
r.
-<> - -."..
SUCTION
DISCHARGE
LOAD
COIolPOUNO SCREW
OVEA'III£W
Fig. 3. Typical Compound Sys- Fig. 4. Typical Compound Screw.

tern Flow Diagram.
The system shown in Fi9ure-3 is identical in operation

to the configuration shown ln figure-2, except single com-
pound compressor and single motor are utilized. Providing
for simple smaller packaged system, with distinct economic
advantage.
Screw Compressor Features
The two-stage compound screw compressor Figure-4 in-
cludes two single-stage screws internally coupled, with sin-
gle shaft-seal and single drive shaft. Each stage is of
conventional 4-lobes male rotor and 6-lobes female rotor
assembly, supported by journal bearings and positioned by
ball type thrust bearings. Each stage has fixed discharge
portings and slide-valve mechanism for capacit¥ control.
Also lube oil and injection oil connections WhlCh are sup-
plied by pump and controlled to assure proper distribution
of cooled and filtered oil.
The standard single-stage and compound models with
their displacements are shown in Table 1. These values and
required oil flow can be estimated by the equations pro-
posed by Scheel {1}.
vth=D*L*U*X (1)
G=O.03*vth+10 (2)
Where Vth-Theoratical displacement in cfm; D=Rotors diame-

ter, inch; L=Rotors length, inch; U=male rotor tip speed, fps
629
Table 1. Standard Single Stage and Compound Screw Compressor
Models: (a) Compound Models, (b) Single Stage.
COMPOUND MODELS SINGLE STAGE
a b
MODEL SRM' ROTOR SIZE DISPLACEMENT
FIRST AND DISPLACEMENT
SECOND STAGE IN CFM
CIA.XLI o [MMJ L(MMJ AT 3550 RPM, CFM
MODELS @3550RPM
125 S 140 133
1ST 160L 438 ~ 127.5
2ND 1255 139 210 208
125 L
1ST 200S 571
2ND 1605 292 1605 180 292
1ST 200M 712
I----
2ND 1605 292 160M 163.2 225 365
1ST 200L 853 I----
2ND 1605 292 160 L 27D "8
1ST 2505 1,110
2ND 200S 571 200 S 225 571
1ST 250M 1,390 f----

2ND 200S 571 200 M 204.0 281 712
1ST 250l 1,670 I---

200 L 337 8"
2ND 200S 571
1ST 320L 3,340
250 S 280 1,110
2ND 2505 1,110
I---
EACH COMPRESSOR IS AVAILABLE 250 M 255.0 352 1,390
WITH EITHER 'L'·'M'·'H' PORTINGS' -
250 L 420 1,670
MODEL ROTOR LENGTH TO
DIAMETER RATIO: S=1.1; M=1.32; L=1,65 3208 355 2,240
-
320 M 321.3 443 2,780
-
320 L 530 3,340
limited to 0.25 mach; X=Cross sectional factor 0.0645; G=oil

flow, gpm varying with operation.
Two dimensional presentation of single stage screw
Fi9ure-5 provides simplified illustration of the re1ation-
ShlP between built-in volume-ratio and discharge partings.
Similarly Fi9ure-6 represents part load operation and ty-
pical variatl0ns in capacity and power at various slide
valve positions, as function of compression ratio at fixed
outlet pressure. To accomplish part-load operation the
slide valve is moved into discharge manifold providing for
internal bypass.
SUC'TIOt~
Vr, . on"
AXIAL POFtT UtnOAO

lOAO ~
--~LT~~--~----~ 40~
>-
>-
Vi~ 5 a YI ~2, 6 CR i ~V'''' u
~
- -+----120 ~
RADIAL
POATS "
20 010 50 60 80 100
!}:{) POAT ~ P'ORT SL.IOE VAL,VE POSITION ....
Fig. 5. Built-in Volume Ratio Fig. 6. Part Load Operation

(Vi) Relationship with Internal Single Stage Screw.
Compression Ratio (CRi).
630
CHARACTERISTICS AND PERFORMANCE
Screw characteristics
The screw compressor volumetric efficiency is func-
tion of internal gas leakages, operating pressures, type
and quantity of lubricant, rotors tip speed, rotors pro-
file and assembly clearances, and gas being compressed.
Stipulating that discharge ports match operating conditions
with minimal hydraulic effect, namely equatin9 the inter-
nal compression ratio with manifolds compreSS10n ratio, as
expressed in the following equation.
K P2i P2
CRi=vi pH ~ CR PI (3)
where CRi=internal compression ratio; vi=built-in volume

ratio; P2i=internal discharge pressure before portings are
exposed; PIi=PI=suction pressure; CR=manifolds compres-
sion ratio; P2=discharge pressure.
Internal leakages illustrated in Figure-7, show the
multi-paths of gas leakage from high pressure to low ~res
sure zones. Of course these leakages effect volumetr1c ef-
ficiency as function of the variables depicted in Figure-S.
15 r ----,----r---,
P1 SUCTION VE:%
CR oP, P,
DRIVE
SHAFT PO=P, . P, PSIO
13r--+-
/
MODEL AT 3550 RPM
WITH OPTIMUM V,
C::
PRESSURE
t=!~CY TRANS OIJCEA ~ 11
P~H . APPAO)CIMAtIO:t OF 'JE
FOR
~ AT CA:l:''& PO. I00
a: 340l HAS VE =9S'.
DISCHARGE P2 5 91--- -H" - 4 HiD!. vE. 87-.
AT CA lliS to. PO=200
in l:!'DL. MAS 'JE ; 9D".
<Il
W 16(]l VE=83',
g: 7 t -- -'li--
:l1
o
u
5r--iV~tr.·
320l
...J
UJ
SUCTION 250l Cl
u 0
END a:
....
:l1
w 200l ;:
:l1 UJ
:::> a:
...J u
0 lS0l <Il
>
Fig. 7. Typical Internal Gas Fig. 8. Full Load Volumetric

Leakage Paths -1-2-3. Efficiency 'VE'vs'CR' and 'PO'.
Showing that for given operating conditions the larger mo-

dels with higher tip speed have higher volumetric efficien-
cy than smaller models. Similarly the increased speed of
small models will improve volumetric efficiency . Kishi {7}
proposes that:
631
Cb Na
VEb=l-(l-VEa)Ca * Nb (4)
where change of speed N from Condition-a to Condition-b,

effects VE in conjunction with sonic velocity-C, which de-
pends on gas to lubricant mass flow rates. Namely, increase
of oil flow rate will retard internal 9as leakages due to
the increased-C thus improve VE signif1cantl¥.- However, the
power penalty due to increased oil flow requ1res more inves-
tigation, due to the obvious hydraulic effect at the dis-
charge zone. these hydraulic effects occur due to trapping
of oil between the rotors as shown in Figure-9. There is a
significant hydraulic hammer effect which occurs at the
pinch-point shown in step 3, perfect close condition. This
hydraulic impact was measured by pressure transducer present-
ed in Figure-7, and results shown in Figure-l0 for helium
200L-L screw model.
,
Mil l e Rotor Female Rator
<\"0
3.5
3..0
- Cas i Cl.i
!'
Steo .1 ~
2..5
w
a:: 2.,0
=>
'"
'"a:
w
Q.
Stoo.2
~. n-' T I ME ( .... e)
"~
' ~' u~ s uctio n pressurc
discharge P['cssure 1.7?

O. ~5 IMP,,)
(lWa )
I . ~), rV con vent iona l D~cl1inc 3000 (rpR)
.....>0, I ·,
~rfect close cond i t ion c.pcrinC HLal lachi ne 1000(rp~)
Fig. 9. Hydraulics Effect. Fig. 10. Oil Compression

Typical Phenomenon.
The trapped oil hydraulic effect occurs 4 times per

revolution as function of male rotor 4 lobes, or 240 impacts
per second. This trapped oil will be relieved to suction
thru the cleanrance between the rotor-ends and discharge
casing, also blow-hole and radial clearances. Thus thehy-
draulic impact forces imposed on the rotors, will transmit
the axial component to the thrust bearings, resulting in
unpredictable bearing races and balls fatigue failures.
Obviously the reduced magnitude of these impacts by modify-
ing discharge porting to allow oil relief thru the discharge
ports to discharge manifold, as illustrated in Figure 10,
will contribute to shaft-power reduction and increased thrust
bearing life with significant improvement in compressor reli-
ability. These improvements have been applied on some helium
and hydrogen screw compressor systems installations.
632
Screw Performance
The improved power demands for two-stage helium test
screws with modified discharge portings operating at vari-
ous speeds are shown in Figure-II. These data apply to the
helium 200L screw model with 3.6(M)vi portings on low-stage
duty and helium 125L screw model with 3.0 vi on high-stage.
120
~ 120 l'
""
~110 ~ 11 0 fsTANDEROI
a:
~ 90
0
100
~1~~ PO;/~
"- 80
r-
:;: 70
~ SO ~OIFIED
Ie
:';: 70 PORT
Ie
UJ 60 60
PORT UJ
50 50
3.03.54.04.5 3.03.54.04.5
X1cY (rPm) X1cY (rPm)
NUMBER OF REVOLUTIONS NUMBER OF REVOLUTIONS
HIGH STAGE MACHINE LOW STAGE MACHINE
suction pressure o. 1 [MPa]

internal pressure 0.45 [MPa]
dj scharge pressure 1. 72 [MPa]
Fig. 11. Discharge Ports Modification Effects on Power.
Performance of helium compound screw models including

modified features are shown in Table-2, indicating for mo-
del 2016-C Isothermal efficiency of about 10% higher than
measured by Peterson {3} for conventional configuration with-
out intermediate or side load. At identical conditions the
use of smaller hi9h-stage with two-separate stages in series,
provides 5% additl0nal lncrease in efficiency as indicated
for model 200L + 125L in Table-3. Both tables have ± 3% var-
iances in power and capacity, and isothermal efficiency shown
does not include oil pump power.
Table 2. Compound Helium Screw Compressor Unit Performance.

HELIUM HELIUM HELIUM
Inlet Revol ut i on Inlet Discharge Shaft HELIUM Volumetric Isothermal Oi I Pump
MODEL Temp. Speed Pressure Pressure Power FIOIW Eft i ci ency Eff iciency Power
( K) (rpm) (atm a) (atm a) (HP) (g/s) (I) (I) (HP)
1612C 298. 3550. 1.0 21.0 144. 29.8 83.3 52.9 1.8
2016C 298. 3550. 1.0 21. 0 281. 59.5 90.6 54.3 3.6
2520C 29B. 3550. 1.0 21. 0 545. 11B. 92.4 55.B 7.5
3225C 29B. 3550. 1.0 21. 0 1076. 241. 94.1 57.5 9.5
The increse in screw size further increases the Isother-

mal efficiency, but due to required intermediate flow add-
ition in most cases, computer performance program should be
used for optimized screw selections.
633
Table 3. Separate Single Stage Helium Screw Units Performance.
HELIUM HELIUM HELIUM
Inlet Revol ution Inlet Oi scharge Shaft HELIUM Volumetric Isothermal Oi I Pump
MODEL Temp. Speed Pressure Pressure Power Flo", Efficiency Efficiency Power
( K) (rPII) (atm a) (atm a) (HP) (gls) (X) (X) (HP)
2005+125S 298. 3550. 1.0 21.0 85.9+77.8 38.4 87.5 60.2 2.7+2.0
200L+125L 298. 3550. 1.0 21. 0 130.+117. 57.5 87.5 59.8 2.7+2.7
250S+160S 298. 3550. 1.0 21. 0 157.+160. 76.6 89.9 61. 9 4.8+2.7
250l+160L 298. 3550. 1.0 21. 0 239.+241. 115. 89.9 61. 6 4.8+4.8
320S+20OS 298. 3550. 1.0 21.0 323.+314. 158. 92.0 63.6 9.5+6.8
320L+200L 298.. 3550. 1.0 21. 0 484.+471- 236. 92.0 63.3 9.5+9.5
SUMMARY AND CONCLUSIONS

The adaptation of recent developments for enhancement
of the two stage oil injected helium screw compressors ef-
ficiency and reliability relate to reducing power demand
by modifying discharge ports and sizin~ each stage for to-
tal lowest energy input. Also increas1ng volumetric effi-
ciency by minimizing internal ~as leakages accomplished by
the use of proper lube and by 1ncrease of low stage speed.
Magnitude of expected improvements in performance attribu-
ted to each key factor is estimated at ·11% lower BHP for
reduction of hydraulic effects; 6% lower BHP for best lo-hi
displacement ratio; 3% higher VE for optimization of lube
supply; and 3% higher VE for increasing low stage tip speed.
Based on present technology isothermal efficiency of 60% is
readily attainable. The continued investigations of these
matters in the lab and field should provide new innovations,
some proprietary, for new helium screw systems design and
applications with isothermal efficiency approaching 65%.
REFERENCES
{1} L. F. Scheel and M. Y. Dreksler, "A technique for rating
rotary compressors", AS ME 76-Pet-104 (1976).
{2} W. M. Toscano and L. C. Hoagland, "The thermodynamic per-
formance of oil flooded helium screw compressors," ASME
76-WA/PID-15 (1977).
{3 } T. Peterson, M. Hentges, A. James and M. Urso, "Measure-
ments of Satellite refrigerator compressor power con-
sumption and efficiency," Fermilab Technical publication,
TM-1276 (1984).
{4 } J. A. Satti and P. Andrews, "Satellite refrigerator com-
pressors with the oil and moisture removal system," Ad-
vances in Cryogenic Engineering, Plenum Press, Volume
29, p. 453 (1984).
{5} C. Pallaver, "The design, fabrication, operation and main-
tenance of (40) 400.HP-600Scfm-5 years operation re-
port, Fermi National Accelerator Laboratory," Technical
publication, TM-1561 (1988).
{6} M. Y. Dreksler, Compound screw compressors low tempera-
ture applications, International Institute of Ammonia Re-
frigeration (IIAR) (1991).
{7} T. Kishi, N. Ino, M. Kudo and H. Iisaka, High effici-
ency, high reliability oil injection type helium screw,
Cryogenic Engineering Conference (CEC), (1991).
634
THERMAL OPTIMIZATION OF THE HELIUM-
COOLED POWER LEADS FOR THE SSC
J. A. Demko, W. E. Schiesser, R. Carcagno,

M. McAshan, and R. McConeghy
Superconducting Super Collider Laboratory·

Dallas, Texas 75237-3997
INTRODUCTION
The optimum thermal design of the power leads for the Superconducting Super Collider (SSC)
will minimize the amount of Carnot work (which is a combination of refrigeration and liquefaction
work) required. This optimization can be accomplished by the judicious selection of lead length and
diameter. Even though an optimum set of dimensions is found, the final design must satisfy other
physical constraints such as maximum allowable heat leak and helium vapor mass flow rate. A set
of corresponding lengths and diameters has been determined that meets these requirements for the
helium vapor-cooled, spiral-fin power lead design of the SSC.
Early efforts by McFee! and Mallon 2 investigated optimizing power leads for cryogenic applica,-
tions with no convection cooling. Later designs utilized the boiled-off helium vapor to cool the lead.
One notable design for currents up to several thousand amps is presented by Efferson 3 based on a se-
ries of recommendations discussed by Deiness. 4 Buyanov 5 presents many theoretical models and design
formulae but does not demonstrate an approach to thermally optimizing the design of a vapor-cooled
lead.
A method for optimizing superconducting magnet current leads is described by Maehata6 et al.
The approach assumes that the helium boil-off caused by heat conduction along the power lead into
the low-temperature helium is used to cool the lead. The optimum solution is found when the heat
flow at the cold end is minimized.
In this study, a detailed numerical thermal model of a power lead design for the SSC has been
developed. It was adapted from the dynamic model developed by Schiesser. 7 This model was used to
determine the optimum dimensions that minimize the Carnot refrigeration and liquefaction work due
to the leads. Since the SSC leads will be cooled by supercritical helium, the flow of vapor is regulated
by a control valve. These leads include a superconducting portion at the cold end. All of the material
properties in the model are functions of temperature, and for the helium are functions of pressure and
temperature. No pressure drop calculations were performed as part of this analysis. The diameter that
minimizes the Carnot work was determined for four different lengths at a design current of 6600 amps.
* Operated by Universities Research Association for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.
Supercollider 4, Edited by J. Nome

POWER LEAD DESCRIPTION
The lead structure is made up of two sections as depicted in Figure 1. The power lead core
diameter was varied between 1.2 and 4.0 cm. There are approximately 3.15 fins/cm. The fin thickness
and spacing are 0.159 cm. The depth of the spiral fins was held constant at 1.18 cm, providing
a constant cross-sectional area of 0.188 cm 2 for the helium flow. An Nb-Ti superconductor cable
surrounded by a copper casing is included in the first 30 cm of lead length. The cable has a rectangular
cross section given as 0.813 cm x 0.30fi cm, resulting in a cross-sectional area for the Nb-Ti of 0.248 cm 2 .
An assumed copper-to-superconductor ratio for the cable of 1.67 was used in this model. The remainder
of the lead core is solid copper. The total length of the lead was varied over 60 cm, 80 cm, 100 cm,
and 120 cm-not including the end hardware used to connect to the power supply or to the cryogenic
device. The whole lead is convective-cooled by liquid helium that enters at 4K at a pressure of 4 atm.
The cooling passages spiral around the outside of the core, removing heat from the lead.
Figure 1. Spiral-Fin, Helium-Cooled Power Lead Design.
MODELING EQUATIONS
The power lead transient thermal analysis was modeled by applying an energy balance to the
solid conductor and the helium. These equations are for the solid:
(1 )
and for the helium:
o (Pv€v dv )= .m-{)
n
{)hv
ds + hconvP(Tv -l~)dx. (2)
ot oS
636
In these equations, Cc and r" represent the internal energy of the conductor and the helium vapor,
respectively, Pc and pv refer to the density of the conductor and helium vapor, hv is the helium vapor
enthalpy, h eonv is the convective heat transfer coefficient, P is the convect.ion surface area per unit length
of the lead, m is the helium mass flowrate, and Tv and Tc are the helium and conductor temperatures,
respectively. The quantities dv, dx, and ds are the differential conductor volume, differential distance
measured along the axis of the lead, and the differential distance measured along the flow path of the
helium, respectively, for which the energy balance is written. The joule heating is determined from
the following relation:
Q. _ Pelee[2 dx
(3)
J - AeJec 1
where the quantities Pelee and Aelec refer to the electrical resistivity of the lead and the cross-sectional
area for electrical conduction. The first term of the right-hand side of the helium energy equation can
be written in terms of x using the relation:
ds = (1/(7r x pitch x D)) dx. (4)
This formulation can be simplified further. It was determined that the heat capacity of the helium
was very small compared to the solid conductor, so it may be neglected. By applying an upwind
finite difference technique to the enthalpy derivative, the helium energy equation becomes an algebraic
expression for determining the helium liquid temperature.
The heat transfer coefficient is determined from a Dittus-Boelter relation when the flow is tur-
bulent (Re > 2000) given in Wilson," and for laminar flow the Nusselt number based on the aspect
ratio of the channel is a constant from Kays.9 The heat transfer coefficient is given by the following
for these two flow regimes:
N Ulammar == hconvDhyd = 6.6 and (5)

kv
0.716
, _ hwnvDhyd _ 0.8 0.4 ( Tv )
1\ Uturbulent - - 0.0259 x Re x Pr x -
kv Tc
In the model, the convection area included two sides of the channel and the channel base. The
channel sides were assumed to be isothermal in the radial direction. The resulting algebraic expression
becomes:
m
. (hi -~x
hi-I) == h con v P (Tc - Tv )i· (6)
The enthalpy term and the convective heat transfer coefficient are evaluated using the conductor
temperatures at a given time. This equation can then be used to solve for the new values of the
helium vapor temperature along the lead since hi =
hi(1~). This approximation is acceptable due
to the small difference between the helium and conductor temperatures (usually less than 0.5%), and
because the thermal capacitance of the helium is very small in comparison to the lead. Thus the helium
temperature changes almost instantaneously with a change in the lead temperature.
The mass flowrate of the helium was determined using a proportional-integral (PI) control algo-
rithm which adjusted the mass flowrate of the helium to maintain the temperature at the top of the
lead at a set point of 300K. The equation which describes this is:
(7)
The second term represents the proportional controller with a gain of Kg =

0.1, and the third term is
the integral action with a time constant T =
500 sec. The gain and integral time were set at values for
which the dynamic model ran smoothly and efficiently, since only the steady state results were used
in the optimization study. This additional ordinary differential equation was integrated as part of the
solution to provide the mass flow at each time step.
To solve the resulting system of partial differential equations, the numerical method of lines
approach was used as discussed in Schiesser. 1O In this method, the first term on the right-hand side
of Eq. (1), which represents the net conduction heat transfer along the lead, was cast into a finite
difference form given by:
637
.i.
ax
(kA aTe) = (kA).
e ax '+,
(Ti+1~x- Ti)
1 _(kA)..-, (Ti - Ti_l) .
1
~x
(8)
The temperature-dependent thermal conductivities at the i + 1/2 and i - 1/2 locations were computed
based on the geometric mean as recommended by Patankar .11 The expression for this is:
(9)
The lead conduction cross-sectional area is constant except for the prototype lead, which was used to
take experimental measurements and had a hollow section at the top. For this case, the area of the
hole is simply subtracted from the lead cross section. At the interface region, the proper area is used
for the i + 1 and i - I locations.
The material properties of the conductor were taken for OFHC copper with a residual resistivity
ratio (RRR) of 40. The properties of the Nb-Ti superconductor were based on a copper to supercon-
ductor ratio of 1.67. Material properties were determined from a database used at the SSC Lab,12
from Coliings,13 and from Reed and Clark.14
Special consideration is also required for the superconducting section in calculating the composite
thermal conductivity and the electrical current flow when the superconducting transition occurs in this
region. The thermal conductivity of the composite region was based on a simple area-weighted average
of the thermal conductivities.
When the Nb-Ti cable is in a superconducting state (i.e., T < Terit), it is assumed that the
electrical resistivity becomes that of the copper stabilizing the superconductor (based on a copper to
superconductor ratio of 1.67) and that all of the current flows through the superconducting cable. The
critical temperature was taken to be 1O.2K assuming no magnetic field. 14
At the cold end, a fixed temperature was specified based on measured values. The warm end
temperature was controlled to 300K (room temperature), so no heat transfer occurred from the external
environment. A convective boundary condition was specified at the warm end of the lead for the
comparisons with experimental data. The value of the overall heat transfer coefficient was determined
through calibration with the available experimental data. Assuming an environment temperature of
300K, the value of the overall convective heat transfer coefficient was 2.2 W /K. This value was relevant
only to the comparisons with experimental measurements since the warm end temperature was 253K.
For the initial temperature profile, a linear distribution was specified that ranged from the cold end
boundary condition to the desired set point temperature. The current was ramped up at a rate of
6 A/sec until the design current of 6600 A was reached.
Upon making the preceding algebraic substitutions for the spatial derivatives and applying the de-
scribed boundary conditions, a system of ordinary differential equations in time results. The LSODES
integrator was used to solve the resulting system of ordinary differential equations. To ensure that
steady state had been reached, the solution was run for a total of 5000 sec.
MODEL·VALIDATION
The model validation was based on comparisons with the experimental measurements recently
made at Fermilab for the existing prototype lead. The test data were taken on June 19, 1991, for a
current of 5007 A. The model used the experimentally measured helium flowrate and end temperatures
as inputs. Since the measured top end temperature was below ambient at 253K, it was necessary to
calibrate the heat leak into the top end by adjusting a convection coefficient in order to match the
measured top temperature at the measured helium flowrate.
The thermal conductivity and electrical resistivity of copper are sensitive to the RRR. Several
calibration runs in which the RRR was varied were made to obtain the best fit with the data. Pre-
dictions made using RRR = 40 yielded a good fit with the temperature data. Figure 2 shows the
temperature distributions along the lead that resulted from the calibration. It is evident from this
figure that there is good agreement between the predicted and measured temperatures, particularly
near the cold end.
Figure 3 contains comparisons of voltage drop measurements. In order to obtain the best fit with
the data, the superconducting transition point was set to match the measured location of X =27 cm.
The comparisons show reasonable agreement for X < 40 cm. The overall voltage drop across the lead
is drastically different (predicted voltage drop is 51 m V, measured 24 m V). Also included in Figure 3
are voltages that were calculated using the experimentally measured temperatures for RRR = 40.
These calculated voltages agree fairly well with the model's predictions. The difference is partly due
to electrical conduction by the fins, which would reduce the current density. Since there is only one
638
voltage measurement near the top, an accurate comparison of the voltage distribution is not possible.
Additional measurements are being planned to provide better resolution of the voltage distribution
along the top half of the lead.
OPTIMIZATION BASIS
The optimum dimensions of the lead were determined by searching for the minimum value of the
Carnot work parameter given by:
We arnot = 7000 x mHo

.
+
(TTOP - T COLD )
To
COLD
x QEND· (10)
The first term represents the ideal work required to liquefy the helium vapor used to keep the warm
end of the lead at room temperature, and it is proportional to the helium mass flowrate. The second
term is the ideal refrigeration work required due to heat conduction by the lead at the connection to
the cryogenic component (i.e., the cold end of the lead), and it is proportional to the cold end heat
conduction. This choice of an optimization parameter is very appropriate for a thermal optimization,
since it characterizes the overall system impacts of the power lead design. For the optimization study,
a current of 6600 A was assumed for the operating current. The warm end was controlled at 300K.
The cold end was assumed to remain at the helium bath temperature of 4K.
RESULTS AND DISCUSSION

Figure 4 shows the Carnot work variation with lead diameter for lead lengths of 60 cm, 80 cm,
100 cm, and 120 cm. In general, the Carnot work decreases as the diameter increases, until a minimum
is reached. As the diameter is increased past where the Carnot work minimum occurs, the Carnot
work increases sharply. The power leads tend to operate more efficiently at the smaller diameters since
the Carnot work is much lower than for the larger diameters. Figure 4 also shows that the optimum
power lead diameter increases with increasing total length. The higher values of RRR require smaller
diameter leads as well. There is a minimum diameter limit that occurs due to burn-out of the lead
if the current density is exceeded. Figure 5 shows the dependence of the mass flowrate of helium on
diameter and RRR for the different lengths. The SCDR15 budget of 0.4 g/sec is exceeded only for the
longer leads with small diameters. The flowrate of helium vapor decreases with increasing diameter
due to the increased conduction along the lead transferring the heat to the low temperature end and
the decreased joule heating along the top end of the lead. The heat transfer conducted along the lead
into the ayogenic space is shown in Figure 6. There is no heat conduction for the smaller diameters
because of the high flowrate of helium required to remove the heat dissipated at the top. If the
diameter were to be decreased further, a condition could be reached where the helium cannot remove
the heat generated through dissipation (joule heating) fast enough, and the lead would overheat. As
the lead diameter is increased, the heat conduction through the end of the lead increases, which in turn
increases the refrigeration load sharply. As the diameter is decreased, the liquefaction load increases
due to the increased mass flow of helium required to maintain the top of the lead at the set point
temperature of 300K. The minimum Carnot work corresponds to diameters of 1.5 cm, 1.7 cm, 1.9 cm,
and 2.0 cm for lengths of 60 cm, 80 cm, 100 cm, and 120 cm, respectively. In Figure 7 a summary
of the ac(eptable design space is shown. The SSC has a conduction heat load limit given as 7.9 W,
and a helium mass flow limit of 0.4 g/sec for each lead. 15 The heat transfer constraint requires that
the diameter be less than the limit shown in the figure. The mass flow constraint requires that the
diameters be greater than the lower mass flow constraint line. Between these constraints lies the locus
of diameters that minimize the Carnot work for a given lead length. Since too small a diameter would
provide no thermal transient stability, diameters less than 1 cm were not considered.

The thermal model of the spiral fin current lead design has been compared to experimental data
with reasonable agreement in temperature and less satisfactory agreement in voltage drop. Com-
parisons made with voltage drops computed using RRR = 40 and measured temperatures were in
agreement with the predicted results. The model has been used to determine the optimum length and
diameter for minimizing the ideal Carnot work needed to liquefy the helium vapor used as coolant and
to provide refrigeration used to offset the heat leak caused by the connection of the lead to a cryogenic
component. Also, the design must meet constraints on heat leak to the cold connection and helium
639
300 ,[~,~~,~'C"""TI~-'-'-'----r-'-'-'--'C"""TI--,--r--,.---r''-I-'-r-T~j
L
~
250r I Measurements
Conductor +
Helium <>
~
CD
200
Wv = 0.2537 g/s
::E.. RRR =40
CD
:; 150 Analysis Conductor ( - - )
I
E
Helium (- - - )
~ 100
50
X (cm)
Figure 2. Test Lead Temperature Distributions.
0.06
Predictions - - - -
0.05
Measured + + +
Calculated I
0.04 -
I
Wv = 0.2537 g/s
RRR=40
1!l
0 0.03
>
I +
I
0.02 -
I
0.01 l
0
0 60 80 100
X (cm)
Figure 3. Test Lead Voltage Distributions.
20000
x
i <>
! 15000 " /
~ T
~ /
T
, I I " , I , , , I I ,
1.5 2.0 2.5 3.0 3.5
Diameter (cm)
Figure 4. Power Leads Carnot Work.
640
0.6
Length
0.5,
60 em
80 em
0.4
'" "'''8 100em
~;t
0
0.3
~.
+-c -c
120 em 0
-= 0- - 't
't
E
.~
a;
:r: 0.2 C-
0.1 .- '11
~ 1
r ,
Oi. .... L_J ~~~~~.L.J~~L j'~ I -~)i( j_~
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Diameter (em)
Figure 5. Power Leads Helium Mass Flow.
200 I
I
I
Length t
L I 60 em +
150 U 80 em I
1
100 em
W 120em 0
I
'cil t
.!. = 40(--)
-'" 100- RRR = 100 (- - -)
'"
~
RRR
~ , /
iii
,
Q) /
:r: T
/
, /+ I
50'- --j
O~-.'''''''II::'I''''Ii=~io@''''''''''''',,="~1 ! ~_ J ___ ~~.

1.0 1.5 3.0 3.5 4.0
Diameter (em)
Figure 6. Power Leads Conduct.ion Heat. Transfer.
2.50
2.25 -
E
.9-
Q; 2.20 ::-
0;
E
'"
'0 1.75 L.
"0
~'"
"0
Q)
N 1.50
'1'"
a
0
1.25 f-
1.00
60
Figure 1. Opt.imized Power Lead Dimensions.
641
cooling mass flow. The conduction constraint is violated as the diameter is increased beyond some
limit for a given length. The mass flow can exceed its constraint if the diameter is too small for a
given length. The combination of this information can be used by designers to determine an optimum
power lead design for the SSC.
REFERENCES
1. R. McFee, "Optimum Input Leads for Cryogenic Apparatus," The Review of Scientific Instru-
ments, Vol. 30, No.2, February 1959.
2. R. G. Mallon, "Optimum Electrical Leads of Aluminum and Sodium for Cryogenic Apparatus,"
The Review of Scientific Instruments, Vol. 33, No.4, April 1962.
3. K. R. Efferson, "Helium Vapor Cooled Current Leads," The Review of Scientific Instruments,
Vol. 38, No. 12, December 1967.
4. S. Deiness, "The Production and Optimization of High Current Leads," Cryogenics, Vol. 5,
October 1965.
5. Yu. L. Buyanov, "Current Leads for Use in Cryogenic Devices. Principle of Design and Formulae
for Design Calculations," Cryogenics, Vol. 25, February 1985.
6. M. Maehata, K. Ishibashi, M. Wake, A. Katase, and M. Kobayashi, "Optimization Method for
Superconducting Magnet Current Leads," Cryogenics, Vol. 28, November 1988.
7. W. E. Schiesser, "A Dynamic Model of the SSC Power Leads," SSC Laboratory Informal Report,
August 1990.
8. M. N. Wilson, Superconducting Magnets, Clarendon Press, Oxford, 1982.
9. W. M. Kays, Convective Heat and Mass Transfer, McGraw-Hill, 1966.
10. W. E. Schiesser, The Numerical Method of Lines Integration of Partial Differential Equations,
Academic Press, San Diego, 1991.
11. S. V. Patankar, Numerical Heat Tran.~fer and Fluid Flow, Hemisphere Publishing Corporation,
1980.
12. A. Devred, SSC Laboratory, Internal Communication.
13. E. W. Collings, Applied Superconductivity, Metallurgy, and Physics of Titanium Alloys, Plenum
Press, 1986.
14. R. P. Reed, and A. F. Clark, Materials at Low Temperatures, American Society for Metals, 1983.
15. J. R. Sanford, and D. M. Matthews, Site-Specific Conceptual Design of the Superconducting Super
Collider, SSCL-SR-I056, July 1990.
642
DEVELOPMENT OF AN ULTRA LOW HEAT LEAK CRYOGENIC VALVE
FOR THE SSC TUNNEL SPOOL PIECES
E. J. Baker and C. Rista
SAES Pure Gas, Inc.

San Luis Obispo, California
G. E. McIntosh
Cryogenic Technical Services, Inc.

Boulder, Colorado
ABSTRACT
Satisfactory performance of the ASST spool pieces for the Super

Collider requires that the Cooldown/Warrnup, Quench, and Recooler LHe
bellows sealed cryogenic valves exhibit heat leak characteristics not
available with current industry technology. Typical solutions for controlling
heat leak of cryogenic valves are thwarted by the requirement to reduce
envelope size and by restrictions in the use of typical heat stationing
techniques. Going beyond the conventional approach, composite structures
have been devised for the valve bonnet and stem extensions resulting in
heat leaks nearly a factor of ten lower·than a conventional I 112 inch vacuwn
jacketed cryogenic valve. To our knowledge, the configuration of these
radiation resistant composite structures has not been used previously on
cryogenic valves.
Included in this paper is a background review of a typical vacuwn
jacketed valve, our design approach and heat leak analysis, design details,
and test results of prototype valves.
INTRODUCTION
The typical vacuum jacketed cryogenic valve consists of the valve

body (usually 300 series stainless steel), a stainless steel tubular bonnet
extension wrapped with MLI, conventional topworks, a solid or tubular
extended stem communicating with the valve plug, thermal radiation
reducing packing between the stem and bonnet extension, and a vacuum
jacket pipe surrounding the assembly. Because over 95% of the heat leak
is by solid conduction, design focus is on the thermal and mechanical
aspects of the bonnet extension, stem, and cold packing. The bonnet
extension tube has the greatest thermal impact because it is the largest
element and must be sized to withstand both the maximum valve operating
pressure and to resist stem closing forces without excessive displacement.
The stem must have sufficient column strength to close the valve
against full pressure with added force to make a tight seal and have enough
tensile strength to pull the plug up against pressure and seat sticking forces.
Packing between the stem and bonnet helps to center the stem and blocks
thermal radiation in this annular space.
In conventional vacuum jacketed valves, stem and bonnet extension
mechanics and economic sizing effectively govern heat leak. Reducing heat
leak is mainly a matter of increasing stern length and/or adding a liquid
nitrogen-cooled heat intercept (heat station). Increasing stem length is a
common inexpensive practice but is subject to space limitations and
mechanical stability if the topworks is heavy. Heat stationing is very
effective on the bonnet extension tube but is difficult to use on the
removable stern assembly and adds substantially to the valve and system
cost.
Typical vacuum jacketed valve features are illustrated by a
commercially availablel product. In the I 112 inch pipe size, this valve has
a 2 5/8 inch OD X 0.049 inch wall bonnet extension with a thermal length of
11 inches. The 112 inch diameter 17-4 PH stem is 12 inches long and the
solid Teflon packing is pinned to the stem over a length of 9 inches. Solid
conduction of this assembly between 4.2 and 300 K is calculated as follows:
Bonnet extension 2.708 W

Stern .1.229
Teflon packing 0.839
Total heat leak 4.776 W
This valve has an excellent record of reliable performance in industrial

and laboratory installations but, as shown, its heat leak is undesirably high
for continuous use in liquid helium systems. This paper will describe the
design, development, and testing of a new valve with similar physical
dimensions but with a heat leak approximately one-tenth that of the standard
valve.
DESIGN CONCEPT
The bonnet extension tube is the key element because it is the major
heat leak. Aside from low heat leak, this tube must withstand pressure and
axial mechanical loads and be vacuum tight. An early design proposal was
644
to use a very thin stainless steel liner overwrapped with filament wound
epoxy/fiberglass. However, in a previous application,a fabrication of the
liner (0.003 inch wall) and the filament winding proved to be expensive and
delicate from a manufacturing perspective. Use of a fiberglass/epoxy tube
slipped over a thin liner was rejected for reasons of dimensional fit and
perceived difficulties in achieving axial rigidity. (In fact, axial stretch
proved to be the major driver in the design selected). As shown in Figure
1, the concept selected utilizes off the shelf thin wall (0.006 inch) flex tube
as the liner, an overwrap of 5-2 fiberglass and epoxy, and heavy tubular ends
for welding to the valve body and topworks.
S- 2 GLASS/EPOXY
FILAMENT WOUND
OVERWRAP
LOW HEAT LEAK

TEFLO SPACERS
FILAM ENT WOUND

COMPOSITE STEM
STAINLESS STE EL
FLEX BELLOWS
Figure 1. S2/bellows bonnet extension.
To understand the design, it is necessary to consider the function of each

component. The stainless steel flex serves only as a pressure boundary and
its heat leak contribution is reduced by less than one-half due to the
convoluted length. The filament wound overwrap is only mechanical. It
transfers axial loads between the end stubs without bearing on the flex while
providing sufficient restraint to prevent squirm. Since the end stubs are not
considered part of the thermal resistance, they are of heavy construction to
provide rigid terminations for the filament wound structure. Machined end
645
rings facilitate welding assembly arid removable chill rings are used to
protect the epoxy from overheating.
The stem assembly is a straight forward application of composite
technology. The stem itself is an S-2 fiberglass/epoxy filament wound tube
with sufficient bearing area and moment of inertia to function as a column
for valve closing forces. Slotted end connections provide epoxy area
needed to develop required tensile strength and constrain the tube ends to
prevent dissociation under heavy compression load.
Teflon has low thermal conductivity and solid pieces are relatively
inexpensive to fabricate. However, as calculated above, teflon packing heat
leak of the standard valve is greater than the design goal for the entire low
heat leak valve. The concept valve utilizes machined Teflon discs inside the
valve stem and rings in the packing space which provide multiple surface
contacts, extended thermal path lengths, and reduced heat flow areas. The
result is a calculated contribution of 0.082 W for both Teflon thermal paths
which is down by a factor of ten from the original.
A valve incorporating the features described above is shown in Figure
2. This assembly, for which a U.S. Patenta has been applied, has a calculated
heat as follows for 10 inch thermal lengths:
Bonnet extension (0.075" wrap) 0.349 W

Stem 0.034
Projected heat leak of this valve is 0.5 to 0.6 W with a contingency

allowance and MLI contribution.
Figure 2. Prototype valve with ultra low heat leak features.
646
DEVELOPMENT AND TESTING
A valve utilizing the concept design was fabricated almost

immediately after it was proposed. In general, this valve proved the
concept but it's axial rigidity was unsatisfactory. Problem areas were excess
flexibility in the overwrap due to a filament winding angle too far from axial,
welding heat damage to the epoxy terminations, and too much stretch due
to the combination of low modulus (8 to 8.5 X lOB psi) and a thin (0.025 inch)
layup. Corrective actions included decreasing the winding angle to + /- 10°
from axial, reinforcing the end winding detail, adding fusion welding lips
with provision for removable chill rings, and increasing the filament winding
thickness for low stretch. The last change was the key improvement. In
cycle testing it was determined that good seat sealing performance was
dependent on achieving rigidity similar to that obtained with a stainless steel
bonnet extension. This meant that the overwrap thickness would have to be
nearly four times that of an equivalent stainless steel tube because of the
modulus ratio of about 3.5. Since the S-2 glass and epoxy composite is
considerably stronger than stainless steel (Actual uniaxial ultimate is in
excess of 150 ksi or 1,340 MPa.), stretch, not strength, becomes the driver
for design of the overwrap.
A prototype valve incorporating all of the above features has been
fabricated and tested recently. A photo of the revised bonnet extension is
shown in Figure 3 and Figure 4 is a photo of the complete valve.
Figure 3. Revised bonnet extension to the latest sse specifications.
647
This valve remained bubble tight through 4,000 pressure cycles at 78 K with
a seat differential of 0.5 MPa and a closing force of 2,856 N (642 lbf). In a
second test, the valve remained bubble tight through an additional 1,000
cycles at 78 K with a seat differential of 2 MPa and a closing force of 4,911
N (1,104 lbe). Prior to cycle testing, the valve seat was mass spectrometer
checked and found to be tight to less than 1 X 10-9 atm cc/s of helium for
pressure either above or below the seat.
Figure 4. Photograph of prototype ultra low heat leak valve.
ADDITIONAL VALVE FEATURES
For further heat leak reduction, the bonnet extension can be made in
two elements in series with a 78 K heat station between them. For Carnot
work/factors of 2.881 W /W for nitrogen and 70.056 W /W for helium,
minimum total refrigeration work is realized when the heat station is located
36.2% down from the warm end. Although the actual lengths may differ, if
a heat station is added to the dimensions previously calculated, the revised
heat leak becomes:
Bonnet extension 0.062 W

Stem 0.034
648
Heat stationing combined with optimization of the composite design
offers the potential for cost effective, low heat leak valves.
One of the SSC design requirements is for radiation resistance to an
accumulated dose of 10-9 Rad. The wet-wrap epoxy used in the filament
winding operation has been specially developed4 and tested for radiation
exposure in excess of the specifications.
CONCLUSIONS
A low heat leak, mechanically rugged cryogenic valve has been

developed and tested. Although further work to reduce manufacturing costs
is in progress, the valve presently shows good potential for cost effective
reduction of SSC system heat loads.
REFERENCES
I. SAES Pure Gas valve Model CV8-S12-5WPY2.

2. G. E. McIntosh, et al, Bayonet for superfluid helium transport in space, in:
"Advances in Cryogenic Engineering," Vol. 33, Plenum Press, New Yor
(1988), pp. 885-891.
3. U. S. Patent application Serial No; 071711, 629 dated 06/06/91.
4. CTD-IOI Epoxy Composite Technology Development, Inc., Boulder,
Colorado.
649
CONCEPTUAL DESIGN OF A SUPERCONDUCTING
MHD PROPULSION SYSTEM
H. Gurol, L. Nguyen, and H. H. van den Bergh

P.O. Box 85990,
San Diego, CA 92138
ABSTRACT
In this paper, we discuss the conceptual design of the superconducting magnets for
a magnetohydrodynamic thruster for a "generic" full-size submarine. The advantages of
this propulsion system are the elimination of the screw, and gears, and the acoustic
signature they produce. The major magnet design issues are: low magnetic signature,
robustness, and light weight. Low external magnetic signature was achieved by a unique
toroidal magnet array that produces an azimuthal magnetic field around the hull, and still
allows conducting sea water to enter the channel. This configuration also results in
maximum use of the field to produce thrust. Detailed design of the coil ends, as well as
the use of shielding materials may allow further reductions in the magnetic signature.
Robustness of the coil is achieved by the use of a CICC 80 kA conductor, that uses NbTi
at 4.2 K. Low magnet system weight is achieved by use of a filament wound structure for
the coil, honeycomb steel for the vacuum vessel, and detachable leads.
INTRODUCTION
The use of magnetohydrodynamics (MHD) for submarine propulsion offers the
potential advantages of increased acoustic stealth, since the screw and all rotating
machinery associated with propelling the vehicle are eliminated. This is replaced by a
MHD thruster that consistS of a superconducting magnet and a system of electrodes that
induce a current in the sea water perpendicular to the magnetic field. This results in a
1 X .B. force that propels the submarine. Even though a potentially significant reduction in
acoustic noise appears possible, the MHD thruster concept brings with it some new
concerns. These concerns include: magnetic signature, coil and cryogenic system weight,
and robustness of the magnet under service conditions.
General Dynamics Space Systems Division (GDSS) has been working with A YCO
(Textron), funded by DARPA, for the last 1.5 years, to develop a conceptual design of a
thruster system for a "generic" submarine l . System analysis, and channel design were
performed by Textron, and the superconducting magnet conceptual design was performed
by GDSS. The work done to date is of a general enough nature so that the application to
any specific platform requires a greater level of detail than achieved in this study.
However, the study serves to investigate geometric configurations and channel parameters
that optimize thruster efficiency while minimizing acoustic, electric and magnetic field
signatures. The superconducting magnet is probably the highest risk component in the
thruster and is the focus of this paper. The approach taken to address the concerns of
magnet signature, weight, and robustness is summarized below.
Supercollider 4. Edited by I. Nonte

- Magnetic signature is minimized by using a toroidal geometry with an "8-post" design.
The current flow geometry, coil gaps, and post pieces are chosen to yield a rapid magnetic
field drop-off rate.
- The coil system mass is minimized by the use of a composite coil structure, lightweight
steel vacuum vessel design, relatively high current density conductor, moderate field level
(6T), and detachable leads. A total magnet system mass (coil, cryogenics and all
associated systems) of 1000 MT results in a neutrally buoyant vessel.
- All the selections have been made to make the system as robust as possible to ensure
survivability. The chosen conductor concept (CICC) has a very high stability margin; the
conductor current is 80 kA, and the average channel field is 6 T, with a peak field on the
coil of about 9 T, allowing the use of NbTi superconductor; He-I cooling is used to
simplify the cryogenics and minimize the refrigerator power; a moderate field level
ensures the superconducting strands are held in place tightly to reduce relative motions that
could lead to a coil normalization.
Even though the resulting magnet pushes the state-of-the-art in terms of its size (15
m length), stored energy (13.5 GJ), high stored energy per unit mass (13.5 J/gm), and
magnetic stealth (earth's field in -100m), it nevertheless achieves these numbers without
resorting to any new superconductor technology, shielding, or cryogenics. The things
that are unique are the lightweight structure and vacuum vessel (pressure hull), the toroidal
geometry, and the use of detachable leads. The conceptual magnet design will be
discussed in the next section.
CONCEPT DESCRIPTION
Figure 1 shows the resulting magnet configuration. The arrows indicate the current
flow directions. The magnet consists of 16 separate saddle-type coils, assembled end-to-
end to result in the 8 posts shown. Current flows along the straight sections, producing
the toroidal channel field. When the current reaches the coil end, it makes a 90 degree tum
then goes to the post and flows in the radial direction. The total current in the posts is
twice the current in the straight sections, since two coils come together at each post. The
current then flows on a return path on the opposite side of the channel. For a 6 T channel
field, a peak field of about 9 T occurs at the coil ends where the conductors see the
sharpest bend radius. The coil length is 15 m, coil inner radius is 6 m, separation between
electrodes 1.5 m, and the coil outer radius is about 8 m.
---
_I
Figure 1. The Baseline Coil Geometry Using the Eight Post Design. The Arrows Indicate the Direction
of Current Flow.
Each of the 16 magnets can be wound off-site, then shipped to the docks for
assembly into the toroidal configuration. The coils will be connected into 4 sectors for
redundancy, so that if there is a coil out condition in a quadrant, there would still be field
from the remaining sectors for propulsion. The structural requirements and resulting
652
signature from a coil out condition will need to be investigated in future studies. The
conceptual design discussed in this paper assumes operation of all the sectors. It is
anticipated that splices would be needed only where the coils are connected to each other.
A variation of the 8 post design shown in Figure 1 is seen in Figure 2. This is a
much simpler magnet to manufacture for two reasons; first the large bend radius at the
ends makes winding a high current conductor (80 kA) much easier; second the dimension
of the coil in the azimuthal direction is the same at the coil inner and outer radii; this results
in a much simpler winding platform. There is a fairly large gap between coils at the outer
radius. This was of concern initially; however, magnetic field calculations show that this
configuration actually results in a lower far-field signature since the field leaking from the
gaps is opposite in direction to the field from the coil ends, resulting in a partial
cancellation.
'-( ~
Figure 2. Variation of the Eight Post Design Results in a More Easily Producible Coil
The coil structure must withstand the Lorentz forces generated by the coil when it is
carrying current. Conventional steel structures would be too heavy and would not result
in a neutrally buoyant vehicle. Therefore, it was decided to use filament-wound
composites for the structure. Figure 3 shows the coil, coil support structure, and the
lightweight vacuum vessel, which also must withstand the external water pressure.
Internal support rods are used to connect the vacuum vessel sections in the channel to the
sections on the outside. The "holes" that are produced by the posts have been found not to
affect the magnetic signature as long as care is taken to conserve azimuthal symmetry.
However, these posts will degrade the properties of the composite materials used in the
coil structure which the posts pass through.
Figure 3. Magnet Cross Section Showing the Use of Composite Structure and Honeycomb Vacuum
Vessel.
653
A number of composites are being considered for use in the structure. Since the
structure next to the hull is under compressive loads, a composite that has good stability
characteristics needs to be used there; on the other hand, the structure at the outside radius
is under tension and hence will require the use of a composite that has good stren&th
properties. More than likely, an optimum design would use two types of composites.
There have been a fair number of large structures that have been built using filament
winding techniques, and at this time this technology does not appear to be a show-stopper.
Some examples of large composite structures include the case for the Space Shuttle
booster, and long (-45 m) wind turbine blades. The most important thing to keep in mind
is to use realistically degraded properties for the composites. Table 1 shows the properties
of selected composites for the inner and outer radius support structures, as well as
stainless steel for comparison. The outer support structure weight is dictated by its
strength since it is under tension, while the inner support structure is under compressive
loads and hence its weight is dictated by its stability. The total structure weight is the sum
of the inner and outer weights in this table. Obviously the steel structure is not feasible
because of its mass. It is seen that two different composites may yield a lower mass
system.
Table 1. The Properties of Selected Materials Showing the Thickness of Structure

Required Dictated by Stability and Strength, the Inner and Outer Support Structures.
Outer Support Structure (Limit Load 14330 psi) Strength Determined
Carbon-Epoxy Aramides-Epoxy Polyethylene St. Steel
IMG iHIM Kevlar 49 S nectra 1000 17 - 4 PH
Strength Thickness (In) 3.17 3.7 3.0 2.5 2.9

(00/900)s
Strength Weight (Ton) 95 llO 87 41 430

(0°/90°)
Inner Support Structure (Limit Load 2417 psi) Stiffness Determined

Carbon-Epoxy Aramides Boron Ultra-High E St.Steel
IMG UUM Carbon-Epoxy
PIQOS_ 17 - 4 PH
Stability Thickness (In) 13.0 11.2 10 9.5 5.3 9.6

(00/900)s
Stability Weight (Ton) 319 260 484 309 147 lloo

(O°/90 0 )s
In addition to the coil structure, the other major weight contributions come from the
vacuum vessel and the conductor. An overall conductor current density of 18,000 A/cm2
is used to minimize conductor mass. The vacuum vessel for this coil is also a pressure
hull, since there is water on the outside surfaces. If a conventional steel hull were used for
the vacuum vessel, calculations show that it would be heavier than the combined weights
of the conductor and structure. For this reason, a honeycomb steel vacuum vessel design
is used (Figure 3). A composite vacuum vessel has not been chosen because it is
considered a high risk item requiring much development (mainly due to the complexities of
the ends). The basic design philosophy was to minimize as much as possible items that
need significant development. However, if a composite vacuum vessel eventually
becomes feasible, it will offer additional weight savings. The total weight for the magnet
mass is 1,000 MT, which allows the submarine to be neutrally buoyant.
654
CRYOGENIC SYSTEM
In this section the conductor and the associated cryogenic system will be discussed.
Choices were made in ways that would reduce mass, increase survivability, and minimize
compressor power. To achieve these goals, an 80 kA cable-in-conduit-conductor (CICC)
was chosen as the baseline conductor for its high stability margin and low mass;
detachable leads are used to minimize compressor power; and sea water dump electrodes
result in a low weight energy dump system.
The 80 kA baseline CICC conductor has an average current density of 18,000
A/cm2 , uses NbTi superconductor strands with a ratio of copper to superconductor of
about 2: 1, and an overall ratio of copper to superconductor in the cable in the range of 10: 1
to 20:1.
The refrigerator capacity required is determined by the cold mass support heat leak,
thermal radiation, the number of lead pairs (we have 4) and their type. There are 3
different types of leads; these are: a) permanently attached, always carrying current, b)
permanently attached, carrying current only during coil charging and discharging, and c)
detachable leads. Detachable leads are attached to the coil leads only during coil
charging/discharging and during a coil dump. During normal operation they are detached
from the coil, and hence do not have any resistive losses. Furthermore, detachable leads
are designed such that there is no direct path from the 4.2 K region to room temperature,
meaning no heat load in the leads. Hence, with detachable leads the only heat load will be
due to support heat leaks and thermal radiation. Detachable leads are the option with the
lowest compressor power (and resulting refrigerator weight), and are our choice as the
baseline. This is an item requiring development work, since there has not been a need for
high current detachable leads up to now. However, they are used routinely, albeit at much
lower current levels, in Magnetic Resonance Imaging (MRI) machines. The required
compressor power has been estimated to be 182 kW; this translates into a refrigerator
weight of about 11 MT.
Because the coil is segmented into 4 sectors, there are 4 pairs of detachable leads;
hence, the energy stored in these 4 segments can be charged and discharged
independently. The charge and discharge rates can be adjusted to minimize inductive
coupling between segments. In case of a coil segment quench, the coil energy can be
dumped into sea water by way of small electrodes located at the ends of the thruster.
Voltage taps in the coil detect a coil normalization. Once the voltage drop is detected, the
dump electrode leads are connected to the coil leads. Since, the resistivity of water is high
(....Q.25 ohm-m), only a fairly small electrode surface area is needed to give reasonable L!R
time constants. The detachable leads are in the vacuum vessel and hence do not lead to
any heat leak during nominal operation. However, during a quench they carry all the 80
kA to the electrodes. Current flowing between the electrodes results in energy dissipation.
The heat load on the LHe during a coil energy dump occurs for only a short time, hence
the resulting temperature rise can be brought back to 4.2 K by the existing 182 kW
refrigerator.
To get an idea of the electrode size required, one must select a maximum coil dump
voltage. A dump voltage of 15 kV was chosen for this study, because of the inherent
stability and safety of the CICC concept from a voltage breakdown point of view.
Additional safety can be achieved by using the voltage protection scheme developed for the
Superconducting Magnetic Energy Storage (SMES) coil design 2 . This design uses
insulation and conducting paint to ensure that the voltage difference between any two
surfaces in the coil pack does not exceed the breakdown voltage for He gas under all
conditions of temperature and pressure. Using the coil inductance of 1.06 H for a
quadrant (total coil energy = 13.5 GJ), gives an external resistance of 0.19 ohms, which
results in a coil time constant of 5 seconds. This is fast enough to avoid damaging the coil
in case of an energy dump. Assuming an electrode spacing of 1.5 m, and using sea water
conductivity of 4 ohms gives the electrode surface area of about 1.8 m2 . This is a small
electrode area compared with the main electrodes and can be mounted at any convenient
location.
655
MAGNET DESIGN TRADE STUDIES
The major trade studies for the coil involved design approaches to achieve a low
magnetic signature external to the coil as well as keep peak fields in the coil to levels below
about 9.5 T, which is about the maximum field in which NbTi superconductor can be
used. The initial concept is seen in Figure 4a. This configuration has large gaps between
the 16 coils, which results in a large magnetic signature as well as a high field ripple in the
channel. To close up the gaps, the next design configuration was a 4 post design (Figure
4b), with pairs of nested saddle-type coils. This results in a total current in the posts that
is 4 times the current in the straight section of each of the 16 coils. The result of so much
concentrated current is that the peak magnetic field in the post region is unacceptably high.
In order to reduce this peak field, we went to the 8 post design that is the present baseline
(Figure 1). The current in the posts is now one-half of the 4 post design. Another way of
Figure 4. Trade Studies Performed to Minimize External Magnetic Signature as Well as to Minimize
the Peak Fields Included (a) Initial "Leaky" Concept, (b) Four-Post Design, and (c) Twelve-Post Design.
reducing the post current density would be to increase the post dimensions; this however
would increase the hydrodynamic drag. The resulting peak field in the post was calculated
to be less than 9.5 T. We also investigated the possibility of a 12 post design (Figure 4c),
as a means of further reducing the peak field. The peak field was reduced by about 20 %
over the 8 post design. However, this comes at the expense of greater complexity;
therefore, the 8 post design was chosen as a reasonable compromise. As far as the
magnetic signature is concerned, there was very little difference between the 4, 8, and 12
post designs.
656
MAGNETIC SIGNATURE ANALYSIS
The baseline coil geometry with the current flow directions and dimensions used for
the magnetic signature analysis is shown in Figure 5. There are 16 individual saddle-type
coils, arranged such that each coil has its mirror-image as its neighbor. This results in 8
radial posts. The number of radial posts is important for two reasons. First, it determines
the field contribution of the radial currents. And second, it keeps the peak magnetic field
in the coil below about 9.5 T so that NbTi superconductor can be used.
/ - ~---
/
/
~+r------------/-~"~~~
'/_~J-~-----1S~
~ --~~~
<...,
"/
Figure 5. Baseline Coil Geometry Showing Dimensions Used for the Magnetic Signature Analysis.
If the field signature was due only to the radial currents, then the field would
drop off faster as the number of radial posts increases. However, the magnetic signature
at a given point in space is the result of the sum of contributions from the radial posts,
the coil main body, and the azimuthal currents at tge ends. The ~agnetic fields were
calculated using two 3D magnetic field codes: EFFI , and TOSCA which is also able
to handle the effects of permeable materials. The field contribution of the azimuthal
currents is found to be much less than the other two. However, the field from the posts
and the coil main body produce fields away from the submarine that oppose each other.
This leads to a characteristic "field-reversal" as seen in Figure 6, which shows the
magnitude of the field as a function of radius r (r=O is the center of the sub). The field
is nonnalized to the central channel field. The "dip" observed at about 25 m corresponds
to this reversal point where the fields from the posts and the main body just cancel each
other. At distances less than 25 m, the dominant field contribution is from the coil main
body; for r>25m, the field from the posts dominates.
10 - I
0
cc
.......
10 -2
cc
C
...J
w 3
;:: 10-
C
w 10--
N
:J
«
:::I:
a: 10- 5
C
Z
10- 6
0 5 10 15 20 30 35 _0 45 50
RADIUS (m)
Figure 6. Magnetic Field as a Function of Radius at the Magnet Axial Centerline, Normalized to the
Average Channel Field.
657
To verify these effects and also to prove the validity of the magnetics computer
modeling, a 1/50 th scale model of the coil was wound. The coil used 14 gauge copper
wire, carrying a current of 18 amps, with an average channel field of 112 Gauss, and was
cooled by ambient air. Figure 7 shows the experimental coil. The fields were measured
with a Hall probe accurate to about 1 milligauss, that could be rotated, and moved in the
axial and radial directions. The fields were measured and compared with computer models
of the coil. It was found that the measured "near-fields" are in qualitative agreement with
the computer models, showing the field reversal point, as well as the broad peak following
this point. The magnitude of the "far fields" (r > 50 m) are in good quantitative
agreement.
The effects of permeable materials on the coil signature are being evaluated using
the code TOSCA. This includes the effects of materials already there (eg., the hull), and
permeable material shields that will divert the flux into the channel instead of letting it leak
out, further reducing the field signature. There are three main issues related to permeable
materials: first is the effect of the hull on the earth's magnetic field in the vicinity of the
submarine; second is the effect of the hull and vacuum vessel on the coil magnetic field;
third is the effect of shielding materials at the coil ends on the magnetic signature. All this
leads to the question of how much signature should the magnet be allowed to project? It
seems that the best way to answer this question would be to estimate the signature
resulting from all systems other than the magnet, and to keep the magnet signature below
this value.
MAINTENANCE AND INTEGRATION ISSUES

There are no expected maintenance requirements on the coil itself. However, the
component that will require regular maintenance is the compressor. In fact, this is the
single most common reason for magnet shut-down in land-based magnets. Therefore, the
compressor system must have redundancy built into its most break-down prone
components. The only other component that has moving parts and may require regular
inspection and maintenance are the detachable leads. There is no data on high current
detachable leads.
The current concept is basically a retrofit unto an existing platform and requires
integration only of the inner support structure to the hull. Since the inner support structure
can carry all of the radial loads that are generated by the coil, there will be no Lorentz
forces transmitted to the hull. This means that the only mechanical integration
requirements are attachments between the coil and the submarine hull that can withstand
gravity, and external vibration loads.
Since there are 4 pairs of leads on the coil, there will also be 4 pairs of detachable
leads. These leads can be grouped and routed through a single hull penetration. If the
detachable leads are positioned inside the hull, then the vapor-cooled leads will likely be
inside the hull as well. This means that the hull penetrations will have to accommodate
cryogenic He. A separate cryogenic penetration will be required for the He feed lines to
the coil from the refrigerator.
SUMMARY
The conceptual design developed for this study had one major emphasis from the
beginning. It had to be of a design that was as producible as possible with existing
technology,. or with technology that was not too far away or risky. The thinking was that
if we had to build this magnet today, it could be done with a detailed design and
development program. Therefore, choices were made which were conservative on the one
hand, but at the same time brought the concept back to reality. For example, the 6 T
channel field was chosen since there are no known materials (including real-world
composites) that can withstand the loads of higher fields and still be light enough to
maintain a neutrally buoyant submarine. We believe that the coil is windable since the
bend radii have been chosen large enough to accommodate a high current conductor. High
658
current conductor is used to minimize dump voltages making the system safer and more
reliable. Since GDSS has recently finished testing a 200 kA conductor for SMES, the 80
kA conductor is expected to require a similar test program. Other items requiring
development are the use of composites in the structure, development of high current
detachable leads, shock isolation of the coil, a reliable cryogenic system, and a verification
of the energy dump at sea concept.
REFERENCES
1. Captain Richard Sharpe, OBE, RM, Jane's Fighting Ships, 1988-89, Jane's Publishing Company
Limited, London, England.
2. R.J. Loyd, T.E. Walsh, and E.R. Kimmy, "Key Design Selections for the 2004 MW-hr. SMES /
ETM"; IEEE Transactions on Magnetics, Vol. 27, No.2, March 1991.
3. S.J. Sackett, EFFI - A Code for Calculating the Electromagnetic Field, Force, and Inductance in Coil
Systems of Arbitrary Geometry. Lawrance Livermore Laboratory Report, VCID - 17621, May
5, 1977.
4. TOSCA - Total Scalar Potential- A 3-D Vector Fields code for calculating fields in the presence of
permeable materials
ACKNOWLEDGEMENTS
Work performed under Textron (AVCO) subcontract RL264037 sponsored by the

Defense Advanced Research Projects Agency (DoD), Naval Technology Office,
Submarine Technology Program, DARPA Order No. 6660 issued by DTRC under
contract N61533-89-C-0067.
659
RA W MATERIALS AND EARLY MONOFILAMENT ANALYSIS
FROM THE VENDOR QUALIFICATION PROGRAM
J. M. Seuntjens, V. A. Bardos, D.W. Capone II, F. Y. Clark,

E. S. Coleman, M. J. Erdmann, and B.A. Troupe

Dallas, TX 75237
INTRODUCTION
Eight cable vendors are presently fabricating materials as part of Phase I of the
Superconducting Super Collider (SSC) cable qualification program. This program is
designed to prepare each vendor to compete for the supply of superconducting cable for
dipole and quadrupole magnets for the SSC magnet subcontractors. Phase I is split into a
final R&D effort (Phase IA) of approximately 4000 kg, and a baseline production process of
3500 kg (Phase IB). As part of the R&D effort variables for most cable vendors,
approximately half of the Phase IA billets are being fabricated using NbTi alloy from the
cable vendors' "primary" alloy vendor and the remainder from a "secondary" alloy vendor,
as the vendor sees appropriate. All cable vendors have chosen Teledyne Wah Chang-Albany
(TWCA) as their primary alloy source. Two cable vendors have chosen Toho Titanium, and
four have chosen NRC, Inc., as their secondary alloy source.
The purpose of this paper is to report the typical values quoted on material supplied in
this effort as well as verification measurements made at the SSC Laboratory (SSCL). The
final product chemistry, grain size, hardness, and radiographs from the alloy vendors will be
reported along with grain size, hardness, and relative homogeneity measured by EDX at the
SSCL. Normalized monofilament extrusion data are also discussed. Data will be analyzed
for significant differences in raw material properties. These differences are to be noted
during future analysis of the final strand manufactured from these material. Only at that
time will it be possible to determine positive or negative impacts of specific properties of
the raw materials.
No. DE-AC35-89ER40486.
Supercollider 4, Edited by J. NonIe

EXPERIMENTAL
Chemistry data on raw materials are as reported to SSCL. Specification SSC-Mag-M-

4000A for NbTi alloy and specification SSC-Mag-M-4001 for niobium barrier require the
submission of data for raw materials. Samples of NbTi alloy supplied to SSCL were either
disk samples of ingot or post extruded rod. Samples of Nb sheet were also supplied.
Hardness measurements were made with a Leco M-400-G 1 micro hardness tester with a
300-g load and a Vickers indentor. Grain size was measured as a function of radius on
samples etched with a 1: 1: 1: 1 mixture of hydrofluoric acid, nitric acid, lactic acid, and
water, and photographed with a Mitutoyo FS-110 light microscope. The relative
homogeneity of the NbTi alloy was measured on as-polished sections using a Link eXL
energy dispersive X-ray analysis (EDX) on a JEOL 6100 scanning electron microscope
(SEM). An X-ray spectrum collected over an approximately I-mm 2 area was used as a
standard and assumed to be equivalent to the average composition specified by the vendor.
Each run consisted of a line along the specimen radius comprised of 101 points spaced
10 !lm apart. Two runs were made on each sample, one line near the center of the sample
and the other near the outside. Integrated peak height ratios were corrected for atomic
number, absorption, and fluorescence (ZAF) in the Link program. Tensile measurements
were made using a Instron 4302 Tensile Testing system with flat serrated grips.
Transmission radiographs of NbTi disks were made taken by Bonded Inspection, Inc. Image
analysis was used to enhance the brightness and contrast to reveal features difficult to detect
in a standard image of a X-ray radiograph. Image analysis was performed using the Link
eXL system connected to either the SEM or a Hitachi KP-MIU video camera. A complete
description of image analysis routines will be published elsewhere.
RESULTS
The heat numbers for the NbTi alloy and Nb sheet used at each of the cable vendors
for Phase I were checked for commonality between vendors. One cable vendor has two
heats in common with two other cable vendors. No common heats of secondary alloy
sources were found among the cable vendors.
Table 1 contains averages for the alloy compositions and hardness, as well as typical
ASTM grain size numbers for the heats of the primary and secondary sources for each of the
cable vendors. Averages and coefficients of variation (C of V, the standard deviation
divided by the mean) are calculated for each alloy vendor's material. Only a subset of the
elements listed for analysis in the SSCL specifications is given. Elements required by the
specifications (but not listed here) vary little between heats or between alloy vendors.
Table 2 contains select reported Nb chemistry data as well as yield strength and
elongation for the four Nb sheet vendors used by the cable vendors. The number of heats is
too small to calculate C of V for each vendor. Table 3 contains an average and range of the
reported copper RRR for can materials of each cable vendor.
Table 4 shows selected homogeneity, hardness, and grain size data measured at the
SSCL from each of the alloy and sheet vendors. The reported hardness and grain size from
the alloy and sheet vendors are given for comparison. The samples chosen reflect the range
of diameter and heat treat conditions supplied by each cable vendor.
Figure 1 shows relative titanium concentrations along I-mm-Iong radial lines for the
same samples from each alloy vendor as in Table 4. The radial line was approximately 1 cm
from the surface of each disk, and the standard deviation for each data set is given in
Table 4.
662
Table 1. Reported data on NbTi alloy from three alloy vendors.
Cable Mean Mean Mean Mean Mean Mean Grain Hardness

Vendor Ti% 0 C Fe Si Ta Size Hv
ppm ppm ppm ppm ppm ASTM
46.5 550 45 36 25 828 3.7 130

2 46.6 511 39 36 25 893 4.0 129
3 46.7 525 40 42 62 937 3.7 132
4 46.5 581 98 65 100 827 2.0 133
5 46.4 542 74 47 57 884 3.3 133
6 46.7 555 32 44 45 938 2.5 131
7 46.5 603 41 43 44 798 3.2 132
8 46.7 560 40 35 25 905 2.5 128
TWCA ave. 46.6 553 51 44 48 876 131
C of V (%) 0.2 5.7 47.1 22.6 53.9 6.4 1.2
47.6 543 88 60 3 47 4.3 131

47.7 608 112 85 5 56 6.0 140
47.8 533 97 63 6 31 5.0 133
47.4 422 106 45 10 95 5.0 125
NRC ave. 47.6 527 101 63 6 57 132
C of V (%) 0.4 14.7 10.4 26.1 49.1 47.5 4.7
47.5 343 40 116 20 696 6.3 127

47.7 346 102 107 53 777 2.5 126
TOHOave. 47.6 345 71 112 37 737 127
C of V (%) 0.3 0.6 6L7 5.7 63.9 7.8 0.6
Table 2. Reported data on Nb sheet sorted by Nb sheet vendor.
Nb Mean Mean Mean Mean Mean Grain Yield Elongation

Vendor 0 C Fe Si Ta Size MPa Percent
ppm ppm ppm ppm ppm
Cabot 200 20 10 10 140 8.0 98 40

NRC 81 11 20 20 200 6.1 92 31
DENKAI 80 40 10 20 900 4.0 81 24
TWCA 65 25 50 40 690 8.0 108 46
663
Table 3. Reported copper RRR values for monofilament
and multifilament cans.
Vendor Min Max Average CofV

RRR RRR Percent
345 404 384 4

2 148 301 183 16
3 301 358 336 9
4 317 346 328 4
5 269 317 285 7
6
7 317 343 322 4
8 282 288 286
Table 4. Homogeneity, hardness, and grain size data on NbTi alloy and Nb sheet.
Alloy Reported Measured Reported Measured Reported Measured

Sample wt.% Ti Std. Dev. ASTM ASTM Hardness Hardness
Homogeneity Major Grain Size Hv Hv
Grain Size
NRC 200 47.6 0.40 3 3.5 141 134

TOH0200 48 0.42 2.5 4.5 124 134
TWCA200 46.9 0.58 3.5 130 140
NRC 250 47.4 0.85 2.5 3.2 144 135
NRC 150 47.6 0.87 4 4 138 138
TWCA 150 46.2 0.38 4.5 3 131 138
TOHO 150 46.8 0.71 5.2 6 129 132
Sheet thickness Reported Measured Ingot Sheet

Vendor mm Grain size ASTM Hardness Hardness
Cabot 0.6 8 8.5 74 83

Denkai 1.5 6 4 56 74
NRC 1.4 7.2 3.2 65
TWCA 0.6 8 7.2 46 80
TWCA 1.54 8 8.8 56 79
664
TOHO 150
NRC 200
TOH02oo
r-.."'---"'~."'I'vJ'-...,...,...--.,~.............,..,..---,.,..,....,.- 1WCA2oo
r... ~....
/", l' "'- r '...
70 "'--'.f ,--,/".,.~ "VOl,.,........,: _ .......... NRC 250
~~....J.r~"'--.,~ NRC 150
1WCA 150
60~~--~--~--~~--~--~~------~
o 200 400 600 800 1000

Microns
Figure 1. Relative Ti concentrations along I-mm radial lines in representative material from each alloy
vendor. The legend gives the alloy vendor and the material diameter at the time oftesting, rounded to the
nearest 10 mm.
Figure 2 shows representative micrographs of the grains of each alloy vendor. The
NRC grains appear to contain grains and "subgrains". The subgrains are low-angle grain
boundaries resulting from the plasma melting process. l The subgrain size is the reported
grain size in all of NRC data.
Figure 3 shows representative radiographs of each alloy vendor's material. While all
raw materials are nominally homogeneous to less than ± I %, the images yield some
information about the deformation after casting. All alloy vendors' material meets
homogeneity requirements, with the exception of one cable vendor's Toho alloy. Some
alloy vendors' radiographs were less uniform in brightness than others, which limits the
contrast enhancement for the detection of very faint defects.
Monofilament dimensions, as well as pre-extrusion compaction processes, vary widely
among cable vendors. The standard deviations in monofilament diameters, NbTi core
lengths, running pressures of monofilament extrusion, and extrusion reduction ratios are 31
mm, 124 mm, 186°C, 183 MPa, and 35, respectively. The area fraction ofNbTi at the mid-
cross section of the billet at extrusion size has a range of 10%. The SSCL strand
specification calls for 1-llm nominal spacings on 6-llm nominal filaments, resulting in a
nominal LAR of 0.5 (66 area percent NbTi). The average billet extrusion local area ratio
(LAR) for all cable vendors is 0.589. The cross sections of each cable vendor's billets are
similar enough to make a comparison of the extrusion data to a modified form of the simple
extrusion formula given in Dieter: 2
P'= k In (R),
where R is the area reduction ratio of the extrusion and P' is the pressure on the front
of the extrusion ram. Figure 4 contains a plot of k vs. temperature. Six of the eight vendors
were used to form the curve. The other two vendors' processes apparently do not ensure a
uniform billet temperature from a billet surface temperature measurement.
665
Figure 2. Grain Boundary line traces from nominally 50-X micrographs showing the grains of the samples in
Figure I. From top left to lower right: TOHO 150, NRC 200, TOHO 200, TWCA 200, NRC 250, NRC 150,
TWCA 150.
Figure 3. Representative radiographs for each alloy vendor' s material. From top left to lower right: TOHO
150, TOHO 200, TWCA 200, NRC 250, NRC 150, and TWCA 150.
666
280
260
240
'2
c.. 220
;5
6
tl 200
.s
-'"
180
160
140
200 300 400 500 600 700 800 900
Centigrade
Figure 4. Curve representing k factor vs. temperature data for Phase IB monofilament.
DISCUSSION
Table 2 shows significant differences in the chemistry (wt. % Ti and ppm weight for
trace elements) between alloy sources as well as variations in the material from one alloy
vendor at different cable vendors. The average titanium concentration of TWCA material is
reported at 46.6 wt.%, while Toho and NRC are 1% higher at 47.6 wt.%. Average oxygen
contents are considerably lower for Toho alloy (345 ppm wt.) than for TWCA (553) or
NRC (527). The average oxygen content of NRC material varies significantly between cable
vendors, from a low of 422 to a high of 608 ppm wt. The higher NRC heat numbers tend to
have lower oxygen contents than lower NRC heat numbers. The average silicon content of
NRC alloy (6) is considerably lower than Toho (37) or TWCA (48). Similarly, the average
Ta content is lower for NRC (57) than for Toho (737) or TWCA (876). The hardness for all
alloy material is similar at approximately 130 Hv. The grain size for NRC material is finer
than either TWCA or Toho of similar billet size. Monofilament billet size differences do not
allow direct comparison across all billets.
The average copper RRR varies significantly between cable vendors over a range of
183 to 384. However, the coefficient of variation at each cable vendor is typically a few
percent. This translates into a low temperature resistance variation of a factor of
2. Variations in RRR are important in checking for correlations to strand-to-strand contact
resistance in finished cables.
The chemical homogeneity data in Table 4 indicate that all alloy vendors' material has
a standard deviation of I % or less. The hardness data for the alloy agree reasonably well
with the alloy vendor (overall average of 136 Hv). The measured sheet hardness is
considerably higher than the reported ingot hardness, but are quite similar for all vendors
(average of 76 Hv). All ofthe sheet material had equiaxed grains with the exception of NRC
sheet, which had an aspect ratio of approximately 4. The circle template method used for
grain size measurement (ASTM standard EI12-85) averaged the dimensions in the rolling
and transverse directions.
Figure 1 shows that all alloy material show some composition variations on a spacing
of 200-300 f.lm. The grain size of X-ray negative film is in the range of 40-80 f.lm, and
therefore this chemical inhomogeneity is on a scale at the limits of X-ray negative
resolution. The X-ray negatives in Figure 3 do show compositional variations over a larger
scale. The "tree-ring" nature has a size scale on the order of 10 mm. In the case of Toho
667
alloy, Ti freckles are clearly visible from about 0.2 to 0.8 of the sample radius in one cable
vendor's material, while the other cable vendor's material supplied by Toho has no freckles
and only traces of tree rings. In general, the tree ring nature is less prevalent in TWCA
material than in NRC material. However, some heats of both TWCA and NRC material
show no tree rings, while others' heats show some tree rings. Tree rings tend to be most
prevelant at about 0.75 of the material radius. In some NRC material, the tree rings have a
four-fold symmetry. Since cooling on solidification is radial, a square-like tree-ring pattern
is thought to result from nonuniform metal working during deformation after casting.
Correlations to the NbTi core uniformity in monofilament at restack size will be studied in
later work.
Tree rings represent a variation in the Ti content on the order of 1% over a distance of
a few hundred microns. Such a variation will affect the Ti precipitation kinetics locally to
some extent. The SSCL has observed severely tree-ringed alloy material in past work. In
that case, filament distortion (non-roundness), filament roughness (non-smoothness of the
NbTilNb or Nb/Cu interface), or sausaging (axially varying cross-sectional area) was found
not to be significantly worse than other composites. Analysis of monofilament quality is
ongoing. Early reports of monofilaments with non-round cores have been reported for all
alloy vendors' materials. Future work by SSCL will check for correlations to features in
X-ray negatives with the monofilament morphologies.
The k factor data in Figure 4 cover a temperature range of nearly 600°C, and ram
speeds of 0.3-4 cm per second. Because the correlation coefficient to this curve was 0.98,
variations in press design, die design, and ram speed apparently do not strongly affect the k
factor for TWCA Phase IB alloy. The data available from the Phase IA billets made from
secondary alloy sources show that NRC and Toho alloys have k factors similar to TWCA
alloy.
It is likely that the higher titanium content materials from NRC and Toho in Phase IA
will respond differently (both electrically and mechanically) to heat treatments than TWCA
material in each cable vendor's baseline process. While it is clear that the alloys from NRC,
Toho, and TWCA are significantly different, there are no strong drawbacks at this early
stage to using any of these alloys for 6-llm strand fabrication.
ACKNOWLEDGEMENTS
Metallographic work by D. Washburn and V. Levy is acknowledged. The SSC
Laboratory is operated by the Universities Research Association, Inc, for the
U. S. Department of Energy under Contract number DE-AC35-89ER40486.
REFERENCES
1. R. Balliett, private communication.
2. G. E. Dieter, Mechanical Metallurgy, pp. 620-22, McGraw-Hill. 1986.
668
CABLE KEYSTONE ANGLE OPTIMIZATION FOR
40-MM SSC QUADRUPOLE MAGNET DEVELOPMENT
D.W. Capone, V.A. Bardos, D. Christopherson, F. Clark, S. Graham,

C. R. Hannaford, T. D. Reed, Jr., and J. Seuntjens

Dallas, TX 75237
INTRODUCTION
A series of cables was fabricated meeting all aspects of the latest Superconducting Super
Collider (SSC) specification for 3D-strand outer cables I with the exception of varying the keystone
angle between the current specification value of 1.2° and a maximum of 2.2°. In addition, a series of
cables with an increased pitchlength is presented. These fabrication experiments have been
performed in order to provide an expanded design envelope for the development of quadrupole
magnets by the Babcock and Wilcox/Seimens (B&W/S) team. By pushing the cable parameters
beyond currently acceptable limits, the B&W/S team will be provided with the data necessary to
perform a trade-off between performance, cost, and manufacturability. At the maximum keystone
angle of 2.2°, a coil cross section is possible that eliminates the need for wedges in the coil and
provides for a simplified magnet manufacturing process. This experiment focuses on the differences
identified within each of the cables as the keystone angle is increased. The sections below describe
the fabrication details of the cables produced and the metallographic features of the cables. There is
also a summary of the electrical and mechanical properties of the as-fabricated cables.
CABLE FABRICATION
All of the cables produced for this study were fabricated from a single batch of strand produced
by Oxford Superconducting Technology for use in the dipole magnets for the Accelerator System
String Test (ASST) program at the SSe. The properties of 3D-strand inner and 36-strand outer cables
produced for the ASST program have been described in a companion paper by Christopherson et
al. 2 All of the strand for all of the cables was loaded onto the 48-strand cabling machine at
Lawrence Berkeley Laboratory3,4,5 at the outset of the runs. A new set of turkshead rollers was
installed to bring the initial keystone angle of the cables in line with the midpoint of the SSC
specification. (Previous runs, as shown in Ref. 2, have been produced near the minimum keystone
allowed by the specification in order to minimize the degradation observed in the ASST cables.) In
addition, care was taken to seat the rolls properly within the set of roller bearings.
No. DE-AC35-89ER40486.

The cables were manufactured sequentially, with an effort made to maintain the cable parameters
at the midpoint of the specification tolerances for both the cable midthickness and the cable width.
The initial runs were l80-m lengths of standard 30-strand outer cable that were supplied to Seimens
for a series of winding experiments to initiate its magnet development program. These cables are
identified by the standard SSC cable ID convention, SSC-2-0-00052 and SSC-2-0-00053.
Following these initial runs, a series of five cables was manufactured using the same cable pitch-
length as the standard cable. These five cables were made with the following nominal keystone
angles: 1.2°, 1.4°, 1.6°, 1.9°, and 2.2°. These cables are identified as TST-A12, TST-A14, TST-A16,
TST-A19, and TST-A22. In addition to the five cables produced with the standard pitch length of
73 mm, three cables were produced with a pitch length of 81 mm. The final three cables had
nominal keystone angles of 1.6°,1.9°, and 2.2°; they are identified as TST-P16, TST-P19, and TST-
P22. See Table 1 for properties of the various cable samples.
Table 1. A summary of the cable parameters and variations measured during these experiments.
Cable ID Keystone Std. Dev. Mid-thick. Std. Dev. Width Minor

Angle (0) n (in.) (in.) (in.) Edge (in.)
SSC-2-0-00052 1.193 0.005 0.04591 0.00002 0.38370 0.04192
SSC-2-0-00053 1.188 0.005 0.04591 0.00003 0.38375 0.04193
TST-AI2 1.199 0.013 0.04590 0.00007 0.38357 0.04189
TST-AI4 1.389 0.009 0.04590 0.00006 0.38325 0.04125
TST-AI6 1.563 0.037 0.04591 0.00005 0.38348 0.04068
TST-AI9 1.905 0.013 0.04588 0.00004 0.38398 0.03950
TST-A22 2.219 0.018 0.04593 0.00003 0.38352 0.03850
TST-PI6 1.601 0.010 0.04589 0.00006 0.38370 0.04053
TST-PI9 1.911 0.011 0.04592 0.00007 0.38387 0.03952
TST-P22 2.213 0.021 0.04592 0.00007 0.38357 0.03851
Table 1. A summary of the cable parameters and variations measured during these experiments
(continued).
Cable ID Minor Edge Ie (4.22 K, 5.6T) Degrad. Modulus

Packing Factor (A) (%) (MPa)
SSC-2-0-OO052 0.958 9181 --0.8

SSC-2-0-00053 0.958 9263 -1.7
TST-AI2 0.959 9179 --0.8 7170
TST-AI4 0.973 9115 --0.1 6740
TST-AI6 0.987 8970 1.49 7070
TST-AI9 1.017 8819 3.15 7180
TST-A22 1.043 8545 6.16 6280
TST-PI6 0.099 8953 1.68 6630
TST-PI9 1.016 8846 2.86 6510
TST-P22 1.043 8665 4.84 5910
670
In order to verify the degree of success at modifying only the keystone angle during this
experiment, both the midthickness and the width of each of the cables fabricated for this study are
plotted in Figure I. The total axis height, for both width and midthickness, is approximately the
same as the range allowed by the SSC specification for these parameters. The data for the width of
the cables are complicated by the fact that the CMM cannot apply a sufficient load while clamping
on the edges of the cable to accurately measure the exact cable width. The measurement is a
compromise between obtaining as accurate a measurement as possible and preventing buckling of
the cable across the face during the measurement. As a result, the variation in the width represents a
worst case situation. (Note: The present version of the CMM software performs the measurement of
the cable dimensions in standard inch-pound dimensions. A new version performing SI
measurements will be released prior to the start of pre-production fabrication.) The evidence
indicates that this cable set does indeed represent a controlled set of samples in which the keystone
angle is the only parameter that has been varied.
0.0461 0.3850
---0-- Avg Mid T (in)
--0-- Avg Width
0.0460 0.3845
c c
I- 0.0459 0.3840
.....
.t:::.
"C "C
L 3:
en en
> 0.0458 0.3835 >
00« 00«
0.0457 0.3830
'"
N t') N \0 0'> N \0 0'> N
If) If) N N
0 0 <:( <:( <:( <:( <:( 0.. 0.. 0..
0 0 I I I I I I I I
0 0 ~ ~ ~ ~ ~ ~ ~ ~
I I ({) ({) ({) ({) ({) ({) (f) (f)
0 0 ~ ~ ~ ~ ~ ~ ~ ~
I I
N N
I I
u
(f)
u
(f)
CABLE ID NUMBER
(f) (f)
Figure 1. Average cable midthickness and average cable width plotted for each of the cables produced for
this study.
METALLOGRAPHIC EXAMINATION
Representative micrographs (at a magnification of 50x) of the cables from these experiments are
displayed in Figure 2. Figures 2a and 2b are images of the major and minor edges, respectively, of
the standard 40-mm outer quadrupole cable that meets all aspects of the SSC specification. The
cable has been sectioned near the location of the highest compaction of the strands on the minor
edge. Features to note include the radius present at the corners of the minor edge strands, the
absence of a gap between the strands near the minor edge, and the presence of deformation on the
strands at the major edge. These features are all typical of standard SSC cables. Figures 2c and 2d
are images of the major and minor edges, respectively, of the cables fabricated using an 81-mm
pitch length and the nominal 2.20 keystone angle. This image was selected because it represents the
features that concern magnet designers. Features to note include the sharp corner apparent on one of
the strands at the minor edge of this cable, the same absence of a gap between the strands at the
minor edge, and the near absence of any deformation on the strands near the major edge. The region
near the minor edge, in which the gap between strands is closed, extends into the cable for only a
few strands in the 1.20 keystone angle case. In the 2.20 case, there is essentially no gap between the
strands through approximately half the cable width.
671
Figure 2a. Micrograph of cable major edge Figure 2b. Micrograph of cable minor edge
TST-A 12 taken at a magnification of of 50X. TST-AI2 taken at a magnification of SOX.
Figure 2e. Micrograph of cable major edge Figure 2d Micrograph of cable minor edge
TST-P22 taken at a magnification of 50x. TST-P22 taken at a magnification of 50x.
ELECTRICAL PROPERTIES
The critical current density of each of these cables has been measured at Brookhaven National
Laboratory. Each of the cables was measured at several fields between 5 and 6 T in liquid helium.
The critical current is reported at the specification field of 5.6 T, by interpolation using a linear fit,
and at a temperature of 4.22 K, by correcting for measured changes in ambient pressure above the
sample. The cables were mounted in the holder so that the magnetic field generated by the transport
current flowing in the cable was aligned, thereby adding to the applied field at the location of the
minor edge. 6 Thus the worst-case current degradation within the cables was measured. The cable Ie
has been corrected for the self-field of the cable. The cable degradation is calculated based on
measurements of virgin samples of the wire used in the cable. The wire measurements have not been
corrected for self-fields. Thus the total absence of degradation would manifest itself as a reported
degradation of approximately -4.0%.6
The electrical behavior of the cables as a function of increasing keystone angle is shown in
Figure 3a, and the resulting Ie degradation as a function of increasing keystone angle is shown in
Figure 3b. Cables with the standard pitch length of 73 mm are displayed as open circles; cables with
the lengthened pitch of 81 mm are displayed as closed circles. Of note are the extremely low values
of degradation achieved for the three cable samples fabricated to the latest sse specifications.
A comparison of these data with the 36-strand cables produced using the same batch of wire on
the same cabling machine2 reveals that equivalent values of degradation have been achieved in the
30-strand cables and in the 36-strand cables. This somewhat surprising result, it is believed, is
primarily due to the optimization of the rollers dimensions prior to the start of these runs. In
672
addition, it demonstrates that once sufficient experience is obtained with a particular cable type,
manufactured on a particular machine, it is possible to reduce degradation to similar levels,
independent of the aspect ratio of the cables. This result reinforces the validity of the cable scaling
guidelines described in Ref. 3, and the learning curve effects demonstrated in Ref. 4.
9400 .-...-r-"""""'--'-""-I""""T--..,-.-r.......,
o
073 mm o 73 mm
9200 o .81 mm 6 • 81 mm o
I-
10
vi 9000
o
~4
G>
•
~.
o
N
~ 8800 De 2
u 8600
Ic Spec.
•
0
o
D
0
8400~-L--L-~--~~--~~ -2 D
1.01.2 1.41.6 1.82.02.22.4 1.0 1.2 1.4 1.6 1.82.0 2.22.4
Keystone Angle Keystone Angle

Figure 3a. Critical current measured at 5.6 T Figure 3b. Degradation plotted against cable
and 4.22 K plotted against cable keystone keystone angle for both the 73-mm and 81-mm
angle for both the 73-mm and 81-mm pitch pitch lengths.
lengths.
Cable degradation has often been discussed in terms of the minor edge packing factor (PFt),
which relates the compaction of the strands along the minor edge to the area of the virgin strand. The
expression for the packing factor is given by the formula:
PFt = 2rrDlt, where D is the wire diameter and t is the minor edge
thickness of the cable.
The degradation data plotted against the minor edge packing factor is shown in Figure 4. Of
interest are the significantly smaller values of degradation achieved in this case compared with the
data obtained on similar 30-strand cables. 2
Figure 4. The degradation data plotted against the minor edge packing factor.
673
Upon analysis of the cable data for the cables used for the ASST, it is obvious that a parameter
besides the minor edge packing factor controls the degradation. The complete set of data is
presented elsewhere in this conference proceedings. 2 As shown in that publication, the level of
degradation achieved for a given manufacturer's wire is directly related to the design of the strand
cross section.
Of particular importance to the degradation is the distance the outermost filaments are from the
outer surface of the strand. The closer to the outside of the strand the filaments are, the less copper is
available to absorb the deformation of the strand resulting from the cabling operation. In addition,
the filaments near the outside of the wire are subjected to higher bending and twisting stresses as
they get closer to the outside surface in the strand design. We have parameterized this effect by
using the ratio of the minimum distance the outermost filaments are from the outside surface of the
strand, r, and the diameter of the strand itself, D.
The data presented in this paper and the data of Ref. 2 lead to the deduction that the degradation
vs. PFt is, in reality, a family of curves each having the same general shape. However, the design of
the strand, described through the parameter riD, serves to shift values of degradation to either higher
or lower values of PFt, depending on whether riD is smaller or larger, respectively.
MECHANICAL PROPERTIES
A IO-stack of the test fixture was used to measure the modulus of IO-stack specimens of the
(uninsulated) cables produced in this program. In this fixture, 10 short samples of cables were
loaded in such a way as to alternate the major and minor edges of the cable. This geometry allows
the total sample to remain approximately rectangular in shape. The geometry also simplifies the
tooling required compared to that necessary for the curved geometry that would result in aligning the
narrow edges within the fixture.
All tests reported here were performed on uninsulated cable, with no heat treatments to simulate
any type of coil curing cycle. All of the modulus measurements were performed at room
temperature. All of the reported data were obtained under condition of increasing compressive load.
The following test procedure was utilized in the determination of the modulus of the cable
IO-stacks: 1) The gage bar was inserted into the fixture, and a load was applied equivalent to a
pressure of 6.34 MPa at the load surface. The gage bar was verified to be at approximately the
height of a lightly compressed lO-stack (12.7 mm in this case). 2) The deflection gages were zeroed
at this point, the machine was unloaded, and the gage bar was removed. 3) The lO-stack cable
sample was loaded into the fixture. 4) A stable load, equivalent to 6.34 MPa of pressure, was applied
while recording the deflection gage readout (used later to determine the initial stack height). 5) The
applied load was increased at a rate of 4.448 kN/s to a maximum load equivalent to 102 MPa. The
stress data were collected during the ramp at a rate of 1 data point/so 6) The machine was unloaded,
and the sample removed and archived for later examination. 7) Finally, the gage block was
reinserted, the applied load was increased to 6.34 MPa pressure equivalent, and the deflection gage
zero was checked for comparison to the initial value. This procedure was repeated for each of the
measured samples.
The behavior of the IO-stack moduli as a function of increasing keystone angle is shown in
Figure 5. Several trends can be deduced from the data: First, there is a noticeable decrease in the
IO-stack modulus for keystone angles beyond 1.9° for both pitch lengths examined. Second, the
IO-stack modulus is lower for the cables produced with the longer pitch length for the range of
keystone angles examined. Finally, for both pitch lengths, one can infer from the data that, in
general, the IO-stack modulus is approximately constant up through keystone angles near 1.9° for
both pitch lengths.
Work is continuing in the area of understanding the mechanical properties of the cables as a
function of keystone angle. Particular attention will be focussed on the effects of thermal treatments
that simulate the coil curing process. These results will be published in future work.
674
SUMMARY
This paper demonstrates the capability of expanding the design window available to magnet
designers. Under controlled conditions, a series of cables were successfully produced that provide
the ability to form wedgeless coil cross sections. Through the entire range of keystone angles
examined, electrical and mechanical properties within acceptable limits were achieved. Several
issues of concern when moving into the range of large keystone angles have been identified.
However, based on these results, acceptable quad cables can be produced with angles near 1.9°, and
perhaps beyond.
In determining the properties of these cables we have demonstrated that strand design, in addition
to the minor edge packing factor, plays a major role in determining the level of degradation achieved
in SSC-type cables. The mechanical properties of these cables have been found, thus far, to vary in a
systematic and gradual manner, so that only moderate changes in coil properties are expected as a
result of the changes.
Work is now underway to develop alternate 2D cross sections for the SSC quadrupole magnet
that take advantage of the increased design flexibility available. Cables will be produced during the
coming months for the fabrication of several model magnets using cables with increased keystone
angles. Results of this work will be published.
10000
9500 o 73 mm Cable PitcH
9000 • 81 mm Cable Pitcr,;
8500
ro 8000
a.
I: 7500
7000[ 0 0
• •
(f)
::I 0
6500 Eo
::I
0
•
u
0 6000
I:
5500
5000
1.2 1.4 1.6 1.8 2.0 2.2 2.4
Keystone Angle .
( )
Figure 5. Cable lO-stack modulus plotted against keystone angle for both the 73-mm and 81-mm pitch
lengths.
ACKNOWLEDGMENTS
The authors gratefully acknowledge discussions with J. Waynert and W. Nick from the B&W/S
team, along with the support of the SSC quadrupole product manager, S. Shapiro. The authors are
also grateful for the tremendous effort made by M. Garber of Brookhaven National Laboratory in
providing the cable Ie test results in such a short amount of time.
675
REFERENCES
1. SSC Cable Specification M30-000287, Superconducting Cable for SSC Quadrupole Magnets.
2. D. Christopherson, D.W. Capone II, c.R. Hannaford, S. Graham, and D. Pollock, "Summary of the
Superconducting Cable Produced for the Accelerator System String Test Program."
3. J. Royet and R. M. Scanlan, "Manufacture of Keystoned Flat Superconducting Cables for Use in SSC
Dipoles," IEEE Trans. on Magnetics, MAG-23, 480 (1987).
4. R.M. Scanlan and J. Royet, "Recent Improvements in Superconducting Cable for Accelerator Dipole
Magnets," Preprint.
5. S.A. Gourlay, M. Garber, J. Royet, and R.M. Scanlan, "Degradation Studies of Fennilab Low Beta
Quadrupole Cable," IEEE Trans. on Magnetics, MAG-27, 1815 (1991).
6. M. Garber, A.K. Ghosh, and W. B. Sampson, ''The Effect of Self Field on the Critical Current
Determination of Multifilamentary Superconductors," IEEE Trans. on Magnetics, MAG-25, 1940
(1989).
676
EDDY CURRENT INSPECTION OF SUPERCONDUCTING
CABLE DURING MANUFACTURING
D. A. Bein, J. Zbasnik, and S. Graham

2550 Becldeymeade A venue
Dallas, TX 75237
R. Scanlan

1 Cyclotron Road
Berkeley, CA 94720
INTRODUCTION
The downstream failure of cable during winding, insulating, coil winding, and coil
assembly is a significant issue in magnet production. The impact of these failures is costly
both financially and because of the time needed to recover from them. The current approach
to cabling has been to visually inspect the cable for any gross defects during cabling. To
date this has been effective in finding small defects such as crossovers, which drastically
reduce the mechanical integrity ofthe strand and thus of the cable itself. But because of the
large volume of cable that will be manufactured, an automated flaw detection system that
can inspect the cable and detect these types of defects will be needed. We have recently
done an on-line experiment using an Eddy current system and specialized Eddy current
probes to inspect cable during manufacturing.
We will present the results of our inspection, demonstrating detection of crossovers
and cold welds. This will include a description of the instrumentation, probe drawings and
their setup, and a synopsis of the experiments performed at Lawrence Berkeley Laboratory
(LBL) to obtain these results.
*Operated by the Universities Research Association, Inc .• for the U. S. Department of Energy under Contract
No. DE-AC35-89ER40486.

EXPERIMENTAL OBJECTIVES
Our work had two primary objectives:

1. To evaluate the signal-to-noise ratio of the smartEDDY 3.0 system and
specialized Eddy current probes in situ during a cabling run.
2. To determine the sensitivity of the Eddy current probes-that is, their physical
resolution.
EXPERIMENTAL WORK
The groundwork for this experiment was done by Ron Scanlan, ofLBL, and Yuehda
Krampfner, of Failure Analysis Associates (FaA), where the first probes were designed and
the ability of the smartEDDY 3.0 system to locate defects in the cable was verified. The
present work utilized this smartEDDY 3.0 system and an iterated probe design that FaA
engineered for this application.
The Eddy current system, smartEDDY 3.0 (available from SE Systems, Inc., Hayward,
CA), combines specially designed software programs with internally-mounted, eddy-current
instrument modules to convert any PC-based computer, including 80486 computers, into a
test, measurement, and imaging system. The smartEDDY 3.0 system provides multiple-
frequency, low-noise detection of cracks, corrosion, and other defects in metallic
components as well as measurements of thickness, magnetic permeability, hardness, and
physical dimensions. The system is ideally suited for production-line applications.
50-mm Outer Cable Experiments. The first studies were done on cable driven by a
pair of motorized take-up spools. It was extremely difficult to evaluate the signal-to-noise
ratio of the system and the system's sensitivity because ofrather high background noise. We
feel that the source of this noise was due to a "jerkiness" in the tension applied to the cable
by the take-up spools. The tension in the cable causes the individual strands to lie in a very
uniform geometric position, giving a very periodic structure to the cable. But when the
tension changes, this periodic structure is disturbed. These effects could show up in the
Eddy current signal as large indications. We assume that the transverse conductivity across
the wide face of the cable is very dependent on the compaction of the strands: the lesser the
tension the lesser the compaction and, therefore, the poorer the conductivity from strand to
strand. Thus, an increase in the cable tension causes a major change in the conductivity of
the cable, resulting in a large disturbance in the Eddy current fields and a large change in the
Eddy current signal. The variation of the tension caused the Eddy current signal to be very
noisy and to generate false indications. Since the tension between the Turk's-head and the
cable-measuring machine is quite constant, we decided to conduct another series of
measurements with the Eddy current system mounted on the cabling machine. This also
allowed us to verify that no new noise sources would be generated by the cabling machine
itself.
We modified the probes used in the previous LBL experiment to allow complete
coverage of a single strand. Referring to Figure 1, each probe element was 2.5 mm in
diameter, and there were four elements in each probe. We aligned the probe elements
parallel with the pitch of the strand to improve signal-to-noise ratio within each ofthe probe
holders.
The probe elements were wired in series, then the probes were wired in a differential
fashion.
678
,
0.460
36 STRAND CABLE 'A' \.lIRE IS 0.025'

Figure 1. Eddy current probe diagram.
This bridge configuration allowed us to inspect an entire strand; if there were some
difference in the signal seen by either of the probes, an output occurred across the bridge
indicating some type of defect. This configuration was used for both the top and the bottom
set of probes. The probe holders were made of Delrin, with a shallow recess designed for
the cable to run through. In addition, there were thru-holes for the probes to be mounted in,
with a pair of tapped holes drilled perpendicular to these thru-holes for screws to hold the
probes at a fixed stand-off. We used a probe stand-off distance of 0.1 mm, which gave us an
adequate signal-to-noise ratio but left us with acceptable sensitivity.
In the first attempt to use the inspection system, we placed the probes as shown in
Figure 2, directly behind the caterpillar and before the Cable Measurement Machine.
Figure 2. Probes behind caterpillar.
To provide additional stability, we placed the probes between the Delrin rollers, which
were clamped into position by small pneumatic cylinders that applied around 10 psi to the
cable. Initially we had the probes mounted on the top and bottom of the cable directly facing
each other. However, we had to install the probes in a staggered fashion, as shown in
Figure 3, because we found that the Eddy current fields generated by the probes had deep
enough penetration into the cable to cause interference between their signals.
679
Figure 3. Staggered probe set-up.
We found that both the cable's position and tension oscillated, perhaps due to the
Cable Measurement Machine. The result was a rather noisy signal.
We then moved the transducer banks between the Turk' s-head and the caterpillar. The
tension in this region was very constant--on the order of 200 lb-which resulted in a very
steady Eddy current background signal. The set-up, with staggered probe position, is shown
in Figure 4.
Figure 4. Probe arrangement between Turk' s-head and caterpillar.
The probes were initially arranged this way without the proper mechanical fixturing.
But with the transducer holders "riding" on the oscillating cable, they performed
surprisingly well. The holders were not held by any rigid fixture, but supported only by the
cable itself and constrained to move with any oscillation of the cable. In this orientation we
began to obtain a low-noise signal output, and it became evident that cold welds also could
be detected.
680
In this test of the Eddy current inspection system there were no crossovers to detect,
and we were unable to manufacture any because of the shortage of outer cable. This run,
though, did have numerous cold welds that were readily detected. The second run, described
below, gave us the opportunity to generate some crossovers and to manufacture some cold
welds.
40-mm Outer Cable Experiments. This cable run was for LBL quads, and it had no
cold welds. The geometry of the orientation of the transducers is shown in Figure 5.
0.383
30 STRAND CABLE 'B'

Figure 5. Orientation of Eddy current probes for inspection of 40-mm cable.
Our earlier run showed that the optimum position to locate the transducers was
upstream of the caterpillar, and we built a simple rigid fixture to accomplish this. The
arrangement is shown in Figure 6.
Figure 6. New probe-locating fixture.
Again the probes were mounted in the carriers that were free to "ride" on the cable.
The same stand-off distance was used (0.1 mm), and the probes were wired in the same
fashion as in the first run.
The original positioning of the transducers again had to be modified, as shown in
Figure 7.
681
Figure 7. Final probe location.
As is shown, the 'bottom transducer was situated in front of the mounting bar, and the
top transducer was situated riding against the back of the mounting bar. This seemed to give
us an acceptable signal-to-noise ratio and a uniform background signal. The mechanical
fixturing still seems to be lacking, in that occasionally the cable seems to stick in the
transducer holders, causing an anomalous signal. In this run we were able to detect very
small surface defects, such as scratches, and small galls in the strand of the cable of the
order of the diameter of the individual probes themselves (2.5 mm). The cognizant LBL
personnel present asserted that these were insignificant flaws. This clearly demonstrated the
sensitivity of the probes to even small defects. Further tests are needed to set the detection
threshold levels to prevent generation of false alarms due to these small, allowable defects.
After the cable run was complete, there was a sufficient amount of strand left over to make
some intentionally defective cable.
The cable was made with several cold welds and several crossovers. A typical cross-
over is shown in Figure 8. A typical cold weld is shown in Figure 9.
Figure 8. Typical crossover.
682
Figure 9. Typical cold weld.
These were then inspected, and clear indications were detected using the
smartEDDY 3.0 system and our specialized probes. The output signal traces associated with
the cross-overs and cold welds are shown in Figures 10 and 11, respectively.
I npedancc SETUP .1
0.00
Fi 1 hi rIO (Sec:)
+
O.~O
)(- 10
= ,010
+ +
)(-hi = ,20
~ y-l0
= .0 10
"cz
0 .0
~-hl = ,20
+ RotJltions (.)
~
+ + + +
~
1
-0.40
+ +
=
Al 3~1 . 00
R2 = 108 . 25
~ -0.80 + R3=SO,25
+ .. ,
~
..... ....... ... .. .+ ... .. + - •• •• -+0 ....... .. ......
=
R4 181.25
0.080 RS = 172 . 2~
~
~
Z Frequencies
C 0.040
1) 2 . 00 11Hz
t:
2) 740 kHz
0 .0
~ Drive Le'Jels
+
1) 2.5 Voll
-0.040
2) 2 . S Volt
-0.080 + • +
AutoSlore OFF
~
Tlne (sec . ) a 2 6 e 10 ~
Hain Menu; 1 [n ~ pec t 2 S oP'"e 3 Pa"<Jrteter s
• Replii1tj 5 Relrieue & UtilltieslE:xit
Figure 10. Trace indicating crossover.
CONCLUSION
The Eddy current system and probes proved useful for inspection of the
superconducting cable during manufacturing. The instrument and probes clearly showed
that the detection of crossovers could be made with a signal-to-noise ratio of 4:1. While the
cold weld indications occurred with a signal-to-noise ratio of the order of 2: 1, we need to
improve this ratio for reproducibility and reliability of cold weld detection. The results
suggest that we can implement the crossover detection both because of the satisfactory
signal-to-noise ratio for these defects and because they degrade the mechanical integrity of
the cable. Our future efforts will first entail optimization of the mechanical fixturing,
followed by work on better cold weld detection, and, finally, work on edge detection.
683
IMPedance SETUP IH
0.010 + + + + + + +
Filters (Sec)
+ + + + + + + +
x-Ie = .10
O.OOSO + + + + + + +
x-hi = .0:>0
+ +
IJI
(J 0.0 ....... + ........ y-Ie = .0:>0
Z y-hi = .0:>0
CI: +
Q Rotations (0 )
IJI 0.0050 + + + + + + +
R1= 80.00
Il. + + + + + + + +
t R2 = lOB .25
... -0.010 + + + + + + + + R3= 50.25
....... + .... .. + .. .... + .. .... + .. .. + ..... .+ ....... + . ..... + ........
...Z 0.20 + + + + + + + +
R4 = 181.25
R5 = 172 .25
IJI
I:l + + + + + + + +
Z 0.10 + + + + + + + Frequencies
CI:
I + 1) 2.00 MHz
(J
0.0 ....... + .. 2) 740 kHz
:0(
Drive Levels
+
1)2.B Uolt
-0.10 + + + + +
2) 2.B Volt
+ + + + + + + +
LJ
-0.20 + + + + + + + +
AutaStore OFF
TiMe (sec. ) 0.0 0.5 1.0 1.5 2.0 2.5
Hain Menu: 1 Inspect 2 Store 3 ParaMeters
• Replay 5 Retrieve 6 Ut i 1 it ies/Exit
Cold Weld
Figure 11. Trace indicating a cold weld.
ACKNOWLEDGEMENTS
The authors are grateful to Scott Graham and Hugh Higley for their operation of the
cabling machine at LBL during this experiment, and to Bill Gioroso for his ingenuity in the
design and construction of the transducer holders and for his help in locating the fixtures.
We also acknowledge the groundwork for this experiment that was done by Ron Scanlan of
LBL and by Yuehda Krampfner of Failure Analysis Associates (149 Commonwealth Drive,
Menlo Park, CA 94025).
684
COLD WELD ANALYSIS IN SSC STRAND AND CABLE
J. M. Seuntjens. D. Christopherson. F. Y. Clark. and D. W. Capone II

Dallas. TX 75237
INTRODUCTION
Cold welds on superconducting strands are simple joints formed by pushing two strand
ends together. Cold welding is commonly used between cable runs to avoid having to
rethread cable machines. However. to increase the quantity of finished strand going into
cable. cold welds have been used within cable for High Energy Physics magnet coils.
Although magnets have successfully been built and tested with cold welds. the presence of
the cold weld increases the risk of cable failure during cable fabrication. magnet fabrication.
and possibly magnet operation. A cold weld failure during the cable run is not repairable
and results in the scrapping of a full-coil length of cable. The Superconducting Technology
Group has experienced approximately 6 coil failures due to cold welds during the past two
years in supplying cable for model and Accelerator Systems String Test (ASST) magnets
for the Superconducting Super Collider Laboratory (SSCL). Four of these cold welds failed
during cable manufacture. One cold weld failed during cable insulation. and another during
coil winding. Cold welds that failed within the splicing between coil lengths on the cabling
machine did not result in a loss of cable and were not included. The cold weld failures
within the coils represents a failure rate of 6% of the coils fabricated and approximately 2%
of the total number of cold welds contained.
sse Inner and Outer cable specifications SSC-M30-000014 and SSC-Mag-M-4148.
respectively. currently require coil lengths of cable to be cold weld-free for 80 m on each
end. Cold welds must be a minimum of 35 m from each other. and the cold weld spacing
within the same strand of the cable must be a minimum of 70 m. This removes cold welds
from the pole and the mid-plane turns as well as the coil ramp splices. and separates cold
welds within the wound coil. The cable length for winding of SSC dipole magnet coils are
650 m for Inner and 885 m for Outer. Therefore. the maximum number of cold welds in a
magnet is 14 for Inner and 20 for Outer. Removal of the cold welds increases cable
reliability at a cost in cable yield from strand. The yield penalty is strongly dependent on the
strand piece length (PL).
Cornish I and Garber 3 have shown that cold weld joints are non-superconducting with
about the same resistance as copper. If we take the resistance of the cold weld per unit area
from Cornish. the resistance of a Inner cold weld is 4 x 10-7 n. This value is equivalent to
the resistance of 1.2 mm of copper of the same 0.808-mm-diameter having a RRR of 100.

This paper comprises two parts. Part 1 analyzes cold welds morphologically and
mechanically. Cold welding procedures are described based on welds made at the SSCL on
Inner conductor. The end preparation and the number of repetitions during the welding are
varied. Part 2 analyzes the cable yield from strand vs. the strand PL to help evaluate the cost
of removing cold welds from SSC dipole magnet coil lengths. A model is developed to
express the yield of cable from strand based on an exponential distribution of PLs. Six
batches of strand procured by SSC for the ASST are analyzed for theoretical yield to
support the model.
EXPERIMENTAL-PART 1
A Huestis hand-held cold welder with 0.0318-in. (0.808 mm) dies was used for this
study. IGC Inner strand number 642-1-17 was used for all cold welds. Strand ends were
either pinch-cut with a Crescent 951-4 102-mm side cutter, or scissors-cut with a standard
8-in. (200 mm) office desk scissors. The procedure for cold welding has been reported
previously,1 and the previous authors' description is repeated here:
Figure 1 shows the various stages of making the cold weld joint. In stages 1
through 3, the die inserts grip the two conductors and then press them
together. The die is recessed to accommodate the extruded material. The
presence of oxide and other contamination between the two butted surfaces
usually prevents the formation of a bond at this stage. The dies then open and
re-grip the conductor further back; the process is then repeated (stages 4
through 6). After stage 6, when the contamination has been extruded out of
the joining region, the two pieces are invariably joined. To ensure a good
joint, we repeat stages 4 through 6 three more times (making a total of 5
pressure applications).
Figure 1. Steps in cold-welding process, taken from Reference 1.
Approximately 10 cold welds each were made for 2, 5, and 10 repetitions of the cold
weld cycle for both pinch-cut ends as well as scissors-cut ends. Samples 5S and lOS are
scissors-cut ends with Sand 10 repetitions, respectively. Samples SP and lOP have pinch-
cut ends with 5 and 10 repetitions, respectively. In addition, pinch-cut samples with
686
5 repetitions were made with a single-sided razor blade inserted between the die faces to
provide a 0.2S4-mm spacing that prevents complete die closure (sample 5PP). The travel
distance for the dies during each repetition is 1.96 mm. The spacer was inserted to simulate
an improper weld fabrication caused by operator error or residue flashing in the dies.
The flashing created during cold welding was bent back and forth along an arc having
an axis normal to the axis of the strand. This usually resulted in a break close to the nominal
strand surface. The joints were then sanded smooth with 320 grit (47 J,lm) sand paper. The
sanding was continued only until the cold weld felt smooth to the operator's fingernail.
Cold welds were examined as formed, and after tensile testing, with a JOEL
6100 SEM. Tensile tests were made using an Instron 4302 testing machine with a gauge
length of 250 mm. Custom-made 50.8 mm (2.00 in.) barrel grips were used to grip the
samples using a 300° wrap. Cold welds were placed at the center of the gauge length, and
tensile tests were performed at a cross-head displacement rate of 10 mm per minute.
Figure 2. Light micrographs at approximately 10 x of cold welds with 5 repetitions (left) and 10 repetitions
(right).
RESULTS-PART 1
Figure 2 contains micrographs of cold welds with 5 and 10 repetitions. The flashing
typically flows out in two arms approximately 180· apart. The thickness of the flashing is
typically about 0.25 mm (0.01 in.). The flashing in the SPP samples was notably thicker, as
much as 0.4 mm (0.02 in.). Occasionally, the flashing extruded strongly in one direction.
Figure 3 contains representative images of the cleaned surface of cold welds. After removal
of the flashing, the surface of the strand is smooth with an occasional crater caused by
pullout of the flashing. In some cases, the cold welds remain as much as 0.04 mm
(0.001 in.) larger in diameter than the stand, and in other cases, a slight misalignment of the
two strands remains. Figure 4 shows longitudinal cross sections for the same samples as in
Figure 2. A SP type sample (top) has a similar flow pattern as seen in 5S, lOS, and lOP
sample longitudinal sections. Note that the material from the filaments on the top left side
flow to the bottom flashing rather than the top. This does not appear to indicate a bad weld.
A SPP type sample shown in Figure 4 (bottom) contains a "dead zone" of material on the
top left side.
687
Figure 3. Surface of representative cold welds after the flashing has been removed and the surface smoothed
with emery paper. Craters on the surface (left) and some misalignment of the strand going into the joints
(right) are sometimes seen.
Figure 4. Longitudinal sections of representative cold welds in this study. A 5P sample (top) and a 5PP
sample (bottom).
Cold welds with 2 repetitions were poorly joined. They often failed during the
cleaning of the flashing, and were therefore no longer studied. Stress vs. strain curves in all
cases have very similar curvature up to an abrupt failure of the cold weld. Failure strain was
not studied since the strain over the entire gauge length tells little about the nature of the
cold weld, which has an axial extent along the strand of approximately 1 mm. The failed
cold weld shows little sign of yielding typically seen in copper tensile specimens. Table 1
shows the ultimate engineering stress of cold welds for the 5 different configurations tested.
688
No significant difference in the tensile strength of the cold weld was found between
predominantly one-sided or two-sided flashing. Tensile tests on the strand without cold
welds are included for comparison.
DISCUSSION-PART 1
No significant difference was found between cold welds with scissors- and pinch-cut
end preparation. Five to 10 repetition welds have similar cross sections. Welds with less
than 5 repetitions were much less reliable. Welds with more than 10 repetitions generated
excessive flashing that interfered with the motion of the dies. The cold welds made with a
0.254-mm limit on the die travel (5PP) formed cold welds with flashing that was more
difficult to remove. Longitudinal sections of cold welds from both scissors- and pinch-cut
ends showed similar flow pattern morphology. Many samples show a flow pattern where all
of the filaments from one strand go to one side of the flashing. This does not imply a poor
quality weld. The cold welds made from incomplete travel of the cold weld dies (5PP) have
a dead zone where the filaments have not been extruded out into the flashing. The lack of
extrusion may result in poor bonding due to remaining oxide from the original strand end
surface. This may explain the low tensile strength and large C of V in these welds. Perhaps
the incomplete extrusion is one of the reasons for cold weld failure on the cabling line. A
general precaution of slowly and completely executing each repetition of the cold weld is
recommended.
Tensile data show that cold welds typically have about 75% of the tensile strength of
the strand itself (625-650 MPa vs. 760-820 MPa). Scissors-cut strand ends tend to result in
cold welds with slightly higher tensile strength. The 5PP samples had a lower tensile
strength and a larger variation. The C of V for each type of end preparation and number of
cold weld repetitions, with the exception of the 5PP samples, was similar (within 3%) to the
C of V for the virgin strand. The tensile load on a strand during the cabling operation
typically does not exceed 50 N, which results in a maximum stress level of 100 MPa. All of
the cold welds fabricated had tensile strengths that greatly exceed 100 MPa. When care is
taken to make complete motions of the cold weld dies (avoiding the partial motion as
simulated in the 5PP samples), the tensile strength of the cold weld routinely exceed
600 MPa.
Table 1. Ultimate stress in MPa for different cold weld processes.
Test no. No Cold Weld 5S 5P 5PP lOS lOP
688 651 609

2 640 360 637 646
3 768 639 624 244 667 604
4 767 615 644 531 619 585
5 822 675 647 320 665 617
6 817 676 646 306 657 637
7 829 705 654 548 682 625
8 831 678 649 419 637 675
9 813 641 645 545 656 627
ave. 807 662 644 409 652 625

std. dey. 27 29 9 120 19 26
CofV 0.034 0.044 0.014 0.294 0.029 0.042
6139
EXPERIMENTAL-PART 2
Six batches of strand pieces from the ASST program were analyzed for the yields of
cable without cold welds. The cold weld-free yields are compared for these batches with
average PLs ranging from 1086 m to 7544 m. If breaks in strand processing are distributed
randomly in a production unit, then the distribution of PLs is exponential. From an
exponential distribution of the form e-aX, an expression for the distribution can be derived
when adequate boundary conditions are known. In this study, the maximum PL was taken to
be 100 kg, which was assumed to be a machine capacity limit for final size strand. The
minimum practical PL was assumed to be 1 kg. The theoretical yield of cable from strand
without cold welds can be calculated as a function of the median mass piece length (the
piece length where half of the total delivered strand mass is longer and half is shorter).
RESULTS-PART 2
Six production batches of SSC conductor have been procured for ASST magnets.
Three vendors, Intermagnetics General Corp. (lGC), Oxford Superconducting Technology
(OST), and Supercon, Inc. (SeN) produced both Inner and Outer billets. 2 The number of
billets in each Inner and Outer batch ranged from 3 to 5. In each batch, the number of full
coil lengths obtainable was calculated along with the fractional coil length remainder. In this
work, the required coil length was specified as 675 m for Inner cable and 900 m for Outer
cable. The cable yield for no cold welds was calculated as the number of coil lengths of
675 m for Inner or 900 m for Outer obtainable divided by the total delivered length of
conductor in the batch. Table 2 shows the average PL, the median mass piece length, and
the cable yield without cold welds. In each case, additional coils can be made with cold
welds distributed among several of the coils so as not to violate the cold weld spacing
requirements in anyone coil. The yield, therefore, in the maximum number of cold weld
situations is estimated to be 98% and approximately independent of PL for PLs that exceed
3 coil lengths.
Table 2. Average and median mass piece length and theoretical cold weld-free cable yield
from strand purchased by SSe.
Vendor Cable Type Avg. PL Median Mass Cold Weld Free

(m) PL (m) Cable Yield (%)
OST Inner 1086 1351 67.5

IOC Inner 1587 2215 77.0
SCN Inner 3225 4710 89.4
OST Outer 1556 2300 69.4
IOC Outer 3963 5805 89.5
SCN Outer 7544 12033 94.3
The 100 kg maximum PL was converted to a value M of 26,000 m for Inner strand and
40,000 m for Outer strand. The minimum deliverable PL was taken to be 1 kg or
approximately 300 m. Assuming that the finished strand has the form of an exponential
distribution represented by y = c-ax, the median mass PL can be expressed by the ratio of
integral equal to 0.5, as follows:
690
J L e-ax dx
300
=0.5, (1)
J Me-ax dx
300
where L is the median mass piece length. The integral has a solution of the form
(e-fJ..X)/( -a).
Eq. 1 can be solved by iteration for a when L is fixed and M is defined as above for
Inner and Outer strand. The solution is:
e-aL = 0.5 (e- aM + e-<x300). (2)
Values of a were determined for L values of 1000, 1500, 3000, 6000, and 10,000 m
for Inner and Outer strand. In addition, an L value of 15,000 m was used for Outer strand.
Inner strands cannot have a median mass length L of 15,000 m in an exponential
distribution with a maximum length of 26,000 m. Figure 5 contains a plot of the resulting
equations for Inner strand.
The SSCL cable specifications imply a specific median mass PL. The
specifications state that 85% of the total mass of a production unit must be in lengths
greater than 1500m. Setting up the integral as before:
M e-<XX dx
J 1500
=0.85, (3)
M e-<XX dx
J 300
the solution is:
e-a1500 = 0.15 e-aM + 0.85 e-a300 , (4)
where M is 26,000 m for Inner and 40,000 m for Outer. Eq. 4 can be solved for ex
through iteration. By comparing this ex to the previous a values calculated above,
minimum median mass piece lengths are approximately 5500 m for both Inner and Outer.
.~ 0.4
- - 10000m
~ -_...__.. 6000 m
0.2 3000m
1500m
1000m
0.0
o 9000 18000 27000
Length
Figure 5. Relative PL probability curves for Inner strand as a function of median mass PL.
691
From these curves, the theoretical yield of cold weld-free strand can be determined.
The area under a region of the curve, divided by the total area under the curve, represents
the fraction of the conductor having the PL defined by that region. For example, Inner
strand lengths between 675 m and 1350 m yield 675 m of useful material; lengths between
1350 m and 2025 m yield 1350 m of useful material, and so on. The calculated yield of
Inner and Outer strand in cable as a function of the median mass piece length is shown in
Figure 6. The data from Table 2 are included as discrete data points; they demonstrate very
good agreement with the actual batches of production material.
DISCUSSION-PART 2
The exponential distribution model for the production strand PLs is validated by the
close agreement to the batches of SSC conductor with median mass piece lengths from
1500 m to 12,000 m. The model is helpful for determining the impact of using cold welds in
cable when a cable vendor's median mass strand PL is known.
The current SSC specification implies a median mass piece length of 5500 m . This
represents a cold weld-free yield of 88% for Outer and 92% for Inner cable. If SSCL's
experienced coil failure rate of 6% due to cold welds is typical, then it is clear that cold
welds are not practical for median mass piece lengths greater than approximately 6800 m
for Inner and 10,500 m for Outer strand. These lengths exceed the minimum target levels in
SSCL specifications; however the cable vendors should be strongly aware of the advantages
of long PL from the perspectives of reducing strand production cost and increasing cable
yield.
100
_-----'C)' --- ----

---
= 90
<U
~
<U
c..
#' '" '"
o
'" -- -- ---
"''>'"
Q)
80 '"
, '" '"
.£ I' '"
I
-§ I
(,)
<U
I ----- Outer theoretical
I
<U
~I
70 0 Inner theoretical
I
0 ASST Outer
'" •
I
Q) I
~ I ASST Inner
,
I
'"
'0 60 I
U I
I
I
50
0 5000 10000 15000
Median mass piece length
Figure 6. Theoretical cold weld-free yield of cable from strand as a function of median mass piece length.
Calculations of theoretical yield from ASST material from Table 2 are included as discrete points.
692
ACKNOWLEDGEMENTS
The authors thank D. Washburn and V. Levy for technical assistance in this work. The
SSC Laboratory is operated by the Universities Research Association, Inc., for the
U. S. Department of Energy under Contract number DE-AC35-89ER40486.
REFERENCES
1. D. N. Cornish, D. W. Deis, and J. P. Zabasnik, "Proceedings of the Seventh Symposium on Engineering
Problems of Fusion Research," IEEE Pub. No. 77CHI267-4-NPS, 1977, p. 1266.
2. D. Christopherson, D. W. Capone II, C. R. Hannaford, S. Graham, D. Pollock, and J. M. Seuntjens,
"Summary of the Performance of Strand and Cable Produced for the 1991 Magnet Program,"
llSSC IV, New Orleans, to appear in Supercollider 4, 1992.
3. M. Garber and W. B. Sampson, "Short Sample Results for Cable SIC 374: The Effect of Cold Welds,"
Eighth Niobium-Titanium Workshop, Asilomar, CA, March 15, 1988.
693
RESULTS OF AC LOSS MEASUREMENTS ON SSC CONDUCTORS
G. T. Mallick, Jr.,I J. M. Toms,I W. 1. Carr, Jr.,I

G. Snitchler,2 V. Kovachev 2, and R. Jayakumar 2
lWestinghouse Science and Technology Center

2Magnet Systems Division, SSC Laboratory
Dallas, TX 75237
ABSTRACT
This paper reports experimental results on ac loss measurements made on cables

which may be suitable for use in the high energy booster magnets. Hysteresis losses are
in good agreement with theoretical predictions. Eddy current losses in the individual
strands are considerably less than hysteresis losses, in reasonable agreement with theory.
Eddy currents between strands were negligible for measured samples.
INTRODUCTION
Since the high energy booster magnets which will be used in the superconducting
super collider must be cycled rapidly. through a considerable range in magnetic field, the
time dependent field effects are of interest. Changing magnetic fields induce eddy currents
in any conductors which may be present. The energy dissipated as heat due to the eddy
currents not only impacts the operating margin of the magnets, but also is a consideration
for the refrigeration system. This work was undertaken to help determine the magnitude
of the losses induced in cable which may be used in the construction of REB magnets as
it is subjected to time dependent magnetic fields.
EXPERIMENTAL SETUP
Four samples were measured in our experiments. Two consisted of cables

manufactured with 6 micron diameter superconducting filaments, and two samples were
made with 2.5 micron filaments. Cable parameters are given in Table 1.

Table 1. Sample parameters.
Sample id 1 2 3 4
Wire diam (mm) 0.808 0.808 0.6477 0.808
Filament diam 6.0 6.3 2.5 2.5
(micron)
Cable resistance 0.46 0.18 0.21 0.22
microohm/cm
Number of strands 30 23 36 30
Filament twist 13.5 13 10 8
pitch (mm)
Cable RRR 42 151 125 105
Cu:Sc ratio 1.5 1.25 1.65 1.36
The lay pitch for the cable was approximately 3 inches. Samples were cut to a 15
cm length. They were subsequently .placed in a coil form which contained two saddle
shaped coils, one of which was a dummy coil which permitted the subtraction of the
magnetizing field in the usual way. A small moveable coil was provided to enable a more
precise nulling of the background signal. The length of the sensing coils was
approximately 5 cm, with the objective of being insensitive to end effects caused by finite
FIELD EXCITATION
FIELD SLPf'LY
I-P E£l3OA
MG sn
BUBBLE COUNTII'G M
t>i>PAAATUS ~ D "" ~
DV M
HP 3457A
....... .. .................... ..""'
SPMPL[
~.. ..
~-- ..
..
~
- o
CtNP
RESISTCR
IEEE 488 BUS
CRYOSTAT
COMPUTER HI SPEED
IBM PS2 DATA ACO
KE ITHLEY
80286 COMPUTER BUS
575
'-------'
Figure 1. Block diagram of the experiment.
696
sample length. The absolute sensitivity of the sensing coils was detennined by substituting
a coil of wire wound on a bobbin the same size and shape as the cable sample, and
measuring the response to a known current in the calibrating coil.
The magnetizing field was produced by a dipole magnet which produces a
substantially unifonn field in its four inch bore over a five foot length, providing more than
adequate space for the sample holder and coil fonns. Fields of up to five tesla are
available in this magnet, with dB/dt of up to 250 millitesla/second. The current to the
magnet, up to 5000 amperes, was supplied by two large motor generator sets, connected
in parallel. Excitation to the generators was varied under the control of the data acquisition
computer, providing the proper ramp rates to conduct the experiment. Voltage induced in
the search coil was measured by a HP3457 A digital voltmeter, which communicates with
the data acquisition computer over the HPIB (IEEE-488 bus). Evolved helium gas was
monitored in some experiments to detennine the time dependence of the energy dissipated,
and to provide an independent check on the absolute magnitude of the loss. A block
diagram of the experimental setup is shown in Figure 1.
I[J 40
(f)
(])
-+--'
20
E
Z
0 0
t-;
f-
«
N
t-;
f-
w -20
z
~
«
L
-40
-J -2 -1 0 1 2 J
APPLIED FIELD IN TESLA
Figure 2. Typical hysteresis plot.
Software was written to control the experiment and collect data. The magnetic field
was ramped up and down (bipolar mode) at a constant rate for the experiments. Voltage
induced in the coil was measured, integrated with respect to time, and stored in the
computer. Measurements were made four times per second over the length of the cycle.
A typical hysteresis plot so generated is shown in Figure 2. The tilt of the plot is caused
by an imperfect null, and is easily taken into account. It should be remarked that the area
enclosed by the curve, which is of primary interest, is not affected by making this
correction. Failure of the hysteresis loop to close upon itself for the most part may be
corrected by assuming that a small constant offset was present throughout the cycle. This
typically was on the order of a tenth of a microvolt, and is probably of thennal origin.
Considerable effort was expended on minimizing this constant offset.
697
Bipolar magnetic field ramp data was taken to 3 tesla and at ramp rates from lO
milliteslalsecond to 160 milliteslalsecond. Several measurements were made to 4 tesla,
confirming the observation that the slope of the M-H curve is sufficiently small above 3
tesla in these experiments that results obtained may accurately be extrapolated to higher
fields for comparison to calculations published previously. The integrated voltage data was
then plotted directly against the current in the dipole magnet, and the area of the resulting
curve was measured with a planimeter. This area was then corrected for the area of the
hysteresis loop known to result from the empty sample holder. The value so obtained
represents the energy expended per cycle. The sample could be rotated through 90 degrees
in order to determine the effect of orientation of the sample with respect to the magnetic
field.
150.00.-~~~~,-~~~-,~~~~~~-
W
--.J 2
U
>-
U
" 100.00~~~~~~~~~o~~~~~-+~-
~
L
::J o
U
_ -0- kJ --
h" --- - 0
50.00~~~~~~~~~~~~~e~m~+-~
-s1t
(f)
(f)
o
--.J
0.00~""""hrnn""~"",,,,'+,,~
0.00 50.00 100.00 150.00
RAMP RATE (mT/sec)
Figure 3. Loss vs. ramp rate for 6 micron cable.
EXPERIMENTAL RESULTS AND DISCUSSION
The object of this work was to obtain estimates of the relative magnitude of the
various components of the total energy loss caused by the application of a time dependent
magnetic field to the cable sample. The total energy per cycle per cubic meter measured
for the (nominal) 6 micron samples is presented in Figure 3. No difference was found for
the loss per cycle for any sample when it was rotated 90 degrees, so we conclude that
losses due to interstrand coupling for these samples is small compared to all other losses.
It is noted that these samples are not under pressure, and that pressure may increase the
strand to strand coupling, and the associated loss. The energy lost per cycle, per unit
volume, varies linearly with the ramp rate, as predicted by a simple model wherein the
conductor is fully penetrated by the applied field. The best fit is shown in Figure 3 for
both 6 micron samples. Sample two has approximately three times the eddy current loss
as sample one, and the hysteresis loss is quite comparable for the two samples.
698
The hysteresis loss dominates these measurements. The 2.5 micron sample exhibits
a hysteresis plot that is similar in shape to that of the 6 micron sample, with the
magnetization being smaller in magnitude by a factor of approximately two. Theory would
predict the hysteresis loss to be proportional to the diameter of the filaments, and these
differ by a factor of 6/2.5 = 2.40.
Measurement of the areas of the 2.5 micron sample hysteresis loops proved to be
difficult because of the noisy appearance of the data, and the failure of the loops to close.
While it is tempting to attribute this "noise" to random thermal fluctuations in the input
signal, it must be realized that this noise was totally absent in repeated measurements of
an empty sample holder under the same conditions and sensitivity, and was much less
prominent in measurements of the 6 micron samples. The "noise" was more prominent for
the lower ramp rates. It is rather thought that we are observing the effects of time
dependent currents that are actually flowing in the sample. One possibility for currents of
this type is that conduction occurs from one strand to another across the boundary where
strands touch (particularly likely at the edge of the cable). Contact resistance for this
process would be critically dependent on the area of contact, as well as the condition of
the surface. Neither was controlled in these samples, but physical examination of the cable
samples shows that the intimacy of contact for the 2.5 micron cable is greater as evidenced
by the significantly greater deformation of the strands at crossover points. Resources did
not pennit further investigation of this interesting effect.
The theory of losses in multifilament superconductors being well developedl •2 , it is

possible to calculate the losses expected and compare them with the measured results.
Considering first the hysteresis losses, following Wilson, the hysteresis loss for a complete
cycle may be written:
Using the parameters which Snitchler used for his calculation, (Bo = 1 tesla, Bm =
3 tesla, Jc(B o )=lOlo amperes/meter, and the wire parameters in table 1) we obtain the zero
field intercept of the dashed lines in Figure 3.The functional dependence of the loss upon
the ramp rate may be estimated using the theory developed by Carrl • The eddy current
power loss in a single strand may be written:
The transverse resistivity is related to the matrix resistivity according to the

relationship:
I-A.
0=0 ( - )
.1 m 1+A.
where there is a relatively insulating barrier between the superconducting filament and the
matrix, which is assumed in our case. The matrix conductivity was calculated from the
cable resistance at 100 K. The result was used at 3 tesla with no correction for
magnetoresistance in the matrix. The calculated eddy current loss as a function of ramp
.699
rate is plotted as the dotted lines in Figure 3 for the two samples studied. The
experimental data is not in serious disagreement with the theory. It should be noted that
the expression used to calculate the transverse conductivity is approximate, and the opposite
limit, which neglects the insulating barrier between the superconductor and the matrix,
yields a value five times greater. The difference in the measured rate dependence for the
two samples is most probably due to a difference in the copper resistivity. However,
because of the different wire geometry between samples #1 and #2 (#1 has a solid center,
#2 does not), and the different Cu:Sc ratio, this result should be investigated in more detail.
Early efforts to measure ac loss in superconducting wires were carried out by

measuring the boiloff of gaseous helium from the sample4 • A new variation on this theme
was devised for this experiment. The new calorimeter traps gaseous helium as it evolves
from the sample as it is heated, and channels it through a chimney. The bubbles which
emerge from the chimney are observed through tiny fiberoptic probes which actually
indicate the presence of individual bubbles. While not yet fully developed, this technique
appears to be quite linear over a reasonable range of power inputs, as measured by its
response to known power input provided by a calibration resisto~. One of its very
important attributes is its very fast response time, which permits us to observe the power
dissipated in the sample as a function of time, without the annoying delay and integration
inherent in the previous instruments.
The real time power loss as a function of applied magnetic field is shown in Figure
4 for a rather typical measurement consisting of a superposition of two cycles of magnetic
field (Sample 2, 18.4 millitesla/second). The peak power measured occurs slightly after
the magnetic field passes through zero, in either direction, even though the rate of change
of field is constant during that time. Conspicuously absent is the liberation of heat during
the time of maximum dR/dt, which occurs at the peak current values. These results show
conclusively that the heating of the conductor due to magnetic fields with constant dHldt
will not be constant during the cycle, but rather will peak strongly near zero field,
paralleling the functional dependence of Jc upon H. This is of considerable practical
importance, inasmuch as it tells us that in the REB magnet, most of the heating occurs at
a time when the margin is the greatest.
0.5,-----------~----------_,
CL
W
o==
o. 0 -t-------l~~
.
~
W
>
H
J
f--
«
~ -0.5
W
CL
o J
FIELD C T e s Ie)
Figure 4. Power dissipated in cable sample as a function of field.
700
CONCLUSIONS
For the samples measured, the predominant loss mechanism is certainly hysteresis
loss in the filaments, and is in good agreement with theoretical predictions for the 6 micron
and 2.5 micron filaments. The measured eddy current loss in individual strands is
considerably smaller. Evidentally, interstrand eddy current losses for these samples was
very small. However, in actual magnet operation, there will be forces pressing strands
together in the cable. The lower resistance paths created in this manner will allow
additional interstrand eddy current losses to occur if steps are not taken to insulate one
strand from another.
We have also directly demonstrated that the hysteresis loss is not constant over the
cycle, but peaks strongly near zero magnetic field.
REFERENCES
1. W. 1. Carr, Jr., "AC Loss and Macroscopic Theory of Superconductors,"

Gordon and Breach Science Publishers, New York, NY, 1983.
2. Martin N. Wilson, "Superconducting Magnets," Oxford University Press,

New York, NY 1983, Ch. 8.
3. G. Snitchler, R. Jayakumar, V. Kovachev, and D. Orrell, "Design and A.C. loss

considerations for the 60mm dipole magnet in the high energy booster."
Proc. Third International Symposium on the Super Collider, 1991.
4. M. Ashkin and G. R. Wagner, "Comparison of calculated and measured hysteresis loss

in multifilimentary superconducting wire," JAP 60:2477(1986).
5. G. T. Mallick, Jr., J. M. Toms, and W. J. Carr, Jr., "Design of a sensitive calorimeter

for use in ac loss measurements." Proc. Third International Symposium on the
Super Collider, 1992.
701
TEMPERATURE DEPENDENCE OF THE CRITICAL CURRENT
DENSITY IN SSC-TYPE SUPERCONDUCTING WIRES
Kaname Matsumoto and Yasuzo Tanaka
Yokohama R&D Laboratories

The Furukawa Electric Co.,Ltd.
2-4-3, Okano, Nishi-ku
Yokohama 220, Japan
ABSTRACT
The critical current density of SSC-type superconducting wire as a function of both

field and temperature is measured by transport and magnetization measurements, for
perpendicular field in the ranges of 0 < B < 11 T and 4.2K < T < Tc. The flux pinning
characteristics of both SSC-type multifilament conductor and monofIlament conductor with
high critical current density will be compared.
INTRODUCTION
For understanding the flux pinning properties in superconducting wires, it is neccesary

to clarify the temperature dependence of them. Meingast et a1. 1,2 have already made a de-
tailed study of the flux pinning characteristics in NbTi superconductor. They reported the
lack of temperature scaling of the gloval pinning force ( Fp ) in optimized NbTi with high
critical current density. However, the specimen used in Meingast's experiment was taken
from a monofilament conductor, so that the fabrication process differs from that of a fine
multifilamentary conductor, which is generally used for the winding of magnets. A multifi-
lament conductor is produced by multi-stacking method, and undergoes both multiple hot
extrusion and heavy cold working after the initial solution treatment of NbTi alloy. SSC-
type superconducting wires are also manufactured by a double-hot extrusion process, and
NbTi filamet diameter should be reduced to around 6 /lm for preventing from the effect of
magnetic hysteresis, 3 resulting in the refining the microstructure in NbTi matrix compared

with that of the monofilament conductor, when heat treatments for a-Ti precipitation are
carried out. Therefore, the optimization process of Jc in multifilamentary composites dif-
fers from those in monofilamentary composites and, moreover, the J c properties are sub-
jected to the effect of the extrinsic factors involving the chemical inhomogeneity, the
filament sausaging and so forth. Actually, the J c values in multifilament conductors are
lower than those in monofilament conductors.
To satisfy the specification of Jc in SSC-type superconducting wires, it is required to
attain the high Jc with around 3000 A/mm2 at ST, in production leve1. 3 For development of
such as high Jc in fine multifilamentary wires, we should know the effect of the difference
of the fabrication processes, between a monofilament conductor with high J c and a usual
multifilament conductor, on the scaling parameters of the flux pinning strength. In the
present paper, we present the variations of J c as a function of both magentic field and
temperature in the multifilametary SSC-type superconducting wire, which is selected care-
fully and well-characterized so as to diminish the effect of the extrinsic factors, and we
compare the results with those in the monofilament conductor.
EXPE~ENTALPROCEDURES
The sample used in this study were taken from a multifilametary composite made from
a high-homogeneity Nb-47wt% Ti alloy. The specification of the wire is shown in Table 1.
Table 1. Specifications of the SSC-type multifilamentary

superconducting wire used for the experiments.
Materials Nb-47wt%Ti / Copper

Copper Ratio 1.80
Number of NbTi filaments 41S2
Figure 1 presents the SEM photogragh of the cross section of the wire. The composite
was fabricated by the single stacking method. 4 The heat treatment repetition number, total
time, and temperature for developing the Jc value were three times, about 100 hours, and
chosen in the range from 648K to 683K, respectively. After final heat treatment, the wire
was reduced up to the final strain ( £r) of S.O and the specimens were selected for super-
conductivity measurements. Workability of the wire was superior and the non-occurrence
of the filament sausaging was confirmed by the surface observation of NbTi fllaments by
SEM. According to the analysis of voltage-current characteristics of the specimen, the n
value was approximately 40 or SO at ST.
The gloval pinning force, Fp = Jc x B, was determined in the perpendicular magnetic
fields in the range of 0 < B < lIT by both the transport J c measurements and magnetization
measurements using a vibrating sample magnetometer (VSM). The transportJc measure-
ments were performed at 4.2 K and the Jc values were defined using an electric field criteria
of 0.1 J.1V/cm. VSM measurements were carried out at a variable temperature from 4.2 K to
above Tc. The Bean model was used to obtain the critical current density.
704
50j.UD
Figure 1. SEM photogragh of the cross section of the SSe-type

superconducting wire.
The magnitude of J e in the optimized NbTi superconductor usually increases by the

final strain, and then decreases after attaining the maximum Je value with the further draw-
ing. Figure 2 shows the changes in Je at 4.2 K and the applied fields of both 5T and 8T, as a
4,000
3,500
., , .
.t:s
~
N 3,000
5T # :0
2,500 "'9"
,,
,.,
'<:!"
iii
. 2,000 ,,
!i.
.
,,
....•.
N
,
E ,,
E 1,500 f:I'
.-:-:,.-r-...... .--
, aT
"-<l: 1,000 ,
..,
u
500 i- .,A."'-"·
oL-~~--L-~~--~~
2 2.5 3 3.5 4 4.5 5 5.5
£ f
Figure 2. Variation in the critical current density with final drawing

strain. ( 0 , . present work; ~,A Meingast et al. ).
705
function of the final strain. The data from the monofilamentary conductor by Meingast et
al. are shown together. In the case of multifilamentary conductor, the maximum Je value, at
5T, was obtained at Ef =4.5 and its value was 2920 Nmm2 • At ST, the peak of Je ( 1231
A/mm2) was achieved at Ef = 4.S. The ratio of Je( 5T )/Je( ST ) decreased from 2.59 to
2.34 according with the change from Ec =4.0 to 5.0. This tendency corresponds to the shift
of the peak in the pinning curve versus field to the higher fields as the microstructure was
refined by the wire drawing. While, the maximum Je in monofilamentary composite was
attained at Ef =5.3, and then there was the difference of O.S in Ef between two cases.
Meingast et al. used the heat treatment conditions of both higher temperature ( 693K )
and longer time ( 240 hours) compared with our experiments for optimizing Je in monofi-
lament conductor. This is attributed to the lower driving force for precipitation in their
specimen than that in our one, since a monofilament conductor is received a lower drawing
strain, in its history of cold work, than that of a multifialment conductor. On the contrary,
our specimen has a higher strain owing to the difference of fabrication process so that the
(l-Ti precipitation tends to proceed at the lower temperature than 693K. If we apply the
condition in monofilament conductor to our specimen, the deteriorate of Je may occur

adversely. We can anticipate that the difference of fabrication process between two con-
ductors generates the different microstructure, including (l-Ti precipitates immediately after
final heat treatment, which casuses the differences of both maximum Je values and E f
values for attaining them.
Next, the temperature dependence of the magnetization curve versus field in our
specimens are shown in Fig.3. The width of magnetization hysteresis loop and Be2 , where
B e2 was determined by the field at which the magnetization loop closed, decreased as the
temperature increased. Figure 4 also shows the dependence of Be2 on temperature. The
measured value of dBeidTe at Te of 2.6 -2.7 T/K was close to that in monofilament conduc-
tor. In Fig.5, the Je value vs. field relationship at several temperature, as determined from
0.06
o
....... 0.04 •
::::J
E o
.......
CD 0.02
c:
0
0
~
N
~ -0.02
c: o
~
~ -0.04
o
-0.06
-10 -5 o 5 10
B (T)
Figure 3. Magnetization hysyteresis loop in sse-type superconducting
0 4.2K; • S.7K; A 7.4K;
*
wire with the final drawing strain of 4.5. (
8.3K).
706
14 I-
12 I- ,.0,
10 ,,
,,
I- , ,,
8 ,
C\I 0,
U \
!II
6 \
\
\
\
\
4 • '«,
'b
2 \
,,
0
. . \
\
0 2 4 6 8 10
T ( K)
Figure 4. The upper critical field, determined by extrapolating F p to zero
for data at 4.2K and the closure of magnetization loop for data at 5.7K, 7.4K,
and8.3K.
30,000
0
10,000 0
.00
~ • 00
C\I 3,000 .b.

•
••
00
000 4.2K
• b. •• 000
E
•• • 0
"-
E
1,000 •
••b.
b.
0
0
« •
0
• b.
S.7K
300
u
...., • •b.
7.4K
100 •
8.3K
30
° 2 4
B (T)
6 8 10
Figure 5. Temperature dependence of the critical current density vs. field,

determined by transport]e and magnetization measurements.
707
1r------.~":!'r"_"T"T""-----.....,
* !> v -*
.-.c'J .V 0
.0
6. .a 4 o
o
0.8
*0
• ~ 0
o
.0
o
*
c.
u.. 0.4 r 6.
ff>
* 0
0.2 r 6..
*0
0
I I
O~--~I----~----~--~~--~
o 0.2 0.4 0.6 0.8 1
B/B c2
*
Figure 6. The reduced gloval pinning force vs. the reduced field.
( 0 4.2K; • S.7K; ~ 7.4K; 8.3K).
the magnetization hysteresis loops, are given, and the reduced pinning curves are compared
in Fig.6 at different temperatures. Here, we adopt an expression of the scaling law asS
FP = const x Bc2m( T )bP( I-b )q (1)
where b = B/Bc2' The parameters p and q for the field dependence, obtained so as to get a
comprehensive agreement iIi the whole field range, are summarized in Table 2. The pin-
ning curve function at 4.2 K was very close to b(1-b) and approached bo.s(1-b)l.s at 8.3 K.
This behavior is similar to the data from monofilament conductor. The slope of the line, at
several reduced fields, from plotting log FP vs. log Bc2 gives the temperature scaling expo-
nents (m), which is also summarised in Table 2. The values are roughly in agreement
Table 2. The parameters p and q for the field dependence and the
temperature scaling exponent m.
Temperature ( K ) P q m
b=0.25 0.5 0.75
4.2 0.82 1.12

5.7 0.69 1.25 1.6 1.8 2.0
7.4 0.59 1.43
8.3 0.62 1.42
708
with results in monofilament conductor, though it is required the more detailed study for a
complete comparison between them.
The present results suggest that there is no intrinsic difference between the flux pinning
mechanisms of multifilament and monofilament conductors. Since extrinsic factors, which
suppress the J e properties, in multifilamentary composite are diminished by the careful
preparations, the changes of F against both temperature and field seems to reflect the
changes of pinning mechanism~ directly. The differences of J e between two conductors
may attributed to the small differences in obtained microstructures owing to different me-
chanical and thermal treatments, such as the total amount of cold work and the repetition
number of hot extrusions received, during their fabrication processes. For development of
Je in SSC-type wires, we should make clears the effects of the delicate difference in micro-
structures on the flux pinning.
ACKNOWLEDGMENTS
The authors would like to thank H.Takewaki and K.Susai of the Furukawa Electric
Co. ,Ltd. for their helpful cooperation.
REFERENCES
1. C.Meingast, P.J.Lee, and D.C.Larbalestier, Quantitative description of a high Jc Nb-Ti

superconductor during its final optimization strain: I. Microstructure, T e , He2 , and
resistivity, J.Appl.Phys., 66:5962(1989).
2. C.Meingast, and D.C.Larbalestier, Quantitative description of a high Jc Nb-Ti super
conductor during its final optimization strain: II. Flux pinning mechanism,
J.Appl.Phys., 66:5971(1989).
3. M.lkeda, H.Ii, S.Meguro, and K.Matsumoto, Recent development of sse cable in
Furukawa, 2nd International Industrial Symposium on the Supercollider, Miami(1990).
4. R.Ii, S.Meguro, T.Suzuki, K.Ogawa, and M.Ikeda, Development of the production
technology for fine filament superconducting cable used in accelerator magnets,
3rd International Industrial Symposium on the Supercollider, Huntsville( 1991).
5. A.M.Campbell, and J.E.Evett, Flux vortices and transport current in type II super-
conductors, Adv.Phys.,21: 199(1972).
709
SSCL QUALITY PROGRAM OVERVIEW
Robert V. Hedderick and Dennis C. Threatt
Quality Assurance Office

Dallas, TX 75237-3946
ABSTRACT
The Quality Program for the Superconducting Super Collider Laboratory (SSCL) was
developed for a number of reasons. The need for a quality program not only is a contractual
requirement, but it also makes good economic sense to implement such a program. The
quality program is the device used to coordinate the activities of different Laboratory
organizations, such as Engineering and Procurement, and to improve operational reliability
and safety. To be successful, the QA Program not only must satisfy Department of
Energy (DOE) requirements and provide for flowdown of requirements to performing
organizations, but must also be flexible enough so that the program is tailored to meet the
needs of each internal organization. The keys to success are management support,
acceptance by personnel, and cost effectiveness. These three items are assured by involving
appropriate management at each step of program development, by personnel training and by
feedback, and by programs to reduce defects and improve qUality. Equally valuable is
involvement of key organizations in program development. We will describe the basic SSCL
Quality Program requirements, how the requirements are tailored to the needs of Laboratory
organizations, and how the effectiveness of the program is validated.
INTRODUCTION
The SSC Laboratory Quality Assurance Program was established on a lab-wide basis
in 1990 and was based on requirements passed down from DOE. It started with the goal
defined in the contract with DOE: to successfully create and operate a world-class, high-
energy physics laboratory.
To create an environment for achieving this goal, several major quality objectives were
established:
• Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

• Meet the needs and expectations of our clients, the DOE and Laboratory users
• Infuse a quality attitude at all levels
• Meet the basic requirements of ASME NQA-l1 and DOE orders 5700.6B2 and
4700.1, Section III, Part D3
• Provide Laboratory-wide quality requirements
• Apply appropriate quality requirements to our suppliers and subcontractors
• Ensure the effectiveness of the SSC Laboratory Quality Assurance Program.
These objectives not only provide the basis for a sound and comprehensive quality
program but they also make good business sense in promoting the efficiency and
effectiveness of the entire SSCL organization. The SSCL QA Program is intended to provide
the catalyst for interface between internal organizations. This is accomplished through
integration of project management, technical, and administrative requirements and by
defining the lines of communication. By serving as the facilitator, the QA Office promotes
dialog between organizations with common interests. Continuation of this effort will serve to
increase consistency and to eliminate redundancy between the various functions of the
Laboratory .
The ultimate objective of the SSCL QA Program is to establish confidence in the safety
and operational reliability of the machine. Infusing a quality attitude at all levels within the
Laboratory is intended to ensure the achievement of that objective through pride in
workmanship and attention to detail. The main philosophy of the program is "doing the right
thing right the first time."
KEY ELEMENTS
The first key element in the design of the SSCL QA Program is meeting DOE
contractual requirements, including the guidance contained in DOE Order 57oo.6B. In
addition, the SSCL has committed to use ASME NQA-l basic requirements, which is the
DOE preferred national consensus standard. These documents, incorporated into the SSC
Project Management Plan and the SSCL QA Program Plan, form the basis for a
comprehensive program.
To further contribute to the success of the SSCL, the Laboratory is working to fonn a
partnership with DOE. This working relationship is also being carried out with our
subcontractors. Through the sharing of resources and cooperation between organizations,
the effort to create a world-class laboratory will go much more smoothly and efficiently.
This partnership is essential to attaining and improving quality and meeting schedules.
A second key element of the SSCL QA Program is management support. This support
starts at the top of the organization and filters down to line supervisors. The Director of the
SSC Laboratory has endorsed the QA Program by issuing a Quality Policy that establishes
the authority and responsibility for carrying out the program. Although the SSCL Director
retains ultimate responsibility for the program, the responsibility for preparing,
implementing, and monitoring the program has been delegated to the SSCL General
Manager. The SSCL General Manager has assigned the responsibility for development and
implementation of the program to the SSCL Quality Assurance Officer. Division Heads
endorse the QA Program through concurrence with the Laboratory Quality Assurance
Procedures.
Above all, for us to be successful all managers and supervisors need to believe in and
adopt the QA Program at every level of work within the SSC Laboratory. Additionally, each
employee of the Laboratory must accept quality as his or her responsibility. To achieve this
acceptance, the SSCL QA Office has embarked on an aggressive training program designed
to ensure that each employee understands his/her responsibility and what it means to achieve
712
quality. This effort is part of overall quality communication, which is considered to be a
third key element of the QA Program.
In summary, through partnership with the DOE and subcontractors, management
support, and acceptance by personnel, it is anticipated that the SSC Laboratory can achieve
an optimum blend of quality, cost effectiveness, resource allocation, and timeliness that will
ensure our success well into the next century.
DEFINITION OF QUALITY
To describe the purpose of the SSCL Quality Assurance Program, the QA Office has
adopted two basic definitions of quality: "conformance with requirements" and "customer
satisfaction." In the first definition, we are concerned with accurately describing the
requirements for products and services to ensure complete understanding and compliance.
With the second definition, we are conveying the philosophy that the requirements are based
on what the customer actually wants. In this case the customer is defined as the recipient of
our efforts, whether they be products or services.
The ultimate customers of the SSC Laboratory are DOE and the scientific community.
The customers of SSCL personnel are their supervisors and the co-workers who receive the
output of their work. By adopting the philosophy that we are all customers and suppliers for
someone else's efforts, we are promoting greater teamwork among all Laboratory personnel.
PROGRAM DOCUMENTATION
sse Laboratory Quality Policy
Establishing the QA Program began when SSCL Director Roy Schwitters issued the
SSCL Quality Policy. The policy is a broad-based document that describes the general
philosophy of the Laboratory and the overall responsibilities for achievement and verification
of quality.
sse Laboratory Quality Assurance Program Plan
The QA Program Plan is the primary document that describes program requirements
and responsibilities. This plan was submitted to DOE and subsequently approved. It
expands the responsibilities delineated in the Quality Policy and describes the basis for
program development and implementation at the Laboratory-wide level and at the division
level. The QA Program Plan describes the 18 elements of the QA Program and requires each
division to have a subordinated Quality Implementation Plan.
Divisional Quality Implementation Plans
The Quality Implementation Plan (QIP) describes the division QA Program and tailors
the requirements to the division level. The QIPs identify the standards, practices, and
procedures that implement the QA Program within the division. This approach allows the
flexibility for each division to concentrate on the program elements that are pertinent to
division activities.
Standards, Practices, and Procedures
The SSCL QA Program is implemented through the development and use of

appropriate standards, practices, and procedures. These documents provide for the
713
achievement of quality. It is important that these documents be streamlined, easily
understood, and flexible to meet the varying requirements of work at the Laboratory. Well-
structured procedures can help to ensure that quality is achieved at the greatest level of detail.
QUALITY PROGRAM TRAINING
There is an old adage that says, "It's not what you don't know that gets you in trouble,
it's what you think you know that just ain't so."
An effective Quality Assurance Program requires an effective training program. We can
put a lot of effort into the development of the quality program. We can do all the planning up
front. We can get everyone excited and do effective verifications. But, even with all of this,
if we don't effectively communicate and train our personnel with respect to the requirements,
the methods, the focus, and the "want to," we will surely fail in our mission.
At the SSCL we are promoting an effective QA training program. Within the first week
of employment at the Laboratory, the employee is introduced to the SSCL quality program
during orientation. The new employee is shown a video depicting the importance of quality
to the success of the Super Collider program; the basis and basic content of the quality
program; and support for the Quality Assurance Program by the General Manager. In
addition the employee is introduced to the flow-down of requirements from DOE, the quality
organization, and communication paths for quality matters at the SSCL.
All employes attend a "Quality Assurance Program and Principles" course that stresses
the importance of prevention activities like planning, using Lessons Learned, and training.
The financial impact of exercising the Prevention mode in lieu of the Corrective Action mode
is conveyed to the employee. In addition the course includes an introduction to the use of
Quality Tools for Problem Solving and idea generation. Major requirements contained in the
"SSCL Quality Program Plan" are discussed.
These introductory training sessions are supplemented with SSC program-specific
courses targeted for specific audiences. Examples of planned or implemented training
courses include Quality Auditing at the SSCL, How to be Audited, Quality Procurement
Requirements, Problem Solving Using Quality Tools, Design Quality, Corrective Action,
and Quality Records.
Each of the SSCL divisions is responsible for identifying requirements and conducting
classes to achieve initial proficiency and enhancement of quality skills. These include skills
such as non-destructive examination, inspection and test methods, and others as needed
within the division. A key element to the success of our efforts centers about providing and
receiving effective instruction and training so that we may achieve the SSC objectives.
PROGRAM IMPLEMENT ATION
The primary guidance for program implementation at the Laboratory-wide level is

provided through Laboratory Quality Assurance Procedures (LQAP). These procedures
describe activities affecting quality that are performed Laboratory-wide. Developing
procedures at this level will provide for more consistency throughout the Laboratory and will
minimize redundancy.
To encourage division participation, all Laboratory Quality Assurance Procedures are
distributed for review to each of the Division QA Representatives, and they receive final
concurrence by the Division Heads. Where more specific direction is required at the division
level, implementing procedures are developed using the LQAPs for guidance. By working
together on program development and implementation, the SSC Laboratory can achieve the
most efficient use of resources and greater assurance that the QA Program will be effective.
714
PROGRAM VALIDATION
As previously mentioned, verification activities comprise an important part of the

Laboratory-wide Quality Program. The Laboratory QA Organization performs the
independent verification function required by most major quality standards. The divisions
are responsible for performing verifications within the division.
Program validation activities are focused in a performance-based direction. Our prime
interest is that the Laboratory and subcontractors meet the requirements that we have
committed to DOE. We are interested not only in verifying compliance, but also in the
effectiveness of quality-related activities. Verification efforts are comprised of audits,
surveillances, and management assessments.
Surveillances are primarily conducted within the division and are targeted and narrow
in scope. Audit efforts are aimed at determining that the Laboratory and subcontractors have
programs in place that are effective, meet all requirements, and are effectively implemented.
During audits we strive to evaluate whether the organization is adequately achieving
quality. If not, we try to identify specific areas requiring improvement. Our focus is not to
play "Gotcha," but to produce items and services that will help us all reach our ultimate
objective-assuring that the Superconducting Super Collider is the most successful DOE
project. Our objective is to work with the divisions and subcontractors to make this possible.
FUTURE DIRECTION
The annual Management Assessment is currently underway at the Laboratory to assess

the effectiveness of the Laboratory Quality Program. We are eagerly awaiting the results of
this assessment so that we can address any weak spots in the program. Over the next few
years we expect the Quality Program to mature effectively in a continuous improvement
fashion. The results of the various tools used to validate the program will provide the data
and ideas we need to produce our desired improvement.
REFERENCES
1. American Society of Mechanical Enginecrs NQA-l, "Quality Assurance Program Requirements for
Nuclear Facilities" (1989).
2. U.S. Department of Energy Order DOE 5700.6B, "Quality Assurance" (9-23-86).
3. U.S. Department of Energy Order DOE 4700.1, "Project Management System" (3-6-87).
715
SSCL MAGNET SYSTEMS QUALITY PROGRAM IMPLEMENTATION
FOR LABORATORY AND INDUSTRY
D.G. Warner

Dallas, TX 75237
D.L. Bever

San Diego, CA 92186
ABSTRACT
The development and delivery of reliable and producible magnets for the
Superconducting Super Collider Laboratory (SSCL) require the teamwork of a large and
diverse workforce composed of personnel with backgrounds in laboratory research,
defense, and energy. The SSCL Magnet Quality Program is being implemented with focus
on three definitive objectives: (1) communication of requirements, (2) teamwork, and
(3) verification. Examination of the SSCL Magnet Systems Division's (MSD) current and
planned approach to implementation of the SSCL Magnet Quality Program utilizing these
objectives is discussed.
INTRODUCTION
The implementation of the MSD Quality Program is based on the concepts outlined in
Department of Energy (DOE) Order 5700.6B. This quality assurance (QA) order provides
the policy, objectives, and requirements for magnet programs currently implemented at four
National Laboratories (including the SSCL), one foreign laboratory, five industry
subcontractors, and more than 100 magnet component suppliers.
The MSD QA organization has been established as a line-management organization
reporting directly to the division director. The QA organization is responsible for
implementing DOE Order 5700.6B through the QA Implementation Plan. However, the
MSD QA Plan embraces all departments of the division by establishment of requirements
•Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

teamwork and of verification policy and procedure. Engineering, production, test, business
management, and procurement all play crucial roles in the implementation of Quality
Assurance.
Figure 1 reflects the MSD flowdown of DOE Order 5700.6B to magnet and component
collaborators and suppliers.
SSCLMSDQA
IMPLEMENTATION PLAN
LABORATORY COLLABORATION
TAILORED MOA
MAJOR SUBCONTRACfORS
MIL-Q-9858 AND MIL-STD-1535
SSCL MANUFACfURED MAGNET COMPONENTS

TAILORED WITH MINIMUM OF MIL-I-45208
Figure 1. Requirements are implemented to enhance quality, production, and cost-effectiveness while
complementing existing systems.
QUALITY IMPLEMENTATION ISSUES
DOE Order 5700.6B is the governing document for the implementation of SSCL
Quality program requirements. It is essential that these requirements be disseminated to all
applicable suppliers and to the MSD Production Department. The question is how to
implement 5700.6B requirements that enhance quality, production, and cost-effectiveness
while complementing systems already in place at National Laboratories and at industry
magnet and component suppliers.
Within the National Laboratories, systems exist that facilitate research and development
programs and are not necessarily geared for extended production of the same product.
However, due to a low turnover in personnel (resulting in excellent management and highly
qualified technical personnel) and an established supplier base, research and development
programs are completed with satisfactory results.
Industry has experience in extended production runs and as a result utilize various
systems to control the baseline standard and end product. Industry, especially those f"rrms
with large defense or energy program experience, have an extensive background in a
MIL-Q-9858 Quality Assurance Program environment.
Potential magnet component and tooling suppliers have been found to have a wide
range of experience with systems and process controls, ranging from ASME code
requirements through MIL-I-45208 to no system at all. The challenge is to implement
systems that the supplier understands (that complement the present system), enhance quality
and production, and are also cost-effective.
Experience has shown that it is sometimes easy to implement an extensive list of QA
requirements, but the result is not always cost-effective or productive. The MSD QA
programs are designed and implemented after an analysis of requirements intended by DOE
718
Order 5700.6B. Communication of these tailored requirements through a team environment,
along with effective verification of results, is the desired objective.
LABORATORY APPROACH
The SSCL, in collaboration with Brookhaven National Laboratory (BNL), Fermi

National Accelerator Laboratory (FNAL), and Lawrence Berkeley Laboratory (LBL), has
established programs for the successful development of superconducting magnets. The
implementation of these programs is through Memorandums of Agreements (MOA) that
identify the research and development requirements for a given fiscal year.
The QA programs for this effort are based on an assessment of technical risk and are
developed into a series of work statements that consider applications of DOE Order
5700.6B. The statements are chosen to fill essential needs in the QA program to enhance
quality-yet permit productivity and cost-effectiveness-while complementing existing
institutional QA systems.
Each of the SSCL programs for MSD at BNL, FNAL, and LBL is tailored through the
MOA. The laboratories agree to specific QA requirements for the duration of the MOA. The
laboratories and assigned SSCL personnel perform the tasks that fulfill the requirements.
Common quality tasks at each lab are configuration management, calibration, process
documentation, test, and problem reporting.
MSD has assigned SSCL QA engineers and/or technicians to assist in the
implementation of Quality Assurance program requirements at the laboratories. In all cases
these QA personnel are integrated team members in the laboratory workforce and have
verification authority. Periodic SSCI,. audits are accomplished as added assurance that
established MOA requirements are being properly implemented.
The current industrialization 1 concept, which integrates the major magnet subcontractor
personnel from General Dynamics, Westinghouse, and Babcock and Wilcox at FNAL,
BNL, and LBL, respectively, allows the QA personnel from these major manufacturers to
gain early magnet production experience. This industrialization concept assists in
establishing these manufacturers as suppliers of superconducting magnets. The
industrialization has been implemented in two phases. The first, technology transfer, takes
place when the collaborating laboratories provide instruction to subcontractor personnel in
the form of lectures, seminars, and hands-on instruction on technology transfer magnets.
The second phase allows subcontractors to produce prototype magnets with laboratory
oversight. This concept ensures that subcontractors have acquired the necessary skills and
knowledge to enter into the next manufacturing stage at specific subcontractor sites.
The SSCL also has development and manufacturing responsibility for model and
production type magnets to be built on-site. A quality system to support these internal
programs has been implemented through the MSD Quality Implementation Plan as derived
from DOE Order 5700.6B.
An analysis of potential DOE Order 5700.6B QA flowdown criteria to smaller
suppliers of magnet components was performed in an effort to enhance quality, production,
and cost-effectiveness, while complementing the existing QA systems of these suppliers. A
survey of smaller suppliers identified MIL-I-45208 as a common QA document. Most
suppliers had little or limited knowledge of what was required by NQA-l or ISO series QA
criteria. In fact, more than 80% of all surveyed laboratory suppliers were familiar with or
had implemented QA plans requiring MIL-I-45208. Based on this assessment, MIL-I-45208
has been established as minimum QA criteria for MSD magnet and tooling component
suppliers. However, MSD has established, by exception, provisions to qualify suppliers
based on historical performance and existing QA systems.
719
The development of a qualified supplier base and methods to verify supplier
performance have been elements in the implementation of the MSD QA Program. Procedures
and policy have been developed to implement criteria requiring procurement only from
qualified suppliers listed on the MSD Approved Supplier List (ASL), QA survey of new
suppliers, QA source/receipt inspections, QA review of all requisitions for magnet and
tooling components, a certified QA auditor program, and a supplier performance rating
system to assist in the recognition of quality products and continuous improvement of
substandard products.
Communication of requirements has been a priority with each new procurement.
However, an effort is underway to emphasize development of a team member environment
with all our suppliers. To accomplish this task we are revalidating our procurement process
to provide better up-front communication of requirements with team members and to
establish QA follow-up surveillance at the time of contract award for development of new or
complex products.
INDUSTRY APPROACH
For complex systems, such as the superconducting magnets, the SSCL has chosen to
impose a MIL-Q-9858 requirement. MIL-Q-9858 (titled Quality Program Requirements) is
extensively used and well known to industry, particularly to those major contractors
involved in defense or aerospace business. Within the framework provided by MIL-Q-9858,
General Dynamics has developed a Magnet Quality Program for the collider dipole
magnets (CDM).
The focus of the General Dynamics Magnet Quality Program is prevention of product
defects. To facilitate the prevention of defects in all phases, the Magnet QualitJ' Program
approach has been designed to:
Develop quality-conscious attitudes and disciplines in all personnel working on

superconducting magnet programs.
Establish cost-effective plans and systems to support magnet verification and
assurance activities.
Involve our suppliers as essential members of the team by building long-term
relationships.
• Ensure that all collider dipole magnets satisfy the requirements of the SSCL.
Actively maintaining quality-consciousness among magnet personnel is facilitated by

an on-going training program and continuous management support. The first step is to attract
a very special workforce of individuals who operate as a team and view quality as an
essential part of their job. General Dynamics has utilized a candidate screening process that
assures potential magnet personnel are qualified and dedicated. This screening process
includes personnel interviews as well as assessment center evaluations. It was used to select
the team of technicians and engineers hired to build the Industry Demonstration magnets at
FNAL.
Magnet factory personnel will complete an extensive training program. The first
160 hours is provided at the Hammond Area Vocational School, a few hundred meters from
the General Dynamics facility in Hammond. The curriculum includes superconducting
magnet basics, quality assurance policies and practices, safety, team-based work
organization and decision-making, metrication, cleanliness, statistical process control, and
defect prevention.
720
Following the training at the Vocational School, job offers are tendered to those
candidates believed to have the technical and interpersonal skills to be successful in the
General Dynamics team-based work environment. These individuals then receive additional
training in the specific job function to which they are assigned. Successful completion of this
training is required prior to hands-on factory work. Refresher training is scheduled at
intervals consistent with the goal of assuring that quality consciousness and awareness
remain at a high level throughout the CDM program, and when individual performance
indicates the need.
To facilitate cost-effectiveness, the General Dynamics Magnet Quality Assurance
(MQA) organization has performed a thorough review of program requirements. All quality
requirements are planned in conjunction with all other affected functions, such as
production, engineering, and procurement. The MQA personnel are co-located with the
magnet production and engineering personnel in the program team area. They actively
participate, along with designers, manufacturing engineers, and analysts, on the concurrent
engineering teams charged with the responsibility of developing the final design of the
collider dipole magnet (Figure 2). Quality engineers have participated in all aspects of the
technology orientation sponsored by the National Laboratories and product development.
Through this approach, they have attained first-hand technical knowledge of the program
and are aware of the critical elements.
Engineering Analysis Support

Engineering
Tooling
Concurrent
Test Engineering
Team
SSCL
/ Procurement
Manufacturing
Engineering
Figure 2. Quality Assurance actively participates in concurrent engineering.
The validation and verification process emphasizes process control to assure

performance requirements and to attempt to maximize yield. Automated and computer-aided
quality verification and data collection systems are being developed and implemented where
cost-effective.
General Dynamics requires full and responsible participation by magnet component and
tooling suppliers, and encourages a partnership relationship with them. A number of major
tooling suppliers are currently under contract with the company. These contracts feature
emphasis on supplier communication not only with General Dynamics, but also with other
suppliers who will supply interfacing tooling or hardware.
For superconducting magnet programs, the General Dynamics objective is to assure
that responsibility for the quality of products and tasks rests with each person doing the
work and with supervision/management. Quality is not the sole responsibility of the Quality
Assurance organization; each employee's effort to provide a quality product, whether or not
he directly handles parts, is essential to assure a successful program.
721
SUMMARY
The SSCL Magnet Systems Division has implemented a structured approach-derived
from DOE Order 5700.6B requirements-to assure that magnets will be reliable and
producible. These requirements are communicated within the National Laboratories and to
industry by flowdown through the SSCL MSD QA Implementation Plan.
The qUality program implementation has been tailored to be a cost-effective approach to
achieving program requirements. Appropriate verification has been designed into the quality
program, with emphasis on the development of a team environment between the SSCL
Magnet Systems Division, collaborating National Laboratories, and industry.
REFERENCE
1. T. Bush, "sse Accelerator Superconducting Magnet Industrialization," 1991.
722
TOTAL QUALITY: A PROVEN APPROACH FOR
MAGNET MANUFACTURE
Olarles E. Owen and Karen A Malone
Quality Assurance
Project Control
Westinghouse Magnet Systems Division
Round Rock, TX 78681
INTRODUCTION
The Westinghouse Magnet Systems Division (WMSD) was fonned in late 1990 when the
Superconducting Super Collider Laboratory (SSCL) awarded WMSD the Follower portion of the contract
for manufacturing collider dipole magnets. The Division's small cadre of management start-up personnel
moved into its office and manufacturing facility in Round Rock, Texas in August of 1991. In January
1992, WMSD won a second SSCL contract to design and build the High Energy Booster dipole magnets.
These contracts presented a rare opportunity: the chance to start with a clean slate and to build, from the
bottom up, a whole new product, with all its required manufacturing processes and all its management sys-
tems. This was the opportunity to "do it right the first time."
With these two contracts, doing it right the fllSt time is the only way that WMSD can succeed. By
mid-I994, WMSD will start its delivery of collider dipole magnets for installation in the Super Collider
tunnel. Each of the magnets, from the fllSt delivery through the last, requires a calculated reliability of
99.99954 percent. This is six sigma performance the fllst time with little or no opportunity for trial and er-
ror, false starts, and continuous improvement. In late 1994, the SSCL will award the full production con-
tract for collider dipole magnets to one of the two preproduction (or • combination of both) subcontrac-
tors, based on performance during the preproduction and low rate initial production phases. There will be
no second chance in the quality, reliability, delivery and cost competition for this contract.
The Westinghouse Electric Corporation and its many divisions have the management, engi-
neering, and production systems and practices required to provide competitive products at a profit. Our
Department of Energy customer for the magnets requires many additional systems and controls to better
assure that the contract requirements are met Westinghouse good management and industry practices,
coupled with the contract's control requirements, have been proven to provide a high level of success in
countless commercial and government projects. However, in few of these projects has success been
measured by the quality, reliability, delivery, and cost standards of the dipole magnets. It is clear to
Westinghouse that a traditional business approach, even with the added controls of the government
requirements, will not be enough. Something more is required, and Westinghouse believes that something
is Total Quality.
THE WESTINGHOUSE TOTAL QUALITY MODEL
In the early 1980's, dramatic changes in the global marketplace caused the Westinghouse leader-
ship to create a new management vision for the corporation. While it was still a world leader in many

fields -- energy. defense. and industry -- Westinghouse was becoming concerned about maturing market
conditions and international competition. Its new vision was based on a management strategy independent
of changes in technology or market conditions. It was based upon the principles and practices of Total
Quality.
Total Quality at Westinghouse is defmed as:
"performance leadership in meeting customer requirements by doing

the right things right the frrst time."
To provide high quality products and services. Westinghouse believes that an obsession for excel-
lence and quality must be the core philosophy of management practice. from design to production. to hu-
man resource management. Westinghouse committed significant capital resources to support the quality
efforts of all its divisions and created the frrst corporate Productivity and Quality Center in this country. It
was the people at this Center who provided a universal management model of the Total QUality process.
which division managers could apply to their businesses (Figure 1).
Total Quality Imperatives
The Westinghouse model for Total Quality is built on four imperatives. Although every operation
will have its own unique way of implementing Total Quality. these four imperatives must always be ad-
dressed. They are Customer Orientation. Human Resource Excellence. ProductJProcess Leadership. and
Management Leadership.
Customer Orientation. From the perspective of Total Quality. everything starts with the cus-
tomer and his perception of value. Customer satisfaction requires an organization to provide a value-to-
price ratio which is equal or superior to world-class competition. At the same time. Total Quality per-
formance requires this to be performed with an above-average value-to-cost ratio. to achieve superior fi-
nancial performance.
OTAL QUALITY IMPERATIVES lWELVE CONDITIONS

OF EXCELlENCE
--... -
Figure 1 - Westinghouse Total Quality Triangle
724
These market realities and imancial imperatives establish requirements -- for products and
services, as well as for the processes by which these are supplied. They are the foundation from which
Total Quality standards must be builL This approach requires competing primarily on the basis of com-
parative value. Competitive cost and price are important but are not the driving competitive factors.
Setting world-class requirements, as dictated by the internal or external customer's needs and perceptions,
is at the heart of "doing the right things."
Human Resource Excellence. As in any endeavor, the most important resource to Total Quality
is people. The key ingredients are motivation, training, and education. Dedication to Total Quality re-
quires a significant effort within an organization, directed toward knowledge and skills acquisition to bet-
ter manage the change process. Much of this effort must be done at the local level, rather than the cor-
porate level, and be tailored to local needs.
Produc:tlProc:ess Leadership. Positioning a business to compete on the basis of Total Quality,

with superior value-to-cost ratio, requires achieving Product and Process Leadership.
Products are all the things an organization supplies to a customer, including services. Processes
include everything an organization does to supply the customer with these products and services. Product
and Process Leadership is the focus for operational Quality efforts, and setting requirements for products
and processes is a key strategic issue.
Management Leadership. Line management must exercise leadership in establishing Total

Quality as a way of life. Special emphasis must be given to training, setting objectives, communications,
incentives, and measurement and feedback.
Twelve Conditions of Excellence
Within these four major imperatives are twelve individual criteria that must be considered. These
are called the "12 Conditions of Excellence. "
Customer Orientation. Customer Orientation, which is both a Condition of Excellence and one
of the four Total Quality Imperatives, is the fll'St and foremost concept comprising Total Quality. With an
organization that understands Customer Orientation, there are always opportunities to satisfy the customer.
Participation. Participation means that all employees join in establishing and achieving Total
Quality. While managers must lead the Total Quality process, employees must participate in establishing
measures and goals and in developing functional quality improvement plans.
Development. All employees, including managers, must be adequately training to assure that
they understand, support, and contribute to achieving Total Quality.
Motivation. 11lrough trust, respect, and recognition, employees can be motivated to achieve
Total Quality. Recognition of employee performance may be accomplished in many different ways, but
recognition in any form is an important aspect of motivation.
Products and Services. To achieve success through Total Quality, an organization's products
and services must be appropriately innovative, and must be reviewed, verified, produced, and controlled to
meet the customer's requirements. This is the classic view of quality.
Processes and Procedures. Processes and procedures used to prepare and deliver products and
services must be developed as an integrated, verified, and statistically controlled system.
Information. The clear, accurate, timely, and useful flow of information is an important factor in
a Total Quality environment, and must be integrated with products, services, processes, and procedures.
Suppliers. Suppliers are considered partners who are selected, measured, controlled, and rec-
ognized based on their potential and actual contributions to meetings requirements for Total Quality.
725
Culture. Establishing a Total Quality culture requires that management establish a value system
in which individual and group actions reflect an innovative ''Total Quality Frrst" attitude.
Planning. Total Quality must be recognized as a primary business objective and incorporated
into strategic business and fmancial planning activities.
Communications. Verbal and non-verbal communications should be two-way, clear, consistent,

and forceful. Follow-up communication should be timely, and actions should match words.
Accountability. Performance measures for Total Quality should be developed, posted, analyzed,
and effectively used to chart continuous improvement throughout the organization. Employees should
participate in the development, maintenance, and interpretation of performance measures.
A "BENCHMARK" FOR TOTAL QUALITY SUCCESS
Within the corporation, there has been a steady and, at times, dramatic stream of stories about the
success of individual divisions adopting a Total Quality philosophy, turning their businesses around and
attaining world class performance in their markets, products, and profitability. Three different divisions
have been judged as fmalists in the Malcolm Baldrige National Quality Award in 1989, 1990, and 1991.
The most notable story of success through Total Quality is that of the Westinghouse Commercial Nuclear
Fuel Division (CNFD), which won the Malcolm Baldrige National Quality Award in 1988. In 1991, in
their first year of competition eligibility, CNFD won the George Westinghouse Total Quality Award for
best in the corporation.
As with superconducting magnets for the Super Collider, CNFD's commercial nuclear fuel rods
must meet exacting standards for quality and reliability, to perform at the required levels in nuclear reac-
tors. A typical nuclear power plant has a core consisting of 193 fuel assemblies composed of over 50,000
fuel rods, which contain over 18 million fuel pellets with 700,000 feet of zirconium alloy tubing. Each of
the rods and each of the pellets must perform reliably for up to five years in an environment of extremely
high temperature and pressure. Reliability of these fuel assemblies in the hostile environment is the single
most important factor in measuring customer satisfaction and in profitability of the world-wide com-
petitors in the nuclear fuel business.
Prior to 1982, the CNFD quality program was directed toward meeting the demanding nuclear in-
dustry regulatory requirements. It consisted of sound management practices overlain by copious proce-
dures, inspections, traceability, documentation and audits; that program was an industry model for its day.
In 1983, however, CNFD recognized that fuel reliability could provide its most significant com-
petitive advantage and that something better than past practice would be required to attain that higher reli-
ability. Consequently, the division made the continuous long-term quality improvement strategy of Total
Quality its number one management objective.
In 1983, the fuel reliability (as measured by the number of in-service leaking fuel rods) was
99.95%. While this would be considered excellent for most products, it was only average for the nuclear
fuel business.
In the next seven years, with a quality program that improved every facet of the business, the
CNFD fuel reliability was 99.9995% -- the best in the world Customer satisfaction, market share, and
profitability also improved and, by 1990, were "world class."
From its experience in quality improvement, Westinghouse CNFD has learned a number of
lessons that they have shared with the Westinghouse Magnet Systems Division. As we begin the task of
creating in five short years the systems that will yield 99.99954% reliability magnets, we intend to
capitalize on those lessons.
THE WESTINGHOUSE MAGNET SYSTEMS DIVISION

TOTAL QUALITY MODEL
As Westinghouse CNFD is quick to point out, Total Quality is not a substitute nor a replacement
for those sound systems of management planning and control common to any well-run business. Our
WMSD Total Quality model includes all those fmancial, engineering, manufacturing, program control,
726
quality assurance, cost/schedule, and other systems. as well. These traditional systems form the founda-
tion upon which we will build Total Quality.
A Common Vision
CNFD, in writing about its Total Quality process, says:
'The Key to our success has been the clarity and consistency of our Vision. the
frequency with which that Vision is shared with employees. and the consistency
with which it is put into practice by managers. "1
Shortly after its formation. the WMSD General Manager established a Quality Council to manage
the Total Quality Process. That council is comprised of the same people who manage the Division's day-
to-day business -- the Division Manager and his staff. It is the Council, rather than anyone person. who is
in charge of directing the course of Total Quality within WMSD.
The Quality Council has prepared the following statements that reflect its vision of the division.
• VISION - For the Magnet Systems Division to become a great enterprise; to become known as the
world leader in the design and manufacture of superconducting components and systems.
• GOAL - Our greatness will come from our demonstration of excellence in everything we do. We will
conduct ourselves with openness. integrity, and respect for the individual. our customers. our sup-
pliers. the corporation. and society.
• QUAliTY POUCY - To perform in complete accordance with customer and Westinghouse require-
ments.
As WMSD begins to build an organization that will achieve some astounding goals in the next
few years. this Vision statement is being communicated throughout the workplace and reinforced with all
employees in a number of ways. It is being communicated through posters and bulletin boards and in our
the annual Quality Plan; it is the core of Total Quality Workshop training that is mandatory for all em-
ployees; and it is the topic for discussion at regularly held all-employee meetings.
As WMSD continues the Total Quality process we expect that each and every employee will be
able to clearly and consistently state who we are. where we are going. and how we are going to get there.
Total Quality Framework
A second lesson leamed from Westinghouse CNFD is that a framework for Total Quality is
absolutely essential. This is the model that keeps everyone focused on higher standards of performance
and continuous improvement.
The WMSD 1992 Quality Plan is now being structured by the Quality Council for March distri-
bution to all employees. With this document. the Council will start building the Total Quality framework.
The Plan repeats and discusses the Vision and builds on the corporate definition of Total Quality to link
the Division's plans and objective to the Total Quality definition.
The Quality Council has identified ten Key Issues for 1992 in the Quality Plan. One or more of
the Staff managers has assumed the sponsorship and accountability for each of the Key Issues. The staff
managers will prepare their own brief Quality Plans to focus their people and activities toward the divi-
sion's common issues. We expect that individuals and teams will become directly involved by establishing
their own objectives and measures that support achievement of the division's Key Issues.
WMSD is firmly convinced that its Total Quality approach to the Super Collider magnet business
is a competitive advantage -- one that we intend to exploit to its fullest. Since many of the Key Issues in
the 1992 Quality Plan address the competitive issue from the view of both our strengths and weaknesses.
those Issues will not be discussed in this paper.
It is obvious. however. that a theme of Total Quality in any newly formed organization must be
"definition" and "awareness." The WMSD Vision and the 1992 Quality Plan provide the broad definition
and the vehicles for creating awareness. However. these definitions need refmement and detail to serve as
the specifications for "the right things right the fust time."
By a recent count. CNFD had in place 3599 Division policies and procedures. 2 While we do not
intend to benchmark: on this number. in 1992 we do intend to create that comprehensive set of procedures
727
required to explain what is expected and required on all tasks. We intend to take a Total Quality approach
to developing these procedures, an approach that will assure that we get "the right procedures right the
fIrst time."
We are developing standard formats and procedure dictionaries. We are training the procedure
authors to write with user-friendly language. In most cases, we require that the user actually be the author.
We are using teams with management involvement to write and critically review the policy and procedure
content and language.
By taking a quality approach to the issue of procedures we can prevent them from becoming the
administrative burden that exists in many established organization. Instead, we make them the defInition
and communication of a Total Quality approach to each of our jobs.
Measures
It is a truism that we pay attention to those things that are measured. It is fundamental that we can
only measure the things we can defIne; control only the things we can measure; and improve only the
things we control. The third Total Quality lesson to be leamed from CNFD is to measure, measure,
measure.
The Westinghouse Magnet Systems Division is in the defInition phase of developing both its or-
ganization and it Total Quality model. The first of the top level measures is being developed and will be
On-time Software and Error-free Software. At this time, these will be two of the important parameters by
which our customer and our management can judge our performance and measure "customer satisfaction."
As each of our departments, sections, and groups establish its own part of the Total Quality
process, it will develop its own objectives and measures in concert with the Division's goals and objectives
for performance leadership and right things right.
Involvement
Westinghouse CNFD discovered in the early 1980's, as did every other organization embarking on
a Total Quality journey, that no Total Quality or quality improvement program can be successful unless
management, employees, and customers are all intimately involved in the process. Four of the
Westinghouse 12 Conditions of Excellence -- Customer Orientation, Participation, Development, and
Accountability - address the involvement question. These four conditions lead to Ownership, which
Westinghouse CNFD describes in their 1991 Quality Plan as follows:
"We believe ownership is taking a personal responsibility for our jobs ... for as-
suring that we meet or exceed our customers' standards and our own. We believe
ownership is a state of mind and heart that is characterized by a personal and emo-
tional commitment to approach every decision and task with the confIdence and
leadership of an owner."3
In our WMSD Total Quality model, we begin to address the ultimate Ownership issue through the
development of our human resource -- our employees. We select all employees, including managers,
based on their proven ability and willingness to work in a Total Quality culture. Can they and will they
accept responsibility? Can they and will they work effectively as a team member? Can they and will they
participate to the point of sharing the ideas?
Second, as our Management LeaderShip responsibility, we provide the training and further devel-
opment of all employees. This begins with a Total Quality Workshop where new employees are encour-
aged to explore the Division Vision, Mission, Policies, and Quality Plan. In small group sessions, these
new employees are exposed to a few of the basic participative and team building approaches to solving
problems.
In subsequent training sessions, all employees are given training in Intemal Customer Process,
Quality Improvement Teams, and, as appropriate, Problem Solving and Decision Making, and Statistical
Methods for problem solving.
728
Commitment
Most companies have found that, while they can quickly advertise "Total Quality" and "Quality is
lob One" in the national media, they can only achieve Total Quality by a long-term commitment to con-
tinuous improvement Total Quality is not a short-term proposition. Westinghouse CNFD, with a twenty-
year running start, took seven more years to attain the National Quality Award level. CNFD continues
their improvement process today.
In contrast to CNFD, the Westinghouse Magnet Systems Division must achieve National Award
levels of performance in just five short years from the time the Division was established. By working
within the Westinghouse corporate Total Quality framework, capitalizing on the lessons learned from
CNFD and other Westinghouse divisions, and with the long-term commitment of our management, we will
meet that challenge.
You can be sure that WMSD will begin to deliver 99.9995% reliability magnets in 1995. We plan
to realize our vision and goal of world leadership in superconducting components and systems through
consistent and non-stop dedication to Total Quality in every aspect of our operations. When we succeed,
we will be a strong contender for the 1996 Malcolm Baldrige National Quality Award.
ACKNOWLEDGEMENTS
Thanks go to Mead D'Amore, Fred Kramer, and Michele DeWitt, Westinghouse CNFD, who provided
input on the success factors of their Total Quality Program.
REFERENCES
1. Westinghouse Commercial Nuclear Fuels Division 1991 George Westinghouse Total Quality Award
Application.
2. George Westinghouse Total Quality Awards Application, 1991 Guidelines.
3. Westinghouse Power Generation Business Unit 1991 Total Quality Improvement Plan.
729
DC DIELECTRIC BREAKDOWN TESTS OF
LIQUID HELIUM AT TEMPERATURES 1.8K TO 4.2K
J. L. Wu and J. F. Roach
Westinghouse Science & Technology Center

INTRODUCTION
Dielectric properties of liquid helium are essential infonnation for designing superconducting
equipment such as the coils for the Superconducting Magnetic Energy Storage (SMES) or the
superconducting magnets for SSC. Data of breakdowns in liquid helium and flashover along the
surface of insulators under the operating conditions are required. For SMES coil, which is designed
nominally to operate in superfluid liquid helium at temperature of 1.8K, breakdown data in the
superfluid state are also required.
Dielectric breakdown phenomena of liquid helium have been investigated for man?)
decades. 1- 14 Gerhold has reviewed the results of many of these investigations in a recent paper.
Among these investigations, several required special mention due to their relevancy to the present
work. Blank and Edwards lO investigated dielectric breakdown of liquid helium at temperatures
from 1.2 to 4.2K and under saturated vapor pressure. Several electrode configurations were used
including sphere-to-sphere, plane-to-plane and point-to-plane. Their finding indicates that the
breakdown voltage generally is independent of temperature for small electrode spacing (0.015 mm)
but decreases with decreasing temperature for larger gaps (0.5 and 1 mm). These findings were
partially collaborated by the investigation of Goldschrartz and Blaisse 11 who obtained liquid
helium dielectric breakdown data at temperatures of 1.32 to 4.2K but for smaller gaps of20 to 100
~m. Their results show that, for electrode spacing of 40 to 80 ~m, the breakdown field decreases
when the temperature was decreasing through the A-point transition, i.e., the transition from normal
fluid state to superfluid state, of about 2.17K. The investigation by Yoshino et al., 12 also showed,
with some scattering, a trend of decreasing breakdown voltage when the liquid helium temperature
decresed from 4.2K to just below 2K. This trend occurred for tests in both uniform and highly non-
uniform field electrode configurations, with 50 ~m electrode spacing and using a pulsed applied
voltage. However, the investigation by Hwang,13 for helium temperature of 1.4 to 4.2K and
electrode spacing of 0.155 mm, did not show a decrease but rather an increase of breakdown field as
the liquid helium temperature decreases through the A-point. In the investiNation of dielectric
tracking on the surface of G-1 0 insulators in liquid helium, Migliori et al., 1 also recorded the
influence of helium temperatures near the A transition region and under the saturated vapor
pressure. The results show that, for small tracking distances of 0.135 and 0.231 mm, the breakdown
fields differ very little between tests at 3.9K and at 2.1K. At larger tracking distances, 0.366 and
0.475 mm, the breakdown field decreases with decreasing temperature.
Even though the database for dielectric breakdown in liquid helium are substantial, the data
were generally obtained in different, specified and unspecified, laboratory conditions, resulting in
discrepancy among the data. In addition, dielectric data of both breakdown across electrode gap
and flashover in the superfluid state liquid helium are still very limited. There is a definite need for
these data as more usages are being envisioned for the superfluid in systems such as SMES.
Therefore, as part of the SMES project and part of a corporate-funded investigation program, we
have perfonned a series of dielectric testing of liquid helium in both nonnal and superfluid states.
The tests include gap breakdown and flashover under essentially unifonn field, non-uniform field
flashover and gap breakdown for a large electrode surface area electrode configuration. These data,
which were used in the conceptual design of a SMES system, are reported here to facilitate in the
dielectric design of superconducting equipment including superconducting magnets for SSe.
EXPERIMENTAL SET-UP
The dielectric testing was conducted with the test electrodes immersed in liquid helium
maintained at approximately constant pressure of one atmosphere but with temperature varied
through cooling by evaporation ofliquid helium contained in a separated chamber. This is achieved
with a test vessel which allows control of pressure in separate regions within the dewar. All the
liquid helium used in testing is commercial grade and no additional filtering process was used to
purify the liquid helium.
Test Apparatus
Fig. I shows the simplified sketch of the test apparatus. It is a double vessel arrangement. A
stainless steel tube enclosure (20.3 cm O.D., 302 cm long) containing the test chamber is placed
within a 30.5 em I.D. stainless steel dewar. Liquid helium can be introduced into the test volume
and the dewar through separate helium transfer tube and maintained at different pressures. Pumping
of the liquid helium contained within the dewar will bring down its liquid helium temperature
which, in tum, cools the liquid helium in the test chamber within the stainless tube enclosure.
Within the stainless steel tube, a filament wound fiberglass tube was hung and supported from
the Lexan cover end plate. The fiberglass tube provides electrical insulation and support for the
grounded electrode and the Lexan insulating guide which centers the high voltage electrode. At the
lower end, the tube was machined to provide large openings to minimize the distortion of electrical
field distribution in the test region.
The test voltage is applied to the electrode through a 2.54 cm O.D. thin wall stainless steel
tube. This tube is attached to a micrometer assembly which provides a precise control of the
position, with an accurcy better than 5.08 ~m of the high voltage electrode.
Electrodes and Insulating Spacers
Three electrode configurations were used in the testing: plane-plane unifonn field with and
without G-lOCR insulating spacer, pin-plane non-unifonn field with G-lOCR insulating spacers,
and a coaxial cylindrical large surface area electrode configuration. These configurations are shown
in Fig. 2. All electrodes were made with aluminum 6061 material and the electrode surfaces were
polished with fine grain silicon carbide sand paper and cleaned with acetone before testing.
Plane-Plane Electrodes (Fig. 2a). The upper (high voltage) electrode has a flat disk
geometry with a diameter of 7.62 cm. The edge of the disk is machined to a 0.64 cm radius to
minimize the local electrical field stress. Similar geometry is maintained on the top portion of the
lower (grounded) electrode so as to produce a fairly unifonn electrical field in the central portion of
the inter-electrode region.
Pin-Plane Electrodes (Fig. 2b). In this configuration, the same lower electrode as in the
plane-plane geometry was used. The upper electrode was replaced with a pin electrode of 3.3 mm
in diameter and with its lower end squarely cut off. The centerline of the pin is located off-axis
from the centerline of the high voltage stainless steel tube. This pennits moving of the contact point
at which the pin presses against the insulator spacer by merely rotating the stainless steel tube. This
arrangement provides an easy way to vary the creepage flashover distance.
732
L· · I--- - - - - -It' gh V o ltage r en ninal
I
ij ~IU-L
~ ~ ( S . !> . Tub)
""~",, "'~""
I· = lIelium T,·ans fcr Tube Po rt
_1~"L~lIIr~over Plate (Lexan)
Tube
(Lexan)
IS.2cm
1-_-+--+l-30.5cmrt-t---lt-t-_'i
274 cm
Insulating Guide Support

(Lexan)
igh Vol tage El ec trode

(Aluminum)
-II'>--II--+-6round El ectrod e
(Aluminum)
I nner Contour
of Dewar
Spring
Figure 1. Simplified drawing showing the liquid helium dielectric test apparatus.
--..1 12.65 em 0
I
7.50 em 0
High Vohage
Electrode
High VoIIage
Electrode Grounded
Electrode
Insulating
$uppcn
1_) Plane· Plane Electrodes (b) Pin· Plane Electrodes (e) Coaxial Cyl,nder Electrodes
Figure 2. Electrode configurations.
733
Coaxial Cylinder Electrodes (Fig. 2c). The high voltage inner cylinder has an outside
diameter of 12.14 cm and a length of 20.32 cm. Both ends were machined to a large radius (1.27
cm) to reduce the electrical stress. Enclosing this inner cylindrical electrode is a grounded outer
cylindrical electrode of 12.65 cm in inside diameter and 26.9 cm long. Both the inner and outer
electrodes were supported by a rigid insulating structure and were assembled and positioning
calibrated before being attached to the lower end of the fiberglass tube. This ensures equal
electrode spacing all along the circumference. The configuration produces an essentially unifonn
electrical stress over an annular region of 12.39 cm in diameter, 2.54 mm in spacing, and 17.78 cm.
long. The stressed electrode surface areas are 678 cm 2 for the inner electrode and 692 cm 2 for the
outer electrode.
Insulating Spacers. The insulating material samples used in surface flashover tests were
fabricated from G IO-CR material. It is a thin circular disk of 2.54 cm in diameter and 1 mm thick.
The samples were machined in such a way that the lamination layer is oriented parallel to the disk
surface.
Instrumentation and Test Voltage Supply
The pressures in the test volume and in the dewar were monitored by two Wallace & Tieman
pressure gauges. Temperatures at six locations, including three in the test electrode region, were
monitored by Lake Shore DT-470 Diode Temperature Sensors. In addition, liquid helium level
sensors were used to monitor the liquid helium levels in the test electrode region and in the dewar.
This ensured that proper liquid helium levels were maintained during testing. A Hipotronic Model
HD 100 Hi-Pot Tester was used to supply a negative dc voltage to the high voltage electrode. Pre-
breakdown discharge currents, as indicated by the Hi-Pot Tester, were recorded.
TEST RESULT AND DISCUSSION
The test parameters used in the dielectric testing are summarized in Table 1. Typically the
temperatures were varied between 1.8 or 1.9K, Le., below the A transition point, to 4.2K, with the
pressure maintained at near 1 atmosphere pressure. The electrode spacings for the gap breakdown
test under unifonn field condition (plane-plane geometry) were varied between 0.254 mm to 1.278
mm at an increment of 0.254 mm. Other tests have fixed electrode spacing as shown. It is to be
noted that for the non-unifonn pin-plane geometry flashover test, the minimum creepage distance
for surface dischargeis the electrode spacing shown plus the additional horizontal distance from the
pin to the edge of the spacer. For the tests conducted, the minimum horizontal creepage distance
was from 0 to 10.6 mm. The test results are summarized and discussed separately below.
Table 1. Summary of Test Parameters
Electrode Test Temperature Pressure Electrode

Configuration K Torr Spacing
mm
Plane-Plane Gap Breakdown 1.93-4.18 684-777 0.254-1.27

Flashover 1.93-4.24 728-783 1.00
Pin-Plane Flashover 1.87-4.20 717-822 1.00
Coaxial Cylinders Gap Breakdown 1.80-4.2 727-762 2.54
734
Plane-Plane Electrode Configuration
The dielectric test results for the plane-plane electrode geometry are plotted in Figs. 3 and 4
for gap breakdown between electrodes, and in Fig. 5 for surface flashover. It was observed that,
more often than not, appreciable pre-breakdown discharge currents of several tens to about one
hundred microamperes were present for times up to a few seconds before voltage collapse. In some
cases sustained discharge currents of a few hundred microamperes to a few milliamperes were
detected without an immediate voltage collapse. Under these conditions, the power supply was
turned off and the voltage was registered as a breakdown voltage.
Gap Breakdown. Fig. 3 shows the dc breakdown voltage of liquid helium at temperature of
1.93 ± 0.02K and pressure of 684 to 708 Torr, plotted with respect to electrode spacings. At least
two tests were conducted for each electrode spacing. As expected, the data shows a general trend of
increasing breakdown voltage with increasing electrode spacing. The breakdown voltage increases
slightly less than linear with the electrode spacing. Substantial variations from test to test for a
given spacing are also observed. For general design guide, a breakdown voltage of about 10 k V for
an electrode spacing of I mm can be assumed.
Dielectric breakdown data between electrodes in liquid helium at different temperatures are
shown as Fig. 4. These data cover tests for electrode spacing of 0.254 to 1.27 mm and temperature
of between 1.93 to 4.l8K. Substantial fluctuations in data are clearly evident. Dashed lines were
added to indicate the general trend of the data. Increase of breakdown voltage with increasing
electrode spacing are generally evident at all temperatures. However, the trend for voltage
dependence on temperature is less clear. Generally speaking, the breakdown voltage seems to
increase for tests with 0.254 mm electrode spacing but decrease for larger electrode spacings as the
temperature decreases through the A-transition of around 2.16K. These trends seem to be consistent
with the results of Blank and Edwards 10 for the larger spacing tests and with Hwang 13 for the
smaller spacing tests.
Symbol Electrode Spacing, mm

Q
0.254
18 0.508
1. 016
1.27
16
o
___ _
_....
o
_6-')-~---
0
14 1/ l:). 0 0 CI £>
@
I
14
J
I
o
...... 0-0 - - -
0_- 0
_ _0 - - 8
/ Q
12 j!
..-
..-~
@
;'
"" ......®
,- -----_....----:
/
/
/
/
/
/----; :
<
/
/
u
ci
o
0/
/
liquid He /g ! __ -o Q
/
D. C. Breakdown Strength ,,~
.... - - 6 - -a-.a- -..-
8
-
( Negative to H. V. Electrodel
T =1. 93 ± 0.02 K
P = 684 - 708 Torr
OL--L__~~__-L__~~__~__~~ OL-__l-I__~___ L_ _~_ _ _ _L_~
o 0.20.40.60.8 1.01.2 1.41.61.8 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Electrode Spaci ng, mm liquid Helium Temperature K
Figure 3. Uniform field dc dielectric Figure 4. Uniform field dc dielectric

breakdown of liquid helium at T=1.93K. breakdown voltage vs. temperature.
735
110r--r--r--r--r--r-;,-,-_,--,--,
I I I I
P = 732-733Torr
12.0 - -
~T=4.24K
11.0 r- -
IO. 0 I I I I I _L I
r4
'---'---~-'---~-'---'----'---'---'---'
,)---r--1
22
110r-'--'--~-r-'--.-~~~~~ 20
18
12. 0
11.0
P = nS-7S2Torr
S; 16
~'14
I~ 12 "/
/.. / ---I-- ---- 1 --
10.0 T =l48K
c::
9.0 L...--l.-_ _L..--l.-_ _L..--L-_ _' - - - ' - - - '_ _....L----I 110
~ 8 Symbol Temp. K Pressure, Torr
'"
u
0
! 4.2 817-822
4
I 1.86-1.81 717-711
o 2 4 6 10
Horizontal Creepage Distance, mm
Figure 6. DC flashover breakdown voltage

for pin-plane electrode configuration.
22
20
I
18
110r--r-,r-~-.--._-.--r--r_,--, 1I
12.0
Electrode Spacing 2.54mm
ll0 Electrode Surface Area 685 cm 2
P = 728 -734 Torr Pressure 721-762 Torr
4
10.0 T=1.93K
9. 0 L-.....J..---lI..-....I---1__..L.--.L.__L...-...J....--L---I OL-__ ~ __ ~ ____ ~ __- L_ _ ~ __ ~
1 2 34567 S 9 10 1.5 2.0 2.5 10 15 4.0 4.5

Test Number Liquid Helium Temp., K
Figure 5. Uniform fioeld de flashover Figure 7. DC dielectric breakdown voltage for

voltage of 1 rnm thick G-IOCR spacer in coaxial cylinder electrical configuration.
liquid heliwn.
Flashover. Fig. 5 shows the test results of dc flashover of I mm thick G lO-CR spacer in
liquid helium of different temperatures, Ten tests were perfonned at each temperature level from
1.93K to 3.48K and five tests at 4.24K. Except for tests at temperature of 3.19 and 4.24K,
substantial variation from test to test were observed. At the lowest temperature (below the A-
transition) the surface flashover for the spacer is between 9.3 to 12.2 kV. This voltage level is
essentially equal to the level obtained for breakdown between electrodes of the same spacing as
736
shown in Fig. 3. This indicates that the presence of the insulating spacer does not reduce the
negative dc dielectric withstand capability of the electrode pair in the liquid helium.
From Fig. 5, we also observed that the variation of flashover voltages with temperature can be
masked by the shot to shot variation. A clear trend of flashover voltage with respect to temperature
can not be determined from these data. For design purposes, however, a minimum negative dc
flashover voltage of 9 kV for 1 mm spacer in a uniform field can be used as general design guide for
liquid helium at temperatures of from 1.93 to 4.24K.
Pin-Plane Electrode Configuration
The results of flashover breakdown of the highly non-uniform field pin-plane electrode
configuration, plotted with respect to the horizontal creepage distance on the upper surface of the
insulating spacer, are shown in Fig. 6 for tests for helium temperatures at 4.2K and at 1.86 to 1.87K.
In the figure the averaged breakdown voltages and the range of deviation from the averaged value
are given. Dashed lines were drawn to indicate the trend of data. The numbers of tests at each
given condition were from two to eight, and the values of breakdown voltage ranged from 10.6 kV
to 21.6 kV under different test conditions.
Similar to the flashover tests in the plane-plane electrode configuration case, the data in Fig. 6
show substantial scattering at a given test condition. This effect was observed in the data obtained
at other temperature as well. The results indicate lower breakdown voltages for 1.87K than for 4.2K
except for the case of zero horizontal creepage distance. Furthermore, the data show a slight
increase in breakdown voltage with the increase of horizontal creepage distance from 0 to about 4
mm, but seems to have no further improvement with additional increase in creepage distance. This
effect is expected and is an indication that the breakdown discharge process was dominated by local
field processes.
Coaxial Cylinder Electrode Configuration
The average breakdown voltages and the deviations at four different temperatures, from 1.8 to
4.2K, are shown in Fig. 7. Five to nine tests were performed at each temperature point. As found in
tests in other electrode configurations, large data scattering was observed in Fig. 7, except for T =
3.46K. A decrease in the breakdown voltage is indicated as the temperature decreases from 3.46K
to 1.8K.
The true effect of large electrode surface area in reducing the breakdown voltage can not be
assessed with confidence by these data. However, a preliminary assessment can be made by
comparing these data with the gap breakdown data of the plane-plane electrode configuration since
the electrical fields in both configurations are essentially uniform field. By increasing the
electrically stressed areas 21.6 times, from 31.65 cm 2 to 685 cm 2 , the large surface area electrode
configuration shows a reduction of 32 to 39% in breakdown electrical field, from 9.84 to 6.02
kV/cm forT=4.2K and 9.06 to 6.19 kV/cm forT=1.8 to 1.9K. However, it is to be noted that part
of this reduction in breakdown electrical field is a result of increase in electrode spacing.
CONCLUSIONS
Dielectric breakdown tests of commercial grade liquid helium at temperatures of 1.8 to 4.2K
and near atmospheric pressure were conducted for both uniform and non-uniform field electrode
configurations. Both breakdown across electrode gaps and flashover on the insulating spacer
surface were obtained in testings. In most tests, there are appreciable data scatterings with a fixed
test condition. However, certain trends, such as the variation of breakdown voltages with respect to
liquid helium temperature and the reduction of breakdown electric field with increase in stressed
electrode surface area were detected. These results provided design guidelines in the SMES
insulation design and may prove to be useful for the designing of other superconducting equipment.
737
ACKNOWLEDGMENT
The authors are grateful to the Westinghouse Electric Corporation and to the Defense Nuclear
Agency for permission to publish this work. We would also like to express our appreciation to
Messrs. V. B. Doshi, R. R. Malingowski, and J. M. Toms of the Westinghouse Science &
Technology Center, and Mr. P. W. Eckels of General Electric Corporation for able assistance in
designing the test apparatus and in conducting the testing. This work was funded in part by
Westinghouse Electric Corporation and in part by the Defense Nuclear Agency under Contract No.
DNAOOl-88-C-0027.
REFERENCES
1. R.J. Meets, "Pressurized-Helium Breakdown at Very Low Temperatures," Proc. IEE, 119:760
(1972).
2. J. Gerhold, "Dielectric Breakdown of Helium at Low Temperatures," Cryogenics, 370 (1972).
3. M.M. Menon, S.W. Schwenterly, W.F. Gauster, R.H. Kernohan, and H.M. Long, "Dielectric
Strength of Liquid Helium under Strongly Inhomogeneous Field Conditions," Adv.
Cryog. Eng., 21:95 (1975).
4. K. Yoshino, "Dependence of Dielectric Breakdown of Liquids on Molecular Structure," IEEE
Trans. E1ectr. Insul., EI-15:186 (1980).
5. D. May and H. Krauth, "Influence of the Electrodes Surface Condition on the Breakdown of
Liquid Helium," IEEE Trans. Mag., 17:2089 (1981).
6. I. Ishii and T. Noguchi, "Dielectric Strength of Large Gaps in Cryogenic Helium," Proc. Int'l
Cryog. Eng. Corn., Kobe, Japan (1982).
7. J. Gerhold, "Helium Breakdown near the Critical State," IEEE Trans. Electr. Insul., 23:765
(1988).
8. M. Hara, T. Kaneko, and K. Honda, "Thermal-Bubble Initiated Breakdown Characteristics of
Liquid Helium and Nitrogen at Atmospheric Pressure," IEEE Trans. Electr. Insul.,
23:769 (1988).
9. J. Gerhold, "Breakdown Phenomena in Liquid Helium," IEEE Trans. Electr. Insul., 24:155
(1989).
10. C. Blank and M.H. Edwards, "Dielectric Breakdown of Liquid Helium," Phys. Rev. 119:50
(1960).
11. J.M. Goldschvartz and B.S. Blaisse, "Electrical Breakdown of Liquid Helium," Brit. J. Appl.
Phys. 17: 1083 (1966).
12. K. Yoshino, H. Fujii, K. Hayashi, U. Kubo and Y. Inuishi, "Electrical Breakdown in Cryogenic
Liquids," Proc. 6th Int. Conf. on Conduction and Breakdown in Dielectric Liquids,
Mont-Saint-Aignan, France (1978).
13. K.F. Hwang, "Dielectric Strength of Helium Vapor and Liquid at Temperature Between 1.4 and
4.2K," Adv. Cryog. Eng. 23: 110 (1978).
14. A. Migliori, R.I. Shermer and M.D. Henke, "Dielectric Tracking in Liquid Helium,"
Cryogenics, 343 (1978).
738
TESTING AND PERFORMANCE OF HIGH TEMPERATURE
SUPERCONDUCTING CURRENT LEADS
J. L. Wu, J. T. Dederer, O. R. Christianson,

and S. K. Singh
Westinghouse Science & Technology Center

INTRODUCTION
Vapor-cooled current leads generally used for transmitting power to a superconducting magnet
can introduce a significant heat leak into the cryostat. Refrigerating and liquefying the vapors
associated with cooling of these leads may constitute a significant portion of the power requirement
of the refrigeration/liquefaction system. Theoretical studies and experiments have demonstrated
that the heat leak introduced by a current lead can be reduced by using ceramic high temperature
superconductor as part of conductor in the current leads. 1-6
A high temperature superconductor (HTSC) reduces heat leak in a current lead by being
superconducting in the temperature range below its critical temperature and by having a low
temperature thermal conductivity which is generally orders of magnitude lower than the copper
alloys commonly used as the current lead conductors. The combination of superconducting and low
thermal conductivity can produce a significant reduction in heat leak and the associated helium
boiloff, as demonstrated earlier by a HTSC lead tested up to 2 kA operating in a self-cooled mode. 1
Application of HTSC to current lead design also permits efficient interception of heat leak at a
temperature substantially higher than 4.2K and therefore further reduces the power requirement
associated with cooling of current leads. 2 ,7-9
The ceramic high temperature superconductors are relatively new materials and their
mechanical, electrical and thermophysical property data are very limited. Employing these
materials in a device such as current lead requires not only detail design analyses but, more
importantly, comprehensive testing using test models and prototypes. The testing needs to include
performance evaluations of the superconductor parts as well as the entire lead assembly. Of
particular importance due to the difficulty in making low resistance ceramic/normal metal joints,
experimental evaluations of the joints in the operating temperature range are also required.
Since demonstrating the utility of HTSC in the 2-kA high temperature superconducting lead
two years ago in a joint effort with Argonne National Laboratory, Westinghouse Science &
Technology Center has continued its work in this technology area with the objective of developing
low heat leak vapor-cooled current leads for cryogenic applications. The technical issue of the
testing of these leads and its associate superconductor parts as well as analysis and discussion of
some of the test results are presented.
TESTING OF HIGH TEMPERATURE SUPERCONDUCTORS
The processing technique of ceramic superconductors has made great strides in recent years
and the performance such as critical current density has increased significantly. However, many of

these development works have been accomplished with small scale samples in research laboratories
and have not yet been widely implemented in commercial operations for large quantity fabrication
of device-size HTSC components with acceptable performance data and at a competitive price.
Application of HTSC to current lead design therefore requires detail testing of the superconductor
components fabricated in the desired geometry to evaluate their performance specifications.
We have tested ceramic superconductor bar samples obtained from various sources. These
bars are multicentimeters long and have a cross-sectional area of from 0.3 to over I cm 2. The bars
are of various material composition and were fabricated by different proprietary processes. The
main testing objective was to obtain the data, at temperature of 77K, of critical current density in the
presence of low external magnetic field, the contact joint resistance and the effects of thermal
cycling. Contact joint resistances at lower temperatures were also obtained as part of current lead
performance testing.
Figure I shows the test apparatus for testing the ceramic superconductor bars in liquid
nitrogen bath. It includes a container for liquid nitrogen, a Helmholtz coil pair for generating a
uniform magnetic field of up to 220 Gauss in the test region and an electrical terminals/support
assembly for conducting the transport current to the superconductor bar and for holding the
superconductor bar in place. DC power supplies were used for testing the bar for currents of up to
400 A. Transport critical current densities were determined by using the conventional four-probe
method and the 1 ",V/cm criterion. The four-probe measurements also produced data which yielded
the contact joint resistance. Thermal cycling tests were accomplished by repetitive cooled-down
and warm-up of the test sample between liquid nitrogen temperature and room temperature.
The details of the testing and the test results will be presented in a separate paper. I 0 It is
sufficient here to summarize the test results by stating that, at 77K, critical current densities of up to
several 100 A/cm 2 were obtained in the presence of external magnetic field of up to 100 Gauss,
contact joint resistance in the micro-ohm range can be achieved, and thermal cycling of up to 60
cool-down warm-up cycles does not degrade appreciably the performance of the superconductor
and its joints. In addition, the joint resistance of properly made HTSC/normal metal joints were
found to have a metallic characteristic with respect to temperature. Reduction of joint resistance by
a factor of three to more than ten was observed when the temperature is reduced from 78 to 4.2K.
Based on these test results, it can be concluded that the current ceramic superconductor
fabrication technology has advanced enough to produce practical superconductor components for at
least low current rating current leads. Additional testing of superconductors may be required for
their applications to current leads of rating in the 10-kA and up range.
Figure 1. Apparatus for testing superconductor bars in LN2 bars.
740
TESTING OF IllGH TEMPERATURE SUPERCONDUCTING CURRENT LEADS
The main objective of the testing conducted on current leads containing high temperature
superconductor was to obtain the helium boiloff rates associated with their heat leaks under steady
state operating conditions. The operating conditions of the leads were monitored by temperature
sensors located at various locations along the lead conductors and by the measurement of voltage
drops across different sections of the lead current path including across the superconductor part and
the overall lead assembly. Due to low thermal conductivity of the superconductor materials, the
temperature gradients near the liquid helium pool, which directly affect the helium boiloff rates, are
sensitive to the liquid helium level. Careful monitoring and controlling of the liquid helium level
during testing is also required. These tasks were achieved with level sensors and valves.
Other objectives of the testing of superconducting current leads include the assessment of the
possible helium vapor flow instability problem in a multiple lead assembly configuration and the
evaluation of zero flow stability of the lead. The latter was conducted with the helium vapor flows
deliberately cut off for a certain duration and the temperature rises of the lead conductor, including
the temperatures of the superconductor parts, were monitored carefully to determine if there was
any sign of an inception to bum-up.
Testing of WI ANL 2-kA HTSC Lead
The 2-kA HTSC lead, originally designed for l-kA rating, was tested for up to 2 kA. The lead
conductor was constructed with multiple superconductor bars and copper wires and was contained
within a fiberglass housing. This lead was thoroughly instrumented with nine temperature sensors
and nine voltage taps along its conductor.
The lead was tested in a circuit shown in Figure 2. A commercial copper lead was used to
provide a return current path. Both the superconducting lead and the commercial lead were
installed vertically through the cover plate of a dewar. The lower ends of these two leads were
connected by a cable of Nb-Ti superconductor soldered to a copper bar stabilizer. Figure 3 is a
photo showing the installation of the HTSC lead (left to the centerline) and the commercial lead
(right to the centerline) under the dewar cover plate. Instrumentation wires, liquid helium level
sensor, and aluminum radiation shield plates are also clearly visible. Figure 4 shows a top view of
the dewar cover plate during testing. The dc electric current was supplied by a dc generator and was
measured by a low resistance shunt. A shunt resistor was connected across the generator terminal to
provide protection to the test device.
The helium boiloff rate of the superconducting lead was measured by a calibrated rotameter
flow meter. The helium vapor was introduced into the flow meter after passing through a copper
pipe heat exchanger which brings the vapor temperature to near room temperature. The output of
the flow meter is exhausted to a vacuum pump through a metering valve. This arrangement
eliminates the variation of instantaneous flow due to minor pressure variation within the dewar.
The pressure within the dewar was maintained at slightly above aunospheric pressure.
Steady state conditions were achieved by adjusunent of the flow rate to such a point that the
pressure difference between the outlet of the lead and the dewar (monitored by a differential
manometer) is stabilized and the temperatures in the lead conductors and of the exit helium vapors
are also stabilized. In this test setup, all the helium boiloff associated with the current leads has to
pass through the flow meter.
The 2-kA lead was tested successfully for current from 0 to 2000 A with 100 A increments.
Testing times were typically 10 to 20 minutes per current step, with some having been run for a
longer time of up to one hour.
Testing of Multiple HTSC Lead Assembly
A vapor cooled current lead assembly consisting of multiple pairs of HTSC leads has been
designed, fabricated, and is currently under testing. Each lead pair is electrically isolated from the
others but is cooled by helium vapors generated at a common liquid helium pool when the lead
assembly is operating in the self-cooled mode.
741
R
To Atmosph ere'--_--t>IQ-_----,:i.
or Vacuum Pump
Commercial Copper
Lead TSC Lead
Dewar
liquid He - - - -
Nb Ti Conductor
G: D. C. Generator H: Heat Exch anger

R: Shunt Resistor ( 400 ~ Q ) P: Manometer
5: Low Resistance Sh u nt F: Flow Meter
(49. 8 ~Q I
Figure 2. Test arrangement of W/ANL 2-kA HTSC lead.
Figure 3. Installation of Y:!J ANL HTSC lead for testing.
742
Figure 4. Top view of dewar cover plate during testing ofW/ANL 2-kA HTSC lead.
The testing of the lead assembly in the self-cooled mode has recently been completed. The
instrumentation, test setup and test procedures for these tests are very similar to the testing of the
2-kA lead discussed earlier, the exception being that additional vapor flow measuring paths,
including heat exchangers and flow meters were installed to allow measurement of flow rate
through some selected lead elements. Figure 5 is a photo of the general test set-up showing the
dewar containing the lead assembly (center), the flow rate measuring instrument panel (left) and the
data acquisition equipment (right).
The testing completed includes the measurements of helium boiloff rates under steady state
condition with all or part of the lead elements carrying currents. The run times for steady-state
helium boiloff measurement were between 1.5 hours to 3 hours per lead connection configuration.
In addition, tests to assess the zero flow stability of the lead assembly have also been completed.
PERFORMANCE OF IDGH TEMPERATURE SUPERCONDUCTING CURRENT LEADS
The heat leak performance of a vapor-cooled current lead operating in the self-cooled mode
can be compared by the specific helium boiloff rate, defined as the liquid helium boiloff rate per kA
of current. Optimum conventional current leads are generally quoted to have a heat leak of 1.16
W/kA, or equivalently a specific boiloff rate of 56.8 mg/kA· S.
Figure 6 shows the specific helium boiloff rates, together with the exhaust helium vapor
temperatures, of the 2-kA HTSC lead. Clearly, the optimum operating current for this lead is
between 1.7 to 1.8 kA. In this current range, the exhaust helium boiloff rates in the optimum
current range are between 34.5 to 37 mg/kA • S, which is 35 to 39% lower than the optimum
conventional leads.
The just completed helium boiloff rates measurement of the multiple HTSC lead assembly
again confirms the utility of HTSC in reducing heat leak in current leads. Preliminary analysis of
the test data shows that, with all lead pairs carrying the rated currents, the specific boiloff rate is
around 31 mg/kA • S. In addition, the multiple lead assembly was shown to have a zero flow
stability of longer than 20 minutes.
743
Figure S. Test set-up of multiple HTSC lead assembly.
744
320
310
300
Vapor Temperature 290
:><:
280
ai
'-
270 ~
10
'-
V'l60 260 Q)
Cl.
E
250 Q)
I-
'-
240 0
Cl.
>'"
230 ::r:
Q)
220 ~
UJ
210
200
190
180
2.0
Current. kA
Figure 6. Specific helium boiloff rates and exit vapor temperatures of the W /ANL 2-kA HTSC lead.
DISCUSSION
As demonstrated in the testing of 2-kA HTSC lead and the multiple lead assembly, employing
ceramic high temperature superconductor as part of the conductor in vapor-cooled current lead can
produce a significant reduction in heat leak. This reduction can be as much as 50% as compared to
the optimum conventional lead when the leads are operating in self-cooled mode. The resultant
reduction in liquid helium boiloffs significantly reduces the consumption rate of liquid helium in an
open system. In a closed system with refrigerator/liquefier, the capacity of the installation as well as
the operating power requirement can be reduced accordingly. Our preliminary estimates have
shown that the cost reduction associated with the reduced refrigerator/liquefier capacity often is
sufficient to offset the expected increased cost of HTSC current leads. The saving from lower
operating power consumption is then the minimum net benefits realized in using these current leads.
Significantly higher reduction in refrigeration/liquefier power requirement can be achieved by
operating the HTSC current lead not in the self-cooled mode but in such a way that the heat leak can
be intercepted at an intermediate temperature between 4.2K and the critical temperatures of the
superconductor. 2 ,7-9 This operating scheme is made possible mainly by the lower thermal
conductivities of the superconductor materials. The low thermal conductivity limits the heat
conduction down the superconductor and allows efficient removal of heat at the intermediate
temperature by a heat sink or by externally introduced helium vapors. Our study has shown that up
to 95% of the boiloffs in the self-cooled mode can be replaced with externally introduced vapors at
an intennediate temperature to cool the lead conductor. The reduction of ideal power requirements
for the refrigerator/liquefier, depending on the intermediate temperature, can be as high as a factor
of four.
745
CONCLUSIONS
Testing of superconducting current leads and the superconductor bars for use in these current
leads has been performed. The test results demonstrated that the fabrication technology of ceramic
superconductor has advanced enough to produce components for HTSC current leads and that a
reduction in helium boiloff rate of up to about 50% were achieved by these leads when operating in
self-cooled mode.
Applying HTSC leads to superconducting devices, either in self-cooled mode or with
intermediate heat intercept, can realize substantial economical benefits in both the reductions in
refrigeration/liquefier capacity and in power consumption.
ACKNOWLEDGMENT
The authors are grateful to the Westinghouse Electric Corporation for permission to publish
this paper. We would also like to express our appreciation to several companies and one laboratory
for providing us with the superconductor samples used in testing, and to the Argonne National
Laboratory for their cooperation in carrying out the cooperative R&D program resulting in
successful fabrication of the 2-kA lead.
REFERENCES
1. J.L. Wu, J.T. Dederer, P.W. Eckels, S.K. Singh, J.R. Hull, R.B. Poeppel, C.A. Youngdahl, J.P.
Singh, M.T. Lanagan, and J. Balachandran, "Design and Testing of a High Temperature
Superconducting Current Lead," IEEE Trans. Mag., 27:1861 (1991).
2. J.R. Hull, A. Unal, and M. Chyu, "Analysis of Self-Cooled Binary Current Leads Containing
High Temperature Superconductors," submitted to Crogenics.
3. J.R. Hull, "High Temperature Superconducting Current Leads for Cryogenic Apparatus,"
Cryogenics, 29: 1116 (1989).
4. F.J. Mumford, "Superconducting Current-Leads Made from High Tc Superconductors and
Normal Metal Conductor," Cryogenics, 29:206 (1989).
5. A. Matrone, G. Rosatelli, and R. Vaccarone, "Current Leads with High Tc Superconductor Bus
Bars," IEEE Trans. Mag., 25:1742 (1989).
6. F. Grivon, A. Leriche, C. Cottevieille, J.e. Kermarrec, A. Petitbon, A. Fevrier and Y. Laumond,
"YBaCuO Current Lead for Liquid Helium Temperature Applications," IEEE Trans. on
Mag. 27:1866 (1991).
7. M.A. Green, "The Role of Superconductor in Reducing the Refrigeration Needed to Cool the
Leads of a Superconducting Magnet," Cryogenics, Supplement, 30:679 (1990).
8. K. Ueda and M. Mimura, "Calculation of Thermal Characteristics of HTSC Current Lead,"
presented at the 1991 Spring Meeting of the Cryogenic Society of Japan.
9. J.L. Wu, J.T. Dederer, S.K. Singh and J.R. Hull, "High Temperature Superconducting Current
Leads for Fusion Magnets Systems," 14th IEEE Symposium on Fusion Engineers, San
Diego, CA,. Oct. 1991.
10. J.L. Wu, "Testing of High Temperature Superconductors for Cryogenic Curret Lead
Applications," to be published.
746
sse 50 MM COILIDERDIPOLE CRYOSTAT
SINGLE TUBE SUPPORT POST
CONCEPTUAL DESIGN AND ANALYSIS
Thomas H. Nicol
P.O. Box 500
Batavia, IT.. 60510 USA
INTRODUCTION
Superconducting Super Collider (SSC) dipole magnet cold masses are
connected to the cryostat vacuum vessel at five places equally spaced along
their length. Five supports limit sag of the cold assembly due to its own
weight to a level consistent with the final magnet alignment specifications.!
The supports currently used in the 50 mm dipoles being built at Fermilab and
Brookhaven are adaptations of the design developed during the 40 mm design
program at Fermilab. The design essentially consists of two composite tubes
nested within each other as a means of maximizing the thermal path length.
In addition it provides an ideal way to utilize materials best suited for the
temperature range over which they must operate. Filament wound S-glass
is used between aOOK and SOK. Filament wound graphite fiber is used
between SOK and 20K and between 20K and 4.5K. S-glass is a better thermal
performer above approximately 40K. Graphite composites are ideally suited
for operation below 40K. The designs for both the 40 mm and 50 mm
reentrant supports are well documented in the literature. 2 - 5 Complete
coverage is outside the scope of this report. Figure 1 illustrates a cross
section through the current design for 50 mm collider dipoles support post.
The current design of the reentrant support has two major drawbacks.
First, it requires very tight dimensional control on all components; composite
tubes and metal attachment parts. Second, it is expensive, with cost being
driven by both the tolerance constraints and by a complex assembly
procedure. It seems clear that production magnets will require a support
structure which is considerably less expensive than that which is currently
used. The unit cost for support posts for 50 rom prototype magnets is on the
order of $2500 in quantities sufficient for 20 magnets (100 supports).
Work at Brookhaven National Laboratory and the SSC Laboratory over
the past two years on RHIC cryostat support structures led to the
development of a support post fabricated using Ultem-2100, an injection
molded, chopped glass fiber, composite material. 6 Due to its apparent cost

1----------228.6 m m - - - - - - - 1
1--------177.8 m m - - - - - - I
I
177 mm
mm
1----------241.3 mm-----------I
1------------292.1 mm------------I
Figure 1. Current 50 mm collider dipole reentrant support post cross section.
advantage, the injection molding concept has found strong support among
industrial contractors working on the design of production collider dipoles
and quadrupoles. It may not, however, be the panacea that it is perceived to
be for sse dipoles. Space is limited in collider dipole cryostats so an injection
molded assembly which provides sufficient strength and low heat load may
be difficult to incorporate in the design. In addition, the sse design requires
an intercept at 20K which RHIe magnets do not utilize, further complicating
the assembly. Finally, injection molded composite materials don't exhibit the
combination of stiffness, strength, and low thermal conductivity required in
the sse collider dipole design.
It seems clear that a design alternate for reentrant support posts will
be required for production dipoles primarily due to their cost. It seems less
clear that injection molded composite materials are the ideal choice.
This report describes the conceptual design for a support post whose
function is identical to that of the current reentrant design, which requires
very few modifications to surrounding cryostat components, is thermally
equivalent to the current 50 mm support post, and is nearly equivalent
structurally. The focus of this work is on a design aimed specifically at
application in sse 50 mm collider dipoles, however, the conceptual design
presented here is applicable to other cryogenic systems.
DESIGN OVERVIEW
The reentrant supports used in 40 mm and 50 mm sse collider dipoles
have shown themselves to be very capable of meeting their design
requirements structurally and thermally, both in prototype magnets and in
individual component tests. Filament wound tubes are superior to tubes
fabricated from cloth lay-ups or prepregs due to the ability, inherent in the
filament winding process, to tailor the material properties in each of three
principal directions. They are superior to injection molded tubes in strength
and stiffness by virtue of their continuous fibers. Injection molded
748
components either use chopped fibers randomly oriented in a resin matrix or
a resin alone.
The method of joining the composite tubes to their metal end
attachments in current reentrant supports has also proven very reliable over
the broad temperature range to which the assemblies are subjected. Shrink
fitting is used to capture the composite tubes between inner discs and outer
rings. 2- 5 Epoxy joints and other forms of mechanical fasteners are less
reliable when subjected to temperature extremes which range from aOOK to
4.5K.
This is the starting point for the subject design. It utilizes filament
wound composite tubes connected to metal end fittings and thermal
intercepts via shrink fitting. The difference is that it no longer utilizes the
reentrant tube concept and so contains fewer metal attachment parts and is
easier to assemble.
DESIGN ANALYSIS
The design analysis for this or any other suspension system
component consists of two parts; estimation of both structural and thermal
performance. In this case each is accomplished by separate finite element
models which are parameterized to allow many different design options to be
considered. The structural model allows the deflections and stresses in the
composite tubes to be estimated and compared with the design requirements
and with the composite material properties. The thermal model allows heat
loads to 80K, 20K, and 4.5K to be estimated and compared with the heat load
budget. Tables 1 and 2 illustrate the structural and thermal requirements
placed on the support structure for 50 mm sse collider dipoles.
Table 1. 50 mm collider dipole structural criteria.

Overall cold mass weight: 11,360 kg
Lateral load requirement: 1.0g
Vertical load requirement: 2.0 g
Axial load requirement: 1.5g
Allowable composite material stress: 50% O"ult
Table 2. 50 mm collider dipole heat load budget per support.

Thermal station Budget heat load
4.5K 0.032W
20K 0.480W
80K 3.160W
The first step in the initial design process for both 40 mm and 50 mm
sse dipole supports was to determine something about the available physical
envelope into which the assemblies would fit. The intent here is not to
launch into a completely new cryostat design, but rather to develop a
replacement which requires as few modifications to other components as
possible. Given that, the overall height of the support proposed here is
identical to that of the current 50 mm collider dipole, i.e. 177.8 mm. There is
some flexibility in the overall assembly diameter. Preliminary analysis
indicated that the largest tube diameter possible would result in the lowest
749
overall thermal conductivity and highest structural strength. Space
limitations inside the cryostat represent a physical limit to what is practical.
In this case, a 235 mm outside tube diameter represents a reasonable
compromise between performance and minimum impact on the existing
cryostat design. It was also clear from work on 40 mm and 50 mm supports
that fiberglass reinforced composite material (FRP) is ideally suited for
operation between 300K and 2DK and that graphite reinforced composite
(GRP) was the material of choice for operation between 20K and 4.5K.3-4 The
difference in material for the operating temperature regions is due to the
thermal conductivity of each vs. temperature. FRP has a lower thermal
conductivity integral between 3DOK and approximately 40K than GRP. GRP,
on the other hand, is a better thermal performer below 40K. Additionally,
GRP provides superior structural strength, allowing smaller material cross
sections leading to even better thermal performance of the structure.
The structural and thermal performance of this and all other
superconducting magnet support systems are generally at odds with one
another. Load and stiffness specifications imply materials with high
strength. Low allowable heat loads imply good insulators which are typically
not good structural materials. The goal of this or any such design is to strike
a reasonable balance between the two.
After choosing a physical envelope, the tube wall thicknesses were
determined using the support thermal model. A single tube can be
machined such that the wall thickness is optimized between thermal
intercepts. Optimized in this case means that the conductive heat load to
each thermal intercept just meets the budget. Figure 2 illustrates the
equivalent thermal model used in the analysis. The tube thicknesses
between each intercept are determined which result in cross sectional areas
that meet the heat loads in table 2. The cold mass cradle AIR is explicitly
assumed to be 0.85 em based on thermal tests of 40 mm cradle assemblies.
Given these parameters, successive iterations of the thermal analysis
resulted in the tube thicknesses and thermal path lengths shown in table 3.
A closed form solution does not exist due to the temperature dependent
nature of the material thermal conductivities.
Table 3. Material, thermal path length, and wall thickness for each
tube section (mm).
Tube ~e!:tiQn M!!t~ri!!1 Th~rmal nath W!!ll thi!:kne~~
300Kt.oBOK FRP 51.9 mm 2.15 mm
80Kt.o20K FRP 36.4 mm 1.64 mm
20Kt.o4.5K GRP 26.0 mm 1.52 mm
4.5K connection
Cold mass cradle A, I, JICdt
Post/cradle interface
20K-4.5K tube A, I, JICdt
20K intercept
80K-20K tube A, I, JICdt
80K intercept
300K-80K tube A, I, JICdt
300K connection
Figure 2. Single tube support equivalent thermal model.
750
ANSYS 4.4A
FEB 211992
16:16:29
PLOT NO. 1
PREP7 ELEMENTS
TYPENUM
XV =3
YV =4
ZV =5
DIST=6.899
YF =3.5
ZF =-2.813
PRECISE HIDDEN
Single tube suppart-235mm OD, 177.8mm high-Feb 1992
Figure 3. Structural model - finite element mesh.
Given the above geometry, the next step in the analysis is to determine
the structural adequacy of this design. The structural finite element model
simulates a complete 3-d cross section of the support. The worst case load is
that which occurs at the cold mass centerline, i.e. 1.0 g lateral applied 359.8
mm above the base of the support due to shipping and handling loads. For
this analysis a 1.0 g vertical load is superimposed on the lateral load to
simulate the cold mass weight. The axial load is ignored due to the fact that
a shipping restraint will be installed at the magnet ends to limit any axial
load on the suspension during shipping. Figure 3 illustrates the finite
element mesh used in the structural analysis. Table 4 contains the
maximum tensile, compressive, and shear stresses resulting from the
structural analysis.
Table 4. Maximum compressive, tensile, and shear stresses in single

tube support composite members (MPa).
Support sect jon q(compr) a.Lt.en.s.2 I(max)
300K-80K tube -133.7 102.6 28.2
80K-20K tube -163.9 120.7 37.6
20K-4.5K tube -188.7 146.8 40.3
The allowable stresses for each of the two composite materials is 138
MPa for the FRP section between 300K and 20K and 207 MPa for the GRP
section between 20K and 4.5K Given the tube thicknesses, materials, and
thermal path lengths determined in the thermal analysis, we conclude that
this assembly also meets the structural requirements of the 50 mm collider
dipoles.
ASSEMBLY DETAILS
Figure 4 illustrates the overall assembly of the completed single tube

support whose geometry was determined above. The term 'single tube
751
4.5K connection
mm
20K oonnection
mm
BOK connection 177 mm
mm
300K connection
1-------235.0 mm---------l
1-----------311.2 mm----------I
Figure 4. Completed single tube support assembly.
support' is somewhat misleading for this design. The term originally

referred to the non-reentrant nature of the design. Still, however, we utilize
two composite tubes assembled as a single tube. The joint between these
tubes is effected by capturing overlapping sections of each composite material
between the 20K disc and ring, both of which are 6061-T6 aluminum. These
two components are sized such that an interference fit exists at assembly.
The contact pressure from this shrink fit joint is 58.6 MPa which is more
than adequate to resist tensile, compressive, and bending loads imposed on
the assembly. Figure 5 illustrates the details of the joint between the FRP
and GRP tube sections. No adhesives or mechanical fasteners are used in
the assembly. Each of the other joints are made identically. The outer rings
at each thermal station are machined such that they are a line to line fit with
the tube OD. The inner discs are machined such that they oversized to a
prescribed amount based on the overturning moment requirements at each
joint. They are cooled to LN2 temperature and slipped into the tube with
tooling that ensures they are lined up with the outer rings. MLI is installed
during the assembly process to minimize radiation heat transfer through the
assembly.
GRP tube section
20Kring
20K disc
Contact FRP tube section
pressure
surface
FiJ!ure 5. FRP/GRP tube connection detail.
752
FINAL DIMENSIONAL AND PERFORMANCE PARAMETERS
Table 5 lists some of the dimensional and performance parameters for
the single tube support described in this report as well as comparable values
for the current 50 mm support post design.
Table 5. Final dimensional and performance parameters for a

single tube support and the current 50 mm dipole reetrant support.
5inlode tuhe IWlUllldi lkentrant 1WIllllldi
300K to ~ section
OD: 235.0mm 241.3 mm
thickness: 2.15 mm 2.16 mm
thermal path length: 51.9 mm 63.5 mm
material: FRP FRP
max tensile stress*: 102.6MPa 60.4MPa
ll)K to3)K section

OD: 235.0 mm 177.8 mm
material: FRP GRP
20K to 4.5K section

OD: 235.0 mm 177.8 mm
material: GRP GRP
Overall assembly height: 177.8 mm 177.8 mm

Cold mass deflection: 2.33 mm 2.76 mm
Heat load to 4.5K: 0.032W 0.033W
Heat load to 20K: 0.482W 0.518W
Heat load to 80K: 3. 164W 3.107W
*:all stresses are calculated at centerline loads of 1.0 g lateral, 1.0 g vertical
Several things are clear from the above comparison of the current
support design and that proposed in this write-up. First, both are capable of
meeting the prescribed heat load budget. The reentrant design is slightly
over budget at 20K due to a slight error in initial calculations. It could easily
meet the budget with a slight reduction in the GRP tube wall thickness and
adjustment in the position of the 20K intercept. Second, the reentrant design
is clearly superior in terms of structural strength. It was initially sized for a
lateral load of 1.5 g which is the reason for such low stresses at 1.0 g. We owe
the superior structural strength to the reentrant design which maximizes
the thermal conduction path without increasing the overall bending
moments on the component tubes. Lastly, the single tube support is
marginally superior in terms of stiffness. Although not explicitly defined as
a design criteria, stiffness is important in the dynamic performance of the
support structure. There is some flexibility in the design envelope for the
single tube support. More detailed analysis may yield design parameters
somewhat different than those outlined here. These, however, serve as a
reasonable starting reference.
753
7
c:I
Q
1 6 40mm supports
""
I;:::l
~ 5
i 50mm supports
S 0
.5 ..... 4
.s ~
Final design
calculated value
1: ..... 3
iGI
U
III 2
a
'C 1
Q
0
o ~ ~ ffi ~ a~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
Applied cold mass centerline lateral load (kg)
i i ~ ;
Figure 6. Reentrant and single tube support bending test results.
The development program thus far has included assembly of four

single tube supports for mechanical testing only. These supports were larger
than the design described above. The tubes were 240 mm on with wall
thicknesses of 2.79 mm, 2.29 mm, and 2.03 mm along the 300K to SOK, SOK to
20K, and 20K to 4.5K sections respectively. The overall height of the test
assemblies are 203.2 mm. Using this geometry enabled us to make use of an
existing filament winding mandrel, saving both time and development costs.
Figure 6 illustrates the results from these tests as well as test results from 40
mm and 50 mm dipole reentrant supports.
Thermal tests have not been conducted to date, but are critical in
validating the calculated heat loads to each thermal station. Sufficient
prototype assemblies are planned to enable us to complete this testing later in
1992. These tests will be documented and published when available.
The above analysis indicates that a support post fabricated from a

single composite tube can meet the structural and thermal specifications
proposed for the sse 50 mm collider dipole suspension system.
Initial estimates from vendors familiar with assembly of reentrant
supports indicate that the total cost of a single tube support in quantities
sufficient for the entire collider dipole production program would be on the
order of $500 to $600, compared with $2500 for smaller quantities of reentrant
supports. It is difficult to imagine a support fabricated from an injection
molded material resulting in cost savings much more significant.
A test program has been initiated which confirms the predicted
structural performance of this assembly. One which confirms the thermal
754
performance is in process. Development work needs to be done to determine
whether or not it is feasible to fabricate a single tube with both S-glass and
graphite fibers in the same structure in quantity. Filament winding or resin
transfer molding may both be viable fabrication techniques.
It is possible that development work at the SSCL or at one of the
industrial contractors will yield a support structure which successfully
utilizes injection molded materials. The above analysis is one more
alternative and one which is based firmly on proven design concepts.
ACKNOWLEDGMENTS
The author would like to thank Mssrs. Michael Kramer and James
Leslie of ACPT, Huntington Beach, CA for their expertise and assistance in
design and prototype development and Mr. Richard Kunzelman of Fermilab
for prototype assembly and testing.
REFERENCES
1. Superconducting Super Collider Laboratory, "15 Meter Collider Dipole

Magnet Prime Development Specification," March 6, 1990.
2. Nicol, T.H., et aI., "A Suspension System for Superconducting Super

Collider Magnets," Proceedings of the Eleventh International
Cryogenic Engineering Conference, Vol. 11, Butterworths,
Surrey, UK, 1986, pp. 533-536.
3. Nicol, T.H., et aI., "SSC Magnet Cryostat Suspension System Design,"

Advances in Cryogenic Engineering, Vol. 33, Plenum Press,
New York, 1987 pp. 227-234.
4. Nicol, T.H., J.D. Gonczy and R.C. Niemann, "Design and Analysis of
the SSC Dipole Magnet Suspension System," Super Collider 1,
Vol. 1, Plenum Press, New York, 1989, pp. 637-649.
5. Nicol, T.H., "SSC 50 mm Collider Dipole Cryostat Design," presented at

the 1991 CECIICMC Conference, Huntsville, AL, June 11-14,
1991.
6. Wolf, L.J. and J. Sondericker, "Alternative Concepts for Structurally

Supporting the Cold Mass of Superconducting Accelerator
Magnets", presented at the 1991 IISSC Conference, Atlanta, GA,
March 13-15,1991.
755
WELDING VISION SYSTEM FOR sse 50MM YOKE-SKINNING
PRODUCTION
J.E.Haggard l , J.Carsonl, J.Jones 2, L.Nelson2, M.

Reynolds 2, D. Rippinger2, W.Robotham l , and J.Strait l
lTS/SC Magnet Production

2rs Welding Department
Fermi National Accelerator Labs
Batavia, 11. 60510
Introduction
Fermi National Accelerator Laboratory has designed and

installed a four camera Gas Tungsten Arc Welding (GTAW) vision
system which employs a unique filtering system for the welding of
prototype superconducting dipole magnets for the
Superconducting Super Collider at its industrial facility in Batavia,
Illinois. The vision system was developed to provide superior
control over the skin welding process, as well as overcome
hazardous working conditions due to space limitations and high
ozone concentrations. The skin welds, for which this system was
developed, make up the liquid helium containment vessel of the
dipole magnet. This containment vessel assembly (called the "cold
mass") is used to cool the superconducting magnet coils within and
forms the structural element for providing the geometric
requirements of straightness and twist. The weld heat and
deposition of filler material must be as uniform as possible so as not
to induce assymetrical stresses which could lead to geometric
distortions. This weld system is integrated into a specialized press
developed at Fermilab for cold mass assembly.
The sse 50mm Magnets
The magnets being produced are 55 ft. (16.7 m) long, 13.4 in.
(34 cm) diameter prototype superconducting dipole assemblies.
They are prototypes of dipole magnets which will be connected in
series in an underground ring 53 miles in circumference. The
magnetic field produced by these magnets will steer two beams of
protons in opposing circular routes, ultimately causing them to
collide at high energies. SCientists will study the results of the
collisions to gain knowledge about the basic constituents of matter.

Skin Welding Operation
The skin welding operation is a critical step in the

production of a superconducting magnet. During this operation.
skin halves are welded over the coil and iron of the magnet under a
vertical preload of 12.200 lb/inch of length within a special press
and tooling. The skin halves are welded to a backing strip
("alignment key") in a double bevel v-groove jOint. The welding
process used is GTAW with cold wire feed.
1 - - -........---.210
.14775
.09375
CENTER LINE OF WELO GROOVE

0102-MB-292112
ALIGNMENT KEY
0102-MD-292111 SKIN
Figure 1. Weld Groove Configuration: Double bevel v-groove with backing strip.
Center groove is for pre-weld alignment with press tooling and carriage alignment
during welding.
The skin serves a dual purpose in the magnet structure. First.

the skin is the structural support for the coils of the magnet. The
magnet coils require an iron mass surrounding them for efficiency
and the iron mass requires the mechanical stability to resist the
intense magnetic forces generated by the coils. In operation. the
magnetic field forces the iron mass away from the coils when
energized. With insufficient mechanical stability and at operating
temperatures. if the coil assembly is incapable of resisting this
force. it will not be superconductingl. Secondly. the skin halves
form the wall of a single wall pressure vessel which contains the
liquid helium used to cool the magnet. The weld quality itself must
be suffiCiently high to guarantee the integrity of the pressure vessel
wall. The weld is produced to the intent of the ASME (American
Society of Mechanical Engineers) Boiler and Pressure Vessel Code
1 Jim Strait. "50 mm Collider Dipole Magnet: Requirements and Specifications". ArtIcle Title:
"Mechanical Design"; Base Une Issue; Rev. #3; Aug. 16. 1991
758
SPR I NG RETURN FOR
UPPER TOOL! NG
UPPER CROSS BEAM
WELDING
CARRIAGE
OPERATORS
CONSOLE
HYDRAUL I C CYL I NDERS
Figure 2. sse 50mm Yoke and Skinning Press with Welder.
which requires thorough calculations and drawings as well as

inspected, traceable, certified materials 2 .
The GTAW skin welding operation takes place in a 60 ft. long
multi-cylinder hydraulic press with unique tooling. Built within the
press is a two carriage linear welding system. The welding torch
carriages ride on parallel ball bushing tracks, driven by a chain
drive, with two torches per carriage mounted in the horizontal
position. The torches are mounted within a framework which is
spring loaded horizontally against the magnet assembly. The
welding key has an additional central v-groove along its entire
length. The carriage has guide rollers which roll in this groove and
adjust the whole carriage in the vertical direction. The torches are
individually adjustable with reference to the center of the guide
wheel groove. Also, the cold wire feed tips are mounted on the
individual torch frames within the carriage. The wire feed tip is
adjustable with respect to the torch tip in both the vertical and
horizontal plane.
System Redesign
The Original welding control and carriage system was designed and
built by an outside vendor to Fermilab specifications. The original
custom controller featured automatic gap control, servo wire feed,
2 Asme Pressure Vessel Code, Section 8, Dlv. #1. 1983
759
current and pulsed current control. The original carriage provided
operators with hand adjustments at the welding carriage to vary
wire feed tip placement and torch position.
This system was plagued by a number of problems. The first
was unexplained shut-downs of the welding system. Welders would
often be in the middle of a seemingly successful run only to have
the system shut-down by unexplainable system faults. Another
problem was high ozone concentration. The welding process itself
occurs in a very confined area within the press. Welders were
required to be very near to the weld in order to make hand
adjustments at the carriage. The amount of air entering this area of
the machine was not suffiCient to bring ozone levels down. Vacuum
proved to be a good temporary solution. A third problem was UV
radiation. The space inside the press was so confined that welders'
helmets could not be practically worn while welding. The resulting
innovation was hand shields and clothing which covered exposed
skin, providing only a temporary solution to the problem. The
decision was made to solve confinement problems by moving the
operators far away from the welding point. Moving the welders
away from the carriages required a remote viewing system. Camera
systems from various suppliers were demonstrated and quoted. All
of these systems were severely limited by the existing amount of
space. Borescope type lensing produced too much distortion.
Pentaprism type reflectors and mirror type reflectors resulted in
partial focus of the weld puddle area due to the steep angle
required. Laser vision triangulation systems were conSidered but
found to be cost prohibitive. The solution turned out to be a
micro camera and a carriage redesign.
Fermilab engineers designed and built new carriages which
featured four small water cooled cameras with remote set-up/weld
LIGHT SOURCE
fiBER OPTIC BUNDLE
AUTOMAT IC GAP CONTROL PUMP
fiLLER WIRE LETOff
TORCH
Figure S. SSC 50mm Yoke and Skinning Press Welder Carriage. There are two
carnages, one per side, which are chain driven on ball bearing rails mounted to the
press bed.
760
jog. set-up lighting. and a unique light filter arrangement. Also.
joystick control of wire feed tips and torch was added. The custom
wire feed servo drives and automatic gap control devices deSigned
for the first generation welding system were re-used in this design
since these devices had good service records and had not needed
replacement. The custom controller was also re-used with the
addition of a first fault indicator which was designed and built by
Fermi engineers 3 •
The redesign of this system incorporated several innovations.
The new system places two operators at a control console where
they monitor the weld progress via quad split screen camera
images and make position adjustments to wire feed tips. torch-
work gap. torch to base distance. and wire feed rate. Small linear
motors which were required to adjust the wire feed tips and torch
at the welding carriage were aVailable. but nothing that would
CABLE HANDLER
WIRE FEED SERVOS
I
I
\ FRONT
, rul~ ,
" \, 0 WHEEL/'
........ - - - - - - - - - - - - - - - - - - ....""
"' SETUP LIGHTS WIRE FEED TIP /"
Figure 4. SSC 50mm Yoke and Skinning Press Welder Carriage. Drawing shows
cramped conditions surrounding work area and camera to weld distance.
withstand near-welding temperatures. The positioning motors

selected (Newport Corp.) for torch and wire feed tips were
modified to be water cooled. Water jackets of aluminum are fitted
around the cylindrical body of the linear motors in the area of the
windings. The positioning motor controller (Newport Corp.) used
has velocity control. joystick control, as well as the ability to be
computer controlled. The cooling system for cameras and position
3 Allen Forni. "sse Linear WeIder First Fault Indicator"; Fennilab; RD Controls Special
Project Note 7.0; Dec. 2. 1991
761
motors is a standard Bernard Model 3500ss water cooler with a
50/50 mix of water and ethylene glycol. Another innovation is wire
feed tips of alumina ceramic which were added to provide a
measure of safety against torch to frame shorts and allow the tip to
be placed nearer the puddle than the previously used stainless steel
feed tips.
The camera System
The welder operators require certain information when

controlling the weld. The welders must know the location of the
skin edge. the key edge. along with the puddle width in order to
control arc gap setting. the current setting. the pulse durations.
and the torch position within the groove. The operators must also
see the welding wire entry into the puddle and must have a way to
review their work in order to judge surface quality and volume. The
camera system provides most of these with the exception being
volume. Volume must be judged by looking at the solidified weld.
With experience. this too is provided to the welder by the camera
image. Each welder monitors and controls two welds. having the
chOice of full screen or quad-split screen weld images. The welds
are also videotaped for periodic review.
The camera used was the Elmo EM 1 0211 CCD microcamera.
It features 574 horizontal x 489 vertical pixel elements which
produce a very clear image. The camera controller has automatic
gain control and auto white balance. two valuable features. The
small physical size of the remote camera head was the reason it was
chosen (4 3/4" 19. x 1" dia. w/connector). The camera controller is
mounted nearly 65 feet away from the camera head itself. The close
proximity of the camera to the weld (about 2.5" away) required that
the camera head be cooled. Custom camera cooling bodies were
obtained from Bartz Technology in Santa Barbara. CA.. The control
system for the cameras is a simple relay affair. During set-up.
shuttering is disabled and the work lights are enabled. While
welding. the shuttering is enabled and the work lights disabled.
A number of filtering schemes were tried during the
development of this system. One method was the use of
enumerated welding filters. (#10.#11.#13 welding filters.etc.) This
worked well in the puddle flash. but lacked a certain amount of
definition in the background. Variable neutral density filters were
too expensive. The final filtering scheme chosen was 1/1000
sec.shutter with a variable polarizing lense combination and a
neutral density center spot. The polarizer combination is used to
decrease the overall transmittance of light to the camera sensor as
the orientation of the grating goes through 90 degrees of rotation
about the axis of the camera body. The neutral density center spot
dot is used to damp the flash of the plasma at the end of the
electrode and keep the camera sensor from saturating. The dot
filter was purchased through Bartz Technology. Bartz Technology
also designed and built a variable polarizer. as well as a number of
other devices which were tried during prototyping. The N-D dot
filter is designed for 20-25% transmission and employs a quartz
substrate which has 470 0 F working temperature 4 . The light
4 Jerold Bartz; Private Communications
762
intensity is further reduced by the iris which is set at about 1/8 of
the maximum open position. No fine tuning of the camera is done
during a successful run. Originally, it was thought that one camera
set-up would cover each of the three different passes; this was
partially incorrect. Remote iris control of each camera is the choice
of other welding vision suppliers for "on the fly" tune-up and a
useful feature which should have been added to our system. The
lack of this type of control has not hampered Fermi welders,
however, as operators have gained the required experience to make
correct adjustments between each welding pass.
Depth perception, required during set-up, is made possible
by backlighting the torch. While in set-up mode, the wire feed tip
and tungsten are clearly visible due to light falling on the torch
from two directions. This particularly aids the welders during
automatic gap set, as shadows of the tips project on the tooling.
The camera image area is about 1 1/8" square using an Elmo JK-
LIS, 15 mm, F=2.0 lense. The camera is mounted about 2 1/2"
away from the weld pool. With the camera sensor cooled, no
adverse effects of heat have been noticed on the lenses and filters
themselves. The image obtained is excellent quality.
Conclusion
The Fermilab-SSC linear welder is producing real magnet

assemblies and has been very successful in improving the quality of
weld. The cost and time savings on the redesign was considerable.
The bulk of this project was completed in apprOximately 7 months.
The cost savings of the camera and monitor system was
apprOximately half the price of a purchased turn-key camera
system. The Fermilab-SSC Welding System for yoke skinning has
served as a model in the industrial technology transfer program
providing production vendors with valuable insights into magnet
welding technology.
763
AXIAL VARIATIONS IN THE MAGNETIC FIELD OF SUPERCONDUCTING
DIPOLES *
A. K. Ghosh, K. E. Robins and W. B. Sampson

ABSTRACT
Periodic variations in the field along the axis have been observed in a large
number of superconducting dipoles. Both long and short versions of the SSC magnet
have been examined as well as prototypes for the RHIC accelerator and coils from the
CBA and ISABELLE projects. This field oscillation is present in all the components
of the field. The period of the longitudinal pattern was found to match the conductor
transposition length over a wide range of pitches. The amplitude of variation increases
with time spent at high field and its decay has a complex time dependance.
Unbalanced currents in the individual strands of the cable appear to be the cause of
this effect and may also give rise to the unusually high sextupole drift observed near
injection field in most dipoles.
INTRODUCTION
When a superconducting accelerator magnet made from multistranded conductor

(cable or braid) is energized an axial variation in the magnetic field is established
which has the wavelength of the conductor transposition pitch. This pattern is present
at all fields and in both the normal and skew components. The magnitude of this
oscillation increases with both the maximum field achieved during the energizing cycle
and the time spent at high field. This phenomenon was first observed in HERA
dipoles in 1990 using a magnetic measuring device sensitive to the sextupole harmonic~
The purpose of this probe was to study the anomalous sextupole drift which occurs
when the magnet is at the injection field. It was noticed that small axial changes in
the position of the probe produced large changes in the field and the unusual
systematic variation discovered. Further investigation revealed that the variation was
approximately sinusoidal and that its amplitude increased each time the magnet was
cycled to high field. Similar variations were soon observed at other laboratories and
it became apparent that such patterns were present in all magnets made with cabled
conductors. This field oscillation had not been seen earlier with more conventional
* Work done under the auspices of the U. S. Department of Energy

magnetic measuring apparatus because of the short wavelength and the averaging
effect of the relatively long pick-up coils used. Since these field variations are small
(typically a few gauss) they are most apparent at low fields and can easily be seen
when the power supply is turned off after a high field cycle in which the current has
been maintained at the peak value for several minutes. Once the pattern has been
established it decays very slowly with time and can only be completely erased by
raising the conductor above its transition temperature.
MEASURING TECHNIQUES
Because of their small spatial extent and relatively high sensitivity, Hall effect
probes are ideal for measurements of this kind. Three types of elements have been
used in these studies SBV603, SBV604 and SBV6132• The first two have higher
sensitivity but are not as rugged as the 613 elements. Most measurements were made
at room temperature in a "warm finger" installed in the magnet aperture. In one
experimental arrangement two Hall probes were mounted on the magnet axis
separated by one half of the cable transposition width with a third probe situated at
the maximum radius and carefully aligned with one of the axis probes. This array was
mounted on a long shaft connected to a computer controlled mechanism that provided
both axial and rotational motions. Such a set-up allows the oscillations in the dipole
field to be observed directly while the harmonic components can be determined by
rotating the assembly and analyzing the difference signal between the axial and radial
probes. A second arrangement of probes in which three Hall elements are positioned
at the same radius but separated by 120 degrees was also used. This has the
advantage that the sextupole component can be measured directly without rotating the
probe holder and rapid scans along the magnet length can be made when measuring
the time variation of the field pattern.
Table I.
MAGNET LENGTH BQRE PITCH TYPE
DC0203 17m 4cm 79mm SSC
DCA207 15m 5cm 86mm SSC
DCA208 15m 5cm 86mm SSC
DCA207 1.8m 5cm 93mm SSC
DSA208 1.8 m 5cm 92mm SSC
DS0214 1.8 m 4cm 80mm SSC
DSKR19 1.8 m 4cm 79mm SSC
KEK501/03 1m 5cm 95mm SSC
HFM#3 1m 3.2 cm 60mm 3 layer flat coil
H.F. Test 1m 8cm 79mm 4 layer CosO
Nb3Sn MK II 1m 8cm 79mm 2 layer
ISABELLE 1m Bcm 316mm 2 layer braid
RHIC DR5006 1m 8cm 74mm 1 layer CosO
766
MAGNETS MEASURED
Superconducting dipoles for accelerator applications have been under

development at BNL for many years making it possible to compare magnets made
from a variety of conductors. To date some fifteen dipoles have been examined. They
are listed in Table I. Ten of these magnets are SSC prototypes of 4 or 5 cm design
and include both short and full-length versions. All magnets examined so far have
shown axial patterns in their remnant field and the wavelength of the variation has
coincided exactly with the transposition length of the inner cable. In cases where the
pitch of the axial field pattern was not precisely equal to the nominal transposition
length, direct measurement of the cable showed exact agreement. A typical pattern
of the remnant dipole field is shown in Figure 1 for RHIC short magnet DRS006.
15 r---------------,-----------------~----------------~
C,!)
0
....I
W 10
~
W
....I
0
0...
0
I-
Z
<
z 5
:::Ii
w
a:::
o u-______________ ~ __________________ ~ ________________ ~
50 100 150
AXIAL POSITION, em
Figure 1. Axial scan of the remnant dipole field in magnet DRS006 after a 30 minute
flat-top at 7000 Amps.
Since Rutherford cable is basically a flattened spiral there were early suspicions
that the effect was related to the solenoidal path followed by the strand currents. To
test this idea a magnet made from braided conductor which has strands spiraling in
both directions, was examined. The field oscillations in this magnet were obvious and
coincided with the very long (316 mm) transposition length of the braided conductor.
The possibility that this phenomenon is caused by filament deformation resulting
from the cable manufacturing process was tested using a dipole made with Nb 3Sn
cable. The filaments are formed by heat treatment after cabling so that any edge
degradation should be minimized. Again the axial field pattern was obvious suggesting
that any conductor made from multiple strands would give rise to field oscillations.
767
Magnets with one, two, three and four layers wound in the cos 8 style all had
similar field patterns corresponding to the inner layer pitch length. Even dipoles
wound with flat coils produced this characteristic pattern. The effect seems to be
independent of magnet geometry or type of conductor and only requires a conductor
made from multiple strands.
All harmonics show the same oscillatory behavior in both the normal and skew
terms. In general the axial phase relation between the two components is random and
the near sinusoidal variation is only apparent when the in phase and out of phase
portions are separated. This is illustrated in Figure 2 for the quadrupole and
sextupole harmonics of magnet DSA207. Note that the amplitude of the variation in
the skew component can be larger than the normal component but averages to near
zero while the average of the in phase component is equal to the value measured with
a long search coil. The amplitude of the oscillation decreases with harmonic number
but unallowed terms such as the quadrupole can be quite large.
5 15
C!)
.....
4-
3 •••• • C!)
L...i 10 p n-
•
..J
2
...J
o O b ,'"
.
0 Q. ,'\ 0
'0 \ •'
Q.
•
:::> :::>
5 , \ I
.,'.. \\. O.
l- 0:
x c
..... 0 -< .: • 0 .'. •i'
.....
I/) :::> ~
a
I- -1 o 0 .... ,,",
." ,
• .~-
z
z-<
-2 .
0·°0
~ ,
,
I-
z
< ,o ~. "
::::E -3
.....
0:
•• Z
::::E
.....
-5
,o \
\
0 -
-4- 0: I
() bd
-5 -10
0 5 10 15 o 5 10 15
AXIAL POSITION, em AXIAL POSITION, em
Figure 2. A portion of the axial scan of the sextupole and quadrupole components in the remnant field
of magnet DSA207. The normal component is shown by the open points and the skew
component by the solid points.
When this phenomenon was first discovered it was thought that a connection
might exist between the presence of splices in the strands (Koldwelds) and the
magnitude of the oscillation. Recent tests however have shown that it is possible to
have a large axial amplitude even in magnets that are known to have every strand
continuous.
Despite the fact that axial variations seem to be universal in all cable dipoles the
differences in the way the patterns grow and decay is quite remarkable. Each of the
fifteen magnets examined so far has shown some unique features and it has not been
possible to predict anything beyond the fact that an oscillation with the appropriate
wavelength will be observed.
768
OSCILLATION AMPLITUDE
As mentioned above the axial oscillations only appear when a magnet has been
held at high field for some time. For a given field level the amplitude grows
exponentially with time much like the voltage on a charging capacitor. The charge-up
time constant is typically 500 sec. but in one case was observed to be well over ten
thousand seconds. Similarly the maximum amplitude varies considerably from magnet
to magnet. The largest peak-to-peak amplitude measured to date is approximately 10
Gauss whereas the smallest is less than 1 Gauss. Determining the charge up time
constant in a magnet with a high decay rate can be complicated by the fact that
amplitude decay begins as soon as the coil leaves the flat top and considerable time
can elapse before the field is low enough to accurately measure the amplitude. At the
present time there is no explanation of these large differences in signal strength and
charge time but it has been speculated that the magnitude of the effect depends on
the uniformity of the individual strands in the cable. This is based on the fact that a
magnet made from a cable which had been fabricated from a single length of wire was
observed to have very small oscillations which required many hours of flat top time
to grow to measurable size.
The situation can in some cases be much more complicated than outlined above.
One of the sse model magnets DS0206 exhibited the unusually behavior shown in
Figure 3. When energized to 6700 amps for 30 min. the axial scan revealed almost
none of the usual structure but at currents 200 amps more or less that this value the
characteristic pattern shows distinctly. The patterns for different currents have been
offset vertically in Figure 3 for clarity. Note that the phase in the axial direction
W
..J
o
D..
o
I-
Z
<
z
:::IE
w
It:
30 40 50 60 70 80 90 100
POSITION. em
Figure 3. Axial pattern in the remnant dipole field of magnet DS0206 after successive cycles to
currents near 6700 Amps.
769
changes by 180 degrees on either side of the "magic" current, 6700 Amps. In this case
there appears to be a destructive interference between two patterns that has not been
observed in any other magnet.
TIME DEPENDANCE
Once an axial pattern has been established it decays in a complex way. Such
decay curves are often plotted using a logarithmic time scale because of an assumed
relationship between thermally induced flux creep and the drift observed in the
sextupole field at injection. The fact that such plots give reasonably linear behavior
seems coincidental since two or more exponentials with a large difference in their time
constants will also give a more or less straight line on a log t plot. This is illustrated
in Figure 4 where the decay of the amplitude of KEK 501/03 is compared with a curve
composed of three exponentials having time constants of 60,2200 and 400,000 seconds.
Experiments have been conducted to see the effect on the axial pattern of raising
small sections of the cable above the transition temperature. Magnet DSKR19 had
12
10
C!J
W
....I
0
D...
C 8
I-
Z
L
<
z
::::I:
-...... -............ --
w
a:::
6 I-
--"'- - ,.. -
4 ~----------~I------------~----------~I----------~
o 1000 2000 3000 4000
TIME, s
Figure 4. Amplitude vs. time for KEK 501/03. The solid curve is composed of three time constants
of 60,2200 and 400,000 secs. and with respective amplitudes of 4.6, 0.4 and 5.9 Gauss.
heaters installed on the conductor in the sections near the current leads and cross-over
between magnet halves. After a pattern had been set up by cycling the magnet, the
effect of the heaters was observed on the decay curve. When the heater is first
energized the amplitude drops quickly to a stable lower value. Increasing the power
770
25
20
"
W
...J
0 15
a..
c
I-
Z
'"
Z
:2
w
10
0:
20 40 60 80 100 120 140 160 180

AXIAL POSITION, em
Figure 5. Axial scans in magnet DSKR19 before and after energizing the inner lead heaters. The hump
at approximately 110 cm is caused by the strain gauge holder.
10
w
C
:::J
I-
6
::::i
a..
:2
'"
W
...J
4
o
a..
c
2 o
~eWY'"e9
o
o 1000 2000 3000 4000
TIME, s
Figure 6. Time dependance of the dipole oscillation amplitude showing the effect of heater power. The
solid curve is the decay after a cycle to 6700. The points are the same cycle with heater
power applied at 1800 seconds.
771
further reduces the signal and effects the local magnetization currents leading to an
increase in the dc level. Figure 5 shows the axial scan before and after heating the
inner coil leads. Outer coil heaters have no effect on the amplitude until the power
input is high enough to begin heating the inner coils. It is interesting that although
intercepting strand currents by this method does reduce the amplitude it does not
make it zero as would be expected if all the unbalanced current had to pass through
each section of cable in order to complete a circuit through the magnet joints. A time
plot of the amplitude is shown in Figure 6 where the step produced by the heater can
be easily seen.
CONCLUSIONS
The mechanism which produces axial field oscillations is poorly understood at the
present time. It certainly appears to be due to unbalanced currents which are set up
at high current during flat-top and retained at low field due to the much longer decay
time constants. The relationship of this effect to the unusually high drift rates
observed in accelerator dipoles is not obvious but it seems probable that they are
related since two strange phenomena in a system as simple as a Rutherford cable is
highly unlikely.
Despite the lack of detailed understanding several conclusions can be drawn from
those measurements:
i) All cable dipoles produce axial field patterns.

ii) All harmonics, allowed and unallowed, skew and normal are present.
iii) The oscillation wavelength matches the cable pitch.
ACKNOWLEDGEMENTS
The authors wish to thank the operators in the magnet test section for their help
in performing these measurements. Special thanks is due to Pam Meehan for
developing the probe transport mechanism. Rae Stiening of SSCL provided many
useful suggestions and constant encouragement.
REFERENCES
1 H. Bruck et al DESY HERA, 91-01, 1991
2 Siemens Components Inc., Iselin, N.J. 08830
772
CORRECTING FIELD HARMONICS AFTER DESIGN
IN SUPERCONDUCTING MAGNETS*
R.C. Gupta

Upton, NY 11973 USA
ABSTRACT
For a variety of reasons the actual field harmonics, as measured, in the supercon-
ducting magnets come out to be somewhat different than what were intended at the
time of design. In this paper we shall discuss the schemes which can be used to correct
them in the magnets. We shall discuss them for both the allowed and non-allowed
harmonics. Since the deviation in field harmonics from their design value is mainly
related to the mechanical properties of the coil cross section, in order for a scheme
to work as planned, the mechanical configuration of the coil should not be changed
significantly while this correction is being implemented.
INTRODUCTION
The field harmonics have been measured in a large number of superconducting

magnets for the Superconducting Super Collider (SSe) and for the Relativistic Heavy
Ion Collider (RHIC). The following relation is used to define the field harmonics:
By + iB", = Bo ~ [b~ + ia~] [cos (nO) + i sin (nO)] (~o) n ,

where Bo
is the field at the midplane at a radius Ro, B", and By are the components of
the field at (r,O) and Ro is the normalization radius which is 10 mm in sse magnets.
a~ are the skew harmonics and b~ are the normal. Only b2k , with k being an integer,
are the allowed harmonics in a magnet having a dipole symmetry.
* This work has been supported by U.S. Department of Energy.

The measured values of field harmonics in the dipole and quadrupole magnets for
SSC and RHIC have been observed to be different than what was intended at the
time of design. This is particularly true for the lower order harmonics like allowed
b2 and b4 in dipoles (bs in the quadrupoles) and non-allowed at in dipoles. One
practical way to obtain the intended values of allowed harmonics is to carry out an
iteration in the original design of the the coil cross section. This change should be
small mechanically for the process to converge in one or two iterations. Therefore,
it would be useful to carry out a theoretical study to examine the flexibility of a
particular coil cross section in terms of its ability to accommodate some changes in
field harmonics without significantly changing its mechanical properties. In principle,
one can also correct these harmonics, in particular b2 , by modifying the iron aperture.
We shall briefly discuss the feasibility of that also in this paper.
The non-allowed geometrical multipole at in the coil, which reflects an up-down
asymmetry in the coil, can be corrected by a deliberate up-down asymmetry in the
iron yoke. The multipole at is also introduced by iron saturation at high field when
the flux lines can not be contained in the iron yoke. This is because of the fact that
the yoke is asymmetrically located inside the magnetic cryostat wall both in the SSC
and in the RHIC dipoles. To correct this multi pole inside the dipole, one has to
introduce another up-down asymmetry in the iron yoke in the opposite direction.
First we shall discuss the schemes for correcting the non-allowed harmonic at and
then discuss the approaches which can be used in correcting the allowed harmonics.
CORRECTING NON-ALLOWED HARMONICS
Among the non-allowed measured harmonics, only at (skew quadrupole) has a

magnitude to be of any concern in SSC dipoles. Therefore, we restrict our discussion
to just al. We shall discuss first the geometric al and then the saturation induced al.
GeoIlletric at
The major source of geometric at in the dipole magnets is the up-down asymmetry
in the collared coil since the iron yoke contributes very little to it. A good correlation
has been observed l in the measured at in the collared coil before and after it was
placed in the iron yoke. This al is related to the tolerances in the manufacturing
process of the coil. This is of random nature and is the major source of the random
variations in the geometric at.
In this paper we propose that an up-down magnetic asymmetry in the collared
coil be compensated by deliberately introducing another asymmetry in the magnetic
length of the yoke between the upper and and lower halves of the magnet. It may
be pointed out that in the two ends of the magnet, the place where this correction
would take place, there is a transition from the low carbon steel laminations to the
stainless steel laminations. Therefore, the difference in the magnetic length between
the top and bottom half can simply be obtained by switching the type of laminations
between the upper and lower half of the yoke on the two sides of this transition plane.
In order to adopt this scheme, one would first measure the al in the collared coil at
room temperature. This would determine the number of laminations to be switched
in the top and bottom half of the magnet. Our preliminary estimates show that to
correct 1 unit of at one would need to switch 25 mm of the magnetic laminations
with the 25 mm of stainless steel laminations from the top to bottom in the two
ends of the magnet. The total amount of either the stainless steel or the low carbon
774
steel lamination does not change in the process. But for the end effects, there would
be theoretically no change in the allowed harmonics. Locally, this creates '" 200
units of al with respect to the central field there (which is 85% of Eo in the magnet
straight section). Other non-allowed harmonics introduced in the process, for example
a3, etc., are'" .01 unit or less - well within the specifications for them. There is
some 10ss2 in the al correction at high field due to iron saturation. This loss is
not expected to be linear with the length of the correction and therefore one would
make a table of length versus al correction both at low and high field and choose
a proper length accordingly. This scheme should be a relatively easy to implement,
particularly in the BNL type horizontally split yoke design where the geometry of the
non-magnetic stainless steel and magnetic low carbon steel lamination is identical. It
may, for example, be implemented in the following manner in a large scale industrial
production environment:
• The usual thickness of a pack of laminations in the BNL built sse magnets
is 3 inch. We propose that in the end region this pack be made t
inch
thick. This is also a natural choice in those designs where t
inch thick
laminations are used. This should be done for both the magnetic and the
stainless steel laminations.
• Paint low carbon steel and stainless steel laminations differently. Note
that the amount of the two types of laminations to be used in any magnet
is independent of the amount of al correction to be applied.
t
• Each inch block in one end corrects ~ unit of al. Decide the number of
the two painted laminations to be distributed in the top and bottom halves
of the magnet based on the measured al in the collared coil. Note that if
al is measured 0 then the paint will change at the same axial location in
the top and bottom half of the magnet.
In the BNL 50 mm dipole design, the length of the coil straight section is '" 585
inch, the length of the space occupied by the low carbon steel laminations is '" 582
inch and the length of the space occupied by the stainless steel laminations in the two
ends is '" 6 inch in each end.
Saturation Induced al
Both the calculations and measurements show a significant variation in al (skew

quadrupole) as a function of current beyond 6 tesla central field in sse dipoles 3 ,4.
This is because of the fact that the yoke is located asymmetrically 93.7 mm above the
horizontal axis of the magnetic cryostat vessel. One starts seeing this al when the iron
yoke is well saturated and the flux lines can not be contained inside the yoke. This
is a systematic effect and is in addition to the random geometric aI, discussed above.
The effect is several times the allowed specification of 0.04 unit for the systematic al·
In sse 50 mm Dipole the computed al saturation is in the range of 0.1 to 0.2 unit.
The measured values also appears to fall in the same range in those long sse 50 mm
dipole magnets for which the preliminary data are examined.
In this paper we discuss a few ways to reduce this systematic al and to bring
it within the allowed specifications. The final choice of a particular scheme may
depend on it's overall impact on the magnet production and the degree of cancellation
desired at all values of central field. A detailed finalized design would require a
confirmation with the measurements and an iteration may be desired to achieve a
proper compensation/ cancellation.
775
1. Placing Conductors in a Specified Location in Buss Work
It has been found that with a proper spacing and polarity of the two conductors
in the buss, the al produced by the proximity of the cryostat wall can be compensated
by the al produced by the conductors in the buss slot. The calculations show that
if the mid-point of the two conductors is placed 5 mm off the vertical axis on either
side, the net al in the magnet stays within the specified tolerance. The direction of
the current in the buss cable should be opposite to the direction of the current in
the coil below it on the same side. A tuning of al cancellation can be obtained by
changing the spacing between the conductors in the buss work.
2. Using a few Non-magnetic Steel Laminations in Upper Yoke-half
If the number of magnetic laminations are different between the top and bottom
half of the magnet, a skew quadrupole term is created. A practical way to implement
this in a magnet would be that some of the magnetic low carbon steel laminations be
replaced by the non-magnetic stainless steel laminations in the upper yoke-half. If the
number of non-magnetic laminations is a small fraction of magnetic laminations and
if they are evenly distributed, the situation can be simulated in a computer program
by using two different packing factors in the top and bottom half of the magnet. The
calculations show that rv 0.1 % difference in packing factor is adequate to bring the
net al within the specified tolerance. Since the thickness of lamination is 16 Gauge
(0.0598 inch) in BNL built magnets, it means that in a long magnet one would need
to change only 9 laminations from magnetic to non-magnetic in the top half. This
scheme has easy tunability - one would simply change the number of stainless steel
laminations.
3. Placing Extra Magnetic Steel at the Bottom of Yoke
Since the saturation al is caused by the proximity of cryostat wall at the top half
of the magnet, a natural solution to this problem would be to put some extra iron
on the opposite side of it. We examined several configuration and ways to put this
extra iron at the bottom half of this magnet. The calculations show that 1 mm thick
iron strip from 180 degree to 360 degree will be adequate to produce the required
compensation. If the strip is put from 225 degree to 315 degree (width = 90 degree)
the thickness required would be 1 cm. This scheme also has easy tunability - one
would simply change the width or thickness (or both) of the iron strip.
General Dynamics is looking5 into placing this extra iron inside the shell (within
the outer diameter of iron yoke) and those schemes have been found to be adequate
according to the calculations carried out by them.
CORRECTING ALLOWED HARMONICS
A 4-fold symmetry has to be maintained while changing (correcting) the values of

the allowed harmonics b2 , b4 , b6 , etc. First we shall consider correcting the geometric
multipoles. They are present at all field level. Then, we shall consider the saturation
induced multi poles which is only a high field effect. However, before one undertakes
the task of removing these systematic effects, sufficiently good statistics must be
obtained by measuring field harmonics in a large number of magnets to separate the
random and systematic variations.
776
Geometric M uItipoles
Geometric multi poles can be corrected by either modifying the coil cross section
or by modifying the iron aperture. Such modifications should be incorporated only
after the mechanics of the manufacturing process and the mechanical dimensions of
cable, etc. are finalized since they may influence the harmonics in the magnet.
1. Modifying Coil Cross section

One way of obtaining a small systematic values of field harmonics in the final
magnets may be to let the coil design program re-optimize the original cross section
to cancel out the measured systematic values in the previous magnets. In such coil
optimization process the computer program can change the dimensions of all wedges
to do, for example, a least square fit optimization to obtain the desired values of field
harmonics. This approach may not necessarily give a solution which is the best for
bringing harmonics within specifications with a minimum change in the mechanics of
the cross section. Therefore, a systematic study of observing the influence of changing
the size of an individual wedge or of more than one at a time in various combinations,
should be useful to steer the optimization process in a controlled direction. The
following study is done on DSX201/W6733C cross section 6 which has all wedges
symmetric in the inner layer but not in the outer layer.
In figure 1, we study the effect of changing the size of individual wedges, one at a
time, on the field harmonics. The original wedge size is changed by ±1 degree in the
steps of 0.1 degree. The pole angle also changes as the size of a wedge changes. The
wedges are counted from the midplane to pole starting from the midplane. In figure 2,
we change the size of 2 or 3 wedges together such that the pole angle does not change.
Since there is only one wedge in the outer layer, the outer layer can not participate
in this scheme. However, there are three wedges in the inner layer. In the first three
cases, only two of these three wedges are changed at a time and the remaining third
wedge is kept at its original value. As the size of one wedge increases the size of the
other would decrease by the same amount. In the last case, all three wedges of the
inner layer are involved. As the size of wedge 1 increases the size of wedge 2 and
wedge 3 would decrease by half the amount of the increase in the size of wedge 1 to
keep the pole angle constant. Obviously one could study many more combinations
on exactly how to change the size several wedges in a coupled manner. In figure 3,
we change the tilt angles of one block at a time (wedge changes accordingly) by ±5
degree in the steps of 0.5 degree. The tilt angles of the blocks closest to the midplane
in the inner and in the outer layer is not allowed to change. The tilt angles 1, 2 and 3
are respectively for blocks 2, 3 and 4 in the inner layer (counting from the midplane
to pole) and the tilt angle 4 is for block 2 in the outer layer.
One can see from these plots that some parameters are more sensitive than others
to produce a large change in one particular harmonic and small change in other
harmonics. For example, in figure 1, the change in the size of wedge 4 produces a
large change in b2 and small change in all other harmonics.
We present the original and re-optimized cross section in figure 4. This cross
section should produce the magnets in which the systematic values of all harmonics
is close to zero provided they are built with the same cable. The change in cable size,
for example due to change in cable insulation, can be incorporated as a perturbation
to this cross section in a pre-determined way to still produce harmonics close to zero.
In this re-optimized cross section we could make all wedges, both in the inner and
in the outer layer, mechanically symmetric. This relaxes the quality control process
during the magnet production.
777
Effect on Field Harmonics of Change in Wedge Size
! -------- Wedge #1 --0- Wedge #2 -----+- Wedge #3 ---0--- Wedge 114!
IS I
0.8
0.6
0.4
b2 b4 o·~t~~~~~~~~~~~
-0.2
-0.4
-5 -0.6
-0.8
-I~-----------------------
-I -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I -I -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I
0.2 0.08
0.15 om
0.1
0.05 0.06
be -O.::L~~~;::~""i~~~~~H:l b8 0.05 .l~:::8:i?;:8::B=t~~..,.I!~8=G:\:pK>-o-<>~

0.04;>;
-0.1
0.03
-0.15
-0.2 0.02
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I -I -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I
Figure 1. The size of all four wedges is changed one at a time by ±I degree to study
the effect of this change on the computed value of field harmonics in prime units. The
pole angle will change in this case.
Effect on Field Harmonics of Changing 2 or 3 Wedges

Pole Angle is Kept Constant
! -------- WI-WZ --O-W2-Wl -----+- WI-Wl ---0--- WI-IW2+W3l/Z!
IS
b2
-I~-----------------------
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I -I -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I
0.2 0.08
0.15 om
0.1
0.06
0.05
be -O.~
-0.1
o.osl.....-=:;~~~~~::-~
b8 0.04
0.03
-0.15
-0.2 0.02
0.01 L -_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I -I -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 I
Figure 2. The size of the three wedges in the inner layer is changed with two or
more at a time such that the pole angle does not change. The symbol WI represents
the wedge 1. The maximum change in the size of any wedge is ±I degree. The effect
of this change on the computed value of field harmonics in prime units is shown here.
778
Effect on Field Harmonics of Changing Tilt Angle
I - - Tilt #1 -G-- Tilt 111. -+---- Tilt #3 ----0-- Tilt #41
IS
0.8
0.6
0.4
b2 b4 0.2~~~~e;g:&c"'8i~~~~~
o?"
.{J.2
.{J.4
.{J.6
.{J.8
-10 L-_~ _____ ~ _ _ __ _IL-_~ __ ~ ____ ~_~
-5 -3 -I -5 -3 -I
0.2 0.08
0.15 0.07
0.1
0.05
0.06
b6 .{J.~t~~~~~j.o,~~~~~~ b8 o·05
0.04
.{J.I
0.03
.{J.15
.{J.2 0.02
.{J.25 '--~--~------~ 0.01
-5 -3 -I -5 -3 -I
Figure 3. The tilt angle of all blocks except those at the midplane is changed one
at a time by ±5 degree to study the effect of this change on the computed value of
field harmonics in prime units.
SC78MM
---~@ L
24"J8""" MM
____
------7
~~
49.71 MM 49,71 MM
Figure 4. The original cross section DSX201/W6733C is shown in the left. Mea-
surements show that this produces non-zero systematic harmonics. The re-optimize
cross section to eliminate these harmonics is shown in the right. This cross section
has all wedges symmetric.
779
2. Modifying Iron aperture
The field harmonics can also be changed by modifying the iron aperture. But
in practice we found that for any reasonable change in iron aperture, the scheme is
mostly effective for changing b2 only and for higher harmonics it is not efficient. To
introduce a change in field harmonics, one can either put some cutout or extrusion
in the aperture or can modify the aperture as a whole, for example by introducing a
little ellipticity in the circular aperture. Both of these changes in the aperture also
bring a change in the iron saturation which must be compensated if the change in b2
is to be kept constant at all excitations. It may be noted that the geometry of the
stainless steel collar may also have to be changed together with the iron aperture.
The only advantage of correcting b2 by modifying the iron aperture would be that it
does not change the coil.
Saturation Induced Multipoles
The saturation in the magnetic properties of the iron brings a change in the field
harmonics at high field. In addition, the coil deformation due to Lorentz forces 4 on
the coil also changes the harmonics at high field. Whereas, the calculations for the
change in harmonics due to iron saturation are fairly reliable the calculations for the
change in harmonics due to Lorentz forces are not as reliable. In addition the change
depends on the gap between the collar and the yoke. In the magnets one would like
to obtain a small values of these harmonics not only at low field but also at high field
by minimizing the combined variation in the field harmonics due to iron saturation
and Lorentz forces. The iron yoke for the SSC dipoles is designed 6 such that the iron
saturation, and in particular the variation in b2 saturation as a function of current,
can be modified by simply changing the size and location of the midplane cutout
in the yoke. One would do that, if need be, in the final design. In the BNL built
magnets, there is a small variation in b2 as a function of current and therefore no
corrective action is required.
REFERENCES
1. P. Wanderer, BNL, Private communication.
2. D. Orrell, SSCL, Private communication.
3. S. Kahn and P. Wanderer, Unpublished reports.
4. R. Gupta, et.al., "A Comparison of Calculations and Measurements of the Field
Harmonics as a Function of Current in the SSC Dipole Magnets", Presented at the
1991 IEEE Particle Accelerator Conference in San Francisco in May 6-9, 1991.
5. H. Gurol, et.al., General Dynamics, Private communication.
6. R.C. Gupta, S.A. Kahn and G.H. Morgan, "SSC 50 mm Dipole Cross section",
Presented at the International Industrial Symposium on Super Collider (IISSC) in
Atlanta in March 13-15, 1991.
780
BARIUM FLUORIDE CRYSTAL CALORIMETER FOR GEM
Ren-yuan Zhu1
Lauritsen Laboratory
California Institute of Technology
Pasadena, CA 91125
ABSTRACT
Barium fluoride (BaF2) crystal calorimeter is one of two options being pursued by
the GEM collaboration as a precision electromagnetic calorimeter. This report summarises
progress of R&D on barium fluoride crystal calorimeter for GEM.
INTRODUCTION
The barium fluoride (BaF2) calorimeter is one of two principal options for electromag-
netic calorimetry in the "Gammas, Electrons and Muons" (GEM) experiment [1]. To
maximize its discovery potential, the GEM detector has been designed to be a precision
lepton and photon detector. A precision electromagnetic calorimeter, as well as a high
resolution muon system, are the two principal ingredients of the GEM detector design.
This report concentrates on expected performance of BaF 2 electromagnetic calorimeter
at the SSC, and progresses made in understanding requirements to the quality of BaF 2 crystals
to be used in constructing calorimeter. Section 2 of this report gives a brief description of the
concept of BaF 2 calorimeter design. The performance of the BaF 2 calorimeter is illustrated
in section 3. The systematic effect on energy resolution from light response uniformity and
radiation resistance are deliberated in sections 4, and 5. The specifications for BaF 2 crystals
are discussed in section 6.
BaF2 CALORIMETER DESIGN
Design
Figure 1 shows the conceptual design of the BaF 2 calorimeter, which consists of two
parts:
• A central barrel calorimeter with an inner radius of 75 cm and an outer radius of 140
cm, covering a rapidity range of 111 I~ 1.32 (30 0 ~ () ~ 150 0 ).
'Work supported in part by U.S. Department of Energy Contracts No. DE-AC03-BI-ER40050.

Plenum Pres" New York, 1992 781
• Two endcaps, located at z = ± 164 em, covering a rapidity range of 1.32 ::; I 11 1::;2.5
(9.4° ::; B ::; 30° and 150° ::; B ::; 170.6°).
The total crystal volume of the BaF 2 calorimeter is 10.6 m 3 , with a total crystal weight
of 51.8 t. Table 1 shows the basic parameters of the BaF 2 calorimeter.
Table 1. Features of the BaF 2 Calorimeter
Detector Barrel Two Endcaps

Rapidity Coverage 1111 ::; 1.32 1.32 :::: 1111 :::: 2.5
Crystal Front /Rear Face (cm 2 ) 3.1 x 3.1 / 5.2 x 5.2 2.3 x 2.3/ 3.1 x 3.1
Crystal Length (cm) 50 50
Crystal NUInber 10,880 4144
Crystal VolUIne (m3 ) 8.4 2.2
Crystal Weight (t) 41.1 10.7
The fine granularity (1111 ~ 11¢ ~ 0.04) and hermetic design of BaF 2 calorimeter provides
a uniform response over the phase space coverage and a small 2.7% dead space for the cable
path of central tracker. In summary, the BaF 2 calorimeter has the following features which
are deliberated in the next section.
• Tilne Resolution: gating time in less than 16 ns (a single beam crossing);

• Position Resolution: ox and oy ~ 1 mm at the front surface of crystals;
• Energy Resolution: (2.0/VE E!) 0.5)%.
Readout
2.5-5cm 0
SOcm
2-3cm -.., "--
Figure 1. Conceptual view of the BaF 2 EM calorimeter.
782
BaF2 CALORIMETER PERFORMANCE
Time Resolution
The intrinsic intensity of the slow scintillation component (600 nsec) in BaF 2 is 5 times
higher than the fast components. Figure 2 shows the emission spectra for pure BaF 2 and
BaF 2 doped with 1% of lanthanum. The peak intensity of the slow component (310 nm) is
reduced by a factor of about five with little change to the fast components (195 and 220 nm).
Since the manufacture oflarge undoped high quality BaF 2 crystals is inherently simpler,
and since the fast and slow scintillation components of BaF 2 peak at different wavelengths,
a UV-selective photo device (optimised for maximum sensitivity in the 220 nm region) is
proposed to partially suppress the slow component. The slow component suppression is com-
pleted by a fast shaper, and by readout samples just before and shortly after the peak of the
pulse. This technique of selecting the fast component does not restrict the dynamic range,
and it allows us to maintain very low noise because of the low capacitance of the vacuum
photodevice.
Figure 2 also shows the spectral sensitivity of a UV-selective 'solar-blind' photomultiplier
(PMT) with a Cesium Telluride (Cs-Te) photocathode (Hamamatsu R3197) and a PMT with
a bialkali photocathode (Hamamatsu R2059). Both PMT's have a synthetic silica (quartz)
windows. The Cs- Te photocathode has a quantum efficiency of 10% around 220 nm, while
the bialkali photocathode has a quantum efficiency of about 18% around 220 nm. It is clear
that the solar-blind photocathode (Cs- Te) is mainly sensitive to the fast scintillation light.
An additional optical suppression of the slow component recently has been achieved by
using new photocathodes, K-Cs- Te or Rb- Te, developed by Hammamatsu [2]. Both cathodes
have about 10% quantum efficiency at 220 nm and provide better suppression of the slow com-
ponent (by around a factor of 2 relative to Cs- Te). Figure 3 shows pictures of the scintillation
light pulses recorded on an HP54111D digital scope using a bialkali photocathode (a and e),
a Cs-Te photocathode (b and f), K-Cs-Te cathode (c and g) and a Rb-Te cathode (d and h).
The rise time of the scintillation light pulse in these pictures was completely dominated by the
2.3 nsec rise time of the PMT's. On an expanded scale (a, b, c and d) the full width at half
maximum of the fast scintillation light is measured to be 4-6 nsec. The optical suppression
factors (F/S) for the slow component, defined as the number of photoelectrons in the fast
30
1.0
~
>-
I-
25 ~
U5 0.8 >-
0
Z
w
I-
20 aJ
Z 0.6 (5
W 15 u::
LL
> W
~ 0.4 10 ~
--l :::>
W I-
a: 0.2
5
Z
«
:::>
a
0
Figure 2. BaF 2 scintillation spectra and PMT quantum efficiencies.
783
components (F) divided by the number of photoelectrons in the slow component (S), are also
shown in the figure.
Position Resolution
The position of an electromagnetic shower without longitudinal sampling, i.e. the coor-
dinate of the impact point of an electron or a photon on the front surface of an electromagnetic
calorimeter, is usually measured by using the center of gravity method. The position reso-
lution of a calorimeter thus depends on its structure, especially the lateral cell size. For a
calorimeter organized in pointing towers, the position resolution as a function of energy can
be parametrized as:
3
cz (mm) = ,jE e°.4D (1)
where E is the energy of the particle being measured in Ge V, and D is the cell size in radiation
lengths.
The optimum lateral segmentation is :::::: 0.04 X 0.04 (6 TJ X 6¢) at the SS C as in the BaF 2
design. This segmentation results a position resolution, calculated according to Eq. 1 for a 20
GeV electron or photon, is 1.2 mm, which is comparable with what has been measured with
L3 BGO calorimeter [3].
Energy Resolution
The energy resolution of an electromagnetic calorimeter can be parametrized as:
(2)
The contributions of each term to the resolution are:
Figure 3. BaF 2 scintillation Light pulse observed by a bialkali cathode (a and e), a Cs-Te
cathode (b and f), a K-Cs-Te cathode (c and g), and a Rb-Te cathode (d and h).
784
•80 is the contribution from electrical noise, summed over a few Moliere radii around the
maximum of the lateral shower distribution;
• 81 is the contribution from the photoelectron statistics;
• the systematic term b has three contributions:
b2 = bn 2 + bG 2 + be 2 (3)
bG represents the geometry effect, including shower leakage at the front, side and
back of the detector and inactive material between cells;
- b n represents physics noise, including fluctuations of the shower and uniformity of
light response etc.;
- be represents intercalibration error.
At low energy, the dominant contribution to the energy resolution is the noise term
(ao), which decreases quickly with increasing energy. The sampling term (at) dominates in
the range of medium to high energies until a high energy limit is reached, where the systematic
term (b) dominant.
In this section, only systematic term bG is discussed. The systematic terms of b n (light
response uniformity) is deliberated in sections 4 and 5.
Light Yield: al It is not very difficult to build a homogeneous electromagnetic calorime-

ter with a small al term. A photoelectron (p.e.) yield better than 10 p.e./MeV would be
enough to provide an al = 1%. Figure 4 shows the measured 137Cs spectra obtained from
the same BaF 2 sample by using four PMTs with different photocathodes (bialkali, Cs- Te,
K-Cs-Te and Rb-Te) for two different gate widths (55 ns and 2 JLs). By comparing the spec-
tra, it is clear that the K-Cs- Te cathode has good quantum efficiency (better than Cs- Te and
Rb-Te) in detecting the fast components from BaF 2 • Measurements at Caltech show that
more than 50 p.e./MeV are obtained from typical 25 cm long BaF 2 crystals by using K-Cs-Te
photocathodes.
o 100 0 50 o 50 o 100
r ,.
..!'o:. Bialkali
,. ~ J\
": Cs-Te
-.... ,.:.:;.
~. K-Cs-Te
..
;.
\ Rb-Te
~
\ :~
...-...; "'¥' '~ ,
'\
.~. ~ '\
\
\
\
\
a) c)
:\ K-Cs-Te
.,.:
1\
.. Rb-Te
\/\
: ~ ..It
.
. \
f) • ~
o 500 o 2000 100 0 200

Channel Number
Figure 4. 137Cs spectra measured with 4 PMTs with different photocathodes for two different
gate widths: 55 ns (a-d) and 2 J-Ls (e-h).
785
The light yield of 50 em long crystals has been measured at Caltech with 137 Cs ,-ray
source and at Fermi Lab with electron beam. For production quality BaF 2 crystals, photoelec-
tron yield was measured to be 20 to 50 p.e./MeV by using K-Cs-Te or Rb-Te photocathode
with good light response uniformity. This corresponds to an al term of about 0.4-0.7%.
Electrical Noise: ao Because of its small capacitance (10 pf) and small dark current
(0.1 nA), it is not difficult for the R4406 vacuum phototriode and the front electronics to
provide an electrical noise level of 2,000 electrons per channel [5). This means an ao term
of 2-5% for 3 X 3 crystal readout, if existing R4406 is used. The ao term can be further
reduced if the proximity focused 5 stage phototube [4) is used. Since it has a gain of >30
under .8 tesla magnetic field, the ao term can be further reduced to 0.4-1%. The electrical
noise introduced by channel-to-channel correlations, however, must be reduced by carefully
implementing the electrical isolation between the readout channels.
Geometry Effect: hG To reduce the systematic effects caused by shower leakage and
inactive material, a precision electromagnetic calorimeter must be designed to contain nearly
the complete electromagnetic shower, over the whole energy range. Since electromagnetic
shower physics is well understood, a complete GEANT simulation would be able to predict
the resolution of a calorimeter with proposed geometry configuration. To reduce the term
b G , an electromagnetic calorimeter must be designed with enough length and have minimum
dead material in the front, between cells and, especially inside the calorimeter.
A detailed GEANT simulation was carried out to estimate the effect of shower leakage
and non-active material for the BaF 2 design [6). Figure 5 shows a prototype of BaF 2 matrix
used in this simulation, which consists of 121 (11 X 11) BaF2 crystals with the proposed
size: 3 X 3 cm2 at the front, 5 X 5 cm 2 at the back and 50 em long. Effects included in the
simulation are:
Figure 5. BaF 2 prototype used in simulation.
786
• 250 p.m carbon fiber wall between crystals;
• shower leakage because of summing a limited number (3 X 3 or 5 x 5) of crystals; and
• 0.30 radiation lengths of aluminum, representing the beam pipe, tracker, and carbon
fiber mechanical support, at the front of the BaF 2 array.
Particles were shoot uniformly at the front surface of the center crystals of the array.
The energies deposited in each crystal, in the carbon fiber walls between crystals, in the
aluminum and leaking out sideways were recorded. The result of this simulation for electrons
with different energies (5, 10, 100 ad 500 GeV), in terms of (1' of the peak, full width at half
maximum (FWHM) divided by 2.35, is listed in Table 2.
Table 2. Energy Resolution (%)
E (GeV) 5 10 100 500

Electrical Noise 0.4 0.2 0.02 0.004
Photoelectrons 0.2 0.14 0.045 0.02
GEANT 0.60 0.43 0.31 0.29
Intercalibration 0.40 0.40 0.40 0.40
Total 0.85 0.63 0.51 0.49
A further study was carried to look the systematic effect of carbon fiber wall. By
varying the thickness of carbon fiber walls from 0 to 250 I'm, the resolution obtained showed
no observable difference [6]. This indicates that the systematic effect of support structure is
negligible.
Sununary of the Energy Resolution Table 2 sununarizes the BaF 2 resolution, in-
cluding the contributions from electrical noise, photoelectron statistics, intrinsic resolution
from GEANT simulation and the intercalibration. A precision of intercalibration of 0.4% is
assumed. Note, in this simulation the light response uniformity was assumed to be under
control. The real effect of light response uniformity is discussed details below. The result of
8
a) L3 BGO Data 4
2.0%/v'Ee O.5% GEANT Simulation
6
2.0%/v'Ee O.5% 3
.........
~
........ 4 2
r.:l
'-....
r.:l
'0 2 1
E
Figure 6. Energy Resolution of L3 BGO calorimeter (a), measured at CERN test beams with
4000 crystals, and BaF 2 calorimeter (b), calculated with GEANT simulation. The solid curves
represent a simple parametrization of 2%/..JE $ 0.5%.
787
the energy resolution is shown in Fig. 6b. It can be parametrized as 2%/VE E9 0.5%, which
is also shown in Fig. 6b as a solid line.
As a comparison, Fig. 6a shows the energy resolution measured with 4000 BGO crystals
(half barrel) in a CERN test beam [3]. In the energy range beyond 20 GeV, the dominant
contribution to the energy resolution is the systematic intercalibration uncertainties. The
resolution of the L3 BGO calorimeter may also be parametrized as 2%/VE E9 0.5%, which
is also shown in the Fig. 6a as a solid line.
LIGHT RESPONSE UNIFORMITY: b n
Experience with the L3 BGO, and other precision crystal calorimeters, has shown that
light response uniformity at the level of several percent over the length of the crystal (except
for the first few and the last few radiation lengths), is important to maintain the resolution.
This uniformity is also needed to maintain good linearity over a large dynamic range, e.g.
from ",10 GeV up to the TeV range at the SSC.
Effect of Light Response Uniformity
The consequence of light response uniformity has been studied with a GEANT simulation
[6], assuming the effect of non-uniform light responses can be parametrized as a normalized
function:
Y = Y 25 [1 + 6(Z/25 - 1)] (4)
where Y25 represents the light yield at the middle (25 cm) of a BaF 2 crystal, 6 represents the
uniformity of the light response of a BaF 2 crystal, and Z is the distance from the small end
of the crystal.
The effect of 6 values of 5%, 10%, 15%, and 20% has been simulated. Fig. 7 shows the
Effect of Light Uniformity (3 x3)
-
........
~
r::
..... 92.5
0
95.0
8c 10 Gev
.....
CJ
@ 20 Gev
III
s... o 50 Gev
rz.. # 100 Gev
I>. 90.0
tID • 200 Gev
s... + 500 Gev
w
r:: 87.5
r.::I
-
........
~
It:I
C'J
N
1.0
8c 10 Gev
@ 20 Gev
o 50 Gev
# 100 Gev
"
::::!'l 0.5 • 200 Gev
~ + 500 Gev
0.0
-10 0 10 20
Deviation o (%)
Figure 7. Effect of light uniformity predicted with GEANT simulation.
7BB
result of the simulation for sum of 3 x 3 crystals. As seen from Fig. 7, the response of energy
deposition decreases when 5 increases, while the energy resolution degrades. It is clear from
the figure that it is vital to maintain the light uniformity within 5%.
Light Response UniforIllity of 50 CIll Long Crystals
In order to achieve light uniformity of a 50 cm long crystal-pair which make up the

prototype BaF2 detector, two important technical difficulties must be overcome:
1. Find a glue with good UV transmission down to at least 200 nm. This is needed to
optically couple the two crystal pieces together, as well as to couple the crystal-pair to
the photodevice.
2. A technique of crystal surface treating: wrapping, or UV-reflective coating, to combine

good light collection efficiency with overall uniformity.
Transmittance of Grease and Glue

100
/
,.-...
~
-----
(J)
C)
l=1
....,cd
....,
3. Silicones UISC-60M
..... 50 4. KEI03 Glue
S
rn I 5
5. PATE 431B
6. SILGEL 601 Glue
l=1 I
I
cd I 7. Dow Corning X3-6141A
r... iI I
8. Dow Corning X3-6141B
E-< I·
iI 9. Dow Corning X3-9114
I 10. Dow Corning 710
11. Dow Corning Q2-3067
0
200 300 400
Wavelength (nm)
Figure 8. The Transmittance of 500 /Lm thick layers of various grease and glues.
KEI03 Glue While the best UV-transmitting optical coupling material (down to 190
nm) is Dow Corning 200 Silicone fluid, the best viscous grease, which can be used for good
long-term joints if set up and used with care, is GE Silicone based UISC 600M. Although
grease of this type have been used successfully on large BaF 2 crystals in experiments: by
Woody at BNL, and by the TAPS Collaboration [9], there is no doubt that a glue is more
secure.
Recent measurements by Kobayashi et ai. at KEK [7] indicated that KE103 (an RTV
glue) [8] is a candidate ofUV transparent glue. Note, KE103 was also measured to be radiation
hard up to a level beyond 10 7 Rads [7]. The transmission results were measured at Caltech
with hard-cured samples of KE103, as summarized in Table 3 below. Thin joints of KE103
789
were made by pressing two pieces of quartz together (typically few tens pm), and the UV
transmittance was measured, normalized to the results obtained with Dow Corning 200 fluid
between two similar pieces of quartz. The results for the GE grease UISC 600M are shown
for comparison. As shown in the table, the transmission of KE103 adhesive is nearly as good
as the Dow Corning fluid down to 200-205 nm, and it shows good transmission down to 190
nm.
Table 3. UV transmission (%) relative to Dow Corning 200 fluid, for thin layers of KE103
RTV and GE UISC 600M grease
Wavelength (rnn ) 190 195 200 205 210 215 220 225 230
KE103/1 97.2 87.6 90.5 94.6 97.8 98.3 99.9 99.4 99.4
KE103/2 83.7 88.5 95.3 97.6 98.8 99.8 99.4 99.9 99.8
UISC 600M 90.0 94.4 96.4 96.2 97.5 97.3 97.5 97.7
However, it should be pointed out that if the layer of KE103 glue is too thick, e.g. 500
J-Lm, the joint is no longer transparent at UV. Figure 8 shows the transmittance measured for
500 J-Lm thick layers of various adhesives. The cut-off at 220 nm of KE103 shows clearly that
a bad joint of crystal pair would destroy light response uniformity.
Crystal Surface Treating A simple wrapping technique, in which the region near the
phototube is covered with aluminized mylar painted black to suppress direct light was used
for 49 test crystals. The recipe of our wrapping is:
1. 3 to 4 layers of TETRATEX Porous PTFE film (1.5 mil thick) covers the small end to
24.5 cm from the small end;
2.0
Y/Y 25 = 1 + <5(X-25)/25
<5 = 0.4% X2/DoF =0.2
Q)
.....:>
....,
cO
......
Q)
~ 0.5
row: 2 col: 2
0.0
o 10 20 30 40 50
Distance from Small End (em)
Figure 9. Normalised pulse height obtained with a 50 cm crystal pair measured with muon
beam at Fermi Lab.
790
2. 2 layers of aluminum foil (0.5 mil thick) covers 24.5 cm to 35 cm;
3. 2 layers of aluminized mylar painted black (1 mil thick) covers the rest of the crystal,
i.e. from 35 cm to the large end.
Similar techniques have been used to obtain good light uniformity in BaF 2 crystals by
B. Winstein et al. at Fermi Lab. By adjusting the wrapping, a uniformity at the level of ~ 5%
or better over the length was obtained for 50 cm long BaF 2 crystals with a good glue joint.
A result measured at Fermi Lab with a muon beam hitting a 50 cm long BaF 2 crystal pair
transversely is shown in Figure 9.
Note, good uniformity was obtained in spite of the fact that there were substantial
variations in the light output of the first crystal batches produced in 1990. It is also known
that aluminum foil in contact with air is not a high-efficiency reflector for 200 nm light.
For the final BaF 2 system, one of the main R&D issues is investigation of a UV-reflecting
coating, such as MgF2 (used routinely for BlCH Counter mirrors at 170 nm). The MgF2 would
be applied in a vacuum tank, immediately followed by a sealing layer of quartz. A masked
pattern of reflector would be used to achieve the uniformity. This technique, which is not
available during small-scale prototyping at moderate cost, would improve our light collection
efficiency as well as the uniformity during mass production.
BaF2 RADIATION RESISTANCE
All above discussions have not touched an important issue of BaF 2 calorimeter for the
SSC: the radiation resistance. In this section we discuss the phenomena of radiation damage
100
50
,--...
~
Q)
<:>
.::oj 100
-+-'
Os
rn
.::oj
H
E-<
50
200 400 600 800

Wavelength (nm)
Figure 10. Transmittance spectra, obtained from 25 cm long BaF 2 crystals from SIC/BGRI
(a) and Merck (b), before (1) and after 60CO '"(-ray irradiation with dosage of 100 (2), lk (3),
10k (4), lOOk (5) and 1M (6) Rads.
791
of BaF 2 crystals. For more detailed discussions in this issue, see, for example, papers and talks
presented in BaF 2 Workshop at Shanghai [10], May 1991, and in GEM BaF2 panel meetings
at SSCL [11,12], Beijing [13] and Shanghai [14], December 1991.
Radiation Resistance of 25 crn long Crystals
The first batch of 25 cm long BaF 2 crystals delivered by SIC/BGRI in 1990, however,
show substantial radiation damage, as measured at BNL [11] and Caltech [12]. Figure 10
shows the transmittance spectra measured for two 25 cm long crystals before and after 60Co
irradiation with dosage in step of 100, 1k, 10k, lOOk and 1M Rads. Figure lOa shows that the
transmittance of a crystal from SIC/BGRI is reduced from 70-80% to 10-20 % at 220 nm
(fast scintillation components) after 10 kRads of 60Co i-ray irradiation. The saturation after
100 kRads is observed. Similar reduction of transmittance was observed for a crystal from
Merck, as shown in Fig. lOb. The situation of unsatisfactory radiation resistance of large size
BaF 2 crystals was first extensively reviewed in May, 1991, in BaF 2 Workshop at Shanghai,
and then in early December, in GEM BaF 2 panel meetings at SSC Laboratory, Beijing and
Shanghai.
It is interesting, however, to note that the spontaneous recovery of radiation damage
has very long time constant at room temperature, in the wavelength range of fast scintillation
100
Days after
(a) SIC 39 Days after
50 BGRI 28 Days after
11 Days after
o Day after
100
Days
Days
50 28 Days
11 Days
o Day
200 400 600 800

Wavelength (nm)
Figure 11. Transmittance spectra, obtained from 25 cm long BaF 2 crystals from SIC/BGRI
(a) and Merck (b), before and after 1 MRads of 60 Co i·ray irradiation. Also shown are spectra
after 11, 28, 39, 69 and 108 days after irradiation.
792
components. Figure 11 shows the recovery of transmittance of the SaIIle two crystals, as
shown in Figure 10. The spectra shown were obtained before and after 1 MRads 60Co ,-ray
irradiation with a time interval between the irradiation and measurements ranged from 0 to
108 days. For each crystal, the spectra obtained from 0 to 108 days after irradiation are
ahnost undistinguishable between 200 to 400 nm.
Looking into the details, Fig. 12 shows the transmittance data measured at 200 and 220
nm as function of time after irradiation for these two crystals, in an expanded scale. It is
clear that the spontaneous recovery at these two wave lengths has very long time constants,
especially for the crystal from SIC/BGRI. This observation indicates that a stable detector
may be constructed by using BaF2 crystals even with some level of radiation damage. The
key issue of a precision BaF 2 calorimeter thus is to control the damage before saturation.
It is also interesting to note that the radiation damage of BaF2 does not have strong
dependence on the rate of irradiation. Figure 13 shows two sets of transmittance spectra of
a 25 cm long BaF 2 crystals from SIC/BGRI with different dose rate. The crystal was first
irradiated up to 1 MRads with fast dose rate. It was then annealed in an oven under 500°C
for three hours, so that it was fully recovered. The crystal then was irradiated again with a
slow dose rate. Figure 13a shows the measured transmittance spectra with dose rates of 1 (for
100 Rads), 11 (for 1k and 10k Rads) and 24 (for lOOk and 1M Rads) Rads/sec. Figure 13b
shows the measured transmittance spectra with dose rates of 0.03 (for 100 Rads), 0.3 (for
1k Rads), 3 (for 10k Rads) and 20 (for lOOk Rads) and 15 (for 1M Rads) Rads/sec. The
small difference between the spectra obtained after the SaIIle integrated dosage, but with 30
times different dose rate, can be attributed to the uncertainty in measurement of integrated
15
g t t t i-
t
nrn
<I) a) SIC/BGRI
()
r:: 10
.... 12.5
ttl
·s
~
{/l
r::
ttl
r-.
E-<
10.0
7.5 b) MERCK
o 50 100
Time after Irradiation (Day)
Figure 12. Transmittance data at 200 and 220 nm, obtained from 25 cm long BaF 2 crystals
from SIC/BGRI (a) and Merck (b), are ploted as function of time after 1 MRads of 60Co
,-ray irradiation.
793
dosage. In general, this measurement indicates that the damage of BaF 2 has almost no rate
dependence. The full recovery of the annealing also indicates that there is no permanent
damage in BaF 2 crystals.
Since May, 1991, substantial progress have been made at SIC and BGRI (see below for
the details). Figure 14 shows the transmittance spectra measured for two 20 cm long crystals
before and after 60Co irradiation with dosage in step of 100, 1k, 10k, lOOk and 1M Rads.
Both crystals were produced recently at SIC with improved processing technique. Figure 14a
shows that the transmittance of a crystal is reduced from 70-80% to 50-60% at 220 nm (fast
scintillation components) after 10 kRads of 60Co I-ray irradiation. The saturation after 100
kRads is also observed. Similar reduction of transmittance was observed for another crystal,
as shown in Fig. 14b.
100
~
50
~
Q)
C)
.::
....,ro 0
"6 rn 100
.::
ro
1-0
E-
50
o
200 400 600 800
Wavelength (nm)
Figure 13. Transmittance spectra of a 25 cm long BaF 2 crystals from SIC/BGRI before and
after 60Co I-ray irradiation, with dosage of 100, 1k, 10k, lOOk and 1M Rads for fast (a) and
slow (b) dose rate.
Note, both crystals were irradiated in Shanghai for 1 MRads before delivery. They were
annealed at Caltech. Systematic measurement on their radiation resistance was carried out
at Caltech by using a radiation facility at JPL. The dose rate is identical to the slow rate
indicated in Fig. 13b. Comparing with Fig. lOa, it is clear that these two crystals have much
better radiation resistance.
As discussed before, the chinese scientists at BGRI, SIC and Tongji university have made
significant progress in understanding the radiation damage mechanism of BaF 2 • They are
confident in providing solutions for the problem. It is expected that 10% loss of transmittance
may be achieved for 25 em long crystals.
794
Consequences of Radiation Damage
There are three consequences in BaF 2 radiation damage:

1. radiation introduce color centers which cause absorption of scintillation light and thus
reduce the light attenuation length;
2. radiation introduce phosphorescense which causes random background;
3. there might be a damage of the scintillation mechanism, and thus reduce the light yield.
Most early radiation damage studies were concentrated on the light absorption, i.e. the
transmittance loss. Recent studies [11,16], however, reported observations of radiation intro-
duced phosphorescense, which in many cases prohibits measurements of light yield of large size
BaF 2 crystals by using a conventional 1 37 CS 7-ray source. Although causing inconvenience for
laboratory study, the phosphorescense can be treated as a random noise. Since its energy scale
is very low, particularly since it is mainly in wave lengths longer than the fast components,
we do not expect phosphorescense would degrade the precision of BaF 2 calorimeter.
It, however, is not clear if the radiation causes damage to the scintillation mechanism
of BaF 2 crystals. Controversial results were reported in literature, e.g. [15,17], by comparing
measured loss of light yield with simulated result of radiation-introduced decrease of light
collection efficiency. It should be pointed out that the surface effect is not negligible in light
collection simulation, and it is difficult to simulate. To reduce surface effect, test result of
large samples are prefered in comparison with simulation.
By using a ray-tracing program originally used in BGO radiation damage study [18], a
preliminary study done at Caltech showed that, to the first order, the measured loss of light
(b) 3.3 2 x 20 X 4.0 2 (em)
200 400 600 800

Wavelength (nm)
Figure 14. Transmittance spectra of two 20 cm long BaF2 crystals, recently produced at SIC,
before (1) and after 60Co 7-ray irradiation with dosage of 100 (2), 1k (3), 10k (4), lOOk (5)
and 1M (6) Rads.
795
yield can be explained by radiation introduced decrease of light attenuation length [12].
One important input for a ray-tracing program is the light attenuation length which
usally is a function of wavelength. Figure 15a shows refractive index of BaF 2 crystal as
a function of wavelength [19]. The theoretical transmittance of BaF 2 crystals, assuming
the crystal has infinite light attenuation length, is shown in Fig. 15b, taking into account of
multiple bouncings of light between two faces. It is clear that at the UV edge (",200 nm), both
the refractive index and the light attenuation length have a stronger wavelength dependence.
Note, if a measurement is done in air, result below 200 nm is usualy not accurate because of
the UV absorption of the air.
By using following formula, we calculate light attenuation length corresponding to a
measured transmittance (TnI) for a 25 em long BaF 2 crystal:
(5)
where T. is the theoretical transmittances with infinite light attenuation length (Fig. 15b).
Using a ray-tracing program [18], incorporating correct geometry, light attenuatIon
length of the fast component at 220 nm (l), 2.5 em mass attenuation length of 0.662 Me V '"(-
ray from Cs source in BaF 2 , light collection efficiency can be calculated, and can be compared
to measurements. Table 4 shows calculated light collection efficiencies (17) at 220 nm before
and after irradiation, with dosage of 100, lk and 10k Rads, for two 25 em long BaF 2 crystals
1.60
a
1.55 BaF 2 Refractive Index
1.50
1.45
b
93
~ 92
E-< BaF 2 Transmittance
91 (No Light Absorption)
90 ~~ ~~
_ _L I __ L-_L-~~~__ I~_
200 400 600 800
Wavelength (nm)
Figure 15. Refractive index (a) and theoretical transmittance (b) with infinite light attenua-
tion length of BaF 2 crystal as functions of wavelength.
796
Table 4. Comparison of Measured and Calculated Light Yield
Crystal Dose (Rads) T 220nm l (cm) 1] (%) 1]n Data

0 84.9 354 44 1 1
SIC/BGRI 100 78.0 162 35.4 0.80 0.81
1,000 52.1 45 16 0.36 0.32
10,000 16.5 15 3.0 0.07 <0.20
0 75.9 137 33 1 1
Merck 100 69.2 91 27 0.81 0.79
1,000 49.0 40 14 0.42 0.30
10,000 20.4 17 4.0 0.12 <0.23
from SIC/BGRI and Merck. The light collection efficiency normalized to before irradiation
("7n) and corresponding data measured at Caltech are also shown in the table.
It is clear frOIn Table 4, that, to the first order, the measured loss of light yield can
be explained by radiation introduced decrease of light attenuation length. Further work is
needed to refine the program for an accurate prediction. Note, the data measured after 10
kRads is not accurate because of phosphorescense background and severe absorption of these
two crystals.
100
(a)
~
-i-
---
~
E-<
80 -i-
-i-
60 -i-
-i- -i-
-i-
(b)
~ 200
S -i-
---
()
100
-i-
-?- -i- -i-
a
100
~ 80
~ 60
;..;
~ 40
20
Dose (Rads)
Figure 16. Transmittance (a) and light attenuation length (b) at 220 nm and the light yield
(c) measured with 2 different gate width for a 20 cm long BaF 2 crystal from SIC are shown
as function of dosage.
797
The BaF 2 crystals recent produced by SIC has much less absorption and phospho-
rescense, we can thus measure light yield with l37Cs source_ Figure 16 shows the trans-
mittance (a) and light attenuation length (b) at 220 nm, as function of dosage. Also shown
in the figure is the normalised light yield (c) of fast comonents and total, measured with gate
of 30 ns and 2 J1,S respectively, as function of dosage. The fact that we observe saturation in
all these measurements after 100 kRads indicates that the light loss indeed is caused mainly
by absorption.
Table 5 lists theoretical transmittance (T .uf, i.e. Fig. 15b), specification from BaF2 col-
laboration without radiation damage requirement (T.peo ) and corresponding light attenuation
length (t.peo). Assuming that a 5% light uniformity may be achieved with an adequate light
attenuation length (t.pel) at 220 nm of longer than 100 cm, Table 5 shows that a transmit-
tance (T.pe!) of 10% less than original specification of BaF2 collaboration may be allowed.
Note, the light attenuation length specification of TAPS collaboration is 75 cm [20]. However,
the effect on light uniformity of BaF2 crystals with t = 100 cm must be further studied with
simulation and measurements in technical proposal stage.
Table 5. Transmittance and Light Attenuation Length of BaF 2 Crystals
A(nm) T. u ! (%) T.peo (%) t.peO (cm) T.pel t.pel

200 90.6 75 130 66 79
220 91.1 80 190 70 95
550 92.7 85 290 77 135
BaF2 SPECIFICATIONS
The specifications of the BaF 2 crystals finally will be accepted for GEM BaF 2 calorimeter
are summarized in this section. The dimensional tolerances, especially for a 50 cm long crystal
pair, the UV transmittance and radiation resistance are the three vital requirements.
The strigent dimensional tolerartces is required to achieve a thin glue joint, e.g. few
tens of J1,m of KEI03, and thus maintain the light uniformity. Good UV light transmittance
:through the crystal is vital to guarantee adequate light attenuation length for a uniform light
response, and thus maintain the intrinsically high resolution of the BaF 2 calorimeter. We
have specified the transmittance requirement in terms of the minimum fraction of the light
passing through a 25 cm long BaF 2 crystal at specified wavelengths. For simplicity, these
specifications include ",8% loss at two interfaces between air and BaF 2.
The radiation resistance requires that after a saturated radiation dosage the BaF2 crys-
tals have more than 100 cm light attenuation length for the fast component at 220 nm. We
thus propose the following specifications:
• DiInensional tolerances are dictated by the need to mount the crystals safely in the
support structure, with a minimum of dead space between the crystals:
from -300 J1,m to 0 J1,m in the transverse dimensions (typical deviations from nom-
inall00J1,m)
from -400 J1,m to 0 J1,m in length
- less than 50J1,m in planarity of all faces after coupling two 25 cm long half crystals .
• Transparency before irradiation, specified in terms of the minimum fraction of the

light at specified wavelengths which passes through 25 cm of crystal length:
798
- ~ 75% at A = 200 nm
- ~ 80% at A = 220 nm
- ~ 88% at A = 550 nm
• Transparency after 1 MRads irradiation, specified in terms of the minimum frac-
tion of the light at specified wavelengths which passes through 25 cm of crystal length:
- ~ 66% at A = 200 nm
~ 70% at A = 220 nm
- ~ 77% at A = 550 nm
ACKNOWLEDGEMENTS
I am grateful to many useful discussions with Drs. P.J. Li, H. Newman, S. Suzuki,
Z.Y. Wei, C. Woody and Z.W. Ying.
References
[1] GEM Collaboration, Letter of Intent to the SSCL, November 30, 1991.
[2] The K-Cs-Te cathode has been implemented in R4406 triode for BaF 2 readout. The
Rb-Te cathode has been implemented in R4480 PMT for BaF2 readout. Both tubes use
quartz window and are commercially available.
[3] L3 Collaboration, Nucl. Instr. and Meth. A289 (1990) 35.
[4] Hamamatsu, Proposal to Develop a Remote Processed Prozimity Focused Phototube, June
19, 1991.
[5] The ATT preamplifier has 1200 electron noise for an input capacitance of 10 pf. See M.
Levi, in Symposium on Detector Research and Development for the SSC, Fort Worth,
Texas, October 15-18 (1990) 3.
[6] R.Y. Zhu, Physics Requirements to the Precision of Electromagnetic Calorimeter at SSC,
GEM TN-91-06, August 1991.
[7] M. Kobayashi et al. Transmittance and Radiation-Resistivity of Optical Glues, KEK
Preprint, 1991; Also see M. Kobabyashi et al., KEK Preprint 90-130, October, 1990.
[8] KE-103 glue is manufactured by Shin-etsu Chemical Co., Japan.
[9] W. Doering et al., Total Absorption Photon Spectrometer (TAPS) Collaboration Report
No.3, March, 1988.
[10] P.J. Li et al., in these proceedings.
[11) C. Woody, A Report on Radiation Damage in Barium Fluoride to the GEM Collaboration,
presented in GEM BaF2 Panel meeting, December 4th (1991).
[12] R.Y. Zhu, On Radiation Damage of BaF2 , presented in GEM BaF 2 Panel meeting, De-
cember 4th (1991).
[13) G. Chen et al., in these proceedings.
[14) L.M. Wang et al., The Optical and Radiation Damage Properties of Barium Fluoride
Crystals, talk presented in Shanghai BaF 2 Workshop, Shanghai, May 1991.
L.Y. Chen et al., The Radiation Damage of BaF2 Crystals, talk prepared for GEM BaF 2
Panel meeting, Shanghai, December 1991.
[15) C. Woody et al., Radiation Damage of Barium Fluoride Crystals, paper presented at
IEEE Nuclear Science Symposium, Santa Fe, New Maxico, Nov 5-8, 1991.
799
[16] Z. Wei et al., A Study on Radiation Damage of Barium Fluoride Crystals, to be submitted
to Nucl. Instr. and Meth ..
[17] A. Murakami et al., Nucl. Instr. and Meth. A301 (1991) 435.
[18] The ray-tracing program was originally written L.Gatingnon and D. de Lang of University
of Nijmegen, and was substantially modified by R.Y. Zhu in 1986.
[19] I. H. Malitson, J. of Optical Society of America Vol 54 No.5 (1964) 628.
[20] R. Novotny, Performance of BaF2 -Crystals, (TAPS) Collaboration Report No. 10, Febru-
ary, 1990, and No 10/1, October 1990.
800
RADIATION DAMAGE IN LARGE BARIUM FLUORIDE CRYSTALS
P.J. Li, Y.Y. Xie, Y.L. Zhao and Z.W. Yin
Shanghai Institute of Ceramics

Chinese Academy of Sciences
Shanghai 200050, China
ABSTRACT
The radiation damage of barium fluoride (BaF2) crystals was investigated through the
changes in their optical transmission before and after -y-ray irradiation. The correlation of
radiation damage with crystal growth conditions revealed that the radiation damage is im-
purity and defect related. As impurity is concerned, when the contents of cationic impurities
decrease to a certain tolerated level, oxygen and hydroxyl become most harmful. High density
of defect such as scattering centers in crystal also resulted in radiation damage. Thanks to re-
cent progress in BaF 2 crystal growth, an encouraging result of radiation test on large crystals
has been achieved. On the mechanism of radiation damage, it is likely that the oxygen ion
came from dissolved oxygen during irradiation is the cause of 190-250 nm absorption band
and the formation of F color center is the origin of 500-600 nm absorption band. Another
hypothesis by L.Y. Chen et al. suggests that the formation of U center after irradiation is
responsible for the VUV absorption at 204 nm and IR absorption at 793.75cm-l respectively.
INTRODUCTION
Considerable attention has been devoted to BaF2 crystal since the discovery of the fast
component in its scintillation light [lJ. BaF 2 is the fastest scintillator known with its fast
decay time of 600 ps at 220 nm wavelength, and has long been believed to be one of the most
radiation resistant inorganic scintillators. In addition, BaF 2 has excellent time resolution,
good detection efficiency and is not hygroscopic. All these advantages make BaF 2 favourable
for the applications involving very high counting rates and severe radiation such as at the
SSC.
After irradiation with high energy particles or other radiation sources, the coloration
occurred in BaF 2 crystal resulting in the reduction of both its optical transmission and scin-
tillation light output which is so called radiation damage. Therefore the radiation damage in
BaF2 crystal becomes an extremely important problem for its applications where the radiation
dosage is rather high.
Although some research work [2J have been done in understanding the mechanism of
radiation damage in BaF2 crystal, many points still remain unclear. The purpose of present

article is to investigate the mechanism of radiation damage in BaF 2 crystals, especially for
large size crystals from summarizing the results of various radiation tests with the hope to
help improve the radiation hardness of BaF 2 crystal.
EXPERIMENT
The radiation hardness of BaF 2 crystals were investigated through the changes in their
optical transmission before and after '"(-ray irradiation. Two categories of samples were used.
One category samples are in truncated pyramidal shape measuring 3x3, 4x4, 25 cm3 , which
were prepared for the first prototype of GEM detector for SSC project. The other is 2.54cm
dia.x 2.54cm long cylinder. All these samples were cut from ingots grown in a graphite
crucible in a vacuum furnace at 10- 5 torr level, then lapped and polished.
The irradiation of samples were carried out with a 60Co source at Shanghai Nuclear
Research Institute, Chinese Academy of Sciences. A Shimadzu UV-265 spectrophotometer
was used to measure the optical transmission of small samples. For crystals more than 10 cm
in length, a home made large compartment which has been cross-checked was used.
The influence of impurities on radiation damage
A radiation test was performed on five samples a,b,c,d and e, each 2.54 cm long, taken
from various places of a 2.54 cm dia. X 30 cm long cylindrical ingot along the longitudinal
axis. (see left bottom of Fig. 1) The radiation dose is 1x10 5 rad. The transmission spectra
for all these samples shown in Fig. 1 indicate that before irradiation all of them except sample
e have good transmission with similar shapes in the region from visible to UV. However,
after irradiation all samples were coloured in different purple tint and consequently various
decrease occurred in their optical transmission that means the radiation damage for each
sample is not identical. A distinguishable tendency is that along the longitudinal axis of
ingot, the radiation damage is gradually increased. At the top corresponding to the end of
crystal growth, it was even in dark purple and sharp absorption band was observed. Similar
100n-------,-------~------r-----~
z
o
C/)
C/)
50 .............................................. ~
~
.............. ~~ ...................
C/)
z ---a
«
a: ab cd e
-·-·-b
---- c
I- I I I I I I I II -d
·······e
°2~0~0~--~3~0~0----~4~00~----5~0~0~--~6~00
WAVELENGTH (nm)
Figure 1. Transmission spectra measured before and after '"(-irradiation (lx10 5 rad) for five
samples, .pI" xl" long each, taken from a long crystal.
802
Table 1. Analytical results of BaF 2 crystal by AAS technique (ppm)
Impurity Sample 1 Sample 2 Sample 3

Fe 0.50 0.44 0.41
Co 0.50 0.85 0.65
Cr 1.5 2.0 1.8
Mn 0.10 0.20 0.20
Ni 2.2 3.2 3.8
Ca 8.5 12 14
Mg 2.1 6.5 19
Sr 900 980 1300
Sample 1: taken from the bottom of crystal

Sample 2: taken from the middle of crystal
Sample 3: taken from the top of crystal
results have been obtained in radiation tests for other large crystals. These experimental
results strongly support that the impurity is the main course of radiation damage. It could
be explained by the crystal growth method we used i.e. the Bridgman-Stockbarger method
which is based on a zone- refining principle. The raw material is placed in a graphite crucible
which is then placed inside a vacuum furnace. The furnace has a temperature gradient which
varies from above the melting point (1280°C) at the upper part to blow the melting point
at the bottom. The crystal growth proceeds from the bottom to the top as crucible descents
down. At the same time the impurities are forced to the top of ingot and concentrated over
there. When the crystal was irradiated, its top part will naturally be coloured in dark tint
due to the high impurity concentration thus resulted in large amount of radiation damage.
By using the AAS (Atomic Absorption Spectrometry) technique, we analysed cationic
impurities such as transition metals and alkaline earth metals in three crystal samples which
were taken at various sites from a 30 cm long ingot. The results listed in Table 1 show that
the contents of transition metals such as Fe, Co, Cr, Mn, and Ni are within same magnitude
of several tenth ppm to several ppm in whole ingot. It could be explained that while these
elements are in a tendency to migrate towards the top of ingot during crystal growth, they
~~30r-TT--------r-------'
Z'E
O~
i= r- 20
a... Z
a: W
OU
(/) u::: 10
roLL No.2
<t:W
o
U oL-~-===~~==~
200 250 300
WAVELENGTH (nm)
Figure 2. UV absorption of BaF 2 crystal with different oxygen content. No.1: low oxygen-
containing sample; No.2: high oxygen-containing sample.
803
easily escape from the top because of the low boilling points of their fluorides. It is rather
difficult at present to determine how much influence do these trace transition metals give on
radiation damage of BaF 2 crystal. The contents of alkaline earth metals such as Ca, Mg and
Sr in the crystals are relatively high and increase as crystal growth proceeds. It is, however,
understood that due to their similarity to barium, other alkaline earth metals even in a rather
higher contents may not have substantial influence on radiation damage of BaF 2 crystal.
For rare earth ions, we have tried to correlate the content of cerium in BaF 2 crystal
measured by using ICP (Inductively Coupled Plasma) technique with the absorption peak
at 290 nm because someone [4] has suggested that it is caused by cerium of several ppm or
less. However, our analytical results on BaF 2 crystals with such an absorption peak show that
no cerium has been found within the determining limit of 0.6 ppm. We have also not found
intrinsic absorption band in UV region for other rare earth ions.
Among the anion impurities, oxygen and hydroxyl ions which are the most harmful
impurities for radiation hardness of BaFa crystal could be easily introduced into the crystals.
Fig. 2 shows the absorption of two crystal samples with different oxygen contents. It is evident
that significant difference between these two samples is existed especially in the UV region.
For high oxygen- containing sample, it is almost cut off at 220 nm.
The influence of defects on radiation damage
The transmission spectra of a tapered BaF2 crystal in size of 3x3, 4x4, 25 cm3 after
"'(-ray irradiation were measured through polished side surfaces at three points a,b and c. (see
left part in Fig. 3) In general, curve b should locate in between curves a and c. However,
the transmission of curve b is actually lower than that of curve c. And strong coloration
could be found in the area around location b where high density of macro defects such as
a large number of scattering centers could be observed even by naked eye. (Fig. 4) It has
been identified by microscopy that these scattering centers are 'negative crystals' with definite
shapes and orientations. (Fig. 5) The perturbation of temperature in crystal growth furnace
or impurity effect is directly leading to the increase of density of defects.
Mechanism of radiation damage
From above discussion, it is clear that the presence of impurities and defects in crystal
100r_------~--------r_------~------_,
,.. .... """.-..... _._._._._._._._._._._._ ..!._ .........

z
a .I
en
en
~
en
z b
~
t- I, p .I
a b c
~oo 300 400 500 600
WAVELENGTH (nm)
Figure 3. Transmission spectra measured at three different locations of crystal (3x3, 4x4,
25cm3 ) after 10 5 rad "'(-ray irradiation.
804
Figure 4. BaF 2 sample with macrodefects.
are the major reasons for radiation damage. As impurity is concerned, we also found that
when the contents of cationic impurities decrease to a certain tolerated level, oxygen and
hydroxyl become most harmful.
We have carried out a diffraction study with an electron beam in a diameter of 1000
angstroms focused on a crystal. Fig. 6a shows a perfect BaF 2 diffraction pattern taken just
after starting a bombardment. This perfect pattern can be found nearly everywhere in the
crystal indicating a low density of dislocation. After a few minutes, however, diffraction rings
appear as shown in Fig. 6b. These rings have been identified as BaO. It is presumed that
the oxygen atoms or molecules previously dissolved in BaF 2 crystals being transformed into
oxygen ions during bombardment which then displace a fluorine atom at a lattice site and
combine with barium ion into BaO.
When BaF 2 crystal was irradiated with 60Co source, two broad absorption bands usually
appear in 190-250 nm and 500-600 nm regions respectively. It is likely that the oxygen ion
came from dissolved oxygen during irradiation is the cause of the 190-250 nm absorption
band. During irradiation, the fluorine vacancy in the crystal is easily to trap one or two
electrons forming a F color center. The formation of F color centers is probably the reason
for 500-600 nm absorption band.
t;I ~
•4' .
..
<."
e;
..~ II
.. . .. .,- ~~
~ •
..,
~ (" •
eJf ' (,
•
• •
.••
-
.
• f:
Figure 5. 'Negative crystals' in BaF 2 crystal (x 100).

lOOJLm
'------'
805
Figure 6, Electron diffraction patterns of BaF 2 crystal. (a) just start bombardment of electron
beam on crystals; (b) after certain time of bombardment of electron beam, pattern shows
appearance of BaO rings.
It is already known that the damaged crystals could be recovered with exposure to UV
light from a mercury lamp, sunlight or roomlight. It could be explained that the electrons
trapped by F color centers which formed during irradiation escape from the centers with the
energy of UV light thus destroy the F centers and the crystal is bleached. There is another
probability that during the exposure to UV light, the irradiation induced oxygen ions lose
their electrons thus return to original oxygen atoms or molecules.
Recent study [5] revealed the effect of OH- on radiation damage in BaF 2 crystal. A
piece of good sample from Shanghai Institute of Ceramics, Chinese Academy of Sciences was
heated in moisture at 700°C for 12 hours. The surface hydrolysis and subsequent migration
of OH- ion into crystal produced an OH- doped sample.
20 ~--~~--~----~-----r-----r----'
I
I - Before Irradiation
I
'7 16 I - - - After Irradla ion
Z E \
o.s. \
,......,
\
i= ~ 12 \
\ I I
~!!! \
\
I
I
\
\
~ 52 B
rot±:
\ I \
«w
\
-
/ \
\
o
()
4 \
"
~ ~20~--~15~O~--71B~O~~~2~'0~==~2~40~==~2~770==~300
WAVELENGTH (nm)
Figure 7. UV absorption of BaF 2 :OH- crystal.
806
20r--r--r----.-----r----,-----~--__,
z E
-
' 7 16 - Before Irradiation
--- After Irradiation
o .s.
- I- 12
I- Z
a:_
D..UJ
o (,)
cn- 8
celL.
«It
o(,) 4
~20 150 180 210 240 270 300

WAVELENGTH (nm)
Figure 8. UV absorption of undoped BaF 2 crystal.
The comparison of absorption measured before and after ,-ray irradiation between this
sample and two undoped samples showed that OR-doped sample exhibits obvious absorption
band. (Fig. 7 and Fig. 8) It was at 192 nrn before irradiation and shifted to 204 nm after. It
was suggested that OR- was decomposed through radiolysis and an U center was formed:
OH-->H?+O- or H;+O?
H? is interstitial hydrogen atom; H; is substitutional hydrogen ion; O? is interstitial
oxygen atom. The U center is the product of H- occupied in a fluorine vacancy. This was
further confirmed by an IR absorption at 793.75 cm- 1 originated from the U center.
CONCLUSION
From the information presented in this article, it is evident that the radiation damage
of BaF 2 crystal is impurity and defect related. The intrinsic radiation hardness of pure BaF 2
may be quite good. As impurities is concerned, the contents of cationic impurities such as
transition metals, rare earth and alkaline earth metals should be lowered to a certain amount
100n-------~------~----~------_,
z
o
en
(/)
~
(/)
z
«
a:
~
°2~O~O----~3~O~O----~4~OO~--~5~O~O----~600
WAVELENGTH (nm)
Figure 9. Transmission spectra for BaF 2 samples, 3x3, 4x4, 20cm 3 (dash) and 3x3, 4x4,
26cm3 (solid), before and after 1 Mrad ,-ray irradiation.
807
1o0n-------~----_,r_----_r------.
z
o
U)
(f) 50
~
(f)
Z
«
a::
I-
°2~0~0r---~3~00~----4~0~0----~50~0~--~6~00
WAVELENGTH (nm)
Figure 10. Transmission spectra for sample (2x2, 3x3, 25cm3 ) of previous batch, before and
after 1 Mrad I-ray irradiation.
for which there is no criterion yet at present. For anionic impurities, much attention should
be given to oxygen and hydroxyl, although the detection and elimination of trace of them
are rather difficult. On the other hand, measures should be taken to eliminate the defects as
possible as we can. There is no doubt that the less imperfection in crystal will make it more
harder to radiation. Recently thanks to the progress in our BaF 2 crystal growth experiment,
an encouraging result of radiation test on large crystals has been achieved. (see Fig. 9 and
Fig. 10) We will make further effort along this direction to improve the radiation hardness of
large BaF 2 crystals.
ACKNOWLEDGEMENT
The authors are indebted to the AAS Group and ICP Group of SIC for performing
chemical analysis measurements and to Dr. Song Xiangyun from TEM Lab. of SIC for the
help with diffraction study. Many useful discussions with Drs. R.Y. Zhu and Z.Y. Wei from
CALTECH are greatly appreciated.
References
[1] M. Laval et al., Nucl. Instr. and Meth. A206 (1983) 169-176.
[2] A.J. Caffrey et al., IEEE Trans. on Nucl. Sci. 33 (1986) 230-234.
[3] S. Majewski et al., Nucl. Instr. and Meth. A260 (1987) 373-376. P.K. Chakraborty and
K.V.Rao., J. of Mater. Sci. 22 (1987) 587-591. A. Murakami et al., Nucl. Instr. and
Meth. A301 (1991) 435-444. C.L. Woody et al., paper presented at the IEEE Nuclear
Science Symposium, Santa Fe, New Mexico, Nov. 5-8, 1991.
[4] H. Xiao et al., paper presented in GEM BaF2 Panel Meeting, Beijing, Dec. 12, 1991.
[5] L.Y. Chen et al., paper prepared in GEM BaF2 Panel Meeting, Shanghai, Dec. 14, 1991.
808
RADIATION DAMAGE OF RARE EARTH IONS DOPED
BARIUM FLUORIDE CRYSTALS
G.Chen, H.Xiao, S.Q.Man, J.Q.Zhang and S.X.Ren
Beijing Glass Research Institute

1 Dongdadi, Chongwenmenwei
Beijing 100062, China
ABSTRACT
The radiation resistance of rare earth doped barium fluoride (BaF 2 ) crystals was in vest i-
gated. Stable trivalence elements, such as Y, La and Lu, have no effect on radiation resistance.
While the elements which changes its valence from +3 to +4 under irradiation, such as Ce and
Pr, are harmful to the radiation resistance of BaF 2 , the elements which changes its valence
from +3 to +2 under irradiation, such as Sm, Eu, Dy and Yb, are useful in eliminating color
centers in visible range, but may introduce some additional color centers in UV range.
INTRODUCTION
Barium fluoride (BaF 2 ) crystal has been actively studied as a new kind of scintillator
material for electromagnetic calorimeter for several years[1,2). However, the radiation damage
of large BaF 2 crystal prevents its use in high radiation environment in future high energy
accelerators. Studies of large BGO crystal showed that several ppm doping of europium
can improve its radiation resistance [3). It was expected that the rare earth ions would also
influence the radiation resistance of BaF 2 crystals. The damping of rare earth elements in
BaF 2 crystals were systematically studied, which helps us in understanding the mechanism
of radiation damage in BaF 2 • The stable trivalence elements, such as Y, La and Lu, have no
effect on radiation resistance. While the elements which changes its valence from +3 to +4
under irradiation, such as Ce and Pr, are harmful to the radiation resistance of BaF 2 , the
elements which changes its valence from +3 to +2 under irradiation, such as Sm, Eu, Dy and
Yb, are useful in eliminating color centers in visible range, but may introduce some additional
color center in UV range. This report summarizes our results of this study.
EXPERIMENT
For doped crystal samples, the raw material was doped with various rare earth fluorides
with several ppm in weight, which were first obtained by transforming oxide reagents into

100~----~------------.------------,
°1~9~0----~3~00~--------~5~0=0-----------7~00
WAVELENGTH (nm)
Figure 1. Transmission of BaF2 crystals: 5 ppm La doped before irradiation (solid line) and
after irradiation (short dashed line); undoped before irradiation (long dashed line); and after
irradiation (dot dashed line).
fluoride compounds. As a reference, a undoped crystal was grew in the same furnace at the
same time. Therefore, except for the difference of the rare earth doping,. these two samples
were grown in the same conditions, and there should be no differences in metal impurities,
oxygen contents. The resulting crystals should also have similar density of defects in the lattice
structure. These ingots were then cut and polished to 10 mm thick samples. The transmission
of these samples were measured before and after the '"'{-ray irradiation of 1 MRads. For some
samples, the fluorescence and luminescence spectra were also measured.
RESULTS
HarIDless Elements
Our results show that some rare earth ions, such as y3+ , La3+ and Lu3+ , are not harmful
to the radiation resistance of BaF 2 crystals if the doping level is of several ppm in weight. An
example of the transmission spectra of LaH doping is shown in Fig. 1.
100n-----~------------._----------_,
300 500 700

WAVELENGTH (nm)
Figure 2. Transmission of BaF2 crystals: 1 ppm Ce doped before irradiation (solid line) and
after irradiation (short dashed line); undoped before irradiation (dot dashed line); and after
irradiation (long dashed line).
810
1oon--------------------------------,
°1~9~O----~3~O~O------------5~O~O----------~700
WAVELENGTH (nm)
Figure 3. Transmission of BaF 2 crystals: 5 ppm Pr doped before irradiation (solid line) and
after irradiation (short dashed line); undoped before irradiation (long dashed line); and after
irradiation (dot dashed line).
Harnlful Elements
Some rare earth elements were found to be harmful to the radiation resistance of BaF 2 .
A strong absorption band was found in 1 ppm Ce3+ doped BaF 2 crystals before irradiation
which is shown in Fig. 2. Its transmission spectrum after the irradiation is also shown in
the figure. We conclude that Ce 3 + doping is harmful to the radiation resistance. Pr3+ ion is
another example of harmful ions to BaF 2 radiation resistance. Fig.3 shows a clear evidence
of this observation.
Useful Elements
The crystals doped with Eu 3 +, Yb3+ and Dy3+ show complete different result. These
elements can eliminate color centers in visible range which are usually found in reference
samples after irradiation. Figures 4, 5 and 6 show colorless crystals after 1 MRad irradiation
with several ppm doping of Eu, Yb and Dy, respectively, while the undoped crystals were
brown in color with the same dosage.
However, these three elements create additional absorption bands in UV range after
irradiation, which were different from that of undoped samples. While the position of the
100n------,-------------.-------------,
- I
/ ;'
I," /
" \" //"
/.
v \/ lJ
°1'±9~O----~3~O-=O-------------,5~O~O--------------'700
WAVELENGTH (nm)
Figure 4. Transmission of BaF 2 crystals: 5 ppm Eu doped before irradiation (solid line) and
after irradiation (dot dashed line).
811
100~----,------------,----------~
.-
v /': /
\I
,.,-----
1/\j /\.'j
°1~9~O----~30~O~--------~5~O~O----------~7~OO
WAVELENGTH (nm)
Figure 5. Transmission of BaF 2 crystals: 5 ppm Yb doped before irradiation (solid line) and
absorption bands are related to the kind of dopants, the strength of these bands are related
to the level of the doping. Sm3+ is unique in this family in which it creates absorption bands
in both visible and UV range, as shown in Fig. 7.
DISCUSSION
In general, the radiation damage of the BaF 2 crystal is caused by formation of color
centers in the crystal after irradiation. These color centers absorb scintillation light thus
decrease the light yield of crystal. According to our studies in influence of alkali metal ions,
alkali earth metal ions, transition metal ions and oxygen [4), we refer the color centers in
BaF 2 crystals to "Complex Color Centers" (CCC), and assume they are produced based on
the following processes:
where Mn+ is metal ion, 0 is oxygen ion, Va is vacancy of fluorine ion and F is "F" type color
center which is unstable at room temperature. The underlined part in the formula are CCCs.
100n------r------------~----------_,
~-.---
~~90~--~300~----------~5~OO~--------~7~OO
WAVELENGTH (nm)
Figure 6. Transmission of BaF 2 crystals: 2 ppm Dy doped before irradiation (solid line) and
812
Table 1. Electron Configuration of some Rare Earth Ions
Element Electron Configuration Normal Valence

Y 3do4so +3
La 4fO +3
Lu 4(14 +3
Ce 4(1 +3 +4
Pr 4£2 +3 +4
Sm 4F +2 +3
Eu 4fi +2 +3
Dy 4f9 +2 +3
Yb 4f13 +2 +3
It should be emphasised that the M n + in the formula must be a changeable valent ion
which can offer one or more free electrons to form CCC during the irradiation. At the same
time, its valent orbits can affect energy levels of simple color center, such as F or F- centers,
and make them stable at the room temperature. Based on this analysis, we categorize the
rare earth elements to three families, as shown in Table 1.
All rare earth dopants in our experiment are trivalent. The valence of y 3+, La 3 + and
Lu 3+ ions are stable. Therefore, they have no effect on the formation of CCC which needs
a free electron through Mn+ -> M(n+l)+ during irradiation. This can explain why these
elements have no effect to radiation resistance of BaF 2 crystals.
When Ce3+ or Pr 3+ doped BaF 2 crystals were irradiated, a free electron might be pro-
duced through valence change from +3 to +4 of these ions, which forms a CCC. This indicates
that rare earth elements, which may change its valence from +3 to +4 under irradiation, is
harmful to the radiation resistance of BaF 2 crystals. On contrary, Sm3+, Eu3+, Yb 3 + and
Dy3+ might absorb a free electron under irradiation, since their valence might be changed
from +3 to +2. This process would eliminate some CCCs and their corresponding absorption
peaks, as shown in Figures 4, 5, 6 and 7.
The emission spectra of Sm3+ doped BaF 2 crystal are shown in Fig. 8. The intensity of
emission peaks observed in the figure is proportional to the the concentration of Sm 3+ ions.
After irradiation, the valence of Sm3+ changed to Sm H , and thus reduced the intensity of
these emission peaks.
The variablity of valence of Sm, Eu, Yb and Dy ions from trivalence to divalence is dif-
100~-----.-----------.----------~
300 500 700

WAVELENGTH (nm)
Figure 7. Transmission of BaF 2 crystals: 5 ppm Sm doped before irradiation (solid line) and
813
400 500 600 700
WAVELENGTH (nm)
Figure 8. Emission Spectra of BaF2 crystals: 5 ppm Sm doped before irradiation (solid line)
and after irradiation (dot dashed line).
ferent. Part of Eu 3 + can change into Eu2+ even before irradiation. Its transmission spectrum
is shown in Fig. 9. Note the difference between Fig. 4 and Fig. 9 can attributed to the level
of doping.
The result of Dy3+ doping in BaF 2 crystal must be further discussed. There are two
absorption bands in Dy3+ doped BaF2 crystal after irradiation (Fig. 6). These two bands, at
about 190 nm and 270 nm, are away from 220 nm where the fast component of BaF 2 crystal
is located. The luminescence spectra of Dy3+ doped samples are shown in Fig. 10. We did
not observe obvious difference between light outputs of fast component of Dy 3+ doped and
undoped samples. It seems Dy3+ doping may be one of several approaches for producing
radiation hard BaF 2 crystals.
At the end, we briefly discuss the combined effect of the oxygen and the rare earth
doping. It is considered that a lower vacuum during crystal growth will cause a higher oxygen
content in the crystal [4]. Figure 6 shows that 2 ppm of Dy3+ doping can eliminate eees in
visible range. But the crystal with 5 ppm Dy3+ doping, grown under the low vacuum 5 x 10- 2
Pa, showed strong absorption at visible range, in addition to the two UV bands, as shown
in Fig. 11. It is clear that 5 ppm Dy3+ doping can not eliminate all color centers caused by
high oxygen contamination. This indicates that the oxygen content are at several ppm level
in normal BaF 2 crystal. Similar results were also obtained from Yb3+ doped BaF2 crystals.
--
100~----~----------r----------'
/""---------
I
;
1'\ j
\ I', !
V'·\ .I
U
300 500 700
WAVELENGTH (nm)
Figure 9. Transmission of BaF 2 crystals: 100 ppm Eu doped before irradiation (solid line)
and after irradiation (dot dashed line).
814
o ppm: LT = 4690s, RT = 4700s, Is/It= 8:1
Peak Channal91 dao, Peak Counts 13874
I =
2 ppm: LT =8155s, RT =8170s. Isllf 8:1:1
Peak Channal 90 dao, Peak Counts 19364
Time (ns)
Figure 10. Luminescence spectra of BaF 2 crystals: 2 ppm Dy doped before irradiation (solid
line) and after irradiation (dot dashed line).
100n------r----------~----------__,
O~19~O~--~3±OO~--------~50~O~--------~700
WAVELENGTH (nm)
Figure 11. Transmission ofBaF 2 crystals: 5 ppm Dy doped before irradiation (solid line) and
ACKNOWLEDGEMENT
Many useful discussion with Professors Z. W. Yin, P.J.Li, S.K. Hua and Z. Z. Xu are
acknowledged. We also want to thank Prof. W.H. Liu, Dr. X.G. Li and Prof. G.J. Lin for
their help with measurements.
References
[1] P.Schotanus, C.W.E.van Eijk et al., IEEE-NS 34(1987)272.
[2] M. Murashita, H.Saitoh et al., Nucl. Instr. and Meth. A243 (1986) 67.
[3] Z.Y. Wei, PhD thesis, Shanghai Institute of Ceramics,1988, unpublished.
[4] G. Chen, S.X. Ren, S.Q.Man, H.Xiao and J.Q. Zhang, Radiation Damage of Barium
Fluoride Scintillator, paper presented at GEM BaF2 Panel meeting, Beijing, Dec. 1991.
815
IMPROVED PROCESS FOR MANUFACTURE OF RADIATION HARD
N-CHANNEL JFETs FOR DETECTOR ELECTRONICS
Larry A. Rehn and Daniel E. Roberts
InterFET Corporation
322 Gold Street
Garland, Texas 75042
INTRODUCTION
Input transistors for SSC cryogenic calorimer preamplifiers must have very good low
noise performance while remaining tolerant to the high radiation found inside the
calorimeter near the interaction region. N-channel Junction Field Effect Transistors (JFETs)
combine the best features of low noise operation and high tolerance to radiation to a greater
extent than bipolar or MOS transistors. New designs and manufacturing methods are
described which reduce the degradation of JFET noise performance due to neutrons. Two
basic premises governed the choice of design and manufacturing methods. First, the JFET
was designed with smaller features and spacings in order to minimize silicon in the active
volume of the transistor. This reduces the interaction cross section with neutron radiation.
Second, reduced feature size should increase the transconductance of the transistor, thereby
improving noise performance.
DESCRIPTION OF EXPERIMENT
The object of this study was to explore new designs and wafer fab processes to
minimize the transistor dimensions. The most important goals were to reduce channel
volume and increase transconductance of the JFET. The reduction of channel volume would
reduce the degradation of the signal-to-noise ratio of the system due to lattice damage from
neutrons. Increasing gm/C ratio by increasing the device gm is favored for the calorimeter
application since total input capacitance is largely governed by the detector. The process also
involved the use of thin oxides for passivation. It has been shown by Watanabe, et. al.,l & 2
that the use of very thin oxides and double dielectric layers can be used to reduce the effects
of ionizing radiation. Also, a suitable guard ring structure was included to prevent the
formation of parasitic inversion layers. Dopant concentration at the siliCOn/oxide interface at
the conclusion of processing is designed to exceed the critical amount of 1 x 10 18/cm- 3 to
prevent inversion from ionizing radiation (Sandia National Laboratory, Albuquerque 3 ).
To accomplish these goals, a test matrix of both JFET design parameters and silicon
epitaxy specifications was formed. The wafer fabrication process was developed to

minimize lateral and vertical device dimensions. Several custom wafer lots were run to
define the fabrication process. Mter this was successfully completed, a suitable wafer was
identified from the final lot and samples were fabricated in TO-18 metal case packages.
Their electrical and noise performance was measured and characterized as a function of the
design parameters. Finally, these results were compared to standard production JFETs.
DESIGN ISSUES
The design matrix consists of sixteen combinations of gate-width and SID pitch
dimensions. These range from about 50-100% of the typical production JFET design rules.
With the emphasis on those transistors with reduced dimensions, the test structurres were
loosely termed short-channel FETs, or SCFETs, to distinguish them from production type
JFETs. A standard interdigitated structure was chosen with multiple source, gate and drain
stripes repeated on a common centerline spacing, or pitch. This was implemented by
designing a photomask set with four cells in an array. The cells each have four JFETs with a
given gate-width and four different gate-lengths and SID pitch. The four gate-widths were
chosen to be similar to existing, production InterFET designs: NJ26, NJ72, NJ99, and NJ132
(See Table 1).
Table 1.
Gate-Width Comparable Standard
Array Device (microns) Inter FET Device
Cell "660" 660 NJ26 (2N4416)
Cell "1790" 1800 NJ72 (U31O)
Cell "2500" 2500 NJ99 (U310 Modified)
Cell "3350" 3350 NJ132 (2N6451 & IF1320)
Four JFETs are present in each cell sharing the common gate-width. These have a
selection of gate-length and SID pitch that is described in Table 2. The 8 and 10 micron SID
pitch designs are both considerably tighter than the normal InterFET production process. In
addition, the 1.5 micron line width used for the gate and SID stripes is smaller than the
normal negative photoresist process would allow. The object was both to bound the standard
production design dimensions, and to push the limits tighter than existing InterFET product
designs.
Table 2.
Design Gate-Length SID Pitch
5.0 microns 20 microns
Wafer Process
The wafer process flow was similar in the basic concept to the standard process used at
InterFET, but differed in specification requirements, control of critical dimensions, and
methods to introduce dopants to form the transistor structures. The process involved the use
of an n-type (phosphorus) epitaxial layer over a heavily doped p-type (boron) substrate. The
epitaxial layer forms the channel of the JFET, and the p-type substrate forms the backside
gate, or lower boundary defining the channel. This produces an n-channel JFET whose
principle structures are diffused into the epitaxial layer.
818
The object of the newer design is to end up with the proper selection for the electrical
parameters, but with much smaller design features. This has the greatest impact upon the
gate pinchoff voltage, Vpo, and breakdown voltage, BVgss. Very tight features and spacings
means that the effects of diffusion, both lateral and vertical, must be very carefully
controlled and minimized. Lateral diffusion affects all the silicon structures defined by
photomasks and vertical diffusion is important to control junction profiles from the surface,
and up from the highly doped substrate.
The wafer fabrication process was chosen to minimize junction diffusion. Ion
implantation rather than high-temperature gas deposition was used to dope the silicon.
Boron was implanted for both the guard ring and gate structures. Phosphorus was implanted
to form the SID overdope regions. The starting epitaxial silicon thickness was chosen to
allow a minimum amount of implant anneal and total junction diffusion during processing.
The initial thermal oxidation thickness was 2500 angstroms, about a third of the normal
production process. This reduces both the diffusion of the heavily doped substrate up into
the channel, and the sensitivity to ionizing radiation. The desired junction profiles of the key
transistor features are shown in Fig. 1.
1021
I I
1020 Gate Implant (Bl1 )
a' 5XlOE14@ 100 keY P+ Substrate
'"\
c 10 19
...=
.2
...c
~
/' /
i""""
V
10 18
~
<;J
c0 Epitaxial Channel
J
10 17
\ /
U
,
""
Y
10 16
10 15
1
V 2 3 4
microns
Figure 1. Profile of Gate Implant Into Epitaxial Silicon
An assumption was made that better noise performance would be achieved by the use
of an epitaxially formed channel for the transistors. This was based upon the fact that an
epitaxial layer has only a single species of dopant atoms, which can be controlled to obtain a
relatively uniform doping profile independent of the substrate silicon. Conversely, a channel
formed by diffusion must reverse the starting polarity type by adding a greater number of
dopant atoms (compensation) to the lattice. In addition, if an implant process is used to form
the channel, further lattice damage is likely.
It was necessary to specify a thinner layer of epitaxial silicon in order to minimize both
lateral and vertical junction diffusions. Five different channel resistivities were used to
determine possible effects on performance after radiation; 0.25, 0.5, 1.0, 1.6, and 5.0 ohm-
em target values. In each case two different thicknesses were used to increase the chance for
the proper match of final device characeristics for the least amount of total processing
diffusion. Thicknesses were targeted to achieve 2.0 and 2.5 microns of "flat zone" in the
doping profile prior to wafer fabrication.
819
RESULTS
Two-dimensional modeling was used to reduce process development time. Even so,
several wafer fabrication lots were necessary to achieve the proper final device electrical
parameters. The actual device doping profile turned out to be very close to that shown in
Figure 1. It is important to note that the combination of channel doping concentration and
distance between the top and back (lower) gate structures determines the transistor pinchoff
voltage (V po) - a fundamental control parameter in the process. It was found that the
transistor pmchoff voltage was very sensitive to variations in the epitaxial thickness. Only
one combination of epitaxy resistivity and thickness yielded the relatively low values of Vpo
that was desired among the tests run. Among two fab lots, four epitaxy groups resulted in
Vpo ranging from 1.4 to 12.0 V, indicating the sensitivity to starting epitaxy thickness.
Besides achieving electrical parametric results, the wafer fab process also needed to
provide suitable resistance to inversion of the guard ring under prolonged irradiation.
Spreading resistance profiles on some of the wafers indicated that the minimum required
concentration of> 1 x 10 18 /cm 2 was achieved.
Electrical & Device Characterization
The best match to desired electrical parameters was from the silicon epitaxy group
with resistivity equal to 1.1 Q-cm. This range of resistivity is typical for the manufacture of
a general purpose JFET. A channel resistivity of 0.25 to 0.50 Q-<:m would be a better choice
to counter neutron damage effects due to higher dopant concentration. In addition, standard
production experience would predict that 0.50 Q-<:m resistivity should increase gm around
25%. Somewhat surprisingly, even the devices made with the tightest spacings yielded good
functioning JFETs. As expected, both Idss and gain improve significantly as the gate-
length/pitch is reduced. Also, both Vpo and gm increase with decreasing gate-length/pitch.
The change is rather large between the two tightest spacings, indicating, perhaps, that we
were near the minimum workable limit for this process/material combination.
35
30 ~
----
25 ~ i-- 3350
Vgs=0V, Vds=5V
E 20
effi 15
2500
1790
-
10
5 660
08 20
10 12 14 16 18
SID Pitch, microns
Figure 2. Transconductance vs. Pitch. SCFET Families.
Fig. 2 shows gm as a function of SID pitch for each of the four gate-widths. As
expected, gm increases for greater gate-width and reduced pitch. Evidence of the minimum
spacing limit was also seen on the breakdown voltage. There is a bi-modal distribution of
BV gss as pitch is changed from 10 to 15 microns. The two wider spacings have
approximately 25 V BV gss, while the two tighter designs run around 12-15 V.
820
- L-----
7
---- -
6
120
~ 5 lIS
Q.4 18 & 110
cJ Std NJ26
3
2
SID Pitch in microns
1
Vds=5 V
o o
-6 -5 -4 -3 -2 -1
VgS'V
Figure 3. Input Capacitance vs. GateVoltage. SCFET 660 Family.
In Fig. 3 input capacitance is plotted against Vgs for the 660 micron gatewidth family
of JFETs. Notice that the characteristic curves are somewhat flatter than the production NJ26
JFET (same gate-width and SID pitch of 18 microns).
--------
1.4
I
~ r--
,.
~
Std NJ26 Device ""-
~
5 1.0
cJ
--= 0.6
0.Il
Vgs = 0 V, Vds = 5 V
I
10 12 14 15 18 20
SID Pitch, microns
Figure 4. gm/C Ratio. SCFET 660 Family.
High gm/C ratio is important for good noise performance and high switching speeds. In
Fig. 4 gm/C is shown for the 660 micron family as a function of SID pitch. As expected, the
highest gm/C occurs for the tighest spacings. The NJ26 production part, with similar channel
resistivity, falls somewhat above the curve of the SCFETs.
5
3350 Vgs=0V, Vds=5V
~
" ...
2500
~"
f-
___ 2500
1
660
10 12 14 16 18 20
SID Pitch, microns
Figure 5. gm/C Ratio. SCFET Families.
821
In Fig. 5 the gm/C vs. pitch curves are shown for all four gate-width cases. The gm/C
ratio improves with increasing gate width. This is probably due to the more efficient use of
active transistor area in the larger designs. There is a small, fixed separation (necessary for
adequate breakdown voltage) between the end of each SID finger and the adjacent isolation
guard ring. The larger gate width devices use longer gate stripes, which means that these end
effect losses are reduced.
lk /8
110 SID Pitch in microns=
N
/15 &/2Jl ~/IO
-z== 100
~
~ 10 NJ26 --.... f-- ~
~ ~
~
1
10 100 1k 10k lOOk
Frequency, Hz
Figure 6. Noise vs. Frequency. SCFET 660 Family.
A very interesting result is shown in Fig. 6, which plots room temperature equivalent
noise performance, eN, as a function of frequency and compares the test JFETs to the similar
NJ26 production device. At very low frequency (10 Hz) the 8 and 10 micron gate lengths are
considerably noisier than the 15 and 20 micron JFETs. However, for frequencies in the range
100 to 100 kHz there is not much difference. Also, the standard NJ26 JFET is as much as 30
times less noisy than the short-channel JFETs at 10Hz, but is only about twice as good at f =
100 kHz. This difference is apparendy due to the difference in the basic process, particularly
the use of ion implantation to form the transistor structures in the test JFETs.
"'" --- -
4.5 I I
Vgs=0V, Vds=5V
r-. 3.5
~
e 2.5
NJI32 Standard
~ 1.5 /
0.5
8 10 12 14 16
0'
18 20
SID Pitch, microns
Figure 7. gmlC Ratio. SCFET 3350 Family.
In Figs. 7 and 8 grrJC vs. pitch and noise vs. frequency are plotted for the larger 3350
micron gate width JFET. Fig. 7 also compares gm/C of the SCFETs to the standard,
production NJ132 device, which shares the same gate width. Notice that the standard part
falls slighdy below the curve of the test JFETs. Also, the 8-micron pitch has nearly a five-
fold gm/C advantage over the equivalent, standard JFET.
822
lk
:::::::::::::::
====
-
18 SID Pitch in microns
100
:---
N
110& 115 & 120
==
-Z ...................
;> f------
':10 F== NJ132 ~
~
1
- -- .....................
............:
0.1 1 1 1 1 1111 1 1 1 1 1111 1 1 1 I I III I I I I ""

10 100 lk 10k lOOk
Frequency, Hz
Figure 8. Noise vs. Frequency. SCFET 3350 Family.
In Fig. 8 the noise performance of the 3350-micron test devices are compared to the
standard production JFET. They are similar in behavior to those shown in Fig. 6 for the 660-
micron parts. For f < 10 kHz the implanted JFETs are at least 10 times noisier than for the
standard process. However, for 100 kHz, and presumably higher frequencies, the noise is
within a factor of 2 of the standard JFET. This difference is supported by cooled noise
measurements made by Madden at Lawrence Berkeley National Laboratory.4 Noise spectra
over the range -195 C to room temperature showed noise peaks characteristic of
generation-recombination sites .. This behavior is generally attributed to residual radiation
damage and/or problems in the conduction channel. Since the SSC applications anticipate
very short data collection times, there may not be much penalty in the use of the newer
process. The noise performance of the 3350 family JFETs departs somewhat from the data
shown for the 660-micron case in that there is a clear noise penalty at all frequencies for the
8-micron pitch case compared to the others. It is possible that this indicates the onset of hot
electron injection, as is sometimes seen in short-channel silicon MOSFETs and GaAs PETs.
---
1k
660/20 SID Pitch in microns
N
~100 ~ 1790/20
~
= r--- 3350/20
2500/20
- --
~--...........
~ 10
1 . .'
10 100 1k 10k lOOk

Frequency, Hz
Figure 9. SCFET Noise vs. Frequenccy. Best Case of Each Gate-Width.
The noise performance which compares the best case for each gate-width is shown in
Fig. 9. There is not much difference for the three largest gate-width cases. There is a fairly
uniform increase in noise of the 660-micron case by about 2X over the other three part
823
families. The reason for the difference is not clear, but may be due to the SID fmger end
effects that were mentioned previously.
SUMMARY
The results of the SCFET project have been very encouraging. A process which uses a
very thin, epitaxially formed channel and ion implantion to form the isolation, gate, and
source/drain regions has been demonstrated. The precision of the epitaxial thickness appears
to be an area which could greatly affect gate targeting and process yields of the production
process. Improved methods of manufacturing thin silicon epitaxy would be a great benefit.
The reduced device line widths will require the use of a tightly controlled, positive-
photoresist process, or more sophisticated direct write photolithography. The achievement of
better JFET performance and higher production yields will require production control
improvements in both thin-film epitaxy and fine-line photolithography,
Electrical results are also very positive. Electrical gain and gm/C performance can be
significantly improved by reducing gate length and SID pitch. The gm and gm/C ratio were
improved by up to 50% and 400%, respectively, for the 3350 micron gate-width over the
equivalent, standard production JFETs. The equivalent noise performance is improved as the
device gate width is increased. Noise performance is less sensitive to changes in pitch
dimension, but the tightest 8-micron spacing is somewhat noisier in all cases. This could
indicate the onset of hot electron injection. The 8 and 10 micron pitch cases also showed
reduced BV gss' to around 10-12 V, which must be considered in the circuit design.
Compared to the equivalent, production JFETs, the noise performance of the short-channel
JFETs is significantly worse for frequencies in the range 10Hz-10kHz. However, at f = 100 kHz
equivalent noise is within 50% of the production JFETs. Since SSC applications require very
short signal processing times, there may not be much noise penalty. At this time SCFET
performance with respect to radiation hardness is not available, but the results are expected
to be favorable. SCFET prototypes are currently being evaluated by Brookhaven National
Laboratory to determine whether the reduced channel volume will result in less noise
degradation due to neutron radiation compared to the currently available JFETs.
ACKNOWLEDGEMENTS
This work fulfills contract obligations under Brookhaven National Laboratory contract
number 484547. Thanks are due to Paul Brown and Joni Hunter of RF Monolithics for
processing help, and to Phillip McIntire at InterFET for the gathering and reduction of data.
REFERENCES
1. Watanabe, K., M. Kato, T. Okabe, and M. Nagata, "Radiation Hardened Silicon

Devices Using a Novel Thick Oxide," IEEE Trans. Nuc. Science, NS-32, 3971-3974, 1985.
2. Watanabe, K., M. Kato, T. Okabe, and M. Nagata, "Radiation Effects of Double

Layer Dielectric Films," IEEE Trans. Nuc. Science, NS-33, 1616-1222, 1986.
3. Dawes, W., Sandia National Laboratories, Albuquerque, private communication.
4. Madden, N., Lawrence Berkeley National Laboratory, private communication.
824
OVERVIEW OF SSC ACCELERATOR REQUIREMENTS
G. Dugan

Dallas, TX 75237
INTRODUCTION
This paper will present a general overview of the requirements of the Superconducting
Super Collider (SSC) accelerators. Each accelerator in the injector chain will be discussed
separately, followed by a discussion of the collider itself. In conclusion, the top level
requirements of the overall accelerator system will be presented.
For each accelerator, the primary operating parameters will be presented in tabular fonn.
A brief narrative discussion of the principal technical features of each machine will be given.
Finally, the principal technical design challenges for the machine will be noted, together with
the currently planned solution to these challenges.
LINAC
The parameters of the Linac are presented in Table 1.
Table 1. Linac Parameters.
Machine Parameter Requirement Units
Input Energy 35 keV

Input rms normalized transverse 0.2 nmm-mrad
emittance
Output Energy 600 MeV
Output rms normalized 0.9 X 10- 6 eV-sec
longitudinal emittance
Output rms normalized transverse 0.3 nmm-mrad
emittance
Cycle rate 10 Hz
RF frequency 428,1284(SCL) MHz
Peak current 25 rnA
Pulse length (max) 35. IJSCC
Length 146 m
Availability 98.4 %
No. DE-AC35-89ER40486.

Beam (in the form ofH- ions) will be delivered to the rrrst element of the Linac from a
35-keV magnetron ion source. This first Linac element is a 2.5-MeV, 428-MHz RFQ. Beam
is accelerated in the RFQ and injected into a 428-MHz drift-tube structure (DTL) which
accelerates the beam to 70 MeV. Both the RFQ and the DTL are powered by 4 MW klystrons.
The DTL contains 14 quadrupoles for transverse focusing.
Beam from the DTL is matched into a side-coupled linac (SCL) operating at 1284 MHz.
The side-coupled linac accelerates the beam to 600 MeV. It is powered by
10 20-MW klystrons, and it contains 61 quadrupoles for transverse focusing. Subsequent to
exit from the SCL, the beam is transported to and injected into the Low Energy Booster
(LEB).
The principal technical challenge for the Linac are the preservation of the tight transverse
emittance specification noted in Table 1. This requires care in the design of the matching
sections between the ion source and the RFQ, the RFQ and the DTL, and the DTL and the
SCL. The details of the SCL lattice design are also important.
LOW ENERGY BOOSTER
The parameters of the LEB are presented in Table 2.
Table 2. Low Energy Booster Parameters.
Input Energy 600 MeV

emittance
Output Energy 11.1 GeV
Output rms normalized 2n X 10-3 eV-sec
emittance
Cycle rate 10 Hz
RF frequency 47.5-59.8 MHz
Rotation frequency 417-524 KHz
Protons/bunch 10 10
Bunch spacing 5 m
Circumference 570 m
Betatron tunes(HIV) 11.65111.60
Transition gamma 22.14
Availability 99 %
H- ions from the Linac will be charge-exchange injected into the LEB at 600 MeV. The
LEB is a rapid-cycling synchrotron with superperiodicity 3. The lattice of the .machine is
configured to place the transition gamma above the operating range of the machine, thus
avoiding the need to cross transition. The rf system is required to tune over a substantial
frequency range (see Table 2); this is planned to be achieved through the use of rf cavities
826
tuned via perpendicularly biased ferrites. The synchrotron's main magnetic fields are provided
by conventional resistive magnets: 48 4-m dipoles and 90 0.6-0.7-m quadrupoles.
The principal technical design challenge in the LEB is the preservation of the beam
transverse emittance during the low energy end of the machine's accelerating cycle. This will
be difficult because of the substantial space-charge tune spread. If the baseline design fails to
meet the emittance growth budget, the principal option for dealing with this is the provision
for increasing the injection energy: the civil construction of the Linac tunnel will be performed
in such a way as to accommodate easily a future upgrade to a I-GeV Linac. Another option
under consideration is the use of a higher harmonic rf system to reduce the peak bunch
current.
Other technical challenges include preservation of the emittance during acceleration
(made easier by the lattice design, which avoids transition); compensation for eddy current
sextupole fields during ramping; and the control of transverse and longitudinal multibunch
instabilities.
MEDIUM ENERGY BOOSTER
The parameters of the Medium Energy Booster (MEB) are presented in Table 3.
Table 3. Medium Energy Booster Parameters.
Input Energy 11.1 GeV

emittance
Output Energy 199.1 GeV
Output rms normalized 2-4n X 10- 3 eV-SfX

emittance
Cycle period 8
Rotation frequency 75.7 kHz
RF frequency 60 MHz
Protonslbunch 10 10
Bunch spacing 5 m
Circumference 3960 m
Betatron tunes (H/V) 25.4/25.4
Transition gamma 23.27
Availability 93.5 %
The MEB is a large, slow-cycling synchrotron which accepts 11.I-Ge V beam from the
LEB and accelerates it to 200 GeV/c. Transition crossing in this machine occurs at an energy
near 25 GeV. The machine contains a conventional rf system providing a peak voltage of 2.1-
MY. The accelerator's main magnetic fields are provided by 340 6.5-m dipoles and 206 2.4-
m quadrupoles; all these magnets are resistive. Substantial power systems for the magnets are
required: the dipoles, for example, require 16 5200-A, 16-kV power supplies.
827
The MEB delivers fast-extracted 200 GeV/c-beam to the High Energy Booster (HEB).
In addition, it will deliver slow-extracted 200 GeV/c-beam to experimental areas for use in
producing test beams for SSC detector calibration.
As in the case of the LEB, one of the technical challenges in the MEB is transverse
emittance preservation during the lowest energy part of the operating cycle. However, the
issue in this case is not so much space charge as the quality of the machine's magnetic field.
This implies care in the design of the magnets to limit the influence of remnant fields at
minimum magnet excitation. An additional technical challenge involves transition crossing,
which must be performed with a minimum of transverse emittance growth.
HIGH ENERGY BOOSTER
The parameters of the HEB are presented in Table 4.
Table 4. High Energy Booster Parameters.
Input Energy 199.1 GeV

emittance
Output Energy 2000 GeV
Output rms normalized 0.004-0.035n eV-sec
emittance
Cycle period 515 sec
RF frequency 60 MHz
Protons/bunch 1010
Bunch spacing 5 m
Betatron tunes (HIV) 39.42/38.415
Transition Gamma 34.47
Availability 89.1 %
The HEB is the last accelerator in the injector chain. It receives 200-GeV/c beam from
the MEB and accelerates it to 2 TeV for injection into the Collider. Because of its high
operating energy, it is most economically realized as a superconducting accelerator.
This machine will be a slow-cycling synchrotron. In order to optimize the beam transfer
into the Collider, the machine will operate in a bipolar mode. It will have both clockwise and
counterclockwise injection, extraction and abort systems. The principal superconducting
magnet systems will include 512 13-m dipoles and 2318 1.6-m quadrupoles. Both dipoles
and quadrupoles will have 50-mm coil diameters. This is required both to provide sufficient
linear aperture at injection and to maintain the option (not implemented in the baseline design)
for slow-extracted 2-TeV beam to the test beam areas. The superconducting systems will be
828
cooled to cryogenic operating temperatures using 2 LHe plants, each providing about 6kW of
cooling at 4° and 7kW of cooling at 20·.
As in the other machines in the injector chain, one of the principal technical challenges in
the HEB is the preservation of the beam's transverse emittance during the operating cycle. As
noted above, this challenge is addressed by providing a sufficient coil diameter (50 mm) for
the superconducting magnets. In addition, the bipolar nature of the machine's operation cycle
is unprecedented for a machine of this type.
COLLIDER
The parameters of the Collider are presented in Table 5.

Table 5. ColJider Parameters.
Input Energy 2000 GeV

Input nns normalized transverse 0.8 nmm-mrad
emittance
Output Energy 20000 GeV

Output rms normalized 0.233n eV-sec
Output rms normalized transverse nmm-mrad
emittance
Cycle time 24 hr
RF frequency 360 MHz
Protons/bunch 0.75 X 1010
Bunch spacing 5 m
Betatron tunes (HJV) 123.4285/123.265
Transition Gamma 105
Availability 81.7 %
The Collider is a pair of superconducting synchrotrons which also function as storage

rings at their top energy. The two rings share a common circumference and are located one
above the other, with a vertical separation of 90 cm.
The basic geometry of the rings is that of two arcs joining two long, straight sections.
The long axis of the resulting figure points roughly north-south. Counterrotating beams of
protons in the two rings are brought into intersection at four locations around the
circumference. These points, called interaction points, are located in the long straight sections,
two on the east side and two on the west side. At these points, proton-proton collisions occur
with a center-of-mass energy of 40 TeV and are observed by experimental detectors.
The Collider contains a large number of superconducting magnets: there are 7704 15-m
dipoles, 136 13-m dipoles, and 1564 5.2-m quadrupoles in the arcs alone. The required
cryogens for the collider systems are provided by 10 LHe plants, each supplying about
6.5 kW of cooling at 4· and 14.4 kW of cooling at 20·. The associated power systems for the
829
magnets consist of both ramping and holding supplies capable of 7 kA. Extensive quench
protection and energy dump systems, as well as beam abort systems for each ring, are
required to protect the superconducting magnets. Finally, special magnetic optics is required
in the vicinity of each interaction point to reduce the beam size to a small value in order to
enhance the probability of high-energy collisions when the proton beams intersect.
Among the principal technical challenges of the Collider is the requirement of
preservation of the transverse emittance, both at injection energy and during the steps leading
to operation with intersecting beams. This requirement drives the aperture of the
superconducting dipoles in order to achieve the field quality needed for sufficient linear and
dynamic aperture for the beam. It also dictates the required correction systems needed for
control of persistent currents, and it plays a role in determining the alignment requirements for
the machine.
Additional technical challenges arise in connection with the intersection points. At these
points, the proton-proton collisions will generate substantial amounts of high-energy
radiation, some of which will be converted to heat in the adjacent quadrupole magnets.
Maintaining these magnets in the superconducting state in the presence of this heat load is a
difficult problem. The field quality requirements on these quadrupoles is also more stringent
than those in the remainder of the machine.
Control of beam losses is an important issue in any superconducting accelerator because
of the sensitivity of the magnets to beam-induced quenches. In a collider, the problem is
compounded by the sensitivity of th~ experimental detectors to small amounts of beam loss.
In addition, in this machine, because of its high energy and large size, the stored energy in the
beam is more than 100 times greater than in any previous machines of this type. The
consequences of a catastrophic local loss of the entire beam are dramatic and must be guarded
against with great care.
Finally, these proton synchrotrons will be the first for which the proton energy is high
enough that the phenomenon of synchrotron radiation becomes an important design issue.
Photodesorbtion of hydrogen from the cold bore tube by synchrotron radiation, and the
consequences for the bore tube vacuum, are important new considerations in this machine.
The proper treatment of these issues represents an additional significant technical challenge.
SUMMARY
The pam meters of the overall accelemtor system are presented in Table 6.
In meeting the luminosity goal given in Table 6, the principal technical challenge is that
of transverse emittance preservation, both in the injector chain and in the Collider.
For the injectors, this requirement determines the choice of injector chain energies; the
design of injector lattices and associated components to provide sufficient linear aperture; the
impedance budgeting for each machine, and the implementation of damping systems as
required; and the use of diagnostics and control in beam transfer to provide adequate
matching.
For the collider, transverse emittance preservation influences the design of collider
lattice, power systems, and correction magnet systems to provide sufficient linear aperture;
the design and fabrication of magnet systems to provide adequate field quality and reliable
performance; impedance budgeting and implementation of damping systems as required;
implementation of diagnostics and the control of orbit, tune, and chromatic features during
injection, acceleration, and squeeze; and control of the bore tube vacuum.
In addition to the difficult task of emittance preservation, perhaps an equally difficult
requirement is the availability specification. This challenge will be met by the design of
technical subsystems with redundancy and/or highly reliable components, the practice of
design risk reduction whenever possible, and the use of proven technologies.
830
Table 6. Overall Accelerator Parameters.
Parameter Design Goal Units
Initial luminosity .1033

Center-of-mass energy 40
Bunch spacing 5
Particles/bunch 0.75 X 10 10
Number of bunches 17424

Total number of particles 1.3 Xl0 14

Collision frequency 60 MHz
RMS normalized transverse emittance nmm-mrad
Beta at IP 0.5 m
Beam rms transverse size at IP 5 11m
Beam-beam tune shift (total) 0.007
Sync. Rad Pwr/ring 8.75 kW
ReliabilitylAvailability 80% Scheduled time
Stored Energy 400 Mloule
831
THE SSC LINAC: STATUS OF DESIGN
AND PROCUREMENT ACTIVITIES
L.W.Funk
Superconducting Super Collider Laboratory *

2550 Becldeymeade A venue
Dallas, TX 75237
ABSTRACT
The Linear Accelerator 1•2 (Linac) is the first element in the injector chain for the Superconducting
Super ColIider (SSC). It must be operational by October 1995 in order to commission the Low Energy Booster
(LEB). The physics design of the Linac is now essentially complete, detailed engineering is well advanced,
and manufacturing has begun. This paper will review the physics design with emphasis on recent significant
improvements, summarize mechanical design and construction status, and describe recent achievements in the
INTRODUCTION
The top-level technical requirements for the SSC Linac are given in Table 1. To assist with
understanding of the more detailed descriptions which follow, Figure 1 shows a block diagram of major Linac
subsystems and their interrelationships.
ION SOURCE AND LOW ENERGY BEAM TRANSPORT
The ion source must take hydrogen gas and, in a plasma, break up the molecules and attach a second
electron to some of the hydrogen atoms, forming H- ions. These ions are then electrostatically extracted from
the plasma and accelerated into the Low Energy Beam Transport (LEBT). The voltage chosen for the SSCL
ion source is 35 kV. To make up for unavoidable losses and emittance growth in the accelerator chain, the ion
source must produce 30-mA beam pulses with a normalized rms transverse emittance ~ 0.18x mmomrad. The
ion source pulse length will be 100 JlS, and provision will be made in the LEBT to switch the best part of this
pulse into the rest of the Linac for acceleration.
*Operated by the Universities Research Association, Inc., for the United States Department of Energy under
Supercollider 4. Edited by J. NonIe

Table 1. SSC Linac Main Technical Requirements.
Parameter Value
Particle H-
Output Energy 600 MeV*
Output Current 25mA
Pulse Length 2-35 jlS
Pulse Repetition Frequency 10Hz
Output Transverse Emittance (n, rms) ~ 0.3 1t mmomrad
Output Energy Spread ~ l00keV
Basic radio frequency 427.617 MHzt

Scheduled availability (Collider filling) ~98.8%
*Ability to upgrade to 1 GeV must be preserved.

t9th harmonic of LEB injection rf preserves bunch-to-bucket transfer option.
Historically, ion sources have proved to be one of the principal sources of unscheduled downtime, so a
long-term development program has been started to improve source reliability and maintainability. Two
technical options are being studied: the magnetron source,3 and the volume source. 4,5
The main characteristics of the magnetron source are 1) it utilizes a magnetic field, transverse to the arc
electric field (a "magnetron" configuration) to maximize the efficiency with which the plasma is generated; 2)
H- ions are created on the surfaces, defining the plasma volume, when plasma protons strike the surface and
pick up two electrons; and 3) a thin layer of cesium, with its low work function, is plated onto the ion-
producing surface to increase the probability of picking up two electrons. This technology is very attractive,
since it is in operation at both Fermi National Accelerator Laboratory (FNAL) and Brookhaven National
Laboratory (BNL), and there is considerable technical support available. It represents our main line of attack.
There are two main difficulties with the magnetron source: 1) over time, the cesium also plates out on other
parts of the system, reducing voltage holdoff capability and leading to sparking, and 2) the source requires
several days of conditioning to achieve satisfactory levels of stability and reliability.
A magnetron source, constructed with the assistance of the Texas Accelerator Center (TAC), has been
operational on a development test stand at SSCL since late 1990, and considerable progress has been made in
understanding and overcoming some of the problems associated with operation of a cesiated source. Its
performance has been quite encouraging. We have been able to reproduce the low-cesium, low-breakdown-
rate mode of operation utilized at FNAL and BNL. At the end of 1991, we completed a 15-day continuous run
at design current and voltage during which a transverse, normalized rms emittance of 0.12 1t mmomrad was
measured.
Magnetron 4-Vane RG+CG(PMQ) Side-Coupled

35 kV 1.8 EK 1.4 EK 1.0EK
HESQLEBT 0.6MW 4x3MW 9x16.3 MW
0.2m 2.2m 23m 112m
Ion
Source j
roo
RFQ
428
MHz I j
-
DTL
428
MHz
I-~
•
CCL
428
MHz j -
. ToLEB
E(n, long)
0.035 MeV 2.5 MeV 70 MeV 600 MeV n:mmemrad
30mA 27 rnA 27mA 27mA 10**-7 eVes
E(n, trans) 0.18 0.20 0.20 0.23
8.2 10.1 9.4
Figure 1. Block Diagram of Major Linac Subsystems.
The main characteristics of the volume source are 1) the plasma is generated by a high-power rf
discharge. A magnetic filter separates the plasma volume into a high-temperature region, where ionization
takes place in the rf fields, and a low-temperature region, where slower-moving electrons are re-attached to
protons and hydrogen atoms throughout the volume to create the ions, which are then electrostatically
834
extracted from the production zone; 2) it does not require cesium, so that commissioning times are shorter, and
operation is more stable; and 3) the low-temperature of the source region implies that lower emittances are
possible, and indeed, some have been measured below 0.1 2t mm·mrad. The two main difficulties of the
volume source are that it requires significantly higher gas pressures, posing problems for the vacuum system,
and that large quantities of electrons are also extracted, often more than 20 times the ion current, which is both
a power supply and a beam transport problem.
An rf volume source, built under contract by Lawrence Berkeley Laboratory (LBL), passed acceptance
testing at the end of 1991 and is now in the process of being instaIled in a development test stand at SSCL. An
emittance measuring system is being assembled for long-term testing of this source.
The characteristics of the ion source beam are not such as to allow immediate injection into the next
acceleration device, the Radio Frequency Quadrupole (RFQ). A short section of transport line, designed to
match the beam from the ion source into the RFQ, is required. This matching section, the LEBT, is a critical
element of the flJ'St sections of the Linac, and two distinct technical approaches are being studied.
The flJ'St of these is the classic Einzellens. 6 The advantage of this solution is its relative simplicity. It
can be segmented to provide a beam steering or switching capability. Its disadvantage is that it is weak
focusing, requiring voltages approaching the beam voltage in decel. mode, and up to twice the beam voltage in
accel. configuration. A decel. Einzel lens has been built with the assistance of T AC. In preliminary tests at
SSCL, the lens showed signs of considerable aberration, in general agreement with simulation. Further tests
are planned, and LBL has proposed a variation of an accel. configuration, the "ring" lens,7 which shows
promise of improved performance.
The second approach is a Helical ElectroStatic Quadrupole (HESQ),8 which offers the advantage of
stronger focusing (lower voltages) and intrinsic steering capability. Its disadvantage is that it is relatively more
difficult to align. A version has been built, once again with the assistance of T AC; preliminary tests have been
reasonably consistent with simulation, but once again with considerable emittance growth, thought in this case
to be due to serious misalignment Further tests are planned.
RFQ AND RFQIDTL MATCHING SECTION
Table 2 gives the main technical characteristics of the RFQ, which is being constructed by Los Alamos
National Laboratory (LANL) and is on schedule for delivery to the SSCL by August 1, 1992. The four-vane
structure is made up of two 1.1-m-long sections. The vanes for each section were separately machined and are
being electroformed together to form the rf and vacuum envelope. Detailed design of all remaining mechanical
components of the resonant assembly is well advanced. Special end walls, incorporating thin vacuum isolation
valves, tuning stubs, and provision for instrumentation are being designed at SSCL.
Table 2. RFQ Technical Characteristics.
Parameter Yalue
Frequency 427.617 MHz
Energy: Input 35 keY
Output 2.5MeY
Design Nominal Current: Input 30mA
Output 28mA
Design Peak Current: Input 70mA
Output SOmA
Transverse Emittance (n, rms): Input ~.20 2t mm'mrad
(at nominal current) Output ~.20 2t mm·mrad
Longitudinal Emittance (rms): Output 8 x 103 eY.s
Length 2.20m
Maximum Peak Surface Electric Field 36 MY1m (1.8 EK>
Total Peak rf Power 355kW
835
The rf power system, based on that used for the Ground Test Accelerator (GTA), is also being
constructed at LANL, largely by SSCL staff. Acceptance tests are scheduled to take place at LANL before the
middle of March. Low-level rf systems and supervisory controls are based on the GTA designs and have been
delivered to the SSCL. Design of remaining subsystems, such as support stands, vacuum pumps and gauges,
instrumentation, and the temperature control system, are in detailed design. All elements of the RFQ are
scheduled to arrive at SSCL this summer, with the intent of operating the unit in temporary quarters in the
Central Facility at Waxahachie. Preparation of SAR documents in support of an application for operating
approvals has begun.
The change in longitudinal and transverse focusing properties between the RFQ and the Drift Tube
Linac (DTL) necessitates an intervening matching section. The physics design has been completed for this
matching section, which consists of two buncher cavities (EoTL = 214 kV) and four movable (± 2 mm),
variable-gradient (50 TIm S; IGI S; 100 Tim, flGI.dl S; 4 T), permanent magnet quadrupoles, plus space for a
limited amount of instrumentation; and mechanical design has begun. The engineering approach chosen calls
for an outer shell to provide the vacuum envelope, as this appears to offer the prospect of making more space
available on the beamline for essential components and instrumentation. Specifications for the quadrupoles are
in preparation and have been completed for the rf amplifier systems for the bunchers. We are attempting to
achieve delivery of the completed matching section during the summer of 1993.
DRIFT TUBE LINAC AND DTL/CCL MATCHING SECTION

The main technical specifications of the DTL9 are given in Table 3. The contract for fabrication of the
DTL tanks was awarded to AccSys Technology, Inc., in October 1991, for delivery starting in late 1993. The
order for the 4-MW klystrons for the DTL was also placed in late 1991 with the Thompson Tubes
Electronique division of Thompson CSF, with delivery of the first klystron scheduled for November 1992. The
specification of the line-type modulators for the klystrons is complete, and the bid package has been issued.
Final date for bid submission is set for early April, and it is hoped that the award can be placed shortly
thereafter. Preparation of specification and requirements documents for the balance of the DTL subsystem has
begun. Our schedule calls for start of DTL commissioning in February 1994.
Table 3. Drift Tube Linac Technical Specifications.
Parameter Value
Number of tanks 4
Input Energy 2.5 MeV
Design Nominal Current 25mA
Design Peak Current 50mA
Transverse Emittance (n, rms) Input ~.20 It mm·mrad
(at IlOminai current) Output ~.20 It mm·mrad
Drift Tube Diameter 80mm
Bore Diameter 16mm
I/G·OII 4.64T
rf Pulse Length/Pulse Repetition Frequency loo~lOHz
Tank # Length # Drift # Post Gradient Bout Power

(m) Tubes Stabilizers (MV/m) (MeV) (MW)
4.515 55+2/2 27 2.423-4.600 13.364 1.00

2 5.951 39+2/2 39 4.600 32.766 1.75
3 6.062 29+2/2 29 4.600 51.491 2.00
4 6.230 25+2/2 25 4.600 70.149 2.00
836
Once again, there are major changes in the longitudinal and ttansverse fields to which the beam is
exposed as it transits between the DTL and the Coupled-Cavity Linac (CCL), and a matching section is
required to ensure smooth handling of the beam and minimization of emittance growth. The D1L/CCL
matching section design is not yet complete, but its major features have become clear. It will consist of six
electromagnetic quadrupoles (101 :S 40 Tim) and two buncher resonators. The bunchers will operate at
1282.851 MHz, the same frequency as the CCL. The tota1 voltage (EoTL) required from the second buncher is
about 40% larger than that required from the frrst, but by making tank length proportional to the voltage it is
required to develop, we can make both buncher cavities have the same response to transient beam loading,
thus allowing the two bunchers to be driven from a single rf source.
The schedule for delivery of the components of the DTl./CCL matching section is spring of 1994.
COUPLED-CAVITY LINAC
The most fundamental design changes that have taken place in recent months have been those that have
affected the CCL. (See Table 4 for the main technical specifications.) The original designlO aimed at low cost
and high reliability through proven technology, a low DTL-CCL transition energy, and conservative design.
The event that led to the changes was the discovery of unacceptable levels of higher multipoles in the field of
the RFQ. It was possible to modify the RFQ design to correct the problem, but the solution was constrained by
the manufacturing approach and led to an unavoidable increase in longitudinal emittance. Since longitudinal
emittance is blown up by several orders of magnitude during adiabatic capture in the LEB, this was not
considered a serious defect, and the design was accepted. It was later discovered, however, that the
combination of an increased longitudinal emittance and the design of the CCL conspired to greatly increase
the transverse emittance growth in the CCL, to the extent that we would not have been able to achieve the low
growth factors implied by the values in Figure 1. While reviewing the design to reduce emittance growth, we
attempted to find ways to further reduce the cost.
A number of changes resulted from this effort:
1. All tanks were redesigned to have the same number of cells. This eliminated a biperiodicity in
longitudinal focusing.
2. The number of cells per tank was reduced. Since all cells in a given tank are designed for the same
value of B = vic (for ease of manufacture), there is an unavoidable phase slip through each tank.
The shorter tanks have less slip and therefore reduce the exposure of the beam to non-linear rf
defocusing forces that lead to emittance growth.
3. Division of each module into 8 tanks, rather than 6, provided more opportunities for insta1lation of
focusing magnets, which reduced the average beam size and further reduced the non-linear rf
defocusing forces. This allowed a reduction in the diameter of the beam bore hole and an increase
in the cavity shunt impedance.
4. The ramped gradient in the first two tanks of module 1 was not found to be essential for low
emittance growth and was eliminated, which also slightly increased the acceleration efficiency of
theCCL.
5. Intermodule spaces were made identical with local intertank spaces, which further smoothed both
longitudinal and transverse lattices. The loss of diagnostic space implied by this was compensated
for by the addition of small diagnostic spaces between tanks.
6. Details of the cavity geometry were modified to improve the shunt impedance and ratio of peak-to-
average electric field for the high-B cavities.
7. Side coupler dimensions were modified to minimize variations in the air-vacuum frequency shift
and to reduce peak surface fields.
8. The cell-to-cell coupling constant was increased from 5% to 6.8% to reduce problems anticipated in
driving long coupled structures. This reduced the anticipated shunt impedance.
9. The above reconfigurations and efficiency improvements/reductions, combined with a modest
reduction in the rf power safety factor, were sufficient to enable us to eliminate one klystron stand,
for a considerable cost saving.
837
Table 4. CCL Technical Specifications.
Paramenter Value

Number of modules x Number of tanks/module 9x8
Design Nominal Current 25mA
Design Peak Current SOmA
Transverse Emittance (n, rms) Input g).20 1t mm·mrad
(at nominal current) Output g).20 1t mm·mrad
Beam Bore Diameter 20mm
Peak Power Required!Available 16.3/l0MW
Peak Surface Electric Field 32 MV/m (1.0 EJ()
rf Pulse Length/Pulse Repetition Frequency 70~lOHz
Final cavity geometries have been determined for module 1, and detailed engineering design is
underway. Our plan is that all the CCL tanks, bridge couplers, waveguide adapters. windows, and mechanical
support stands will be provided by the Institute of High Energy Physics (IHEP). Beijing. under the terms of an
international agreement. Several members of the IHEP staff have worked at the SSCL on the design and
engineering of the CCL. Manufacture of the first four tanks of module 1 will serve to qualify their
manufacturing capability. We will have the remaining tanks of module 1 manufactured locally, to provide a
backup position. All of module 1 is to be complete by the end of 1992, and the remaining modules are
scheduled for delivery by IHEP at the rate of one module/month. starting in early 1993.
The quadrupole magnets, about 100 of which are required, are part of a package of contributions being
planned by the Centre for Advanced Technology, Indore, MP, India. A mechanical engineer from the Indian
institute has been invited to SSCL to assist with final design.
TRANSPORTrrRANSFER LINEll
The Linac facility has been designed so that space is reserved for a future upgrade of the Linac energy
to 1000 MeV. should this be the most cost-effective way of improving the Collider luminosity. The line
required to conduct the beam from the 6OO-MeV point to the nominallOoo-MeV point is called the Transport
Line. This line gradually transforms the transverse optics to those required for the Transfer Line that follows,
and it provides a location for the Energy Compression (EC) cavity. The 6OO-MeV beam emerging from the
CCL has an energy spread of about 0.5 MeV. Injection into the LEB requires an rms energy spread of only
about 0.1 MeV, so after sufficient drift to translate energy differences into phase differences, a buncher cavity
is installed that reduces the energy spread. The design of the EC cavity will follow closely that of the final
tank in the CCL. The quadrupoles will be provided by India as an international contribution, and design is well
advanced.
The Transfer Line is being designed to conduct the beam from the Linac to the LEB, to place it on the
closed orbit and strip it. or to transfer it to the various beam dumps, as necessary. Provision has been made in
the transfer line for instrumentation to measure both longitudinal and transverse characteristics of the beam.
Power supplies. dipole and quadrupole magnets, vacuum systems, and support stands are being
specified and designed.
CONTROLS AND BEAM INSTRUMENTATION
The master schedule calls for commissioning and operation of the Linac before the Global Accelerator
Control Systems will be available to support it. In addition, installation and commissioning of Linac
subsystems is naturally handled in a modular fashion. Finally, the supervisory controls for the rf systems will
handle more than half the total control and monitor points in the associated Linac module. As a result, a
controls architecture has been developed that provides local supervisory intelligence for each module. based
on rf supervisory controls.l 2 · 15 Lab-wide communication facilities and protocols are then used to tie these
modules into a functional Linac structure, and ultimately to tie the Linac into the rest of the Injector Complex.
The first test of this approach will come with the commissioning of the RFQ during the latter part of 1992.
Emittance preservation puts extremely taxing demands on low-level rf controls. system alignment, and
beam diagnostics. The first low-level rf control system that will be used to attempt to achieve the tight phase
838
and amplitude tolerances f<x' cavity fields (±OS, ±O.5%) is the ~led "I and Q" system,16.11 which derives
in-phase and quadrature error signals for application to the input rf signal. We are presently developing
detailed plans for commissioning that will use specialized reloca1able diagnostics elements for accurate initial
set-up, and for calibration of a more restricted set of diagnostics that will be available during routine operation.
The designs of all systems are being constrained to provide as much access to the beam as possible. Actual
population of these access ports with instruments will be determined by available funds and operational needs.
BUILDINGS AND FACILITIES

Title II design of the conventional facilities is complete, and the design has been sent out for bid. We
anticipate award of the construction contract in late May, and beneficial occupancy by the end of April 1993.
We are including features that will permit the servicing of a proton radiotherapy facility by 70-250 MeV
beams from the Linac during periods when we are not fIlling the LEB.
REFERENCES
1. J. Watson, "The SSC Linac," Proceedings of the 1990 Linear Accelerator Conference, September 10-14,
1990, Albuquerque, NM, LA-12004-C, pp. 31-35.
2. J. Watson et al., "Design of the SSC Linac," Proceedings of the 1991 Particle Accelerator Conference,
May 6-9,1991, San Francisco, CA, 88-647453, 91CH3038-7, pp. 572-74.
3. C.W. Schmidt and C.D. Curtis, "A 50 rnA Negative Hydrogen Ion Source," IEEE Trans. Nucl. Sci.,
NS-26 (3),4120 (1979).
4. K.N. Leung et a1., "Optimization of H- Production from a Small Multicusp Ion Source," Rev. Sci. Instrum.
60(4),531 (1989).
5. K.N. Leung et al., ''Optimization of the RF Driven H- Ion Source,", Proceedings of the 1991 Particle
Accelerator Conference, May 6-9, 1991, San Francisco, CA, 88-647453, 91CH3038-7, pp. 1919-21.
6. C.R. Chang, "Design Studies of SSC Low Energy Beam Transport System Using Einzel Lenses,"
Proceedings o/the 1990 Unac Conference, September 10-14,1990, Albuquerque, NM, LA-I2004-C,
pp. 399-400.
7. O.A. Anderson et al., "Low-Energy Injector Design for SSC," Proceedings of the 4th International
Conference on Ion Sources, Sept. 30-Oct.4, 1991, Bensheim, Germany, in press.
8. D. Raparia, "HESQ A Low Energy Beam Transport for the SSC Linac," Proceedings of the 1990 Linac
Conference, September 10-14, 1990, Albuquerque, NM, LA-I2004-C, pp. 405-7.
9. G.H. Neuschaefer et al., "SSC Drift Tube Linac Physics Design," Proceedings of the 1990 Linac
Conference, September 10-14, 1990, Albuquerque, NM, LA-I2004-C, pp. 351-4.
10. C.R. Chang et al., "Design Studies of SSC Coupled Cavity Linac," Proceedings 0/ the 1991 Particle
Accelerator Conference, May 6-9,1991, San Francisco, CA, 88-647453, 91CH3038-7, pp. 2993-5.
11. R.K. Bhandari, E. Seppi, and S. Penner, "Design Characteristics of the Linac-LEB Transfer Line for the
SSC," Proceedings of the 1991 Particle Accelerator Conference. May 6-9, 1991, San Francisco, CA,
88-647453, 91CH3038-7,pp. 351-3.
12. G. Martinsen et al., "Initial Control of the H- Ion Source at the Superconducting Super Collider
Laboratory," Proceedings of the 1991 Particle Accelerator Conference, May 6-9, 1991, San Francisco,
CA, 88-647453, 91CH3038-7, pp. 1365-7.
13. Kewisch et al., "Accelerator Simulation and Operation via Identical Operational Interfaces," Proceedings
of the 1991 Particle Accelerator Conference, May 6-9,1991, San Francisco, CA,
88-647453, 91CH3038-7,pp. 1443-5.
14. AJ. Kozubal, D.M. Kerstiens, J.O. Hill, and L.R. Dalesio, "Run-Time Environment and Applications
Tools for the Ground Test Accelerator Control System," Proceedings of the International Conference on
Accelerator and Large Experimental Physics Control Systems, Vancouver, BC, Canada, OcL 3O-Nov. 3,
1989.
15. W.H. Atkins, G.D. Vaughn, and C. Bridgman, "An Automated Vacuum System," Proceedings of the
1991 Particle Accelerator Conference, May 6-9,1991, San Francisco, CA, 88-647453, 91CH3038-7,
pp.1522-4.
16. S.P. Jachim, "Some New Methods of RF Control," Proceedings of the 1990 Linac Conference, September
10-14,1990, Albuquerque, NM, LA-12004-C, pp. 573-7.
17. C.D. Ziomek, S.P. Jachim, and E.F. Natter, "Design of a Multivariable RF Control System Using Gain-
Shaping in the Frequency Domain," Proceedings of the 1991 Particle Accelerator Conference, May 6-9,
1991, San Francisco, CA, 88-647453, 91CH3038-7, pp. 1329-31.
839
AN UPDATED OVERVIEW OF THE LEB RF SYSTEM
Jimmy D. Rogers, Jim H. Ferrell, and Jim E. Curbow

Dallas, TX 75237
Carl Friedrichs

ABSTRACT
Each of the Low Energy Booster (LEB) rf systems consists of the following major
subsystems: a vacuum tube final rf amplifier driven by a solid state rf amplifier, a ferrite-tuned
rf cavity used to bunch and accelerate the beam, a low-level rf system including rf feedback
systems, a computer-based supervisory control system, and associated power supplies. The
LEB rf system is broadband with the exception of the rf cavity, which is electronically tuned
from approximately 47.5 MHz to 59.7 MHz in 50 ms.
The design and development status of the LEB rf system is presented, with particular
emphasis on the cavity and tuner, and the tuner bias power supply.
REQUIREMENTS
The LEB rf system is required to raise the beam momentum from 1.22 GeV/c at
injection to 12 GeV/c at extraction. The beta change experienced by the particles in the LEB
from injection to extraction is 0.8 to 0.97. The frequency shift must be proportional to the
beta change. The bunch spacing, which is set at 5 m in conjunction with the beta change,
determines the frequency change required in the LEB cavity. The most challenging design
requirement of the LEB system is the design of the cavity and tuner, which must track the
excitation frequency with close precision from 47.5 to 59.8 MHz over a 50-ms accelerating
cycle.
The maximum voltage of the LEB during this acceleration cycle is 765 kV per tum.
Assuming that n cavities are operating in the LEB, each cavity must operate at a maximum gap
No. DE-AC35-89ER40486.
Supercollider 4, Edited by J. Nonie

voltage of (765/n) kV. In addition to the frequency and voltage requirements, higher-order
mode suppression, tuner frequency response, and cavity length are important design
considerations. Space for 16 cavities, each 1.5 m long, has been allocated in the LEB lattice.
Reliability is of the utmost importance. All LEB rf systems are required to have a combined
availability greater than 99%.
SYSTEM DESCRIPTION
Each rf system for the LEB consists of the major subsystems depicted in Figure 1. The
power amplifier consists of a high power tetrode and associated input and output matching
sections. The 5 KW driver amplifier is a solid state unit with less than 50 ns group delay.
The low-level rf subsystem includes feedback loops to maintain amplitude and phase stability
of the accelerating voltage. The rf system is broadband over the 47.5 MHz to 59.8 MHz
frequency range, except for the accelerating cavity, which is ferrite-tuned over the frequency
range in 50 ms.
Figure 1. Rf System block diagram.
The LEB Cavity is a quarter-wave resonant structure with a ferrite-filled tuner

comprising the high-current end. The rf amplifier, which provides the excitation energy to the
cavity, uses a 4CWI50,OOOE tetrode. The amplifier circuit employs tight capacitive coupling
to the cavity such that the cavity becomes the tuned output circuit for the tetrode amplifier. The
resonant frequency of the cavity is varied over the required frequency range by controlling the
magnetic field bias applied to the ferrite. This bias determines the permeability of the ferrite
and therefore controls the resonant frequency of the cavity. The tuner bias supply is controlled
by a "feed-forward" signal derived from previous pulses to within 10% of the desired
frequency. A feedback signal derived from the cavity is then used to closely track the desired
resonance frequency.
842
CA VITYITUNER
Traditionally, synchrotron cavities have been operated with the ferrites in the parallel
bias mode. The LEB cavity will employ "perpendicular bias." The distinction of this mode of
operation is in the perpendicular direction of the applied control bias magnetic field with
reference to the rf magnetic field in the cavity.
While the bias requirement per unit volume of ferrite is substantially greater with
perpendicular bias than with parallel bias, the high rf efficiency and the resulting volume
reduction make perpendicular bias an inviting choice.
One ferrite being considered for use in the LEB tuner is an aluminum-doped yttrium-
iron-garnet (YIG) material with a saturation magnetization of 0.08 T. The material will have
an electric Q in excess of 10,000. The magnetic Q, which is a function of the permeability,
will have an average over the acceleration cycle substantially higher than 10,000. This sort of
material will require a bias change from 0.1 to 0.25 T to change the cavity frequency from
47.5 to 59.8 MHz. The relative permeability of the ferrite will be changed from 3.5 to 1.4 as a
result of the bias change. The relative dielectric constant of the ferrite is approximately 14.5.
Three methods of tuner cooling are being considered: liquid cooling, air cooling, and
solid conduction cooling. The maximum gap voltage that can be applied to the LEB cavity will
depend on the type of cooling selected. The gap itself is not voltage-limited, but the electric
fields in the tuner are proportional to the gap voltage for any given configuration. The
breakdown voltage of the dielectric seems to be the limiting factor for the air-cooled design,
and heat dissipation becomes a consideration in the solid conduction-cooled design. Trade-off
studies of the tuner cooling are presently being performed.
A typical LEB cavity design is shown in Figure 2.
Coupling
CapacItor
Accelerati
Gap
Figure 2. A Typical LEB Cavity Design.
The design shown in Figure 2 uses a liquid-cooled tuner. The tuner contains four ferrite
disks, each of which is 30 cm outside radius by 16 cm inside radius by 2.5 cm thick. The
coupling capacitor is approximately 13 pF, and the voltage step-up ratio from the tube anode
to the accelerating gap is 8.5. The length of this cavity without a higher-order mode damper is
100 cm. The circuit Q is about 5000, and the shunt impedance is about 160 kQ.
843
Figure 3 is a superfish computer model of an LEB cavity with a solid conduction-cooled
tuner and a higher-order mode (HOM) damper. The cavity of Figure 2 has been modified to
increase the gap capacitance and to facilitate the possible addition of other HOM dampers or
tuners at a later time. The HOM damper shown in Figure 3 has recently been proposed by
SSC engineer Linda Walling as an improvement of the Smythe damper.l The input section of
the HOM damper is filled with carbon-loaded aluminum nitride.
~ ~
- f)MJ
I j
Figure 3. Geometric Model for Superfish Analysis of LEB Cavity.
LEB TUNER BIAS POWER SUPPLY AND CURRENT REGULATOR
In the recent past, bias regulator designs employed inefficient linear transistors or low
signal bandwidth switchers operating with low-frequency pulses. The preferred state-of-the-
art designs presently available from industry employ Pulse Width Modulated (PWM)
regulators operating at frequencies in the range of 80-120 kHz. Such designs routinely
provide wide signal bandwidth on the order of 5-10 kHz. In a complementary fashion, wide
signal bandwidth enables achievement of regulator tracking accuracies to within 200 ppm of
the input reference waveform. The PWM regulator scheme has been chosen for the LEB tuner
bias supply.
Figure 4 is a block diagram of the bias current regulator.
The power supply connected to the input of the bias regulator provides an unregulated voltage
source. The DC output of the power supply is tap-selectable to yield voltages of 150 VDC,
200 VDC, 250 VDC, or 300 VDC. The power supply is capable of providing 1500 A of
output current. The electrical characteristics of the power supply are listed below.
Primary. 480 VAC, ±3%; 3 phase; 57-63 Hz, wye-connected with taps to provide the
necessary DC output voltages.
Secondary. The output of the power supply is 150 VDC, 200 VDC, 250 VDC, or
3()() VDC selected by tap connections on the primary. The secondary is provided with one
delta and one wye winding which feed two 3-phase rectifier bridges. The output ftlter is a
shunt capacitor yielding a ripple of no greater than 0.1 %.
Impedance. 10% nominal at 60 Hz.
Construction and Materials. The framed housing conforms to ANSIINEMA 250-

1985, Type 2 Enclosure for indoor use. The windings and buswork in the supply are copper.
The cooling pipework may be either copper or 304 stainless steel.
Temperature Rise. The transformer assembly must not exhibit an average

temperature rise greater than 50·C when operating in an ambient of 45·C.
844
480VAC,3
Phase 60Hz
Unregulated
~_...L._--,
DC Input Bus
Unregulated Regulated
DC Power Output
Supply Current
•
•
100 KHz Synch, •
Reference Fanout
~.......~ Regulatocl--r.f-4-t_ _ _ _ _ _•
Rack#S
System
Control Thner
Inductor
Reference
Feedforward
~ Phase Error Feedback Signal
-----. Control Signal

. . . Power Connection
Figure 4. LEB Bias Current Regulator Block Diagram.
845
Each regulator rack consists of six power field-effect transistors connected in parallel,
configured for series current regulation. Each transistor is rated to provide SO A of
continuous drain current with no derating at 100kHz, and is capable of withstanding drain-to-
source voltages of 400 VDC. Five regulator racks are parallel-connected to yield a total output
current of lS00 A. Current feedback around each regulator rack ensures equal load current
sharing among the four amplifiers.
The regulator is globally controlled from the system control block. System control
provides synchronization pulses to each regulator rack oscillator and input signal correction.
The error signal for input correction is derived from a highly stable zero flux DC current
transformer, which will be replaced at a later date with an rf cavity phase detector signal.
PWM regulation of the output current is accomplished using fixed-frequency operation,
with variable duty-cycle pulses. Fixed-frequency operation simplifies output filtering and EMI
filtering. Careful attention has been directed at minimizing propagation delays through the
signal chain to ensure rapid response. The PWM control circuit, for example, is capable of
operation at 1 MHz with only SO ns of signal propagation delay.
LEB DC POWER SUPPLY REQUIREMENTS

LEB Power Amplifier Anode Power Supply
The anode power supply is an oil-filled, transformer/rectifier rated at 4S0KV A with the
following characteristics:
Primary. 480 VAC, ±3%; 3 phase; 57-63 Hz. The primary windings may be
connected in either delta or wye.
Secondary. Dual secondary with delta-wye windings, provided with a no-load tap-
changing switch that allows selection of 16 KVDC, 18 KVDC, 20 KVDC, 22 KVDC, or
24 KVDC output voltages with the primary connected in delta. This configuration is
connected to cascaded three-phase, full-wave bridge rectifiers providing an output voltage
with 720 Hz ripple.
Impedance. 10% nominal at 60 Hz.
Bracing. The core, coils, tap switches, and all parts are capable of withstanding
40 times normal current for 50 ms since the power supply must withstand an estimated
1000 crowbars during testing and an average of 20 per month for the life of the unit.
Materials. The windings and buswork are all copper.
Temperature Rise. The total assembly must not exhibit a temperature rise greater
than SO°C when operating in an ambient temperature of 4S°C. The operating temperature
range is 30°-45°C.
Controller. The controller consists of all necessary circuitry and controls for safety
and operational interlocks. Included in the controller is a vacuum contactor and an SCR
controller/switch. The SCR controller/switch provides a soft start and ±l % regulation around
a set point. The set point is determined by an input voltage of ~ 10 VDC.
LEB RF Power Amplifier Screen Power Supply
The rf power amplifier screen power supply is a rack-mounted, air-cooled DC power

supply, adjustable from ~2000 V, with output current up to 1000 rnA. The output current
846
will be pulse duty since the screen of the power amplifier tube will draw current only during
the 50-ms time interval when rf drive is applied.
Input. 208 V AC, ±1O%; 3 phase; 57-63 Hz.
Output. 0-2000 VDC. The output current may be from 0-1000 rnA pulsed or
continuous load. The power supply ripple must be less than 0.1 %.
LEB RF Power Amplifier Bias Power Supply
The rf power amplifier bias power supply is a rack-mounted, air-cooled DC power

supply, adjustable from 0-600 V, with output current up to 500 rnA.
Input. 120 VAC, ±1O%; single phase; 57-63 Hz.
Output. 0-600 VDC, with output current from 0-500 rnA. The power supply ripple
must be less than 0.1 %.
SUMMARY STATUS
The cavity and tuner are being designed at SSC with support and/or collaboration with
numerous other labs and consultants, including Los Alamos National Laboratory, TRIUMF,
and INP Novosibirsk, Russia. It appears likely that two different prototype designs will be
built and tested.
The power amplifier will soon be procured from industry on a "build-to-print" contract.
A prototype solid-state driver has been ordered via a competitive bid and will soon be
delivered.
The ferrite bias power supply is in the procurement cycle. SSC is designing and plans to
build the current controller for the ferrite bias power supply.
The prototype supervisory control system and the low level rf systems are being
developed in cooperation with Los Alamos National Lab.
All other elements required for testing a prototype system are already in place at SSe.
REFERENCE
1. W. R. Smythe, C. C. Friedrichs, and L. S. WaIling, "Proton Synchrotron rf Cavity Mode Damper Tests,"
IEEE Particle Accelerator Coriference, May 6-9, 1991.
847
ENGINEERING DESIGN CONSIDERATIONS FOR THE ENERGY STORAGE INDUCTORS USED IN
THE lEB POWER SUPPLY
Dennis Pavlik
Pittsburgh, PA. 15235
ABSTRACT
Energy Storage Inductors (ESIs) play an important part in the
operation of the power supplies for the Low Energy Booster (LEB) ring of
the SSC. There are a series of 12 ESIs connected in series for the LEB
ring. Successful operation of the power supply requires that the ESIs
operate in a linear fashion with a narrow tolerance on inductance varia-
tion (plus or minus 1.25%).
This paper describes the engineering process for selecting the
appropriate design configuration for these inductors and the optimization
criteria used to evaluate the different design concepts. Three distinctly
different concepts were eva luted. These include: a solenoid in a box
design, a toroidal "air-core" arrangement, and a two-legged core picture
frame concept. The advantages and disadvantages of each are discussed.
BACKGROUND
The magnetic field of the bending and focusing magnets in synchrotrons
must vary in precise synronization with the velocity increase of the par-
ticles being accelerated. This field variation is achieved by changing
the current through the magnets at the proper amplitude and frequency
needed to keep the particle on orbit. The energy variation in the magnetic
field during each acceleration cycle represents a large reactive load to
the power source. This means that the power supply must be large compared
to the actual power supplied to the magnets.
In 1956, White and others[l] introduced a resonant charging scheme
which: 1) reduced the cost and complexity of the magnet power supplies,
2) minimized the reactive power requirements from the electrical grid and,
3) kept the magnets in synchronization. The basic White circuit shown in
Fig.1 uses an inductor and capacitor resonant at the accelerator operating
frequency and a d.c. power supply to pass the bias current. Energy is trans-
ferred back and forth between the magnet, capacitor, and inductor in pre-
cise synchronization with the particle acceleration. Power is fed to the
system via a secondary winding on the inductor. Because the excitation
energy oscillates between the magnet and inductor the power supply does
not see this large reactive load and need only be sized to make up for the
losses in the system. This results in a considerable savings and minimizes
pertubations to the power system grid. Synchronization is provided by
connecting the individual cells in series.
ENERGY STORAGE
INDUCTOR
\
BEAM BENDING
MAGNET
RESONATING
CAPACITORS
~
/'\
S~
Figure 1 Schematic diagram for a basic White ciruit resonant charging

circuit module.
SYSTEM SPECIFICATIONS
The Low Energy Booster (LEB) ring of the Super Collider uses a varia-
tion on the basic resonant power supply, however, the operating principles
are as just described. The LEB magnet system consists of 12 resonant
cells comprising 7 dipole and 9 quadrupole magnets, an [SI, capacitors,
and a power supply[2]. Functionally the ESl must be capable of storing
the appropriate energy and injecting it back into the sytem as needed.
The best configuration for an ESl is determined by the specific sys-
tem requirements for the device. The system specification can be broken
into three general catagories: 1) Performance requirements, 2) Environ-
mental conditions, and 3) Economic limitations. Although important to the
final selection of a specific inductor, it is not necessary to consider
all of the environmental or economic limitations when determining the best
physical configuration for the storage inductor. Only those aspects rele-
vant to that decision were considered in this work.
The performance requirements specify the operating conditions for the
inductor and how it matches to the power supply. This includes the term-
inal voltage and current, total stored energy, and allowable variations.
Environmental conditions specify the operating temperature, available
resources for cooling, ambient conditions, and limitations on the
materials used in the construction of the inductor.
The performance requirements and operating environment specification
for the LEB E51 as defined by the Superconducting Super Collider Labora-
tory (SSCL) are given in Table 1. The ESl has two windings; one for
energy storage connected directly to the magnet, and the other for inject-
ing makeup a.c. power to the magnet system. The operating requirements of
the load winding, designated as coil 2 drive the design.
Items 1 thru 11 in table 1 are the performance parameters which
define the interface of the ESl with the LEB ring and its power supply.
These items establish the basic size of the ESl for any specified physical
arrangement. Items 11 thru 17 are a partial list of the environmental
restrictions and define the interface of the ESl to its surroundings.
The size of an energy storage inductor is driven first by its energy
storage requirement and next by the current density at which the winding
can operate. The energy storage is fixed at the outset based on the
operating current and inductance limits. The operating current density is
850
fixed by either an efficiency limit or the acceptable cooling schemes. The
limitations on materials, availability of power and water, and the
reliability requirements specified by SSCL[3] restrict the LEB ESIs to
some type of oil-cooled, natural convection configuration. Although there
is an efficiency goal, the thermal limitations are the dominant factor for
this system.
Table 1. LEB Energy Storage Inductor System Performance and

Environmental limitations.
Description Value
1. Operating Frequency 10 Hz.
2. Current waveform (coil 1) -100oSin(62.8 ot) amperes
3. Current waveform (coil 2) 2000 + 1250·Sin(62.S·t) amperes
4. Current Peak (coil 2) 3250 amperes
5. Current RMS (coil 2) 21S6 amperes
6. Fault Current surge, peak (coil 2) 4000 amperes
7. Voltage across terminals peak (coil 2) 3300 volts
8. Voltage to ground, peak (coil 2) 1700 volts
9. Inductance (coil 2) 40.0 mH
10. Inductance tolerance (coil 2) ± 0.5 mH
11. Inductance nonlinearity (coil 2) 1%
12. Maximum stray field @ 10 meters 0.001 tesla
13. Maximum coil temp rise above ambient 65 °c
14. Maximum hot spot temperature rise SO °c
15. Maximum ambient temperature 45 °c
16. Minimum ambient temperature -25 °c
17. Maximum relative humidity 100%
a) SOLE OlD IN A BOX b) TWO-LEGGED P ICTURE
WINDING
MAGNET IC
FL UX LI NES
c) IDEAL T OROID
Figure 2 Schematic representations of the different energy storage induc-

tor concepts evaluated. a) Solenoid in a box, b) Air-core
toroid, c) Two-winding picture frame.
851
DESIGN EVALUATIONS
The three different design concepts were selected for evaluation
were: 1) Solenoid in a box, 2) Toroidal "air-core" arrangement, and 3)
Two-legged core, picture frame configuration as illustrated in Fig. 2.
These concepts were selected because of their linear inductance charac-
teristics, simplicity, ruggedness, reliability, and relatively low stray
field characteristics. The details of each evaluation study will now be
discussed.
Solenoid In A Box
In this concept, a solenoidal air-core inductor is enclosed in a
magnetic frame. Because the field contains alternating currents, the
frame is made of thin silicon steel laminations. The requirement on
linearity (item 11 Table 1) and stray field (item 12 Table 1) establish
the maximum operating flux density of the surrounding frame at 1.47 T
maximum.
For comparitive sizing purposes, the inductance can be based on the
assumption that magnetic field approximates that of an infinitely long
solenoid over the winding height. The flux density as a function of
radial position, is given by Eq. 1 for the area inside the winding (i.e. r
~ r 1) and Eq. 2 for points inside the winding region (i.e. r 1 ~ r ~ r 2).
B(r) Bz(r) = #·N·I/1e (1)
B(r) Bz(r) = [#·N·I/1e].(r 2 - r)/(r 2 - r 1 ) (2)
where N is the number of turns in the winding, I is the current, 1e is the

effective magnetic path length, and r1 and r2 are the inner and outer
radii of the winding. Equations 1 and 2 are integrated to determine the
total flux which is used in conjunction with the core limiting flux
density to size the steel outer box. The inductance is related to the
energy stored in the magnetic field, E, by
L = 2·E/I 2 (3)
Using the design specifications, Eqs. 1,2, and 3, and limits on oper-
ating current density an algorithm was developed for sizing solenoid in a
box energy storage inductors. A series of parametric studies were made
for different current densities as the number of turns and window flux
density were varied.
A plot of the variations in coil volume, total weight, and coil loss-
es as a function of flux density for a fixed number of turns, N= III and a
fixed coil current density of 1650 Amps/in2 is shown in Fig. 3. Observe,
that there is a definite minimum to the loss and total volume of the
acitve parts of the inductor. Both volume and loss decay rapidly and then
begin to grow slowly as the optimum flux density is raised. A second study
was conducted with the flux density held constant and the number of turns
varied. The total weight of the inductor asyimtotica11y decreases to a
minimum value. However, the losses increase without limit as the number
of turns are increased as illustrated by Fig. 4.
These studies were repeated over the practical range of current den-
sities and compared. The best combination of minimum weight and loss
seemed to occur between 100 and 200 turns at flux densities between 0.6
and 1 T. The minima for both loss and weight were most pronounced at the
852
lower current densities. The flux density at which mlnlmum loss and
weight occured increased with increasing current density. A footprint for
the best design selected as a result of these studies is shown in Fig. 5.
It has III turns, operates at a flux density of 0.6 T in the gap. 'The
total weight of coil, core, tank, oil, and all fittings is 108,000 lbs.
The 12R loss is 90 kw with an added 25 kw for eddy currents, core loss
glvlng a total of 115 kw. This is marginally higher than the targets
specified in the Site Conceptual design[3], however, the inductance is
also 15% higher.
80 200
~
en
c
0 ,
IiJ Westinghouse STC
70 ,, TURNS = 111
,, = AI sq 3
I
f-
60 , Jden 1650
150 ~
~ &,,
W
50 ,, PAC. FAG. = OA
Vi
3: 15\., Vi
.. ' .
~ 40 100 0
s·~.B_~~_e-~_8_~~·B-~-~ ---1
"'"
$2 30 ---1
«
'---
C 20 ..,, ~,
'G
-"().-@_€>-oo()o_e_...()o_9 .. eo--<1--e- Q - 9 .. €l-..o 50
f-
0
". '&-e.- -o--A_~
~
f-
w 10 -.!r" 6- 9 -Al- -6- -~- 6--t:r-.!I--6
::::;:
::::>
--'
0 0 0
> 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
FLUX DENSITY (T)

--6-- coil --E>-- core - weigh --8-- loss
Figure 3 Variation in inductor size, weight and losses as the operting

flux density is varied.
----c
(!)
100
Westi~e STC
250
.....
0
'-../
FLUX DEN. = 0.6
80 Jden = 1650 AI s 200

f-
I PAC. FAG. = OA
Cl
w 3
60 150 ~
3 '-...../
~ Vi
Vi
----
.Y 40 100 0
S2 ---1
::;----
c 20 50
w
::::;:
::> 0 0
--' 0 50 100 150 300 350 400
0 200 250
>
NUMBER OF COIL TURNS
--6-- coil --E>-- core _ weigh .... G.. ·· loss
Figure 4 Variation in inductor size, weight, and loss as the number of

turns is varied with flux and current densities held constant.
853
r
115.5
1 11.875
L--------.-----~
III I
li-oII4~---115.5----~~ 62.125
Figure 5 Solenoid in a box dimensions for the best combination of losses

and weight based on a current density of 1650 A/in2 and III turns.
TOROIDAL INDUCTOR
Continuously wound toroids of cylindrical form using both single and
multiple layers of conductor were investigated. The inductance of a
toroid having major radius T, minor radius R, and N turns is given by
(4)
Using a method developed by Murgatroyd[4], the maximum inductance for

a single layer toroid for a given length of conductor, d, is given by
Lc = Lo 0 [ 0.27217 0(w/d)3/2 + O.25 0(w/d)] (5)
where Lc is the desired inductance, Lo is a baseline inductance equal to

~d/n, and d is the diameter of the conductor (including insulation). The
diameter of the wire is determined from the operating current and a speci-
fication on current density. Using 1650 A/in2 and back substituting into
Eqs. 5 and 4, the minimum size and shape of a single layer toroid which
met the specificaiton required a box 28 ft square by 16.25 ft high just
for the toroid, not counting cooling fluid, tankage, or fittings. The
copper weight alone was 52935 lbs and had 361 kw of loss. This is clearly
not competitive with the solenoid in a box design.
Equation 5 was extended to include multiple layer toroid configura-
tions, assuming the radial winding build to be negligible. For the con-
ductor dimensions of interest in this work, that limits the usefulness of
the formulation to toroids with less than 4 layers. Subject to the above
assumptions the maximum inductance for a multiple layer toroid is given by
Lc = Lo 01 1 / 2 o[ 0.272l70(w/d)3/2 + 0.25 0(w/d)] {for 1<4} (6)
where 1 is the number of layers. Equations 6 and 4 were used to determine

the copper weights and basic dimensions for inductors using different num-
bers of layers from 1 to 4. The current density had to be reduced from
854
1650 A/in2 to 1000 A/in2 to keep the losses competitive with the solenoid
in a box. A graph of the major dimensions, copper loss, and the combined
weight of copper and winding form as a function of the number of layers is
shown in Fig. 6. Note that this analysis doesn't account for the radial
buildup of the winding layers, therefore, these numbers should be viewed
as minimum values. A 4 layer toroid produces the lowest loss inductor,
but even so both its losses and weight are significantly greater than the
equivalent solenoid in a box.
If a segmented or cage type toroid[5] of "Princeton 0" type construc-
tion were used instead of a continuously wound toroid, the inductance for a
given length of wire could be increased by about 15% according to Chuboy[6].
However, the fringing flux would also increase and some external shielding
may be required for this marginal gain .
..--.. 250 300

'"
(0
C WestqT>ouse STC
R
(T)
0
....
.".................. .................
..--...
-+-'
'--"
250
I- 200 ~
I ..::t:.
'... .. ..................
Q ' 200
i.E------ T~
'....
w (f)
s. __________ ....
.
3: 150 (f)
a:: ----------- 150 0
0
•.....•......
-
.....J
'. '.
~ 100 ...
'. '.
........... .....J
-... __ .... -
'().
'. '
----- .. _-- 100 <x:
Z • • • • • • • • o()o ..
0 l-
v; "- .....
z 0
UJ 50 I-
:::I; 50
a
--'
...
~
c 0 0
a::
1 2 3 4 5
NUMBER OF LAYERS ( t)
--A--
--e·· major - weight --s-- lass
minor
Figure 6 Basic dimensions, weight, and loss as a function of the number

of layers in a multilayer toroidal inductor.
TWO-LEGGED PICTURE FRAME

Two separate winding connections can be used in the two-legged
pictu~e frame construction; flux aiding and flux opposing. In the flux
opposlng arrangement, the fluxes generated by each winding are oppositely
directed within the core and the net inductance is defined by the leakage
flux paths. Although the window region leakage flux predominates, it is
not as well defined as for the solenoid in a box making calculations much
more difficult. Furthermore, a substantial portion of the leakage flux
is external to the core which must be contained if the field is to be
limited to 0.001 T at 10 m from the device. This means that some form of
magnetic shielding will be required on the tank walls (eddy current
shields aren't practical at the LEB frequency). The tank becomes a
defacto containment magnetic core. The net coil weight is somewhat less
than for the solenoid in a box, but this is offset by increased tank
weight and complexity. Because the inductance is controlled by the
leakage field both internal and external to the core window this design
configuration will be sensitive to variations in construction, positioning
of the winding, and positioning withing the tank making it much more
difficult to control the inductance within tolerance limits.
855
The flux aiding arrangment with continuous magnetic cores can't be
used for energy storage inductors. "Air-gaps" must be placed in the legs
of the core to allow for energy storage and to obtain a linear inductance
characteristic. This gapped core, two-winding arrangement offers some
advantages over the solenoid in a box. Two small diameter coils are easier
to build than one large diameter coil, their combined weight will be less
than the large coil, and the losses will be lower. The core manufacture
mayor may not be more difficult depending on the final core dimensions
and manufacturing procedures. Alternate core structures were investigated
by Pav1ik[7] for accelerator energy storage inductors. Whether or not the
two-legged, flux-aiding arrangment is better than a solenoid in a box
depends mainly on economic and manufacturing factors as they are compar-
rab1e in size and weight. If eddy currents are significant such as in the
Sasaki[8] high power density design, the two-legged flux aiding arrange-
ment can reduce the losses significant1y[7] at the price of increased
weight.
OTHER FACTORS
Economically, the best inductor is not necessarily the one with
the lowest inita1 purchase price. For instance, even a free inductor may
in fact be very costly if doesn't meet the allocated availability require-
ment of 0.999973 and reliability of 0.38 failures per million hours. The
cost of down time must be factored into the price evaluation criterion, as
well as the availability of repair facilities and service technicians.
Ut il ity companies typically evaluate purchases of capital equipment
on a total owning cost basis. They consider the inita1 purchase price,
the cost of anci11iary equipment costs, operating and maintainence costs,
reliability requirements, and the cost impact on the system. This
approach should also be applied to the selection of equipment for the
Super Co 11 i der.
CONCLUSIONS
Three different concepts for energy storage inductors were considered
in this paper. Based on the system requirements imposed by the SSCL the
best choice for these inductors is the solenoid in a box arrangment. It
is readily analyzed, has a low stray field, has the most controllable and
repeatable inductance. Its only potential drawbacks are that the windings
will be of relatively large diameter and the core will be deep by industry
standards. This makes manufacturing and assembly more difficult.
The other viable alternative is the two-legged, flux aiding picture
frame arrangment. This design is more sensitive to tolerances, especially
in the small magnetic material gaps. Core construction can become compli-
cated and must be evaluated against the winding and core construction
problems in the solenoid in a box arrangemnet. The choice between the two
will be based on manufacturabi1ity, shop practice, and material costs as
both can be made to meet the specifications.
REFERENCES
[1] M.G. White, F.C. Shoemaker, & G.K. O'Neill, "The 3 BEV High Intensity
Proton Synchrotron," CERN Sympos i urn of 1956, Vo 1. 1, P 525.
[2] J.R. Sanford, and D.M. Matthews, Site-Specific Design of the Super-
conducting Super Co11ider Section 4.2.3.3 Low Energy Booster
Magnets, July 1990.
856
[3] C. Jach, "Energy Storage Inductor Specification," ESI Preliminary
Design Review, SSCL document, July 1991.
[4] P.N. Murgatroyd, et. al., "Economic Designs for Single-layer Toroidal
Inductors," lEE Proceedings, Vol. 132, Pt. B. No.6, Nov. 1985.
[5] P.N. Murgatroyd, "Some Optimum Shapers for Toroidal Inductors," lEE
Proceedings, Vol. 129, Pt. B, No.3, May 1982.
[6] H.C. Chuboy, "The Effect of Modularity on the Inductance of Toroidal
Magnetic Energy Storage Coils," Thesis for Masters of Science Deg-
ree in Electrical Engineering at University of Pittsburgh, 1991.
[7] D. Pavlik, "A Review of Energy Storage Chokes For Rapid Cycling Synch-
rotrons," TRIUMF KAON Facility Design Notes, edited by A.J. Otter,
University of British Columbia, Vancouver 1989.
[8] H. Sasaki, T Adachi, H. Someya, & I. Sakai, "A DC Biased Rapid-cycling
Magnet System Operating in a Dual Frequency Mode," Proceedings of
the KAON PDS Magnet Design Workshop, 1989, TRIUMF, Vancouver, B.C.
Canada
857
MEASURED CONTROL CHARACTERISTICS OF THE HALF-CELL
40MM APERTURE MAGNET STRING
A.D. McInturff,l J.G. Weisend 112,

C.E. Dickey,2,3 R. Flora,l and D.B. Wallis 2
Accelerator Division
lFermi National Accelerator Laboratory*
Batavia, Illinois 60510
2Superconducting Super Collider Laboratory
Dallas, Texas 75237
3Duke University, North Carolina
ABSTRACT
The data presented here were obtained in the course of operating a five 40mm aperture
dipole string. The eighty eight meter long string of dipoles was assembled to test the various
proposed operational scenarios of the SSCL collider. As reported earlier,l there had been a
short control and system (data procurement) checkout run performed on an abbreviated two
dipole string. The problems that were then uncovered with the exception of the high 20K
shield heat load, were corrected. There has been over 5000 hours of running time on the
system, 3000 hours on the five magnet string alone, and to date, no major problems or
incidents have occurred.
The quench (superconducting to normal transition) performance of the magnet string was
excellent, with the exception of four premature quenches that occurred during power supply
commissioning. The operational parameters were all found to be manageable or equal to or
greater than design. The operational heat loads were within the budget with the exception of
the 20K circuit which was a factor of three too high. The relative internal voltages of the
magnets have been higher than previously measured in the shorter string by a significant
amount. There will be a discussion of concerns and problems plus their possible solution.
INTRODUCTION
A Superconducting Super Collider "ssc" size tunnel section with a prototype control room,
"ER-4" which is capable of operating multiple cells of the "ssc" lattice has been constructed
by Fermilab and SSCL personnel near the Tevatron's E-4 service building. There are
presently installed and operating five dipoles and a "SPR" spool piece (1/2 cell). The data,
however, that will be presented in this paper are for the recently completed five dipole run
only. The ER-4 operating system for this particular run consisted of A) computer controlled
and monitored 5 Atm ab helium gas refrigerator producing up to 0.6 kilowatt in a 3.8K to 5K
temperature range which used a reciprocating cold compressor; B) an 8.0kA SSCL prototype
inverting computer controlled 2 power supply creating prototypical accelerator ramps; and C)
a quench protection monitor or "QPM"3 which determines the status of the superconducting
elements (resistive or not) and their associated systems (i.e., heater firing circuit, security
*Operated by Universities Research Association, Inc., for the Department of Energy

gates, power supply, refrigerator, etc.). In addition, to this "ACNET"4 based accelerator
control system, there is a stand alone "UNIX" operating environment, Research
Instrumentation Data Acquisition System (RIDAS)5. This system receives an independent
signal from different quench detection circuits "QDC's" which can operate independently
by firing the magnet protection heaters in parallel with the "QPM" through a TTL "OR"
gate. This data acquisition system can obtain both long term and transient data. The long
term system typically measures 140 - 180 plus devices every five minutes. The operator can
continuously plot or recall them at his leisure. The high speed circular buffers presently
contain 96 channels, although it is possible to expand this in blocks of 32. The final phase of
this experiment is currently underway and includes a spool piece with a recooler, safety
leads, quench relief valve, and vacuum break, in addition to the five dipoles. The system
protection consists solely of the strip heaters which raise the superconductor in the magnets
above their transition temperature.
There were several operational scenarios envisioned for collider operation which were
novel and different than those of the present superconducting machines (Tevatron, HERA).
Pressures, voltages, temperatures, valve positions, and flows are needed to validate or
invalidate the ideas and the various computer code simulations assumptions.
EXPERIMENTAL PROCEDURE
The primary test objective is to verify that "SSC" magnets can be installed, leak checked,
cooled to LHe temperatures, energized in an acceleration cycle, and quenched repeatedly as
a string in a tunnel type environment. There were secondary objectives which involve
various sub-systems, developments, operating parameters, and scenarios, and above all,
gaining operational system experience with these devices. The experimental objectives can
be divided as follows: A) refrigeration requirements and control, B) powering parameters
and control, C) installation, and D) quench control and prevention. The operating
accelerator control system was similar to that of an upgraded Tevatron, except where there
were systems far enough along in development by SSCL to be employed. An example of this
was replacing the FNAL low '13' quad prototype with an SSCL 8kA inverting power supply.
The string magnets have been individually tested. After reception at ER, they are
mechanically and electrically checked at room temperature and compared with the previous
test results. These tests include vacuum, physical inspection, high voltage (insulation
integrity), continuity, inductance and resistance checks, as well as voltage tap position
verification. The magnets were then modified for string use. The modifications included
installation of magneUmagnet interconnect busses and instrumentation busses and the
unblocking of cooling passages (removed cross flow plugs).
The magnets were only roughly aligned (about an order of magnitude less accurate than
required for machine operation) for mechanical purposes, in that it was not envisioned to
accept beam. Once the magnets were in place and electrically connected, the first of three
electrical checks were made. Then after passing the first electrical check, all leads were
strain relieved, checked for contraction clearance, then insulated, with the interconnect
bellows pulled into place, but not welded. At this point, the power circuits were high potted and
the second complete electrical check out was made. The various cryogen circuits are then
welded and the vacuum vessel is closed as a high pressure (21.5 Atm) pneumatic test of the
cold mass/liquid helium return line is performed. The circuits are simultaneously vacuum
leak checked and certified tight.
The cooldown to final operating parameters of 4.35K, 5 Atm ab, and 50g/sec of gaseous
helium was performed uneventfully. The rate is governed by a 100K temperature gradient
limit across any device. The daily power-up check list is similar to an operating accelerator
including personnel lockout procedures. The string was powered in a stepwise fashion.
This enables the operator to estimate fairly accurately, the worse case scenario to be
encountered during the next current increase using the data from the last. The cryogenic
load data were continually refined throughout the run. The observed thermal equilibrium
time constants were measured in hours and/or days. The heat load data was obtained using
the VXI5 based data acquisition system. It recorded temperatures, pressures, flows, and
strains every five minutes throughout the five month run.
The various cryogenic circuit heat loads were determined by observing the temperature
gradient across the string and point to point, at a given gas flow rate. Flow rates were
860
determined or cross-checked in two ways. The primary method used a precision gas volume
meter versus time. The second was to put a known local load into one of the systems resistors
and note the increase in temperature. The various interconnects were instrumented to
determine their gradients. The heat loads of interest were with the cryostat circuits near or
at operational temperatures, i.e., SOK at SOK, 20K at 20K and cold mass at 4.3K There were,
however, different cryogenic loads determined with the various circuits at non-operational
temperatures.
There were five spontaneous quenches during the course of testing. These were: 5700A
D027, 6145A D026, 6290 D0201, 6370 D027, and 6550 D027. D027 quenches were in the outer
coils. There were a total of 19 strip heater induced quenches that were operator initiated. A
total of 29 quenches or power trips (phase back) occurred during the run. Some of these
occurred while exploring various operating parameter's ranges like lead flow rates. In all
but one case, there were data acquired by three high speed (two 1 kilohertz and one 10 hertz)
CAMAC Data Loggers and a four channel 100 kilohertz storage scope. There is also a 60 Hz
circular buffer in the QPM system. It recorded all of the events with the exception of a few of
the power supply phase backs.
The main electrical power circuit contains four terminals. Normally, the two leads at the
return can were shorted. All of the dipoles were in series on the same buss.(NOTE: this
configuration should result in the highest voltages). The "QPM" monitored coil halves to
decide when and where a quench had occurred. The "RIDAS" system monitored all but one
of the quarter coil voltages. The system was ramped at an accelerator cycle rate 3 times
faster than the one proposed for the collider. Therefore the number of cycles obtained in a
given time period increased. Collider peak energy stability studies were done. The criteria
used was that the system came to equilibrium and stayed there for longer than 4 hours. There
were a series of above peak operating field induced quenches. The valve configuration for
relief of the quench pressure was systematically changed in order to explore different
operating scenarios. Each of the operating relief valves had a pressure transducer and
thermometer near their outlets. (NOTE: the coded safety valves are a different set of
valves). See Figure 1. These valves had the following operating modes: A) they opened
when the internal pressure exceeded the backing pressure (Kautzky 6 type valves), and B)
they could be operated by venting the backing pressure on the signal from the QDC's. Those
modes could be additionally modified either to stay open or to reclose once the internal
pressure is less than the backing pressure. The valves had two different outlet circuits.
They could vent into either: A) the warm helium gas return line to the suction side of the
compressors, or B) the 20K helium gas circuit at either the return can or to the 20K point in the
refrigerator heat exchanger. The number of valves activated varied from all six to one
valve only for the entire half cell.
Quench recovery was also briefly studied. The Tevatron's Central Refrigerator's liquid
transfer line could be opened after the quench pressure subsided. The transfer simulates the
enthalpy of the remainder of the machine (collider) being utilized to speed up recovery.
RESULTS
The deficiencies encountered during the two dipole assembly specifically, the
longitudinal stability of the shield and the multilayer insulation (MLI) were apparently
solved as evidenced by their absence in the five dipole run. The interconnect that was
disassembled for the installation of the "SPR" spool for the present run was in the same
configuration as before the start of the run. The stability and suitability of the interconnect
was also evidenced by the very small temperature gradients measured across the
interconnects. There had been concerns about the vacuum in between layers in the "MLI".
Therefore, thermometers were placed in between "MLI" layers of the two SOK shield
blankets. The temperature gradient appeared to be linear with layer number. Therefore, the
heat flow was dominated by conductance (through the aluminized mylar).
The cooldown rate was primarily determined by experimental constraints. The first was
the lOOK maximum gradient across any given magnet. The second was the hold at lOOK and
20K for electrical measurements. The actual cooldown to SOK was 3.3 days for the 70 ton
string. No voltage taps and only three devices failed during the cool down cycle. The system
went through various calibration tests and electrical verification steps at temperature. The
thermal load was determined for the various circuits, i.e., SOK and 20K, but not the cold mass
861
SUCTION
IMIYJ
VENT
.
if
!
...
~ ...""
Q
.j
..
~
0: 0:
~ IL~
IL
"" ""
0
0
0 0
0
0
..... .....
EVSD
.,
L
48
..
~
Z
0::
::J
C C I- Z
...J ...J IJJ 0::
IJJ IJJ 0:: ::l
::c -::c VI
I-
IJJ
VI VI «(!) 0::
::.:: ::.:: Q)
0 0 Q) ::c
(l) N ::c ...J
5 ATM He
Figure 1. This is a schematic drawing of the valves and circuit flow in the return can.
These circuits are to the left of those in Figure 2 and connected one to one, i.e., 20K to 20K
shield. The "Kautzky" valves ("KV") are the quench activated valves as described in the
text. The "SV" valves are the "coded" relief valves and as such not operation valves.
862
DOWNSTREAM
SVSN END CAN
80K NITROGEN
®
l'
SVSH
GAS RETURN
LIQUID RETURN
D201 S002
JUNCTION JUNCTION
Figure 2. This is an instrumentation diagram of a part of the 40mm aperture 112 cell. It
contains most of the types of devices used. (HW = heater; PI = pressure indicator, 70 series are
cold sensors all other PI's are room temperature); SD = silicon diode; SG = strain gage; TI =
vapor bulb thermometer; TL = cryogenic linear temperature sensor; TP = platinum
resistance thermometer; TR = thermometers (30 series are Ge resistors, 40,50, and 60 series
are carbon resistors), or C series are carbon glass resistors, G series are germanium
resistors, SDR series are silicon diodes located in the recooler.
(although it was bracketed just to determine the measurement range). The various
cryogenic circuit heat loads resulted in gradients across the string of degrees. The cold
mass circuit was an exception where it was about l.5K with an uncertainty of a quarter of a
degree. The temperature variation at a given point was equal to or less than a couple of
tenths of a degree. The temperature variation at the inlet or the outlet was on the order of a
degree. The largest measurement uncertainty was the flow determination which was cross
checked by introduction of a load and observing the temperature increase. This check
resulted in a discrepancy of about 20%. This maybe a result of incomplete mixing, i.e.,
stratification. In the cold mass, only the ''known'' load method was used to determine the
flow. The heat load data are presented in Table l.
The typical instrumentation package of the string is shown in Figure 2. It can be seen that
there was ample redundancy in most sensors, to enable the operator to identify an erroneous
device and correct it.
The string field level was raised in a step wise fashion to the peak operating value.
Therefore data from previously induced strip heater quench could be used to predict the
parameters during the next induced higher current quench. Unlike the previous run,
spontaneous training quenches were encountered during this phase. These premature
863
TABLE 1- Heat Load Summary (Heat load represents 1 dipole plus interconnect)
SOKSHIEI.D
This is with the 20K shield operating at 20± 1K Tav (lBK - 23K). The cold mass temperature
was between 4.5 to 4.3K The temperature range of the BOK shield was between 99K and 131K
Budget Flow D017 0026 0019 0027 D201
27W 3.3g1sec 1B.2±2W 17.2±2W 31.0±2W 23.0±2W 2B.0±2W
20KSHIELD
This is with the BOK shield operating at 7B - BOK and the cold mass temperature between 4.5K -
4.3K The temperature range of the 20Kshield was between 20K- 25K
Budget Flow D017 0026 D019 0027 D201
3.3W O.85g1sec 12.9±3.0W 6.4±2.0W 7.0±1.5W 9.0±2.0W 6.0±2.0W
Cold Mass Range Estimate

The BOK shield at 79K - BOK and the 20K shield between 19K - 22K The cold mass temperature
was 4.6K
Budget Flow Average load = Magnet + Interconnect
0.2BOW* (4.B4 - 2.4 glsec) 0.3W ±.2W
*This does not represent the major load which is an order of magnitude larger, namely the
synchrotron radiation load.
transitions resulted in the highest operational voltages recorded to date. All but one of these
quenches however occurred at <!:95% of maximum operational field. The QDC's and QPM's,
both diagnosed the presence of resistance accurately, and early. The "QPM" initiated the
protection strip heater firings. The resulting resistance brought the current down rapidly
enough not to unduly heat the windings. In Table 2, the events (data taken) are listed as well
as their "MIlTS" (million ampere squared seconds) if applicable. The delay time is the
period between heater firing and the coil volume under them becoming resistive. The
voltage listed is either with respect to ground or if it is in parenthesis 0 with respect to another
coil.
The primary quench experiment objective was to measure the pressures produced in the
system and their time profiles. The other parameters of interest included voltage to ground,
coil temperatures, and coil to coil voltages. Toward the end of the studies, a brief look at
recovery was taken. Recovery means the cold mass parameters are such that powering to
full field is possible. Figure 3 shows the intermagnet pressure as a function of time. There
are various valve scenarios plotted. They are: (see Figure 1) A) all "Kautzky" valves (K
type) opened and held open for 1 minute, KVUL, KVDL, KVDS, KVUS, feedean cold Kautzky,
=UCKV", DCKV; B) all valves (K type) opened at the feed can and held open for 1 minute; C)
both cold valves (20K shield, UCKV and DCKV) are opened but not held open; D) both cold
valves are opened and held 1 minute; and E) one cold valve, (DCKV) opened and held for 1
minute. The important observation here is that it is adequate to have only one cold quench
relief valve per half cell. The maximum pressure differential across the five magnet string
was less than 140kPa and the maximum pressure was 1.25MPa. It appears if there are two
valves of this type, and one fails and only one valve per cell opens, it will be sufficient. The
reason for these low pressure increases is explained by the temperature measured with the
exhaust line diodes. The 20K line diode has lowered to <5K after 300ms (about coincident
with pressure peak). It only takes about 10 seconds for the warm header diode to reach <6K
near the outlet. These very low temperatures create a high conductance through the valves,
therefore, the low pressure drops.
The operational result that presents a problem is the high voltages encountered during the
spontaneous quenches. This is partially due to the small coil volume affected by the
protection strip heaters. The coils had distributed spot heaters along the heater strips. This
requires the quench to propagate both longitudinally and tangently. The 50mm magnet
heaters have double the number of turns and larger lengths of each turn under the heater. It
is also planned, in the accelerator, to split the dipoles - three on upper buss and two on the
lower buss. This combination should lead to lower overall voltages.
The only comment that can be made about quench recovery was that it expedited recovery if
the quench valve was left open at least for 1 minute after the quench. When the peak liquid
864
TABLE 2 - QuenchlPhase Back Event Listing
Event Current Description MIlTS Delay Voltage
(Sec) Ground or
(Relative)
1 1.3kA Power Supply Trip 1.5 .500 (36)
2 2.0kA Power Lead Trip 3.4 .25 (60)
3 3.0kA Strip Heater Initiate 3.0 .14 (130)
6 2.5kA Power Supply Induced 3.4 .22 (65)
7 5.7kA D27LO Spontaneous 10.0 - .3 (350)
9 6.2kA D26UO Spontaneous 8.9 -.03 1075
10 6.3kA D201LO Spontaneous 9.1 -.03 (356)
11 6.4kA D27UO Spontaneous 9.2 -.03 975(1300)
12-14*
l5 5.5kA Strip Heater Initiate 5.3 .09 626
16 6.0kA Lead Trip Inducted 5.1 .08 623
17 6.4kA Transd Failure 6.5 .07 663
18 6.6kA D27LO Spontaneous 8.7 -.02 861
19 6.5kA Refrigerator Trip 5.4 .07 635
20-29*
*COMMENT: Events #12 - 14 and #20 - 29 the first group,were strip heater induced at a
current of 6.5kA, and 6.6kA in the second. The thermal diffusion time was between 60 and
70 milliseconds. The number of "MIlTs" varied from 5.3 to 7.2. The voltage to ground was
in the 500V to 750V range.
flow rate from the central "Tevatron" liquefier was about 500 ]/hr, recovery time was about 3-
4 hours. This represents a moderate rate of turnaround (6+ MJ stored), and not the
maximum rate (minimum recovery time).
During the disassembly of the interconnect between the last dipole and the return can, the
"MLI", the taping, and the shields all seemed to be "as assembled" when compared with the
assembly photos. Inside the cold mass a large amount of dust was found which is thought to
be grinding wheel remnants and stainless steel chips. In the future only cutting tools should
be used for disassembly after individual cryogenic testing. In summary, the operation of the
half cell has not encountered any problems that appear difficult enough to prevent the
operation of the SSC Collider.
ACKNOWLEDGEMENTS
The authors would like to recognize the contribution of the various groups and many
individuals for the completion of the facility and the actual performance of the various
phases of the experiment. There will not be an attempt to order as of merit. The Fermilab
technicians, engineers, and scientists include,R. Brooken, T. Cross, S. Hayes, R. Lively, J.
Lockwood, M. MacPherson, K. Martin, R. Orr, P. Payne, A. Rehbein, G. Saevent, T.
Savoied, J. Theilacker, D. Wolf, and B. West. The SSCL staff includes: R. Applegate, R.
Carlone, W. Clay, J. Darby, D. Freeman, K. Goetze, J. Gran, M. Hentges, F. Kelly, J.
Morris, D. Musser, R. Nehring, T. Neustadt, T. Reeves, B. Smith, W. Smith, G. Sorenson,
G. Tool, D. Voy, and C. White. The data has been independently analyzed by S.
Augustynowicz, M. Levin, and W. Robinson of SSCL. The instrumentation and
installation of interconnect insulation blankets was done by R. Kunzdmen, and R. Tuskey,
directed by W. Boroski (FNAL) There were many others who contributed before the authors
were involved and therefore unknown to them. We apologize for any oversights, a general
thanks to them as well.
865
Cold Mass Quench Pressure
kPa
1260
A
1120
980
840
l~C~
700
\~E~ ~
560
420 \ B,\"" ~C~ f:::::;E-

O---=:::::'-
"B-, 0V-- 0-
280
~ _B
"'-v--
140 \ ........ A_
B-
A A
o
-25.00 0.00 50.00 100.00
Seconds
Figure 3. These are plots of pressure measurements, PI75, a cold transducer located in the
junction between dipoles D026 and D019. The valve scenarios are as follows: A) KVUL,
KVDL, KVUS, KVDS, UCKV, and DCKV quench activated and held open 1 minute; B)
KVUL, KVUS, UCKV, activated and held 1 minute; C) UCKV, DCKV activated; D) UCKV,
DCKV activated and held 1 minute; and E) DCKV activated and held 1 minute. See text and
Figure 1 for more detailed explanation and system schematic.
REFERENCES
1. A.D. McInturff, R. Flora, B. Norris, J. Theilacker, D.A. Wolf, S.A.

Augustynowicz,C.E. Dickey, G. Tool, D.B. Wallis, and J.G. Wei send II, "Half
Cell "SSC" 40 mm Aperture Magnet String", Proceedings of the Twelfth Int'I.
Conference on Magnet Technology, Leningrad, USSR, June, 1991.
2. O. Calvo, R. Flora, M. MacPherson, "Real Time Control of the SSC String Magnets",
IEEE Transaction on Nuclear Science Yol NS-34 No 4 August, 1987.
3. R.H. Flora and G.S. Tool, "Doubler Tevatron IlP Quench Protection System", lE.E.E
Transaction on Nuclear Science Yol NS-26 No 3 June, 1979.
4. F.J. Nagy, "The Fermilab Accelerator Controls Networking System", Nuclear

Instruments and Methods, A247 (1986), P. 208.
5. C. Dickey, D. Wallis, J. Weisend, K Low, M. Allen, "A VXI Based Data Acquisition
and Control System for Super Collider Magnet Systems and String Tests", Pro-
ceedings of the Test Engineering Conference, Atlanta, GA., June, 1991.
6. Private communication, Hans Kautzky, Tevatron Relief Valve, which was redesigned
and built to ASME Type Safety Code for HERA. The cold relief valve is a
modification and redesign for cold operation by D. Allspach and J. Theilacker
(FNAL).
866
THE PROTECTION SYSTEM FOR THE SUPERCONDUCTING
ELECTROMAGNET RING OF THE UNK
o. V. Afanasiev, A. P. Dubasov, A. N. Erochin, V. I. Gridasov,

B. V. Kazmin, O. V. Kumaev, K. P. Myznikov, M. V. Prima,
A. N. Sytin, V. A. Sytchev, G. V. Tichin, L. M. Vasiliev,
O. M. Veselov, and N. N. Yarygin
Institute for High Energy Physics

Serpukhov, Russia
INTRODUCTION
The superconducting stage of the electromagnet ring of the UNK consists of

2194 dipoles and 472 quadrupoles placed in the 21-km long tunnel. 1 The operating current
of the electromagnet ring is 5250 A and the stored energy is 1.5-109 J. The minimum
temperature reserve in a superconducting (SC) dipole is 0.7-0.8 K. Such a small reserve
makes it highly probable that some areas of the SC cable may quench due to a local heating
caused by such factors as beam losses or malfunctioning of the cryogenic system. The main
purpose of the quench protection system is a timely detection of the normal phase in the
SC cable and taking appropriate measures that ensure the safe removal of the energy stored
in the ring electromagnet and protection of the magnet coils.
GENERAL DESCRIPTION OF THE QUENCH PROTECTION SYSTEM
The quench protection system for the electromagnet ring of the UNK is based (similar
to the Fermi National Accelerator Laboratory [FNAL] machine)2 on dividing the whole ring
into cells, shunting each cell with a thyristor switch placed at the exterior of the cryostat at
ambient temperature, and removing energy from the remaining part of the ring into dump
resistors.
The SC dipoles and quadrupoles of the UNK are integrated into one sequential power
circuit (see Figure 1). Twenty-four power supplies PSl.. .PS24, each having a voltage of
650 V and nominal current of 5250 A, power the circuit. The power supplies are placed
equidistantly around the ring. Each power supply is switched into the circuit with two
thyristor switches, one shunting the power supply and the other one connecting the dump
resistor.

Figure 1. Block Diagram of Power Supplies and Protection of the Ring Electromagnet for the UNK.
The power circuit between the adjacent power supplies is broken into a few quench
protection units (QPU), each consisting of 12 dipoles and 2 quads (see Figure 2) and
separated from each other by quench stoppers containing a pair of safety leads. Each half
cell is shunted with a quench bypass switch (QBS). The QBS switches and the initiating
electronics are placed in the tunnel alcoves in order to be protected against irradiation. The
alcoves are 200 mm in diameter and 4 m long holes. All SC dipoles have strip heaters
placed on the second layer of the coil.3
CBS
OBS
Figure 2. Block Diagram of QPU.
868
The status of the SC circuit is monitored by the quench protection monitor system
(QPM). Whenever any element goes normal from superconducting, the system performs the
following operations:
- confines the quench cell containing the quenched element;

- sends the command to turn off and shunt all power supplies and turn on the
dump resistors;
- sends the command to the QBSs and the heater fIring units (HFU) of the
cell where the quench was observed. As a result, the current in the magnets
of the quenched cell drops in less than 1 sec and the remainder of the
current of the ring decreases with a time constant determined by the ratio of
the circuit inductance and value of the dump resistors.
The voltage at the dump resistors is nominally 950 V. In this case the decay time
constant is 25 sec and the integral of load into the safety leads and QBSs is
340-1Q6A2 sec. To reduce the heat load, the dump resistors are manufactured from material
possessing a strong resistance-temperature dependence. With the maximum voltage of
950 V, the load integral is reduced to 260-106A2 sec. To reduce short-term overvoltages
from transients affecting the switches, nonlinear resistances limit the voltage to 1.7 kV are
used.
QUENCH PROTECTION MONITOR SYSTEM
The system performs the following functions:
- monitors the status of the power circuit elements including dump switches,
QBSs,HFUs;
- detects a quench in the elements of the SC circuit;
- whenever a quench is detected, it releases the instructions to the executive
blocks; and
- records the signals from the elements of the SC circuit both in the normal
operating mode and during energy removal.
For reliable operation and convenience in service, the system should be capable of
performing self tests, be monitored from the main control room, and contain the minimum
number of active electronic elements placed in the radiation area of the ring tunnel.
The quench protection monitor system for the UNK consists of 24 QPMs placed
equidistantly along the ring in the same buildings with the power supplies of the ring
electromagnets (see Figure 1). Each QPM monitors eight quench protection units (QPU)
and is connected to "its" corresponding power supply. All QPUs are connected by one
network and also to the central control room.
The major component of the system is a magnetic modulator (MM). The MM
possesses a high sensitivity, good zero stability, and high radiation sensitivity.4 Figure 3
shows the scheme of connecting MMs to a QPU. As is seen, there are two groups of MMs.
In the first group, MMs are connected to the diagonal of the bridge circuit (BC), one arm of
which is formed by two half-coils of each dipole or quad. Another two pairs of MMs are
connected to half cells. One MM of this pair is connected to the diagonal of the relevant BC
and another one is functioning as a linear amplifIer and measures the voltage of a half cell.
The signal from the output coil of the MM goes to the relevant module of the QPM, from
which the signal of the second harmonic proportional to the normal phase voltage is
extracted.
869
~:;:::::~.r.::::;::~-/;::=:==::::: to inp20/i
frOM AMp.i
Figure 3. Scheme of Connecting Magnetic Modulators to the SC Components of QPU.
Figure 4 shows the structure of the QPM. Each QPM contains eight amplifier
conditioners, AMP(i), eight modules of cells, eight counters (their number corresponds to
that of QPUs protected by one QPM) and also some modules common for all cells: a timer,
microprocessor, RAM, interface, and input/output module. AMP(i) supplies ac voltage to all
magnetic modulators of the i-th cell (the carrier having a frequency of 1 kHz and amplitude
of 40 V). A cell module is used to convert signals from the output coils of all magnetic
modulators of the i-th cell. Each conversion channel contains a scheme to detect breaks of
wires in the input circuits of MMs, a differential amplifier, demodulation scheme, filter, and
voltage-frequency converter (VFC). The signals from the outputs of all VFCs are converted
into a parallel binary code with the help of 20-channel counter modules, whose
measurement time, 20 ms, is specified by the timer.
All status signals are collected and the logic signals are sent to the cell modules and
executive blocks through the input/output module.
The operation of all blocks of the QPM is monitored by a Kl81OBM86
microprocessor whose external memory is 1 Mbyte (two modules of RAM each having a
capacity of 0.5 Mbytes).
The logic signals from the module go to the executive blocks through the safety logic
module. The module is configured such that if any block of QMP goes faulty, it
automatically releases the instruction to all relevant devices to function in the safety mode.
Testing all channels of the QPM is foreseen. In this case, when the instruction "Test"
is released, currents of 0.5 rnA are applied to all output coils of MMs. The binary code of
the counter outputs corresponds to this signal, is read and analyzed by the microprocessor.
The hardware and software of the QPM is designed to detect all actual failures.
Whenever an AMP(i) fails or there is a break of the cable connecting it and an MM, the cell
module releases the signal "status 1" recorded by the microprocessor through the
input/output module.
870
If in an MM channel connected to the half-cell, a quench signal higher than some
threshold occurs, the hardware releases a signal (DNFi) directly to the inputs of the safety
logic modules. This prevents dangerous situations in the case of software failures or
microprocessor malfunctioning.
The bridge circuits will be balanced at alternating current. In this case the main power
supplies are turned off the current bus, and the string of magnets between two adjacent
buildings is powered from a sinusoidal-shaped power supply having a voltage of -600 V
and frequency 50 Hz. The same power supplies can be used for remote localization of
ground shorts with an accuracy of one magnet.
.-----HOST
f.-_ _ connection line
With another OPS
~p,
20
I~PUTS
2/1 N l£ST ADR OUTPUT DNFl INPUTS l£ST ADR OUTPUT DNFS
20/1 VOLTAGE' STATUS 2/8- VOLTAGES STATUS
20/8
MODULE 1 CELL MODULE 8 CELL
2kHz OUT.CUR. 2kHz
AIlp.s
output current AMp.l output current AMp.8
to coils MM 1 cell. 1, =lkHz to co~s MM 8 cell,

18 =lkHz
Figure 4. Block Diagram of QPM.
The microprocessor with its RAM, the input/output modules, timer and counter
modules are placed in the same VME crate with a bus (it is an analog of BUSl). The
modules of cells, amplifiers-shapers and the modules of the safety logic are also placed in
individual VME crates. All crates are powered from individual redundant power supplies,
energized from a reliable power supply line. All power supplies furnish fault signals to the
input/output module.
The prototypes of all electronic blocks have already been manufactured. A prototype
of the system, which is to be tested first with a four-magnet string of SC dipoles, and then in
the experimental area of the UNK tunnel containing eight QPUs, is at the production stage.
871
REFERENCES
1. V. A. Yarba, "The Status and Development of the UNK," Peoc. of 2nd European Part. Accel. Conf., Nice,
France, 1990, Vol. 1, p. 60.
2. R. Flora, Y. Saravirta et aI., "The Energy SaverlDoubler Quench Protection Monitor System," IEEE
Trans. on Nuclear Sci., 1981, Vol NS·28, N3, P. 3289.
3. A. I. Ageev, N. I. Andreev et at., "Study ofUNK Superconducting Dipole Magnet," Particle Accelerator,
Vol. 27, p. 181, 1990.
4. I. M. Bolotin, A. N. Erochin et at., "The Quench Detector on Magnetic Modulator for the UNK Quench
Protection System," Paper presented at this Symposium.
872
THE MAGNET DATABASE SYSTEM
MJ. Ball, N. Delagi, B. Horton, J.e. Ivey, R. Leedy, X. Li,

B. Marshall, S.L. Robinson, and J.e. Tompkins

Dallas, TX 75237
ABSTRACT
The Test Department of the Magnet Systems Division of the Superconducting Super
Collider Laboratory (SSCL) is developing a central database of SSC magnet information that
will be available to all magnet scientists at the SSCL or elsewhere, via network connections.
The database contains information on the magnets' major components, configuration
information (specifying which individual items were used in each cable, coil, and magnet),
measurements made at major fabrication stages, and the test results on completed magnets.
These data will facilitate the correlation of magnet performance with the properties of its
constituents. Recent efforts have focused on the development of procedures for user-friendly
access to the data, including displays in the format of the production "traveler" data sheets,
standard summary reports, and a graphical interface for ad hoc queries and plots.
FUNCTION
The superconducting magnets for the Superconducting Super Collider

Laboratory (SSCL) accelerator complex will be provided by the Magnet Systems
Division (MSD). Some of these magnets will be built "in house," but most will be built by
industrial vendors under contract to MSD. The Data Management Group of the MSD Test
Department is developing and operating a data management system to integrate all the data that
is relevant for the analysis of test results on these magnets. An important aspect of this system
is provision of user interfaces that facilitate access to this data from local workstations or
network connections.
The data management effort relies primarily on the use of the SYBASE relational
database management system in a UNIX environment, plus tools for database design and data
access, and file management strategies for archiving the large quantities of test data. The focal
* Operated by tbe Universities Research Association, Inc, for the U.S. Department of Energy under Contract number
DE-AC35-89ER40486.
Supercollider 4, Edited by 1. Nonte

Plenum Press, New York:, 1992 873
point of the data management system is a central database that we have named MagCom. This
database will include test results for critical component materials, configuration information
(serial or batch numbers) for critical components in all magnets, data collected at major
production steps, summaries of warm and cold test results on all magnets, and reference
"pointers" to the computer files that have the full data. It will serve as the source for the
standard information needed by the magnet analysis programs, the QA (Quality Assurance)
monitoring system, the Accelerator Division controls databases, and any ad hoc requests for
information.
The size of the data file generated by a single test run is expected to range between 60
and 800 kilobytes, depending on the particular test. The data for the total set of magnets (from
R&D through full production) is estimated to be 100 gigabytes. We plan to store the full test
files outside the database, with only summary data and pointers in the database itself. We plan
a user-friendly interface to allow simple access and manipulation of any magnet data, whether
it is in the database or in the external files archive, so that users will not need to know the
directory structures or physical locations. Consolidating the data in one data management
system allows development of a unified storage approach and integrated applications
interfaces (which can be sophisticated, flexible, and easy to use).
There are multiple advantages to a central database that is available to all. It ensures that
everybody is using the same information for decisions on magnet performance and for input
to their analysis programs. The same data is available for magnet builders, accelerator
designers, and accelerator operators (when that phase is reached). Having the relevant data on
all magnets, from different designs and even different vendors, in the same database allows
easy comparison between magnets. This facilitates study of trends and correlation of
performance to design or manufacturing features. Having all the data managed from a central
system will allow us to maintain good security and backup procedures. We need to protect
against loss due to hardware, software, or human malfunction.
HARDWARE & SOFTWARE
The current workstation setup of the Test Department consists of 2 servers and more
than 35 client workstations on the SSCL ethernet. All of the workstations are SUN-
compatible machines, running SUN OS v. 4.1 (a combination of BSD and System V UNIX)
with a choice of window environment (SunTools, XllR4, and OpenWindows). The servers
are a SUN 4/280 (grumpy.ssc.gov) and a SUN 4/470 (cranky.ssc.gov) on Internet; grumpy
is also connected to DECnet as node name SSCSUN. We plan to upgrade to a faster server in
FY92, to add mass storage capability via optical or magnetic disks, and to convert to use of
SUN Solaris 2.0 (based on System V release 4). A discussion of the philosophy of the
system architecture and security considerations is given in another report.}
The MagCom database is implemented using the Sybase relational database management
system (version 4.01) on grumpy. Sybase modules provide interactive window (dwb) and
batch command file interfaces (isqI) for knowledgeable users. In addition, we have
implemented a simple text menu system that can be used for local or network access to
standard reports. This is described in the next section.
Our software approach is to acquire multi-user floating licenses, keep the software on a
server, and use automount to make it transparently available from any workstation. The
database software, mail, and printing facilities reside on grumpy; users' accounts and general
analysis software reside on cranky. We are using C and C++ (and related utilities) for our
system and data manipulation programming, but most of the users' analysis programs are
written in FORTRAN.
We are using some design tools that were developed by the Data Management Group of
the Information and Computing Sciences Division at Lawrence Berkeley Laboratory (LBL).
874
ERDRAW2 is an X-window based tool for graphical specification of the entity-relationship
model of the database, including key relationships, and SDT3 is the related Schema Definition
Tool. ERDRAW includes a meta-database feature for entering descriptions of the entities and
attributes. SOT generates a SYBASE schema definition, including the triggers for maintaining
referential integrity when adding, deleting, and updating data. A graphical browsing tool
called QUEST4 is expected in the next year; it will use the metadata from ERDRAW to guide
users through the database.
We are also looking into the use of the ISTK5,6 (Integrated Scientific Tool Kit) UNIX
tools being developed for the SSCL Accelerator Controls system. These include a self-
describing data structure (SDS) that provides a general way to store test data files so that they
are easily used by analysis programs on either SUN or VAX systems. We are now using the
graphics interface tools in the ISTK set for some simple graphical displays. However, we
expect to develop applications interfaces to connect to more complex tools such as the
graphical visualization product PV-WAVE or the LBL QUEST browser.
USAGE
There are several alternative ways to connect to grumpy. Network users can connect via
the UNIX rlogin command or the V AXlVMS set host command. Macintosh users at the
SSCL can obtain a set of macros that facilitate the connection ~d file transfer over our local
ethernet. External users can dial in to the SSCL central system (214-708-9500) and then
specify "connect grumpy."
If the user does not have an account on grumpy, he can use our public account, user
name mdbuser. This account is recommended for users who are not familiar with UNIX
because it has very restrictive permissions and will not allow any actions other than reading
from the database and running the standard reports; also it has the appropriate environment
setup for using the database. To leam the password for the mdbuser account or to register for
a regular account, call our system manager (214-708-2968) or send an email message to
[email protected]. The UNIX operating system is case-sensitive, so all commands
must be typed precisely as they are specified.
The database menu system is accessed by typing the command magcom. Some help
messages and the main menu will then appear. There is a tree of menus that allow selection
from our standard reports (see Figure 1). The user is given the option of seeing the screen
display of the selected report, writing a file copy of the report, or both. The file version is
written to a directory called results in the login account (either mdbuser or the user's own
grumpy account); it can then be transferred by ftp to the user's host computer system for
printing. 7
We do not allow the data suppliers to enter data directly into the database or archive.
Data files from the magnet building or testing operations can be transferred to a special
directory on grumpy8 and then processed by our input procedures.
PRESENT STATUS
Past effort9,10 has focused on development of database structures, data import

procedures for R&D data, and standard reports available through the rnagcom menu. The
reports are based on formats of the data tables in the travelers used at the labs and other
formats requested by the analysis staff. Detail reports give results for one item (for instance, a
cable, coil, or magnet), and summary reports give results for a series of items beginning at a
specified identifier. Sample summary and detail reports are shown in Figures 2 and 3.
875
SSC CABLE DATABASE MAIN MENU
1. Data by Cable to
2. Data by Coil 10
3. Data by Magnet 10
4. Strand data
5. Documents
6. Help menu
X. EXIT
1. CABLE MENU
1. Configuration summary. cable-coil-magnet

2. Electrical data summary
3. Electrical data detail
4. Magnetization data summary
5. Magnetization data detail 2. COIL MENU
6_ Mechanical data summary
7 _ Mechanical data detail 1. Configuration summary. coil-cable-magnet
8_ Cable production summary 2. Winding detail
9_ Cable production detail 3. Cable detail
10. Coil detail 4. 17M Coil size data summary
11. Cable strand map detail 5. 17M Coil size data detail
X. Ex~ to main menu 6. Koldweld summary
7. Koldweld detail
3. MAGNET MENU
x. Ex~ to main menu
1. Configuration summary. magnet-co ii-cable

2_ Cable properties summary 4. STRANO MENU
3. Configuration detail
4. Coil and cable detail 1. List id: billet vendor job no. or strand spec.
5. Magnet features menu 2. Strand ic summary
6. LBL dipole menu 3. Strand ic detail
7_ Quench summary 4. Strand production summary
8_ Magnet instrumentation menu 5. Strand production detail
9. Koldweld summary 6. Strand specijication detail
10. Koldweld detail X. Exit to main menu
11. Transfer tunctions detail
12_ Collared coil mole measurements detail
13. Magnet mole measurements detail
14_ Coil RRR summary 6. HELP MENU
x. Ex~ to main menu
1. General info. and how to transfer files

5_ DOCUMENTS MENU X. ExH to main menu
1. List available documents

2. Retrieve a document
3. Store a document
x. Ex~ to main menu
Figure 1. MagCom menu level I.
The data available includes cable production (including strand data), cable electrical and
magnetization tests from Brookhaven National Laboratory (BNL), configuration (cable to
coil to magnet), quench run summaries, and R&D magnet design features (baseline and
variants). Work in progress is focusing on strain gauge data (including cold mass assembly
steps and test runs), quench run data, magnet test instrumentation descriptions, selected coil
production traveler data sheets, and selected cold mass assembly traveler data sheets.
At the present we are concentrating on importing the data for the R&D series of
50-mm-aperture dipole magnets. These include seven magnets constructed at Fermi National
Accelerator Laboratory (FNAL) by a team from General Dynamics and five magnets
constructed at BNL by a team from Westinghouse. These two dipole series have the same
magnetic cross section but differ in collar, yoke, and end design. There are also several
quadrupoles constructed at LBL. The five dipoles and one quadrupole for the Accelerator
Systems String Test (ASST) to be conducted in fall 1992 will come from this set. Each of the
876
KoldWeld Summary by Magnet
Search strings: magnet: DCA209
Cable Pos Welds

Magnet
-------
0CA209
Coil
-------------
DCAI-200S
---------------
SSC-3-o-00044
-------
Inner 6
0CA209 DCAI-2006 SSC-3-o-00044 Inner 6
0CA209 DCAo-200S SSC-4-o-00030 OUter 13
0CA209 DCAo-2006 SSC-4-o-00031 OUter 14
Figure 2. Sample summary report: number of Koldwelds per coil.
Koldwelds Reference Location by Magnet

Search strings: magnetid-DCA209
Magnet Coil RefType Turn MinDist MaxDist Side Coun

-------- ----------- -------- -------- --------
11. 7S
0CA209 DCAI-200S REF-BNL 6 11.417 RIGHT
0CA209 DCAI-200S REF-BNL 10 7.S RIGHT
0CA209 DCAI-200S REF-BNL 12 7.37S RIGHT
0CA209 DCAI-200S REF-BNL 16 .9479999 RIGHT
0CA209 DCAI-2006 REF-BNL 4 3S.33300 LEFT
0CA209 DCAI-2006 REF-BNL 7 34.31200 LEFT
0CA209 DCAI-2006 REF-BNL 10 37.SS2 LEFT
0CA209 DCAo-200S REF-BNL 4 29.74000 RIGHT
0CA209 DCAo-200S REF-BNL 6 18.16700 LEFT
0CA209 DCAo-200S REF-BNL 7 32.7S RIGHT
0CA209 DCAo-200S REF-BNL 10 34.8S399 RIGHT
0CA209 DCAo-200S REF-BNL 20 2.041999 RIGHT
0CA209 DCAO-200S REF-BNI. 22 44.S4200 LEFT
0CA209 DCAo-200S REF-BNL 24 S.7S RIGHT
Figure 3. Sample detail report: Koldweld locations in specified coil.
877
three labs provides a final package of data summarizing the construction and test results for
each magnet; however, the reporting formats are different from the three labs. We are
developing input procedures, both manual and automated, for consolidating all this data into
our database structure.
We are investigating approaches to development of a graphical user interface (GUI) with
pull-down menus that will allow ad hoc selection of data to be displayed in tabular or plot
format. An example based on the X windows and ISTK tools is shown in Figure 4. This
version allows access to only one table (type of data) at a time.
~
~
c:::J
c:::::J
G:J
c::J
c:::J I ,...th; I\mp ~I~
~
..•."'.
I epo.m Balm I I c-_ I 1~~C1~""'~""i'PO=!'IN~:::=;:~
CE:J 1 ......... 0"11 .. 9*7 11 So,"'" II ......... ' I
~ I ,,",00<0lt 0" I I apoaoo __ I
1...'·0fT1
I !?o>e!!r eQ. 1
I .... SQ. I
I "'" eQ. I
()O'· &o··· .... ·--c
<>-----C DCI-045
<>- - _ ..., DCI-047
ze
t
9 .0
7.0
5 .0
.3.0
1.0
-1.0
-.3.0
-5 .0 __ ~~~~~~~~ __ ~~~~~~~a
0 .0 200 .0
distance
Figure 4. Screen display for user interface based on ISTK tools.
Weare also developing interfaces between the database and the standard analysis
programs used by Test Department physicists. The goal is to get the data into the database as
soon as it is released by the magnet builder, and to have a computer procedure that runs the
standard analysis programs and generates the tabular reports and graphs used by the magnet
analysis staff. An example of a graph of the cable short sample quench current versus the
magnet quench current is given in Figure 5. We are developing an interface to the strain
gauge analysis procedures; the user specifies the desired parameters via an interactive
program, which then extracts the data and formats it for PV wave plots.
878
Quench Current Comparison
Green Parametrization •• 4.35K
8000.0
7800.0
g • DCA201
ooc....
i
7600.0 oOCA2Ol
6DCA3tl
VDCA3tZ
U t>DCA313
! ..
ClDCA314
+DCA315
7400.0 • 0CA31'
Ii:
~
7200.0
7000.0
7000.0 7200.0 7400.0 7600.0 7600.0 8000.0
Extrapolated Short Samplo (AI
Figure 5. Comparison of component property with magnet property.
LONG RANGE GOALS
In the longer term, we will incorporate data from the industrial vendors and the SSCL
"MAKE" magnets. This will require development of format specifications and transfer
protocols to retrieve the desired data from their computer systems and enter it into our data
management system. It is desirable to have this as automatic as possible, with programs that
initiate file creation and network transfer during the night. We will develop programs to
validate the files that are received, including checking data ranges and referential integrity,
before entering the data into our database.
It is desirable to have a general graphical user interface (GUI) that will allow users to
make ad hoc selections of data items from mUltiple tables and to specify the tabular or plot
output that is desired. We hope the data selection portion of this GUI will be served by the
QUEST tool being developed by the Data Management Group at LBL Computer Science
Division. QUEST is a general query system that will allow users with no knowledge of
database structure or SQL query language to browse the database and select the items they
want. It works with metadata for entities and relationships that is maintained by the LBL
database definition programs (ERDRA Wand SDT), which we have been using. We will
develop the connections between the data selected by QUEST and the input to the selected
report or graphics processing.
The goal is to make all magnet data easily available for processing through any of the
standard analysis procedures or for input to the user's personal program.
REFERENCES
I. James Ivey, Xiaoyu Li, Joe Garbarini, and Penny Ball, "System Design Overview for the Magnet Test
Database System," IISSC 1992.
2. Emest Szeto and Victor M.Markowitz, "ERDRAW, A Graphical Schema Specification Tool," Lawrence
Berkeley Laboratory, LBL_PUB_3084 (1991).
3. Victor M. Markowitz and Weiping Fang, "SDT, A Database Schema Design and Translation Tool,"
Lawrence Berkeley Laboratory, LBL-27843 (1991).
4. Victor M. Markowitz and Arie Shoshani, "QUEST, A Query Specification and Translation Tool,"
Lawrence Berkeley Laboratory, LBL-31155 (1991).
5. Erika Lutz, "The ISTK Overview Document," Lawrence Berkeley Laboratory (1991).
6. Chris Saltmarsh, Lawrence Berkeley Laboratory; programmers group email to [email protected]
7. Nancy Delagi, "User Information for MagCom Database," SSC Laboratory Magnet Test Department,
MTL-004 (1992).
879
8. Xiaoyu Li, "Transferring Magnet Data to SSC Lab," SSC Laboratory Magnet Test Department, MTL-005
(1992)
9. M.J. Baggett et aI., ''The Magnet Components Database System," Supercollider 2, M. McAshan, ed,
Plenum Press, New York (1990).
10. P. Baggett et aI., "The Magnet Database System," Supercollider 3, M. McAshan, ed., Plenum Press, New
York (1991).
880
mE QUENCH DETECTOR ON MAGNETIC MODULATOR
FOR mE UNK QUENCH PROTECTION SYSTEM
I. M. Bolotin,l A. N. Erokhin,2 A. V. Enbaev,l

V. I. Gridasov,2 M.V. Priyma,2 V. A. Sychev,2 and L. M. Vasiliev2
lRadio Engineering Institute, Moscow

2Institute for High Energy Physics, Protvino
INTRODUCTION
When designing and constructing superconducting high energy accelerators, the

development of the Quench Detection System (QDS) for superconducting (SC) magnets
becomes an important and critical problem. At present there is experience in developing
such systems for the Tevatron (FNAL, USA)l and HERA (Hamburg, Germany)2. The
machines for more than 3 TeV-the UNK (Russia)3 and SSC (USA)4, which are presently
under construction, have very large circumferences, 21 and 87 km, respectively. The
QDS's, similar to those of the Tevatron, require a larger part of the active components of the
electronic equipment be placed in the machine tunnel close to the magnets, and protected
from irradiation or additional surface buildings will have to be constructed. In either case
the cost of such a QDS increases. In addition the former ones reliability decreases and
maintenance becomes more difficult.
For such machines, a QDS in which the quench signal, in each superconducting
magnet (SCM) or groups of SCM's, is extracted with the help of a bridge circuit (BC)
appears to be more suitable. The half coils of SCM's are connected as two arms of the
bridge and the resistors placed in the vacuum vessels of the magnet cryostats are connected
to the other two. The off-balance signal of each BC is enhanced with the help of magnetic
amplifiers.
A magnetic amplifier whose output signal is isolated from the power circuit of SCM's,
possesses an appreciable amplitude and low output impedance. This allows a noise-immune
transmission of signals at distances of greater than 1 km to the technological buildings
where the recording and processing electronics will be placed.
This note describes the experimental prototype of a bridge-type Quench Detector (QD)
based on a magnetic amplifier Magnetic Modulator (MM) type, allowing one not only to
detect a quench in a SCM, but also making feasible a wider system capability, namely:
- to record the signals from all SC elements during a quench for further analysis of its
causes;
- to check the presence of short circuits of the ring electromagnet bus relative to the
cryostats and to localize their position.

The QDS described, is a part of the protection system for the SC electromagnet ring of
theUNK.5
DESCRIPTION OF QUENCH DETECTOR
When developing the QD, the following imposed requirements were considered:
- radiation resistivity of all components of the QD placed in the machine tunnel;

- the quench detection threshold should be less than 100 mV;
- the response should be less than 10 ms;
- the transfer characteristic should be linear;
- the energizing and output windings should be high-voltage insulated from the input
electrically connected to the SCM (the test voltage is 5 kV);
- the possibility of precision balance of the bridge circuit;
- the possibility of a remote testing the QD;
- the possibility to detect open circuits in the QD.
Figure 1 presents a simplified scheme of the QD comprising a bridge circuit formed by

half coils of the SCM's, 112 L M , and resistors R1-R4, a magnetic modulator (transformers
T1 and T2) and electronic module. Resistors R3 and R4 make the precision balance of the
BC. Fuses F1-F3 are used to protect the input loops of the QD during emergency
operations.
Each transformer of the magnetic modulator contains three identical coils, each having
100 turns wound onto a toroidal core made from an amorphous metal having the hysteresis
loop as shown in Figure 2. The components of the BC and MM are enclosed into a steel box
which provides a magnetic shield. The box is fixed directly at the SCM cryostat close to the
potential leads of the magnet.
The energizing windings of the MM, (W3, W6) and the output, (W2 and W5), are
connected to the electronic module by two 1-km twisted pairs. The MM is powered from
the amplifier-shaper (Ampl) by the voltage of the carrier, shape of an amplitude of 80 V and
frequency of 1 kHz. The R5 = 1 kQ resistor restricts the current running through the
windings of the MM when its cores are saturated. DA is the input differential amplifier
whose load is a demodulator (DEMOD). It extracts even harmonics from the 1 kHz voltage.
The signal coming from the DEMOD output is filtered with an active filter F and amplified
up to the output voltage required. The operation of the electronic module is synchronized
against the reference frequency standard. An open circuit in the input windings of the MM
(W 1, W 4) leads to an abrupt variation in the value and nature of the output impedance of the
MM. This is taken advantage of when checking to ensure that these windings and fuses F1-
F3 are intact.
The testing of the QD is based on the fact that the input and output windings of the
MM are identical in terms of the effect on the level of even harmonics. It is tested by
applying with switches Sl and S2 through resistors R a current of about 0.5 rnA of either
direction to the output winding of the MM. This leads to the known voltage appearing at the
filter output.
CHARACTERISTICS OF QD
Figure 3 shows the transfer characteristic of the QD. Its nonlinearity does not exceed
2% in the output range voltage of ±1O V and input current of 6 rnA. The ambiguity in the
correspondence of lin and Uout caused by decaying subcircuits at an input current larger
than 20 rnA, present on the transfer characteristic, is eliminated by using in the demodulator
the data on the rectified current of the circuit of the energizing winding (see Figure 4).
882
Tunnel
Magnetic shield " I Electronic module
r--------l~ r---------l
I F1 R1 T1 I~ ... >C>QI Output I
W2 Voltage
112Lm I 2000 I··· I I
I R3 2kHz J1Il I
3300 "Test·/
I II '"'" I
112Lm I I -1km , o-&.+1SV I
F2 R2 I / I RS2
II 2000 I I
... >C>QI
Fa I~... I 1kHz JUl I
II •
T2 W6
~nkH.,' n
I II 80V T I U U L
I
---"!-----II L _________ J
Figure 1. Block diagram of QD.
CD
CD
W
8(T) ...
0.8
0.6
0.4
0.2
....
-1 0.2 H(a/m)
-0.2
-0.4
- 0.6
-u.8
Figure 2. Hysteresis loop for amorphous steel.
884
U out(V)
12
10
8
6
4
-32 -30 -28 -26-24 -22 -20-6 -4

4 620 22 24 26 28 30 32mA
lin (mA)
-6
-8
-10
-12
Figure 3. Transfer characteristic for QD.
40
.. Eamp (V)
o --....
1.0 2.0 t(ms)
--40 f-
lamp (rnA) lin = 0

--40
10 t-------6
-1g~--------t=====~~~1~.0~-----t======~~~2n.O~t:(ms~)
--40
amp (rnA) I·In> 30ma
40
o
1.0 2.0
~
...
t(ms)
--40 f-
Figure 4. Oscillograms of voltage and current of the magnetic amplifier for different input currents.
When currents of -a hundred rnA affect the MM input, which may be the case during
an asymmetric quench of the magnet coil, the effect of magnetic memory, meaning that the
cores demagnetize within a prolonged period of time (up to 15 minutes), is observed. The
input/output time delay of the whole MM is 2.5 ms. The duration of magnetic memory,
sensitivity, linearity and time response are determined not only by the correct choice of
ampere-turns, dimensions of cores, frequency and value of the energizing voltage but also
by the value of R5 restricting the current in the energizing winding.
885
In our case, the optimal value of R5 is I kn. At larger values of the power current the
sensitivity goes down, while at smaller ones the linearity deteriorates and the magnetic
memory effect in enhanced.
The QD with a magnetic modulator was tested with 6-m dipoles for the UNK both at
the ambient temperature and 4.2 K. A major result was that the ac balance of the bridge,
performed at the ambient temperature, remains even when the magnet is cooled down to
4.2 K. Moreover, the MC, balanced optimally from a sinusoid voltage power supply, does
not require any additional correction when a SC magnet is powered with 1500 A triangular
cycles at a ramp rate of 400 AlS. Thus, the study showed that the QD BC can be balanced,
prior to putting the magnet into the tunnel from any sinusoid voltage power supply, e.g.,
from the mains.
To illustrate the operation of the QD, Figure 5 shows the signal at its output during
emergency current removal from the SC dipole. When energy is removed, the power supply
is turned off and an additional resistance of 11.25 mn is switched into the circuit. In this
case the maximum extraction rate is about 600 AlS. The negative pulse lasting about 12 ms
and having an amplitude of 700 mV in the beginning of emergency removal is caused by
actuation of thyristor switches turning the power supply off and the extraction resistor on.
The amplitude of the positive signal caused by an insufficient compensation of the inductive
component of the voltage on the SC dipole does not exceed 200 mV and corresponds to the
12 mV disbalance of the bridge circuit.
The test results allow one to believe that a high-efficiency, reliable, sensitive and
noise-immune QDS can be developed on the basis of the QD, incorporating magnetic
amplifiers.
Quench 13 04-APR-91 CPSE04 10:33

343
188
33
7== 122
DUMP
~
Imax = 2.4kA
Rd=11.25mil
277 i - 600 AlS
432
588 0
296
Sec *-3
Figure 5.
886
REFERENCES
1. R. Flora, Y. Saravirta et al., "The Energy Saver/Doubler Quench Protection Monitor System," IEEE
Trans. on Nucl. Sci., v. N5-28, N3, 3289 (1981).
2. K. H. Mess, "Quench Protection at HERA," Proceedings o/the 1987 IEEE Particle Accelerator
Conference, March 1987, p. 1474.
3. V. A. Yarba, "The Status and Development of the UNK," Proc. o/the 2nd Part. Accel. COnj., Nice,
France, v. 1,60 (1990).
4. "Site Conceptual Design of the SSC," SSCL-SR-I056, 1990.
5. O. A. Manasiev, A.P. Dubasov et al., "The Protection System for the Superconducting Ring
Electromagnet of the UNK," report at this symposium.
887
REQUIREMENTS CONTROL SYSTEM APPLIED TO
COLLIDER DIPOLE MAGNET DEVELOPMENT
D.A. Brouillette
General Dynamics
Systems Engineering (MZ: CI-8771)
P.O. Box 85990
INTRODUCTION
A number of Myths abound in the area of requirements development. The

Requirements Control System (RCS) developed and used by General Dynamics on the
CDM contract aims to dispel these myths and ensure comprehensive requirements
flowdown to elements of the Collider Dipole Magnet (CDM) and the supporting Analysis,
Demonstration, Quality Assurance and Master Test Plans. RCS also provides requirements
traceability information by providing the source for each requirement, whether it be from a
higher level specification, or from a supporting analysis or trade study. The computer is
put to work doing what it does best, keeping track of details and checking for consistency
and completeness.
The myths that the use of RCS attempts to drive away are:
I have to design it first so I can determine what the requirements are.
I don"t have time to write down the requirements. I'm too busy designing.
I know what the requirements are. I don"t see why I should write them down.
If I have to wait for requirements, I'll never get done.
The requirements are changing, so why document them until they are finalized?
For the CDM program, the source requirements documents are the SSCL 15 Metre
CDM Magnet System Specification and the SSCL CDM Interface Control Document.
RCS has been used to generate:
CDM Prime Item Development Specification
CDM Prime Item Product Fabrication Specification
Prime Item Product Function Specifications
[Cold Mass, Cryostat, and Interconnect Region]
Critical Item Specifications
[Beam Tube Assembly, and Superconducting Inner and Outer Cables]
Requirements have been further developed for major subassemblies such as the Cold
Mass Support Post and other major items of tooling, test equipment, and transportation
equipment.

Requirements validation, flowdown and traceability place heavy demands on the
requirements database tools used by the systems engineer. For CDM we have prepared 73
requirements sets and 25 specifications using RCS.
RCS facilitates real time requirements development and strict requirements
configuration control. Requirements analysis and review have been conducted in real-time
using electronic optical projection in a conference environment. Specifications have been
generated directly from the database.
The use of RCS has enabled us to accomplish more work in a shorter time than
conventional methods would allow. All data resides in one database giving us the
capability to audit flowdown/traceability without confusion, track requirements analysis
history including rational for changes, and facilitate requirements configuration control by
eliminating duplicate data sources.
REQUIREMENTS CONTROL
The Computer Integrated Systems Engineering (CISE) project at General Dynamics

has developed the Phase II Requirements Control System (RCS), an automation tool which
provides for the collection, analysis, review and documentation of the requirements
analysis information associated with a program. It was developed with the goals of
providing on-line accessibility to this information, providing a simple, intuitive user
interface, allowing multiple simultaneous views of information, and enhancing the
requirements analysis capabilities of GDSS. The system was designed to facilitate more
effective communication and cooperation between engineering groups. Requirements can
be collected from many sources, integrated and refined in a central location, and made
available for review.
The tool has a user-friendly graphical user interface accessible from Macintosh and
IBM PC compatible computers. Phase II RCS can be run as a stand-alone system on a
local workstation or accessed by multiple users on a network.
Phase II RCS System Capabilities:
Define Architectures
Define Operational Requirements and Design Constraints
Link Architectures, Functions, and Requirements
Define Requirements Traceability / Flow Down
Define Requirements Verification Methods & Descriptions
Define Applicable Documents per Architecture
Baseline & Archive Architecture Configurations
Import MIL-STD-490A Specification Sections 2 & 3
Generate MIL-STD-490A Specification Sections 2, 3 & 4
Output Multiple Reports
Provide Multiple Levels of Access to System Information
Minimum System Requirements:
IBM PC Compatible with Microsoft Windows 3.0 or

Macintosh II with System 6.04
3MB Disk Storage (Application)
10MB Disk Storage (Data), Size increase for larger programs
Blyth Software's Omnis 5 Run-time ("N" user version)
File Server (if multiple simultaneous user access required)
890
Res METHODOLOGY
Requirements in RCS must be associated with an architecture. The starting point for
the CDM program was the definition the requirements tree. Each node on the requirements
tree represents a collection of requirements for a specific CDM program element or
architecture. Figure 1 illustrates the requirements tree for the CDM program with each
architecture highlighted for which a specification has been developed.
:15M~tr~66Mn
no
.,.,.,.
Systyem Spec
M80-000001 .....
t } ~g~
,': MRn_nnnM~
ICD
PIDS
CIDS
Interface Control Document
Prime Item Development Specification
Configuration Item Development S pacification
PFS Product Fabrication Specification
I SSCL SSCL
I I
COM COM COM
PIDS Tooling CIDS Test Eqt CIDS
M8A-l00002 M8A-l001XX M8A-l002XX
I GDSS GDSS GDSS

I I I I
Cold Mass Cryostat Interconnect Region Interface Control
Prime Item PFS Prime Item PFS Prime Item PFS Specification
M8A-l00003 M8A-l00004 M8A-l00005 M8A-l00009
I I I I
I
GDSS GDSS GDSS GDSS
COM COM Elements COM

Critical Item PFS Requirements Detail
M8A-l0000X Lists [48J Drawings
GDSS GDSS GDSS
Figure 1. CDM Requirements Tree
Top level requirements from SSCL CDM documents were initially loaded into the
Phase II RCS database to begin the process. Next, requirements were developed at various
levels of the CDM architecture. Verification and traceability information were then
defined and added to the database for each requirement in each architectural element. This
was done repetitively until the fully developed requirements database withstood audits for
traceability and verification accuracy and completeness.
All of these steps can be iterated multiple times to define more detailed information
and refine information, as necessary. Reports are available to print the traceability of
requirements between a pair of architectures or the full traceability of one requirement
throughout the architecture hierarchy. Numerous reports can be printed to review and
analyze information which has been entered into the database. Sections of MIL-STD-490A
formatted specifications can be generated from the database [Section 2 Applicable
Documents, Section 3 Requirements, and Section 4 Verification Matrix] as well as
requirements lists for each architectural element.
DATABASE ENVIRONMENT
Window Processing. Omnis 5 windows are used to display data entry screens,
display lists of data, and prompt users for information (refer to Figure 2). Multiple
windows may be open at the same time. You may switch between the windows by clicking
anywhere on the window you wish to bring to the foreground. [Note, in certain cases,
windows do not allow other windows to be brought to the foreground until a specific
sequential operation which cannot be interrupted is completed.]
891
Data Processing. Omnis 5 works on the concept of the Current Record Buffer. You
work with one record at a time. You may only edit or delete one record at a time.
Data Manipulation. Omnis 5 applications are designed such that the information
displayed on the screen is "read-only" by default. If you attempt to move to a field and
change it, the system will ignore your actions. To edit or insert information, you must first
click on an appropriate button ("Edit" or "Insert") at the bottom of your screen. Then, you
may modify or create the data on the screen. When you are done, click on the "Ok" button
to save your changes. If you make a mistake, you may click on the "Cancel" button to
abort the attempt.
[jjJ
~~"m$£
ReYie'al
Traceability
Figure 2. Sample Graphical Menu Screen (Main Menu)
SCREENS
Screens are used for data entry or review purposes and have a common set of
command buttons (refer to Figure 3). The six buttons on the lower left part of the screen
(Find, Next Previous, Insert, Edit, & Delete) are standard on many of the data entry
screens. The 'first three are used to find a specific record and the next three are used to
change the record.
Architecture Definition Screen. Allows the user to browse or update the architecture
information. Phase II RCS will automatically retrieve the top level architecture defined for
the program (if one has been entered into the RCS database). The user may insert new
architectures and edit existing ones.
892
Requirement Definition Screen. Allows the user to browse or update the
requirement information. It is intended to be used by program personnel to document their
requirements. The pool of requirements collected with this screen will be processed later
with the Convert Requirements screen and the Edit Specification screen available to
subsequent requirements reviewers and/or approvers.
Edit Specification Definition Screen. This is one of the most useful screens within
Phase II RCS. It is used to create, update, and delete requirements associated with one or
two specified architectures (refer to Figure 4). It can be used to compare the requirements
of two related architectures to ensure consistency between them.
Figure 3. Sample Data Entry Screen
Convert Requirements Screen. This screen is used to convert raw requirements

input from design personnel or other personnel not responsible for the set of requirements
associated with an architecture. It allows the principle engineer or a systems engineering
group to validate requirements from outside groups prior to incorporating them as
requirement paragraphs in the RCS database.
Requirements Traceability Definition Screen. This is also one of the most

important Phase II RCS screens. It is used to create / remove traceability links between the
requirements (usually design constraints) of a pair of architectures. See Figure 5. It may
be used to identify the flow down of requirements from parent architectures (or
specifications) to child architectures (or specifications). It may also be used to identify
horizontal traceability between two sibling architectures. You may even link two
requirements associated with the same architecture.
893
Figure 4. Edit Specification Screen
Figure S. Requirements Traceability Definition Screen
Requirements Verification Definition Screen. This screen is used to create the

Verification Matrix associated with an architecture. The requirements will be listed down
the left hand side of the screen and the appropriate verification methods will be displayed
in the matrix. To change the verification methodes) to be used against a particular
requirement, simply double-click on the requirement and then click on the appropriate
check boxes in the lower left hand corner of the screen.
Reference Document Definition Screen. This screen allows the user to associate
applicable documents with an architecture. The list of applicable documents may later be
generated for Section 2 of a MIL-STD-490A specification.
894
REPORTS
Numerous reports are available from Phase II RCS (Printer, Screen, File, etc). Refer
to Figure 6 for a display of the report selection screen.
Figure 6. Report Selection Screen
SPECIFICATION GENERATION
The Specification Generator creates intermediate ASCII files in Rich Text Format
(RTF) from the RCS database; one file for each applicable section of a specification. The
text files created by the generator contain page layout information such as the footers, page
numbering, and headers (specification title) in the RTF format and are not to be modified
directly. Instead, the database is to be loaded or changed and these files regenerated
automatically from that new information. This will ensure that the database holds the
master version of the data and it is up-to-date with the current specification changes.
Configuration control and verification efforts associated with the specification are greatly
reduced by having to update only one source of the data.
The following section(s) can be created under one generation:
SECfION 2 Applicable Documents
SECfION 3 Allocated Requirements
SECfION 4 Verification Descriptions for Section 3 Paragraphs
SECfION 4 Verification Matrix
Specification Section 2- Applicable Documents
Section 2 of the specification (Applicable Documents) is divided into two types of

document listings (government, and non-government). Both types will list documents
which are referenced by the either the architecture or its allocated requirements with a
usage code "C" (for Compliance). The following information is listed for each document:
(T) For tailoring has been identified
Document ID
Document Version (date)
895
Document Revision (in parentheses)
Title
Specification Section 3 - Requirements
Section 3 of the Specification Generator lists all allocated requirements for the selected
architecture. If the User information calls for only approved items, the latest approved
requirements will be retrieved; otherwise, the latest requirements regardless of approval
status will be used. Inactive requirements are not included.
Specification Section 4 - Verification Matrix
The Verification Matrix is only part of Section 4 of the specification generator

(Quality Control). The matrix lists all of the verification methods for the allocated
requirements of the selected architecture. The matrix places an X under each of the
applicable columns for test and assessment. The Verification Matrix in Section 4 contains
the following for each allocated requirement:
Paragraph Number
Test Methods (Functional, Environmental, Proof)
Assessment (Similarity, Analysis, Instance, Demonstration, Validation)
Remark.
Specification Loader
RCS supports the loading of documents into the database. The documents may be in
one of two formats: documents using style headings for each paragraph or documents in
MIL-STD-490A format where the paragraph titles are bold.
TRADEMARKS
Omnis 5 is a trademark of Blyth Software.

Microsoft Windows is a registered trademark of the Microsoft Corporation.
Macintosh is a registered trademark of Apple Computer, Inc.
CONTRACT
The use of RCS to define CDM program architectural element requirements and
generate specifications as described herein is being accomplished under contract SSC-91-
B-01701 to the Universities Research Association in support of the Superconducting Super
Collider project for the U.S. Department of Energy.
896
A DYNAMIC MODEL FOR BEAM TUBE VACUUM
EFFECTS ON THE SSC CRYOGENIC SYSTEM
R. H. Carcagno, W. E. Schiesser, H.-J. Shih, X. Xu, and A. Yiicel

Dallas, TX 75237
Abstract: The dynamic interaction between the beam-gas scattering induced energy
deposition and the SSC cryogenic system is studied by integrating a cryogenic system
dynamic simulator with an adsorption model and a beam-gas scattering and energy dep-
osition model. Simulation results are obtained for a I km long SSC arc section where the
beam tube pressure in one of the dipoles is increased up to 150 times its nominal operat-
ing pressure. The beam-gas scattering induced heat loads arising from such high pres-
sure perturbations can be very high locally but do not overload the cryogenic system.
They are more likely to induce a magnet quench as they result in coil temperatures that
largely excecd the desired operating limit. Simulations are also carried out for the case
where a large extemal heat load of 1 W/m is imposed on the coldmass of a half-cell in
the section. Results show that the coldmass temperatures rise significantly with the
added heat load but the energy deposition in the coldmass due to beam-gas scattering
remains low despite the increase in the beam tube temperature.
1.0 INTRODUCTION
It has been noted that a local disturbance in the SSC beam tube pressure would initiate a pro-
cess that may eventually make it necessary to shut down the collider due to a local overload of tlle
cryogenic system before any effect is seen on the beam lifetime. In the SSC baseline design, the
beam tube vacuum is maintained at a nominal pressure level of 1.3xlO-10 torr and the beam tube at
4.17 K. During the operation of the collider, synchrotron light photons continuously strike the beam
tube sUlface and release H2 molecules and other gases. The gas molecules are pumped or adsorbed
by the cold beam tube surface, but are re-.lesorbed easily by the flux of photons. The resulting den-
sity of gas molecules in the beam tube scatters the protons in the beam, which not only rcduces the
beam lifetime, but also produces particle showers which deposit energy in the coldmass [I J.
In the event that the beam-tube vacuum is degraded at a small section such as one dipole, the
increased density of H2 molecules will yield increased energy deposition in the coldmass, causing
the coldmass temperature to rise. As the beam tube temperature increases, the gas equilibrium pres-
sure and consequently the gas density increases, which further raises the beam-gas heat load on the
* Operated by the Universities Research Association, Inc. for the U.S. Department of Energy
under Contract No. DE-AC35-89ER40486.

Table 1. Dipole Radial Configuration
Yoke
o o
Radial range
Material
(em)
o. - 1.64 vacuum
1.64 - 1.85 bore tube (Fe)
1.85 - 2.50 liquid helium

2.50 - 3.70 inner coil (NbTi & Cu)
3.70 - 5.00
5.00 - 6.80
outer coil (NbTi & Cu)
collar (Fe)
~ Main Helium
Channels
P
6.80 - 16.4 yoke (Fe)
Figure 1. Schematic of the
coldmass assembly
cryogenic system. If the cryogenic system can not maintain the beam tube temperature below a cer-
tain level, the above mentioned phenomenon could continue and cause the collider to shut down.
The purpose of this work is to analyze this phenomenon and evaluate its potential effect on the
cryogenic system operation. 'Three models have been integrated: the beam-gas interaction model
provides the energy deposition for each proton-proton collision resulting from beam-gas interactions
each second at every meter of the magnet; the adsorption model provides the density of the hydrogen
molecules available to collide with the proton beam; and the cryogenic system dynamic simulator
simulates the cryogenic system response to the total energy deposition for a given section of the
magnet string, obtained from the other two models.
An increased heat load on the coldmass due to a source other than the beam tube vacuum deg-
radation can also raise the temperature of the cold mass and the beam tube, and trigger the above
described phenomenon. Therefore, in addition to the problem of a local perturbation in the beam
tube pressure, we also analyzed the problem of a perturbation in the external heat input to the cold-
mass.
2.0 BEAM-GAS INTERACTION INDUCED ENERGY DEPOSITION MODEL
This section describes the simulation of energy deposition in a regular half cell due to beam-
gas collisions. Protons in the circulating beams interact with the gas in the beam pipe. Most of the
particles produced in beam-gas collisions enter and interact with matter in the magnets, creating cas-
cades of particles and depositing energy. Since the predominant gas is hydrogen in the cold section,
beam-gas collisions are in effect proton-proton collisions.
Particle production in proton-proton collisions is simulated using the Monte-Carlo program
ISAJET [2]. Five thousand events are generated at the center-of-mass energy of 194 GeV. The colli-
sion points are uniformly distributed along a unit length of one meter. Particle cascades and the asso-
ciated energy deposition in a half cell are simulated using the Monte-Carlo program MARS 10 [3]. In
the simulation, a half cell consists of 5 dipoles each of length 15.815 meters and a quadrupole of
length 5.85 meters. The spool pieces are ignored. The radial configuration of the dipoles are shown
in Table 1 and in Fig. 1. Dipoles are curved with a bending angle of 0.086 degrees. The magnetic
fields, with strengths of 6.6 Tesla in dipoles and 2.056 Teslalmeter in the quadrupoles, are present
only in the vacuum. The superconductors are composed of 62% Cu, 19% TI and 19% Nb.
Shown in Fig. 2(a) are the energy depositions per meter in a half cell and in different layers of
the magnet for one beam-gas collision (one event). The energy deposition in the helium in the annu-
lar channel is negligible. The first peak, about 8 meters downstream, is caused by the charged parti-
cles from beam-gas collisions. The second peak is caused by the neutrals and should be near the
point where the neutrals hit the beam tube which is about 19 meters downstream according to the
. geometry, j2;;. Here x is the inner radius of the dipole beam tube and r is the dipole radius of cur-
vature.
898
2.0 10
I I I I
0: Tolal 0: Tolal
0: Beam Pipe B f- 0: Beam Pipe -
1.5 0: Inner Coil 0: Inner Coil
":
~
~ Outer Coil ):(: Outer Coil

"S
"-
+; Collar
x: Yoke ...>" 6 f-
+; Collar
X; Yoke -
>
..." 1.0
d
:3·iii
d0
0
4 -
:::l
·iii C
D.
" ~
0
D. ~
c" 0.5 ...
OIl
>-
...
OIl ..,"c 2 -
..,"c I
o
40 o 20 40 60 BO
Longitudinal Position. Meter Longitudinal Position, Meter
(a) (b)
Figure 2. Energy depOSitions from beam-gas scattering in

(a) one event per meter, (b) one event per dipole.
Fig. 2(b) shows the energy depositions per dipole in different layers of the magnet for one
beam-gas collision. obtained by integrating Fig. 2(a) over the length of one dipole (15.8 m) and
dividing by the length. Since we consider one dipole as the basic unit, the values in this figure are
used in our model.
3.0 ADSORPTION MODEL

From the beam-gas interaction model, we are able to obtain the amount of energy released
from a single event in one dipole. In this section, we will develop a model that calculates the number
density of H2 gas molecules in the beam pipe, p, as a function of the beam pipe pressure and temper-
ature. The number density will then be used to determine the number of collisions and consequently,
the total energy deposition from beam-gas scattering.
The adsorption model is based on the BET equation for the physical adsorption process [4].
Consider a beam tube of diameter D and length L. For a given temperature T and pressure P, the total
number of molecules in the tube. Nt. can be written as
N, = Na +Nv' (1)
where N. is the number of adsorbed molecules, and Ny the number of gas molecules.
Na is given by the BET equation:
zx (2)
Na = TtDLnm (I-x) [I + (z-l)x]'
where x = P/Psat , z = eB/T, nm is the number of molecules in a monolayer per unit area, and e is
related to the energy of adsorption. The gas adsorption parameters nm and e are determined from
experiments. For hydrogen gas with stainless steel, we have estimated the following values based on
the experiments of Benvenuti et al. [5]:
nm =0.66 x 10 16 cm- 2, 9 = 41 K.
The saturation pressure is given by:
/n(P = 4.7222-90.S93/T+2.1269/n(T),
SQ ' ) (3)
and Ny is given by the ideal gas equation:
(4)
where k is the Boltzmann constant
During a transient, as the beam-tube and the gas temperatures increase due to increased heat
load, the relative compositions of adsorbed vs. gas molecules will be affected while the total number
of molecules, Nt, remains unchanged. Under the quasi-equilibrium process assumption, the new
equilibrium pressure P' corresponding to temperature T' at a given instant is:
899
P' = x'p.a,(T') , (5)
where x' is obtained by solving the equation: N, = N,' = No (x', 1") + Ny (x', 1").
The number density of Hz gas molecules (moleculeslm3), p, is subsequently calculated from
Ny P
(6)
P = (nLD2/4) = kT'
The number of interactions per second per meter, Nint, can be computed from
N in , = 2paNbea J,o,' (7)
where (J is the proton-proton scattering cross-section (5xlO-3o cmz ), Nbeam is the number of protons
per beam (1.3xlO I4), and frot is the rotation frequency of the beam (3.44 kHz).
The heat deposition due to beam-gas interactions is proportional to the number of beam-gas
interactions and is given by:
(8)
where Elevent is the beam-gas scattering energy distribution from a single event, given by the beam-
gas model.
4.0 SSCDYSIM, v2.0
4.1 The SSC Cryogenic System

The sse requires an extensive cryogenic system that not only accommodates the operating
conditions of the superconducting magnets, but also the transient conditions such as cooldown,
beam acceleration, magnet quench and magnet maintenance. The sse cryogenic system contains
twelve equal sized helium refrigeration plants, ten for the collider and two for the high energy
booster. Each plant on the collider provides cooling for a 8.6-lan sector of both rings. A more
detailed description of the cryogenic sector structure can be found in reference 6.
Fig. 3 illustrates the flow diagram of the cryogenic system for a portion of the collider rings.
Single-phase helium at 4.ISK and 4 atm from a refrigeration plant is pumped into the magnet
strings. It flows through the magnets in series and is recooled at intervals of one cell to maintain the
magnets below 4.35 K. Encompassing these pipes are the 20K and 84K shields with multilayer insu-
lation, and all these are contained in a vacuum vessel to form the cryostat
Ring number 1 cryostat

84 K Shield sub-cooled Iiguid nitrogen
•
Dipole Q S Dipole Q S
...
Dipole Q S Dipole Q S
20 K Shield
DipoleQS DipoleQS
...
I
Ring number 2 cryostat
I
r-
Magnet single
phase helium ~
tJA------
Dipole Q S Dipole Q S Dipole Q S Dipole Q S Dipole Q S Dipole Q S
• 20 K Shield
Figure 3. The flow diagram of the sse cryogenic system
900
4.2 The Flexible Flowsheet Dynamic Process Simulator, SSCDYSIM, v2.0
To study the operation of the SSC cryogenic system under various transient conditions, the
SSC Central Design group and Air Products and Chemicals, Inc. have developed a dynamic process
simulator, SSCDYSIM, v1.0 [7]. This dynamic process simulator is based on a nonlinear, homoge-
neous, lumped-parameter physical model. It is written in FORfRAN, and has a modular structure. It
uses the LSODES [8] integration package, and incorporates Air Products' implementation of the
NBS helium thermodynamic model and two additional simplified models for the helium properties.
Since SSCDYSIM, v1.0 is a fixed-flow sheet simulator, it is difficult to make modifications
when a new problem is to be studied and a new flowsheet is to be built Thus, we have upgraded ver-
sion 1.0 to make version 2.0 a flexible-flow sheet simulator.
Unlike version 1.0, a flowsheet for SSCDYSIM v2.0 can be built with a small set of object
types and connection types, and it is easy to add or remove objects from the flowsheet without
changing the source code. The structure of SSCDYSIM v2.0 also allows for easy addition of new
object types and/or connection types. The majority of the code in SSCDYSIM v2.0 was written in C
programming language, except the LSODES integration package and the Air Products' helium prop-
erty package.
SSCDYSIM v2.0 uses the C facility "structure" to represent each type of object, such as vol-
ume, recooler, pump, valve, etc. This is similar to the convention of lumped-nodes inSSCDYSIM
vl.O. The flows between these objects are called "streams". Any physically feasible connection
between two objects can easily be accomplished through these streams. Adding or deleting the
objects therefore is very convenient
In addition to simulating the process flow, SSCDYSIM v2.0 also has the capability of solving
heat transfer problems such as conduction, convection, and radiation. Hence, the heat transfer prob-
lems in the magnets can be coupled with the process flow of the cryogenic system.
5.0 MODELING PROCEDURE
To study the effect of beam-gas scattering energy deposition on the SSC cryogenic system, we
have integrated the beam-gas interaction model and the adsorption model into the cryogenic process
simulator. Based on the two-dimensional energy distribution from the beam-gas model and the num-
ber density of the H2 gas molecules, we can calculate from Eq. (8) the energy deposition for all the
beam-gas collisions. Taking this energy deposition as the dynamic heat load to a dipole in our
SSCDYSIM f1owsheet, we can solve various process flow problems and calculate the response of
the cryogenic system to that heat load, such as the variations in recooler duty, the helium pressure
and temperature in the magnets, and in the liquid return and vapor return lines.
The study was conducted on one section of one string, about 1 km long, as shown in Fig. 4.
The section has 6 cells, and each cell has IO dipoles. The helium in the main channels of each dipole
is lumped into one node and is represented by a dot in Fig. 4. The coldmass configuration is shown
in Fig. 1. As shown in Fig. 1, we used 6 nodes in the coldmass of each dipole to account for heat
conduction; the coil, collar and yoke each has 2 nodes. The energy deposition in each layer (Fig. 2)
is assigned to the corresponding node.
4K refrigeration
4Kfeed
Figure 4. The flow sheet for the beam-tube vacuum adsorption study
901
The current flowsheet does not include any nodes for the beam tube and for the helium in the
annular channel. Thus, the effects of convection in the annular channel are not accounted for. There-
fore, the synchrotron radiation load and the beam-gas load to the beam tube are allocated directly to
the inner coil node. In order to obtain the beam tube inner surface temperature to be used in the
adsorption model, we have developed a separate three-dimensional model to solve for the detailed
heat transfer problem in the coldmass, including the axial flow in the annular helium channel along a
15.8 m dipole. Using this model, the beam-tube temperature along a dipole can be calculated given
the heat load distribution and the temperature of the helium in the main cooling channel. To this end,
we have used an iterative procedure outlined below:
• For a given heat load, apply the simulator to solve for the helium temperature in the main
channel.
• Use the thermal resistances obtained from the detailed heat transfer model to compute the
beam-tube surface temperature. To be conservative, we have used the temperature distribu-
tions at the end of the dipole, where the beam-tube temperature is higher and hence renders
higher beam-gas load.
• Calculate the beam-gas heat load from this beam-tube temperature and supply the load into
the simulator which in tum gives new helium temperature in the main channel.
As a first step in the simulation, we obtained the steady state condition under the nominal
static heat load and dynamic heat load (including synchrotron radiation load and beam-gas scattering
at p= 1.3x 10- 10 torr). We then impose a perturbation in the beam-pipe pressure or a perturbation in
the external heat input. Tho scenarios were considered: a) a local beam-tube pressure bump in a
dipole, and b) a local heat load disturbance in a half cell.
6.0 DISCUSSION OF RESULTS
6.1 Beam Thbe Vacuum Degradation in a Dipole
As shown in Fig. 4, the degradation in beam-tube vacuum is assumed to occur at the ninth
dipole of the second cell where the overall temperature here is the second highest in the cell. Most of
the scattering energy is deposited in the next dipole, which constitutes the worst case scenario. After
running in steady state for 10 minutes, we perturbed tlle beam-pipe pressure in one dipole and
obtained the results for a range of test conditions.
Figs. 5(a)-(b) show the temperature profiles in the helium channel and in the inner coil along
the 1 kIn long string section at different times, with the pressure bump being 150 times the nominal
value of -3x1O- 1O torr. It can be seen from Figs. 5(a)-(b) that this pressure bump causes only a mod-
erate increase in the helium temperature (-1l0 mK), but a relatively large increase in the inner coil
temperature (-420 mK) in the tenth dipole. This is somewhat obvious by examining Fig. 2(b). Most
of the beam-gas scattering energy is deposited in the second dipole, and radially, about half of the
total energy is deposited in the inner coil.
4.40
b
Time: Time:
0: 10 min o 10 min
4.35
g
+ : 20 min
0: 30 min g 4.6
+
o
20
30
min
min
X : 40 min x 40 min
[;l 4.30 '"::>'"
0 : 60 min
l:, : 120 min ... o 60 min
120 min
~
~ 4.25
i2
'"0..:>l 4.4
..,'"...
:>l
~
~ 4.20 (3
CJ
::l
[;j '"'"Z 4.2
4.15 ~
20 40 60 40
DIPOLE NUMBER DIPOLE NUMBER
Figure 5. Temperature profiles along the 1 km section for 150 fold

pressure bump: (a) helium, (b) inner coil.
902
Figs. 5(a)-(b) also show that the pressure bump effect is confined to the second and the third
cells. The beam-gas scattering heat load is far less than the recooler capacity, and the temperatures of
the dipoles in the fourth cell remain unchanged. TIle heat duties of the second and the third cell
recoolers are 78 W and 51 W respectively, up from the nominal value of 39 W.
While the cryogenic system capacity is not exceeded for such a high pressure perturbation,
the inner coil temperature of the tenth dipole in the second cell (T ... 4.68 K) is far above the desired
operating limit of 4.35 K. Therefore, the increased heat load due to beam tube vacuum degradation
in a dipole is more likely to initiate a magnet quench before levels that would overload the cryogenic
system can be attained. We also studied cases with beam tube pressure perturbations of 100 and 50
times the nominal value: the corresponding inner coil temperatures for the tenth dipole are 4.44 K
and 4.33 K.
6.2 Heat Load Disturbance in a Half-Cell
The effect of an sudden increase in the heat load of a half cell due to an external source was
studied. As before, after 10 minutes of steady state operation, an additional heat load of 1 W/m was
imposed on the first half-cell of the second cell. This load is assumed to originate from outside the
coldmass (e.g. due to a leak in the cryostat insulating vacuum) and is therefore assigned to the outer
layer of the yoke in the model.
Fig. 6(a)-(b) depict the resulting temperature profiles along the I km long string section at dif-
ferent times. The largest temperature increases occur in the last dipole (about 220 mK for the helium
and 240 mK for the inner coil). Despite the increased beam tube temperature levels in the second
cell, the corresponding gas equilibrium pressures in the beam tube do not reach the levels to produce
significant beam-gas scattering effects. The recooler duty for the second cell increases to 119 W as a
result of the added external heat load. The effects of the disturbance are localized, resulting slightly
higher temperatures in the next cell and a small increase (from 39 W to 43 W) in the recooler duty
for the next cell.
4.5
4.5
0: 10 min
+ : 20 min
g 4.4 o : 30 min g 4.4
X: 40 min
'"
0:
~
0: 60 min '"
0:
....
~
'" : 120 min <
~ 4.3
0:
'"
0.. '":>!
0.. 4.3
:>!
'":>!
.... '",...,
....
~ 4.2
au 4.2
::l 0:
'"
X
'"~
Z
4.1
0 20 40 60 20 40 60
DIPOLE NUMBER DIPOLE NUMBER
(a) (b)
Figure 6. Temperature profiles along the I km section for the heat load
perturbation of 1W1m: (a) helium, (b) inner coil.
7.0 CONCLUSIONS
We have studied the dynamic interaction between the beam-gas scattering energy deposition
and the sse cryogenic system by integrating a cryogenic system dynamic simulator with an adsorp-
tion model and a beam-gas scattering and energy deposition model. We have obtained simulation
results for one sse string section that shows the cryogenic system evolving toward new states as a
result of a) a local disturbance in the beam-tube pressure in a dipole, and b) a local disturbance in the
heat load of a half-cell.
903
In the case of a perturbation in the beam tube pressure of a dipole, the cryogenic system capac-
ity can easily accommodate the increased energy deposition in the cold mass due to beam-gas scat-
tering. The beam-gas scattering effect does not propagate than the following cell. TIle beam-gas
scattering has a more immediate impact on the coil temperature than on the cryogenic system capac-
ity and can provoke a magnet quench for a sufficiently high perturbation in the beam-tube pressure.
In the case of an externally imposed heat load on a half-cell, the beam-gas scattering induced energy
deposition remains low despite high beam tube temperatures.
8.0 REFERENCES
1. Edwards, H.T., "Study on Beam Thbe Vacuum with Consideration of Synchrotron Light, Potential
Liner Intercept, and Collider Quad/Spool Coil Diameter," SSC Report SSCL-N-771, August,
1991.
2. Paige, F.E., and Protopopescu, S.D., "ISAJET 6.24: A Monte Carlo Event Generator for pop and
pbar-p Reactions," BNL Document, 1990.
3. Mokhov, N. V., "The MARS 10 Code System: Inclusive Simulation of Hadronic and Electronmag-
netic Cascades and Muon Transport," Fermilab Report FN-509, March 20, 1989.
4. Brunauer, S., Emmett, P.H., and Teller., E., "BET Model for Physical Adsorption," J. Am. Chern.
Soc., Vol. 60. P 309. 1938.
5. Benvenuti, c., Calder, R.S., and Passardi, G., "Influence of Thermal Radiation on the Vapor Pres-
sure of Condensed Hydrogen (and isotopes) between 2 and 4.5 K," J. Vac. Sci. Technol., Vol.
13, No.6, Nov/Dec 1976.
6. Site-Specific Conceptual Design of the Superconducting Super Collider, SSCL-SR-1056, July,
1990.
7. Hartzog, D.G., et al., "Dynamic Modeling and Simulation of the Superconducting Super Collider
Cryogenic Helium System," Supercollider 1, Plenum Press, New York, 1989.
8. Hindmarsh, A.c., "ODEPACK, A Systematized Collection of ODE Solvers," Scientific Comput·
ing. R.S. Stepleman et al. eds., North-Holland, Amsterdam 1983, pp. 55-64.
904
A FACILITY DESCRIPTION OF THE
SSCL MAGNET TEST LABORATORY
Mark W. Coles

Dallas, Texas 75237
INTRODUCTION
The Magnet Test Laboratory (MTL) at the Superconducting Super Collider Laboratory
(SSCL), presently under construction at the Nl5 site west of Waxahachie, will be used to
cold test superconducting magnets prior to their installation in the high energy booster
(HEB) and the main ring ofthe collider. This paper describes this facility, the activities and
apparatus planned for use within it, and the schedule for completion and phasing in of the
operating facility. More detailed information regarding the physical facility and its
functional requirements is given in References I and 2. Additional references are given
within this paper to apparatus that has been designed for use in the MTL and that may be of
interest to those building similar facilities.
Properties of superconducting magnets will be tested in resistive and superconducting
states. Ten single-magnet cold test stands, a three-magnet string test area, and a vertical
dewar test area will all be installed within the MTL. Eight of the single-magnet cold test
stands are intended for production testing, while the remaining facilities are more heavily
instrumented and are intended for detailed research investigations of superconducting
magnets, cable, and wire. It is anticipated that the MTL will have a production test rate of
between one and two magnets per day. (Asymptotically, roughly 10% of all magnets
provided by vendors and 100% of all magnets built by the SSCL will be cold tested.)
Typical warm magnet tests to be performed in the MTL include hi-potting, electrical
continuity check, test of electrical isolation tum-to-tum and conductor to cryostat, check of
alignment of the beam tube and the magnetic field with respect to magnet fiducials,
measurement of the magnetic field harmonics at low current, and measurement of coil
prestress. Warm testing within the MTL will be conducted only on magnets that are also
scheduled for cold testing; routine warm tests for all magnets are planned for the proposed
Magnet Inspection and Storage Facility (MISF). Typical production cold testing activities
will be vacuum testing, verification of the electrical isolation of the conductor when a
*Operated by the Universities Research Association, Inc, for the U. S. Department of Energy under Contract
No. DE·AC3S-89ER40486.
Supercollider 4, Edited by 1. Nonte

magnet is cold, verification of the operation of the quench heaters, quench testing,
measurement of field characteristics at constant current, fundamental field angle orientation
(field angle for dipoles, field angle and center for quadrupoles), and integral product of field
times length. More detailed R&D activities planned for the single magnet test stands are
quench evolution and propagation studies, measurement of stress due to Lorentz forces,
determination of magnetic field characteristics at various operating currents and
temperatures from 2.5 K to 4.5 K, determination of detailed quench characteristics (critical
current, quench initiation location, propagation velocity, etc.) as a function of temperature,
and time-dependent field measurements. Similar activities will also be carried out on short
test magnets in the vertical dewar facility. The string test studies will include repetitive
thermal and power cycling, heat load tests, evaluation of magnet interconnect performance,
and quench studies.
Figure 1. Flow of magnets to be tested in the Magnet Test Laboratory.
In general, production magnet tests in the MTL will be done to determine systematic
or process-dependent performance characteristics. Random defects (shipping damage,
dimensional tolerance, workmanship errors, etc.) will be identified through testing in the
MISF. Figure 1 shows the anticipated magnet flow into and out of the MTL. It is anticipated
that all magnets delivered to the MTL will arrive from the MISF, and all production
magnets exiting the MTL will return to the MISF. It is presently proposed that the MISF be
constructed adjacent to the MTL. Anticipated flow of magnets within the MTL is shown in
Figure 2. Common bridge crane coverage of the cold test floor area, warm test area, and the
cable and vertical test area simplifies circulation of magnets from the receiving area. The
bridge crane has a 25-ton lift capacity and a 20-ft hook height.
906
cold"", ~
cable/vertical test
Figure 2. Flow of magnets and cable samples through the M1L.
FACILITIES DESCRIPTION
The floor plan of the MTL is shown in Figure 3. The warm test areas, on either side of
the central cold test area, provide space to determine warm measurements as part of the
study of warm-cold correlations in magnet behavior. Of special interest are correlations
between warm and cold magnetic multipole measurements. Since the fields being measured
are so small, power provided to this area is well isolated from the high-current DC power
supplies and various motor power requirements of the MTL.
The cold test area, in the center of the high bay, is clustered around the central
cryogenic distribution box mezzanine. Cryogens enter each magnet from the distribution
box through the feed end, flow through the magnet to the turn-around end can, then return
through the magnet to the feed can and into the refrigerator room. Additional space is
provided near the end of each test bench in order to insert a measurement probe and to
transport it through a magnet. This requires approximately 10 ft beyond the tum-around end
can and about 5 ft beyond the feed can end. Vacuum pumps and the magnetic measurement
data acquisition electronics will be located on cars that can be rolled from stand to stand.
Power supplies for the single-magnet cold test are located in the basement under the
cryogenic mezzanine, within the central area of the high bay. Seven power supplies are
shared among the ten single-magnet stands and the three-magnet string. There are dedicated
supplies for the R&D test stands and the three-magnet string; the eight production stands
share the remaining four supplies. At least one of the supplies provides an output of 8000 A
at 40 VDC; the remaining power supplies will be rated at 10,000 A. (Recent advances in the
performance of the superconducting cable have necessitated the high current rating in order
907
co
o(Xl
Probe Shop
Matellals ProcesSlngl
Slaglng
Control
Mach. Compressor
EQuIp d;-.;:'", - ':';:: ~
l'
. .. . ,0'. c 11Jl tl<llJ U <I II U II 11
.. :l _... :: :: =:J
oJ q:: :d:: ::;; .... :' ;; H :; <:filii 0<111 U II 11 1I II "
~- :. ~:;:!:~ - .. :~ ;.o
=t; --:-:-:- , :.; ::-:: t • • • II :: c:n 1<1 tl <I II 11 11 1111 II II

Cable & Warm Tes!s ...... ';1/ rIO
~ .. : == ~ J;l
Vertical :r:::: .... " Warm Tests
Tesl :1=." :0
. "
Cold Tesls c: ~:;::-:: "
~ .
A' 01 ••••
,,:: :: ::~:;:::
............ , "or ' ..... .. II
u" ,. U U"LIo .. ~::: ~: t.: If:'; :: ~:.tJ
...... .......
-,- -
Lila Cycle TeSI COld Tes! s
sasemen t 7
Figure 3. Floor plan of Magnet Test Laboratory showing layout of magnet test stands and central cryogenic
distribution boxes.
to conduct quench studies at temperatures in the region of 2.5-3.5 K.) The 8000 A power
supply specification is given in Reference 3. Power to the production stands is routed
through link boxes that manually switch the power from one stand to another. Two of the
production test power supplies are also equipped with reversing switches for use with HEB
magnets. The reversing switches are also connected via link boxes. (Details of the link box
design are given in Reference 4.) Also located in the basement are the dump switches.
Water-cooled buswork connects the link boxes to the power leads in the magnet feed cans.
Use of a basement to house this equipment allows the test floor to be clean and free of
overhanging wires and cabling that would otherwise interfere with rapid installation and
removal of a magnet under test. Note that the basement has an access hatch so that the
power supplies may be installed and removed using the overhead bridge crane.
The feed can, the magnet under test, and the turnaround end can all sit on a cold test
stand. (Specifications for the feed can and the turnaround end can are given in References 6
and 7, respectively.) These stands, located on either side of the cryogenic distribution box
mezzanine, are interleaved so that a long (approximately 22 m) warm bore may be inserted
in either end of a magnet prior to cool down when it is needed to make magnetic
measurements. Also, the stands are spaced so that there are alternating wide (12 ft) and
narrow (8 ft) aisles on each side of each magnet. (Details of the cold test stand design are
given in Reference 5.) The test stand is designed to accommodate all anticipated magnet
lengths. The data acquisition electronics are located underneath the cold test stands in
portable racks (so they can be moved when magnets of different lengths are tested). Details
of this equipment (isolation amplifiers, MVMEl47 single board computers, data loggers,
strain, cryogenic sensor readout electronics, quench detection electronics, UPS backup for
the quench detection system, and the heater firing units) are given in Reference 8. It is
anticipated that all data-acquisition electronics will be located in the immediate vicinity of
the cold and warm test stands. Only digital lines will connect the test floor to the MTL
control room. Trenches located in the floor of the high bay separately conduct conditioned
and unconditioned power, digital, and other communication lines to the individual test
stands and to the power supply controls in the basement.
Within the control room are 11 individual workstations for controlling each cold test
stand and the three-magnet string. Warm test area activities will not be controlled from the
control room. The software architecture used for data acquisition and control (see Reference
9) allows any workstation to control any test stand, including the regulation of the flow of
cryogens through the feed can and into each magnet. A separate console for control of the
refrigeration system is also provided. The other workstations do not control the internal
operation of the refrigerator.
Workstations will be linked via Ethernet to the local area network (LAN), which is
also connected to the data acquisition processors, the file server, and a LAN router. A
window in the control room provides a limited view of the high bay. Supplemental TV-
camera coverage will be used as necessary. The control room has a raised floor in order to
facilitate cable routing.
The computer room will house two file servers and the LAN router. A "jukebox" disk
storage system and a UPS power supply will also be located in this room. One file server
will archive all warm and cold magnet test data; the other will handle instrumentation
calibration information and system configuration data.
The probe shop will be a testing, maintenance, and calibration laboratory for magnetic
field measuring probes. All probes used to determine the quality of the magnets will be
calibrated and certified in this lab prior to their use on the magnets. Probe development and
enhancements will be performed in this facility. The probe shop will also be used for
normal electronic instrumentation support and development activities associated with the
magnet test operations. Magnetic multipole measurement equipment supplied to the magnet
subcontractors will be periodically recalibrated here..
909
The cable and short-magnet test area will contain two short-magnet test dewars, one
cable test dewar, one wire test dewar, and one pre-cool dewar. The magnet test dewar, the
wire test dewar, and the cable test dewar may operate simultaneously. Sufficient space is
available to add more short-magnet test dewars if circumstances later require it. Ajib crane,
covering the entire dewar test area, will be used to lift and insert short magnets into dewars.
The short-magnet test area has a dedicated data acquisition system and a file server that is
separate from that of the long-magnet test facility. The short-magnet control room is
exclusively for this area and will provide four control and data display stations. Anyone of
the control stations will be able to run an experiment at anyone of the vertical dewars.
Space is provided adjacent to the dewars for staging of the magnets and test samples. The
dewars will have access to liquid helium supplied at atmospheric pressure from the MTL
refrigerator.
Data acquisition and control electronics are located in racks directly behind the
de wars. The facility will contain four 1O,000-A bipolar power supplies. There will also be
one 15,000-A supply. (Details regarding these supplies can be found in Reference 10.) In
addition to the supplies, there will be four energy dump systems that will each contain an
SCR circuit to remove energy from a magnet.
Engineering labs in the MTL will be used to study the material properties of
superconducting cable. Individual areas will be dedicated to physical testing, non-
destructive testing, a metallurgical laboratory, a scanning electron microscope room, and an
optical lab. A small machine shop area will be provided to support maintenance and
installation activities within the MTL. An instrument repair shop (for maintenance of
electronic equipment) is also located within the MTL. Individual office space will be
provided for 30 people. The total population of the MTL (the sum of all three shifts) will be
approximately 100 people.
The refrigerator compressor is located in a separate, sound-isolated building abutting
the Accelerator Systems String Test (ASST) compressor building. The cold box room abuts
the MTL high bay to limit the length of the transfer lines that run from the cold box to the
distribution boxes. The refrigerator ll can supply 2000 W of refrigeration while liquefying
20 glsec of liquid helium at a combined flow rate of 500 glsec to the magnets under test.
Other important infrastructure features of the MTL are the low-conductivity cooling
water supplied to the long and short magnet power supplies and to the calibration magnet in
the probe shop (approximately 300 gpm), the 25-ton bridge crane that provides coverage of
the entire high bay, oxygen-deficiency hazard monitoring through the high bay area, and the
provision of transformer-isolated instrument power in order to facilitate very sensitive
electrical measurements.
It is possible that the cold test capability of the MTL could be expanded. The optimal
level of production sampling will be determined dynamically through sampling statistics
accumulated during production experience. Expansion, if required, will occur to the west of
the high bay. The layout of the present building facilitates expansion through the
implementation in the design of two access doors, bridge crane motion in the direction of
expansion, easy conversion of the conference room to additional control room space, and
adjacency of the ASST and N15 refrigerators so that additional refrigeration capacity could
be added if required.
DEVELOPMENT PLANS
The principal milestones for completion of MTL construction are given in Table 1.
First cool down of a long magnet will be early in 1993, following the commissioning of the
MTL refrigerator. Initially, the MTL will have only a single cryogenic distribution manifold
with five distribution boxes, two cold test stands, and a single 8000-A power supply. Once
910
Table 1. Construction and cryogenic refrigerator installation milestones for the Magnet
Test Laboratory.
Activity Description Start Finish
Design Title I (30%), Value Engineering 1/91 3/91

CCB Change Control 3/91 4/91
Design Title II (100%) 5/91 8/91
Procurement A ward-foundations 12/91 12/24/91

Construction Pit and drilled piers 1/92 2/20/92
Procurement Award-general buildings 1/92 2/5/92
Construction MTL steel erection 3/92 5/92

Construction MTL building enclosed 5/92 6/92
Construction MTL substantial completion * 7/31/92
Construction MTL subcontract completion 8/31/92
Construction Compressor bldg. steel erection 2/92 4/3/92

Construction Compressor partial beneficial occupancy 4/30/92
Construction Compressor substantial completion 5/15/92
Construction Refrigerator bldg. steel erection 2/92 3/15/92

Construction Cold box installation 3/16/92 3/22/92
Construction Refrigerator bldg. partial BOD 4/15/92
Construction Refrigerator bldg. substantial completion 4/30/92
*High bay and low bay control room, computer room, and electrical equipment room may be utilized for their
intended purposes at this time.
these equipment items are evaluated, additional instrumentation will be procured so that
four single-magnet test stands and the three-magnet string test will be operational in the
second half of 1993. The additional six distribution boxes will be procured for delivery in
1994.
REFERENCES
1. A-E/CM Work package authorization for the Magnet Test Laboratory and Compressor Building, SSC
Control No. C42-000003.
2. Contract documents, Issue for Bid (JFB) set of 13 Dec., 1991 for the MTL, PBIMK document control No.
CPB-001068. This may be purchased from PBIMK with SSCL approval. These documents are also
available for reference through SSCL document control.
3. Specification for a 40 Volt, 8000 Amp DC power supply for the Magnet Test Laboratory, SSC Control
No. A40-000001.
4. Linkbox 2A, part no. M60-000953, and Linkbox 2B, part no. M60-000962.
5. MTL Cold Test stand, part no. M60-000901.
6. Feed Can Equipment Specification, SSC Control No. M62-000025.
7. End Can Equipment Specification, SSC Control No. M62-000034.
8. Electrical Instrumentation for Cold Test of SSC Magnets, J.E. Dryer and Y. Liu, presentation IX-4 of
these proceedings.
9. Software Design Philosophy for the SSCL Magnet Test Laboratory, P.A. Craker et aI., presentation IX-25
of these proceedings.
10. Specification for a 6 VDC, 15,000 Amp Power Supply, SSC Control No. M30-000400, and Specification
for a ±lO,OOO Amp ±12VDC Power Supply, SSC Control No. M30-ooo294.
11. Refrigerator Specification-Request for Proposal, MTL Cryogenic System, RFP No. SSC-90-A-Ol107.
911
TRANSPORTATION STUDIES: 40-MM COLLIDER DIPOLE MAGNETS
E. Daly

Dallas TX 75237
ABSTRACT
Several fully functional 40-mm Collider Dipole Magnets (CDM) were instrumented
with accelerometers to monitor shock and vibration loads during transport. The magnets
were measured with optical tooling telescopes before and after transport. Changes in
mechanical alignment due to shipping and handling were determined. The mechanical
stability of the cryogen lines were checked using the same method. Field quality and dipole
angle were measured warm before and after transport to determine changes in these
parameters. Power spectra were calculated for accelerometers located on the cold mass,
vacuum vessel, and trailer bed. Where available, plots of field quality and dipole roll both
before and after were created.
Shipping loads measured were largest in the vertical direction, where most of the
structural deformation of the magnet was evident. It was not clear that magnetic
performance was affected by the shipping and handling environment.
INTRODUCTION
CDMs will be transported by truck from vendor production facilities to the acceptance
location prior to installation both in the Accelerator Surface String Test (ASST) and in the
Collider Ring. It is important to ensure that the CDMs can be "safely" transported. "Safely"
is defined here as maintaining comparable mechanical integrity and magnetic performance
before and after shipment.
Early work} consisted of transportation and vibration computer models that attempted
in part to predict the response of a CDM to transportation loadings. Earthquake Engineering
(EQE) used input for these models that were formulated from actual test data taken from
road tests conducted on large shipping containers (gross weight roughly 39,900 kg) at
Sandia National Laboratories. 2 Maximum accelerations for several different road conditions
were less than 1 g for all events, and RMS loads were not greater than 0.25 g. The report
No. DE-AC35-89ER40486.

recommended that actual testing of the final trailer mounting configuration was required to
verify this work.
An in-depth transportation study was then conducted by Vibration Engineering
Consultants (VEC) at Fermi National Accelerator Laboratory (FNAL) on a 40-mm dipole
DSIflP.3 DSIDP is a dedicated transportation and vibration prototype, with accelerometers
and strain gages permanently installed, intended for structural and dynamic studies. This
test was geared toward measuring shock and vibration loads during transport. Several
mounting schemes were investigated. A set of mounts that attached the magnet to the trailer
were built, as well as an axial restraint that served to couple the cold mass to the vacuum
vessel at the return end of magnet.
Initial results showed that the existing method of transport would indeed compromise
mechanical alignment of the magnets, but correlation in terms of vibration response
between the actual test data taken by VEC and the computer models developed by EQE was
poor. The mechanical data taken showed deformation of the magnet, but in a qualitative
sense only. Deformations were observed, but the magnitudes were not determined, and the
mechanisms causing the deformations were not identified.
Another series of transportation studies were conducted in the interest of learning more
about the magnet's response to shipping and handling. Particular attention was paid to
quantitatively determining the deformation of the magnets due to the transportaJion
environment. Magnets DC0202, DC0201, and DDOO28 were transported between FNAL
and Brookhaven National Laboratory (BNL). Studies on these three magnets were
conducted not only to measure dynamic loads during transport and to qualitatively identify
structural changes caused by transport, but also to quantify changes in magnetic field
quality if possible.
Results from these studies are contained within the body of this report. Acceleration
spectra are consistent with earlier test data. Natural frequencies of the magnet-trailer system
correspond well with those predicted by computer models created at SSCL.4 Mechanical
deformations of the vacuum vessel are as great as 1 mm in the vertical direction.
Measurements of the change in location of the cryogen lines show displacements of the
order of 5 mm. Warm magnetic field quality and dipole angle measurements were taken
before and after transport for DC0202. Values of the magnetic multipoles and dipole angle
were determined versus position along the length of the magnet.
TESTING EQUIPMENT AND PROCEDURES
The magnets were instrumented with 14 accelerometers placed on the cold mass (CM),
vacuum vessel (VV), and trailer bed (TB) to monitor accelerations in as .many as three
directions. The vertical direction (V) is aligned with gravity, the axial direction (A) is
aligned with the magnet centerline, and the horizontal direction (H) is orthogonal to the
vertical and axial directions. The accelerometers were attached to both the lead and return
ends of the magnets as shown in Figure 1.
The accelerometers were PCB 308 units with a nominal output of 100 mV/g and a
resolution of O.OOlg. PCB 483A07 amplifying power units were used to supply power to
and amplify output signals (xl0) from each of the accelerometers. The amplified signals
were then fed to a Kyowa Dengyo RTP-650A 14-chann~1 tape recorder, which converted
the time-varying voltages to frequency-modulated data. The signal-to-noise ratio on the PM
band used was 41 dB nns .
In order to power the transducers, amplifiers, and data logger, a 2.2-kW portable
gasoline-powered generator was attached to the trailer bed. The units that required 110 VAC
were connected to the generator through a surge suppression strip and were contained in a
weatherproof box mounted to the trailer bed.
914
Figure 1. Accelerometer Locations During Transport.
The accelerometer signals were recorded to tape during the entire trip of -950 miles,
covering a variety of road conditions and speeds. The tape-recorded time signals were fed
into an HP 3567A Spectrum Analyzer controlled by an IBM PC. The time histories were
converted to power spectral densities (PSOs) using a uniform window over a 4-sec time
record. The analyzer was set for a range of 0--200 Hz using 800 frequency lines, resulting in
resolution of 0.25 Hz. Using a peak hold type of measurement, the analyzer recorded the
largest value of acceleration for each frequency line over the loo-Hz range for the entire
trip.
The magnet was attached to a 55-ft air ride trailer using a pair of magnet mounts
developed at FNAL, consisting of a combination of inflatable air bags that supported the
magnet at its external feet locations. These mounts were used for early transportation
studies, moving magnets from building to building on site at FNAL, and cross-country trips
between FNAL and BNL. The total weight of the magnet and trailer was less than
20,000 kg.
An axial restraint that rigidly coupled the cold mass to the vacuum vessel during
transport was used as well. This restraint was designed to allow the long-term, or slow,
motions such as thermal contraction or expansion while restricting the relatively short-term,
or fast, motions such as impulsive forces associated with shipping and handling.
A pair of endbells sealed the magnet from the environment. BNC feed-throughs routed
the cold mass accelerometer cabling from the sealed interior of the magnet to the tape
recorder. The magnet was backfiIIed to positive pressure with gaseous nitrogen and set to
purge during transport.
None of the magnets were temperature controlled. The shipment dates and average
temperatures are shown in Table 1. The upper limit on temperature for the magnets is 30°C.
It is obvious from looking at the temperature ranges that this was not a problem.
A series of tooling balls were placed at the five reinforcing rings, numbered 1 through
5 from return to lead end, on the vacuum vessel, and the end rings on both the lead and
return ends. The cross section (Figure 2) shows seven tooling balls at each location, A-F. At
the return end of magnet 000028, four tooling balls were welded to the cryogen pipes and
were labeled H, J, K, and L. The positions of the tooling balls were measured vertically and
laterally using several jig transits and a pair of tooling bars. The same workers and set of
instruments were used to measure the magnet both before and after transport.
A tooling ball network was generated from each data set. Two points from each data
set, points 4F and 2F, were chosen as coincident, and a third point, 4B, was chosen as co-
planar. All other points in the data sets were rotated accordingly.
Two types of magnetic-field quality probes called moles were used to perform these
measurements. At FNAL prior to transport, the B2 mole logged multipole values and dipole
915
. Table 1. Magnet Temperature During Transport.
Average Maximwn
Magnet Dates of Travel Temp. (0C) Temp.(OC)
DC0202 12/19-12/20/90 0 5
DC0201 1/29-1/30/91 5 10
DD0028 3/19-3/20/91 0 5
D
o
c E
o o
B 0 o F
x
o o
A G
Figure 2. Tooling Ball Locations at Return End of DD0028.
916
angle at 28 positions along the magnet's length. At BNL after transport, the Dl mole logged
the same values at 17 points along the length. One major difference between the moles was
the dimensions of the rotating coil that senses the magnetic field. The B2 mole (FNAL) has
a 24-in coil, while the Dl mole (BNL) has a I-m coil. Another difference is the diameter of
the rotating coil. The D series had a smaller diameter due to more insulation between the
coils and the measuring device.
RESULTS AND DISCUSSIONS
Table 2 illustrates the types of tests conducted on each of the three magnets. It would
have been most desirable to conduct all of the tests listed below on all three magnets, but
schedule constraints and function at destination precluded conducting the full array of tests
on the trio of magnets.
Table 2. Magnet Measurement Matrix.
Measurement DC0202 DC0201 DD0028

Dynamic Loads
Mechanical Alignment
Cryogen Line
""
x
"
x
x
"
""
Alignment
Warm Magnetic
Measurements
Cold Magnetic x
" "
x
"
x
Measurements
As mentioned earlier, the accelerometer time histories were converted to PSDs using
the spectrum analyzer. A PSD is a graphical representation of values of power density
versus frequency, and it represents the distribution of vibration energy with frequency. The
PSDs are important results, and they provide useful information to the designer of a shock
and vibration isolation system.
Generally, the transportation environment was comparable to the studies mentioned in
the introduction. Acceleration spectra measured in the vertical direction were the largest,
roughly ± 1 g incident to the cold mass. The dynamic displacements of the entire structure
on the trailer bed were of the order of several inches. This displacement was due to the air
bag suspension system that held the magnet to the trailer bed. Axial and horiwntal spectra
were of the order of ± 0.5 g, which were well below design parameters. Specifications for
the 40 mm design B cryostat required ± 2 g vertical, ± 1.5 g axial, and ± 1.0 g lateral.
However, this dynamic environment caused the vacuum vessel to change shape. This
suggested that another mechanism caused the magnet to deform.
Tooling ball locations were measured with respect to a pair of tooling bars both before
and after shipment. Three points on each magnet were rotated to reside in the same plane to
facilitate a comparison of changes in mechanical position. This mathematical operation
minimizes errors associated with instrument set-up, and allows for comparison of position
of magnet components both before and after shipment.
The overall change in elevation of the magnet is shown in Figure 3. The magnets
exhibited a bowing downward of roughly 0.25 mm at the midpoint. The final shape was
coincident with the first vertical mode of vibration of the magnet, as can be seen in the PSD
plots given earlier. This shape change was most likely due to the method of support used
during transport. Note that the beam tube, point 0, was displaced by roughly 1 mm from its
original position. This change was attributed to the installation and removal of the axial
shipping restraint that was attached to the return end of the magnet.
917
I--POINTB - - - POINT 0 I
2.5
1.5
Y (mm)
0.5
-0.5 I
RETURN END I I
LEAD ENOl
Z (m)
Figure 3. Change in Vacuum Vessel & Beam Tube Vertical Position.
Questions were raised regarding the integrity of the folded path posts during transport.
It was thought that the shock and vibration environment would cause the posts to fatigue
and yield, resulting in misalignment of the cold mass within the vacuum vessel. Figure 4
illustrates the change in radial position of the beam tube centerline with respect to the
vacuum vessel exterior at post #1. The beam tube centerline and consequently the cold mass
position within the vacuum vessel were preserved, although the overall position of the
reinforcing ring was changed. Changes in lateral position were within the tolerance of the
measurement devices or slightly greater. Little change was expected due to the minimal
dynamic inputs experienced during the trips.
1-- Initial Position - - . - - Final Position
500 o
400 c _......... - .... --,_ E
,a:--- ---
~,--
300 ~ ,
B
20~'
,
"0
-200
"
'
"
,. F
-800
Y(mm)
-300
NOTE: Change in Position is Scaled by 700
x (mm)
Figure 4. Change in Radial Position at Post 1.
918
The overall change in location of the cryogen lines (see Figure 5), as much as 6 mm
vertically at the end of the magnet, was due in part to the change in shape of the vacuum
vessel. Structural instability-twisting of the thermal shielding-accounted for most of the
permanent displacements. The cryogen lines and shields experienced permanent
deformations of as much as 4.5 mm vertically and 0.5 mm laterally with respect to the beam
tube centerline. These motions were independent of the change in shape of the vacuum
vessel.
16 Before Transport .. After Transport
400.0
300.0
200.0
100.0
Y(mm)
0.0 .....f---I~o:---+---+--~:---="iH~--+-...-I
(200.0)
(300.0)
Note: Change in position scaled by '0
x (mm)
Figure S. Change in Cryogen Line Location at the Return End of DD0028.
Measurements of magnetic field quality were conducted both before and after
transport. Figure 6 shows the results of a warm Z-scan conducted both before and after
shipment. These data were taken from the DC0202 study. WandererS indicated that these
changes should be viewed as upper limits to changes in multipoles due to transportation.
B1 VI, Position ; BeforelAfter

Transport
8'C_.,j ~I I
-10" -5 0 5 10
POSITION (meters from magnet center)
•• , 11 DEC 90 • FNAL: Worm z-scan, yoked, Go 11 Nor 91 • BNL: lit waim Z _

cryostatecl, B2 mole befae lit CD, 01 Male
Figure 6. Warm Z-scan Before and After Shipment
919
Figure 7 plots dipole roll angle both before and after shipment along the length of the
magnet. There is a definite shift between the data sets of roughly 10 mrad. It was most
likely an instrumentation bias due to the differences in construction between the two moles
that were used to obtain the measurements.
Dipole Roll Angle ¥s, Position;

Before/After Transport
DIPOLE ROT.
ANGLE
(mradians)
I
·10 ·5 o 10
POSITION (meters from magnet center)
.1. 11DEC 90. FNAL: Warm z·scan. yoked. .Q..ll Mar 91 • BNL: 1st warm Z-5car1
cryostated, B2 mole before 1st CD, 01 to4oIe
Figure 7. Dipole Roll Angle Before and After Shipment.
CONCLUSIONS
The 40-mm design B vacuum vessel design is compromised by the shock and
vibration environment experienced during transport. The magnitudes of the PSDs show
relatively low vibrational energy transfer to the magnet when compared with earlier studies
conducted on DSHIP and DDOO 18. The vertical accelerations were largest-of the order of
1 g incident to the cold mass.
The deformation of the magnet in the vertical direction is greater than acceptable
limits. This deformation, primarily the vacuum vessel, is due to the present trailer-mount
configuration and weld relaxation at the reinforcing rings.
The compact cryogenic support posts withstood the transportation environment and
maintained the cold mass position with respect to the vacuum vessel reinforcing rings at
each of the five support post locations within 0.125 mm.
The present shield assemblies do not withstand the transportation environment and
deform by as much as 6 mm in the vertical direction.
Magnetic measurements have provided an upper limit on the changes of multipoles
due to transportation. To better understand these changes, measurement equipment with
improved accuracy must be developed.
FUTURE WORK
All transportation studies including this one have been conducted with 40-mm
magnetic aperture CDMs. The baseline design for the ASST and consequently the Collider
Ring has been changed to 50-mm aperture CDMs. Therefore, studies should be conducted
with 50-mm CDMs, which are 25% heavier with a different vacuum vessel design. These
studies should begin with an improved method of support of the magnet during shipment.
Transportation issues will be extended to the many other types of magnets that will be
installed in the Collider and High Energy Booster as well. Studies conducted on other
magnets will provide insight into the variety of shipping problems that may be encountered.
920
ACKNOWLEDGEMENTS
The author would like to thank both M. Ruschman of the InterGalactic Engineering
Lab, who was involved with the technical aspects of many of the early studies conducted at
FNAL, and D. Sorensen for their assistance with setting up the studies.
REFERENCES
1. BJ. Benda, R.D. Campbell, D.J. Doyle, and J.J. Johnson, "SSC Dipole Magnet System: Stress Analysis
for Seismic and Transportation Loadings," EQE Engineering, January 1991.
2. R.E. Glass and K.W. Gwinn, ''TRUPACT-I Over-the-Road Test," Sandia National Laboratories,
SAND87-0513, October, 1987.
3. Vibration Engineering Consultants, Inc., "SSC Magnet DSHIP Transportation Vibration Test," 1989.
4. A.R. Jalloh, R. Viola, and E. Daly. "The Mechanical Response of the SSC Dipole Magnet to Ground
Motion," Proceedings from the 1991 International Industrial Symposium on the Superconliucting
Super Collider.
5. J.T. Foley, M.B. Gens, and C. F. Magnuson, "Current Models of the Dynamic Environment of
Transportation," Proceedings of the Institute for Environmental Sciences, pp. 35-44 (1972).
6. C.M. Harris and C.E. Crede, Shock and Vibration Handbook., McGraw-Hili, 1976.
7. c.F. Magnuson, "Shock and Vibration Environments for a Large Shipping Container During Truck
Transport (part I)," Sandia National Laboratories, SAND77-111O, September 1977.
8. P. Wanderer, "Limit on Multipole Changes in a 17-m Dipole due to Transportation," Brookhaven
National Laboratory, SSCL-N-769, August 1991.
921
EXPERIMENTS TO IMPROVE MATERIALS FOR SSC MAGNETS
E. Gregory,l H. Liu,l G.M. Ozeryansky,l

M.D. Sumption,2 K.R. Marken Ir.2 and E.W. Collings 2
llGC Advanced Superconductors

1875 Thomaston Ave
Waterbury, cr 06704
2Battelle Columbus Laboratories
505 King Ave
Columus, OH 43201-2693
INTRODUCTION
The standard 6 /lm diameter filamentary SSC conductor is now produced, with Ic's as
high as 3000 Nmm 2 at 5 T, in relatively long piece lengths. The design of the billet and the
thermomechanical treatments which are given to the strands were established some years
ago 1. The chief concern in recent years has been the achievement of long and reproducible
piece lengths while maintaining the required Ic level.
Since the early work was carried out, 1 several significant investigations have taken
place and important findings have been made. These are:
1. The determination, in 1986, that the material reported in reference 1 exhibited 3450
Nmm 2 at 5T and 1350 Nmm2 at 8T, when the filament diameter was 3.6 /lID. 2
2. The determination of the importance of filament spacing. 3,4
3. The effective decoupling of closely spaced filaments by the addition of Mn to the Cu
matrix between the filaments. 5
4. The effect of increasing the number of heat treatments on Ic and on the strain at
which the peak Ic appears. 6
5. The development of a modified heat treatment procedure which employs smaller
spacings between the intermediate heat treatments. 7
6. Under high temperature (410 C) heat treatment conditions, there appears to be
some evidence that relatively thin barriers which some workers have used to
prevent diffusion, begin to break down. 8 & 9
In 1987 data on Ic began to be plotted against strain after last heat treatment. 10 &11 In
reference 6, the importance of such curves was pointed out on the basis of a 15 filament
material given a series of multiple heat treatments in order to develop high Ic's.
Many of these findings have yet to be explored in depth. In this paper we report
preliminary results on such explorations.
The ultimate aim of this work is to determine if conductors with closely spaced
filament arrays, a range of filament sizes, a thick barrier and a CuMn matrix, given multiple
closely spaced heat treatments to develop a very high Ie with low proximity coupling, can be
made in the necessary long piece lengths. This preliminary work was carried out on small
diameter billets and for this reason no meaningful data on piece length was obtained. There
was, however, no evidence of a decrease in ductility as the Ie was improved by increasing the
number of heat treatments.

EXPERIMENT AL PROCEDURE
Billet Preparation
Two 50.4 mm diameter billets were assembled using material which had a Cu 0.5 wt
% Mn matrix enabling the filaments to be brought very close to one another without proximity
coupling. The material also had an 8% Nb barrier which enabled the employment of relatively
aggressive heat treatments in small wire sizes.
The two billets differed from one another primarily in the spacing/diameter (sid) ratio
of the filaments; one (S 1) had an sid of 0.15 and the other (S2) an sid of 0.06.
The material was obtained from stock as round rod at 1.63 mm (0.064") diameter.
This was drawn down to a hex with flat to flat of 1.38 mm (0.054"), straightened and cut into
101.6 mm (4") length pieces. The billet that was to have an sid of 0.15 (designated S1) was
packed with material that was at the above size. In order to produce a material that was to
have an sid of 0.06 (designated S2) the rods were etched in sulphuric:peroxide (18% :8%) at
41C (106F) for about 30 minutes. This removed 140 /lm from the flat to flat dimension. The
copper filler rods were made to have the same sizes as the clad filaments in each of the billets.
S 1 was loaded with 555 filaments and S2 with 582 filaments.
These billets were E.B. welded, HIP'd at 750 C under 30 ksi for 0.5 hour and, to
insure that they were indeed fully annealed, the two HIP'd billets were heated to 800 C for
1.5 hours in a vacuum of Ix 10-4 torr and then gas quenched. They were then extruded from
50.4 mm (1.985") to 12.7 mm (0.5") at 600 C. An SEM photograph of S2 in a fine wire
size, is shown in Figure 1 at 2500X. The filaments have a very irregulary shaped cross
sections but they do not appear to be sausaged. That was confirmed by an SEM examination
of the extracted filaments from both S 1 and S2.
Thermomechanical Treatments
The extruded rods S 1 and S2 were examined after four different thermomechanical
cycles. All heat treatments (HTs) were carried out at 375 C for 40 hours and the sizes where
they were applied are illustrated in Table 1, in which filament diameter (fil. 0), filament
spacing (spa), incremental strain (E) and accumulated strain (A strain) are also included.
S2 material was given three such heat treatments with a strain (E) of 0.92 between
each heat treatment. The J c values obtained were relatively low and the J e peak occurs at a
strain after last HT of around 5, where the filament size was -3.6/lm. The SI material was
not tested after this heat treatment but the decision was made to test both SI and S2 after four,
six and eight heat treatments, each closely spaced by a strain of 0.46. It was also decided to
start heat treatments at two die sizes above that used in the 3 HT experiment. The idea behind
this was that it would allow a larger number of HT's and raise the size at which the Jc peaked
to one where the filament diameter was closer to 5 or 6 /lm, the SSC size, as well as at smaller
diameters. The results of Jc at 5T and 7T versus strain after the last heat treatment in materials
given 4, 6, and 8 HT's, are reported in Table 2, together with the strain at which the
measurements were taken. An analysis of the results is given in a later section.
Image Analysis of Filaments

In order to determine the effect of the spacing on the variation of the filament cross
section, back-scattered electron images of the filament cross-sections were obtained by SEM
examination of sectioned and polished samples of the 8 HT wires at 0.16 mm (0.0065")
diameter. These images had sufficient contrast to allow quantitative analysis of filament size
and shape using a Tracor-Northern 8500 Image Analysis System. This analysis included
filament area, minimum, maximum, and mean filament diameter, and aspect ratio. Standard
deviations were calculated for each of these parameters. The number of filaments analyzed
was 274 in wire SI, and 336 in wire S2. The average filament area measured in the wide
spaced (S1) was 13.58 /lm2 with a standard deviation of 1.81. In the close spaced (S2) the
924
Figure 1. SEM of material from billet S2.
average filament area was 13.43 11m2 with a standard deviation of 1.13. The filament area
distribution histograms are shown in Figures 2a & b for SI and S2 respectively. These show
once again the advantages of close filament spacing in the reduction of sausaging even in
filaments with irregular shapes as shown in Figure 1. This difference in cross section was not
discemable by an examination of the extracted filaments for sausaging.
Table 1. Heat treatment schedule.

Wire0(in) fiI0(J.lm) Spal (nm) Spa2(nm) € AStrain S2,3H Sl, 2 4H Sl, 2 6H Sl, 2 8H
0.2294 135.9 20481 8155 0.23 4.32 1 HT 1 HT 1 HT

0.2043 121.0 18241 7262 0.23 4.55
0.1819 107.8 16241 6466 0.23 4.78 1 HT 2HT 2HT 2HT
0.1620 96.0 14464 5759 0.23 5.01
0.1443 85.5 12884 5130 0.23 5.24 3HT 3HT 3HT
0.1285 76.1 11473 4568 0.23 5.47
0.1144 67.8 10214 4067 0.23 5.71 2HT 4HT 4HT 4HT
0.1019 60.4 9098 3622 0.23 5.94
0.0907 53.7 8098 3224 0.23 6.17 5HT 5HT
0.0808 47.9 7214 2872 0.23 6.40
0.0720 42.7 6428 2559 0.23 6.63 3HT 6HT 6HT
0.0640 37.9 5714 2275 0.24 6.87
0.0582 34.5 5195 2068 0.19 7.06 7HT
0.0529 31.3 4722 1880 0.19 7.25
0.0481 28.5 4293 1709 0.19 7.44 8HT
0.0105 6.3 934 372 10.49 Jc Peak

0.0096 5.7 860 342 0.17 10.66
0.0089 5.3 794 316 0.16 10.82
0.0082 4.9 735 293 0.15 10.97
0.0076 4.5 678 270 0.16 11.13 Jc Peak
0.0070 4.2 627 250 0.16 11.29
0.0065 3.8 579 230 0.16 11.45 Jc Peak
0.0060 3.6 536 213 0.15 11.60 Jc Peak
925
Table 2. Jc values in A/mm 2 of S I and S2 materials after 3,4,6 and 8 Heat treatments
Strain3H S1.5T3H S2.5T3H S1.7T3H S2.7T3H Strain4H S1.5T4H S2.5T4H S1.7T4H S2.7T4H
3.1 1827 1020 3.07 952 803
4.03 2265 1287 4.02 2229 1313
4.5 2417 1463 4.21 2237 1336
4.97 2471 1577 4.78 2411 2542 1515 1606
5.39 2254 1445 5.42 2233 2228 1450 1444
5.78 1938 1161 6 1958 1932 1293 1292
6.2 1757 1139 6.53 1634 1683 1121 1111
Strain6H S1.5T6H S2.5T6H S1.7T6H S2.7T6H Strain8H S1.5T8H S2.5T8H S1.7T8H S2.7T8H
3.1 2264 2365 1217 1264 3.05 2453 2592 1373 1386
3.48 2375 2496 1332 1382 3.53 2624 2827 1394 1567
3.86 2829 1510 4.01 2826 2980 1671 1713
4.03 2589 2844 1529 1670 4.48 2637 2797 1515 1664
4.5 2653 2861 1556 1752 4.97 2353 2549 1467 1522
4.97 2529 2705 1529 1677 5.48 1968 2012 1116 1335
5.71 2131 2427 1286 1032 6.02 1171 1765 799 1054
50
'"
u
u 40
c
~
='
30
~
>.
u
c 20
J 10
06 .0 8.5 11.0 13.5 16.0 18.5 21.0

Area, Jlm 2
Figure 2a. Histogram of filament area of NbTi with widely spaced fIlaments.
80
'"
u
u 60 -
c
Ii
u
r r-
8
>.
40 r-
-
5='
l 20 r- r- "-
o
6.0
il
8.5
_I 11.0 13.5
rh
16.0 18.5 21.0
Area, Jlm2
I<'igure 2b. Histogram of filament area of NbTi with closely spaced filaments.
926
Magnetization Measurements
In order to detennine the losses that can be expected in pulsed operation in such
closely spaced filamentary material, magnetization work was carried out to detennine the
extent of proximity coupling at different wire sizes. SI and S2 wires, where the average
spacing was calculated to be in the range between -350 nm and -140 nm were examined.
These magnetization measurements were carried out to determine at what filament spacings
and diameters (and at what wire sizes) the proximity coupling occurs with Cu 0.5 weight %
Mn in the matrix.
Magnetization measurements were made using a Princeton Applied Research Vibrating
Sample Magnetometer (VSM) in conjunction with a ±1.7 T peak field iron core electromagnet.
Magnetization as a function of applied field strength M(H) was obtained for a variety of peak
field sweeps up to the 1.7 T maximum. The sample configuration was an epoxy impregnated
bundle of short straight strands. These bundles were cylindrical with a diameter of 3 mm and
approximate length of 6 mm. The samples were mounted in the VSM with the cylinder axis
(hence the strand axis) perpendicular to the applied magnetic field.
In order to compare the behavior of the strand material with that of the bare NbTi
filaments and thereby separate matrix effects from filament behavior, similar potted samples
were also prepared from strands with the matrix material removed by chemical etching.
Samples denoted "clad" have strand as-drawn, while samples denoted "bare" are bare NbTi
filaments with the matrix removed. .
DISCUSSION OF Jc AND MAGNETIZATION RESULTS
Jc Results
Table 2 clearly shows that, at the same strain after last heat treatment, material with the
closer spacing is generally superior in J c (peak) to that of more widely spaced filaments. The
difference between the two materials increases as the number of heat treatments goes from 4 to
6 and the resulting peak J c s increase. The Jc of the closely spaced 8 HT material is also
above that of the more widely spaced 8 HT material and the Ie in the former reaches 2980
A/mm2 at 5T. This is a high value considering that the material has 3.8 )lm diameter
filaments, an 8% barrier, Cu 0.5 wt. % Mn between the filaments and was /3 quenched at
50.4 mm (-2") diameter, i.e. immediately before extrusion.
The other results to note are that the strain after last heat treatment at which the J c peak
occurs is lower, the higher the number of closely spaced 375 C 40 hour heat treatments,
Figure 3. This is the same effect that has been noted previously for 15 filament wires,6
subjected to the more widely spaced 375 C 40 hour heat treatments.
Magnetization Results
Figures 4 and 5 show comparisons of the M(H) hysteresis loops for the clad and bare
samples at two wire sizes. The Nomenclature used is explained below:
SDS8 (S2) wire diam::::().19 mm (0.0075") fil. diam.=4.5)lm fil. spacing=270 nm.
SDW24 (SO wire diam=0.06Imm (0.0024") fil. diam.=1.4)lm fil. spacing=212 nm.
Note that, in each case, the hysteresis loss, as measured by the area enclosed by the
loop, is greater for the bare filaments. This effect has been observed before and attributed to
differences in flux pinning at the filament surfaces. 12 Despite this difference in loop area, it
is clear for the larger size of wire that the M(H) curves for clad and bare are roughly
"parallel", that is, the loops do not cross. However, at a sufficiently small strand diameter a
second peak appears in the M(H) loop for the clad sample, causing the loops to cross. This is
clearly seen in Figure 5 for the SDW24 clad sample.
927
3500
8 HT
3000
N
<
E 2500
E
~
..,u 2000
1500
1000
3 4 5 6 7
Strain after last HT
Figure 3. Ie of 0.06 sid material at 5T with 4,6 & 8 HT's showing shift in peak Jc
It has been shown previously that one clear manifestation of proximity effect coupling
of filaments in short samples is an excess magnetization, tending to a sharp peak at low fields,
in the flux trapping branches of the M(H) loop (that is, in the tITst and third quadrants). 13,14
Thus the additional peak observed in the SDW24 clad sample is clear evidence for
interfilamentary coupling in this sample. On the other hand the larger wire shows no such
excess magnetization; this is a clear indication that interfilamentary coupling is not present in
sample SDS8. We may conclude that in these short samples, interfilamentary coupling sets in
for interfilamentary spacings between approximately 270 nm and 212 nm. This is in the same
size range as was found in previous investigations using Cu 0.5 wt. % Mn between the
filaments. 15
CONCLUSIONS
The results show that despite the facts that the material was annealed and 13 quenched
at the 50.4" diameter billet size, the filaments had an 8 % barrier, which was considered part
of the filament, and that the matrix was Cu 0.5 wt. % Mn, excellent Je properties were
obtained. -3000 Nmm 2 at 5T was achieved in -4.0 I-lm diameter filaments by using eight
HTs each separated by a strain of only 0.46. This was in material with an sId of 0.06, which
showed significantly higher Ie's under all heat treatment conditions than the material with the
more usual sId of 0.15.
These results also showed that, as the number of heat treatments increased, the peak Je
occurred at a lower strain after the last heat treatment. This confrrms the results previously
found on 15 filament material. 6
Image analysis showed that these higher Je's are probably due to a smaller spread in
filament area. This technique is assumed to be a very sensitive method of measuring
sausaging. The normal examination of the extracted filaments showed no evidence of
sausaging in either material.
The magnetization measurements showed that, under the optimum condition exhibiting
2980 Nmm2 at 5 T with a spacing of 230 nm, the material had negligible proximity coupling.
Although the small billets did not allow a reliable estimation of the effect of the
thermomechanical treatments on piece length, even the most aggressive HTs appeared to have
little effect on ductility. In future work larger billets will be used to force Ie to higher levels
and to determine the effect of such treatments on piece length.
928
80
B Bare
60 C Clad
,....
(J 40
(J
........
-.
::l
.,
E 20
..,
C
0 0
0
N
-20
cg,
0
::E -40
-60
-80
-20 -15 -10 -5 0 5 10 15 20
Field (kOe)
Figure 4. Comparison of M(H) hysteresis loops for the SDS8 clad and bare samples. No Coupling.
25
20
B Bare
C Clad
,....
(J
15
(J
........ 10
-.
::l
.,
E 5
..,
c 0
0
0
N -5
c
g, -10
0
::E
-15
-20
-25
-20 -15 -10 -5 0 5 10 15 20
Field (kOe)
Figure 5. Comparison of M(H) hysteresis loops for the SDW24 clad and bare samples. Evidence of
coupling.
929
ACKNOWLEDGEMENTS
The authors would like to acknowledge the assistance of: M. Hebert of IGC/ASl's
QA Dept., who performed the Je testing, D. Birdsall of the R&D Dept. who prepared the
specimens and P. Valaris and H. Kanithi of the Eng. Dept. who supplied the monofilamentary
stock from IGC/ASI SBIR Phase II Contract # DE-AC02 89 ER80714. R. Smith and P.
Schumacher also assisted at Battelle. The work was carried out under a Phase I grant from
the Small Business Innovative Research Office for DOE SSe.
References
1. T. S. Kreilick, E. Gregory and J. Wong, "Fine Filamentary NbTi Superconducting
Wires", "Adv. in Cryo. Eng.," A. F. Clark and R. P. Reed, eds., Plenum, New
York, 32, pp.739-.745,1986.
2. E. Gregory, T. S. Kreilick and J. Wong, "Fine filamentary materials for accelerator dipoles
and quadrupoles", in Proc. ICFA Workshop, P. Dahl, ed., Brookhaven National
Laboratory, Report BNL 52006, pp.85-88, 1986.
3. E. Gregory, T. S. Kreilick, A. K. Ghosh and W. B. Sampson, "Importance of spacing in
the development of high current densities in multifilamentary superconductors",
Cryogenics, 27,4, pp.178-182, 1987.
4. T.S.Kreilick and E. Gregory, "Further improvements in current density by reduction of
fJlament spacing in multifilamentary NbTi superconductors", Cryogenics, 27, 7,401,
1987.
5. T. S. Kreilick, E. Gregory, J. Wong, R. M. Scanlan, A. K. Ghosh, W. B. Sampson and
E. W. Collings, "Reduction of coupling in fine filament Cu NbTi composites by the
addition of manganese to the matrix", in: Adv. in Cryo. Eng., A. F. Clark and R. P.
Reed, eds., Plenum, New York, 1988,34,895.
6. E .Gregory, "Recent advances in commercial multifilamentary NbTi wires in the United
States". In: R.P. Reed, Z.S. Xing and E.W. Collings (eds) Cryogenic Materials '88,
Vol. 1. Shenyang International Cryogenic Materials Conference, Boulder, CO, pp.
361-371,1988.
7. P.J. Lee, J.e. McKinnell, and D.e. Larbalestier, "Restricted Novel Heat treatments for
Obtaining High Jc in Nb 46.5 wt. % Ti", Advances in Cryogenic Engineering. eds,
R.P. Reed, and F.R. Fickett, Vol 36A, (1990), Plenum Press, New York, N.Y., pp.
287-294.
8. K.J. Faase, P.J. Lee, J.e. McKinnell and D.C. Larbalestier, "Diffusion Reaction Rates
through the Nb Wrap in SSC Multifilamentary Wire Composites", Paper AX-4
CEC/lCMC Huntsville AL June 1991.
9. Y.E. High, P.J.Lee, J.C. McKinnell and D.e. Larbalestier, "Quantitative Analysis of Nb-
46.5 wt.% Ti Multifilament Sausaging in SSC Composites", Paper FP-14 CEC/lCMC
Huntsville AL June 1991.
10. Li Chengren and D.C. Larbalestier,. "Very High Current Density Niobium Titanium
Composites" IEEE Trans. MAG -23, 2, pp. 1646-1649, 1987.
11. Li Chengren and D.C. Larbalestier, "Development of high critical current densities in
Niobium 46.5 wt.% titanium", Cryogenics, 27,4, pp.171-177, 1987.
12. M.D. Sumption, K.R. Marken Jr. and E.W. Collings, "Hysteretic Surface Effects in
Multifilamentary NbTi Wires Exposed to Transverse Applied Fields", IEEE Trans.
Magn. 27, 2, 2166-2169, (1991).
13. E.W. Collings, K.R. Marken Jr., M.D. Sumption, E. Gregory and T.S. Kreilick,
"Magnetic Studies of Proximity-Effect Coupling in a Very Closely Spaced Fine-
Filament NbTi/CuMn Composite Superconductor", Advances in Cryogenic
Engineering, eds, R.P. Reed, and F.R. Fickett, Vol 36A, (1990), Plenum Press,
New York, N.Y., pp. 231-238.
14. M.D. Sumption, K.R. Marken Jr. and E.W. Collings, "Position and Amplitude of
Proximity Effect Peaks in the Magnetization Curves of NbTi/Cu and NbTi/CuMn
Multifilamentary Strands", IEEE Trans. Magn. 27, 2, 1129-1132, 1991.
15. E. Gregory, T.S. Kreilick, J. Wong, E.W. Collings, K.R. Marken, Jr. R.M. Scanlan &
e.E. Taylor, "A conductor, with uncoupled 2.5 11m diameter filaments, designed for
the outer cable of SSC dipole magnets," IEEE Trans. Magn. 25, 2,1926-1929,
(1989).
930
SUBSIDENCE OF THE PIT SLAB AT SLC
EXPERIMENTAL HALL
Junich Inaba,l Yoichi Himeno,l and Yutaka Katsura2
1 Nuclear Power Division

2 Institute of Technology
Shimizu Corporation
Seavans South, Shibaura 1-2-3, Minatoku, Tokyo 105-07, Japan
INTRODUCTION
Detectors installed at particle accelerator facilities are quite heavy, weighing thousands of
tons. On the other hand, ground subsidence caused by the installation of a detector adversely
affects the beam line alignment of the collider. It becomes, therefore, very important to figure
out the expected amount of ground settlement by means of adequate evaluation methods in
advance.
At Stanford Linear Accelerator Center (SLAC), a 1700 Mf (metric tons) Mark II detector
was replaced with a 4000 Mf SLD detector in Stanford Linear Collider (SLC). The exchange
started in December 1990 and lasted until March 1991, and the amount of ground settlement was
measured by SLAC during that period.
We performed simulation studies to evaluate the subsidence of the pit slab using several
analysis methods. Parameters used for the analyses were decided based on the information of
the SLC structure and the ground conditions at the SLAC area. The objective ofthis study is to
verify the applicability of several simulation methods by comparing the analytical results with
the actual subsidence data obtained by SLAC.
STRUCTURE OF THE FACILITY
Collider Experimental Hall (CEH) of SLC consists of an underground pit containing the
detector and a steel-framed building covering the pit. The layout of the CEH underground pit
is shown in Figure 1.
The pit slab is a 2 ft thick reinforced concrete plate directly placed on the ground, and is
disconnected from the surrounding external walls. Steel floor plates are embedded in the pit slab
to support the weight of detector assembling, moving and installation. Each detector is
supported by four steel columns. Two concrete block walls which shield the Mark II detector
are also on the CEH pit slab and weigh 920 Mf each.

GROUND CONDITIONS AT THE FACILITY
The ground at the CEH consists of a Tm stratum (MIOCENE) and a Te stratum

(EOCENE). The Tm stratum is on top of the Te stratum. The Tm stratum is 400 to 500 ft thick
around the CEH, but the thickness of the Te stratum is not known.
Although the Tm stratum is called "sandstone," severnl test results show that the
characteristics of the stratum are closer to the sand stratum of diluvium. These tests include the
standard penetration test (the result was 50 - 100 blows!ft), the unconfined strength test (4 kg!
cm 2), the tri-axial strength test, and shear wave velocities (300 - 600 m/sec). Thus the Tm
stratum is more appropriately evaluated as soil (diluvial sand stratum).
The tests of the Tm stratum were performed only for the ground above the pit slab, and the
ground data for the Tm and Te stratum deeper than the pit slab were not available. Therefore,
for the simulation analyses of the subsidence, it was assumed that the Tm stratum continues to
the infinite depth which has the same characteristics with the ground above the pit slab.
MEASUREMENT OF PIT FLOOR ELEVATIONS
SLAC measured the subsidence behavior of the CEH pit slab during the exchange of the
detectors.} Measurements started before the shielding block walls were removed, and continued
until the SLD was rolled in. The water level systems and the optical level techniques were
applied for the measurements.
SLAC had also carried out the measurement of the settlement during the roll-in oftheMark
II detector. The same water level systems were also used for this measurement.
H 318 71 018.4 233''() 318

~ .:=...-------~-"'-'--------~
PIER BLOCK (Moll< II) V e.m Une
••1 n , _ ,'12
A.OOR Pl.ATE!" 1- - : '
.1
SUPPORT (SLDI$?= . =It ~

18+
SUPPORT 1_ R)
ft·: :
r T
.10+ I
, ,
::
SHIELDING BlOCK (Mall< II) ) ,I
I
tj ~= t=~ ~
I
I
I
I
I
I
.5., ' III " ..,13
PI.AH (El .16.998.7)
,, ,,
r ---,
,, ,,
V El - 16998.7
SECTION I A - A I
Figure 1. Layout of the CEH underground pit.

+ shows the water level. The level #3 and #16 were placed on the wall.
932
Measuring with Water Level Systems
To measure the elevation change of the entire CEH pit slab, 17 water levels were placed
at various points indicated in Figure 1. Each water level was connected with a water tube and
an air tube.
The observation records showed that the pit slab, as a whole, rebounded after the block
walls were removed and the Mark II detector was rolled out, and that the pit slab subsided after
the SLD rolled in. The records of the wells located far west of the pit slab, however, indicated
that the rebound also occurred during the roll-in of the Sill.
Relative displacements were quite small, 1 mm or less, at most of measuring points except
several points close to the SLD. Since increase of the subsidence was also very small after the
SLD rolled in, the amount of relaxation after the detector transfer is considered to be negligible.
Measuring with Optical Level Techniques
The classical optical level techniques were employed to measure the local deformations
around the SLD supports since water levels were not able to be located very close to the SLD.
Elevations were measured prior to and after the SLD was moved on beam line in order to
estimate the level changes around the SLD.
The amounts of the relative subsidence in the vicinity of the SLD supports were from 4 to
7 mm as shown in Figure 2.1
SIMULATION STUDIES
Study Policy
The CEH pit slab has experienced 7 different load stages during its construction and the
completion of the detector exchange. The load history of the pit slab is shown in Figure 3.
The real subsidence of the CEH pit slab is considered to be closely related to the load to
which the pit slab has been subjected in previous stages. In this study, however, the load history
was not taken into consideration since the ground is nearly equivalent to the sand stratum and
sinking due to consolidation is negligible. Each load stage was, therefore, handled as an
independent phenomenon.
NORTH
111+0.63 111·1.20
-1.55111 111-1.62
32'
-t~l-
_111-4.31
-3.81 - _itn
WEST
EAST
-6. 19
-6.70
1_ _1-2.76
-2.84
-5.10 - - -4.37
111-2.45 111-2.16
111-1.41 -1.42 1iII+O.6

SOUTH
Figure 2. Subsidence around the SLD supports.

Map of CEH floor showing locations of points whose elevations were measured prior to and after
SLD was moved on beam line. Elevation changes are indicated in millimeters. The locations of the
SLD foot shims are also indicated.
933
Following four simulation methods were chosen for the simulation analyses. The results
obtained were compared with the actual subsidence data mentioned before, and the applicability
of each method to future accelerator facilities was evaluated.
a) Theoretical method (approximate calculation)
b) Theoretical method (exact calculation)
c) Two-dimensional finite element method (FEM)
d) Axisymmetric FEM
Parameters for Simulation Analyses
Geomechanical Parameters The modulus of deformation,.Es, can be obtained by the

Equation (1). This equation was proposed by Shimizu Corporation based on past experimental
data. 2
Excavation and construction

I I
Shielding wall blocks
(for Mark II detector)
ij (920 MT)i i(920MT)1
Mark n detector assembly

I
Mark II detector
(1720MT)
~ ~ ~QI
Mark n detector roll-i n Moved on to the control line
(the east shielding wall was temporarily removed)
~1~=====,~~afk~:======lI::::::::I
SLD detector assembly I
r-
SLD detector (3090 MT )
Q~ Q ~
Mark n detector roll-out Removed the east and west shielding walls and
,Qr7~!
SLD detector roll-in
Mark n detector removal
Figure 3. Loading history of the CEH pit slab.
934
ES = (0. 50'v' 100yo' + 0.1). EyS (1)
Where ES: modulus of deformation (kg/cm 2)

EyS: modulus of deformation obtained from the seismic survey (kg/cm 2)
O'y': vertical stress of ground after excavation (kg/cm 2)
O'yO': vertical stress of ground before excavation (kg/cm 2)
Poisson's ratio, v =0.3, was used for the simulation analyses. This value is frequently used
for static analyses of sand stratum. The unit weight, 'Y, was estimated as 2.0 g/cm3 based on the
ground data obtained at the facility.
Structural Parameters The underground pit slab is the only structure which should be
considered in the simulation analyses. Physical properties of the reinforced concrete used for
the pit slab were determined based on the design criteria of the CEH. Concrete strength was fc'
= 4,000 psi, Young's modulus was Be = 3.6 x 1()6 psi, and Poisson's ratio was v = 0.167.
Simulation by Theoretical Method (Approximate Calculation)
As the simplest evaluation method, Equation (2) based on the Steinbrenner's theory was
used for the estimation of the settlement under the SLD supports. The adopted equation was
originally established for the immediate settlement of a foundation on the sand stratum.
(2)
Where SE: immediate settlement (m)

q: mean load of a foundation (t/m2)
A: foundation bed area (m 2)
E: ground modulus of deformation (t/m2)
IlH: coefficient determined by Poisson's ratio, ground thickness, and founda-
tion bed shape (t/m2)
The immediate settlement of each layer Si and the total settlement SE are shown in Figure
4. Since the result corresponds to the mean settlement of a floor plate, the estimated value is
considered to match well with the actual settlements, 4 to 7 mm, shown in Figure 2.
Thus, this method is simple and suitable for predicting the approximate amount of the
settlement. This method, however, cannot take into account the settlements caused by other
supports, nor evaluate ground stresses.
772.5 ton i3090 ton/4)

8,= 3.91mm
10 It
20 It
ttt ttt
E, = 16,300 (tim) v = 0.3 8 2= 1.68mm
20 It E2 = 16,300 (tim) v = 0.3 83= 0.38mm
20 It E3= 39,500 (tIm) v = 0.3
8 4 = 0.28mm
20 It E4 = 47,000 (tim) v = 0.3
8s= 0.25mm
20 It Es= 52,000 (tIm) v = 0.3
8 6= 0.0 mm
20 It E6 = 81,000 (tIm) v = 0.3
Total 8E= 6.50mm
Total 110 It
Figure 4. Conditions and results of the approximate calculation.
935
Simulation by Theoretical Method (Exact Calculation)
To understand the subsidence behavior of the overall CEH pit slab, the distribution of
settlements over the entire pit slab was calculated. The analysis was, as the approximate
calculation, based on the Steinbrenner's theory, and the rigidity of the pit slab was taken into
account.
The same six-layer model was used for the ground as the one used for the approximate
calculation. The pit slab placed on the ground was modeled with beam elements having the
equivalent rigidity as shown in Figure 5.
Stages III - VI in Figure 3 were analyzed in order to compare the analytical results with
the measured settlements. The relative settlements of Stage VI to Stage IV correspond to the
actual measurements obtained by the water levels, and the relative settlements of Stage VI to
Stage V to the optical levels. The results for the Stage VI are shown in Figure 5 as an example.
The maximum relative settlement of Stage VI to Stage V was 6.7 mm immediately below the
SLD support. It corresponds well with the measured settlement, 4 to 7 mm, shown in Figure 2.
Thus, this method is relatively simple compared to finite element methods, and suitable for
evaluating the overall subsidence tendency of the ground. This method, however, cannot
evaluate ground stresses.
Simulation by Two-dimensional FEM
Two-dimensional FEM was applied to the north-south section of the CEH pit slab. The
input load to the model was evaluated as the uniformly distributed load which is the load of a
SLD support divided by the support's projected area.
The result is shown in Figure 6 (a). The maximum settlement obtained by the analysis was
20.4 mm at the pit slab beneath the support. This value is almost three times as large as the
measured settlement, 4 to 7 mm, around the support. This is due to the overestimation of the
input load. Although the actual load was considered as the concentrated load, it was inevitably
modeled with the line load for the two-dimensional FEM, producing much larger load than the
reasonable one.
Thus, the evaluation of the input load is difficult when applying this method to individual-
type supports.
Simulation by Axisymmetric FEM
Axisymmetric FEM was applied to the circular area around one of the SLD supports. The
input load to the model was evaluated in the same way as the two-dimensional FEM.
The result is shown in Figure 6 (b). The maximum settlement obtained by the analysis was
5.4 mm at the pit slab under the support. In the axisymmetric FEM, settlement of any point
BEAM ELEMENT SLD MARK II

/ I I \ ~
1
Loads
(MT)
",2.5 r-.. 7'12.
P......... 4 '--"": f""r-..
t I
-
Symmetrical
~ .1 ~6 3.
Setllements -r-., 7.'f ~ 5.

(mm)
'I
3.2 3.5 2.
I.D 1.0 1.
I.. 233ft BEAM LINE .1
Figure 5. Analysis conditions and results of the exact calculation. (Stage VI: when the SLD roll in)
936
within the analysis area can be obtained according to the distance from the center. Therefore,
the settlements at the other three supports are determined as follows.
Support in the north: 0.6 mm
Support in the east: 0.2 mm
Support at the diagonal: 0.0 mm (out of the analysis area)
Since the analysis was linear, the total settlement including the contributions from other
supports becomes 6.2 mm by adding all the above values. The calculated settlement corresponds
well to the measured settlements, 4 to 7 mm, near the supports shown in Figure 2.
Thus, the axisymmetric FEM provides with good results in a relatively simple manner.
This method is applicable to individual-type supports, and is able to evaluate ground stresses.
CONCLUSIONS
We performed simulation analyses of the subsidence behavior of the pit slab of the Collider
Experimental Hall at Stanford Linear Collider of Stanford Linear Accelerator Center. Four
analysis methods were applied: two theoretical analysis methods (approximate and exact) and
two finite element methods (two-dimensional and axisymmetric). Parameters used for the
analyses were determined based on the ground data and the structural data of the SLC.
(a) 2-d i m. FEM (b) Axisym. FEM
t.. (!!
! ~
ad =:l.n
.: ·:::··.~:·:···.-:'T"
.......... ··T
..... J .. 11.
--
.C.,J....
..:
o I- . .......... ._ ..--_ .. _----
....
0.
0.
<
Approx. 39f t.
Approx. 82ft.
Figure 6. Results of the two-dimensional FEM and the axisymmetric FEM.
937
Table 1. Analytical settlement obtained by each method, and the measured settlement.
Analytical settlement (mm)

Measured
Theoretical Theoretical settlement
Two-
analysis method analysis method Axisymmetric (mm)
dimentional
(approximate (exact FEM
FEM
calculation) calculation)
SLD
6.5 1 7.2 20.4 6.2 4-7
roll-in
MaIkII
roll-in
- 3.5 - - 2-3
IThe result does not include the contributions from other supports.
The maximum settlements obtained by the four simulation methods are summarized in
Table 1. The actual settlement data measured during the exchange of the detectors are also
shown in the table for comparison. From the simulation studies, the characteristics of the four
analysis methods are summarized as follows.
• The theoretical analysis methods (approximate and exact) and the axisymmetric FEM
can estimate the maximum settlement value fairly accurately. On the whole, the
theoretical analysis method (exact) and the axisymmetric FEM are considered to be
more effective than the theoretical analysis method (approximate) .
• When applying the two-dimensional FEM to the individual-type supports, it is
important to evaluate the input loads reasonably.
Based on this study, selecting and applying the appropriate simulation method according
to the conditions of facilities will help estimate the settlements and the ground stresses
accurately and effectively.
ACKNOWLEDGMENTS
It is our pleasure to acknowledge the help of many members of SLAC and Shimizu
Corporation in compiling this report. In particular, we would like to thank Dr. Gerhard E.
Fischer for conducting the subsidence measurements and compiling the data.
REFERENCES
1. G.E. Fischer. Collider Hall Pit Floor Elevations During the Made II and SLD Moves, SLAC-lN-91-4,
(1991).
2. K. Tamaold, Y. Katsura. Settlement of Bearing Strata during Construction (Part 1) - Measurement of
Settlement, (Part 2) - Modulus of Deformation, Summaries of Technical Papers of Annual Meeting
(Architectural Institute of Japan), (1990).
938
DEVELOPMENT OF FINE FILAMENT SUPERCONDUCTING WIRES AND CABLES
FOR ACCELERATORS
H. Ii, R. Takagi, S. Meguro, S. Shiga, K. Susai, I. Inoue,

K. Ogawa, T. Suzuki and M. Ikeda
The Furukawa Electric Company, Ltd., Tokyo, Japan
ABSTRACT
High performance and high uniformity are required for

superconducting wires and cables to be used for recently planned
accelerators such as SSC (Collider and High Energy Booster), RHIC and
LHC. These wires and cables have been developed by Furukawa Electric
to have high critical current densities in spite of fine NbTi filaments
and, at the same time, to have high dimensional, mechanical and
electrical uniformity. On the other hand, realization of the long
piece length of the wire is the one of the most crucial requirement
for the economical process. The problem of the wire breakage
originates from various causes. In this paper, we analyze the problem
from the viewpoint of wire structures such as wire diameter, matrix
ratio, filament diameter, filament spacing and so on, based on our much
experience with fabrication of accelerator wires and other wires having
similar structure.
INTRODUCTION
The recent specifications of the wires for accelerators 0i~2n

require the very f.ine NbTi filament diameter of a few microns
Moreover, they require the definite structural parameters such as
filament spacing and volume fraction of the central copper core or of
the Niobium diffusion barrier thickness around the NbTi filament. These
structural parameters of the wire have the close correlation
with workability during wire processing and, as a r€jult, with the wire
breakage. Al though some efforts have been made to quanti tati vely
analyze the deformation of multifilamentary wires, the optimum condition
have not yet been obtained for such a complicated wire for accelerator
use.
Then we tried to statistically analyze the wire breakage data on
various configurations. In many cases, the cause of the wire breakage
is the internal defect of the wire. Analyses were made paying special
attention to the relation between the wire diameter at the breakage
and the other structural parameters of the wire.

WIRE S1:RUCTUAAL PARJ\METERS
In the past years, we have been fabricating various kinds of wires

for accelerator use. From the viewpoint of analyses here, structural
parameters of relevant wire are summarized in table 1. Although
matix metal is copper for most of the wires for accelerator, some wires
with mixed matrix metals were examined for comparison. These wires
were fabricated for the superconducting power applications and have
similar filament numbers, diameters and spacings. These parameters
are used in the analyses in the next paragraph.
Table 1. Wire structural parameters
Parameter Unit SSC inner SSC outer SSC HEB SSCHEB

inner outer
Wire diameter mm 0.808 0.648 0.808 0.648
Matrix ratio (Cu/NbTi) 1. 3/1. 5 1.8 1.5 1.8
Matrix ratio (CuNi/NbTi) 0 0 0 0
Number of filaments 7,236 4,152 39,601 22,686
Filament diameter urn 6 6 2.5 2.5
Filament spacing urn 1 1 0.5 0.5
Billet stacking hexaqonal hexagonal hexaqonal hexaqonal
Parameter Unit RHIC LHC inner LHC outer
Wire diameter mm 0.648 1. 39 0.800

Matrix ratio (Cu/NbTi) 2.25 1.65 1.8
Matrix ratio (cuNi/NbTi) 0 0 0
Number of filaments 3,552 20,485 6,339
Filament diameter urn 6 6 6
Filament ·spacing urn 1.2 1 1
Billet stacking hexagonal hexagonal round
Parameter Unit MX-1 MX-2 MX-3
Wire diameter mm 0.37 0.580 0.580

Matrix ratio (Cu/NbTi) 1.7 1.5 1.5
Matrix ratio (CuNi/NbTi) 1.5 1.0 1.0
Number of filaments 10,000 8,600 12,222
Filament diameter urn 1.8 3.5 2.8
Filament spacing urn 0.37 1.1 0.9
Billet stacking round round hexagonal
WIRE BREAKAGE ANALYSES
When wire breakage is caused by internal defects whose size does

not change during drawing, important parameters which seriously affect
the frequency are thought to be the NbTi filament diameter and the
filament spacing. Then, the influence of those parameters on the
frequency was studied with Breakage Sensitivity (A) which is defined
by the empirical formula. (1) • Generally, the larger the Breakage
Sensitivity, the higher the breakage frequency.
2
1 ( Do
A LOG (1 )
d x 1 D
d (urn) NbTi filament diameter at final 'dire diameter
940
1 (urn) NbTi filament spacing at final wire diameter
o (mm) Wire diameter at wire breakage
Oo(mm) Final wire diameter
Relations between Breakage Sensitivities at wire breakages and

the breakage frequencies in our recent products are shown in Figure
1 and 2. There are two types of distributions. The one has a peak
of the breakage frequency at a certain value of A. The other shows
monotonically increasing relation between Breakage Sensitivity and the
breakage frequency. The former corresponds to the cases of SSC, RHIC
and MX-3 and the latter to the cases of LHC Outer, MXl and MX-2. The
difference between them originates from multifilament billet assembling
methods. Hexagonal monofilament rods were used for the former
multifilament billet and round ones for the latter. When both the
high packing factor in the multifilament billet assuring the perfect
metal bonding of materials and the prevention of monofilament rods I
crossovers in the multifilament billet are considered, the former design
is better. When the billet assembly is perfect, the cause of wire
breakage is not complicated and i t can be seen clearly as the peak of
the breakage frequency. The main cause is the internal defects whose
size is most sensitive to wire breakage. However the imperfect billet
assembly brought some additional causes of wire breakage such as poor
metal bonding and i t becomes difficult to analyze the cause. In
the products whose matrix material is only OFC (SSC and RHIC), peaks
of the breakage frequency are about minus one. It means that the
internal defects affect most sensitively when Breakage Sensitivity of
wire is minus one and if Breakage Sensitivity of the wire at final size
is designed to be less than minus one, the breakage frequency will
decrease.
In case of the products whose matrix materials are OFC and CuNi,
Breakage S~nsitivities at the peaks of the breakage frequency are larger
than those of OFC matrix products. It suggests that the wires with
the same structure and different matrix materials will lead do different
breakage frequency. The breakage frequency of wire with cuNi matrix
is smaller than that of OFC matrix wire because of higher mechanical
strength of CuNi than OFC.
50
40
'"
~
'H~
30
0* nne r
'"
0'"
20
"' ....
'" '"
"'.., °u
0'0
"''''
........ 10
k.'H
0
-2 -1. 5 -1 5 o . 5
LOG { l / (d 1) (Do/D) 2)
Figure 1. Wire breakage frequency vs wire structural parameter

(Breakage Sensitivity)
941
50 , - - - - - - - - - - - - - - - - - ,
MX-3
OOl
c" 20
Ol:;1
:;1....,
0'0
Ol'"
.... 10
~ .....
-1.5 -1 5 0 . 5
LOG { l / (d 1) (Do/D) 2}
Figure 2. Wire breakage frequency vs wire structural parameter

(Breakage Sensitivity)
SUMMARY
Breakage frequencies of recent superconducting wires with fine

NbTi filaments were studied in relation with the wire structures. The
following matters were cleared out.
(l) There are two types of breakage frequency to Breakage Sensi ti vi ty.
The one has a frequency peak at a certain Breakage Sensitivity
and the other does not have such a peak.
(2) The breakage frequency peaks are observed only on the wires whose
multifilament billets were composed of hexagonal monofilament rods.
And the peak shows that there is only one breakage cause (internal
defects) which affects the breakage frequency at a certain Breakage
Sensitivity.
(3) Wires with OFC matrix have the breakage frequency peaks when their
Breakage Sensitivities are about minus one. It means that the
wire structure design (for example, less than minus one of Breakage
Sensitivity of wire at final size) will decrease the wire breakage
frequency.
(4) In the case of wires whose matrix are OFC and CuNi, Breakage
Sensi ti vi ties at the peaks of the breakage frequency are larger
than those of OFC matrix products. High mechanical strength
of matrix material prevents wire breakage.
REFE:RENCES
1. SSC-Mag-M-4l46, NbTi Superconducting Wire for SSC Dipole Nagnets

(Outer).
2. RHIC-Mag-M-414l, NbTi Superconducting Wire and Cable for RHIC
Dipole and Quadruple Magnets with 8 cm Aperture.
3. B. Avitzure and S. Talbert, "CRITERIA FOR THE PREVENTION OF
FILAMENT AND WIRE BREAKING DURING THE FABRICATION OF
MULTI FILAMENTARY COMPOSITE SUPERCONDUCTING WIRE," IISSC,
SUPERCOLLIDER 3, 341-354, 1991.
942
DEVELOPMENT OF HIPH STRENGTH ALUMINUM-STABILIZED SUPERCONDUCTOR FOR THE SDC DETECTOR
I. Inoue l • T. Suzuki I.M.Ikeda 2.A. Yamamoto S and T.Kondo s
ISuperconducting Wire Development Department. Nikko Reseach

Laboratory.The Furukawa Electric Co.• Ltd.• 500 Kiyotaki.
Tochigi.321-14.Japan
2Superconducting Products Department. The Furukawa Electric Co.
Ltd•• 6-1 Marunouchi 2-chome.Chiyoda-ku.Tokyo.Japan
SNational Laboratory For High Energy Physics. 1-1 Oho.Tsukuba.
Ibaraki.305.Japan
ABSTRACT
Development of high strength aluminum-stabilized superconductor for the SOC

superconductor which has appropriate RRR has been carried out. Requried properties
for aluminum matrix (yield strength at 77K >6.5kgf/mm2 and RRR ~500) have been obtained
by using Al-Zn alloy and cold-working. In the production of 1/3-sized superconductor by
way of experiment. the reduction ratio of cold-working has been restricted below 15%.
Based on investigations of Al alloys and 1/3-sized experiment. full-sized superconductor
for the SDC detector has been designed and produced by way of experiment. The properties
of aluminum matrix obtained satisfy the specification after final curing heat treatment.
INTRODUCTION
Thin solenoid for the SDC detector. for which aluminum-stabilized superconductor
is used. has been designed. [1.2.3.4J This solenoid. having 1. 84m-radius and 8. 8m-length.
will provide the highest field of 2T in thin superconducting solenoid in the world-wide.
In this solenoid. calculated stress intensity in the coil is 5.2kgf/mm2. therefore
aluminum matrix with high mechanical strength is required.
For the development of aluminum-stabilized superconductor for SDC thin solenoid. it

is one of the most important items to improve (or optimize) the mechanical strength of
aluminumu stabilizer in the superconductor with keeping appropriate RRR. Recently.this
effort has been made in the R&D program for superconducting magnets in the particle
astrophysics magnet facility project. ASTROMAG[5. 6J. 4N7 high purity aluminum has been
used in these conductors. and in order to improve the mechanical strength cold
working process has been adopted. By using the same techniques as above mentioned. R&D
of aluminum-stabilized superconductor for the SDC detector has been planned.

Plenum Press, New York. 1992 943
In this paper, developments of aluminum-stabilized superconductor for the SDC
detector particulary properties of aluminum stabilizing materials, cold working process
and production of full-sized conductor by way of experiment are described.
DESIGN OF THIN SUPEROONDUCTING SOLENOID FOR THE SDC DETECTOR

Design parameters for the SOC solenoid are shown in Table 1. [3] The stored energy
of this solenoid is 166MJ at 10,000A and 2T (in the central field). The ElM ratio in
the various superconducting solenoids used in high energy physics are shown in Fig.!. [3]
The ElM ratio defined by [magnetic stored energy]/[cold mass] of the solenoid has been
introduced as a simplified parameter to know the lightness and transparency of the
solenoid. Aluminum-stabilized superconductor with large aspect ratio will be wound in
the edge-wise direction and be cooled indirectly through coil former by LHe.
Table 1. Design parameters for the SOC solenoid.
Dimensions
Cryostat Inner radius 1. 70 m
Outer radius 2.05 m
Half length 4.39 m
Coi I Effective radius 1. 84 m

Half length 4.12-4.18 m
Coi I + OSC thick. 44 + 31 mm
Tracking Aperture radius 1.7m

Apert. half length 4.39 m
Transperency Rad. thickness 1. 2 Xo

Int. length 0.25 AO
Erectrical Parameters
Central Field 2.0 T
Int. B.dl (7}=0) 3.4 T. m
Nominal Current 8,000 A
Inductance 4.6 H
Stored Energy 146 MJ
ElM 7.3 kJ/kg
Mechanical Parameters
Effective cold mass 20 tons
Total weight 25 tons
Mag. radial pressure(@z=O) 1. 6 MN/m2
Axial total compressive force 11 MN
944
10
6
2EUS • • a:F • ALEPH
-
~
w 4
TOPAZ •
• CLEO·II
• H1
-
2
•VENUS -
Stored Energy [MJ]

Figure I. ElM ratio in thin superconducting solenoid used
in high energy physics experiments.
DESIGN OF ALUMINUM-STABILIZED SUPERCONDUCTOR FOR THE SOC DETECTOR

Design parameters of aluminum-stabilized superconductor for the SDC detector are
shown in Table 2. The characteristics of this superconductor are as follows.
(1) Large current capacity (Ic~16.000A at 5T).
(2) Large cross sectional area and large dimensional aspect ratio.
(3) Cable type NbTi-Cu conductor inbedded in aluminum stabilizer.
(4) High yield strength with apropriate residual resistace ratio (RRR).
0.2% yield strength at 77k >6.5kgf/mm2 RRR ~ 500
(5) Long conductor.
unit length >2.000m (1.150kg)
Compacted stranded NbTi-Cu cable has been inbedded in aluminum because of the two
reasons. One of them is requirement for long unit length and another is to achieve high
critical current density of NbTi. Comparison to the conventional aluminum stabilized
superconductor for the detector are shown in Table 2.
DEVELOPEHENT OF ALUMINUM MATRIX WITH HIGH MECHANICAL STRENGTH

Now. several superconducting thin solenoids for the detector are successfully
operated using the aluminum-stabilized superconductor composed of NbTi-Cu composite
with high purity aluminum (e.g. 5N. 4N-upgraded). The stored energy of these solenoids
945
are smaller than that of the SDC solenoid. therefore electrical magnetic force to the
conductor are acceptably small. However that force is much larger in the SOC solenoid.
so that if the mechanical strength of the aluminum matrix is poor defomation of the
superconductor will happen.
Table 2. Design parameters of the aluminum-stabilized superconductor for the

SOC and TOPAZ detector.
AI-stabilized superconductor TOPAZ SDC
Cu/NbT i Insert
type mono I ith cable
dimension thickness mm 1.8 2.2
width mm 3.3 6.4
corner R mm 0.2
Cu/NbTi ratio 1 1
filament diameter ].tm 50 20
twist pitch mm <30 27
RRR of Cu >120 >100
AI-stabilized superconductor
dimension thickness mm 3.59/3.61 4.42/4.32
width mm 18 43.8
corner R mm 0.3 0.4
position of Cu/NbTi central central
Ic kA >7 at 2.4T >16 at 5T

RRR of Al stabi I izer >1000 500
yield strength of Al kg/mm2 >6 at 77K
(expected) >6.5
share strength of AI-SC >1 >2
kg/mm2
To improve the 0.2% yield strengh of the aluminum. two methods are effectiva One
of them is to reduce the purity of aluminum. and other is cold working for size
reduction. RRR and yield strength of pure aluminum is shown in Fig.2. It is clear that
in the usual pure aluminum to obtain satisfactory RRR and yield strength required for
the SOC conductor simultaneously is impossible. Therefore. we have investgated several
kinds of aluminum alloys being added a small quantity of elements.
The elements added in aluminum have been selected on condition as follows.

(1) Base metal is 5N-grade high purity aluminum.
(2) Metals to be added have small !::,.p (increasing ratio of erectrical resistivity)
(3) Metals to be added are soluble.
Therefore. Si. Ag. Cu and Zn have been selected and investigated as additive element in
aluminum. The results of the investigation are as follows.
946
The properties of AI-Si. AI-Cu. AI-Zn and AI-Zn alloy
We have investgated the yield strength and RRR of Al alloys added small quantities
of Si. Cu. Ag and Zn. Test samples have been extruded by the small extrusion press and
and have been drawn. The dependence of the yield and RRR of Al alloy on cold reduction
ratio is shown in Fig.3 in case Si and Zn. As quantities of Si and cold reduction ratio
increase. yield sterength of the alloy is increasing. but RRR decreasing. The
relationship of the yield strength of AI-Si and AI-Zn alloys to the quantities of added
element are shown in Fig.5. The dependence of quantity of added elements on RRR and
5000 0 5N-Al
c:::
•!:::.
4N8-Al
4N3-Al
-.. ---------0-
'--0
c:::
c:::
1000 ~
~
500 ~~
100
::.::
c-
c-
...., 6
co
:5
be
c::
Q)
....,h
en
-0
.......
...,
Q)
>- 0
o 10 20 30
Cold reduction ratio (%)

Figure 2. The dependence of yield strength (77K) and RRR about usual
high purity aluminum on cold reduction ratio.
947
10
Si 119ppm
.,
••
A
Si 106ppm
Si 57ppm
N
8
• Zn 2l2ppm 800
~
'1-0
Yield strength
=
bD
6 600
""
c-
c- --0-_
..., ........ ---0 _ _ _ -S;)
c:::
c:::
CQ c:::
...,
..c::
---~-
be
c
4 400
...,r...
QJ
en - <)-- --<>- - --L>_ - - - 0

"0
-<
.....
QJ
RRR
>-
2 ---n--- A.... _ _ ~ - - - 6 - 200
o o
o 5 10 15 20
Figure & The dependence of the yield strength (77K) and RRR on
the cold reduction ratio for AI-Si and AI-Zn•
... Si 119ppm
800 •• Si 106ppm
Si 57ppm
• Zn 212ppm
600
0::
0::
0::
400
~
200 .to .to • • •
0
0 Z 4 6 8
Yield strength at RT (kgf/mm2)
Figure 4. The relationship between the yield strength (RT) and RRR
for Al-Si and AI-Zn alloys.
948
8
<'>
~
....
"-
6
.-
til
-'G .t.
•
~
f0-
e::
...,
<0
~ 4
•
~
iii
....
...., Zn
en 0 Ag
••
.....
-0
Cu
.....
c:I
2
>- Si
50 100 500 1000

Quantity of element (ppm)
Figure 5. The dependence of the yield strength (RT) on the quantities of
added elements after 10% cold-worked.
1000
500
~~
~
•
a:;
Zn
0 Ag
.A Cu
100
•• Si
50 100 500 1000

Quantity of element (ppm)
Figure ~ The dependence of RRR on the quantities of added elements
after 10% cold-worked.
949
yield strength after cold wokring is shown in Fig. 6. and Fi~7. In case of AI-Zn alloy.
becase of being smaller ~p than other elements. RRR is lower. Based on these data
AI-Zn alloy seems to satisfy the requirements and be more suitable for the aluminum
matrix material for the SDC superconductor than AI-Si. AI-Cu and AI-Ag.
RiD OF CO-EXTRUSION AND COLD-WORK PROCESS

Bonding Quality is very important property for aluminum-stabilized superconductor.
especially for the SOC superconductor. because cold working is indisipensable for the
improvement of mechanical strength. interface of the aluminum matrix and NbTi-Cu must be
well bonded. If bonding Quality between aluminum and NbTi-Cu is poor. then separation
off between aluminum and NbTi-Cu would happen.
Otherwise. because that NbTi-Cu is cabled. it is difficult to cold work the
composite. In order to solve the above problems. we have tried to co-extrude and to draw
the 1/3-sized aluminum-stabilized superconductor similar to the SDC superconducto~ The
bonding quality between the NbTi-Cu and aluminum has been enough to withstand cold
working. Moreover. co-extruded sample have been cold worked by drawing with 10 to 25%
reduction ratio. After drawing. cross sectional views of each samples have been observed.
Fig.7 shows the part of cross-sectional photographs. In the case aluminum-stabilized
superconductor is cold-worked more than 20% reduction ratio. the cross-sectional area of
central part of NbTi-Cu strands have been reduced anomalously. The reasons of this
phenomenon is assumed as follows. If composite is cold-worked over the 15% reduction
ratio. the breakages of diffusion-layer between Al and Cu would happen and consequently.
in the twin strands which is stacked at the position of up and down in the central part
of two-Iayerd cable. anomalous transformations and reducation of cross-sectional area
would occure due to excessive tension which added by cold working.
As the facts above mentioned. it is proposed that the cold reduction ratio of
aluminumu-stabilized superconductor for the SOC should be restricted within 15%.
PRODUCTION TRIAL OF ALMINUM STABILIZED SUPERCONDUCTOR FOR THE SDC DETECTOR
Based on results of investigation mentioned above we have produced several hundreds

meter full-sized aluminum-stabilized superconductor for the SDC detector by way of
experiment. The cross-section of the conductor is shown in Fig.B. This trial has been
carried out by co-extruding and cold-drawing process. The aluminum alloy which contains
about 200ppm Zn has been used. and cold-reduction ratio has been about 12%.
The dependence of mechanical strength and RRR of aluminum matrix on the cold
working reduction ratio are shown in Fig.9 and Fig. 10. Through the curing heat treatment
process (130~x15hr) which is the final stage of the production process. yield strength
has been reduced and RRR increased. It seems to be sure that the targetted specification
values in final stage will be achieved by using AI-Zn alloy and optimizing cold working
reduction ratio.
Aluminum matrix in the superconductor for the SDC detector has been worked on
various production stages from raw material to cooled magnet. Based on this development.
properties of aluminum matrix on each stages are shown in Table 3 and Figure 11.
950
(1) as extruded
(2) after 10.9% reduced
(3) after 23.2% reduced

Figure 7. Typical cross-sectional photograghs of 1/3 sized
aluminum-stabilized superconductor for the SOC detector
as extruded and after cold-drawn.
951
Figure 8. Cross-sectional photograghs of full-sized aluminum-stabilized
superconductor for the SDC detector by experimental production.
RRR
- -~----a.--t:...... ""-
........ 600
'"~ ..... ....... ....-6_
;;::; " ....... _l:l,.
....
bO
-'" 10 --
- --6.-
""
<-
<-
-oJ --b- 400
co
...,
..<::
.-b--~ c:>::
bO c::
c: c::
...,r....
OJ
5
Before cured
CIl After Cured
".....
.......
OJ
200
>- Yield strength
o
o 0 10 20
Figure 9. The dependence of the yield strength (77K) and RRR of aluminum matrix
on the cold reduction ratio for the full-sized SDC conductor.
952
After Cured
600
0:::
§§ 400
Before cured
200
o
o 4 6 8 10
Yield strength at 77K (kgf/mm2)

Figure 10. The relationship between the yield strength (77K) and RRR of aluminum
matrix on the cold reduction ratio for the full-sized SDC conductor.
Table 3. Mechanical strength and RRR of aluminum matrix on production stages
Step at RT(kgf/mm2) at 77K (kgf /mm2) RRR

erB erO.2 erB erO.2
1.5N (based material) 2.3 ....... 3 2.500

2.Alloyed (200ppmZn) 13.6 ....... 15.5 2.6 .......3.6 680
3. Extruded 5.2 2.3 15.7 2.9 630
4. Cold-worked (12%) 6.6 6.5 15.9....... 17 7.5 .......8.9 490
5.Curred 6.3....... 6.6 5.3 .......5.6 15.6....... 16.1 6.5 .......7.3 530
6. Aged at RT(after 6months) same as above
953
5N
800
Alloyed
600 Cured
Extruded
400 Cold-Worked
200
o
o 2 4 6 8 10
Yield strength at 77K (kgf/mm2)

Figure 11. Yield strength (77K) and RRR of aluminum matrix on production stage.
CONCLUSION
(1) Aluminum having high mechanical strength together with appropriate RRR has been
developed. Yield strength was above 6kg/mm2 (at 77K). and RRR was about 500. This
aluminum matrix has been obtained by using AI-Zn alloy and being cold-worked.
(2) In the exPeriment of the 1/3-sized conductor fabrication. it has become clear that
bonding quality between Al and Cu in the composite is very important and even if
good bonding at co-extruding process would have been obtained the cold reduction
ratio of the composite should be restricted below than 15%.
(3) Several hundreds meter full-sized production of aluminum stabilized superconductor
for the SDC detector has been produced by way of experiment. The aluminum raw
material which contains about 200ppm Zn has been used. and cold reduction ratio has
been about 12%. Aluminum properties at each production stages have been investigated.
Through the curing heat treatment stage the properties of aluminum matrix have
reached to the designed values.
(4) This developement will be useful for the production for the SOC R&D magnet programs.
ACKNOWLEDGEMENT
The authors would like to thank for the invaluable assistance and support of many
colleagues.
954
REFERENCES
[1] Expression of Interest by the Solenoidal Detector Collaboration
(SOC) for the SSC experiment.,May, (1990)
[2] Letter of Intent by the Solenoidal Detector Collaboration, Nov. (1990)
[3] A.Yamamoto et al. "DESIGN STUDY OF AN AIR CORE THIN SOLENOID FOR THE SOC DETECTOR"
Proceedings of the International Workshop on Solenoidal Detectors for the SSC
KEK, Tsukuba., Japan, Apri I. (1990) P141
[4] A. Yamamoto "Progress of the SOC Superconducting Magnet R&D" Proc. of the SDC
collaboration meeting at KEK, Tsukuba, May, (1991) P79
[5] A.Yamamoto et al. "AI based superconductor and a design for the ASTROMAG test coil,"
KEK-Astromag-TN-90-1 (Internal) (1990)
[6] Y. Makida et al. "Development of An ASTROMAG Test Coi I with Aluminum Stabi Ii zed
Superconductor." IEEE Trans.-MAG Vol. 27 No.2 (1991) P1994
955
CONNECTIVITY AMONG COMPUTER-AIDED ENGINEERING
METHODS, PROCEDURES, AND TOOLS USED IN DEVELOPING THE
SSC COLLIDER MAGNETS
Nick Kallas and Abdul R. JaUoh

Dallas, TX 75237
ABSTRACT
The accomplishment of functional productivity for the computer aided engineering

(CAE) environment at the magnet engineering department (ME) of the magnet systems
division (MSD) at the Superconducting Super Collider Laboratory (SSCL) involves most of
the basic aspects of information engineering. It is highly desirable to arrive at a software and
hardware topology that offers total, two-way (back and forth), automatic and direct software
and hardware connectivity among computer-aided design and drafting (CADD), analysis
codes, and office automation tools applicable to the disciplines involved. This paper
describes the components, data flow, and practices employed in the development of the CAE
environment from a systems engineering aspect rather than from the analytical angle. When
appropriate, references to case studies are made in order to demonstrate the connectivity of
the techniques used.
INTRODUCTION
More than 10,000 dipole and 2,000 quadrupole superconducting magnets, divided
among approximately 20 different designs, are envisaged for the SSC project. These
magnets will be installed in the collider ring, high energy booster, and interaction region
subsystem over a period of 10 years. After the initial commissioning of the machines, a
magnet failure rate of 2% yearly is probable, which may call for some rudimentary design
enhancements.
Therefore a CAE environment is required that will satisfy the application software
connectivity among CADD and the magnetic, structural, and thermal analysis codes utilized
at the SSe. Also, further integration is required among the SSC codes and the software
platforms used by the industrial collaborators and other national and international
"Operated by the Universities Research Association, Inc., for the U.S. Departtnent of Energy under Contract
No. DE-AC35-89ER40486.
Supercollider 4, Edited by J. NonIe

laboratories. These tasks are essential to the timely design and production of the magnets
since a large amount of analysis runs of mid-sized-3,OOO finite elements and upwards-
models have to be generated where interaction among the magnetic, structural, and thermal
phenomena is present
The design effort of the CAE system is aimed at the seamless integration of the
analyses activities at both ends of the application envelope, i.e., at the input side either from
CADD or other CAE codes and at the output data sets generated by the software to other in-
house and collaborator packages.
CAE HARDWARE PLATFORM
A number of objectives were established at the early stages of the CAE local area
network (LAN) hardware platform development, including the following:
1. All of the latest versions of the selected analySis codes are to run preferably under
a common operating and licensing server system.
2. Mathematical, spreadsheet, publishing, and presentation software of choice that
accepts data from the analysis codes oUght to run on the same platform.
3. The platform should take advantage of distributed processing among loosely
coupled CPUs in a server-client type of architecture where each computing node may be
both a server or a client to another node.
4. Fast and large disks-above 1.0 GByte-are to be used because FEA codes are
input/output intensive and they produce large intermediate and final analysis meso
5. Future growth and upgrades should be accomplished only by CPU and graphic
board upgrades and not by replacement of existing hardware.
6. Ability to migrate to a multi-processing architecture for each node within the
restrictions of Item 5 above should be available since CAE tasks, unlike CADD applications,
are tackled more effectively with multi-processing computers.
7. Color and good graphics capabilities are essential in FEA analysis. Therefore
accelerated graphics adhering to the XII standards and wider monitors are to be used.
8. Each workstation should have both good mips and mflops ratings, since the codes
utilize both integer and floating-point operations.
9. Since software licensing costs are to be kept at a minimum, large CPU multi-user
machines should be avoided because they have higher costs/mips.
10. A Unix! operating system with multi-threaded extensions and pre-emptive multi-
tasking capabilities should be sought that accommodates all required software.
11. Price/performance is not necessarily the driving function for the hardware platform
design, but adherence to standards and user software friendliness that increases functionality
and productivity are of prime consideration.
12. The CAE LAN should integrate well with the rest of the Laboratory following
guidelines established by the computing division and site-specific conceptual design l
requirements.
A hardware topology, as shown in Figure 1, has been designed, procured, and
installed. It is operating and meeting the objectives to a satisfactory degree.
CAE SOFTWARE PLATFORM
A wide variety of software tools is used within this CAE LAN. They can be thought of
as magnetic, structural, thermal and CADD software, all of which need to be integrated as
shown in Figure 2.
958
Figure 1. CAE LAN Hardware Topology.
SSC/MSD/MEICAE Collaborators
+-+ CADD/CAE
SSCIMSD/MEICAE ( Magnetics ) Environment
spreadsheet visualization +---+ Structural )
publishing mathematics functions SSCIMSD/ME
Thermal +---+ CADD
Figure 2. Software Modules Connectivity.
The two main electromagnetics analysis tools utilized for the cold mass design of the
magnets are Pe2d 2 and Tosca,3 which are used for the 2d and 3d field computations in the
optimization of coil turns and wedges, and for multipole field error minimization at the coil
ends, respectively. Also, in-house developed codes such as the field errors and ac losses
code Aharmc7, 12 the 2d coil optimization with Pe2d and Ansys links code Copal,14 and the
3d optimization for the end turns code associated with Tosca, Bend,13 complete the magnetic
analysis loop for the cold mass. The optimization routines of the 2d coils and wedges for a
set of required harmonics in Copal utilize the Ims121 library of numerical algorithms.
The majority of in-house structural cold mass and cryostat analysis is accomplished
with Ansys,4 but Nastran 5 is also used by the collaborators; therefore, it is required to be
able to transfer models from one to the other.
At the thermal analysis side Sinda'85/Fluint,6 Trasys,7 and the Convair thermal
analyzerS (Cta) are being employed to design the multilayer insulation (MLI) system, thermal
shields, helium and nitrogen flow channels, support post, and vacuum vessel of the
cryostat. The augmentation and presentation software for the thermal codes are Itplot9 for
Sinda and Trasys, and Sdrc-Ideas,16 Tmg,ll and Disspla20 with Cta. The specific heats and
enthalpies of the solid metallic and alloy materials at low temperatures, as well as the
properties of the cryogenic fluids from their triple points to high pressure and temperature
are taken from the NBS-based Cryodata8 databases Cppak and Gaspak.
959
The CADD database at MSD is Unigraphics n.1O In order to bridge CADD with CAE
at the database level rather than via the initial graphics exchange specification (Iges) datafile,
UGlPatran is employed. The Patran17 applications interface is utilized in a number of ways:
right after UGlPatran with bridges to Ansys and back-Patans and Anspat-and/or as a
bridge to Sinda and back-Patsin and Sinpat.
Virtually all of the codes mentioned so far also produce ASCn calculation results ftles,
and most of them facilitate the postscript interpretive graphics description hmguage output. It
is then quite simple to incorporate and link the analysis results to the electronic publishing
tool Framemaker,.23 the numeric calculations with graphics and text spreadsheet Wingz,22
and for more sophisticated visualization and mathematical operations PV-Wave 19 and
Mathematica,15 respectively.
CADD TO CAE CONNECTIVITY
Figure 3 refers to the transition from CAE to CADD and back for the MSD/ME
environment and does not include similar transitions between SSC and the collaborators.
These are mentioned in the following sections on the individual analysis disciplines.
CAE
( Magnetics
Structural +--+1 CADD
I
( Thermal
Figure 3. CADD to CAE Linle
As shown in Figure 4 the more detailed description of the steps required in order to
migrate from CADD to CAE and back involve Iges and/or UG/Patran. The most
sophisticated tool here is the UnigraphicslPatran software, which facilitates the translation of
geometric data from the Unigraphics database to Patran, thus bypassing the Iges translation
pitfalls. Patans converts a Patran neutral ftle into an Ansys Prep7 input deck by translating
the design model derived from Unigraphics that may include fInite element data, material
properties, and loading conditions. Ansys results are translated into Patran-compatible ftles
and Ansys fmite-element data in Patran neutral-ftle-format via Anspat.
PAT/ANS
ANS/PAT
Applications
Interface PAT/SIN
SIN/PAT
Figure 4. Unigraphics to CAE Link.
Similarly, Patsin converts fmite-element model data into Sinda finite-difference model
supporting nodal, conductor, and source mnemonic options. The Sinda analysis results are
translated with Sinpat so that they may be post-processed in Patran for viewing on the fmite-
element model geometry. The Sinda results can also be used as a loading condition for
960
thermal stress analysis. Then the Patran applications interface modules supply directly the
Ansys and Sinda pre-processing files. The advantage of using UOIPatran is that work is
done in one single geometric model where Patran operates directly on the UO geometry,
thus providing unified access to the CADD tools.
The implications here are that the design engineer can accomplish first-order analysis
with a high degree of automation in Unigraphics, and the analyst can perform full function
analysis operating directly on UO models, thus enabling a concurrent engineering
environment. Furthermore, the CAE data is stored in the UO database and the model is
displayed in UO geometry; therefore, a full association between CAE and model data is
feasible.
Since Pe2d and Tosca do not provide for an Iges interface, and the Patran application
interfaces do not accommodate this transition, the Sdrc-Ideas fmite element modeling module
has an Iges interface for access to CADD system geometry, and it produces an Sdrc
universal ASCII file.
This file is combined with a Tosca auxiliary file to form a Supertab input file to the
Tosca analysis algorithms. The analysis results can be directed back to the database and/or to
a specified Supertab format universal file, then to CADD via Iges.
MAGNETICS CONNECTIVITY
The electromagnetic connectivity referred to in Figure 5 is largely associated with the

cold mass. The thermal interaction due to heat leaks manifests itself through the sources of
thermally induced structural deformations. The connectivity required here most is between
the structural stresses arising during manufacturing, assembly, and operation of the
magnets.
CAE
CADD
Figure 5. Magnetics Layer.
Figure 6 depicts a typical cold mass analysis methodology. The 2d design iterations
begin with the optimization of coil turns and wedges so that the sum of squares of the
difference between desired and actual harmonics is minimized. These first-cut calculations
are checked for field qUality in Pe2d. Copal automatically produces a Pe2d input file (Comi)
containing the geometry and the boundary conditions. Output from Copal can be directed
both to Ansys and Pe2d. At this stage in Ansys the collar(s), yoke, and shell geometry may
be added, and a full stress and deflection analysis performed. The deflected coils can then be
translated to another Comi file and reiterated through Pe2d. Coil size and location
perturbations due to manufacturing deviations, persistent currents, and ac losses during
ramping are modelled via Aharmc7. Eddy current analysis due to varying magnetic fields
during quench are carried out in Pe2d. When this analysis loop is stable a coil geometry file
via X-graph is supplied to CADD.
961
ENDGEN2
,------ TOSCA
Figure 6. Magnetics Connectivity.
A similar path to the Copal, Pe2d, Aharmc7, and Ansys takes place for the 3d field
calculations, which are primarily done for the end parts of the cold mass. Here the strain
optimization code Bend is used to derive the end turn shapes. The geometry of these is
supplied in the Centroid and Corner files, which are used to generate Unigraphics drawings
for the end parts. The file Corner can be read by Ends3, which adjusts the straight section of
a group of windings between two wedges in order to minimize the end tum harmonics. The
translation program Endgen2 also reads these two files and generates the Tosca geometry.
Precise field calculations, Lorentz forces, and their distributions can then be modelled in
Tosca. The Tosca/Ansys automatic link is not trivial, and if an equitable trade-off between
development effort and utilization is established, then further work is warranted.
STRUCTURAL CONNECTIVITY
The motivation to establish structural to magnetic to thermal connectivities as shown in

Figure 7 is being discussed to some extent in the magnetics and thermal sections. The
structural to thermal CAE connectivity arises from the fact that there is, for instance, thermal
bow due to the asymmetric cooling of the 80-K and 20-K shields.
CAE
CADD
Figure 7. The Structural Layer.
Also, differential thermal contraction between the as-built and operating dimensions
and the impact of slow and fast cooldown rates impose stresses on the suspension and other
cryostat components.
The mechanical deflections of the coils during energization are due to the Lorentz
forces, which react in a horizontal-outward and vertical-inward displacement. These forces
are calculated in Ansys and are passed to Pe2d via the Comi file translator.
Figure 8 shows in more detail the transitions between MSD/ME and collaborator
CADD/CAE systems. In particular it is desirable to migrate model solution results from
Ansys to Nastran via Sdrc-Ideas, which is common on both sides.
962
MSD/ME MSD/ME &
SIN/PAT ANS/PAT
1+----+/ ANSYS Analysis 1+----+/
PAT/SIN PAT/ANS Data Sets
Figure 8. Structural CAE Links.
Ansys models, including the model solution files containing the analysis data sets, are
transferred to the collaborator's Nastran so that post-processing can be done there using the
data loader utility in the fmite-element module of Ideas. The collaborator, after design review
and modifications, may then supply Unigraphics database-compatible CADn files via Iges
that can be directly converted via the Patran applications interface to Ansys and Sinda, thus
closing the SSCL to collaborator technology transfer bridge loop. It is understood that model
transfer from Nastran to Cta is also possible.
THERMAL CONNECTIVITY
The cryostat may be thought of as a stable and simple low heat leak support of the cold
mass. Both static (infrared, support conduction, and interconnect) and dynamic (synchrotron
radiation, splice IR heating, beam microwave, and gas) heat loads are contributing to the
total heat leak budget for the 80-K, 20-K and 4.5-K thermal zones of the magnets. The main
analysis connectivity, as shown in Figure 9, revolves around the thermal radiation, cold
mass support conduction, and residual gas conduction in the cryostat. The MLI system
limits heat leaks from thermal radiation and residual gas conduction. It must span at least
25 years of quality-assured operation in an environment that exhibits irradiation, periods of
poor insulating vacuum, and a large number of thermal cycles. The MLI blankets are
installed on the cold mass and both 20-K and 80-K thermal shields. It has been recognized
from the outset that the Cta code is particularly suited to analyze these residual gas
conduction phenomena.
CAE
CADD
Figure 9. Thennal Layer.
A typical analysis technique to model the residual gas conduction for the MLI, between
the MLI and thermal shields, and MLI and vacuum vessel is shown in Figure 10.
and/or Supertab
Figure 10. CTA Data Flow Connectivity.
963
The Sdrc-Ideas software package is the anchor of this methodology. The model
geometry can be created using Geomod, or Supertab, or preferably can be imported from
CADD via Iges. The solid geometric model may also contain material and physical properties
data. Supertab facilitates the breakup of the Geomod model into structural grid points and
finite elements. After completion of the finite element mesh within Supertab, the thermal
model generator (Tmg) is invoked to read the finite element mesh database, calculate the
element volumes and cross-sectional areas between the elements, and define thermal
boundary conditions.
The thermal analysis database is created by defining additional temperature-dependent
heat transfer coefficients, view factor calculations, and thermal node mergings. This data is
usually generated from within the interfaces/integration section of the finite-element module
by superimposing the Tmg menus onto the regular options.
Tmg is run to calculate the node temperatures of the finite element model by
transforming it into a finite-difference model where elements become thermal nodes and
finite difference capacitances, convective conductances, view factors, and radiative
conductances. Then Tmg calculates the temperatures. Following the generation of the Tmg
files, the Tmg to Convair thermal analyzer-TmglCta-conversion program is invoked; it
reads the Tmg node and conductance data and writes a Cta input file. General Dynamics has
ported both the Tmg/Cta and Cta codes to the MSDIME hardware platform. At this stage, if
necessary, relevant material data from the NIST Cryogenic database can be pasted into the
Cta input file. The results may be viewed in two ways: either directly from the Cta output file
back to the Supertab post-processing utilities or via Disspla. Both methods supply color
contour plots.
The dynamic heat leaks as well as the thermal radiation analysis requirements are
tackled by utilizing the methodology shown in Figure 11. Here the system's improved
numerical differencing analyzer with the fluid integrator is used in order to couple
thermal/fluid problems.
NIST
Cryogenic Data
Figure 11. ConductionIRadiation CAE Connectivity.
Conduction/radiation models with Sinda are simulated in the transient or steady state,
where the energy storage and flow in a system are modeled as a "thermal circuit." Externally
provided fluxes and radiative interchange factors are used.
The fluid system simulator solves for the transient and/or steady state mass, energy
storage, and flow in a system modeled as a "fluid circuit." The purpose of the fluid system
co analyzer is to provide the same flexible modeling power for internal fluid systems that
Sinda'85 provides for thermal structures. The approach is that fluid systems are analyzed as
networks of basic elements analogous to but different from thermal elements. Sinda'85
models may consist of fluid circuits and/or thermal circuits, and all active circuits are
analyzed simultaneously.
964
The radiation environment calculations performed with Trasys determine radiation
conductors between surfaces and heat fluxes on surfaces from direct or reflected sources.
The output is radiation conductor blocks of various shapes. As shown in Figure 11, Sinda
may be fed Trasys models with minor changes. The vacuum vessel to MLI radiation is
modeled this way, and usually the axial conduction component is used rather than
circumferential or radial temperature gradients. Heat is transferred to the vacuum vessel
through a resistance due to free convention, thus allowing the detection of cold spots. For
the support posts the primary heat transfer path is conduction through the tubes. Parts of the
NIST cryogenic database can be incorporated for material and gas properties to both Trasys
and Sinda input files. Trasys also calculates radiation environments for the Cta code. The
input and/or output math models for Trasys are pre- and post-processed with Itplot, a stand-
alone code whose internal parameters can be accessed via a configuration file.
CONCLUSIONS
The CAE connectivity at MSD as described in this paper has existed in some form for
about 18 months. It has been demonstrated that by utilizing a distributed hardware topology
and a number of finite element/difference analysis codes, a high degree of flexibility has
been achieved. Further, all of the CAE tasks that face the MSD/ME department in the
future----either in-house or with collaborators~an be tackled with confidence because the
abundance and strength of the CAE tools and methods cover the majority of analysis
scenarios that may be encountered. The design, development, procurement, and installation
stages are complete. The environment is stable, having met the functional productivity
objectives given at the outset by the successful integration of a number of hardware and
software tools and techniques.
FURTHER DEVELOPMENT
Faster and more accurate analysis techniques are always in demand. In addition, more
extensive coupling of the analysis sets databases among the various packages from different
vendors via some exchange standard analogous to Iges for CADD will speed up the iteration
loops from one to the other. The loop from Ansys to Tosca and back can also be improved.
Work in the years ahead for hardware and software tools is associated more with
system administration, maintenance to maximize LAN uptime, and software and hardware
updates. The more important tasks are in the area of utilization of this CAE platform to
analyze, design, and troubleshoot the superconducting magnets of the SSC project
throughout its lifetime.
REFERENCES
1. J.R. Sanford and D.M. Matthews, "Site-Specific Conceptual Design of the Superconducting Super
Collider," SSCL, Dallas (1990).
2. Vector Fields Ltd., Pe2d package for electromagnetic analysis, Kidlington (1982).
3. Vector Fields Ltd., Tosca, 3d electromagnetics code, KidJington (1984).
4. R. Jayalmmar et al., Conference on Magnet Technology, MT-12, Leningrad (1991).
5. Swanson Analysis Systems, Inc., Ansys, Houston, Pennsylvania (1989).
6. N. Kallas et al., "A program linking Pe2d and Ansys for SSC magnet design" (1991).
7. J.M. Cook, IEEE Trans. on Magnetics, Vol. 27, No.2 (1976).
8. Imsl Inc., Imsl, math, stat, and special functions Fortran libraries, Houston (1991).
9. McNeal·Schwendler Corp., MSClNastran, FEA program, Los Angeles (1990).
10. Martin Marietta Space Systems, Sinda'85IFluint, Denver (1985).
11. Analytix Corp., Trasys, thermal radiation analysis system, Timonium (1990).
965
12. General Dynamics, "Convair thermal analyzer computer program P4560F," engineering manual and
user's guide, Report no. GDSS-SP-85-012, San Diego (1991).
13. Analytix Corp., Itplot, interactive Trasys plot system, Timonium (1990).
14. Structural Dynamics Research Corp., I-deas for Engineering, Milford (1991).
15. Maya Heat Transfer Technologies, Tmg, thermal model generator, Ontario (1991).
16. Computer Associates International, Inc., Disspla, graphics library, San Diego (1991).
17. Cryodata Inc., Cppak and Gaspak, properties of materials and fluids, Niwot (1985).
18. McDonnell Douglas, Unigraphics, 3d CAE/CAD/CAM database, Hazelwood (1991).
19. PDA Engineering, Pattan application interfaces, Costa Mesa (1991).
20. Frame Technology Corp., Framemaker, publishing software, San Jose (1991).
21. Informix Software Inc., Wingz spreadsheet software package, Menlo Park (1991).
22. Precision Visuals Inc., PV-Wave, data display and analysis software, Boulder (1991).
23. Wolfram Research Inc., Mathematica, Champaign (1991).
966
CONTROL AND DATA ACQUISITION SYSTEM
FOR THE B&W VERTICAL DEWAR FACILITY
A. Billingsly, J. Beoddy, B. Cantor,

P. Hlasnicek, J. P. Kelley, G. Pare', and C. Rey
Babcock & Wilcox
P.O.Box785
Lynchburg, Va. 24505
ABSTRACT
The hardware components and software architecture of Babcock & Wilcox's

(B&W) Vertical Dewar Facility are presented. The major areas of interest are the control
system, the data acquisition system, the quench detection and protection system, and
the magnet power supply. Each area is discussed in detail.
INTRODUCTION
The intent of B&W's Vertical Dewar Facility (VDF) is to quickly bring on line a
facility to perform prototype tests of Collider Quadrupole Magnet (CQM) cold masses.
The commissioning of the VDF will be at least a year prior to the commissioning of the
B&W Magnet Test Facility, which will have the capability to test cryostated magnets
and will eventually incorporate the VDF. The major components of the VDP (Figure
1) include an 8.84 m (29') tall dewar for testing vertically mounted cold masses and
collared coils, a vacuum pumping system to reduce the pressure over the helium in the
dewar to permit operation at temperatures below 4.2 K, an 8,000 Amp power supply,
a dump resistor and switch for magnet quench protection and helium conservation,
lifting fixtures for magnet insertion, a helium supply dewar, a chilled water system,
and a data acquisition and control system. This paper will only be concerned with the
data acquisition and control system.
Present plans include an initial test of a full size quadrupole magnet built and
previously tested by Lawrence Berkley Laboratory (LBL). This will allow B&W to gain
experience in handling a full size magnet and, if the magnet is unaffected by shipping,
will permit data comparison with LBL obtained results to confirm B&W equipment
setup and format.
B&W will parallel the work of the SSCL's Magnet Test Laboratory (MTL) to reduce
development time and cost as well as ensure consistency of test results and data
compatibility. Therefore, the data acquisition and control system will be comprised of
hardware and software such that integration of the test results with the SSCL will be
as seamless and as consistent in form and content as possible.

B&W VERTICAL DEWAR
FACILITY FO COM9§
OVERHEAD BRIDGE CRANE
COLD MASS LIFTING FIXTURE
COLD MASS
MAIN FLOOR LEVEL
COLD MASS UPENDING FIXTURE
QUENCH PROTECTION
UNIT
BASEMENT
FLOOR
POWER SUPPLY
Figure 1. Vertical Dewar Facility
968
CONTROL SYSTEM
The control system will be comprised of three Sun SPARCstation IPX workstations
(Figure 2). The workstations will utilize the Sun Operating System (SUN aS) in the
UNIX environment. B&W has chosen Sun workstations as the development and
control computers for the dewar facility to be compatible with the SSCL's MTL software
development team. The Sun workstations will be used for development of the magnet
test process control and data acquisition. When the development is complete, a
workstation will be used as an operator interface for monitoring and controlling the
process (i.e. temperatures, mechanical strain, power supply data, and quench data)
and analyzing data from the dewar facility. Software packages for data analysis and
graphic user interfaces (gui) are now being evaluated. Test data will be available to
the operator interface from the VME crate via local area network (Ian).
The file server for B&W's control system will use a Sun workstation configured as
a SPARCserver IPX. The file server will include a magnetic tape storage device,
compact disk reader and a laserjet printer. The file server will be connected to the
workstation and the VME via the Ian. Test data will be collected by the file server and
stored using the Sybase data base. The SSCL Magnet Systems uses Sybase for
interfacing data from the various magnet test facilities and laboratories. Therefore,
using Sybase increases B&W's compatability with the SSCL. Test data storage will be
on magnetic tape and archived for future analysis by B&W or the SSCL.
Test Facility Control System
File Server I Operator Operator

Operator Station Station Station
[OJ [OJ [OJ

[ Printer
SUN IPX J I SUN IPX J I SUN IPX J
qa
Ethernet
l~
I.
-
"
Magnetic
Test
Cart , Field
Detection
Equipment
VME VME VME
System System System
Figure 2. Test facility control system.
969
The software will be designed and analyzed using Computer Aided Software
Engineering (CASE) tools. The CASE tool will create an environment which will allow
the engineer to design and analyze the real time system in small components. The
system is first broken down into the control software and externals (equipment which
will communicate with the system software). The externals communicate with the
software via data and control lines. The software is then broken down into the
processes which will communicate with and control the externals. The benefits of using
CASE tools are:
• The final product provides a source of software analysis and design
documentation.
• Modifications and changes are more easily implemented.
• Software can be debugged in a more analytical process.
• CASE allows the engineer to analyze the project without regard to the actual
hardware which will be used.
DATA ACQUISITION
The data acquisition architecture (Figure 3) will operate on the Versa Module
Eurocard (VME) bus. A 68030 microprocessor (MVME 147) utilizing VxWorksTM, a
real-time multitasking operating system, will be the data acquisition controller.
Located on the VME will be a General Purpose Interface Bus (GPIB) interface card, a
Programmable Logic Controller (PLC), and an interface to the voltage tap isolation
circuit.
IOATA ACQUISITION I SUN

Workstation
Anunciator
Panel
Ethernet
VME Crate
MVME PLC-5 GPIB- Voltage

147S 1014 Tap
Isolation
Interface
+ t +
+
Cryogenic:
~~ 1 !
Quench
Temp. Strain Hall Magnet Quench
Pressure, Temp, 4B P.S. Detection
3 Probe Protection
Flow, HE bOiloff, sensors sensors 1 Control
Vacuum, Voltage sensor
JUt
Figure 3. Data acquisition system.
970
A communication path is established from the workstations to the MVME 147 via
Ethernet. The MVME 147 then communicates with the GPIB, PLC, and voltage tap
interface over the VME bus. The purpose of the PLC is to control the cryogenics flow,
vacuum, temperature, etc. The GPIB interface card will communicate commands from
the MVME 147 to the strain gage, Hall probe, and dewar temperature sensor
instrumentation and retrieve data from the instrumentation to pass back to the MVME
147. The interface to the voltage tap isolation circuit will store data collected from the
quench detection circuits (buffer amplifiers) acquired during a quench.
Data will be collected at two different rates, slow and fast. The slow rate will range
up to 1 KHz and the fast rate will be at 5KHz. The voltage tap isolation circuit will be
the only circuit collecting data at the fast rate with all other circuits collecting data at
the slow rate. Test data will be dynamically stored in the VME system until the host
or file server sends a request for the information.
The PLC is designed to monitor and control multiple input/output (I/O) points
and is well suited to monitor and control the flow, vacuum, pressure, voltage
monitoring, and boil-off of the cryogenics. The VME PLC operates independently of
the MVME 147 but responds to reads and writes for data and programming from the
MVME 147. By using the PLC we can consolidate control of the various parts of the
cryogenics (flow, pressure, level, etc.) and reduce the number of cable runs between
the VME system and the control room. The fact that the VME PLC can communicate
with other PLCs will simplify communications with the magnet power supply, which
comes supplied with a PLC of its own for internal control. A separate cable run will
be made from the PLC to an annunciator panel to give a visual and possibly audible
indication of a limit or alarm situation within the cryogenics.
The magnetic field measurement system (Warm Finger) is equipment that will be
supplied and calibrated by the SSCL. It will be a stand alone system that will be able
to communicate with the workstation via Ethernet.
Table 1 is a list of the different types of sensors, the number of sensors we anticipate
as well as other information.
Table 1. Measurement types, number, range and other information.

Type Number Range Other
Temperature 3 1.4 - 325K In Magnetic Field (slow rate)
Hall 1 1.5 -370K & Transverse (slow rate)
-150 to +150 k
Gau~s
Strain 48(max) to 6000 psi Accuracy up to 30 psi (slow rate)
Pressure 2 0-50 psia Dewar internal(slow rate)
Flow 2 up to 425 SLM Mass flow (slow rate)
Liquid Level 2 one 46" long At top
one 20' long 2" off the bottom to 20' mark
(slow rate)
Voltage Taps 64 max 10 micro Volts (fast rate)
to 1000 Volts
971
QUENCH DETECTION AND PROTECTION
Quench Detection Circuit
The quench detection circuit will be capable of detecting a magnet quench and
determining, through the use of software at the operator interface, the location of the
quench in the magnet. Additionally, the detection circuit will be able to initiate the
proper response to the quench, such as firing the quench protection circuit and
signaling the host computer of the quench condition.
The detection circuitry will have sensitivity from 10 micro volts up to the
determined quench threshold and have circuit protection to 1000 volts. The sample
rate for data acquisition will be 5KHz with an upgrade capability to 10 KHz. The
threshold sensitivity for quench detection will be selectable up to a determined limit.
After the quench is detected an adjustable time delay of several milli seconds is initiated
prior to firing the quench detection circuit. This delay permits the study of quench
propagation and the determination of quench location. The delay will give the memory
buffer a picture of the voltage tap readings before and after a quench, the amount of
information after the quench being determined by the time delay.
Quench Protection Circuit
The quench protection circuit will use spot heaters and be capable of inducing a
sufficient amount of heat into the magnet coils to cause the magnet to return to a normal
conducting state upon detection of a quench. The quench protection circuit will also
be capable of initiating a quench, for testing purposes, when requested by the operator
at the operator interface.
Power Dissipation (Dump) Circuits
The power dissipation circuit will be capable of absorbing the stored energy of the
magnet during a quench (150 KJ at 6500 Amps or 225 KJ at 8000 Amps) in a timely
manner so that there will be no damage to the magnet, power supply, or accompanying
circuitry as well as reducing the helium losses. The dump circuit is similar to the circuit
built by A.T.Visserl. The dump circuit communicates with the quench detection
circuitry via one control line (i.e. one fire signal). The dump circuit also has automatic
safety features that cause it to fire if a fault condition is detected. Two of these fault
conditions are a low voltage fault on the switching power supply and the opening of
a power supply panel during operations.
MAGNET POWER SUPPLY
Interface Requirements
The magnet power supply will have an interface capable of communicating

between the supply's resident PLC (SLC-500) and the VME PLC. The host computer
will download various ramping programs consisting of a current ramp rate per second
and a maximum current to be obtained. This will give the capability of ramping to a
specified current level and holding until a decision is made to continue ramping up or
to ramp down. The interface device will also periodically report the status of the power
supply. The power supply is also locked to the power dissipation circuit in the event
972
that there is a solid state component failure or if the voltage rating of the switch is
exceeded because of a high dump resistance selection.
CONCLUSION
The Vertical Dewar Facility will have a data acquisition and control system that
will be capable of controlling the cryogenic system, collecting data from the various
components at various rates and generating reports as required by the operator for
internal use as well as reports for the SSCL.
REFERENCES
1. A. T.Visser in: "Design Note of a 10,000A, 1,OOOVdc Solid State Dump Switch for the
Magnet Test Facility" : TM-1692 Fermi National Accelerator Laboratory (October 1990).
973
CURRENT OSCILLATIONS IN SUPERCONDUCTING WINDINGS
A. Gurevich and S. Kopylov
Applied Superconductivity Center

University of Wisconsin-Madison
Madison, WI 53706
INTRODUCTION
Practical use of superconductivity in high-current magnetic systems, motors, generators,

magnetic energy storages, etc. is often complicated by a problem of cryostatic stability of
current-carrying elements with respect to their transition into normal state due to nucleation
and propagation of normal zone caused by thermal perturbations. Criteria of cryostatic
stability as well as a steady - state thermal propagation of normal zone along a
superconducting wire carrying a ftxed current I were studied by many authors 1.2.3. However,
the regime of ftxed I is an idealization of real electric circuits usually consisting of several
superconducting windings having signiftcant inductances. In this case the appearance of
normal zone in some part of a superconducting winding is accompanied by a redistribution
of currents over its different sections or by a current sharing between the winding and normal
shunt. This can result in complicated dynamic regimes, in particular, self-sustained
oscillations of normal zone which have been observed in different superconducting systems2.3 •
Therefore a superconducting system connected to a d.c. power supply can be an active
element of electric circuit generating current oscillations of large amplitude. In this paper a
numerical study of such oscillations have been carried out within the framework of an
approach proposed in 3.
MODEL
We consider the normal zone oscillations in a superconducting coil of inductance L

shunted by a resistor of resistance r. (Figure 1). The circuit is assumed to be connected to a
d.c. constant current power supply 10 , In this case the length of normal zone D(t) and the
current I(t) in the superconductor are described by the following quasi-stationary equations
d1
L tit + (r+Dp)/ = rIo (1)
dD
di =2v(l). (2)

Figure 1. Scheme of the electric circuit.
where v(I) is a velocity of normal zone propagation and p is a normal resistance per unit
length of a superconducting wire.
The oscillations described in detail in 3 can be outlined as follows. If the current 10
exceeds the critical current of the superconductor I", a normal zone appears in the coil and
the coil current begins to decrease transferring into the shunt r. After a time, dependencies
on the parameters of the circuit, the coil current I became less than Ie and the
superconducting state is restored and the coil current will increase until I again exceeds Ie.
A relaxation type of oscillation will occur with a time constant Llr and a period T given by3
L 10 (3)
T=-In--
r 10 - Ie
For the further computer modeling we use a model in which all thermal and electric
parameters of the superconductor except for I" are assumed to be independent of temperature.
In this case the normal zone propagation velocity v(i) is described by the following
formulas2•3.4:
(4)
(5)
Here 8r = l-i/2, i = III.:, a is the Stekly parameter, Vb is a characteristic thermal velocity, C

is the heat capacity, X is the thermal conductivity, Pb is the electric resistivity, P is the
perimeter and A is the area of the cross-section of the superconducting wire, h is a heat
transfer coefficient to the coolant having temperature To, and Te is the critical temperature
of the superconductor. In this case equations (1), (2) become
di (1 ). .
- (6)
dT:+ +8'=',
0
p dg = 1-2~ (7)
dT: ..j~(1-~)
where the dimensionless currents i, io' time t, length of the normal zone g(t) and parameters
f3 and ~ are given by
(8)
976
(9)
As equations (6), (7) do not depend explicitly on t, the set (6), (7) can be reduced to one
first- order differential equation for phase trajectories g = g(i) :
(10)
The nonnal zone appears when the current through the superconductor I equals r." hence
the initial conditions to equations (6), (7) are as follows: g(O) = 0, i(O) = 1. For computer
simulation the values of the parameters in (6), (7) have been taken to be a. = 100, V h =103m/s,
p = 2 mQ/m 1,2,3, therefore f3 = 0.25 r [Q]IL [H]. It will be shown that the nonnal zone
oscillations occur provided that f3 S; -Va..
NUMERICAL RESULTS
Figure 2 shows the dependence g(i) calculated by (10) for different f3 and io. The curves
g(i) are highly asymmetrical and end at some value I = Ij-Va. corresponding to the minimum
current at which the nonnal zone can still exist. Shown in Figure 3 is a similar dependence
of the voltage across the nonnal zone U = pD(I)I as a function of I. Figure 4,5 show time
dynamics of D(t) and U(t). Likewise to Figure 2 the curves D(t) end at some time t when the
current I(t) drops below ~ and the normal zone disappears. Results presented in Figure 2-5
describe the dynamics of periodic switches of the superconductor into nonnal state when the
normal zone originates, expands and collapses during a time 1:" - L-Vf3/ra,1/4 which proves to
be much shorter than the period of oscillations given by (3) if f3 < 1.
Shown in Figure 6 are dependencies of maximum length of nonnal zone Dm and voltage
Um across the superconductor as a function of 10 , The dependencies Um(Io) and Dm(Io) have
been found to be linear, in accordance to both analytical fonnulas obtained in 5 and
200.00
9
150.00
100.00
50.00
0.00
0.00 0.40 0.80 1.20
Figure 2. Dependencies of the dimensionless normal zone length g(i) upon i for different f3
and io: f3=0.001, io = ik (curve 1); f3=O.OOI, io=1.01 (curve 2); f3=0.01, io = ik (curve 3); f3=0.1,
io = ik (curve 4).
977
100.00
80.00
60.00
40.00
20.00
0.00
0.00 0.40 0.80 1.20
Figure 3. Dependencies of the dimensionless voltage across the nonnal zone u(t) = g(t)i(t)
upon i for different /3 and io: /3=0.001, ~ = it (curve 1); /3=0.001, ~=1.01 (curve 2); /3=0.01,
io = it (curve 3); /3=0.1, io = it (curve 4).
experimented data observed on Nb-Ti composite 5. Notice that the lines Dm(Io) and Um(Ia) end
at some current It above which the oscillations disappear. Analysis of our numerical data
indicate that the current it can be described by the approximation it = 1.42/.J/3a1l4, which
gives It in the fonn
(11)
200.00 . . , - - - - - - - - - - - - - - - - ,
9
150.00
100.00
3
50.00
4
tit 1
a .00 -Fr."-"r-.-rrr."-""",rrr,,.,,-,,T"ri
0.00 0.40 0.80 1.20
Figure 4. Time evolution of the nonnal zone length g(t) for different /3 and io: /3=0.001,
io = it (curve 1); /3=0.001, io=l.Ol (curve 2); /3=0.01, ~ = it (curve 3); /3=0.1, io = ik
(curve 4). Here tl = L.J/3/r.
978
100.00
80.00
60.00
40.00
20.00
tit 1
0.00
0.00 0.40 0.80 1.20
Figure 5. Time evolution of the voltage across the normal zone u(t) = g(t)i(t) for different
/3 and io: /3=0.001, io = ik (curve 1); /3=0.001, io=1.01 (curve 2); /3=0.01, ~ = ik (curve 3);
/3=0.1, io = ik (curve 4). Here t1 = L....J/3/r.
This formula coincides with qualitative estimation of Ik given in 3 with an accuracy to the
numerical coefficient. Thus the oscillations occurs within the current interval Ie < 10 < Ik.
As follows from 3, the period of oscillations f(~) decreases with 10 and reaches the
minimum value Tm at 10 = Ik with Ik given by (11). Since the current Ik at /3 < 1 turns out to
be much larger than ~, the value Tm can be found by expansion of (3) in power series in
Ijlk < 1, which yields
120.00
Om/Do
80.00
Um/Uo
40.00
lo/Ie
0.00
1.00 11.00 21.00
Figure 6. Maximum values of u(t) and g(t) as functions of io obtained by means of curves
similar to those shown in Figures 3,4 for different /3. Here Do = 2r/p....J/3 and U o = r~....J/3.
979
(12)
In this case the period Tm is independent of r. For Vb = 103 mis, a. = 100 and p = 2 mQ/m
fonnula (12) gives Tm [s] "" 0.1 0 12 [H].
Thus the superconducting winding plays the role of a non-linear electric element which
periodically switches from superconducting to nonnal states and vice versa. This results in
a relaxation self- oscillations of the current I.(t) through the shunt, the current decaying
exponentially
I (t)
IS
=I e-I / T
0 '
(13)
within each period of the oscillations (0 < t < T), where t., = Ur. Here the time t = 0
corresponds to the end of normal zone collapse when a characteristic current through the
superconductor I - Ij-Va. is well below 10 at a. :> 1. At t = T the current in the superconductor
reaches Ie' which results in the appearance of the normal zone and the reiteration of the
process.
Notice that at ~ :> Ie the applied electric power mostly dissipates at the shunt. In order
to show that, we consider a ratio T\ of the mean electric power dissipated in the shunt within
the period of oscillations to the maximum power rI02:
1] =- 1 fT lIS2 (t) dt , (14)

TI2 0
o
where I.(t) is given by (13). Performing the integration in (14) and using (3), one finds
2io -1
1] = . (15)
I
2i2ln __ 0_
o t-1
o
The efficiency T\(lo) monotonously increases from 0 to 1 as the current io increases from
1 to At io = IJlc :> 1 formula (15) reduces to
00.
(16)
The coefficient T\(~) reaches the maximum value at 10 = Ik, where Ik is given by (11).
In conclusion, we have studied numerically the dynamics of normal zone oscillations
in superconducting windings and the region of the parameters, when such a regime exists has
been found.
REFERENCES
1. M.N. Wilson."Superconducting magnets," Publisher, Oxford University Press(1983).

2. V.A. Altov, V.B. Zenkevitch, M.G. Kremlev, and V.V. Sytchev."Stabilization of
Superconducting Magnetic Systems," Publisher, Plenum(1977).
3. A.V. Gurevich, and R.G. Mints. Rev. Mod. Phys.,S9, 941(1987).
4. L. Dresner. IEEE Trans. on Magn.,IS, 328(1978).
5. V.P. Baev, A.V. Gurevich, R.G. Mints, and M.S. Ushomirsky. IEEE Trans. on Magn.,19,
236(1982).
980
EVALUATION OF TECHNOLOGY TRANSFER ON COLLIDER
QUADRUPOLE MANUFACTURE AT LBL
J. Boer 1 , H. Fechte1er 1 , H. Moryson 1 , F. Sommer 1 ,

H. Gruneberg 2 , R. Kreutz 2 , D. Krische1 2 ,
W. Bensiek 3 , and B. Ryan 3
lSiernens KWU Mu1heirn, FRG

2Siernens KWU Bergisch G1adbach, FRG
3Babcock and Wilcox Lynchburg, VA
SUMMARY
As part of the contract on the collider quadrupole
magnets a technology transfer to Siemens Power
Generation Group (KWU) was performed at Lawrence Berkeley
Laboratory. Berkeley in September 1991. One inner and
outer 1 m long coil each should be manufactured under the
surveillance of LBL staff to become familiar with the
coil production facilities available at LBL. In addition.
KWU had the possibility to observe the production process
of 5 m quadrupole coils. The work eas successfully
completed and provided additional information for the
further hardware operations at the Siemens site.
WINDING OF 1 M COILS
The winding mandrel is prepared with an appropriate
parting compound to avoid sticking to the mandrel. In
both head zones. one piece of separating film is placed
on the mandrel. A self-adhesive Prepreg-fleece is applied
to this film. being cut to template. It is again
protected by a separating film for winding.
The cable consists of 30 fine strands being transposed

anti-clockwise. The insulation is wound clockwise to the
conductor, contrary to the conductor transposition. One
layer Kapton sheet, 50% overlapped and one layer prepreg
tape gap-free. The conductor end is preformed in a
clamping fixture and placed on the winding mandrel. 50 to
70 mm behind the radii, on each side Al-profiles hold
down the conductor when winding.
The first turn is wound with ca. 70 Ibs preload, the
remaining turns with 50 Ibs. When winding the first turn,
the conductor should be severely held down to get a
radius without nearly any gap. This procedure is the same
when winding the second and third turn. Al-round profiles
with teflon are used for blocking with hard-plastic
hammers. The winding mandrel is axially rotatable to
assure proper guidance of the conductor in the radius
areas. The conductor is slightly turned and passe~ below
the Al-profile to be pressed into the correct pos i tion.
At the radii the conductor is wound gap-free with one-
s idedly res in-coated Kapton sheet; coating at the
outside. To prevent offset when winding the next turns,
the conductor is pressed radially to the vertex of the
radius. The thrust bar used is always placed on the last
turn or the spacer.
The compensating and end spacers are also pretreated
with epoxy resin at the conductor contact faces to assure
better adhesion. The spacers in the head zones are
inserted with a protective film over the tips to avoid
damage to the conductor insulation; these films are
removed after positioning.
The Cu-wedges are produced oversized; the insulation
consists of 2 layers Kapton sheet. The bottom layer is
wound with a gap of ca. 1 mm and with the second layer
abutting. One end is closed by insulation, the other is
cut to length with pliers (no chips) after adapting to
the coil. The cutting faces are rounded with a file and
reinsulated with Kapton sheet. For installation purposes,
the Cu-wedges have a continuous axial groove at the top
edge; this should face the coil inner side. Having wound
the coil, the conductor end is clamped to avoid
982
loosening. The conductor is cut off about 1.50 m from the
coil end.
PREPARATION FOR CURING
Plates are bolted laterally to the winding mandrel.

Between plates and coil, segments are inserted being held
by taut strips. Following this procedure, the coil is
pressed to the mandrel by hammering. The taut strips are
retightened again. After having removed the holding down
devices, the complete mandrel with coil and segment
spacer is wrapped in a separating film, removing the taut
strips one after the other. Finally, a metallic hood of
1.5 mm high-grade steel is placed above the mandrel and
coil. The hood is locked from the outside by bolts. The
winding mandrel is moved to the curing mould and c~red in
a hydraulic press. For curing, the end spacers were
axially pressed at 1000 psi with heavy clamp. This
pressure is applied all along curing to avoid axial
elongation.
The preparation and all the steps for winding and
curing the top coil were familiar from the bottom coil.
SIEMENS KWU employees observed the production of the last
5 m coils for the SQC 405 and made familiar with the
machines and tools required for the 5 m coils. Sequence
and handling were known from the 1 m coils. Following
they wound the 5 MIC 49 and 5 MIC 50, assisted by LBL-
employees, and placed the coil into the curing press. The
pressing result of the coils was adequate according to
LBL.
PREPARATION OF COILS FOR COLLARING
Having removed the individual coils from the curing

mould, they are measured with respect to specified
measuring points; the dimensions are to be recorded. At
first, the loose prepreg taps are fixed with Sicomet
(fast acting bonding agent),
then the straight legs
insulated with 2- Kapton-u are resin-coated on the
inside. They are bonded to the coils by using a soldering
983
iron. The joints are exactly on the Cu-keys. Teflon tape
is applied the joints on the upper side of the coil and
Kapton-film to those on the bottom side. After
preinsulation, the coils are stored on a rack. In the
course of the visit to LBL, KWU had the opportunity of
clamping the SSCQP 404. This Q-pole cannot be compared to
a standard quadrupole due to the many special measuring
pOints.
CONDUCTOR CONNECTION, COIL CONNECTION END
The strand end insulation is removed, the ends to be

soldered finish-ground and the conductor end pulled tight
with pliers. strands are coated with flux agent and
tinned. It has to be noted that the conductor is
continuously tinned. The tinned conductor is heated again
and pressed into the plate, groove step by step. The
second conductor end is preformed in the same way. Having
cut the ends to length they are combined by soldering in
the groove. During step-wise soldering, the heated points
are repeatedly cooled with Methyl-Alcohol. After
soldering in the groove, the conductor connection is
removed and excess materials melted off. These activities
have only been demonstrated on an experimental plate.
ACKNOWLEDGEMENT
The whole program got full support from the staff of LBL,
magnet laboratory headed by Clyde Taylor. Open technical
discussions and instructions on the handling operations
on LBL tooling and other material helped to get a
complete insight in the specifics of the LBL CQM design.
We like to explicitly acknowledge and thank for LBL' s
open and kind collaboration.
984
INTERIM REPORT ON TECHNOLOGY TRANSFER ON COLLIDER
QUADRUPOLE MANUFACTURE AT LBL AND SSCL
Robert F. Ryan Ir. and William F. Bensiek
Babcock & Wilcox

P.O. Box 785
Lynchburg, Virginia
24505
ABSTRACT
The B&W/Siemens CQM technology transfer for coil winding, coil curing, coil collaring,
and yoke/skinning and cryostat assembly procedures began in mid-August 1991 at Lawrence
Berkeley Lab (LBL) and Superconducting SuperCollider Lab (SSCL) and is continuing at
publication of this paper. The purposes and objectives of the effort are addressed and the methods
being employed are discussed. The accomplishments to date are outlined. Conclusions are drawn
with respect to applicability for an industrialized production of the collider Quadrupole Magnets.
Completion of technology transfer anticipated in early summer of 1992.
PURPOSE
The purpose of a technology transfer program is for the knowledgeable party to communicate
to the recipient information on all phases of the subject operations. The knowledgeable party is
responsible for insuring that the technology has been effectively transferred and recipients are
responsible for demonstrating that they have learned and have correctly applied the lessons. To
this end, the technology transfer between B&W/Siemens and the Labs consists of three major
parts: observation of operations performed by the Jabs and solicitation of experience-based
suggestions for process improvements, hands-on performance of operations on actual hardware
under the coaching and watchful eye of Lab personnel, review of the procedures and observations
of the processes used by B&W and Siemens.
Acquiring the technology required accomplishing the following tasks:

- Review the fabrication and test facilities at the labs to determine what equipment,
personnel and operating environment would be needed in CQM fabrication

- Make use of the reference design, fabrication experience, and test methods either directly
or as a basis for further process development and refinement.
- Industrialize the fabrication and test processes and methods using the Labs' work as a
reference point with low cost and high quality manufacturing as the goal.
OBJECTIVES
The objective of the technology transfer is to capture the SSC CQM design, fabrication and
testing technology developed at LBL and the SSCL in the following areas:
L.BL SSQ.
Coil Winding End Dome Welding
Coil Curing Beam Tube Installation
Yoke Shell Welding Support Post Attachment
Cold Mass Assembly Thermal Shield Installation
Cold Mass Testing MLI Blanket Installation
Cold Testing Cryostat Piping
Cold Mass Shipment Vacuum Vessel Completion
Participation, by industrial magnet suppliers, in the above activities at the labs will insure that
the lessons gained in the transference are taken to heart and applied to later resultant processes and
procedures. People learn best by doing. Actual hands on manufacture and assembly operations
performed by technicians and engineers from the recipient company will heighten the sensitivity of
these individuals to the complexities of the fabrication and design process.
A second objective is to heighten industrial sensitivity to the various issues concerning CQM
design and fabrication. It is imperative that manufacturers grasp the subtleties of the design and
fabrication process and apply that knowledge to procedures and processes used in the fabrication
of magnets. Using technology transfer in this way will shorten the overall learning time of
manufacturers by improving the awareness of past program difficulties in their concurrent
engineering efforts.
MEmODS
As one would expect a wide variety of methods are employed to transfer and capture the
available CQM technology. Methods used in this effort are itemized below:
Visits to National Labs Participation in Lab Meetings

Shop Floor Discussions Reviewing Sample Hardware
Taking Photographs Progress Reports
Video Taping Work in Progress Notes and Trip Reports
Resident Engineers at Labs Hands On Experience
Lab Review of Manufacturing Procedures
Visits to Labs
B&Wand Siemens have visited Brookhaven National lab, Fermi National Lab, Lawrence
Berkeley Lab, and the SSCL During the visits personnel attended meetings, special briefings, and
observed manufacturing of SSCL dipole and quadrupole magnets.
986
Shop noor DIscussioDs
Personnel were allowed almost unlimited access to the shop floor equipment and Lab
personnel during visits to the labs. They were able to talk with the engineers, shop floor
supervisors, and workers engaged in the design and fabrication of SSCL magnets. Discussions
were useful in -gaining additional information on topics such as:
Equipment Used In-Process Testing Performed

Procedures Used Thoughts on Scaling Process Up
Process Details Process Improvements
Tooling Design Use of Automation
Process Limitations and Pitfalls Lessons Learned
B&Wand Siemens participating personnel have broad industrial experience and shop floor
discussions between participants and Lab personnel have yielded benefits for the labs in helping
solve some of the Labs problems also. As an example B&W noticed that the pole shim on the
LBL fabricated quadrupole magnet did not extend to the end of the winding. The last wire of the
inner coil was thus unsupported and free to move and potentially cause quench problems. This
was corrected by changing the shim design.
Pbotography and VideotapiD&
The labs have allowed unlimited use of all types of photographic media by visiting personnel.
Personnel took numerous still photographs of the various operations and tooling and equipment
involved in the manufacturing process. Video tapes were also used to capture the actual
individual actions and sequences of actions required in many of the processes observed. The
photos and video tapes effectively capture many details of the written procedures and are very
useful in later training of manufacturing personnel who are not all able to view the operations first
hand at the Labs. Videotaping in particular captures the time sequence of operations and actually
captures added detail that the observer may miss while watching the process. Video tapes have
the added advantage of being editable to become the basis of training programs.
Note Taking and Trip Reports
In addition to photography engineers and technicians observing the lab operations took
extensive notes on a wide variety of topics:
Answers to Questions Equipment Capacity

Problems Encountered Vendor Names & Numbers
Lab Contacts & Numbers Equipment and Tooling Information
Types of Testing Used As-Built Dimensions and Weights
Materials Used Expected Performance Results
Test Results Process Times
Step-by-Step Operations
Engineers and Technicians visiting the Labs prepared trip reports which included a basic
review of everything seen and discussed during their visit Notes taken during the visit are helpful
in generation of the reports. They provide a basic flow of the visit and also particulars from the
list above. Trip reports are invaluable tools in communication of the information gleaned during
the visit to other personnel not in attendance. These reports in addition to the lab written
procedures are used as a basis for internal procedure genaation. Copies of trip reports are
provided to the Labs which provides feedback on their assembly processes and procedures which
further allows the Labs to gauge the effectiveness of technology transfer.
987
Participation in Lab MeetiDp
Participation in Lab meetings provides unique insight into the problems encountered during
fabrication, means of their resolution and alternatives for future operations. Participating
personnel are able to take notes, learn of past experiences, and ask questions of Lab personnel
during the meetings and are thus able to learn valuable lessons. This participation with lab
personnel will help prevent duplicating costly mistakes by learning their lessons now.
LessoIL'i Learned
Through conversations with shop personnel and reviewing test data and fabrication results
engineers are able to learn from iterations made by the Labs in design and fabrication of CQM
magnets. These lessons are then applied to their own design and production process
documentation and thereby push the manufacturing company down the learning curve quickly and
at relatively low cost.
Sample Hardware
The Labs have provided extensive sample hardware to visiting personnel such as:
G-lO spacers MultiLayer Insulation Superconducting Cable

Yoke Laminations Kapton Collar Packs
Copper Spacers Prepreg Tape Collared Coil Sections
Collar Keys Backing Strips Shield Extrusions
These parts are useful as training aids and for experimentation. The parts are also helpful in
explaining to prospective subvendors how parts are made and the interrelationship between parts
they may make and other parts of the magnet. This is extremely helpful in expanding the vendor
base of subcomponent fabrication.
Resident Engineer
In addition to visiting engineers and technicians, B&W has in place a dedicated Technology
Transfer Engineering Manager. His responsibility is to track progress at LBL and SSCL and to
coordinate trips by visiting personnel to the labs and insure they are able to observe critical
assembly operations without delaying lab schedules. He is also in the unique position of watching
fabrication of magnets from start-to-ship and learn from all the problems encountered along the
way. In addition he provides the following services:
Continuity of Effort Technical Information

Answers to Questions Spokesman for B&W
Lab Schedules Communication Link
Assistance to Labs
Progress Reports
The resident engineer generates a progress report each week which includes a synopsis of
activities at the lab during the past week, highlights of information learned and provides a
schedule of upcoming activities. These reports provide the basis for travel plans by other
personnel needing to observe specific manufacturing or testing procedures. These reports also
provide a basis to assess the contribution of technology transfer. The reports are circulated within
the CQM organization and provide further insight into the challenges and problems encountered in
magnet fabrication, thereby aiding the technology transfer process.
988
Hands on Experience
The second major step in the technology transfer process is for the recipient to perform work
at the Lab under the guidance of Lab personnel. Instruction and observation can only take the
transfer process so far. It then becomes necessary to test how well the lessons have been learned.
Original plans called for B&W and Siemens to perform all aspects of magnet fabrication on at
least one magnet. B&Wand Siemens were to perform all early cold mass operations including:
coil winding, coil curing, coil collaring, and yoke shell welding at LBL. All subsequent cold
mass operations and assembly of the cryostat were to be done by B&W at SSCL. This work is
currently underway.
Lab Review of Manufacturing Procedures
The final proof of effective transfer is shown in the manufacturing procedures generated for
production hardware by the recipient company. During the model magnet program B&W and
Siemens plan to provide manufacturing procedures to SSCL. They will review the procedures for
technical content and labor intensity. The procedures will be distributed within the SSCL
organization and scrutinized by personnel with magnet experience at all the major labs.
Comments will be addressed and concerns resolved by revisions to the procedures as needed.
ACCOMPLISHMENTS TO DATE
Siemens technicians wound both 1 meter and 5 meter inner and outer coils on LBL
equipment. The Siemens technicians performing the work: at LBL had been involved in the
winding and curing of coils to support the HERA program and were able to bring that experience
to the winding of CQM coils. The winding of CQM coils also demonstrated the differences in
winding between CQM and HERA designs and the additional requirements of winding long
magnets.
B&W technicians participated in yoke shell welding and cold mass assembly at LBL. They
learned to operate the equipment involved and gained an appreciation for the level of effort needed
to route and terminate the power and instrumentation leads in the cold mass end.
B&W participated in the installation and splicing of all splice plates and of coil leads to the
power bus, as well as installation of the expansion loop and associated hardware for routiog of the
cable. This effort sensitized B&W engineen to the importance and difficulty of making good
superconducting splices.
B&W engineers also participated in completion of the cold mass for the lead string test CQM
magnet. This work included welding of end domes, insertion of the beam tube, welding of the
bellows, and welding of the beam tube to the cold mass. Lessons were learned in the installation
of the beam tube, and the importance and difficulty of placing and retaining the position of the
beam tube spacers. This has provided valuable information to B&W 's current design efforts of
beam tube spacers.
B&W participated in other phases of cryostat assembly including installation of the 20K and
SOK shields and welding of those shield segments, installation of cryogenic piping, cradle and
post installation, and MU installation. This portion of the work went quite smoothly and
reenforced lessons learned during review of Lab procedures.
Design engineers became aware of the importance of coil restraint and the impact it had
relative to magnet performance. In particular the impact of restraining movement of the collared
coil within the yoke iron is crucial (See Figure 1). As a result, clamped and unclamped design
989
COIL COLLARS COLD MASS SHELL
COIL WINDINGS
YOKE IRON POTENTIAL SHIM

LOCA TIONS 8 PLACES
Figure 1 Quadrupole Shim L<xation
tests have been included in the B&W model magnet program to test the impact of both design
alternatives to determine which produces the better magnet performance.
B&W and Siemens have established valuable personnel contacts at the Labs and now know
who is cognizant in the various design and fabrication areas at LBL and SSCL. These contacts
will be extremely beneficial in later problem resolution.
OUTLOOK
The technology transfer process will be completed before the start of summer. This will
allow B&W to incorporate the technology transfer lessons in processes and procedures for
fabrication of magnets as part of the model magnet program. B&W still has two important items
to complete as part of the transfer process:
1. Hands on fabrication of the cold mass and assembly of the cryostat and installation of the
magnet in the vacuum vessel and
2. Review of B&W procedures and processes remotely and on site by SSCL cognizant
personnel.
990
Item I should be completed by summer. Item 2 will occur during the model magnet
fabrication effort which starts in July of '92. Completion of item 2 will insure that the technology
has been successfully transferred from the Labs to B&Wand Siemens and will end the technology
transfer process.
INTERIM CONCLUSIONS
The processes used by the labs provide a basis for repeatable, industrialized magnet
manufacturing and testing.
Economics of scale will be realized in the industrialized process used to make magnets of a
fmalized design based on transferred technology.
991
A CIRCUIT THEORY APPROACH TO QUANTIFYING THERMAL
PERFORMANCE OF CRYOSTAT DESIGNS FOR SSC DIPOLE MAGNETS
T. Kupiszewski
ABSTRACT
Design of cryogenic cooling systems for superconducting magnets requires a priori
knowledge of the total heat transferred from ambient to the cold mass as well as to its
surrounding thermal shields. To estimate this net parasitic heat leak through the cryostat of
the Collider Dipole Magnet (CDM) an equivalent thermal circuit representation of the CDM
cryostat is developed to numerically simulate the multimode, spatially-averaged heat flow
occurring after thermal equilibrium is established. In this paper, the formulation of the
proposed circuit model is summarized. The model was implemented in a computer program
which was run to compute representative values for the net steady-state heat flow expected to
occur within a CDM cryostat. A breakdown of this heat leak on a component basis is
provided in tabular form. The effects of inner shield temperature and pressure are simulated
and the results of these simulations are plotted.
INTRODUCTION
During the design of a superconducting magnet cooled by forced flow convection, the
capacity and input power requirements of the cryogenic coolant refrigeration system must be
determined. Under steady-state conditions of thermal equilibrium, the net flow of heat from
ambient to the magnet cold mass serves as a lower bound value for both the required
refrigeration capacity as well as the product of refrigeration input power times the coefficient
of performance (COP). Since the value of COP for a Camot cycle liquid helium refrigerator
is « 1, it is apparent that magnet operation becomes power intensive if the parasitic
inflltration of heat into the cold mass is not limited by the magnet cryostat to a suitable value.
This fact is a critical design consideration for a physically large system like the Collider
Dipole Magnet (CDM) ring of the Superconducting Super Collider (SSC) where thousands
of 15 meter long dipole magnet coils must be maintained at a temperature of 4.35 K. Clearly,
when one is confronted by the prospect of refrigeration loads equaling the net heat leak for
several kilometers of magnets, the quantification of heat leak for an individual magnet
becomes crucial to the optimization of the cryostat design from the standpoint of reducing
power input to the refrigeration system.
In this work, a calculational tool for estimating the steady-state magnet heat leak was
developed. The technical approach was to invoke an electrical circuit analog 1 to heat transfer
in which lumped mass temperatures correspond to the constant voltage nodes of a resistor
lattice and steady-state heat flow corresponds to dc current flow thru the resistors. For the
CDM, since there exist well defined paths for steady-state heat transfer between isothermal,

macroscopic zones, it is possible to construct a conceptual network of thermal resistors
having values determined by geometric and material properties. This network constitutes a
thermal equivalent circuit model of the CDM cryostat. One possible version of such an
equivalent circuit model is presented in this paper, along with the key mathematical
expressions used to parameterize the heat flow. The model was "exercised" to predict static
heat load and simulate the effects of inner shield temperature and pressure within the vacuum
vessel. The results of these simulations are also included.
(COLD I.CASS I
L . He R£TU'lN
TueE
SOK SHIELD
ASSY
REENTRANT POST
ASS'f
Figure 1. Cross Sectional Illustration of Super Collider Dipole Magnet within Cryostat
THERMAL EQUIVALENT CIRCUIT MODEL

The topology of the proposed thermal equivalent circuit model was based upon a
cross-section configuration of the baseline CDM design. For reference purposes, this
configuration is illustrated in Figure 1, which depicts the 4.35 K dipole magnet (aka cold
mass) set atop one of five reentrant support posts within the cryostat vacuum vessel. Most of
the insulating value of the cryostat is derived from evacuated blankets of MLI surrounding
two aluminum shields, one of which is thermally anchored to the 20K gaseous helium pipe,
the other being thermally anchored to an 80K liquid nitrogen pipe. The support posts are
equipped with thermal intercepts at 80 K and 20 K which result from low resistance
connections to the 80 K and 20 K shield assemblies.
The thermal equivalent circuit model, which corresponds to the Figure 1 cross section
of the CDM, is shown in Figure 2 and consists of a series-parallel combination of 25 thermal
resistances. A principal assumption underlying the model is that thermal end effects in the
vacinity of the interconnect regions can be neglected for magnets connected in a string so that
axial temperature gradients are effectively zero. Three modes of heat transfer are
994
accomodated in the model. These modes are conduction, infrared radiation, and residual gas
conduction. For the support posts, it was assumed that the conduction mode of heat transfer
was dominant. Hence, a grouping of five, parallel conduction resistances is interposed
between the 300 K ambient, 80 K shield, 20 K shield, and 4.35 K cold mass temperature
nodes. For the MLI blankets surrounding the 80 K and 20 K shields, all three modes occur
simultaneously to some degree, depending upon vacuum level and compression of MLI, so a
grouping of three parallel resistors form a second thermal link from 300 K to 80 K and from
80 K to 20 K. A shunt resistor is also included as a third link from 300 K to 80 K and from
80 K to 20 K to simulate the effect of localized MLI compression and/or optical windows
associated with support post holes and inadvertent tears and punctures in the MLI blankets.
From 20 K to 4.35 K, in the case where no MLI blanket surrounds the cold mass, heat is
transferred by conduction thru the support posts as well as infrared radiation and residual
gas conduction across the space between the interior surface of the 20 K shield and the
external surface of the cold mass shell. Thus, two parallel resistors connect the the 20 K
node to the 4.35 K node in addition to a grouping of five parallel conduction resistances.
Reentrant Support Posts
Cold Mass
(4.35 K)
Tunnel
Ambient .-"""'-.--'V\f\r-i.
(300 K)
V
n -T SDK
Shield
...I...
---t.~ Static Heat Flow Between Isothermal Zones

..0. Static Heat Load for Cryogen Refrigeration System
Figure 2. Equivalent Circuit Model of Cryostat Heat Leakage During Nominal Operation at
Thermal Equilibrium
Calculation of the heat load for the cryogen refrigeration system is' straightforward with
the aforementioned thermal equivalent circuit model. Conservation of energy dictates that the
sum of all net heat flows entering and leaving a given temperature node must equal zero at
thermal equilibrium. Accordingly, the heat loading, Qload, on the cryogen refrigeration
system at 80 K, 20 K, or 4.35 K, is given by an equation of the form
(1)
where Qin is the input heat flow in watts and Qout is the outgoing heat flow in watts for the
80 K, 20 K, or 4.35 K node. Calculation of these net heat flows between temperature nodes
995
is accomplished by totaling the heat "current" flowing along each thennal path connecting
two given nodes. Because the resistors connecting any two nodes are all in parallel,
steady-state values of these heat flows to 80 K, 20 K, or 4.35 K are given by an equation of
the fonn
(2)
where Q is the net heat flow between nodes, ~T is the positive temperature difference in
degrees Kelvin between nodes, Ri is a thennal resistance value(K!Watt), and Ui is a thennal
conductance valueCWatt/K). Parametric expressions for these thermal conductances are given
in the following section.
RELATIONS FOR HEAT TRANSFER AND THERMAL CONDUCTANCES

Support Post Conduction
The nominal heat leak: associated with the five reentrant support posts is due primarily
to heat conduction in the axial direction of each thin wall composite tubes. Therefore, the
flow of heat in the post between any two temperature intercepts (or nodes, in reference to the
circuit model) is proportional 2 to a thennal conductivity integral according to
(3)
where Qpost is the conduction heat flow(watt!post), A is the cross section area perpendicular
to the tube axis, L is the length of tube between thermal intercept rings, k(T) is the
temperature dependent thennal conductivity of the tube material, and TI & T2 are thennal
intercept temperatures. Since the support post may be constructed from one or more
composites, more than one k(T) function may be needed. For the purposes of computer
implementation of the model, plots of thennal conductivity data 3 versus temperature for
both G 11 and GRP were curve fit to 5th order polynomials and these polynomials were
integrated analytically. Equation 3 can be rewritten in a fonn which is more compatible with
the circuit model. This alternative fonn is given by
Qpost = Upos t (~T = T2-Tl) (4)

where
Upost = (r..(~Tl))fT~k(T)dT
(5)
Infrared Radiation
Radiative heat transfer to the cold mass and within the cryosat in general is treated as
radiation exchange between concentric axisymmetric surfaces. The radiative heat flow
between any two such nested surfaces within the cryostat can be expressed mathematically as
(6)
where Th is the temperature(K) of the "hot" surface, Tc is the temperature(K) of the "cold"
surface, and Urad is a radiative thennal conductance(Watt/K). An expression for Urad is
given 4 by the following equation.
996
(7)
Here, 0" = Stefan-Boltzmann constant, eh = emissivity of the hot surface, ec = emissivity of

the cold surface, Ah = area of the hot surface, Ac = area of the cold surface, and Fch = the
optical view factor from the cold surface to the hot surface. Since the cold surface by
definition is always nested completely within the hot surface, Fch is set equal to one.
Residual Gas Conduction

Under conditions of low temperature and degraded vacuum quality, residual gas
conduction can dominate the radiative component of the overall heat transfer. Thus, the
magnitude of residual gas conduction can strongly influence the performance of the cryostat
from the 20 K shield to the 4.35 K cold mass. In general, assuming the gas mean free path
length exceeds the characteristic surface spacing in the 20 K to 4.35 K region, the residual
gas conduction heat transfer, Qgas' between a hot enclosure surface and a cold enclosed
surface can be expressed mathematically as
(8)
where Th is the temperature(K) of the hot surface, T c is the temperature(K) of the cold
surface, and U gas is the residual gas thermal conductance (WattIK). U gas is given 5 by
y+ 1
U gas =
y- 1
(9)
where y is the specific heat ratio, R is the specific gas constant equal to Ru/M, Ru is the
universal gas constant, M is the molecular weight of the gas, T is the temperature(K) of the
gas pressure gauge, p is the absolute pressure(Pa) of gas, Ac is the area of the enclosed cold
surface, and Fa is the accomodation coefficient factor. The accomodation coefficient factor is
defmed according to
(10)
where Ac is the area of the enclosed cold surface, Ah is the area of the hot enclosure surface,
ac is the accomodation coefficient of cold surface, and ah is the accomodation coefficient of
the hot enclosure surface. For the case of MLI being wrapped around the cold mass, the
resulting decrease in U gas is approximated, under the assumption that the value of
accomodation coefficient does not change appreciably, by dividing Fa by (N+ 1) where N is
the number of blanket layers.
Multi Layer Insulation (MLI)
The expression for steady-state heat flow through MLI also has a similar form to that of
radiative heat flow. This expression is given by
997
(11)
Here, Th is the hot boundary temperature(K) of the insulation, T c is the cold boundary
temperature(K) of the insulation, and Umli is the MLI thermal conductance(W/K), which is
defmed according to
(12)
where k t is an apparent thermal conductivity (W/m-K), Ah is the area(m2) of the hot

boundary of the insulation, and ~x is the thickness(m) of the MLI blanket. The apparent
thermal conductivity, kt, is given 5 by the following mathematical expression.
(13)
Here, N is the number of aluminized layers, hc is the solid conductance (W/m2-K) for the
spacer material, and e is the emissivity of the aluminum coating. For a 32 layer blanket using
Reemay spunbonded polyester spacer material with double aluminized mylar layers at a
density of 3.61 layers/mm, the published 6 experimental k t value of 0.52 x 10-6 W/cm-K
from 300 K to 80 K translates to a value of 0.13 W/m2-K for hc when e = 0.05. Equation
14 has much utility since it can be used to simulate thermal shunts in the MLI blanket due to
localized effects such as (1) slight increases in layer density due to blanket compression, (2)
increases in he due to cryo deposits, and (3) increases in emissivity.
RESULTS
A computer program was written in Basic to automate calculation of the thermal
conductances, heat flow, and refrigeration loads according to equations 1 to 13. To run, this
program requires input of values for fifty two variables. The input data for surface areas was
derived from dimensions indicated on SSCL drawing M30-000090 showing the 27" dipole
cryostat with 50 mm aperture. Dimensions for the support post were obtained from
information 3 provided during the Technololgy Transfer Sessions. Table 1 displays the
predicted steady - state heat flow to the 80 K shield, to the 20 K shield, and to the cold mass
Table 1. Predicted Values of Static Heat Flow in Cryostat for:

pressure = le-7 Torr, MLI emissivity = 0.03, MLI k t = 0.52e-6 W/cm-K,
62 layer MLI blanket on 80 K shield, 32 layer MLI blanket on 20 K shield,
no MLI blanket on cold mass
4.35 K 20K 80K

Static Heat Flow (Watt per Cryostat)
Infrared 0.0078 0.036 4.6
MLI Spacer Conduction 4.76 11.25
Gas Conduction 0.0228 0.0008 0.0007
Support Conduction 0.126 1.6 11.3
Total 0.16 6.4 27.2
998
under nominal conditions. For an inner shield temperature of 20 K, which is the required
operating temperture of the inner shield, the thermal equivalent circuit model predicts 27.2
Watts, 6.4 Watts, and 0.16 Watts for heat flow to the 80K, 20K, and 4.35K intercepts
respectively. The predicted heat flows correspond to theoretical refrigeration loads of 20.8
Watts, 6.24 Watts, and 0.16 Watts to the 80K, 20K, and 4.35K intercepts respectively.
Relative to the heat load budget 7 of 37 Watts, 5.055 Watts, and 0.363 Watts at 80 K, 20 K,
and 4.35 K repsectively, these results suggest that the thermal performance of the cryostat
design analyzed by the circuit model would fall within the budget, except for the heat leak to
8,....----------------.,.0.4
6 0.3
5
Inner Shield Cold Mass
Refrigeration 4 0.2 Refrigeration
Load (Watt) Load(Watt)
3
--
2 0.1
O+-~~~~~~~-r~~~-r~~~~
10 15 20
--- Inner Shield
Cold Mass
25 30
0.0
Inner Shield Temperature(K)

Figure 3. Predicted Variation in Refrigeration Loads as a Function of Inner Shield
Temperature.
20 K, which is predicted to be approximately 23% greater than the required limit. Also, note
that the difference between the theoretical and the budgeted refrigeration load values at 4.35
K is a conservative one since the model neglects the contact thermal impedance that exists
between the cold mass cradles and the tops of the support posts. Predicted variations in
refrigeration load at both the cold mass and the inner shield as functions of inner shield
temperature are shown in Figure 3. Figure 4 shows a plot of the predicted relative
contribution of residual gas conduction to the total cold mass heat flow input at 4.35 K as a
function of pressure within the cryostat vacuum vessel. From inspection of Figure 4, it
becomes apparent that for pressures of le-7 Torr or less the residual gas conduction
represents a small percentage of the overall heat flowing to the cold mass. As vacuum
degrades to le-6 Torr, the results indicate that heat flow to the cold mass is almost entirely
due to residual gas conduction. This plot dramaticlly illustrates how important maintaining
high vacuum quality is to maitaining adequate cryostat thermal performance. It also serves to
underscore the need for a cold mass MLI blanket to provide additional impedance for heat
flow via residual gas conduction.
999
0.5-r----------------:-----.
0.4
0.3
Heat Flow (Watt)
to4.35K
0.2
0.1
---- Total
- Gas Conduction
O.O+--~--r-~---""T""--~--r--~~
O.OOe+O 5.00e-7 1.00e-6 1.50e-6 2.00e-6
Pressure (Torr)
Figure 4. Predicted Static Heat Flow to Cold Mass, w/o MLI, as a Function of Pressure
CONCLUSION
A simple thermal equivalent circuit model of steady-state heat flow in the Collider
Dipole Magnet was developed. This equivalent circuit model is a robust analytical design
tool in that it facilitates rapid simulation of the effects of a wide variety of design parameters
upon cryostat heat leaks under equilibrium conditions.
REFERENCES
1. E.R.G. Eckert and R.M. Drake, jr., Analysis of Heat and Mass Transfer,
McGraw-Hill, Inc. (1972)
2. W. Frost, Heat Transfer at Low Temperatures, Plenum Press, New York
(1975)
3. T. Nicol, "Component and System Design - Suspension Design," presented at the SSC
Collider Dipole Magnet Technology Transfer Meeting, Cryostat Session, Dallas,
(5/23/91)
4. J. P. Holman, Heat Transfer, 5th ed., McGraw-Hill Book Co., New York
(1981)
5. R.F. Barron, Cryogenic Systems, 2nd ed., Oxford University Press, New York
(1985)
6. W.N. Boroski, T.R. Nicol, and C.J. Schoo, "Design of the Multilayer Insulation
System for the Superconducting Supercollider 50 mm Dipole Cryostat,"
Proceedings of the 3rd IISSC Conference held March 13-15, 1991, in Atlanta,
Plenum Press, New York
7. T.R. Nicol, "Design Development for the 50MM Superconducting Super Collider
Dipole Cryostat, " Proceedings of the 3rd ISSC Conference held March 13-15,
1991, in Atlanta, Plenum Press, New York
1000
A METHOD OF PREDICTING THE RELIABILITY
OF CDM COIL INSULATION
A. Kytasty, C. Ogle, and H. Arrendale
General Dynamics
Department 8760(MZ: CI-8770)
P.O. Box 85990
INTRODUCTION
This paper presents a method of predicting the reliability of the COM coil insulation
design. The method proposes a probabilistic treatment of electrical test data, stress analysis,
material properties variability and loading uncertainties to give the reliability estimate.
APPROACH
Designing reliability into the COM magnet requires a means to assess the design from a
reliability viewpoint. This has traditionally been a qualitative analysis, based on deducing
possible failure modes, and then documenting what design features, tests or process controls
"mitigate" the failure mode from occurring. A "based on similarity" argument is then often
applied to relate historical failure rate component data to give a "quantitative" probability of
success prediction. This approach can be used for highly standardized and mass produced
items such as electronic components, but quickly loses credibility when addressing
mechanical components that are neither standardized in design or application. Further, the
type of failure data information (number of failures, and total operating hours), cannot
properly address the types of failure mechanisms one would expect with mechanical items,
such as fatigue or wearout.
The approach taken to predict reliability of design related failure modes of the Collider
Dipole Magnet is to form analytical models of the various possible failure modes and their
related mechanisms or causes, and then statistically assess the contributions of the various
contributing variables. The probability of the failure mode occurring is interpreted as the
number of times one would expect certain extreme situations to combine and randomly
occur. One of the more complex failure modes of the COM will be- used to illustrate this
methodology .

The failure mode analyzed in this paper is the mechanical breakdown of the multi-
layered polyimide film insulation of the magnet coils. Loss of the electrical insulation around
the coils at any location would cause a short between the coil and the grounded collars or
short between coils or turns, and necessitate magnet replacement. The highest stresses have
been predicted to occur at the midplane. Other areas of potential electrical insulation failure in
the magnet are at the beam tube, bus leads/supports, and end turns.
The SSC system availability requirements and likely maintenance scenarios allow only
73 dipole magnet replacements over the operational life of the SSC. Since there are 7964
CDM's, each with many other potential causes of failure, the allowed probability of anyone
failure mode quickly reduces to the order of I in 100,000. Reliability prediction is the task of
assessing whether or not a failure mode in question will meet this number. If it does not,
then it provides rationale as to what has to be adjusted (i.e., designed in) to give the
capability to achieve this number.
The first step is isolating the causes of the failure mode to a level where process controls
or design features can be implemented. For instance, the insulation can fail because; 1) it is
mechanically overstressed, 2) it weakens due to age or radiation exposure, 3) it is inherently
flawed or inadvertantly damaged during manufacture or in-process testing, or 4) it is broken
or punched through by contamination introduced during manufacture or testing. Thus, there
are four issues that must be checked and resolved. The design aspect of the problem
discussed in this paper is related to the first cause only, that of experiencing high local
mechanical stress. The other issues are being addressed under contractual work of the CDM
Reliability Program.
Mechanical stresses on the coil and insulation are essentially due to the manufactured
coil size, and collar design dimensions. The collar is specifically designed to introduce
sufficient prestress to keep the coils in compression through cool down to liquid helium
temperatures, despite the differences in thermal contraction caused by the disparate materials
of the collars and coils.
However, variations in the nominal prestress level can be expected due to the
manufacturing variations in collar as well as coil dimensions. A collar with internal
dimensions at the small limit of tolerances, and an inner coil with larger than normal
azimuthal dimension, when assembled, can obviously result in a region of much higher than
expected stresses. It is in these areas that the polyimide film must be able to resist
compressive overstresses. To form a probabilistic model of this failure event, statistical
information is required on; 1) collar dimensions, 2) coil dimensions, and 3) insulation
mechanical/electrical capabilities. The general reliability equation for this failure mode can be
expressed as:
Unreliability = Probability (coil stress> polyimide fIlm capability) (eq.l)
where coil stress and film capability are distributed random variables, and
unreliability = 1- reliability.
The tIrst interaction, that of collar and coil interference and the resultant stresses is
modeled with a 2-D finite element model of the cold mass section. Figure 1 shows the
expected stress contours of one quadrant of the cold mass coils. The largest coil stress is
observed at the inner edge of the inner coil midplane. Since this stress value is computed
using mean or expected values for the contributing variables, it itself is then the mean or
average value in some distribution of stress at this point. The first objective in this portion of
the analysis is to to gain a reasonable model for the distribution of stresses at this point.
1002
OUTER
COIL
INNER
COIL
Figure 1. 2-D Finite element cross section model results of CDM cold mass. Nominal case where level 1 =
108 MPa compression, with 3.73 Mpa per level decrement.
The approach taken follows the work given in references 1 and 2. The objective is to
treat a complex analysis model (such as Finite Element Analysis) as an experiment with
controllable inputs and observable responses_ By carefully selecting the input variables and
their combinations, and running "experiments" (in this case finite element runs) at various
parameter settings, one can deduce the influence of input variability and fonn a response
function of stress as a function of the random variables. Table 1 summarizes a factorial
design of experiments with 2 factors (reference 3) and the results. It is assumed that the
variations in modulus of elasticity and Posson's ratio of the collar and wedge material is
negligible in comparison to collar geometrical effects and coil size/modulus variations. Thus,
natural variation in stresses due to manufacturing capabilities are modeled to be solely due to
the upper coil to lower coil midplane interference created by azimuthal dimensional
differences between the coils and collars, and room temperature coil modulus. The selected
input ranges given in the table span the nominal values of 10000 MPa for the coil modulus
and .3248 mm for the coil to coil midplane interference dimension.
Analysis of the results in Table I (stress) can be accomplished using Yate's method as
given in reference 3 or by partial regression analysis methods (ref. 4), with identical results.
The resultant linear model relating stress to the random variables coiVcollar interference (I)
and coil modulus (E) is:
Coil stress at midplane (MPa) = 108 + .OO5(E-E) + 401(1 - I) (eq.2)
where: E and I denote mean levels.
This equation can now be used to determine the distribution of possible stresses at that
point using Monte Carlo methods with E and I represented as random variables with defmite
distribution types and values based on achievable manufacturing tolerances, and updated
experimental data on coil modulus.
1003
Table 1. Experimental setup and Finite Element (FE) results
FE RUN COIL IIIDPLANE INTERF. (11111 E,COIL IIODULUS (IiPAl IIAX. STRESS (IiPAl
1 .18711111 11260 51
2 .18711111 107311 54
3 .46251111 11260 157
4 .46251111 107311 1611
The second part of the reliability equation is the insulating film mechanical/electrical
strength capability. The data for this will be generated by conducting tests to determine the
ultimate mechanical punch forces under an applied high voltage (2kV nom.). The proposed
test setup is shown in Figure 2. The experiment will be repeated for 10 samples following
the cycling stress regime shown until failure occurs (current leakage noted). The data will
then be fitted to a probability distribution using Wei bull techniques.
FORCE
FAILURE WHEN
LEAKAGE
INSULATION AT
CURRENT
7
IIIDPLANE
IIPA
CTED
INSULATED
CABLE
CYCLES
Figure 2. Schematic of proposed test setup to determine polyimide punch through characteristics
The resultant probability distribution of the polyirnide film capability, as configured at

the midplane, will be compared with the derived stress distribution function (eq. 2) , and the
probability of failure computed using equation I, employing Monte Carlo methods for
solution. Figure 3 shows a parametric study of possible results assuming; I)the film
capability to be nonnally distributed for various levels of variation (Coefficient of Variation
(COV) = standard deviation/mean) in the test results, and 2) the stress distribution function to
be normally distributed with COY = 5% in both of its variables.
1004
1.00E-{)1
1.00E-{)2
1.00E-{)3
'-
............... ...
-'- COV 10%
PROB.
1.00E-{)4
..........
·1...... ~
...
OF 1.00E-{)5
...........
.,
FAILURE ...............
1.00E-06
1.00E-{)7
............ -......... COV5%
"" "-
1.00E-08
1.00E-09
1.00E-10
....... COV2.5%
1.00E-ll
140 144 148 152 156 160
MEAN VALUE OF POLYIMIDE CAPABILITY (MPa)
Figure 3. Insulation reliability as a function of possible test results.
REFERENCES
l. D.K. Lloyd, Long Range Service Life Estimate Procedures, 1. Spacecraft, Vol. 14, No.6, June 1977.
2. C.W. Johnson, R.E. Maxwell, Reliability Analysis of Structures - A New Approach, Proceedings 1976
Annual Reliability and Maintainability Symposium, 1353.
3. M.G. Natrella, "Experimental Statistics", National Bureau of Standards, Handbook 91,1963. Chapter 12.
4. Rl... Anderson, T.A. Bancroft, "Statistical Theory in Research", McGraw-Hill, 1952. Chapter 14.
ACKNOWLEDGEMENT
The WOIX described herein is being accomplished under contract to the Universities
Research Association insupport of the Superconducting Super Collider project for the U.S.
1005
SOFTWARE DESIGN PHILOSOPHY FOR THE
SSCL MAGNET TEST LABORATORY
J. David Lambert, l Paul A. Craker, 1 Joe P. Garbarini Jr., l James C. Ivey, l

Wei-chuan Li, l Chris G. Saltmarsh,2 and Igor Syromyatnikov 1
1 Magnet Systems Division

SSC Laboratory*
Dallas, TX 75237-3946
2 Lawrence Berkeley Laboratory
Berkeley, CA 94720
INTRODUCTION
The Magnet Test Laboratory (MTL) will be required to extensively test thousands of
superconducting magnets prior to their installation in the Superconducting Super Collider
(SSe). Software systems that control measurements made in this laboratory must share the
same high standard of reliability required of the facility as a whole. We describe here our
choices of tools, standards, and methodologies used in developing the services needed to
implement MTL applications.
We have examined the magnet testing methods currently used by Lawrence Berkeley
Laboratory (LBL), Brookhaven National Laboratory (BNL), and Fermi National Laboratory
(FNAL), together with the platforms these laboratories have chosen for implementation.
This investigation has shown great diversity in methods, and evidence that the designs have
been, and still are, largely evolutionary. Also, no long-range standards such as ANSI or ISO
have been consistently applied. We concluded that no single existing system can be consid-
ered as a definitive functional model for the MTL. Moreover, many of the systems are cur-
rently under further development.
In the absence of suitable functional specifications, and considering the dynamic
nature of the testing methodologies, it is evident that a rapid prototyping environment is
required. This type of development environment must afford flexibility to requirement
* Operated by the Universities Research Association, Inc., for the U. S. Department of

Energy under Contract number DE-AC35-89ER40486.

changes while retaining system integrity. Therefore, it has been decided to use an object ori-
ented, tool-based approach rather than the more traditional, application-based, top-down
approach to MTL software design'. A tool-based approach provides us with a rapid proto-
typing platform where significant changes in design may be accommodated with minimal
risk and disruption to schedule. In order for this approach to be successful, a powerful tool-
kit must be assembled.
Application (client)
Acquisition Presentation Configuration

Database
Servers
Figure 1. Client/Server model for a "generic application."
To minimize reinvention, unnecessary risks and support costs, it has been decided that
the tools should be based on industry standards wherever possible. If a required tool is not
available as an industry standard, then a search is made for an emergent standard throughout
industry and within the laboratory community. Existin"g test systems at the nationallaborato-
ries have been re-examined for areas of commonality. A composite "generic application" is
then designed based on a client-server model (Figure 1). This provides the definitions for a
set of tool classes. With these classes defined, a search is made for suitable candidates in
each category. Since we first adopted this approach, we have had the opportunity to engineer
a replacement for the magnetic measurement system at the FNAL Magnet Test Facility
(MTF). The first version of this system was designed, implemented, and installed, during the
last quarter of 1991. Initial results have shown excellent correlation with measurements
taken using the original system (Figure 2). Measurement speed has also been increased by
more than a factor of two. A second application using the field angle probe is currently
under development. Testing of this system is scheduled to commence at the MTF in March
1992.
1008
an vs Position for magnet DCA314 - files 18(HP) and 518(Unix)
a, 'lB. POSmON a. 'lB. POSITION
1.0
f ~
e
T
0.6
0
T
0
:::. 0 :::. 0.0
~
! .
~
• -I 1-
:>
0 •6
-I
-10
........ Poaition (meten)
a. 'lB. POSmON
_...10
-1.0
-10
........
0
Position (meten)
a. 'lB. POSITION
--10
0.4
f
G 0.2
T0
.::. 0.0
i 0.2
e.-
: I
_... --
-0.4
-10
........ POIition (meten) ...... Po81tlon (me ten}
Cold measurement after quenches TCI 16/18/It. corrected

Figure 2. Comparison of multipole measurements taken with the MTL system and the
original BNL system at FNAL. Note that each of the four plots shows two sets
of nearly identical results.
SYSTEM DESCRIPTION
Services defined in the "generic application" (Figure 1) perform specific tasks as

required by the client application. Interfaces between the client and these services are imple-
mented by a thin layer known as the Application Program Interface (API). Usually this API
is no more than a library of function calls, together with a set of data structure definitions. A
description of the implementation of these services follows, together with the tools and stan-
dards employed in their development.
Systems Platform
The MTL test stand organization is shown in Figure 3. The hardware platform chosen
is a networked system based upon SPARe architecture workstations and servers. These
machines are off-loaded from real-time tasks by a VME-based, front-end system controlled
by a Motorola MV147 card, and containing a 25-MHz 68030 microprocessor. The above
hardware choices have significantly influenced the following software decisions.
1009
..-
Instrumentation .. .. Data Ac~uisition
Front nd
.. .. Host Control,
Analysis
& Presentation -
··
i
·
!
··
! !
··
! !
· Data Archive
··
!
! i
!
!
r- Server
· ·
-
! i i
Instrumentation Data Ac~uisition

Front nd -- .
~
Host Control,
Analysis
& Presentation
Instrumentation
VME
Data A~isition
Front d
M68030
Host Control,
Analysis
& Presentation
SPARC
- SPARC
Figure 3. MTL Test Stand Dataflow.
The SPARe architecture of the servers and the workstations makes UNIX the natural
choice for an operating system. To date SUN UNIX has been used, but this could change as
POSIX standards are applied. Wherever possible, we intend to conform to emerging POSIX
standards. UNIX is not yet suitable for real-time, preemptive tasks, so a different operating
system is needed for data acquisition. Wind River's VxWorks2 operating system, embody-
ing the WIND real-time kernel, has been chosen for the front-end machine's operating sys-
tem. This choice of operating system has also been influenced by VxWork's seamless
interface with UNIX, good debugging tools, and some previous user experience within SSC
and LBL. VxWorks, however, does not completely emulate UNIX; some utilities are miss-
ing. Therefore it has been necessary to develop libraries to bridge these gaps, particularly in
the areas of networking, time of day, and the remote shell daemon. Custom extensions to
VxWorks for the M11.. include device drivers, a performance monitor, watchdogs, and a
VME bus mapping tool.
Program Development Tools. The choice of UNIX as an operating system dictates

that the prime system's language be C. It has been decided to adopt the ANSI C standard, as
it affords much greater compile-time code checking than pre-ANSI versions. C++3 is being
introduced in some areas, notably those of graphical interfaces and data design. The Free
Software Foundation's G~ compilers "gcc" and "g++" are used in both the UNIX and
VxWorks environments. The choice of C as the primary systems programming language is
not intended to preclude the use of other applications languages, notably FORTRAN. We
intend to provide a FORTRAN interface library, particularly for the development of post-
test applications. Style guides and associated templates have been used to provide some uni-
formity for code generated by individual programmers. A makefile has been developed to
encourage a standard for the use of development tools within the MTL. Regular design
reviews and code walk-throughs are used to trap potential bugs early in a project. Programs
are documented using UNIX "man pages," plus comments in source code.
1010
Computer systems source code is a notoriously difficult environment in which to con-
trol changes. Several source control systems are available under UNIX. We have chosen to
use RCS, as it is simple to use, and significant expertise in its use exists in the department.
The choice of a debugging system has been largely influenced by Wind River's own debug-
ger, which is based on GNU's "gdb". Gdb is adequate for most tasks as it provides full
source-level support. Saber C and C++ have been evaluated and found to offer significant
advantages in tracking obscure problems. At present the Saber products are being used to
complement rather than to replace the GNU tools. Although any text editor may be used for
source development, the editor of choice is GNU EMACS. EMACS is available on many
platforms including VAX. and PC; it offers an integrated program development system.
Operator Interfaces
Window System. A window system is an environment that uses a single physical dis-
play device for several concurrent applications. XllR4 from MIT has been selected as the
window system for the MTL, as it is the de facto workstation standard. The base level win-
dow manager that is distributed with X (twm), is used in preference to commercial exten-
sions such as OpenWindows or Motif. Since both of these window managers are proprietary
and their use could impede migration to future UNIX platforms.
Test Data Display. One of the prime requirements of a test data display system is
rapid response. This display must provide the operator a current view of the test's progress.
This time criterion compromises to some extent the functionality of the display package.
Therefore it seems reasonable to consider two requirements, one for real-time, and the other
for post-test analysis. ~uring analysis, response time is secondary to presentation ~uality.
For real-time display, we have made use of the ISTK5 graphic libraries GLISTK ,6 and
GLOSS. Figure 4 shows a display using an ISTK tool. Candidates for post-test analysis dis-
plays are currently PV-Wave, MatLab, and ACE/gr.
Control Panels. The test systems are controlled by "point-and- click" graphics panels
on which the operator may enter parameters or initiate actions. The requirements for these
control panels are similar to those of the real-time display server. Again, we have chosen the
GLISTK libraries for these functions. The GLISTK system also has an integrated event
communications system that provides a natural interface with the system's sequencing lan-
guage GLISH5,7.
Communications
The ISO/OSI model defines a layered, platform independent paradigm for communi-
cations systems. This model has been used as a guide during systems development. It is
desirable that communications between processes use the same primitives, whether locally,
or residing on separate machines. The ISTK system provides a sequencing language,
GLISH, which manages the communications of events and small data packets (less than a
few tens of bytes) between processes. At present, GLISH is not implemented on VxWorks,
but work is in progress to allow this enhancement. When transport of larger amounts of data
is required, the data are encapsulated as an SOS8 and passed over Berkeley stream sockets.
A reference to the SOS is sent via a GLISH7 event. This method allows reliable data transfer
and synchronization with minimal overhead. Some extensions to VxWorks have been devel-
oped to simplify the implementation of SOS transfer via sockets.
1011
:11 -n :~
c
.,,
-,~~
'f--------II
-2_ 0.000
11111111111111
......... ao.ooo OJJOO _1ncIII:Ic ,~
D._
D.DOD
Figure 4. Real-time display of test results.
1012
Data Formats
Data that are required to be passed between processes, possibly residing on machines
having different architectures, must contain meta-data to describe their internal representa-
tion. ASCII, combined with an application specific format, is often used to achieve machine
independence. This type of data representation is grossly inefficient in data space and is dif-
ficult to maintain. We have chosen the ISTK SDS 5,8 binary-based format, as it combines
efficiency with a standard formatting method.
Database Interfaces
The MTL will provide a database system containing all magnet data (on the order of
hundreds of gigabytes). This database contains previous test results data and configuration
information for the test stands. Some of these data are required to be retrieved and held dur-
ing a magnet test. As there may be unpredictable delays in retrieving data from across the
network, access is not time deterministic. Hence it is necessary to provide a local caching
system.
Test Results Data. Contained in this data base are the results of all tests, for all the
magnets, and for each of the test stands. This information includes raw data, analysis results,
and a reference to the configuration of the test stand when the test was run. Raw data will be
stored as an image of the test SDS; the rest of the data will be stored as relational tables.
Configuration Data. During a magnet test there must be access to information about
the magnet test stand and the magnet under test (for example, instrumentation being used,
gain settings, and calibration information). The configuration database provides this infor-
mation together with the availability of equipment. The MTL database will be updated with
the configuration held locally.
Exception Handling
All run-time exceptions, such as equipment malfunctions, and all resource conflicts are
channeled through an exception handler library. At present these handler routines simply
log, on the console, the process name, an error message, and the current system state. This
exception handler library is intended to be significantly extended as MTL applications grow.
Ultimately it is envisioned that exceptions will be passed to an expert system that will cate-
gorize the severity of the error within the context of the entire MTL. Corrective action will
then be proposed according to rules derived from previous operational experience. Artificial
intelligence systems currently being investigated are CLIPS from NASA, 02 from Oensym,
and Nexpert.
Acquisition
Data acquisition is performed using instrumentation interfaced to a VME bus, either

directly or via an IEEE-488 bus interface. Data are acquired via software drivers, are assem-
bled into raw data sets, and are passed to the host via the local area network. The data sets
use a self-describing data format, as specified by the ISTK SDS standard.
1013
FUTURE DEVELOPMENTS
The tool systems require some further work, particularly in the areas of the configura-
tion data base and extension of the scope of ISTK into the VxWorks front-end. Also, work
on the expert system for handling exceptions and alarms has just begun.
The mole harmonic measurement system is nearing maturity and will shortly be deliv-
ered to vendors of SSC superconducting magnets. A field angle measurement system is
expected to be completed shortly thereafter.
The next applications to be developed using this system will be:
1) Quench data acquisition and characterization systems.
2) Test sequencing and power supply control.
3) Refrigerator system interface.
REFERENCES
1. Grady Booch, "Object Oriented Design with Applications," The Benjamin/Cum-

mings Publishing Company.
2. Wind River System, Inc., "VxWorks Programmer's Guide."
3. Bjarne Stroustrup, "The C++ Programming Language," (1987), AT&T Bell Labora-
tories Addison-Wesley Publishing Company.
4. Richard M. Stallman, "GNU EMACS Manual," (1987) Free Software Foundation.
5. Chris Saltmarsh and Erica Lutz, "SSCL-MAN-OOI7, The Integrated Scientific Tool
Kit," (1991). An overview of the ISTK system and its components.
6. Matt Kane and Erika Lutz, "The GLISTK Manual ISTK documentation," (1991),
[email protected].
7. V. Paxson, "The Reference Manual for the GLISH Sequencing Language," (1991)
ISTK documentation, [email protected].
8. Chris Saltmarsh and Erica Lutz, "The SDS manual, ISTK documentation," (1991),
[email protected]
1014
SYSTEM DESIGN OVERVIEW FOR THE MAGNET TEST DATABASE SYSTEM
James Ivey, Xiaoyu Li, Joe Garbarini, and Penny Ball

Dallas, TX 75237-3946
ABSTRACT
The Magnet Test & Data Management Department is responsible for testing of the
superconducting magnets to be produced by the SSCL and industry, for data analysis of test
results, and for creating and maintaining a central database of Laboratory magnet informa-
tion. The magnet database will be accessed by magnet scientists at the SSCL, other HEP
laboratories, and industry. The department uses a distributed and networked collection of
UNIX workstations and servers to carry out its mission.
We will present an overview of the UNIX network, along with a history of design
decisions and future plans for the system.
INTRODUCTION
Throughout the production phase in the MSD, the Test Department will be acquiring
a large amount of data for later perusal and analysis. The Data Management Group of the
MSD Test Department is developing and operating a data management system to integrate
all this data. In the system support section, our responsibility is to provide a set of useful
tools to aid in the interaction between the users and the database. At the same time however,
it is necessary to keep our system as flexible and adaptable as possible, so as not to restrict
the pursuits of the users. In fact, to the end-user our framework around the database should
be so familiar and intuitive as to be unnoticed in everyday use. In our approach to this prob-
lem, we have focused on what we consider the four most important aspects of the system:
1) Computing and Network Resources
2) Local and Remote Access
3) Data Storage and Retrieval Options
4) Additional Safeguards and Security Measures.
*Operated by the Universities Research Association, Inc., for the U.S. Department of
Energy, under Contract No. DE-AC35-89ER40486.

The purpose of this paper is to brief the reader on how we carne to our decisions on
the topics, and how we are attempting to implement those decisions. While a detailed
description will not be possible in this short paper, reference documents are available from
the MSD Test Department archive on a number of the topics l .
SYSTEM OVERVIEW
Computing and Network Resources
The applications involved in this system will have very different and possibly very
extensive needs. At the SSCL, a distributed processing environment has been a common
solution to this problem2. To ensure that the requirements of each will not conflict with the
other, or together ruin the performance of the entire system, it is necessary to divide these
processing needs into discrete tasks and assign sufficient computing/network power to han-
dle each task on a regular basis. We have decided to make this division of resources between
the database services and the application programs which use that data. Our servers will be
assigned the task of managing the database and providing file service to the rest of the net-
work, while the individual workstations will be required to handle all user interaction and
data analysis. This ensures that applications will not be in contention for processing time
Database
Server
(TO InterNet)
Desktop UNIX
Workstations
Figure la Current Layout of Network
1016
with the database system. As a side effect, it also allows us to select software and hardware
for our applications without being restricted by the requirements of the database system, and
vice-versa.
Another decision, made to gain flexibility in our choice of application hardware and
software, was to adopt UNIX as our primary operating system and X as the basis for selec-
tion of graphical user interfaces. While both have been criticized for various philosophical
reasons, a greater segment of the commercial and scientific community is adopting them for
their own uses. As such, we view UNIX and X as adhoc standards that give us a broad spec-
trum of choices when we look for products to fit our needs. X in particular has influenced a
number of decisions in our ongoing design of the interfaces used to access our system.
Local and Remote Access
By moving the application processing away from the database server, we have given
the users the full extent of our workstation network's processing power. In the past this ser-
vice would have been available only to users on-site. However, our choice of X enables us
to go a step further and provide the same functionality to users at remote locations. X allows
a user to run an application on one computer with the user interface on a completely differ-
ent system. To achieve this level of transparency to the user, X communicates using a proto-
col that handles the graphical events necessary for displaying the information, instead of the
information itself. In this way, the excessive traffic that would be necessary to communicate
our database information to other sites is removed, and is replaced with a subset small
enough to present a fast interface without a dependency on physical location. Furthermore,
because the underlying X protocol has been implemented by a large number of vendors, the
interface can be displayed on very different architectures with little or no visible change.
One side effect of this decision has turned out to be one of the greatest benefits of the
system. Because of the commonality of X implementations between various architectures,
we are able to give all of our users a consistent, familiar interface. Our applications can be
designed with a single GUI (Graphical User Interface), instead of customized for each spe-
cific environment. This is a tremendous help both to software engineers, who can focus on
more critical design issues, and to end-users, who are not forced to relearn the interface
when they move to a different computer or lab.
Data Storage and Retrieval Options
Although the choice to create this centralized database has a number of advantages
that make it a vital addition to the SSCL,3 it also has a number of inherent problems which
we will need to address.
Size. When fully implemented, the size of our database, estimated at around half a
terabyte (500,000 megabytes) of information, becomes a very restrictive criterion for select-
ing storage systems. While a number of technologies can handle this amount of data, each
has its particular drawbacks that become more apparent when we examine our other storage
problems. To interact fully with the database management system (DBMS), it is desirable to
provide a storage system that will appear as a single entity to the software. All selection of
files from the media must be transparent to the DBMS, allowing the greatest flexibility in
our development of data retrieval and manipulation routines. No currently available storage
option can handle our data requirements without the use of a collection of similar devices,
known either as an array or a jukebox. Such devices must also provide a transparent soft-
ware interface to integrate and manage the collection of media.
1017
Lifetime/Reliability. Once of the strongest arguments for optical storage has been
the extended shelf-life and reliability of its media. It is necessary to ensure that the data will
still be intact at the end of the SSCL's 30-year lifespan. Magnetic storage is not currently
capable of meeting this requirement without renewing the media every few years.
With both alternatives it is still necessary to investigate methods for back up. Once
again optical seems to have an edge, as we would be able to copy the data from one set of
optical media to another for back up purposes. With magnetic drives, however, we must find
a separate system for back up, using magnetic tape or, possibly, a small optical WORM sys-
tem.
Access. Because of the nature of the magnet test database and the uses it will be put
to, it is necessary to investigate methods of making all the data available at once. This leaves
us with two options: 1) Use a set of optical drives (known as a jukebox) to hold the raw data
and keep the database structures and header information on magnetic storage, or 2) Use a set
of magnetic drives linked together (known as a disk array) to hold all the information. The
advantages and drawbacks were fairly clear. An optical jukebox, while the most cost-effec-
tive solution with the longest media shelflife, can be intolerably slow when switching
between media. Also, while the technology has been around for a while, a standard format
has not appeared in the marketplace, leaving us with doubts about the future of any system
we purchase. A bank of magnetic drives, on the other hand, would provide us with the speed
and transparent functionality that are a necessity for our project. All drives are on-line at the
same time (in contrast to the optical storage, which requires switching between media at
times) and may be accessed with no intermediary layer of software - only the normal UNIX
I/O commands (another failing of most optical systems). In fact, the only drawbacks we see
in a disk array are its cost and its need for a separate back up system. At the present time, we
are still investigating both of these alternatives, though we are leaning toward a solution
using a disk array.
Additional Safeguards and Security Measures
A final concern that has arisen during the initial implementation is how to ensure the
integrity of our system, whether against accidental damage or from intruders with malicious
intents. Unfortunately, because of the open philosophy of this system, we cannot simply
close our doors to the outside. We must provide service to a very large user base, while
somehow protecting against trespass. Additionally, we cannot restrict the users' environ-
ment without directly impacting their work.
Currently we are still providing a very open system. An account is available with a
publicly available name and password. While the functionality of this account is restricted
in a number of ways, the user still has the ability to access the database and generate queries
and reports. Additionally, should a user need further access to the system, he may arrange
for a personal account. These are supplied on a one-person!one-account basis, in order to
maintain a line of responsibility for the account.
Over the next year we plan to obtain another server and our mass-storage solution.
At this time we will retire one of our older servers to a role as the remote account server and
mail host. In this way, it will become a "firewall" against the outside, providing a level of
security that can be broken without any adverse effects on the rest of the network. While it
will have the ability to read the database and other system files, it will have no ability to
move to other systems or to write to the database. In all other respects, however, users will
have accounts with the full functionality of a normal UNIX account. We think this will meet
our modest security needs without hampering users.
1018
UNIX File
Server
Gateway
Server
EtherNet Backbone
(To InterNet)
--------------i
I
I
I
I
I
I Database
I
I
I
Server
I
I
__- ,
I
I
I-- -~------------~,
I I
1----___
I
____________ J I
Desktop UNIX
Workstations
Database
Storage
Figure 2 . Planned Additions to Network
FUTURE PLANS/CONCLUSION
Unfortunately, after all our consideration we may still be forced to come to some
compromise because of time constraints. Over the next two years we can expect 40-50
gigabytes from the magnets produced. With a number of computer vendors asking us to wait
just a little longer for their new announcements, we have reached a point where we cannot
afford to stay our decision anymore. Additionally, because of budget constraints, any system
we obtain now must either be expandable to or already have the full capacity we will even-
tuall y need.
In response to the changing and unreliable marketplace, we have to accept a prod-
uct's status as a standard as our primary criterion. In light of this, it is almost certain that we
will pursue some sort of magnetic disk array system as our primary database storage, with
the possible addition in the future of an optical WORM or magnetic tape back up system. In
any event, it seems clear that some combination of the two technologies is necessary to meet
the requirements we have laid out.
1019
Our networking situation shows more promise, with a general acceptance by ven-
dors of a standardization of network protocols and graphical user interfaces. In addition,
with SUN Microsystem (our most common system) migrating to its new operating system, a
System V-based UNIX, we find ourselves rapidly approaching a common environment on a
number of levels. With careful planning in the next two years, we should find ourselves in
an optimal marketplace for fully implementing the computing environment we have
planned.
REFERENCES
1. Magnet Systems Division Test Department, MTL Documents 004-005, sse

Laboratory.
2. sse Laboratory, "Information Resources Management FY 1994/1998
Long-Range Site Plan."
3. MSD Test and Data Management Group, "The Magnet Database System."
1020
INVESTIGATION OF AZIMUTHAL STRESSES DUE TO WELDING AND
COOLDOWN
Melvin Lindner and Steve Mulhall

17 Cornell Avenue
Building 902A
INTRODUCTION
For some time, it has been known that a prestress in the cold mass skin surrounding
the yoke was desirable for the prevention of unwanted "loose" coil effects on the
performance of magnets. From practical experience, it was determined that the hoop
stresses developed by seam welding were adequate to provide the desired preload on
the coils.
The tests reported here were intended to:
1. Determine the stresses developed by welding.

2. Attempt to determine the friction coefficient between the yoke
laminations and the stainless steel shell.
3. Determine the effects of repeated cooldown cycles on the stresses in the
shell.
TEST PROCEDURE
An 18 inch long model of a dipole was constructed using laminations, collaring keys,
coil sections and precision rolled Type 304 stainless steel shells, as similar as possible
to the actual configuration of RHIC dipole magnets constructed at BNL. Strain gauges
were attached to half the outside circumference of the skin at the mid-point of the
model, at intervals of 15 degrees, and the strains at these points were recorded.
The test procedure consisted of welding the skins in the same sequence as the
assembly practice for complete magnets. The skin halves were first clamped around
the assembled core components, and the strain gauges zeroed to establish a baseline.
TIG welding procedures were used for all welding operations, and strain readings were

~ 3r---+----r--_+--~--
x
~T 2~--~~~---+--~~
Z
~
I-
I r---+-~~~_+--~----~--+_--~--_+~~~~~~~~~
00
06 15 30 45 60 75 90 105 120 135 150 165

WELD MIDME
AZIMUTH ANGLE-DEG
Figure 1. Azimuthal Strain in Cold Mass,
50
/ f1
,(10 ~J
40
/ ~
iJ-
30
0 30 K
!.
20
II
U
10
00 I 2 3 4 5
STRAIN-IN/lN .. 10-3
Figure 2. Stress-strain Diagram, Annealed 304 SST.
50
40 "
ii5 30,
""
en
I
.... -....
en 20
~
I-
en
10
00 15 30 45 60 75 90 105 120 135 150 165 180

WELD MIDME WELD
AZIMUTH ANGLE - DEG
Figure 3. Azimuthal Stress in Cold Mass.
1022
taken after tack welding of the shells, after the root welding pass, and after the fill
pass. The weld seams were then smoothed, and strain gauges were attached on the
weld seams. Reference strain gauges were bonded to unstressed pieces of 304 stainless
steel to establish the temperature correction for the gauge readings.
After all welding operations were completed and the strain readings recorded, the
model was lowered into a container and covered with liquid nitrogen. Strain readings
were taken after the temperature of the model stabilized. The nitrogen was then
allowed to evaporate, and readings were taken after the model stabilized at room
temperature. The cycles were repeated for a total of three cycles, and readings were
taken after each half cycle. At the conclusion of the temperature cycles, the skin was
cut open at the mid-point of the uninstrumented side of the model, and residual strain
readings were taken to determine if there was plastic deformation of the shell. No
attempt was made to determine longitudinal strains.
RESULTS
The plotted data of Figure 1 show the strains as measured after the indicated steps in
the experiment. The model behaved very nearly as expected, and the recorded data
showed good correlation from cycle to cycle. As expected, there was some slight shift
during the first cooldown cycle, but the changes were not repeated during subsequent
cycles.
Some anomalies were found in that the measurement did not follow a smooth curve,
but this is attributed to small deviations in the shape of the shell. Large data scatter
is apparent at the 0 degree weld position, but this was found to be due to a local
distortion of the shell at that point. The shell showed a small flat configuration, rather
than a smooth radius, that put the weld seam into bending rather than tension. Data
scatter at this point made this one gauge unreliable. The readings on the gauge at the
180 degree point showed good repeatability and are considered valid. A free condition
strain reading at this seam indicates that plastic deformation occurred at some time
during the tests. Since it was not possible to instrument the weld seam during welding,
it is an educated guess that the deformation was caused by stress levels beyond yield
at cooldown, most likely during the first cycle, since the experimental data does repeat
in subsequent cycles.
The strain data was converted to stress using the stress-strain diagrams of Figure 2, and
is plotted in Figure 3. Accurate data regarding low temperature stress-strain
relationship is hard to find, and the curve shown was derived from elastic modulus and
yield strength data. It should be noted that, while the data is presented for annealed
material, the shells have been cold worked to some extent, and the analysis should be
taken as conservative.
Curve fitting techniques using the relationship:
with tensile inputs (To) applied at two places diametrically opposite, were used to
derive friction coefficient effects, and a coefficient of between 1.0 and 1.2 appears to
be appropriate. This value corresponds to values given in several reference books for
1023
dry steel-to-steel combinations, and is reasonable considering the clean and dry
components and the effects of the fine blanked lamination texture rubbing on the
stainless steel surface.
CONCLUSIONS
Azimuthal (hoop) stresses in the shell vary with azimuthal position around the shell.
This variation is directly correlated to the coefficient of friction between the shell and
the laminations. The stresses are generated by shrinkage during the welding
operations, and are increased by differential thermal contraction during cooldown.
Stresses on the order of 35-40 KSI are easily obtained at room temperature, and may
rise to 45 KSI at low temperatures. Extrapolated data indicates that weld seam friction
coefficient estimate, derived from the experimental data, may be used as an input for
the ANSYS analysis of the yoke and skin of the magnet assemblies.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the constructive suggestions made by M. Rehak

and R. Shutt with regard to analysis of the data, and to R. lackimowicz for doing the
experimental work.
1024
DESIGN OF A SENSITIVE CALORIMETER FOR USE IN AC LOSS
MEASUREMENTS
G. T. Mallick, Jr., J. M. Toms, W. J. Carr, Jr.,

and H. C. Cooper
Westinghouse Science and Technology Center

INTRODUCTION
Losses which may be induced in superconducting conductors are of great

importance in the design of superconducting machinery. These losses result from
magnetic fields which vary in time being imposed upon the conductor. They may be
measured indirectly by measuring the area of hysteresis loops generated over a magnetic
cycle,! or directly by calorimetric methods. 2 Generally, the calorimetric method observes
the boiloff of gaseous helium, and relates the volume of gas generated to the amount of
heat provided by the sample. This article describes a variation of the latter method which
provides great sensitivity in that tens of microwatts of dissipation may be detected. In
addition, it requires little integration time, so its speed of response is quite fast.
CALORIMETRY REVISITED
The usual calorimetric method is described nicely by Wilson. 3 The main problems
associated with it include lack of sensitivity, difficulty of implementation, and a slow
response time. These problems all arise from the necessity of making the required
measurements with room temperature instruments, sensitive pressure gauges, for instance.
These instruments require the conduction of compressible gas from the cryostat through
piping of substantial volume into the laboratory environment. If the amount of helium
gas evolved is small in comparison to the volume of the piping, obviously the sensitivity
of the instrument is limited. Furthermore, the time required for the gas to equilibrate at
room temperature is finite, and this limits the speed of response of the typical calorimeter.
DESCRIPTION OF THE CALORIMETER
If a small carbon resistor suspended in liquid helium suddenly has electric current
passed through it, the effect on the helium is virtually immediate. If one visually
observes such an experiment (carried out in a glass dewar, for instance), it is also evident
that the turbulence observed is indicative of the power dissipated in the resistor. The
turbulence observed, after all, is just that helium gas which is collected in the
1025
conventional calorimetry experiment described above. The bubbles which comprise the
visual turbulence are collected in a gas reservoir, and eventually conducted outside of the
dewar for quantitative measurement.
Turbulence in liquid helium has previously been quantitatively measured to

indicate the presence, and to quantify, two phase flow for magnet cooling systems.4
Fiberoptic cables were used in that measurement, and the average transmission of light
through the turbulent stream was monitored to indicate the relative amount of gas/liquid
present. Eikelbooms used a reservoir for gaseous helium in immediate proximity to the
superconducting sample and LEDs as sensors in an ac loss experiment. Results of that
experiment indicate good accuracy over the range of 1 mW to 100 mW. The requirement
that the reservoir be mechanically emptied on a regular basis is awkward however. This
work was undertaken to circumvent this practical difficulty.
~ TO ELECTROI'ICS
OPTICAL RECEIVER
/
, ; ' FeEROPTIC CABLE
~=======n
° OPTICAL TRANSMTTE~
LIO\X) I - H U M - - o
BUlBLES
SAr-I'Lf CHN'1lER
Figure 1. Pictorial representation of the calorimeter.
Rather than collect the gas bubbles in an inverted test tube, we channel them into
a precisely machined chimney, and observe the individual bubbles, or groups of bubbles,
as they escape from the sample container, as shown pictorially in Figure 1. Bubbles
break a light beam conducted to the apparatus through a fiberoptic cable. Eikelboom
remarks that individual bubbles may be seen fiberoptically. They are quite easy to
observe if a careful attempt is made to do so. Care must be taken to properly align the
fiberoptic cable and the chimney, since the bubbles are very small, and the diameter of
the fiberoptic cable is only 100 micron. The critical alignment was obtained by utilizing
standard fiberoptic connectors. These items are manufactured to close tolerances to hold
fiberoptic cable precisely. The same connectors which hold the fibers in alignment also
serve to protect the rather delicate glass cable. While standard components were used,
they were modified (mainly trimmed to length) to fit the rather small space which the
calorimeter had to occupy. A photograph of the completed unit is shown in Figure 2.
The alignment of the cables is more critical than the spacing of the gap through which
the bubbles rise. The present spacing is 0.103 inch. With that spacing, an optical loss
of approximately 25 db is noted across the gap. The optical loss elsewhere in the system
is negligible. The light beam is generated by a standard (Honeywell HFE-4807-014)
infrared LED source, and detected by a standard (Honeywell HFD 3843-(02) optical
receiver. The fiberoptic cable is standard Siecor 100/140 communications fiber. This
1026
Figure 2. Photograph of the calorimeter.
fiber has been rugged and dependable. No problems have been noted after some dozen
thermal shocks to liquid helium temperature. The only metal in the system is the wire
leads to the calibration resistor, which occupy very little volume. This was of great
importance in the initial application of the device, which involved the measurement of
losses of superconducting cable subjected to a changing magnetic field.
RESULTS OF EXPERIMENT
The electrical signal which results from the passage of a bubble through the beam
is not a simple unidirectional pulse, which might be expected if the bubble simply
occluded the beam. It rather tends to be a bipolar pulse, with the intensity rising above
normal during the entry and exit of the bubble in the beam path. This must be taken into
account in the design of the electronic detection system. Because of the random way in
which bubbles form and migrate, there is a statistical nature to these pulses which we
observe. The frequency of occurrence is related to the number of bubbles per unit time
rising through the chimney. When the pulses are applied to a loudspeaker, the noise
which results is very reminiscent of that which is associated with the detection of
radioactivity with a Geiger counter, and in fact the detection circuitry is similar. The
output of the unipolar pulse is delivered to an integrating circuit which has a time
constant on the order of a second. Thus, the integrator gives a resulting dc voltage which
is proportional to the number of bubbles per second rising through the chimney. Since
bigger bubbles also tend to deliver bigger pulses, the integrator tends to integrate the
volume of helium gas passing through the chimney per unit time. This quantity should
be related to the power dissipated in the sample.
The mechanics of the bubbling process is fascinating. 6 Our experiments indicated

that individual bubbles produced in the boiling process at low power levels are so small
that they are not reliably observed. For that reason, we provided a chamber in the
chimney for the tiny bubbles to collect in before they escape up the chimney as a bigger
bubble, which is easy to see. The process is not unlike that which is observed in the
escaping of CO2 gas from the airlock which is used in the fermentation of (homemade)
beer or wine. In fact, that process served as the inspiration for the present work. The
chamber is provided by tapering a large diameter hole to a smaller one.
1027
r:
:. . .:. .f/::;
8.85
25.8s 58.8s
TIME
CAL OUTPUT VOLTAGE Xl

Figure 3. Response of the calorimeter to a 10 second 25 microwalt pulse. Upper trace is calorimeter output,
uncalibrated. Lower trace is power input.
The prototype unit was tested by applying a measured amount of power to a

resistor contained in the volume monitored by the sensor. Figure 3 shows the response
of the sensor when a 10 second square wave power pulse of 25 microwatts is applied to
the resistor. At this low power level, the time delay is on the order of two seconds, and
the signal is unmistakable. Higher power levels monotonically increase the output level
with decreasing sensitivity as shown in Figure 4. It is possible that the nonlinearity may
be due in part to a change in the mechanism of heat transfer from the calibration resistor
to the helium bath as power increases. The formation of an insulating layer of helium
gas on the resistor would cause this type of behavior.
3.00
W 2 . 50
C)
«
~2.00
o
>
f- 1.50
::J
(L
~1.00
o
0.50
0. 001-r-~~TTT~"T"T"'~"'T"rT~~rrr-T"rT~"
0.00 13.00 20.00 30.00 40.00
POWER INPUT IN MILLIWATTS
Figure 4. Response of the calorinleter as a function of power input.
The response of the system depends on the pressure in the system, and modulating
the pressure also modulates the output Fortunately, rapid changes in pressure in many
systems are not likely to occur, and the system may be easily recalibrated on a regular
basis by applying power to the calibration resistor.
1028
Figure 5 shows results of a practical application of the device, with the dots
indicating the integrated density of bubbles produced by a sample being heated by a
unifonnly changing (bipolar ramp) magnetic field. The statistical nature of the output is
clearly demonstrated, but the average density of the resulting pattern is unmistakable.
Clearly there is a peak in the output near zero magnetic field. This is a consequence of
the characteristics of the material.
O.5,---------~----------~
Cl::
W
:3
o
0.- 0 . 0 +-----+-----;'
W
>
-1.0+.~~~~"h.~~~~~
-J
FIELD (Tes I el
Figure S. Practical data Laken with the calorimeter.
CONCLUSIONS
While the prototype unit is far from being completely developed, the results
obtained so far show clearly that the technique described in this paper is capable of
extraordinary sensitivity, fast response, and wide dynamic range. Relatively little effort
was expended in optimizing parameters, either in the electronics or the mechanical
system. It is likely that some very nice optical calorimeters could be made using this
technique.
REFERENCES
1. G. T. Mallick, Jr., J. M. Toms, W. J. Carr, Jr., G. Snitchler, V. Kovachev, and R.

Jayakumar, "Results of ac loss measurements on SSC conductors." Proc. Third
International Symposium on the Super Collider, 1992.
2. M. Ashkin and G. R. Wagner, "Comparison of calculated and measured hysteresis loss

in multifilimentary superconducting wire," JAP 60:2477(1986).
3. Martin N. Wilson, "Superconducting Magnets," Oxford University Press, New York,

NY 1983, p. 252.
4. T. Harayama, "Optical method for measurement of quality and flow patterns in helium
two-phase flow," Cryogenics 27:450(1987).
5. J. A. Eikelboom, "Test results of an apparatus for calorimetric measurements of ac

losses in superconductors." IEEE Trans. on Magnetics, 28:817(1992).
6. Neil E. Shafer and Richard N. Zare, "Through a beer glass darkly," Physics Today
44:48(1991).
1029
TEMPERATURE DEPENDENCE OF THE VISCOELASTIC
PROPERTIES OF sse COIL INSULATION (KAPTON)
Amy Lautenschlager, Finley Markley, James McAdams,

and Barbara Sizemore
Fermi National Accelerator Laboratory·

Box 500
Batavia,IL 60510
ABSTRACT
Tensile creep of Kapton insulation was measured at 5°C and 55°C.
This was done to investigate the time temperature shift factor. From earlier
work, it was thought that when an instantaneous load was applied the
resultant creep could be divided into two parts. There was a time
independent and a time dependent portion. At two different temperatures,
the time scale of the creep should differ by a constant k, the shift factor. The
shift factor would be used to determine time-temperature equivalents to
extrapolate the stress loss during the storage life of the SSC magnets before
the are put to use and thus whether a refrigerated storage area might be
necessary. We have found that within the methods used for earlier work in
this area, a constant shift factor may not be possible to determine.
Compressive Stress Relaxation was also studied at 38°C and 55°C for the
same purpose, but was expected to be a more difficult measurement.
INTRODUCTION
It is our goal to understand how the compressive stress in collared coils

decays with time, especially during the storage of magnets at room
temperatures or even at higher than room temperatures in an uncooled
warehouse prior to installation. We wanted to determine the time-
temperature conditions acceptable to maintaining enough compressive
stress to prevent coil motion and the quench that might result. If the coil
quenches, it loses it's superconducting capabilities; it immediately heats up
and the accelerator is shut down. The coils are made of superconducting
wire, Kapton and fiberglass, and it was reasonable to expect the Kapton to
undergo the greatest creep.1
• Operated by the Universities Research Association, Inc., for the U.S. Department of Energy.

Although the compressive stress is perpendicular to the plane of film in
the SSC coils, it was thought to be more accurate to measure temperature
dependence in tension, parallel to the plane of film. It is more accurate
because sample length is precise, stress is constant, stress application is
more rapid and compensation may be done for thermal expansion. It was
assumed that the temperature effect in both cases was the same, but we
measured at two temperatures in both cases to be sure.
APPARATUS
Figure 1 shows the fixture design for the creep experiment. The samples
consisted of a strip of .0051 cm thick Kapton .635 cm wide and 38.10 cm long.
The top of the Kapton was epoxied to a lucite rod that was used to secure the
Kapton in the apparatus. The core of a linear variable differential
transformer was attached to the other end of the sample. A .454 gram weight
was attached to the end of the core. The transformer coils were held in place
by a Kapton tube that was rolled and epoxied together. Kapton was chosen
because it canceled any difference in effects of thermal expansion of the
sample and the tube that holds the transformer coils. The whole apparatus
was enclosed in a double walled acrylic tube with circulating water. This
was done to control the temperature inside the apparatus. The water was
supplied by one of two commercial water baths depending upon the desired
temperature of the current test. In the past, there were concerns that the
changing humidity level caused inconsistent results. Kapton can absorb up
to 3% its weight in water at 100% relative humidity. To avoid this, silicon gel,
a drying agent is set in the bottom of the experimental apparatus to ensure a
humidity close to zero at 55°C. The samples are stored in an oven when they
are not being used to reduce the amount of water present within the samples.
These two factors should greatly reduce any change of humidity effects.
THEORY
One theoretical model of tensile creep is that when an instantaneous load
is applied, strain could be divided into two parts. There would be a time
independent strain and a time dependent strain. The time independent
strain could be thought of as the initial strain because it should happen
immediately after the load is applied. The time dependent strain was
thought to be a function oftn where n is a constant and t is time. 2
E(t)=EO+Elt n (1)
Both sides of the equation could be divided by the stress, S, to be in terms

of creep compliance.
(2)
To find the time-temperature shift factor, creep compliance is measured
at two different temperatures. We expected that the resulting equation of the
experimental data would only differ by a factor of k, where k is a coefficient of
time dependent creep.
(3)
1032
LUCITE ROD
CLAMP
LUCITE TUBES
KAPTON STRIP
KAPTON TUBE
L.V.D.T.
CORE
LUCITE ROD
WEIGHT
SILICA GEL
·0· RING
Figure 1. Kapton Creep Apparatus
1033
It is with these ideas in mind that we ran our creep tests and tried to fit
our data.
The experimental data was fit to the theoretical model using CERN's
MINUIT program. It uses the least-squares method of function
minimization. MINUIT consistently gave the best fits to the data so it is used
for all the tests.
Two varieties of creep tests were run. The first is a series of five 60
second tests all completed using one sample within a single day at one
temperature. The sample was loaded into the apparatus at least 24 hours
before the first test was begun. This was to assure that temperature was
stabilized. Tests were run approximately every hour. Data was taken every
0.0167 seconds. Data acquisition was started just before the weight was hung
in order to get a zero position number. The first data recorded after hanging
the weight is not the time independent creep due to the fact that it is
impossible to get an instantaneous load applied. At this point, loading time
has been reduced to approximately 0.4 seconds for the 0.454 kg weight.
Three different samples were tested at both 5°C and 55°C. At 5°C, the
repeatability of creep for tests of the same sample ranged between ±O.O% and
±2.7% for the three different samples. This is calculated as the average and
average deviation of the creep of a given sample at a certain time since the
weight was hung (31.5 seconds). The repeatability between samples was
±1.8% at this temperature. .
At 55°C, the repeatability of the creep ranged from ±0.7% to ±1.4% within
the same sample. The repeatability between samples at this temperature
was ±1.4%.
These results show that there is definitely a difference in creep between
the 5°C and the 55°C tests even at 60 seconds or less, although in our
theoretical representation the difference seems to be mainly in J 0 The tests
have been normalized to zero creep before the load was hung and shifted in
time to match the initial drops. Figure 2 shows those results for three
samples.
To check our curve fitting by MINUIT, we also used the following
method. A series of values of Jo were selected and for each one a data file of
In (J - Jo) versus In t was generated a least squares fit of a straight line was
made to the data. J 1 and n were taken as the intercept and slope of the line.
The sum of the squares of the errors between the line and the data was
computed. Then a graph of the sum of the square of the errors versus Jo was
constructed. This showed a clear minimum in the sum for a particular Jo
which was selected as the best J o. The parameters obtained by this method
agreed well with those obtained from MINUIT.
The second battery of tests were run for approximately one week. Data
acquisition was started just before the weight was hung, as with the shorter
tests, to record the zero creep point. Data was taken every 0.0167 seconds for
1000 data points, then every 30 second for 1000 data points and finally, every
600 seconds for 1000 data points. This caused a weighting factor within the
data that MINUIT was modified to handle.
Figure 3 shows week long tests at both 5°C and 55°C and figure 4 shows the
results for a second sample. Unlike the short tests these show a clear
difference between J 1t n hot and cold as well as a J o differences. Unfortunately
the curve fits yield hot and cold results with different values of n which implies
that a constant value for k does not exist.
1034
=
2.00
= ;V:; :i;:
55 O£G. C
10 1.50 -
I 5 DEG. C
""
""oz
'I(
:::; 1.00
Il..
:::E
o
o
0..
,, ,,
:::: 0.50-
a::
o
,, , I
•..•
---------------------------------.
I I
0.00 ....,.~~ttf'I-
" ...
""'" o 10 20 30 40
TIME IN SECONDS
Figure 2. Short Term Creep
(:;)')..0
~§)00
~Jl
,;flO')....
,. •• ~ •• - - . -
-_... -----,. __ .._- ...-- ...-- -_ ...-'" --- --_ ...-
55 DEC. C
""~ ----------.~ ...

(:;)')..').. .. :,;""
'I(
:::;9
000 ......... .",
Il.. .;''''
:::E I
0 0')..0 I
0 ~§lOO
5 DEC. C
Il..
""a::
IAI (:;),'b
9000
(,)
0,0
9000
,"
90000 0 100000 200000 300000 400000 500000 600000
TIME IN SECONDS
Figure 3. One Week Creep Sample "Jane"
1035
,'),f>
,'),t.
.................................................. ···················55 D£C:··C········
W
~,'),'),
........................... , .................... -
..............
. .... ....
"
-----------_._---------_._
8,'),0
~
....
4-
,.
,.-----------
4-
5 D£G. c:
W
5"~
"f>
" . 0 100000 200000
Figure 4. One Week Creep Sample "Xavier"

TIWE IN SECONDS
400000 500000 600000
STRESS RELAXATION EXPERIMENTS
The apparatus and methods used for our continued investigations on

Kapton Stress Relaxation is the same as was described at last year's
IISSC.1 We chose to run a series of tests at the two selected temperatures
(38 C and 55 C) and compare the resulting data sets, which are shown in
Figure 5.
The function minimization program MINUIT was used to
parameterize the data set to an equation of the form L=LO-LUn . The results
of this curve-fitting is still in progress.
1.00
0.95
"V
0
0 0.90
~
~
£ 0.85 ~
\~ -
--
........
"V
~
0
0 0.80
"-. =====
0.75
0.70
o , 00000 200000 300000 400000 500000 600000 700000 800000
Time (In seconds)
Figure 5. Kapton Stress Relaxation
1036
CONCLUSION
We have found that the model J=JO+Jot n for creep does not give the
same value if n at our two temperatures and therefore cannot yield a
constant time scale factor. A temperature dependence of creep certainly is
apparent from our data, but we do not yet have a good model to explain it.
ACKNOWLEDGEMENT
We would like to acknowledge the work of Jay Kreibich who started the
stress relaxation measurements being continued by one of us (J.
McAdams).
REFERENCES
1. Markley, et aI, "Stress Relaxation in SSC Collard Coils and their
Component Materials", Supercollider 3, Vol. 3, Plenum Press, New
York, 1991, pp 155 - 163.
2. Carson, J.A. and F.W. Markley, "Mechanical Properties of
Superconducting Coils", IEEE Transactions on Magnetics, Vol.
Mag. 21 #2, March 1985, pp. 85.
1037
MEASUREMENT OF TIlE ELASTIC MODULUS OF KAPTON
PERPENDICULAR TO THE PLANE OF THE FUM
AT ROOM AND CRYOGENIC TEMPERATURES
Mark Davidson, Steve Bastian and Finley Markley

P. O. Box 500
Batavia, IL 60510
ABSTRACT
Understanding the short term elastic properties, i.e., the instantaneous
modulus of Kapton is essential in determining the loss of prestress during
storage and operation of SSC dipole magnets. The magnet prestress
contributes directly to the coil response to the Lorentz forces during ramping.
The instantaneous modulus is important in extrapolating short term stress
relaxation data to longer times. Most theoretical fits assume a time
independent component and a time dependent component. The former may
be represented by the Kapton modulus near zero K where all relaxation
processes have been 'frozen'. out. 1 Modulus measurements at 77K and 4.2K
may point to a correct value for the near zero K modulus. Three companion
papers presented at this conference will be: 1) "Stress Relaxation in SSC
50 mm Dipole Coils" 2) "Temperature Dependence of the Viscoelastic
Properties of SSC Coil Insulation (Kapton)" 3) "Theoretical Methods for Creep
and Stress Relaxation Studies of SSC Coil."
INTRODUCTION
Kapton is the primary insulator in SSC dipole magnets. The

understanding of the material properties of Kapton is crucial in determining
the loss of prestress in the collared coils. A general equation for creep (or
stress relaxation which has the same general format, see companion paper
3, for further details on creep equations) is:
E = Eo+Eltn (1)
This equation describes the time dependency of creep which is (stress held
constant and strain varies) and stress relaxation (strain held constant and
* Operated by Universities Research Association, Inc., for the U.S. Department of Energy.
Supercollider 4, Edited by I. Nonte

stress is allowed to vary). This equation has several constants that can be
determined experimentally. The Eo in Equation 1 is the modulus at time
t = 0, and can be called the time independent modulus the E It" term the time
dependent modulus. Eo may be determined experimentally by finding a way
to eliminate the effects of time. There are a couple of ways this can be
attempted. One would be to measure a modulus by loading a piece of Kapton
film in tension as fast as possible while taking stress strain data. 2 The other
method (this author's method of choice) was to find the modulus at several
temperatures close to 0 K (although Du Pont has low temperature modulus
data, this was done in tension and not in compression) and this along with
data collected at other temperatures was used to find modulus as a function
of temperature and the modulus extrapolated to near 0 K. This 0 K modulus
could be synonymous with the instantaneous modulus. This is because the
modulus at this temperature has no time dependent portion, i.e., the time
dependence of the modulus is 'frozen' out at 0 K. In normal modulus
measurements, while a material is being stressed it is also creeping. The
slower the material is stressed the more time it has to creep; the faster it is
stressed the less time it has to creep, therefore, the faster the stress rate the
less creep will show up in the modulus measurements and vice versa. The
freezing out principle attempts to eliminate this recovery while stressing at a
rate that is not considered 'fast'.
TEST MATERIALS AND PROCEDURES
The Kapton samples used were two stacks of 100 12.7 mm squares of
0.127 mm film (stacks R1 and L2). This film was used because it is similar to
the film used in the actual coils (i.e., 0.0254 mm film was used in the actual
coil. We used 0.127 mm film because less were needed to make the stack).
The stacked configuration was utilized because taking stress strain
measurements perpendicular to the plane of a single piece of 0.127 mm film
is very difficult at best. Although Kapton is available in a solid form (Vespel)
this was not used due to the manufacturing techniques of Vespel. Because of
these techniques Vespel's macro structure is ditTerent from the Kapton and
it's material properties would not be representative of the film. The tests
were done perpendicular to the plane of the film because this is how the
stresses are applied to the Kapton in an actual coil.
A ±3.81 mm MTS Corporation extensometer calibrated at 300, 77 and 4.2
K was used to measure the displacement of the stack during loading. The
extenso meter was designed by MTS to operate in a liquid nitrogen
environment, but we calibrated it for use in liqUld helium. (It was calibrated
in an apparatus designed by J. Mashek). See Figure 1.
A specific fixture was designed to hold the Kapton stack during loading.
The walls of this fixture were spring loaded to aHow for the expansion of the
stack perpendicular to the applied force to avoid measurement of the bulk
modulus. The fixture also had a guided loading ram, allowing forces to be
applied to the stack without slipping of the ram (see Figure 2).
The sample was tested on two testing macilines. The first being the
Instron 8500 servohydraulic testing machine, a fully automated computer
operated servo driven testing machine. This rna ::hine was used to measure
the stack height of the Kapton (as explained in tJ Le next section). The second
machine was the Low Temperature Loader (L'I'L), an apparatus that was
designed within Fermilab for cryogenic materials testing. The machine
1040
INSTRON ACUATOR HELIUM/NITROGEN
FI LL LINE
LIQUID LEVEL
DmCTQA
FOAM COVER
AND INSULATION
QUARTZ TUBE AND MYlAR BLANKET

ROD ASSEMBLY
DEWAR
EXTE NSOM ETER
Figure 1. Extensometer Calibrator
Looding RAM
Ex' enso meter
Sample Area
Figure 2. Test Fixture
1041
consists of a G-IO tube within a G-IO tube. The inner tube has two steel plates
shrink fitted to the end and applies the load to the sample, while the outer
tube also has shrink fitted rings on the end. A steel plate is bolted to this ring
where the sample sits during loading. This design is patented and licensed
for use by the Instron Corporation (see Figure 3). The LTL can be fitted with
a dewar that covers the sample area and allows for testing with liquid
nitrogen or helium. The press is also fitted with a 90 kN load cell for
measuring applied forces.
The force data as well as the output from the extensometer were collected
on an IBM compatible PC via an HP3457 A digital multimeter. An HP6227B
dual voltage source was used to power the extensometer as well as the load
cell.
TESTING OF THE KAPTON STACKS
Stack height measurements

Measuring the height of the stack proved difficult because of the slight
inconsistencies in thickness so careful steps were taken to measure the stack
height. The stack of Kapton was dried before the test by storing it in an oven
with silica gel desiccant (to absorb any moisture) at 400 K. Kapton is very
absorbent in regard to water; about 3% water at 100% relative humidity. The
extenso meter was then mounted to the fixture and connected to the HP
DOUBLE ACTION
HYDRAULIC CYLINDER
SUPPORT PLATE
HELIUM I NITROGEN - -t-+-- LOAD CELL

FILL LINE - - ----I
7j.-- - - FOAM INSULATION
DEWAR
+-I=t:=~-1?ff--- LIQUID LEVEL
SENSOR
OUTER G-10 TUBE

INNER G·10 TUBE ---l'---->l1'-1
SAMPLE SUPPORT PLATE
Figure 3. Low Temperature Loader
1042
voltage supply at seven volts excitation. A 12.7 mm cube of steel was then
placed into the testing fixture in place of the Kapton stack to serve as a
calibration block for stack height measurements. A reading of the
extensometer position was then taken at the 12.7 mm point. The actual stack
of Kapton was then placed in the fixture and load displacement data was
taken up to 4.5 kN. The stack height was then calculated by fitting the
straight portion of this data to a linear equation and extrapolating to zero
load. This point was taken as the top of the stack.
The stack height was also found using two other methods: weighing the
stack then using the density of Kapton from Du Pont data to calculate the
stack height, and measuring individual pieces of Kapton with a micrometer.
An average of the three methods was used as the actual stack height (see
Table 1 for further details).
Table 1. Stack Height Measurements

( millimeters)
R1 L2
Density 12.645 12.415
Micrometer 12.985 12.425
Loading 12.187 12.688
Mean 12.606 12.509
Measurement of the Elastic Modulus
The measurement of the elastic modulus was done in the same manner
as the stack height measurements. The only exception was that the test was
done on the LTL. The linear portion of the force displacement curve was
fitted to a straight line equation, and the slope of the straight line was then
converted to modulus. This was done at 300, 77 and 4.2K. The modulus data
was then graphed vs. temperature (note that thiH included the modulus from
data provided by Du Pont; see Table 2). We attempted to fit the data with a
cubic, log-log, modulus proportional to 1fT and a log E curve. The best fit
was found by taking the log of the temperature and doing a linear regression
to the data which had an R (correlation coefficient) of 0.97 which was the best
of'all the fits (see Figure 4 and Equation 2).
1043
Table 2. Modulus Measurements
(Mega Pascals).
300K 77K 4.2K
DuPont * 2966 3517 5517
1.2 2276 3310 4483
R1 2345 3379 5103
*(As expected, the Du Pont Moduli are higher than 1.2 and Rl because Du Pont tests
are done in tension, i.e. the direction of the molecular orientation of the film.)
8000~--------------------------------------------~
• Du Pont
....... + L2 1.0 '"'

'iii
III
o * R2 a.
~ 6000-
fit to data ....0
o
'5 ~ 0.8 III
c
m ~ .2
; ....
Q)
~ 'E
III 4000 •• 0.6 ........
:J
::::J
••
•••••• -t. 1/1
:J
'"0 ••••••
o •••..•••.••. :J
.... -•.................
'U
:::::E ••••••••••••••••• 0.4 0
::.:I
~ 2000- u
III :;:
.2 0.2 III
w 0
G:i
o,,----------.----------.r----------.~--------_r0.0
o 100 200 300 400
Temperature (Kelvin)
Figure 4. Elastic Modulus of Kapton
E=klogT+E
E = Eo+Eltn k = -1337 (2)
Ec = 5879
CONCLUSIONS
Equation 2 implies as T goes to 0, E goes to infinity, therefore a clear
zero K modulus for Kapton cannot be determined from Equation 2. This
follows the reasoning that at zero K there is no molecular motion, but by
definition modulus requires there be motion. The infinite modulus agrees
with the value found for the reciprocal of the instantaneous creep compliance
(companion paper 2). In the future stress relaxation and creep data will also
be analyzed with mathematical models that do not assume a time
independent component.
1044
REFERENCES
1. Caufield, Edward M., "An Investigation of Stress Temperature, and

Time Dependent Strains in a Randomly Oriented Fiber Reinforced
Composite with Special Emphasis Given to Thermal Stress
Situations," Ph.D. Thesis, University of Illinois, Department of
Theoretical and Applied Mechanics, 1979.
2. Michaeli, W., FoIster, Th., Lewen, B., and Aachen, "Simulating the
Long-Term Behaviour of Plastics," 1989.
3. Du Pont Chemicals "Cryogenic Properties of Kapton Polymide Film."
1045
AN OVERVIEW OF RECENT PROGRESS USING LOW-COST AND
COST -EFFECTIVE COMPOSITE MATERIALS AND PROCESSES
TO PRODUCE SSC MAGNET COILS AND ASSOCIATED
NON-METALLIC PARTS
John J. Morena
Composite Materials and Processes

Primary Technical Advisor
SSC Accelerator and Magnet Divisions
ACMLC
The American Composites Building
425 Califomia Avenue
Stuart, FL 34994
INTRODUCTION
Thermoplastic and thermoset polymer systems have been used in high-energy physics
applications throughout the world for many years. Like other industries and industrial
communities, the materials and processes requirements ofthese polymers have recently taken
on new meanings. New accelerators and other machines are pushing all material parameters
beyond limits. New polymeric and composite materials are being developed, invented, and
formulated, as is new process and application equipment. This is a decade of change.
Composite materials are being chosen for performance characteristics and cost-effective
processing as well.
The information that follows will note some of the recent progress in the development of
composite materials and processes for producing low-cost and cost-effective, high-quality,
non-metallic composite components for use in SSC magnets and in other accelerators.
The materials and methods for making composite molds, tools, and structural parts for
magnet coils and other components are demonstrated. New, unique, and innovative
approaches for processing thermoset polymers are presented. The formulated polymer
systems are used to form semi and structural insulators, spacers, supports, coil end parts,
blocks, housings, adhesives, and other composite applications.
"UNIVERSAL POLYMER" THEORY
It would be the ultimate machine design if we could develop a single polymeric

composite engineering material, then compound and formulate it for every application,

especially for every application in the SSC magnets and related systems. It happens that this
is just the approach that resulted in the development of the following materials and processes
information.
The goal was to identify structural thermoset polymer systems that could be used in
many material formats. This "universal polymer" theory suggests tough, highly cross-linked,
user-friendly, cryogenic- and radiation-resistant thermoset materials.
A thermoset material system of choice could be formulated and compounded as thin and
thick adhesives, pultruded continuously into lengths, compression- and transfer-molded,
prepreged, poured and cast, vacuum-encapsulated, coated, and resin-trans fer-molded using
structural preforms or short fibrous fillers.
RADIATION RESISTANCE OF POLYMERS

AT CRYOGENIC TEMPERATURES
The selection of composite materials for the extreme environments found at the SSC and
at similar accelerator or detector applications is a critical issue. Most composite materials will
degrade after many years of exposure in a radiation environment (l09 RADS), or at low
temperature (4.2 K). The intra-molecular and inter-molecular mobility of the organic
materials is greatly affected. The ability to withstand applied stresses decreases drastically,
and crack onset or chain scission of the molecular structure can result. The choice of test
methods to determine degradation then becomes very important. One must carefully select
which mechanical, physical, electrical, and other properties will be important to the successful
operation of an insulator, end part, spacer, support, adhesive, or part being designed, built,
and tested. For instance, some parts may be subject to one or more of impact, tensile,
compressive, flexural, or torsional forces.
Testing only one mechanical property in the operational environment may not result in
the proper material analysis. As an example, if chain scission occurred instead of radiation
bonding, there would be a decrease of mechanical strength and elastic modulus, affecting
tensile, flexural, and compressive strengths. If cross-linking of free-radical molecules
continued or occurred, then many properties would improve. Certainly by determining the
appropriate polymer system, one could extend the life of the polymer system in the radiation
environment by undercuring or by selective "A" and/or "B" stage processing.
Chain scission is the severing of polymer molecular chains by high-energy particles.
The high-energy particle, on the other hand, may also cause the formation of free radicals-
highly reactive polymer species-through ionization or excitement of the main molecular
chains. This in tum can lead to cross-linking between or through other interchain reactions. It
has been proven by Nasa Langley that a toughened epoxy matrix formulated using an
elastomeric co-polymer modifier undergoes further cross-linking when exposed to radiation.
The cross-linking causes degradation by changing the molecular network from a flexible,
shock-absorbing system to a stiff, brittle phase with reduced properties.
It was suggested by John J. Morena at Brookhaven National Laboratory (BNL) in
1989 that the thermoset composite materials being considered for use as SSC and RHIC
insulators, spacers, etc., be undercured and in some instances not post-cured. This would, of
course, be determined after cold irradiation and mechanical property testing. The extent of
cure would be chosen by this testing, and it would be anticipated that the physical and thermal
properties developed after partial cure would allow the magnet or other composite components
to perform as required during the assembly, molding, collaring, and operational cycles. This
would assure that the materials have the required compressive, flex, creep, interlaminar, and
1048
other strength characteristics needed to perform properly. The undercured but physically and
thermally acceptable composite material components and materials should, therefore, last
longer in service than their fully cured and post-cured counterparts. This theory is being
demonstrated in 1992 during testing at the Radiation Effects Facility (REF) at BNL.
The tests at the REF, sponsored by the Accelerator Systems Division of SSC, will
simulate the actual magnet operating environment, thereby revealing the reactive nature of
polymer responses to irradiation. For instance, irradiation of samples at room temperature
ambient, or at low elevated temperature, in an oxygen environment can result in totally
different results from the actual environment (liquid helium at 2.2 to 4.6 K). Complex
oxidation reactions with mobile segments could alter the real test results. For these and other
reasons, designers, engineers, and materials/processes specialists should not choose the same
materials used for satellite, rocket, or other space components and merely apply them to
accelerator or reactor component applications.
The thermoset composite materials chosen for these studies and SSC applications
possessed a large number of main chain ring structures and complex hierarchical morphology.
They were chosen because the ring structures provide delocalization of bonding electrons,
making it unlikely that high-energy particles will break the polymer chain easily. Selective
inorganic additives and fillers can also be used to enhance these material characteristics. Not
all additives improve radiation resistance. For example, it is known that flame retardants such
as chlorine atoms greatly reduce the resistance of the material to radiation. Reinforcement
finishes (Boron, others) can also reduce resistance of a composite to radiation.
Thermal resistance was also linked directly to radiation resistance, as the mobility of
polymer chains is directly related to their level of thermal excitation. The warmer the chain,
the more flexibility observed; the cooler, the less flexibility. It was therefore natural to choose
thermosets and even in some instances thermoplastics with a high heat distortion
temperature (HOT) or glass transition temperature (Tg). The use of thermoplastic materials.
does not, however, allows for the use of a "universal polymer" theory.
Finally, let us note that when composite materials are exposed to radiation in a cryogenic
environment, the radiation effects occurring at the cryogenic temperature may not be detected
until the composite material is warmed up. Chain scission or the formation of the reactive
species, and hydrogen removal result after cold irradiation and subsequent material warming
take place.
CHOOSING THERMOSET RESIN SYSTEMS
A "universal polymer" system must retain a greater percentage of physical and

mechanical properties after environmental exposures, as it will be produced in many material
formats for use in a semi-cured state, thin, thick, reinforced, filled and other forms. This is
especially true for organic polymers in thin sections, such as adhesives.
Tests show that Glycidyl amine resin systems are more stable than Glycidyl ether resin
systems based upon Bisphenol A. Curing agents that contain chains of ether links are
valuable for use in the Cold Irradiation Environment (CIE) only when in combination with
other hardeners. Aromatic molecular structures are therefore more stable than aliphatics when
operating in the CIE. Materials with increased or higher cross-link densities are more stable
also.
For these and other reasons, Cyanate Ester, Bis Maleimide (BMI), and other highly
cross-linked thermoset polymer chemistries were chosen as "universal polymer" candidates.
Of course those chosen were tested for "user-friendly" and "cost-effective" performance
1049
criteria. Standard thermoset polyimide polymer systems were not chosen due to their difficult
and expensive processing parameters, which include imidization and other post-forming
steps.
Trade studies were performed to identify cost-effective BMI, Cyanate Ester, and other
related polymer systems. Safety, processing versatility, ability to form and co-cure, technical
support availability by major material suppliers, cost of the formulated (not neat) resin
system, availability, recent past performance, and other reasons formed the foundation for
establishing material choices.
Safety was the most important aspect for choosing materials, because all U.S. users
and material suppliers are very aware of recent legislation from OSHA, NIOSH, EPA, and
other government agencies that establish controls over the formulation and use of various
materials in all national industries. The liability of materials choices by engineering,
manufacturing, materials personnel, or, in fact, the material supplier finally becomes the
responsibility of the program. Passing the use of carcinogenic or suspected carcinogenic
materials onto a contractor or its sub would not reduce the final liability or ethical
responsibility. In addition to these materials there are, in fact, other epoxy amine hardeners,
curing agents, accelerators, and material constituents that cause permanent health effects if not
handled using proper hygienic precautions. Recent studies have revealed that some materials
previously used in the High Energy Physics community on other accelerator programs did. in
fact, contain carcinogens.
Processing versatility was linked with materials safety to result in cost-effective
materials processing. Closed molding methods, both low-cost and continuous, were selected.
These included pultrusion, structural, and standard resin transfer molding (S-RTM and
RTM), and compression molding. Other low-pressure (5 to 100 psi) methods were also
considered. Closed molding techniques using single component thermosets (BMIs, Cyanate
Esters, others) eliminate the need to meter and mix curing agents, hardeners, etc., just prior to
use. Closed versus open molding contains the reactants of a polymer system during
processing and reduces emissions into the atmosphere.
Co-curing and forming-a very popular method and molding approach used in auto,
bus, truck, aircraft, and other industries-allows re-forming, consolidation, and simultaneous
bonding of magnet assembly or other components. Inserts, tapping plates, thermal brakes,
dividers, and many other attachments can be included during a molding cycle. Consolidation
of a number of items in a single molding cycle reduces cost.
Major material suppliers were selected to support SSC materials engineering programs.
One reason was their willingness to generate data at their own expense, and in a timely
fashion. Many of the suppliers already had the basic recent data to substantiate selection of
material candidates. Choosing major suppliers also assured a commitment to material
availability.
Cost and other issues were studied. Filling, reinforcing, and altering resin systems
reduce final unit cost, so formulated and reinforced resin components were considered as
"cost drivers."
CHOOSING MOLD AND TOOL MATERIALS AND CONCEPTS
In order to mold and shape "universal thermoset" resin systems, cost-effective mold and
tool methods were considered. To prove an engineering concept, it is not always necessary to
use "hard" (metal) tooling. There are many unique, highly filled, epoxy and other mold- and
tool-making materials available to produce a "proof of principle" component or assembly.
Mass cast tooling was used for a number of formed and molded parts and assemblies.
1050
Pole spacer
filler
Inner
saddle ---I--",.''::::I'~
Inner coil
lead or
relurn end
saddle
@ (fI
4l SOlidend
spacers '
Return end
{J
Pole spacer
Lead
Bus duct End
.JI' / Inner coil ~
#/ ~ Pote
. Spacer
Figure 1. Typical SSC and RHIC Composite Coil and Magnet Parts.
Figure 2. Outside (a) and Inside (b) Views of Fiberglass Preform (Copper Composites/GrumrnanlSSCL).
1051
Figure 3. Dipole Coil End Parts and Cross-Section View-Co-cured "B" Stage Saddle. (Courtesy of Fred
Norbrega, SSCL.)
Table 1. SSC Processes.
CNC Machining
Water Jet (Abrasive)
Extrusion
Pultrusion
Structural-Resin Transfer Molding (S-RTM), Low-Pressure
Structural-Reaction Injection Molding (S-RIM), Low-Pressure
Compression Molding
Transfer Molding
4-Axis, Computer-Controlled Braiding
Injection Molding, High-Pressure
Sheet Molding Compound (SMC) Forming
Bulk Molding Compound (BMC) Forming
Hot Melt Applications
Welding, Thermal, Ultrasonic, Superheated Gas, Tape Laying, Other
Preimpregnation of Woven, Unidirectional, and Fibrous Mat Materials (Prepreg)
Others
1052
Existing parts can be used to form and shape "intermediate" molds and tools mass-cast
from various thermoset compounds. In fact, in many instances molds and tools are made
from the same polymer system as the structural part. This reduces Coefficient of Thermal
Expansion (CTE) problems by matching shrinkages, etc., during part material forming and
processing.
A good example is duplication of a Dipole Magnet Support Post. An existing or easily
machined model could be used to mass-cast the "post" mold. If the post were to be formed
using S-RTM molding methods, then a similar tool could be cast and used to form the
selectively reinforced, fiber-placed preforms
Semi and Structural Composite Magnet Coil, Interconnect Region,

Support Post, and Other Thermoset Polymer Applications
As mentioned earlier, many thermoset materials have been used to fabricate non-metallic
components and assemblies for High Energy Physics applications. Common materials such
as G-lO, G-ll, G-IO CR, and G-ll CR are electrical and industrial laminates most of us
know about. In areas of high radiation, at cryogenic temperatures, and under extreme loads,
these materials would perform below expected acceptable performance levels. This is true at
molding, processing, and operating temperatures.
By merely changing the G-IO or G-II matrix resin system used to form the fiberglass-
reinforced laminate, we could end up with a material, when machined, that surpassed the
required life expectancy in the cold radiation environment. Of course, we all consider
machining of finished components to be an uneconomical and time-consuming manufacturing
method.
Newer "high-performance" resin systems, chosen properly, could be used for a number
of cost-effective molding methods.
PHOTOS, SKETCHES, TABLES, DRAWINGS
Engineering proof-of-principle parts were fabricated during the materials studies over
the past four years. Figure 1 shows a number of composite magnet components fabricated
during the studies. Figure 2 demonstrates the inner and outer surfaces of the selectively
placed fibrous preform and saddle model used to make quadrupole magnet coils. Figure 3
shows various views of S-RTM formed dipole coil end parts, including a cross-section of a
coil end. Table I is a listing of some of the composite materials processes used to form coil
and magnet parts.
References and credits for the facilitation and application of the associated materials and
processes techniques are noted on each item.
1053
PRODUCTION, METALLURGY, AND UTILIZATION OF HIGH MANGANESE
NONMAGNETIC STEEL-ITS DEVELOPMENT FOR THE sse MAGNET
Kiyohiko Nohara,! Kagao Okumura,2 Shinji Sato,2 and Hiroshi Sasaki2
ITechnical Research Division

2Steel Technology Division
Kawasaki Steel Corporation
Kawasaki-cho, Chiba, 260, Japan
ABSTRACT
As a candidate of structural material for the SSC magnet cryogenic use, high
manganese nonmagnetic steel (KHMN30L) has been developed to give several favorable
results in terms of its production, metallurgy and usability. It is proved available to make a
mass production of both 1.5mm and 2.5mm thick sheet coil. The suitable balance of
nonmagnetism, strength and cryogenic toughness is achieved along with their sufficient
stability and unformity. The material usability has been investigated to involve the experiment
of semi-perforation / fine blanking of a collar with 2.5mm thick specimen currently
manufactured at a factory.
I. INTRODUCTION
The cold mass of a SSC superconducting magnet comprises principally several

components such as beam tube, super conductor, insulator, collar, yoke, helium shell, end
yoke, etc. High manganese nonmagnetic steel (KHMN30L) developed for collar, beam tube,
(and end yoke) has been studied in terms of the production, metallurgy and utilization
including semi-perforation test with 2.5mm thick steel manufactured, which follows to the
previous papers. 1)2) The summarized list of nonmagnetic material and magnet components is
presented in Table 1 to list up a matter of magnet construction including the design,
fabricaton and performance in additon to the material production, mettallurgy and utilization.
As for the material it is noted that the matching of nonmagnetism and the strength is of great
significance.
2. PRODUCTION OF HIGH MANGANESE NONMAGNETIC STEEL
In order to satisfy the targets mainly for the collar and beam tube mentioned in Table
I, there is a possibility of selecting a nonmagnetic metallic material like austenitic steel, super
alloy or aluminum alloy when the cost performance is also considered. The authors have put

Plenum Press, New York:, !992 1055
Table 1. Summary of material and magnet for the SSC.
Target collar Tube (End yoke)

Item
RT
,, 4K RT ,, 4K
Notes
Alloy designing Lower/Stabler permeability @ @ @ Fabricability

c
0
iii:;::; Production process 0 0 0 Yield
._ u
.. ::::J (Steel making)/hot workability 0 0 0 Cost performance
~] (Descaling)/Pickling 0 0 0
::iEn.
Temper rolling/annealing Hardening @ @ 0
8 Strength YS(RT)~60KSI'" 90KSI @ 0 0 Local micro-yield
>o~ Ductility/Toughness TS(4K)~ 170KSI 0 0 (EMF due to 6.6T)
E'~
0 0
::!li
III ..
Fatigue/Corrosivity
Magnetic permeability
P~1.0020, dP~O.OOO5
@ @ 0
Local stress
concentration
-III
CD.c
:::iEo Thermal coefficient 0 0 0 (Pre-stress)
c
Stamping/Spot welding Collar fabrication 0 0 Yield
j Fine blanking
Burring/Dimpling
-- --------
..!~~~!:,J! -_ .. -- .... ---
0
0
Efficiency
Die wear
_____________ 0
~ TIGwelding
Sensitizing 0 0 0
11 Design Aperture 40':;50""" 0 0 0 Tooling
c
I Fabrication
Performance
Yoke split (Horizantal, Vertical)
Collar thickness (1.5...,2.5...)
0
0
0
0
0
0
Strese behavior
Field property
Table 2. Chemical compositions of high manganese steel.
wt%
C Si Mn Cr Ni N V P S AI O.
Nominal 50.4 SI.0 26-32 ~10.o S2.0 ~o.15 50.2 - - - -
Example 0.11 0.6 28.4 7.1 1.0 0.10 0.06 0.033 0.002 0.02 0.026
*Temper cold-rolling as final process
1.5r--,---,--.---r--r---r--,---,
Hadfield Steel
..... 1.0 \ ® Y
~ \
\
o \
\a+E+Y
0.5 \
\
a \ ,
Mn (%)
Figure 1. Equilibrium phase diagram of Fe-Mn-C at 1100°C and composition range of KHMN30L.
1056
the first priority on the matching of magnetic permeability, Jl (Jl~1.002, its fluctuation
~Jl~0.0005 at 4K) and yield strength, YS (YS~90KSI at room temperature), and have
studied the potential of Fe-Mn-C alloy!) among austenitic steels. As a result the most desired
chemical composition of high manganese nonmagnetic steel is determined to be shown in
Table 2, featured by low C (around 0.1 %), high Mn (about 30%) and the proper addition of
other elements (N, Cr, Ni, etc.). Figure I shows the equilibrium phase diagram of Fe-Mn-C
system to show a proposed area of nominal KHMN30L (to exhibit nearly 30% Mn and low
C). This proves that this alloy steel does not cause a phase transformation from paramagnetic
y (+£)to ferromagnetic a through an extended temperature range, leading to nonmagnetism
and a good magnetic stability. The steel is mass-produced on a process given in Fig.2. Both
1.5mm and 2.5mm thick coils are producible by BA (bright annealing line) / AP (annealing
and pickling line).
Normally, high content alloy or special steel has a difficultly in rolling at elevated
temperatures. The current high Mn steel was no exception. To cope with such manufacturing
problems, the addition of a special element and the balancing of the elements are found
effective. The hot workability against rolling temperature is given in Fig.3 that indicates just
over 60% of RA (reduction of area) even at SOOT. This brings about a smooth processing,
good quality control and high production yield. Another question involved is a drawback in
pickling of hot / cold material. An instance of KHMN30L pickling conditions is illustrated in
Fig.4 where the suitable conditions are applicable to an actual plant. Then the total mass-
production process of KHMN30L shown in Fig.2 has been established.
3. METALLURGY OF HIGH MANGANESE NONMAGNETIC STEEL
As mentioned above, the simultaneous fulfillment of the specifications of both

permeability (Jl~1.002 at 4K) and yield strength (YS~90KSI at room temperature) is of most
importance. As for strengthening the steel, there might be several metallurgical measures --
solution hardening, precipitation hardening, strain hardening and the like. In view of a
dramatic stability in magnf!tization of KHMN30L irrespective of the application of heavy
strain, the strain hardening method was adopted) Considering both BA and AP (including
tandem rolling) processes corresponding to the production of 1.5mm and 2.5mm thick
sheets, both temper cold roIling and temper annealing as strain hardening method was
investigated. Figure 5 shows the change in YS at room temperature as functions of temper
cold rolling ratio and temper annealing temperature. It is seen that the latest specifications of
YS~90KSI can be satisfied by either way. The previous result confirms permeability at 4K to
be below 1.002 even when cold rolling ratio is applied up to as high as 50% (at least 35%).3)
Figure 6 shows the changes in Jl (at 4K) and thermal coefficient, f3 (between SOK and room
temperature) with the Neel (magnetic transformation) temperature, TN that is found to be
related to C and Mn contents by the following equation:
TN (K) = 270/ ( 3.SC% - 0.22Mn% + 6.9 ) (1)
The chemical compositon of KHMN30L is controlled as shown in Table 2, so that a typical
value of TN is 3S0K. This is the reason for the low permeability and small fluctuation
attributed to the steel. 4 ) Thermal coefficient, f3 is also affected by TN with slower tendency in
a change than Jl, and its desired value is still somewhat in question depending on the cold
mass configuration of either horizontal or vertical.
4. UTILIZATION OF HIGH MANGANESE NONMAGNETIC STEEL
High Mn nonmagnetic steel (KHMN 30L) has been applied to various usability tests
as shown in Table I in connection with tooling of the SSC magnet. The results of some
experimental trials are described below.
1057
""" "" LD convertor
L..---'--..,._ _ _---' """"" "Rise in yield strength at Rl
Temperature ('C)
Figure 2. Manufacturing processes of KHMN30L. Figure 4. Dependence of temperature and

concentration of ~SO4 solution on picking
of KHMN30L (time:2min.).
(b)
100r---~---'----~---r----r---,50
90 T,: 1200"C
80 40
70
_ 60
-~ 50
c(
It 40
30
20 10
10
~00~~800~~~~~~~~~~~~~ o
900 1000 1100 1200 1300
T2 ee)
Figure 3. High temperature ductility and strength showing hot workability on the above heat
pattern with as-cast specimen heated at 1200°C for Ihr and water quenched.
1058
RT
15 1.020
Fe-Mn-C
1.015
~ ,, 270
'IS!
" 10 , TN(K)= 3.8C%-0.22Mn+6.9
1.010 ~
x \ "1.
20 ~
\
I-
II: \
, ...
~ \
1.005
.. --- ----
g \
o
I
CQ.
0 2 4 6 8 1012 14 16 18 20
'.. (J.l-TN)
Cold rolling (%)
5 1.000
As 500 700 000 1100 I 1300 0 100 200 300 400 500
rolled Annealing temperature rc) TN(K)
Figure 5. Change in yield strength at room Figure 6. Relations between thermal coefficient ([3) or
temperature of KHMN30L with cold rolling magnetic permeability (p) and the Neel temperature (1N)
ratio (l120°C annealed sample) and annealing in Fe-Mn-C alloy, where TN is expressed by the
temperature (50% cold rolled sample). equation, TN(K) = 270/(3.8C%-O.22Mn%+6.9).
Ca)
Table 3. Separation tensile test of conventional

stamped specimens.
r"3 --"r°:!
Shear test piece Cruciform test piece
at.Q ,JI.-
7.
1 957(2125) 1 955(2120) g~- ::; 0"
2 987 (2170) 2 960(2132)
3 950(2109) 3 946(2100)
ave. 965(2135) ave. 954(2117)

Average tensile load required :
min 9OOI<g (2000 Ibs)
unit: kg (lbs) Figure 7. Separation tensile test chart [(a): shear test
piece, (b): cruciform test piece]
Table 4. Magnetic permeability of TIG welded beam

tube with 1.5mm thick KHMN30L and some other
samples including seamless tube of 316L stainless
steel.
Matrix Weldament
Item RT 4K RT 4K
TIGwelded
KHMN 1.0009 11.0010 1.000911.0010
beam tube
TIGweided
316LN 1.010 11.017 1.019 11.029
sheet
Seamless
tube
316L 1.015 11.025 --
.. . " , ". .... . " , . .
~ Annealed
sheet
304L 1.012 11.019 --
Photo 1. TIG welded beam tube of l.5mm thick
KHMN30L.
1059
4-1 Stamping and Spot Welding
The conventional stamping test followed by the spot welding test was perfonned
using 1.5mm thick KHMN30L (YS : 90KSI grade). The press machine with the power of
100ton was used with the total stamped pieces of 24,500 sampled from a sheet coil; after
12,854 laminations were stamped, the die was removed for sharpening, then the balance of
11, 646 collar laminations were stamped without any further die maintenance. Every 500th
stamping was inspected for burr height (its critical value:100~m) and key width, and the
results were acceptable. Lamination, pairing and modeling were conducted without any
problem after stamping. An in-house spot weld test was perfonned. Every 1,000th welded
pair was inspected for maximum thickness build-up in the weld region, for key width
dimension and for any evidence of weld flash (expulsion). All welded pairs inspected were
found to be acceptable and no evidence of weld flash was noted. Welded pairs were
subjected to a pull test to examine separation load with shear and crucifonn test pieces. The
results are shown in Table 3 and Fig. 7. In each case the tensile separation (removing) load
can be over the required value (900kg or 2000lbs).
4-2 Beam Tubing
The final production method of beam (bore) tubing is not detennined. However, the
tubing by TlO welding has been applied during RID work of the sse magnet. The seamless
piping is another choice with certain probability. The concerns are laid on strength,
penneability and helium penetration of a tube manufactured. KHMN30L of 1.5mm thick
with 90KSI (=YS at RT) was subjected to a trial TlO weld tubing on a conventional mass
production facilities. Photograph I shows the products. The welding was more easily
performed than type 316LN stainless steel containing up to over 0.3%N and was free from
weld defects. A major concern with TlO welded tube is a change in permeability, J.l along
weldaments. The observed data of ~ at RT and 4K are tabulated in Table 4 with a couple of
reference data including 316L stainless steel seamless tube. It is clearly shown that
KHMN30L keeps nonmagnetic with sufficient stability compared with other materials.
4-3 Burring(Bore Expanding)
In order to check the stretchability (stretch flanging perfonnance) the standard burring
(bore expanding) test was carried out in connection with collaring. Experimental procedures
are as follows: (l) stamping for perforation (hole dia.= 14mm, clearance=O.11 mm, hydraulic
press=30ton) and (2) burring (Erichsen test machine=50ton, stroke velocity=50mm/min,
lubricant=Die Rust R80, blank holder pressure=1 ton and 4ton). The result is demonstrated in
Table 5. The burring ratio (R) which is expressed by
R=(D-Do)/Do (2)
where Do is initial diameter and D is diameter after test is smaller in KHMN30L than type
316LN stainless steel. The drawing height is approximately 12mm that is somewhat smaller
than the initial hole diameter. It is said that KHMN30L possesses a certain amount of
stretchability for its extremely high strength compared with a stainless steel. Bore expanded
samples are exhibited in Photo. 2 .
4-4 Semi-perforation (Dimple Fonnation)
So far the R/D work of collaring has been done on the conventional stamping
followed by spot weld for pairing with 1.5mm thick sheet material. Lately another way of
collaring seems to be proposed -- the fine blanking followed by semi-perforation (dimple
formation) with 2.5mm thick sheet material -- for the purpose of the fabrication efficiency
and better performance of a magnet. Hence 2.5mm thick sheet coil having the same
properties as 1.5mm thick steel witS produced on basically the same process as in Fig. 2, and
semi-perforation test was perfonned as it is a key subject of the proposal. Figure 8 shows a
1060
Table S. Burring test results of KHMN and
316LN stainless steel.
Burring (Bore-expanding) Stamping
Drawing Burring Drawing

Specimen Stroke Blank Drawing (exp.nding)
holder force height force
ratio
......u .. (R)'"
po" 5.7 ''''' 12.5 mm 049

KHMN t60 k e /em'
Stop at 4 5.7 12.0 0.44
t - - initiation
of necki"g 4.8 16.5 0 .71
316LN 120
4 4.4 15.0 0.69
KHMN Stop at 4 5.2 (10.0)
~ 10rnmh·· 4 3.3 (10.0)
~
120
* To measure burring(expanding) ratio

** To prepare burring samples 'ree 'rom necking Photo 2. Burring test specimens (blank holder
*** R=(D-Oo)/Oo where Do=initial dia. and O=dia. after test
pressure: 1 ton; A:KHMN30L, B: 316LN).
Table 6. Semi-perforation test conditions on

lOmm dia. dimple with 2.5mm and 1.5mm thick
KHMN30L sample.
Hydraulic NC turret
Press machine
punch press
Press capacity 30ton
.
Stroke before deceleration 12Smm/sec
Spring :!! ....:g speed after deceleration 60mm/sec
'" Keeping time 100msec
Maximum press load lSton

!fl
2.Smm thick O.OSmm
Clearance
Blank
1.Smm thick O.OSmm
holder
.... Test temperature lS"C
Sample ___ __
Lubrication dry
Die material SKDll
* At 2mm above the surface of a specimen
SPring ----~~~~~~p~-1
CD
~ =;:= ==tiiiiilllll!l-.. .. Sample SKDll
~1O. 1
Figure 8. Die dimension for semi-perforation test.
Photo 3. Hydraulic NC turret punch press for

semi-perforation test.
1061
2.5mm thick 1.5mm th ick
2.5mm thick
J ~23"58
1.5mm thick
Photo 4. Appearances of lOmm dia. and O.8mm deep dimples after semi-perforation test with
2.5mm or 1.5mm thick KHMN30L (yS(RT»90KSI).1
1062
Photo S. Etched structure in cross section of dimples semi-perforated with 2.5mm or 1.5mm
thick KHMN30L (yS(RT»90KSI.
Table 7. Characterization of high Mn steel, KHMN30L, compared with stainless steel and AI
alloy.
Target KHMN 316LN 7075-T6

Item
RT:4K RT:4K RT:4K RT : 4K
.
t:
YS(KSI)
TS(KSI)
~90: ~110
:
98 :215 58 : 142 73 :
116:246 68 : 225 64 :
§ EI (%) : 33: 29 40 : 34 11 :
li
.: yE (...I) : 274: 135 290 : 120 :
Toughness
II K~ (MP• • m) : (208): 198 -:190 :
:i!
Fatigue limit (KSI) : 57 : 131 : :
.
t:
Permeability :
I'
~I'
(~I.0020) 1~1.0010 l.ooe: 1.017
(~.0005) ~O.OOO5 :
1.00
8.0 Thormll co<:I.(80- 300K. K) 7.5XIO· II.OX'O IS.OXIO

li : 1200:
Young', Mod (kgfmm1 19100:
,.
100));
,;
Resistivity (pOem) 84.8 78 55
~
Specific he.t (J/k g'k) : (200~ 1.4 (200~ 3.5 :
Stamping ; Burr (11m) 80
Defect No defect
Spot weld
S.par.(kg) ~ 900 965
~ Burcing (ratio) 0.49 0.71
....
li Deleet No defect 810w holt IN
:) TIG weld I' : l OOO9: 1.OO1G 1.014: I.OZ3 :
yE (...I) : 254 : 104 ,ao: 75 :
Sensitize : vE (...I) : 270 : '12 275 : 60 :
Dimpling (2.5 t .I.St) Available
1063
view and dimension of a die used for semi -perforation test: dimple = I Omm dia. and 0.8mm
deep, clearance=0.05mm, comer=90°. The test conditions are listed in Table 6. Actually both
2.5mm and 1.5mm thick specimens were used. Photograph 3 shows a machine used that is a
hydraulic NC turret punch press. The macroscopical observation of the tested samples is
shown in Photo. 4. The square-shaped and 50mm<l> aperture collar-shaped samples can be
furnished with dimples in both thicknesses to accompany no particular defects. Microscopical
observation of the cross section of the samples is exhibited in Photo. 5. The deformed
material flow at the edges of a semi-perforation is seen to a considerable extent and even a
micro-crack is not found at the sharp comers. This means that the steel has an enough
capability of semi-perforation with 1.5mm and 2.5mm thick sheets. 5)
5. CONCLUSION
As a candidate structural material for the SSC magnet cryogenic use, high manganese
nonmagnetic steel (KHMN30L) has been developed to give the following results in terms of
its production, metallurgy and usability:
(1) It is possible to make a mass production of either 1.5mm or 2.5mm thick steel on "LD-
CC-BA/AP-Tempering" process with the properly chosen conditions for steel making,
slabbing, pickling etc.
(2) The suitable balance of nonmagnetism (permeability 11<1.002 at 4K) and strength (yield
strength YS>90KSI at room temperature) is metallurgically achieved along with their
sufficient stability by chemical composition control and temper rolling/annealing.
(3) The usability in close connection with the magnet tooling and fabrication has been
investigated to show favoured results, which involves stamping/spot welding, tubing on
TIO welding, burring (bore expanding), and semi-perforation/dimple forming (fine
blanking) with 2.5mm thick material and 1.5mm thick material as well.
(4) The summarized characterization of KHMN30L containing mechanical property, physical
property and usability is tabulated in Table 7 as compared with type 316LN stainless steel
and 7075-T6 aluminum alloy to show its applicability to the SSC magnets.
* The steel developed has been trially used at SSCL, BNL, FNAL and KEK.
* The stamping/spot welding test was conducted at H&J Tool&Die Co., Inc.,U.S.A.
* The tubing was performed at Trent Tube-Division of Crucible Materials
Corporation, U.S.A.
* The semi-perforation experiment was carried out in cooperation with Amada
Metrecs Co., Ltd., Japan.
REFERENCES
1) K. Nohara, Y. Habu, S. Sato, K. Okumura, and H. Sasaki: "Supercollider 2", Plenum

Press, (1990), 765.
2) K. Nohara, K. Okumura, S. Sato, and H. Sasaki: "Supercollider 3", Plenum Press,
(1991),75.
3) K. Nohara, Y. Habu, "Proceedings of a Conference on Manganese Containing Stainless
Steels", ASM, (1998), 33.
4) I. Schmidt: "Proceedings of ICOMAT-86", JIM, (1986), 497.
5) K. Nohara: to be presented to "ASM Conference on High Manganese Austenitic Stainless
Steels," Chicago, (1992).
1064
MANUFACTURING AND MAGNETIC CHARACTERISTICS OF HIGH
PURITY ULTRA LOW CARBON MAGNET STEEL FOR THE SSC
Kiyohiko Nohara, I Ujihiro Nishiike, I Kazuhiro Suzuki, I

Taisuke Mori,2 Kagao Okumura,2 Hiroshi Terami,3 and Hiroshi Sasaki2
ITechnical Research Division

2Steel Technology Division
3Chiba Steel Works
Kawasaki Steel Corporation
Kawasaki-cho, Chiba, 260, Japan
ABSTRACT
Pure iron sheet or high purity (ultra) low carbon magnet steel sheet is employed as
magnetic shield yoke for the SSC magnets of various types. It is required to have adequate
magnetic properties as well as suited mechanical properties and usability (fabricability). To
meet such requirements the presentation is to be made concerning the status of manufacturing
and research work in view of mass production of high purity low carbon magnet steel, the
relation between magnetic characteristics and metallurgical factors, and the further
development of high purity "ultra" low carbon magnet steel exhibiting lower coercive force
plus its laser cut performance for the corrector special magnet.
1. INTRODUCTION
In the SSC project there is a possibility for pure iron to be used as magnetic shielding
material for both magnets and detectors. In the case of a magnet, there are several different
types of magnets depending on the functions required which need a different quality of
magnet steel. The general specifications of iron yoke material given so far are summarized in
Table 1. A major difference among the magnets on the table is seen in coercive force, Hc
after the application of 100 Oe magnetizing force. A dipole/quadrupole magnet just requests
1.9 Oe of Hc, while a correction magnet requires less than 1.0 of Hc. Under these
circumstances the present paper describes a mass production and quality of the conventional
low carbon magnet steel having over 1.0 of Hc, and then development of "ultra" low carbon
magnet steel showing below 1.0 of Hc in terms of the metallurgical factors influencing
magnetic properties of iron, and the skinpass effect plus laser cutting.

Plenum Press, New York; 1992 1065
Table 1. Specifications of iron yoke material for sse magnets.
Dipole/Quadrupole Correction
Item magnet
magnet
C ~0.005 wt%
c Mn ~0.5 wt%
0
-:t:i wt%
1\1 .-
U 1/1
P ~0.02
.- 0
..E E
0
5 ~0.02 wt%
68 Si ~0.05 wt%
AI ~0.10 wt%
Coercive force, Hc(100 Oe) 1.9±0.250e (1.00~Q65)±0.10 Oe
~~
c 8.
Permeability, at 1 Oe 1500~3400
iK J.I. at 10 Oe 1400~1530
Saturation induction, Ms ~2.12 Tesla (H=5OO~3000 Oe)
Hardness, RB 20~40
1~
~8.
Yield strength, YS (~17kg/mm )
Tensile strength, TS - (~28kg/mm )

~K Elongation, EI (~48%)
(Note) Refer to SSC-MAG-M-4341
Figure 1. Mass production of high purity magnet steel (EPE).
Table 2. Typical chemical composition of high purity magnet steel (EPE).
Thickness C Mn P 5 Si AI N· 0
1.6mm
• 0.002 0.18 0.013 0.008 0.007 0.005 0.0016 0.0058
••
6.0mm 0.004 0.16 0.008 0.004 tr. -- -- --
* Cold rolled/annealed material (EFE-3A) wt%
** Hot rolled/annealed material (EFE-2A)
1066
2. MANUFACTURING OF LOW CARBON MAGNET STEEL
The conventional high purity low carbon magnet steel sheet (coil) (EFE) is
commercially manufactured to be suited to the dipole/quadrupole magnet iron yoke material.
This EFE steel is classified to the three categories dependent on thickness (EFE-I,
plate>8mm thick; EFE-2, hot product=I-8mm thick; EFE-3, cold product <2mm thick).
Their manufacturing process route is shown in Fig. I. The steel is produced by the integrated
steel making processes composed of converter, degassing and continuous casting; the cold
product is featured by the use of high temperature continuous annealing line. The typical
chemical compositions of EFE are shown in Table 2. The carbon contents of I.6mm and
6.0mm thick coils are 0.002% and 0.004%, respectively. The properties of these steels are
seen in Table 3. Coercive force, Hc of 1.6mm thick sheet (EFE-3) and 6.0mm thick sheet
(EFE-2) comes to 1.20 Oe and 2.1S Oe, respectively. These values are typical examples, and
there is a definite possibility of satisfying the specifications by controlling manufacturing
conditions that can be referred to the following section. The magnetization chart or B-H and
Il-H curves (B:induction, H:magnetizing force, Il:permeability) of the 1.6mm thick steel
(EFE-3A, A: as-annealed state) is exemplified in Fig. 2. It is noted that the maximum
permeability reaches 3840 at 2 Oe.
As a magnet yoke is immersed in liquid helium in operation, the yoke material is used
around 4K. The specification of the iron yoke in Table I shows the values at room
temperature. The measurement of magnetization at 4K was carried out with 3.2mm thick
EFE-2A (as-annealed state) to be shown in Fig. 3 compared with the measurement at room
temperature. The test at 4K was performed on a direct current magnetizing machine and a
large sized liquid helium dewer with three layers of vacuum and liquid nitrogen. It is found
that the magnetization at 4K is somewhat lowered in the low magnetizing force region
compared with room temperature, while in the region over 2 Oe a marked difference of B-H
and Il-H curves between 4K and room temperature is not observed.
3. DEVELOPMENT OF ULTRA LOW CARBON MAGNET STEEL
Magnetic coercive force (Hc) below 1.0 is desired for the SSC correction magnet. To
cope with the requirement the study on the metallurgical effects on Hc and the mass
production of developed steel was made, including the skinpass effect and laser cut
performance test.
3- I Influencing Metallurgical Factors on Coercive Force
It is considered that the effective factors on magnetic properties of iron from the
metallurgical point of view are chemical composition and micro-structure. Table 4 shows
these factors, their detailed items and the related manufacturing processes. The primary items
among them might be impurity elements and grain size. It is suggestive to note that Brechna
and Gerold proposed the following equation with regards to induction, B of low carbon
steeJl)2):
B(Tesla)=BFe - (0.3C%+0.032Si%+0.07SMn%+0.033Cr%
+0.07SMo%+0.OI6Cu%+0.068AI%+-0.7N%) (1)
where BFe is induction of 100% iron at magnetic force of 104A/m, because Hc shall be
connected with B to a certain extent. Figure 4 shows the relation between Hc and carbon
content of 3%Si steel and the conventional low carbon steel (EFE) plus the currently
developed ultra low carbon steel. It is likely that the reduction of carbon content even at very
low content region is influential to lowering of Hc. The relations between Hc and grain size/
(I 10) intensity are shown in Fig. S. Grain size of carbon steel affects evidently Hc as a form
of negative correlation, while (110) intensity also gives a negative relationship to Hc with
weaker effect than grain size.
1067
Table 3. Properties of conventional high purity magnet steel.
Item 1.6mm thick 6.0mm thick
Coercive force. He (100 0.) 1.200e 2.150e
II Permeability.
J.l
at 10e 2240
at100e 1480
Saturation induction. Bs (2.18)

2040
1460
(2.14)
Hardness. Rs 40 31
1»
i{
Yield strength. YS 17kg/mm2 14kg/mm2
C:i Tensile strength. TS 28kg/mm2 22kg/mm2
Elongation. EI 46% 54%
22r-----.------,------.-----~
20
3.2mmthick 30000
18
16
"'- 1'4
~ ~ 12
3000
.
:;; 'iii
l
10
2000
I-RTl
~
1000
10 ~----~----~~~~100~--~1~000~0
Magnetizing Force, H (08) Ma~zing force, H (0.)
Figure 2. B-H and p-H curves of 1.6mm thick high Figure 3. Comparison of B-H and p-H curves of
purity magnet steel (EFE-3A) at room temperature. 3.2mm thick high purity magnet steel tested at room
temperature and 4 K.
Table 4. Influencing factors on magnetic characteristics of high purity magnet steel.
Factors Detailed items Related actions in actual process
CD Impurity element (1) Selection of raw material
(C. N. S, Mn,······) (2) Steel making/degassing

1. Composition
(2) Precipitation. Segregation (3) Continuous casting
@ Inclusion (electro-magnetic stirring)
CD Grain size (1) Slab heat; FET. FDT. CT
2. Micro-structure (2) Crystallographical texture (2) Continuous annealing
@ Intemal strain (3) Straining. skinpassing
* Induction B=Biron-~Il.BiCi=Biron-(0.3C%+0.032Si%+0.075Mn%+0.033Cr%+
0.075Mo%+0.016Cu%+0.068AI% +-0.7N%) <Brechna-Gerold>
6B : flux reduction rate. r Il. BiCi : total flux reduction rate
*B1': He .... ~Il.BiCi"': He'",
1068
2.0
RT
Giron (conventinal) (H=100 08)

~u 1.5
• Iron (developed) (H=100 Oe)
l:
..8 .
• 3%Si steel (e=13kG)
. .... .... ,,-----

....-------- _..A--
.2
,..."
GI
·e
>
GI 0.5
_-4"
c3
00 0.005 0.010 0.015

c (%)
Figure 4. Relation between coercive force Hc and C content in iron (present study) and 3% Si
steel (by Brechna).
~ 2.0.--.--,---.--.--,---,----.-----.----,
c5 o
o
~
•
(110) intensity (arbitrary)
Figure 5. Relation between coercive force and grain size or (110) intensity (.conventional,.
developed).
1069
In consideration of those effects of chemical cOmpOSltlOnS and grain size, the
formulation to correlate Hc to compositions and grain size was tried and the following
empirical equation was introduced:
Hc (Oe)=2xl02C%+lxI02C%+2xlOlS%+lx100Mn%-5x104GS(f,1m)+OAO (2)
where GS indicates grain size. Figure 6 illustrates the relation between calculated and
observed values of Hc, using the equation(2). The conventional and developed steels mass-
produced are seen in the figure. These data and analysis are useful to develop the magnet
steel to meet the requirement.
3-2 Ultra Low Carbon Magnet Steel Developed
On the basis of the above results concerning metallurgical study, the magnet steel
targeted for the correction magnet having lower value of coercive force after 100 Oe
magnetizing has been developed on a mass-production scale by means of chemical
compositional alloy designing through the reduction of C, Mn, etc, and grain coarsening.
The decarburization was conducted by a special degassing process, and the grain coarsening
was made possible by the application of a special heat pattern at the high temperature
continuous annealing line (Fig. 1). The chemical composition of the developed high purity
"ultra" low carbon magnet steel is demonstrated in Table 5. The carbon content by the check
analysis was markedly reduced down to 0.0011 % from 0.002% of the conventional steel.
Micro-structure of the developed steel appears in Photo. 1, which gives the average grain
size, 300f,1m compared with lOOllm of the conventional. The magnetic and mechanical
properties of the developed as-annealed steel are listed in Table 6. Hc after 100 Oe of
magnetic field comes to below 1.00e (coil A: 0.76 Oe, coil B: 0.81 Oe), and permeability at
1 Oe is much larger than the conventional steel shown in Table 3. The strength, however, did
not satisfy the specifications. The magnetization chart is shown in Fig. 7. The maximum
permeability of 4800 was obtained around 1.5 Oe, a larger value by 1000 than the
conventional in Fig. 2.
In view of importance of strength which was not satisfactory with the specifications
as shown in Table 6, the effect of skinpass to the as-annealed developed steel on strength and
other properties was investigated. The result is presented in Fig. 8. It is noted that yield
strength, likely to be more important than tensile strength, increases extensively with the
increase of skinpass ratio to exceed 17kg/mm2, a specified value, at about 1.5% of
skinpassing, where Hc stays below 1.0, also a specified value. As for tensile strength
specified to be over 28kglmm2, a heavy skinpassing to reach 10% is needed. This generates
a degradation of Hc-- that implies exceeding 1.0 of Hc, a maximum limit, when judged from
Fig. 8. Therefore the question still remains in this regard using even the developed steel, and
the further study is desired including the reconsideration of the specification3 ).
3-3 Laser Cutting
The iron yokes for the SSC corrector magnet are extensively varied in their shapes
and dimensions in addition to demanding a low coercive force, Hc. This is why laser cutting
is desired besides the use of an iron having a low value of Hc. In this connection, laser
cutting test was performed on a production machine using the developed steel mentioned
above.
The experiment conditions are summarized in Table 7. The machine used was of a
C02 type laser with 02 or N2 as assist gas, vertical or axis flow as an operation mode, and a
pulsing wave as a cutting pattern. The speed, power, frequency, and duty (ratio) were
intentionally varied, and their typical (rather optimum) figures are as follows:
speed=1200mm/min, power (apparent)=1000W, power (actual)=400W, frequency=500Hz,
duty (ratio )=30%. The diameter of laser after focussing is approximately 50llm. Photograph
2. shows the operation of the experiment on a C02 laser plant machine. Macrospical
appearances of laser cut square and yoke (shaped for" one of a SSC dipole magnet) samples
1070
2.o,-- -.------.-
,, --r- -......
AT '
~~ i 0° , / '
10
, .../
1 .5
o
51
.! 1.0 --- -________ CL~ --------- - --
"i
iii
q~ !
Jl
g 0 .5
/ ° ,-------.
:
u
l: / ; 0 :L......._ _ _---.J
,
45'
°0~L.......~0~.5~---I~.0----~,.L5----~
2.0
HC lcah after 100 Oe (Oe)
*Coercive force
Hc (cah =2X10 2 C( %)+1 X10 2 N(%l+2X10'S(%)
+1 XlOoMn(%)-5X10- 4 GS(pm)+0.40
Figure 6. Observed and calculated values of coercive force Hc, related to composition elements
and grain size (.conventional, .developed).
Table 5. Chemical composition of developed ultra low C steel.

(wt%)
C Mn P 5 Si AI N 0
0.0011 0.100 0.005 0.005 0.004 0.005 0.0015 0.0071
200.um
LJ
Photo 1. Micro structure of developed ultra low C steel.
1071
Table 6. Properties of developed ultra 20,-----.---.---nrrrTTT---'---'--'-"""'r"TTrr----''---T"T"..TTTl 5000
low C steel.
Item Coil A Coil B ..

"~'2
16 4000
:..
0.81 De
&
Coercive force, He (100 De) 0.76 De 3000
~
i:t Permeability, I at 1 Oe 4890 4280 m D
H C 8 ~
::<0.
J1 [all00e 1490 1470 ~o
2000 ~
(2.19) :J ~
"
Saturation induction, Bs 2.20
.5 1000
Hardness. RB 21 18
-;;;
.~ .e Yield strength, VS 12.7kg/mm' 12.4kg/mm2
~& Tensile strength, TS 24.5kg/mm' 24.4kg/mm 2
1 10
~~ 44% 46%
Magnet;c ion:<:, H (Oe)
Elongation, EI
Figure 7. B-H and p-H curves of developed ultra
low C steel.
Table 7. Laser cutting experiment conditions on CO2

laser with 1.5mm thick ultra low C steel having low
coercive force.
Machine C02 laser machine
Operation mode Vertical/axis flow
Assist gas 02, H2
Speed SOO-----2000mm/min
Apparent ~2000W
Power
Actual ~2000W (depending on duty)
Cutting mode Pulse
--lr---__ _
1 .4 a---,--.----.--.-----r--.-----r-....--,;----, 1 S Frequency S~2000Hz
0;
3
u
1 .2
1.0
::r 0 .8
(a)
!...... - - . .
. --- =
BlO00
--------- He
0 ----- 13
11
Ii
t;
Duty ratio
Diameter before focussing
of laser after focussing

Variable (0~100%)
15-30mm (depending on power)
-50~m (depending on mirror/lens)

9 ~
Ambient temperature ISoC
0 .6 7
0.4 1--+-~ ___-+---1--+--+--+---+-+-..., S

(b)
5000
"""--~--P.:..~-<>-
....---~____ -._-:...-:...--:a.~
::t 1000 /1.0 00

500
l00~-+-~--+-~~--+--+~___~~
28 (e)
....----...------- 45
"
---
,
>0::,-
TS
_- _c------- YS
til .; }(' .....

I- .; - ..... x
<f".Jf' EI - .......... -... - __
vi
>-
Skinpass ratio
Figure 8. Influence of skinpass rolling on magnetic Photo 2. Experiment of laser cutting with l.5mm
and mechanical properties of developed ultra low C thick ultra low C steel having low coercive force.
steel.
1072
t ~ .... •• ¢J' t:l 4 •• 1 •• ~ I 2' :I .. iii. 1 • II
Photo 3. Appearances of laser cut square and yoke samples «a), (c):02' (b):N~.
'\ '\~ ) _. J . I
-q' :- -
v -<' '. ~ •• ~: •. .,..:
Of i '. · '-' .
.', F
~-
Photo 4. Micro-structure of laser cut sample (a) and sample (side surface by arrow: laser cut
cross sectional) (b).
Table 8. Magnetic property of laser cut surface of ultra low C steel developed to show low
coercive force, measured on vibration sample magnetometer.
Item Laser cut Without laser cutting
Coercive force. He (100 Oe) 0.72 0.760e
at 1 Oe 4910 4890
Permeability. J1
at 10 Oe 1480 1490
Saturation induction. Bs 2.20 2.20
1073
are given in Photo. 3. The slits in the square small samples were established to initiate a
cutting operation. It can be said that the laser cutting is applicable to the developed steel
without any particular problems. At present the cutting surface seems better with 02 assist
gas than N2 assist gas. The cutting of the yoke in Photo. 3 needs about 50 sec when the
cutting speed is set to 2000mm/min. The microscopical images of the surface and cross
section of a laser cut sample are presented in Photo. 4. None of the photographs show any
sign of a change in microstructure by laser cutting--neither grain coarsening nor generation of
heat affected zone. Further the magnetic measurement was carried out using a sliced small
sample weighing less than 200mg installed in a capsule on vibration sample magnetometer.
Table 8 gives the result. There seems no difference between a laser cut sample and a matrix
sample. This corresponds to the result from the microscopical observation in Photo. 4. The
experiment proves that the laser cut of the developed ultra low carbon magnet steel can be
free from deleterious effect on magnetic properties as well as on microstructure. 4)
4. CONCLUSION
As a shielding material for the iron yoke of the SSC magnet, pure iron or (ultra) low
carbon magnet steel (EFE) has been subjected to testing and investigation to give the
following results in terms of the manufacturing of conventional pure iron, metallurgical
study, and development of low coercive force steel:
(1) The conventional pure iron sheet containing 0.002% is manufactured by the mass process
including converter, degassing facility and continuous annealing line to satisfy the
specifications for a dipole magnet yoke.
(2) The coercive force is found to be dependent on chemical composition (C, N, S, Mn,
etc.), grain size and (110) intensity of the steel.
(3) The ultra low carbon magnet steel containing as low as 0.001 %C that is attained through
a special degassing process shows below 1.0 of coercive force after 100 Oe of
magnetizing force. Yield strength exceeds 17kg/mm2, a specified value, by light
skinpass. To satisfy the specifications of tensile strength, heavy skinpass is needed to
introduce a slight deviation from maximum limit of coercive force.
(4) The laser cutting can be applied to the ultra low carbon steel to indicate no detrimental
effect on either microstructure or magnetic properties.
*The conventional pure iron has been trially used at KEK.

*The laser cutting experiment was conducted as a collaboration work with Amada
Metrecs Co. Ltd., Japan.
REFERENCES
1) H. Brechna: "processing of MT--67", (1967), 305.

2) E. Gerold: Stahl and Eisen, (1931), 613.
3) M. Reehak and J. Turner: "Supercollider 3", (1991),147.
4) J. NOVak, F. Edeskuty and J. Bartlit: "Preprints of ICEC9--ICMC", (1982), 321
1074
DIPOLE MAGNET SHUTTLE SYSTEM
Art Zinszer and Steve Pidcoe Kyle Spann
General Dynamics Stellar Industries, Inc.

Space Systems Division 1020 W. 13 Mile Road
Plant Operations (MZ: CI-7430) Madison Heights, MI 48071
P.O. Box 85990
INTRODUCTION
A transport system (transfer shuttle) has been developed to move major magnet
subassemblies between tool stations.
The shuttle system is the key part of an integrated non-synchronous manufacturing

system which has been developed for the Superconducting Super Collider Dipole Magnet
Program. The shuttle system has been designed to address the following problems generally
associated with alternate systems (primarily overhead cranes):
1. Safety
2. System Thru-Put
3. Work Disruption
4. Particulate Contamination
5. Gentle and Precise Handling
Safety
Safety was the initial consideration which made alternatives to the overhead cranes
attractive. To address this concern, the shuttle system uses dedicated transfer bays which are
partitioned off from the work stations and operators. The shuttle system interfaces with the
individual stations only during the pallet exchange operation. The shuttle, pal1et(s) and
product in transit from station to station are isolated within the shuttle bay.
Once the shuttle system has engaged with a given station, a communication link is
established. The link provides positive signals between the station controller and the shuttle
controller to verify the redundant safety system status. The link also provides the
communication interchange needed to coordinate the pallet exchange process.
The shuttle control system is designed to require positive enable signals through the link
in order to proceed with any operation. This system design forces all safety and computer

systems to be fully functioning and providing positive "Safe Condition" signals before
activating any output devices. In this manner the operators and other personnel will be
protected in the event of failure or malfunction of any safety device or computer system.
System Thru-Put
The shuttle system is designed to maximize the overall system thru-put. The dipole
magnet manufacturing system is a series of approximately thirty (30) assembly operations.
Of these, twenty-four (24) are large sequential operations requiring shuttle transfer. The
balance are off-line subassembly operations which do not interact directly with the shuttle
system.
The large work station operations require from one hour to several shifts to complete,
and the design of the magnet is such that each operation must be complete before transfer. In
many cases, the sub-assembly is not structurally sound or able to be transferred until
completion of the operation. Therefore, all the work stations require multiple work cells in
order to produce the required ten (10) complete magnets per day at peak production.
Since the "In Station" cycle times are extremely long, in many instances, actual cycle
times from work cell to work cell may vary. The occurrence of shift changes, scheduled
lunch breaks and work breaks, and non-scheduled interruptions will have a varying effect on
the actual cycle time required for each assembly to pass through a given work cell of a given
work station. Also, the combination of operator and machine will produce a delta from the
nominal station cycle time due to many factors such as operator experience and skill, and
machine performance.
Due to these conditions, a transfer system had to be developed which would allow
the overall system to operate in a non-synchronous manner. The shuttle system is designed
with flexibility and surplus capacity to service work cells on an as-needed basis. This
approach allows the actual work cell cycle start and finish points to vary continuously
without effecting overall system thru-put.
In addition to the flexible capability, the shuttle system is also designed for high
speed transfer to enhance work cell thru-put. Each multiple channel shuttle unit will service a
series of work stations. The shuttle transfers the pallet through the series of stations until the
last station in the sequence removes the product from the pallet. The shuttle then returns the
empty pallet to the beginning of the series of stations. On a typical pass of the shuttle bay,
the shuttle will exchange a pallet with a work cell in each of the work stations. The
individual channels or slots on the shuttle unit will be designated to receive a pallet from one
operation and insert it into the next.
With this arrangement, the shuttle system will carry a small queue of pallets. When a
given work cell calls for an exchange, the shuttle will typically contain a pallet of the
appropriate status to replace the one to be removed from the cell. This is referred to as a
short shuttle exchange. The shuttle will simply withdraw the pallet from the work cell into
an empty slot. It will then move a short distance to position the pallet to be inserted with the
station and drive the pallet into the work cells. In this case, the work cell will be ready to
resume assembly operations within a matter of a few minutes (6 to 7).
Under special conditions, the typical series of short exchanges must be broken to
service a given station. This would occur when two or more work cells of a given work
station call for pallet exchanges at the same time. In this case, a long shuttle exchange may
1076
be necessary. A long exchange is one in which the shuttle does not contain the replacement
pallet for the one removed from a work cells. A suitable replacement must be obtained on the
following pass of the shuttle. However, due to the shuttle system's rapid traverse
capabilities, a long exchange can be a«complished with minimal effect on the station
throughput.
In a typical day (two shifts) which producing ten magnets, a total of approximately
570 pallet transfers will take place. A pallet transfer is defined as a complete sequence of
operations to extract a pallet from a work cell in one station, transport it to another station and
insert it into a work cell. The integrated shuttle system and work station designs are based
three types of pallet (narrow, medium and wide) such that each shuttle unit may service a
wide range of work stations. Since identical shuttles service ranges of stations, a given
shuttle may service a station outside of its normal range to eliminate sequence tie-ups.
Work Disruption
In manufacturing systems using overhead cranes and floor transport systems which do
not use dedicated bays, work disruption is a major concern. Such a method used to transfer
the dipole magnets or subassemblies (50 feet long, weighing from 10,000# to 35,000#)
would cause a great deal of intentional and unintentional disruption to adjacent work cells.
In many instances, workers in stations adjacent to transfer operations will be required

to stop work and clear an area to meet safety requirements. This is especialy the case when
using overhead cranes. If, for example, a number of work cells were placed in a common
crane bay, the assemblies would be transferred over several work cells to the next work
station. All the workers in the cells between the pick-up and deposit locations would be
required to clear the way. Furthermore, workers in cells adjacent to the transfer path may
naturally stop work to observe the large assembly passing by for curiosity or out of concern
for their own safety.
As discussed above under "Safety," the shuttle system has been designed to insulate
itself from the work cell operators. Work cells which are not involved in a pallet exchange at
any given time are physically separated from the shuttle system by a fence with interlocked
gates.
Particulate Contamination
Due to the nature of the magnet subassemblies, particulate contamination is a primary

concern. Particles, whether magnetic or non-magnetic, solid or not, may cause failure of the
product. Therefore, it is a basic requirement of this system to design equipment which does
not produce air borne contaminates.
This condition eliminated all overhead systems, floor mounted systems using air
flotation, and self-powered standard floor type vehicles. An appropriate solution must keep
all moving (particle generating) components below the product at all times. The solution
should also be electrically powered without high velocity escapting compressed air or fumes.
The shuttle system is designed to run on floor-mounted steel rails. All shuttle system
compressed air requirements are supplied by on-board air compressors and all exhaust air
will be collected. The shuttles will be powered by electric servo motors operated by a self-
contained electrical control system which derives power from the plant 480 V AC buss. All
1077
1078
GENERAL. DYNAMIC S
SSC MAGNET ASSEMBLY FACILITY, HAMMOND, LOUISIANA Space SySlems Division
Annex III Program Baseline Layout - - ., .~"- ...-.
-- . .. €I
.... __ .
£o . ".
N-
(')
(, (') (') (9 (£) (') (~ (')
(') o o
o OIAC NQSflCS An !: "
-...J
co
Figure 2. CDM Facility Layout.
bearing and components are designed for a 10 year minimum serivce life without lubrication
or with minimal lubrication to minimize the amount of "loose" lubricants in the plant.
Gentle and Precise Handling
A last, but extremely important concern, is the great deal of care that must be taken to
gently handle the various magnet subassemblies. The shuttle system is designed using
computer con troll electric servo motors to produce all motions. The system will be
programmed to use minimum acceleration and deceleration rates in order to avoid "G"
loading in either the transfer or insertion directions. Furthermore, the derivative of the
programmed velocity curve (the acceleration curve) will be interpolated to blend the inflection
points. This will produce a gentle blending of accelertaion ramps to eliminate jerk or impact
starts and stops. Instead, the shuttle will accelerate from one position and decelerate to the
next position in a controlled precise motion.
SUMMARY
A transport system has been developed to move major magnet subassemblies between
tool stations. The need existed to find a more efficient solution than overhead cranes to
handle large parts. The argument against overhead cranes includes safety concerns, work
disruption, particulate contamination and meeting the assembly rate requirements of ten
magnets per day. The shuttle transport system rtepresents a major effort of coordination
between the various tool suppliers and General Dynamics to design a universal device
capable of bridging the gap from single wound coils to a complete CDM. Effort was directed
to systematically minimize material handling and related equipment by interfacing a
completed assembly directly into the next work station or tool without losing its orientation
or changing pickup points.
The shuttle transport system is made up of a common transport device which can
automatically go to any preprogrammed address on the factory floor. Each station has
unique attachment tooling which can interface with the shuttle and the next assembly station.
The shuttle can also circulate attachment tools back to their point of origin. Additional
benefits of this system include inherent part protection, flow control, reduced banking or
inventory, and potential for automatic control.
ACKNOWLEDGEMENT
Research Association in support of the Superconducting Super Collider project for the
U. S. Department of Energy.
1080
CURRENT DEVELOPMENT OF THE SUPERCONDUCTING WIRES AND
CABLES FOR SSC IN HITACHI CABLE, LTD.
S. Sakai!, G. Iwakil, Y. Suzuki2, H. Nakazawa2, H. Moriai 2, and

K. Kamata3
ISystem's Material Laboratory, 2Tsuchiura Works

3Tokyo Head Office
Hitachi Cable, Ltd.
3550 Kidamari-cho, Tsuchiura-shi, Ibaraki-ken, 300 JAPAN
ABSTRACT
In this few years, Hitachi Cable has developed three types of superconducting wires
and cables for the dipole or quadrupole magnets of the particle accelerator. Single stack
proccesed wires and cables, high critical current density wires and large keystone angle
cables, and low AC loss wires and cables were developed. Single stack process makes the
yield rate high because of the reduction of extrusion. On the industrial scale, superconducting
wires with a critical current density of 3,200 A/mm2 at 5T have been produced. The large
keystone angle cables have been developed for the Cu spacer-less magnet. The low AC loss
cables are achieved by 2.5)lm filament and Cu-Mn matrix. The low AC loss cables are very
useful to reduce loads of the liquid helium refrigerator.
INTRODUCTION
Low initial and operating costs, high stability and reliability are required for the
superconducting magnets of the particle accelerator. To meet these requirment,
superconducting cables should play an important and leading role. On this point of view,
Hitachi Cable has currently developed three types of superconducting wires and cables.
Single stack process makes the yield rate high because the extrusion times are reduced triple to
twice. Large keystone angle cables fit the curvature of dipole or quadrupole magnets without

use of wedges. In large keystone angle cables, there are no degradation of critical currents in
spite of the heavy strand deformation in the cabling process. By the high critical current
density, the superconducting magnets have the sufficient margins to the operating current.
The low AC loss cable have been sucessfully achieved by 2.5~m small Nb-Ti filament and
Cu-Mn alloy matrix. The 2.5~m small Nb-Ti filaments and Cu-Mn matrix reduce the
hysteresis losses and the proximity effect coupling between filaments. The low AC loss
cables are very useful to reduce loads of the liquid helium refrigerator. In this paper, the
auther describes the features and superconducting properties of these three currently
developed wires and cables.
SINGLE ST ACKED 6~m FILAMENT WIRES AND CABLES
In Hitachi Cable's wire fabrication process, the most typical feature is the application of
the hydrostatic extrusion process. The advantages of the application of the hydrestatic
extrusion to the superconducting composites are the uniform deformation during extrusions
because of no frictions between the billet and the container, and the high yield rate because the
billet length to the billet diameter ratio is very large.1,2,3 As the billet diameter is relatively
small, monofilament material sizes are approximately l.Omm (flat-to-flat) hexagonal rod for
inner wires and 1.25mm for outer wires. Each two pre-production billets for inner and outer
wires have been restacked and extruded. Hexagonal monofilamentary rod is straightened and
cut to the appropriate length for restacking in copper cans. The small and long rods are
cleaned prior to assembly and great care is taken to maintain as perfect array as possible
during the assembly operations. The restacked billets are sealed by electron beam welding
prior to cold isostatic compaction for the ensuring a metallurgical bonding between all
components and the reduction of pre-heating tiIl?-es, and then extruded at relatively high
temperature. The extruded rods are drawn using standard rod reduction machines and wire
drawing techniques. The great care should be taken in die design and reduction schedules for
the prevention of filaments sausaging or wire breakage.
(a) Inner wire (b) Outer wire

r Igun:.1. 1111; "1U~~ ~"""Ullal pnolOS or nnal sIze WIres.
1082
Figure 1 shows the cross sectional photos of final size wires. The numbers of filaments
are approximately 7,000 for inner wire and 4,000 for outer one. The composites have been
worked very uniform in the cross section. Table 1 shows the specifications and measured
values of single staked 61lm filament wire and cable (outer cable). The critical current density
of wire is 2,830Nmm2 at 5T. The critical current of cable is 1O,200A at 5.6T, and the Ie
degradation during cabling is approximately 4%. Fig. 2 shows the critical current density vs.
magnetic field on high Jc wires and single stacked 61lm filament wires. Solid circles are for
this single stacked 61lm filament wires.
Table 1. Specifications and measured values of single stacked 61lm filament

wire and cable (outer cable)
Items Units Specifications Measured Values

Wire Wire diameter mm 0.648±0.OO25 0.648
Matrix Ratio 1.8±0.1 1.71
Filament Diameter 11m 6±0.5 6.1
Twist Pitch mm 13±1.5 13.1
Critical Current A at 5.6T :2:286 295
Critical Current Density Nmm2 at5T 2,830
Cable No. of Strand 36 36
Cable Mid. Thickness mm 1.156±0.006 1.156
Cable Width mm 11.68+0.05,-0 11.685
Critical Current A at5.6T ~9,780 10,200
4000r-~--'-----~----~----~------~--~
<>--<> High Jc wire

..-... Single stacked 6mm fil. wire
...,u
f 3000
en
z
w
o
I-
Z
W
a:
a:
5 2000
....J
«
u
i=
a:
u
4 5 6 7 8 9
MAGNETIC FIELD (T)
Figure 2. The critical current density vs. magnetic field

on high Jc wires and single stacked 61lm filament wires
1083
HIGH Jc WIRE AND LARGE KEYSTONE ANGLE CABLE
The critical current density, Ie, is influenced by such parameters as grain-boundaries,

dislocations and precipitations of aTi in the Nb-Ti filaments. This influence of these
parameters varies according to plastic deformation, such as extrusions and drawings, and
according to the intermediate heat treatments. It was clarified that the Ie is greatly influenced
by intermediate heat treatment conditions. Excellent Ie properties have been obtained by
optimization of these conditions. 4 ,5 The conditions of intermediate heat treatments are
expressed by such parameters as temperature, the duration of heat treatment, and the number of
steps in the heat treatment process. For various treatment conditions applied to 4-61lm
diameter Nb-Ti filament wires manufactured on a laboratory scale.
Figure 3 shows the maximum Jc at ST at each step of the heat treatment process. The
duration of the intermediate heat treatments was SO hours in all cases. Fig. 3 shows that Ie
increase remarkably as the number of heat treatments steps expands, and each curve shows a
point at which the maximum Ie is obtained. Fig. 4 shows the change in the Ie at ST during a
multi-step intermediate heat treatment processes, where the temperature of the first step was
708K. The Jc on four steps intermediate heat treated wires is 3,200A/mm 2 at ST. An
extremely high Jc at ST, 3,460A/mm 2 , was thus obtained after six steps of heat treatment.
Based on these fundamental investigations concerning the relationship between Ie and
intermediate heat treatment, high Ie wires have been developed in a production scale. As the
wire diameter for the inner cable is relatively large, only the four-steps intermediate heat
treatments have been applied. At ST the Jc of this wire is 3,200A/mm 2, as same as the Jc of
wires fabricated in a laboratory scale.
TEMPERATURE (·C)
~ 3500.-3,6_0______
4,00_______
4,40__--.
l{)
N f-
E
0{)
3500
E -.;;
"-
« NE
E
o
--,
3000 "'-
<{
>- u
-, 3000
f-
Uj >-
Z f-
W U5
o Z
W
f- 0
i'a:5 2500 f-
z
w
a: a: 2500
:::> a:
o :::>
U
-.J
« ...J
o <{
u
i=
oc t:
a:
o 2000~----~~----~------~ U 2000
620 660 700 740 0 2 3 4 5 6 7
TEMPERATURE (K) NUMBER OF TREATMENTS
Figure 3. Transitions of critical current density Figure 4. Transitions of critical current density at 5T
at 5T and 4.2K as a function of intermediate and 4.2K resulting from intermediate heat reatment at
heat treatment progressively lower temperature
1084
Table 2 shows the various properties of high Jc wire and large keystone cable. The
keystone angle of developed cables is 3.10 twice as the conventional cable. The degradation of
critical current, 5.6%, is slightly larger than that of 1.60 or 2.00 keystone angle cable. Fig. 5
shows the cross sectional photos of the large keystone angle cable. The high Jc wires are
applied for this cable. The cable have been worked very uniform in the parts of the thin edge.
Table 2. Various properties of high Jc wire and large keystone angle cable
Items Units Measured Values

Wire Wire diameter mm 0.844
Matrix Ratio 1.56
Filament Diameter 11m 6.5
Critical Current Aat5T 700
Critical Current Density Nmm2 at5T 3,200
Cable No. of Strand 23
Cable Mid. Thickness mm 1.39
Cable Width mm 9.71
Keystone angle deg. 3.1
Critical Current Aat5T 15,200
Figure 5. The cross sectional photo of large keystone angle cable
LOW AC LOSS WIRES AND CABLES
When the Cu-Mn alloy is fabricated into the matrix for the wire, there is a possibility that
the matrix shows magnetic hysteresis. Fig. 6 shows the hysteresis loops for as drawn Cu-Mn
alloy wires with Mn content of 0.5, 1.0, 2.0, and 5.0at% in bipolar fields between -1 T and 1T
at 4.5K. The sample wires have been prepared by drawing following cold rolling with
grooved rolls from the vacuum melted ingots. The hysteresis loss grows higher with
increasing Mn content for the Cu-Mn alloys, showing stronger irreversibility in the hysteresis
loop. Thus, a Cu-Mn alloy with Mn content over 1.0at% (0.86wt%) should not be taken as
the matrix material, because it gives additional hysteresis loss to that of 72kJ/m 3 for the 2.51lm
filament wire with Cu-0.5wt%Mn alloy matrix. The hysteresis loss for the Cu-0.5wt%Mn
alloy is clearly shown to be negligible.
Figure 7 shows the cross sectional photos of 2.5llm filament wire with Cu-Mn matrix
and cable. The wire are fabricated by double stack process because number of restacked
1085
monofilament rods is very large. The wire diameter are 0.81mm for inner wire and 0.65mm
for outer one. Number of wires are 30 for inner cable and 36 for outer cable. Fig. 8 shows
the comparison of histeresis loops of 2.51lm and 6.51lm filaments strands at 4.5K.
Magnetization values are magnetic moments per unit Nb-Ti volume. No extra magnetization
peak close to zero magnetic field (indicating the generation of proximity effect) appears. The
area of hysteresis loop is substantialy smaller for 2.51lm filaments strand. The hysteresis
losses for 2.51lm filaments and 6.51lm filaments wires are estimated from the hysteresis loopes
are 72kJ/m 3 and 180kJ/m3 per cycle, ±5T.
'5 r---.----o----,---.----.--~
z
g °t----+-----+--~F=---_+_--__+--___I
.s;::::
w
Z ·5
Cl
<
:::;:
., 0 1-----+-:,0---+----f-----I---__+--___I
. ' ~, .':'"5- - ..:-'1.0

::------=-0:'::
.5------:0"'::-
.0 ---:-'
0 .-:-
5 - -- '-'-
.0-----".5
APPLIED FIELD (T )
Figure 6. Hysteresis loops of Cu-Mn alloys at 4.2K
.•..•....
•••
........ ,.
..·i::.·:.-!.·:.\\)):(·:(
\ \,\. \\\\~...
';.
.........:,..
..·:.. . •. ... .·
•••
••••~ ,:
•\\••••••
··I.~
1:1..-
..,::,::-
• ••••• .~..
.•.
•...•...•.•...........
··i:·:·:·:·:·:·:·:·:·····
-11:)::111:
••••••••• ••••• •••
..\ ,,,., \ ., F""lft
(a) Inner wire (b) Outer wire
:.. ',r. ' I• · l( - • '" , ,--' .

-:'f Jl
/
~"
""1('
• • I • I.. • f. • ~ •
'- ---- ;
I •
• J~" ":.' \'.. )' . \ .,\ '\ ~

"" I · • : .. .,J
(c) Inner cable
(d) Outer cable

Figure 7. Cross sectional photos of 2.5llm filament wire with Cu-Mn matrix and cable
1086
80r----,-----,----~----,_----~--_,
+. . . . . . . . . . . . . . . . ;. . . . . . . . ..
~r---~----_+----~----~----~--~
40 ...............
f=" :.:::.:.;.~.~.; .-~~.,;;~~;
.s; -:
i
CU-Mn matrix
2.5 jAITIll..
Z 20 ................. +............................................... ·················r·················
'1- '
~ -20r---~-----+--~~~~~----0---~
-40 r--------i-----+-----W----~----+----~
-~ . . . . . . . . .l". . . . . . . . . . . . . . . . . ··································r···············
-80 L---_ _---'-_ _ _ _--'-_ _ _ _- ' -_ _ _ _...L..____' - -_ _--.l
~ 4 4 0 2 4 6
APPLIED FIELD (T)
Figure 8. Comparison of hysteresis loops of 2.5Jl1Tl and 6.5l-lm filaments strands at 4.5K
CONCLUSIONS
1. Single stack processed wires and cables have been developed in production scale using the
hydrostatic extrusion. Single stack process generally makes yield rate high.
2. By the four-steps intermediate heat treatment, high Ie, 3,200A/mm2 at 5T, wires for inner
cable have been fabricated in industrial scale.
3. Large keystone angle cable was developed for the wedge-less magnet without severe
critical current degradation.
4. Wires and cables with 2.51lm small filament and Cu-Mn alloy matrix have been
successfully fabricated. These wires and cables showed very small AC losses, especially
in hysteresis losses.
ACKNOWLEDGEMENTS
The authors wish to express thanks to professors T. Shintorni and K. Hosoyama of

KEK for their useful discussions concerning the cables for the dipole magnets.
REFERENCES
1. S. Sakai, et aI., Proc. of ICMC'82, 301(1982)

2. S. Sakai, et aI., Proc. of ICFA'86, 95(1986)
3. S. Sakai, et aI., Supercollider 1, Plenum, 669(1989)
4. G. Iwaki, et aI., Supercollider 2, Plenum, 621(1990)
5. G. Iwaki, et aI., Hitachi Cable Review, lO:79(1991)
1087
THE CRYOGENIC SYSTEM FOR THE MTL MAGNET TEST STANDS
Q. S. Shu, D. Hatfield, P. Reddy, I. Syromyatnikov,

R. Trekell, and A. Zolotov

Dallas, TX 75237
ABSTRACT
This paper briefly describes the cryogenic test capabilities of the Magnet Test Laboratory
(MTL). The instrumentation for controlling the operating condition of the magnet cryogenic test and
for verifying the requirements of the sse magnet performance is introduced. The development of
the thermometer system, particularly the He vapor pressure thermometers with differential pressure
transducer, is presented in detail. The lO-kA vapor-cooled power leads were optimized thermally,
with consideration for the different fin shapes, diameters, lengths, and RRRs of the power lead
material. Two mechanical designs are introduced. The anti-cryostats, so-called warm bore and warm
finger, that provide a warm environment to allow the magnetic field-measuring probe to run through
the 4.2-K beam tube are described. The warm finger for sse short-magnet cryogenic tests was
manufactured and successfully used. Finally, the feed and end cans-used to provide cryogens to the
magnet being tested as well as cryogenic vacuum-and the support of other instrumentation are
described.
INTRODUCTION
About 11,000 superconducting magnets of various types, including dipoles, quadrupoles, and
specialty magnets, will be produced by magnet subcontractors and by the SSeL itself for use in the
sse accelerator. I The MTL will be used to test a considerable portion of the total magnet
production in order to control the manufacturing process and to verify the requirements of the
magnet performance. With ten cryogenic test stands, MTL is capable of housing tests for 30 dipoles
and 5 quadrupoles per month. For developmental purposes of the sse magnets, there will be two
R&D test stands, designed to accommodate heavily-instrumented magnets. There will also be one
test stand capable of testing a three-magnet string. In order to accomplish the test tasks, a
comprehensive cryogenic system, including refrigerator, cryogenic instrumentation, vapor-cooled
power leads, anti-cryostats, feed and end cans, and other associated systems, has been or is being
procured, designed, developed, and tested. This paper will briefly discuss the progress to date.
No. DE-AC35-89ER40486.

CRYOGENIC TEST CAPABILITIES
Figure 1 is the general cryogenic test flow schematic at the sse MTL. He is transfered from
the refrigerator by a pumping box through the sub-coolers in the distribution boxes, then into the
feed cans. In order to better control the He temperature to the magnet, there is a temperature-
controlled heater located in the He transfer line between each distribution box and feed can. The He
from the feed can first flows through the single phase return line, by-passing the magnet to the end
can, then turns around to cool the magnet cold mass. This flow arrangement avoids the influence of
heat leaks on the magnet cold mass due to a heat leak through the vapor-cooled power leads.
Feed Can End Can

Refrigerator
2150 W Refrigeration
22 g1s LHe
22 g1s LHe
Figure 1. Cryogen flow schematic of the magnet test at MTL.
The cryogenic test system capabilities at MTL are briefly summarized as follows:
1. There will be a total of 2150 W of refrigeration power at 4.2 K, plus 22 gls LHe.
2. Total single-phase mass flow rate will be maximum 650 g/s.
3. The clean, cooldown, warm-up and purge (eeWP) system flow rate is 110 gls.
4. The lowest temperature of the system is 2.5 K.
5. The MTL is able to cold-test 7 magnets simultaneously at 4.5 K and 100 gls. The cryogenic
test stands are configured as follows:
8 production magnets (dipole length 15.5 m, quadrupole length 5 m)
2 R&D magnets
A three-magnet string
Several short magnet vertical test dewars.
INSTRUMENTATION DEVELOPMENT
General Introduction
In order to control the operating condition of magnet cryogenic tests and to verify the
requirements of the magnet performance, a large number of instruments 2 will be allocated and
installed inside the feed and end cans of the test stand. This instrumentation includes the temperature
sensor system, cold and warm pressure transducers, cryogenic mass flow meter groups, and the
1090
wmng systems for wmng all of the sensors out to reach the data acquisition. Besides the
instrumentation, installed in the cans of the test stand, there is additional instrumentation placed in
the magnet itself: strain gauges, strip and spot heaters, and the voltage taps for quench detection.
Figure 2 is the simplified block schematic of the sensor distribution.
End can
Feed can
Power leads Temperature

Heater leads Magnet (CGR, Pt, vapor)
Temperature Pressure
(CGR, Pt, vapor) (Cold and warm
V-Taps
Pressure Transducer)
Strain gauges
(Cold and warm Differential P
Strip heaters
Transducer) Feed-Throughs
Spot heaters
Differential P Vacuum
Temperature
V-Tap Connection
Feed-Throughs Gas Heaters
Vacuum Connection
Gas Heaters
Figure 2. The instrumentation distribution of the sse magnet cryogenic test stand at MTL.
Design of The Thermometer System
Thermometer Accuracy Requirements. According to the heat leak budget of the sse
magnet, the temperature rise of the cryogens was calculated between both ends of the cold mass, and
the temperature change of cryogens was noted along the magnet 20-K shield and along the 80-K
shield as functions of cryogenic flow rates and static heat leaks, respectively. The results were
presented in detail in a separate paper. 3 The choice of the temperature accuracy requirement for the
sse magnet test is based on the quench test results as well as the potential high-accuracy heat leak
measurement. Temperature accuracy requirements for magnet testing are most important in the
temperature region of 2-10 K. The thermometer accuracy requirements in MTL are given in
Table I.
According to the accuracy requirements, three types of thermometer packages were developed
for the MTL cryogenic tests: package #1 for He line (2.5 K-300 K), 2 carbon glass, 2 platinum;
package #2 for Nz line (77 K-300 K), 2 Platinum; package # 3 for He line (single-phase He),
4 vapor pressure thermometers (the VPT package).
The Design of VPT System. The VPT has a higher accuracy than resistance thermometers,
particularly in measuring the temperature differences between two locations by using a VPT group.
In addition, VPT is also better to measure an average temperature than are resistance thermometers.
The VPTS atMTL will be located across the magnets and connected to pressure transducers, as
shown in Figure 3. To reduce the heat transfer to the VPT bulb the capillary tubing should be
thermally stationed to the 80-K shield and also to the He piping. The VPT mounting detail is shown
in Figure 4. The heat leak to the bulb at 4 K from the single phase piping will be 5 IlW and 1.2 mW
from SO K to the single phase, when stainless tube with 3.IS-mm o.d. and 1.24-mm thickness is
used. The volume of the bulb is 0.3 cm 3 , and the volume of warm system connected to each tube is
70 cm 3. When the system is charged to 52 psi, the VPT can be operated from 3.5 K with SO% of
LHe in the bulb to 5 K with 17% LHe in the bulb. With a Sensotec Super TJE absolute pressure
transducer (50 psi range, 0.05% F.S. accuracy), the T accuracy will be ± 2.5 mK at 3.5 K and
tl1.1 mK at 5 K. If the temperature differences across the magnet or between the upper and lower
nozzle are to be measured, the Sensotec Z differential pressure transducer (2 psi range, 0.25% F. S.
accuracy) will measure ~T = 150 mK with an accuracy of ± 0.5 mK.
1091
Table 1. MTL Thermometer Accuracy Requirements.
Relative
Cryogen Type Mounting TRange Accuracy
He CGR in stream 2.4-10 K ±2rnK

CGR in stream 10-20K ± lOrnK
CGR in stream 20-40K ±50rnK
Pt in stream 4O-80K ±50 rnK

Pt in stream 80-300K ±500rnK
N2 Pt in stream 80-300K ±50rnK
Feed Can End Can
BV
CHV
Cold Mass
1,1
VPfBulb
Figure 3. The Schematic of the VPT system with the differential pressure transducers, used for precise
temperature measurement: DP - Differential Pressure Transducer; BV-Bypass Valve; MP-Mechanical
Pressure Transducer; PT-Pressure Transducer; CHV-Charge Valve.
Note: Wrap entire assembly with MLI.
Solder on
5O-mm length.
He Flow
VPT Bulb, 0.3 cm3 Tubing loop, whole length - 100 mm.
volume.
Figure 4. The VPT mounting assembly.
1092
THE DESIGN AND THERMAL OPTIMUM ANALYSES OF
10 KA VAPOR COOLED POWER LEADS FOR MTL
The optimal thermal design is required for the He vapor-cooled power leads that will supply
10 kA to superconducting magnets in the MTL. A survey of existing designs showed that there are
different approaches to designing vapor-cooled power leads as well as different objectives in
determining the optimum. A design method was followed that meets the requirements for vapor-
cooled power leads to be used at the sse.
The diameter which minimizes the Carnot work for a given lead length has been determined
for a spiral-fin, 100kA power lead design for several different lengths, two different rm geometries,
and two levels of RRR.3 The acceptable lead dimensions for a prescribed set of cryogenic system
refrigeration and liquefaction load budgets have also been calculated. 3 In general, the constraints
permit the use of the optimum design length and diameter. This note provides direction for a power
lead designer in the selection of these important dimensions.
A summary of the acceptable lead dimensions for a heat transfer constraint of 7.9 W to the 4 K
level and He mass flow of 0.5 gls is illustrated in Figure 5. The acceptable diameters shall be smaller
than the heat transfer constraint and larger than the mass flow constraint. The BNL design, with
L=46 cm and D=1.4 cm, falls slightly above the optimum when RRR=IOO, and above the mass flow
constraint when RRR=40. The Fermilab design with L=130 cm and D=2.54 cm is outside the range
of computed data. Even so, it appears that it, too, is within the acceptable design space determined
with this model.
For the MTL a lO-kA lead is being designed, as shown in Figure 6, based on the analysis. The
main difference between design A and design B is that the electric insulation assembly, which
isolates the ground potential parts of the lead from the high power parts, was moved into a warmer
region in design B from design A. Therefore, design B has an advantage over design A in that it
prevents vacuum and He leakage in the o-ring seals due to low temperature. The preliminary
selection of the length and core diameter are 120 cm and 2.2 cm, respectively. For the MTL 100kA
lead, the fins will be soldered to the core.
3.0
2.5
-
2'
u
2.0
1.5
HEAT FLUX UMIT = 7.9 W
~/~ __---..:==::::::::::::=,.....: THERMAL OPTIMUM
'" He MASS FLOW = 0.5 GIS
1.0
40 60 80 100 120
LEAD LENGTH (eM)
Figure 5. The computing simulations of the optimum thermal design of lO-leA vapor-cooled leads as functions
of diameters, lengths, fin shapes, and RRRs for MlL<--- = =
RRR 40; ----- - RRR 1(0).
1093
In order to accommodate this, a slight modification to the fin dimensions is necessary. The fin
length will remain the same, but the thickness and spacing tentatively are 0.127 em and 0.381 em,
respectively. It is observed from these simulations that there is little difference in the lead
performance for this magnitude of change in the fin geometry. The authors will introduce more
details of the design and analyses in a separate paper.
ANTI-CRYOSTATS (WARM BORE AND FINGER) DESIGN
It is necessary to design anti-cryostats, so-called warm bore and war finger, for the purpose of
making magnetic field measurements during magnet cryogenic tests. The wann bore inserts into the
magnet beam tube, which is at a temperature around 4.2 K, and it accommodates a magnetic field-
measuring rotating coil, NMR probe, and Hall probe within its warm space. Besides preventing the
rotating parts of the probe from freezing during the measurement operation, it also reduces the heat
leak that reaches the magnet cold mass through the beam tube. Therefore, the warm bore is a special
dewar in itself, which works between 4.2 K and about 283 K. Contrary to a normal dewar, the
temperature of the outer wall of the warm bore is 4.2 K, whereas the inner space of the bore is
maintained at about 283 K. The thermal insulation vacuum jacket of the bore is very long (19 m in
SSC dipole magnet test) and very narrow (only around 2 mm). The thermal insulation assembly in
the vacuum space of the warm bore must be specially designed to meet the design requirement of
heat leak. A temperature sensor and temperature controller are recommended to monitor and adjust
the warm bore heater, resulting in as little heating as possible for maintaining a warm environment.
It is desired that the magnetic field measurements be performed at the full magnet operating current.
0"
¢ CU )
De ign A Design 8.
Figure 6. The two mechanical designs of a 10 kA vapor cooled power leads.
A warm finger about 3.3 m long, with an outer diameter of 40.56 mm, a space of 2.88 mm for
filling superinsulation, and an inner diameter of 28.85 mm was designed, tested, and used for a sse
short-dipole cryogenic test. 4 The thermal performance is in good agreement with design
specifications. The measured heat leak was 3.2 W. A warm bore for SSC full-size dipole magnet
cryogenic test is under design. The schematic cross section of a warm bore design in which the
warm bore has a vacuum separate from the cryostat vacuum is shown in Figure 7. This design uses
three bellows, two of which compensate for thermal movement during magnet cooldown and warm-
up. The other bellow is used for coaxial alignment.
1094
VACUUM
4.2 K
280 K
DIPOLE MAGNET
END CAN
WARM BORE
Figure 7. A schematic of a warm bore and its simplified cross section.
One of the most important issues in designing a warm bore is how to further reduce the heat
leaks. The heat leak though the warm bore to the beam tube can be calculated by the temperature
change, !lTo, in the nozzles of the magnet cold mass. The other important fact to know is the
temperature change in the annular areas around the superconducting coil, which depends on the
following factors: (1) the heat leak though the warm bore to the beam tube, (2) the portion of mass
flow rate through the annular area to the total flow rate, and (3) the heat transfer distribution to the
He in the annular area, cold mass and the He in the by-pass passages. The temperature increase in
the annular region as a function of the heat leak and single-phase He mass flow rate in a SSC dipole
magnet has been calculated. 3
THE DESIGN OF FEED AND END CANS
In order to provide cryogens to the magnet being tested as well as the cryogenic vacuum and
the support of other instrumentation, a magnet test stand system is employed consisting of a feed
can, an end can, and a cold test stand. The feed can first feeds the cryogens to the end can by-passing
the magnet, then the He turns around into the magnet and cools the magnet to 4.3 K. The used
cryogenic He returns to the refrigeration system of MTL. The feed and end cans also provide all the
electrical wiring connections and power connections to transfer all of the signal and power to and
out of the magnet being tested. The feed and end cans have the necessary mechanical flanges to hold
the warm bore in place and to allow a vacuum between it and the beam tube. Both the cans have
provisions to create a thermal insulating vacuum in the magnet, with thermocouple and cold cathode
gauges to read the vacuum. The magnet and the cans are elevated from the ground by the cold test
stand, which distributes the magnet load, stabilizing the magnet from quench and vacuum forces,
and also aligning the magnet with respect to the cans. Two pairs of feed cans and end cans will be
designed and manufactured for MTL by Fermilab per SSC specifications.
ACKNOWLEDGMENTS
The authors wish to thank T. Peterson, M. Kuchnir, D. Orris, P. Mazur, and R. Fast of
Fermilab, and A. Prodell, G. Ganetis, and A. Manone of BNL for their contribution to the
development of the MTL cryogenic system. We appreciated the fruitful conversation and discussion
with the following colleagues: J. Tompkins and M. Coles of SSCL, R. Gregory of CERN, and
M. Levin of UNK. Also thanks are given to our colleagues at SSC Laboratory. They are: D. Finan, J.
Wei send, 1. Demco, H. Barton, A. Yucel, J. Missig, M. McConeghy, A. Devred, T. Ogitsu, and
R. Dorman.
1095
REFERENCES
1. "Site Specific Conceptual Design," SSCL, 2550 Beckleymeade Ave., Dallas, Texas (1990).
2. I. Syromyatnikov and Q. S. Shu, "Feed and end cans instrumentation design and specification," internal
doc., Test and Data Management Department, SSCL (1991).
3. Q. S. Shu, I. Syromaytnikov, and A. Zolotov, "MTL planning and instrumentation development for heat
leak measurement," Workshop on Testing of SSC Magnet Cryogenic Performance, SSCL, Dallas,
Texas, February 11-12,1992.
4. Q. S. Shu, J. Demco, I. Syromyatnikov, and A. Zolotov, "Optimum thermal design of the 10 kA vapor
cooled power leads for the magnet cryogenic test at the MTL," MD-TA-213, SSC Laboratory
(1992).
5. Q. S. Shu, "Design and Specification of the Warm Finger for Magnet Vertical Cryogenic Test," Test and
Data Management Department, SSCL (1991).
1096
A VERSION OF THE RADIATION-RESISTANT
FAST CALORIMETER
Yu.K. Akimov and B.G. Zalikhanov
Joint Institute for Nuclear Research

Dubna, Russia
ABSTRACT
A new approach to the design of a forward hadron calorimeter is described. It

is based on using fast proportional chambers and remote preamplifiers, which also
serve as fan-ins. Signal wires of the chambers are directed along a beam, and a large
number of them are connected to one fan-in to decrease the gas gain.
The forward hadron calorimeter of the Collider Detector at Fermilab (CDF) is
composcd of proportional tube chambers and steel sheets. l Forward calorimeters at
the Superconducting Super Collider (SSC) and other new colliders have to operate
in zones with essentially higher radiation. 2 This causes wire coating, which results in
loss of gas gain. 3 Naturally, it would be very attractive to use a unity gain readout
technique. Such a tcchnique may be based on using compressed gas and low-noise
preamplifiers. 4 - 7 A version of the forward calorimeter with a 100-atm argon-methane
mixture is considered in Reference 8. There, ionization tubes have a diameter of 4 mm
and a wire diameter of 50 pm. They are placed at a small angle ('" 1°) to the incident
particle direction. According to measurements in Reference 9, the charge collection
time is 60 ns/mm. Although this version of the calorimeter is very attractive, it cannot
distinguish events in two SSC beam bunches following one another in 15 ns.
Proportional chambers with a CF 4 /isobutane gas mixture lO at the normal pres-
sure with a small distance between cathode and anode planes are much faster. The
new technologyll allows a high degree of accuracy in positioning the planes with the
gap of 1.5mm and anode wire spacing of 1 mm. Such a chamber with a sensitive area
of 55 x 60 cm 2 and anode wires 30 pm in diameter was constructed and investigated.
With the radioactive source 90Sr, a time jitter was FWHM = 6 ns. For a time interval
of 15 ns, an efficiency was 98% in the 2700-3150 V interval. It has been shown that
the utilization of a thick anode wire (of diameter;::: 30 /lm) in a narrow-gap chamber
creates, at the time of the discharge, a "quasi-uniform" electric field in which the
discharge proceeds via a sole stage, owing to ionization by electron impact. 12 This is
related, first, to the fact that average value of E / p in the gap of such a chamber is
about an order of magnitude higher than in standard chambers. Secondly, the utilized

gas mixture (80 CF4 + 20i - C 4H 10) totally excludes development of the gas discharge
based on a photon mechanism. The above circumstance makes more preferable the
use of narrow-gap chambers with small steps between the thick sense wires for ioniza-
tion measurements. All technological operations are developed enough to construct
larger chambers. Below we consider a 1.2 m x 1.2 m chamber with wires 0.1 mm in
diameter (D).
Wire coating depends on a charge q collected on a wire unity surface. A value of
q is proportional to M = Kd/ND, where K is the gas gain, d is the gas density, and
N is the number of wires in a certain volume. It is possible to have M even smaller
than for the calorimeter with ion tubes at high pressure (in Reference 8).
Figure 1 is a schematic view of two modules of the calorimeter with proportional
chambers. Steel plates (1) are used as a basis for fixing chamber electrodes (9, 10, 11).
A plate is 2 cm thick, and the distance between the cathode planes (9-1) and (9-2)
is 6 mm. The cathode planes are made of mylar with a conductive carbon surface.
Mylar surrounds the plate (1) and is located over the rods (3). The rod (4) presses
mylar to the plate by means of screws, resulting in the stretching of the cathodes. It
is necessary to note that the diameter of the rods (4) together with the double mylar
thickness is exactly equal to the distance between the rods (3). But this diameter is
Figure 1 is smaller than the real one in order to make the cathode (9-1) fixing easy
to grasp.
A relative disposition of the cathode and anode planes is adjusted by means of
the calibrated rods (3, 6, 7), the distributing plates (5, 8), and the guiding calibrated
rods (2) (2 mm in diameter). The latter are stretched and attached to the plate (1)
with compound "Durmetal AG". The copper-wire spiral with the enamel covering
is wound on the rod (6). This wire, 1 mm in diameter, determines the anode wire
spacing. An anode wire is wound on the rod (6) by rotating the module around the
axis, which passes through its center perpendicular to the plane shown in Figure l.
The second cathode (ll) has a conductive carbon surface on both sides. It is stretched
by means of screws to the movable holder (12), which is attached to the slat (13).
After the cathode is stretched, one installs the distributing plate (8). Then the rod (7),
together with the cathode, is pressed to a side of this plate. In the course of assembly, a
gap between the cathode (ll) and the anode plane (10-2) is determined by a template
on the edges of the anode rods (6) by regulating screw rests (14).
As assembly of nine steel sheets and eight sensitive, 6-mm gas layers between
them constitute one section for the calorimeter. There are additional thin, insensitive
gas layers between the first cathodes and the sheets. Their thickness is 1 mm at one
edge and '" 2.2 mm at the other edge. This increase is provided by using screws of
larger diameter at this edge. It results in some widening of one side of the section
(from 26 cm to 28 cm).
Figure 2 is a schematic view of the calorimeter composed of these sections (I-V).
Installed as far as 10 m from the interaction point, the calorimeter covers a pseudo-
rapidity region of 3 ~ 'r] ~ 5. It consists of two almost identical parts to measure en-
ergies E;S 5 TeV. The first one, discussed below, absorbs hadrons with E;S 200 GeV.
This calorimeter has 16 sections, five of which are shown in Figure 2. There are
small angles Q between the shower axis and the nearest wires. In the central horizontal
plain A, Q :::::: 1.5°. The total hadronic shower width is 25-30 cm. Minimum ionization
losses of relativistic particles are 11 MeV / cm in steel and 5 ke V / cm in CF 4/isobutane.
Taking into account the total thicknesses of the sensitive gas layers and steel
in one section, one can find a relation of ionization losses in them to be equal to
1.2 x 10- 4.
1098
13
12
8
7
6
5
I..j
3
2
14
I------j
0 fOlINII
9-2 91
10-2 10
11
Figure 1. Schematic view of two modules of the calorimeter section.
I
IT m Jl V
I
I
I
~I· f.,PART/CLES
~ (B)
-I
(a)
Figure 2. Schematic view of one quarter of the calorimeter: (a) is the face side, (b) is the view in
the central horizontal plain A. I-V are separate sections consisting of proportional chambers and steel
sheets directed along the beam.
1099
Let us assume that only half of the particle energy E is observed in our sampling
device. In this case 2 x 10 3 electron-ion pairs are expected for AE = 1 GeV in a
shower.
The chamber wires make up lines with an impedance of p = 150 f!. One will
need transmission lines to take electronics out of the high radiation zone. (If the lines
are long, low attenuation in them can be obtained by cooling. l3 ) Each line must have
the same value of p and be connected to a preamplifier through a resistor of value p,
too. The preamplifier can serve as a fan-in, if its layout impedance is much smaller
than p. Such a circuit is shown in Figure 3. 14 A load of n lines for the preamplifier is
R = 2p/n. The lower R is, the higher is the noise, which was measured with signals
5 ns long. Table 1 shows the result of the measurements.
>~m
:~5
-=- p
+12v
Kn99
Figure 3. Fan-in. L is 3 winds of 2.5 mm-diameter on a ferro-magnetic circle.
Table 1. Output pulse height U and r.m.s. noise (T versus R.
R (n) 50.0 25.0 19.0 8.0 3.7 1.6

U (%) 100 99 95 85 74 63
(T (nA) 170 450 500 750 900 1200 1500
The last case corresponds to 180 lines with p = 150 f!. Here the noise is a = a( Q) =
alAI x ~tlsl/e = 4.6 x 10 4 electrons. Of course the lines should be carefully screened
against outside electromagnetic noises.
Wires can be connected in series to decrease the number of outputs from the
chamber. It only increases the dispersion of the signal spreading time along the
chamber wires, but not a (nA). We connect only two neighboring wires. Taking
into account a real signal duration, one can expect a ~ 105 e. About 20 fan-
ins will measure the total energy E of a shower. Hence the total noise at = v'2Q,
a = 4.5 X 10 3 e. A number of electrons in a signal is Q = 2 X 10 3 K E. Let at = Q if
E = 1 GeV. Hence, K = 220.
In fact one should take K = J(TJ) because the more TJ the more E and the
background. One can have K < 220 provided that noise will not be more than the
sampling, path length, Landau fluctuations. It's possible to measure K constantly
by means of Q-particles of 228To with E = 8.78 MeV. This source can be placed on a
steel surface. The energy absorbed in the sensitive gas layer will be about 3 MeV.
This calorimeter has an Af-factor smaller by three orders of magnitude than that
in Reference 1. It is not more than for the version with ionization tubes at 100 atm
in Reference 8.
1100
It seems problematic to use the fast CF 4/isobutane mixture at pressure near
100 atm because of too-high voltage. It will result in high leakage of current and the
corresponding noise, which can be much more than the electronics noise. Besides
that, CF4 is very expensive. But wire aging with CF4/isobutane is less than with the
argon mixture by an order of magnitude.1 5 For low aging it is important to have a
gas flow. Naturally, the normal pressure is the best condition for it.
Summarizing the factors discussed above, we expect radiation hardness of the
calorimeter considered to be comparable with that using ionization tubes at high
pressure (100 Mrad in Reference 8). The main advantage of the version with propor-
tional chambers with very narrow gaps and signal-wire spacing is speed. This is very
important at high luminosities when pile-up can become the main source of pulse
heights fluctuations at energy measurements.
REFERENCES
1. S. Cihangir et al., Nucl. Instr. and Meth., A267 (1988) 249.
2. D. DiBitonto et aI., Nucl. Instr. and Meth., A279 (1989) 100.
3. J.A. Kadyk, Nucl. Instr. and Meth., A300 (1991) 436.
4. V.I. Baskakov et aL, Nucl. Instr. and Meth., 159 (1979) 83.
5. T.A. Amatuni et aL, Nucl. Instr. and Meth., 203 (1982) 179.
6. M. Barranco-Luque et al., BNL 31100, OG 624.
7. D.M. Khazins et aI., Nucl. Instr. and Meth., A300 (1991) 281.
8. Yu.A. Budagov et aL, JINR Preprint El-91-572, Dubna.
9. N.D. Giokaris et aL, Nucl. Instr. and Meth., A291 (1990) 552.
10. J. Fisher et aL, Nucl. Instr. and Meth., A238 (1985) 249.
11. H. Kalmar et aL, Nucl. Instr. and Meth., A307 (1991) 279.
12. B. ZaIikhanov, in Proceedings of Int. Workshop, "Physics at the UNK," Protvino, Septem-
ber 1990 (in press).
13. G.V. Glebovich, Pribory i Tekhnika Eksperim, No.3 (1966) 136.
14. Yu.K. Akimov and V.K. Tyupikov, submitted to Pribory i Tekhnika Eksperim.
15. R. Henderson et aL, IEEE Trans. in Nucl. Science, NS-35 (1988), No.1, p. 477.
1101
HIGH RESOLUTION METHOD OF THE MAGNETIC AXIS LOCALIZATION
FOR MULTI POLE MAGNETS ON THE BASE OF THE GARNET FILMS
TECHNOLOGY
K.F.Gertsev!1 V.L.Gribkov!2
V.A.LiSkov2 and A.J.Chervonenkis2
lInstitute for High Energy Physics

Moscow region
2Physical and Technical Center
Moscow
1. INTRODUCTION
The methods of stretched wires for the localization of

the magnetic axis may be inconvenient sometimes in accelera-
tors and colliders of very high energies because of high gra-
dients, large lengths and small apertures. High gradients may
deform the wires due to the nonzero magnetic susceptibility
and microscopic ferromagnetic particles on their surface.
Long wires have large sagittas and small apertures of magnets
limit the transversal workir~ domains for the measuring de-
vices. Precision optics magnets possess extreme parameters,
in particular. in interaction regions.
The magneto-optic (MO) methods of the measurements pre-
sent some new possibilities for the solution of the above
problems.
The use of MO films for magnetic field visualization and
mapping was proposed first in [1 J, where it was shown that on
the base of Bi- substituted iron garnet films and MO Faraday
effect it's possible to obtain the quantitative vector maps

of complicated magnetic field structure. In [2] this was des-
cribed on a large scale. This method was discussed in [3] in
terms of its applicability to the magnetic axis localization
in quadrupoles of accelerators. In our opinion, the films
technology has great advantages as compared with the collo-
idal solution [4].
In this paper the principles and variants of the films
method are presented and further development of the method
under discussion described.
2. FILMS AND THEIR PROPERTIES
Typical films, used in the experiments had the composi-

tion of YBi2Fe3,7Gal,3012CIII) with typical magnitudes of
specific Faraday rotation of 2°/~, optical absorption coef-
ficient of 0.4dB/~m (~=0.63~) and thickness of 7-10 ~m.
Two versions of the above mathod were investigated.
The first was based on the the use of highly anisotropic
uniaxial films with Hk » ~o' where Hk is the uniaxial aniso-
tropy field and ~Mo is the saturation magnetization. The ty-
pical values of Kk and ~o were 104 Oe and 10-30 G. respecti-
vely. Some pieces of such films were placed inside the quad-
rupole in two orthogonal planes x--y, y--z. while the loca-
tion of the zero magnetic field was determined by the posi-
tion of the boundary between two areas of the f11m having op-
poSite direction magnetization (see fig.1). (The structure of
the magnetic field visualized by means of the MO Faraday
effect was transferred away from the quadrupole by a fiber-
optic wave-guide. Such a guide with a polarizer is the light
source for the film).
Note that in the case G > Gar the smooth domain wall
corresponds to the zero magnetic field isoline. while in the
case G < Gar the two rounding curves touching the ends of the
domain fingers (see fig.2) correspond to +Hs and -Hs ' where
Hs is the saturation magnetiC field of the film. while the
zero field isoline is strictly in the middle between these
two rounding curves. In our experiment Gar was typically in
the region 5-20 Tim.
1104
Figure 1. Smoot.h boundary of t.he wall ( magnet.ic field of UNK
quadrupole, G = 12T/m, scale div . 200!J.m )
Figure 2. Development. of t.he sinusoidal inst.abilit.y caused by

t.he gradient. decrease
1105
Figure 3. Labyrinth structure under low gradient conditions.
Figure 2 shows the picture evolution in the field whose

gradient decreases from left to right. and fig.3 gives the
image labyrinth domain structure in the low-gradient field.
One should bear in mind that the above features are true only
in the case of zero or very small coercitivity He. If He is
comparable with Hs, which is qUite possible when a low-magne-
tization film is close to the magnetic compensation state,
the above strict correspondences will be broken. Our experi-
ments have shown that application of weak, ~ He, alternative
magnetic field eliminates this effect almost completely. The
decrease of the effective Goercit1vity 1s at least several
orders of its initial magnitude.
It is clear from the aforementioned that the domain
structure in h1ghly anisotropic MO films may serve an instru-
ment either for determin1ng the zero field 1so11ne or evalu-
at1ng the gradient.
The best resolut1on this method y1elds 1s now a few m1c-
rons. Note that highly an1sotrop1c un1axial films are insen-
s1tive to the in-plane component of the f1eld.
In the second vers10n films of the same composit1on were
used, but they possessed a weak an1sotropy, Hk ~ ~Mo' with
typ1cal values of Hk and Mo 102C. In this version the films
were placed in the plane normal to the magnet1c axis of the
lens in such a way that the force lines were 1n the plane of
the film. The coincidence of the zero f1eld with the magnet1c
ax1s position was identified as a bright spot on the dark-
background, as is shown in fig.4.
, '06
Figure 4. Magnetic axis visualization with
the help of the weakly anisotropic uniaxial
film ( the real dimension of the bright spot
is ~100fJ.m >.
If a film possesses a weak uniaxial anisotropy, Hk~~Mo'

it would contain a normal labyrinth domain structure in the
absence of the external magnetic f1eld. However, as differed
from highly anisotropic films, ttlese ones are extremely sen-
sitive to the in-plane component of the field: the magnetic
moment vector readily turns to the film plane if even a small
in-plane field is applied. This fact indicates a simple way
of direct determination the magnetic axis position in the
cross section of a quadrupole: one would see a bright spot in
the region of the zero field if the polarizer and analyzer
are crossed strictly because only only in this region the
vector Mo is normal to the film plane and Gci',lc,es the max imum
Faraday rotation. Quite delicate equilibrium has to be kept
between the level of the uniaxial anisotropy field, Hk and
sat.uration magnetization fJ,Mo' bec.au2.e the dimensicns of the
above mentioned bright spct and, hence, the measurement accu-
racy are directly dependent on this equilibrium. In our first
experiments, the diameter of the bright spot obtained was~100
fJ,m. Further development of this ms .hod is now in progress.
One should bear in mind that in weakly anisotropic uni-
axial films the contribution from other kinds of anisotropy
may be significant.. In the case of (III) orientation the
cubic anisotropy manifests itself by the presence of the pat-
tern with third-order symmetry (see fig.4). However. this
contribution does not affect the position of the central
bright sDot a~d, hence, the accuracy of determining the mag-
netic axis position.
1107
3. MEASUREMENT SYSTEMS
Though many problems must be solved at the stage of de-

signing the measurement system with MO detectors. these
solutions are also many. To find the best one. it is neces-
sary to design and test a prototype system. We would like to
note here just a few aspects of the system. An MO detector is
just point-like as compared with the magnet length. We have
to move the detector along the magnet axis in order to get
the integral field picture. In this case there immediately
arise at least two questions: 1) what are the tolerances for
the the degree of freedom of the film during its motion? 2)
how to connect geometrically the detector inside the magnet
aperture with the reference pOints of the measurement system?
TIle answer to the first question does not seem difficult
in terms of technical realization. As to the case when the
film is normal to the magnetic axis. there is actually no
problem in maintaining mechanical tolerances. But this ver-
sion requires further development of the technology. In the
second case the films are to be placed in the main magnetic
planes of the qudrupole. which ar'3 not known a priori, and,
in add it ion. during their mot ion the films displace and turn.
The tolerances are as follows: if the cross section line of
the film for the z coordinate is y = Yo +8x and we measure
the x coordinate of the magnetic axis, then the measurement
error introduced by the parameters Yo and 8 will be bx = 8yo.
So, the mechanics is not compl1cated if we want to obtain the
acceptable value of bx.
As to the geometrical connection of the detector with
the measurement system we think that optical methods can be
applied. But this problem as well as the treatment of the
film pictures are beyond the scope of the present work.
4. CONCLUSIONS
To conclude, we woul(j like to make a few remarks.

1. The MO method can be applied not only to very long
magnets. The mechanics and optics are simpler and the adva~-
1108
tages in terms of high resolution remain. if this method is
applied for short magnets.
2. It is possible to apply the MO methods for multipole
magnets for field visualization in the central region and.
hence. the magnetic axis localization.
3. The dynllinic properties of the investigated films may
be characterized by the velocity of film saturation. which in
some films is as high as 1 km/s. corresponding to the upper
limit of external field of ~108 Hz.
4. The standard and referense magnets with well known
magnet1c axis position are absent in laboratories now. Pro-
bably this design must be as a first aIld very important step
in the MO element applicat ion to trle meaS;lrements of tl1e ac-
celerator magnets.
REFERENCES
1. A.Ya. Chervonenkis. N.F. Kubrakov. Magnetooptic visuali-

zation and topographicsl study of magnetic fields. Sov.
Tech. Phys. Lett. 8(6). 303 (1982).
2. V.V. Randoshkin. A.Ya. Chervonenkis. Applied Magnetoop-
tics. M.: Energoatomizdat. 1990 (in Russian).
3. V.L. Grlbkov, N.N. Klryukhln, V.A. LysKov, L.G. Onoprlen-
ko. A.Ya. Chervonenkis. K.F. Gertsev. Abstracts of
International Symposium on Magneto-Optics. September 1991.
Kharkov, USSR, 10C-30.
4. R. Suguhara. T. Kubo, Y. Oosawa, A colloidal solution of
Fe 04 crystallites to optically locate Hle magnetic centre
j
of multipole magnets, KEK report 89-9. September 1989, A.
1109
CONCEPT DESIGN OF THE HIGH-VOLTAGE TRANSMISSION
SYSTEM FOR THE COLLIDER TUNNEL
Leonard S. Norman

Dallas, Texas 75237
INTRODUCTION
In order to provide electrical service to the Superconducting Super Collider Laboratory

(SSCL) 54-mile-circumference collider of 125 MV A at 69 kV or 155 MV A at 138 kV of
distributed power, it must be demonstrated that the concept design for a high-voltage
transmission system can meet the distribution requirements of the collider electrical system
with its cryogenic system's large motor loads and its pulsed power technical systems. It is a
practical design, safe for operating personnel and cost-effective. The normal high-voltage
transmission techniques of overhead and underground around the 54-mile collider tunnel
could not be applied because of technical and physical constraints, or was environmentally
unacceptable. The approach taken to solve these problems is the installation of 69-kV or
138-kV exposed solid dielectric transmission cable inside the collider tunnel with the
superconducting magnets, cryogenic piping, electrical medium, and low-voltage distribution
systems, and electronic/instrumentation wiring systems. This mixed-use approach has never
been attempted in a collider tunnel. Research into all aspects of the engineering and installation
problems and consultation with transmission cable manufacturers, electrical utilities, and
European entities with similar installations-such as the Channel Tunnel---demonstrate that
the concept design is feasible and practical.
This paper presents a history of the evolution of the concept design. Design studies are
underway to determine the system configuration and voltages. Included in this report are
tunnel transmission cable system considerations and evaluation of solid dielectric high-voltage
cable design.
*Operated by the Universities Research Association, Inc., for the United States Department of Energy under
Supercollider 4, Edited by J. Nonie

PROJECT LAYOUT
The project electrical service will supply two (2) main areas-the West and East
Complexes-and ten (10) service areas equally distributed around the 54-mile collider ring at
the surface.
The West Complex includes the injector facilities, made up ofthe LINAC, Low Energy
Booster, Medium Energy Booster, and High Energy Booster (HEB), the west main
substation, two experimental areas, and the campus. The campus includes technical support
buildings, warehouses, workshops, and emergency services facilities.
E.E
S.A.
S.A ,
E.E.
E.E.
S.A.
l
EAST MAIN
~UBSTATION
E.E.
INTERACTION REGIONS E1.
S.A.
HEB
S.A
LlNAC, LEB, MEB
SA SA
@ COLLIDER SECTOR
• SERVICE AREA (SA)
E.E
• EMERGENCY EXIT EE
AND A.C V (EE)
S.A
EE
Figure 1. SSCL Project Layout.
The East Complex has two large experimental areas with industrial buildings, the east
main substation, and emergency services facilities. The collider ring is divided into ten
sectors, each with its own service area and electrical substation at the surface. These sector
service areas are 5.4 mile apart. The surface facilities include buildings to house control
electronics, magnet power supplies, cryogenic equipment, gas storage tanks, tunnel air
conditioning, and ventilation equipment, and cooling pond. The service area substation
provides all power to the service area and tunnel power for that sector. Halfway between each
sector service area is an emergency exit shaft that provides personnel emergency egress and a
location for air conditioning and ventilation equipment. See Figure 1.
1112
CONCEPT DESIGN HISTORY
1983-1987
During this period, the first studies and concept designs for the requirements of an
electrical system were performed. This was before a specific site within the United States had
been selected for the SSCL. Design groups and consultants looked at all of the technical
requirements of the collider, using as a basis the Fermi National Accelerator Laboratory
experience with its large pulsing and ramping loads. The Fermi collider is approximately one-
half the size of the SSCL HEB. The SSC collider main ring has no equivalent in size or
power requirements; it will be the largest in the world.
Conclusions reached by the design groups were that the minimum system, with an
expected average total system load of 107 MVA and a peak load of 162 MVA, would require
a single utility grid of 230 kV, a 69-kV transmission system around the collider ring, and
13.8 kV for surface distribution. An emergency feed of 69 kV for the collider and HEB
cryogenics systems would be required, as power must be maintained to these systems.
1988-1990
The State of Texas was chosen as the site of the SSCL. Studies and concept designs
were then modified with the now-known transmission system that is available in the area
around the SSCL. Also, total system loads increased to 185 MV A average and 225 MV A
peak.
The concept design proposed that the SSCL be served from 345-kV lines available on
each side of the SSCL project, with transmission around the inside of the collider at 69 kV to
the service areas and 12.47 kV for surface distribution.
1991-1992
Reviews of the SSCL concept design were conducted by design consultants, the utility
company, and an expert advisory board. The design consultants recommended an all 25-kV
system. The utility company recommended an intermediate voltage of 138 kV (between
345 kV and 69 kV) and 12.47-kV distribution voltages in the substations. The advisory
board agreed with the intermediate voltage of 138 kV in the substation, and recommended
138 kV around the collider ring. The board also recommended eliminating the East 345-kV
substation and using existing local 138-kV transmission lines for emergency service.
Figure 2 shows the SSCL concept design with the adoption of the 138-kV intermediate
voltage recommended by both the utility company and the advisory board in the substations,
and 69 kV around the collider ring.
Figure 3 shows the advisory board concept of one substation of 345 kV to 138 kV for
the substation intermediate and the collider ring voltage, with existing local 138-kV
transmission for emergency service.
DESIGN STUDIES
Design studies now in progress are evaluating final load requirements, reliability, short
circuit, load flow, motor starting, operating philosophy, and construction costs. These studies
are now being performed on both systems shown in Figures 1 and 2 as well as with 25-kV
surface distribution. A study for 25-kV distribution for the injectors, only, is also being
performed. The studies are due for completion by late April 1992. The results of these studies
will allow the SSCL to make a final determination of system configuration, collider
transmission, and surface distribution voltages.
1113
LlNAC - LEB
MEB - HEB IR HALLS
CAMPUS EM. SERVICE
IR HALLS
EM. SERVICE
Figure 2. SSCL Concept Design with 69 kV Around Collider.
LlNAC - LEB
MEB - HEB
CAMPUS
IR HALLS
EM. SERVICE
Figure 3. Advisory Board Concept with 138 kV Around Collider.
1114
TUNNEL CABLE SYSTEM
The 69-kV SSCL concept design provides the 54-mile collider with 125 MVA of
distributed power-I 2.5 MV A at each of the 10 service areas around the collider from four
feeders, as shown in Figure 2. If the 138-kV system is adopted, the load around the collider
would increase to 155 MV A, as it would then also serve the East Campus from the two
collider feeders from the west substation (see Figure 3). The normal high-voltage
transmission techniques of overhead and underground transmission could not be applied
because of technical and physical constraints; nor was it environmentally acceptable.
Overhead transmission was not considered viable because lightning strikes could cause line
disturbances resulting in lost experiments. Other-weather related problems such as tornados
and ice storms can also cause long outages. Obtaining a right-of-way for a transmission line
around the collider would be very difficult and would be environmentally unacceptable to the
SSCL and the local community. Underground transmission cable for 54 miles was
considered very impractical, and the cost was prohibitive. The approach taken to solve these
problems was the installation of exposed solid dielectric transmission cable attached to the
collider tunnel roof. The electrical medium- and low-voltage distribution systems and
electronic/instrumentation wiring systems are also installed in the tunnel attached to the roof.
Cable installation design has to be practical, cost-effective, and safe for operating and
maintenance personnel. Various designs were developed, using manhole and duct systems in
the tunnel floor, which proved impractical and very expensive. Cable tray systems are
practical but require excessive access space and are also very expensive for 54 miles.
Exposed high-voltage cables gave some concerns about personnel safety; however, after
explanation of modern cable design, manufacturing, installation procedures, protection
systems, and the policy of very limited access to the tunnel only during maintenance periods,
these concerns were abated.
The SSCL had discussions with various American cable manufacturers on the
application of exposed high-voltage cable in tunnels used not for utilities but for other
purposes, such as transportation or process industry, where limited personnel access was
standard. It became very apparent that such an application has been used very little in the
United States but extensively and very successfully in Europe. There is no code or technical
reason in the United States preventing this application--only custom.
European installations have requirements similar to the collider tunnel installation. After
visits to installations and meetings with European design and installation engineers and with
manufacturers of installed cables in those locations (e.g., the channel tunnel), the design
concept for a solid dielectric high-voltage transmission system for the collider tunnel has been
verified as a good engineering concept. It is a practical, safe and effective application of an
exposed cable system. Favorable engineering and cost recommendations were made by those
involved in the meetings.
Figure 4 illustrates the concept of the cable installation in the collider tunnel.
CABLE DESIGN
The function of insulation in electrical cable, both physically and electrically, is to

provide separation of the conductor from its surroundings. Thus the insulation system must
possess the mechanical strength to support a physical load and the electrical strength to
separate the charges with minimum loss of energy. Also involved are load and temperature
changes. Manufacturing considerations are very important, including the purity and
cleanliness of the insulation material, and the extrusion and curing methods employed during
cable manufacture.
111 5
IR'~lS'.I >.10\ C"ELE
I. 'I4 DOv IRIIlUr.Ctl (AIllC'),' ru I
DISlRIB~" \.tI
6~"'1 OR l.layv
450'1 ,.cHl~~G CU\OUII
'[CeNICAL CAELE
I[CH~jCAL CAlLE
(,~om:, bU<
l[c·mCAL 480'1 cet JU' CAllLE Sl:PPCRI

EVERY 4"0'
U\IS"RUI suppeRI
w 4'-0' CEtllERS
.. rEllLl.1 GAS RE TURN LI\[
2" I~ j'lCGW GAS L ~:: "ROll ( Y

VEHClE
.....;
,
-- ..... :-.-;.-:;.
::::::=:;;-:::~..,J
Figure 4. Tunnel Cable Installation.
A solid dielectric cable is more than simply a conductor and insulation; it is a system that
also includes semi conductive shields to provide smooth isopotential interfaces between the
conductor and the insulation; grounding shields; and metal and extruded jackets to provide
physical protection and to minimize the entrance of moisture.
Pure, high-molecular-weight thermoplastic polyethylene (PE) has the best overall
electrical properties, as well as the highest electrical breakdown strength and the lowest
electrical losses. However, its useful temperature range is limited, and it is sensitive to
moisture. PE has been modified by various means to improve one or more of its
characteristics. However, compromise must always be accepted in the formulation of
materials. Low-density PE can be crosslinked with organic peroxide to form crosslinked
polyethylene (XLPE). Or, it can be copolymerized with monomers and other materials to
make ethylene propylene rubber (EPR) insulation. XLPE can be further modified to make a
moisture-resistant insulation called tree retardant crosslinked polyethylene (TRXLPE). These
three insulations make up the vast majority of cable produced in the United States and the
world.
Cable insulation selection from among XLPE, TRXLPE, and EPR is one of the most
critical determinations to be made for the SSCL's high voltage cable system. Other variables
in cable design are not as controversial as this issue. SSCL findings to date on the features of
the various insulations in cables are set forth in Table 1.
XLPE Best use when no moisture is present

Excellent for high-voltage cables
Cost -effective initial and operating
1116
TRXLPE Best use when moisture conditions are expected
Cost-effective initial (1.1 x XLPE) and operating
EPR Best use when overloading is likely
Best use when flexibility is important
Best use when conditions indicate cost (2 x XLPE) justification
Table 1. Findings Summary.
FEATURE XLPE TRXLPE EPR
Cost 1.1 2.0

(XLPE = 1)
Insulation Material Good Good Many Formulations

Formulation Standard
Insulation Material Good Good Many Producers

Quality Variability
Variability Insulation Good Good Hard to Detect

Material Manufacturing
Cable Flexibility Good Better Excellent
AC Breakdown Good Good Lower

Strength (Initial)
AC Breakdown Lower Good Good

Strength Retention
after Aging
Impulse Breakdown Good Good Lower

Strength
Accelerated Good Good-Excellent Excellent

Cable Life Test
40-Year Good Good Excellent

Expected Life
Coefficient of Thermal Non-linear Non-linear Linear
Expansion
Dielectric Constant Excellent Better Lower
Dissipation Factor Excellent Better Lower
Overload Condition Good Good Better
Insulation Poor Lower Better
Strippable Shield Good Good Poor
1117
SDC DETECTOR FOUNDATION REQUIREMENTS
Jeffrey L. Western and Martin W. Butalla

Dallas, TX 75237
ABSTRACT
The Solenoidal Detector Collaboration (SDC) Detector weighs approximately 32,000
metric tons, and its ability to perform to design specifications is directly related to its internal
alignment The limits of the misalignment tolerance envelope in combination with the detector
weight impose a set of tolerance limits of performance directly upon the foundation structure.
The foundation must accommodate different detector loading conditions during installation,
operation, maintenance, and future enhancements. The foundation must also respond to the
loading conditions within a restrictive set of displacement limitations in order to maintain the
detector's critical alignment, thereby guaranteeing its operational integrity. This paper will
present the results of this study, which has been issued to the Architect Engineer!
Construction Manager as user requirements of design. The total structural system
performance of the combination of both the detector and its foundation will be discussed.
INTRODUCTION
The SDC detector's ability to perform to design specifications is directly related, among
other factors, to its internal alignment. The limits of the misalignment tolerance envelope
impose a set of tolerable limits of foundation performance, outside of which the detector
becomes inoperable. 1 The purpose of this report is to identify the user's requirements of the
tolerable limits of foundation performance that must be imposed upon the detector's
foundation and to present a modest argument in support of the requirements.
SCOPE
The scope of this report includes only the aspect of the detector's limited ability to carry
its own weight by means of its support system under all imposed loading conditions. The
detector's ability to function at all is directly related to the structural integrity of its support
system, which in turn is dependent upon the performance of the foundation under the support
system. 2 The user's requirements for this report constitute the foundation performance
impositions that will guarantee the structural integrity of the detector's support system.
* Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER4086.

DETECTOR DESCRIPTION
For purposes of this report, the SDC detector consists of a barrel toroid outer shell,
a moveable calorimeter inner core and moveable end toroid (forward toroid) that moves
concentrically with respect to the barrel toroid, and the barrel toroid structural support and
alignment system that rests upon the concrete foundation. The approximate total weight of
the detector components is 32,000 metric tons.
The Barrel Toroid
The barrel is a hollow toroid magnet containing and consisting of a series of
transverse metallic plates bolted or welded together} The high strength steel bolts or tie
plates are positioned longitudinally with respect to the axis center line of the detector as
shown in Figure l. The plates are held together in compression by the post-tensioned bolts
or welded plates. The maximum bolt tension or plate stress under barrel toroid bending
contributes to the calculation of the deflection / deformation tolerance envelope of the
detector.
[r 1~~;5~~ - -......... 1.500 FT(

I I
1.752)
( 28.032 )
I I I I I I I I I TT T
LONG
BLOCK (128 TYP)
I
-j- 16.500
SHORT BLOCK
(64 TYP)
I I I I I I I I IT T T T
TTTT T T T T T Tf ~TTI
5X 1.200=6.000
- -.416 - 8X 1.900= 15.200 -
76 - 700 METRIC TON JACKS
5X 1.200=6.000
Figure 1.
Structural Support and Alignment

The support system for transferring the weight of the detector to the concrete
foundation consists of a transverse space frame connecting the angled vertical support legs
shown in Figure 1 and 2. The support legs bear upon two built-up girders that are 28.032
meters long each running the length of the barrel. Refer to Figures I in which are depicted a
series of 38 transverse plates bolted to the built-up girders for absorption of the transverse
tension load component imposed upon the girders by the support legs. Also shown in the
referenced Figures are the 76 hydraulic jacks aligned under the girders for transferring the
girder reactions to the embedded plates in the concrete foundation floor. The jacks provide
the capability of returning the detector from any tolerable deflected condition to its
undeflected configuration. The range of jack adjustment contributes to the calculation of the
deflection/deformation tolerance envelope of the detector.
"20
SDC DETECTOR SUPPORT
MAXIMUM LOADING DIAGRAM
MAXIMUM TOROID LOAD

1759
-2- = 880 METRIC TON/METER
INCLINED PLA TE FORCE

880 MT /m X csc 67.5
= 953 METRIC TON/METER
MAXIMUM PLA TE
TENSION FORCE ~
= 953 cos 67.5 X I.~m

= 219 METRIC TONS
MAXIMUM JACK FORCE
R = 880 MT /m X I.~m
= 528 METRIC TONS PER JACK
Figure 2.
soc - DETECTOR DEFORMATION LIMITS

MAXIMUM DEFORMATION: DEFLECTION CAN NOT EXCEED
2 mm OVER A 28 METER SECTION LENGTH
MAXIMUM RIGID BODY DEFLECTION: +25 mm / -25 mm
OVER THE ENTIRE DETECTOR LENGTH
EXAMPLE: FORWARD TOROID INS T ALLATION
-- -,
,, I
'
,_ J
,
1 _______
I--r--
r---- BARREL
TOROID
1- ~
FO RWARD:
TO ROID ' :' : - TOP O F - -
,,- !- ~ OPERA TING
FLOOR
,
-- - -, ------
f-'-
«:
I II II II II I
~
I I I I" I I I I I I I I I I I I
------- - - -- - - - - -
2 mm J
M AXIMUM --- -- ------- -- I
28.0 m I
Figure 3.
1121
The Forward Toroid
The forward or end toroid is the end of the magnet, which fits directly into the end of
the barrel toroid. Refer to Figure 3 entitled Forward Toroid "SDC - DETECTOR
DEFORMATION LIMITS". It consists of metallic plates bolted together in compression by
post-tensioned bolts. The approximate weight of the forward toroid with attachments is
3,600 metric tons.
Refer to Figure 4 entitled Calorimeter "SDC - DETECTOR DEFORMATION
LIMITS". The approximate weigth of the calorimeter with attachments is 6,000 metric
tons. The calorimeter is shown to be installed in one unit, but consideration is being given
for splitting the calorimeter into halves of 3,000 metric ton each.
soc - DETECTOR DEFORMATION LIMITS

MAXIMUM DEFORMATION: DEFLECTION CAN NOT EXCEED
2 mm OVER ENTIRE LENGTH OF DETECTOR (28 METERS)
MAXIMUM RIGID BODY DEFLECTION: +25 mm / -25 mm
OVER THE ENTIRE DETECTOR LENGTH
EXAMPLE: CALORIMETER INST ALLATION
BARREL
Il
CALORIMETER TOROID
~-i--:-:--~-~
II I I I I II
------- -- II
::: II : : ; -
TOP OF--,
~ - ~ __ ~ _~ __ ~ - ~ - - - - - - - f-::..:-:....:--=-<L-.~LL-l.-Jll----j OPERA TING/!
FLOOR /
I I
Figure 4.
FOUNDATION LOADING CONDITIONS AS REQUIREMENTS
An indirect part of the user's requirements for the operation of the detector is the
concrete foundation's ability to withstand the imposed loads of the detector throughollt all
phases of its life cycle, while restricting load defomlations to fall within the tolerable limits of
the detector's support and alignment system. Basic loading combinations that will test the
foundation performance occur during the installation, operation, and maintenance phases. A
future condition may also result in a unique combination that must be considered. Refer to
Figure 5A and 5B entitled "SOC Detector Floor Loads" for the identification of load cases A
through F, combinations of which constitute the following loading conditions. 4
1122
Om_5m
I. Loads in metric tons per meter.
2. Lood Case:
® Installed stationary Muon Borrel Torald.
f
® Moving toraids and pieces attached to
End T orolds. (Absorber. foward
colorimeter. forward muon chambers
2-5 Cerenkov counter)
CONCENTRA TED ----1--"''"".,

N THIS REGION LOADS ARE
MOVING LOADS ON ROLLERS
CO," '~
IN THIS REGION LOADS
TH~8$~6~iOR
© Installed Muon chambers on the barrel
toroid.
28.03·
Load Case 585
®
r 1 I
D ' '·':rr
~
2
'I
~-----39-------~
~
I -l
@
lo.iG~J·
l 14.02
L -_ _ _ _ _ _ _ _ _ _~~~~----~ I
©
l 149
I
Figure 5A.
2. (CONT.)
@ Moving Colorimeter. tracker.
Superconducttngsolenofd. and port of the absorber.
46 - - -- - --- --l I
CD Support and Jack weight.
3.® Floor load of the entire installed Detector.
Total Welght=32,OOO Metric Tonnes
~7 II
! L-
I.
-1- ----' !
12.27
14.02
Figure 5B.
1123
Operating Condition
The operating condition, represented by load case F, results in loads from the barrel
toroid in place with the calorimeter in operating location. The operation phase constitutes the
steady state condition and will be the governing loading combination for approximately 90%
of the life of the detector.
Installation Condition
The installation condition, represented by the worst combination of load cases A
through E, results from the assembly and construction of the barrel toroid and the calorimeter
upon the support system. It also includes the loadings due to the testing of the support and
alignment system and the initial testing of the calorimeter mobility.
Maintenance Condition
During maintenance, the forward toroid and calorimeter may be rolled out from
within the barrel, the barrel and calorimeter may undergo some repair, and the support and
alignment system may undergo repair, replacement, or realignment of the barrel. The
maintenance condition is represented by either (1) the worst combination of load cases A
through E or (2) the leveling of the detector (load case F) combined with a failure of a
number of jacks. The maximum jack operating load is estimated to be 528 metric tons per
jack. (Jack capacity is approximately 700 metric tons each). A single jack failure would
increase the adjacent jacks to an approximate load of 634 metric tons (20% increase to
surrounding jacks) each for the time duration of the repair. Refer to Figure 2.
Future Condition
The future condition is unknown at the time of the detector design but would be
satisfied by applying a load factor of 105% to any or all of load cases A through D and
combined as above.
FOUNDATION PERFORMANCE LIMITATION AS REQUIREMENTS

A direct part of the user's requirements with respect to the concrete foundation's
ability to uphold the integrity of the detector is the tolerable limits of foundation
deflection/defonnation, outside of which the detector would cease functioning and incur
damage. The loading conditions of the detector upon the concrete foundation will produce
stresses within the concrete floor slab and the slab's support structure. The internal stresses
in the concrete will cause deformation of the concrete geometry. The concrete sub-structure
will deliver its reactive forces to the soil/rock media causing stress-producing deformation in
the soil/rock media. The detector experiences the combination of soil, rock, and concrete
defonnations in two basic types of stress-induced movements. The barrel toroid, along with
the forward toroid, the calorimeter and support system, moves as one rigid body in space
experiencing global translations and global rotations through all six degrees of freedom
relative to any point in space. The second type of movement experienced by the detector is a
relative rotation/translation of one node within the detector's spatial coordinate system with
respect to other internal nodes. This results in flexure, shear, and torsional defonnation of
the barrel toroid, inducing significant internal stresses and strains.
Rigid Body Motion of the Barrel Toroid
The rigid body translations and rotations in space experienced by the barrel toroid
induce negligible internal stresses or strains in the toroid. The tolerance limits of the rigid
body movements pennitted by the concrete foundation range from ± 25 mm which lie within
the support alignment design. The rigid body movements cause little global alignment
difficulties and negligible internal stress-strain factors for the detector.
1124
Flexure, Shear, and Torsion of the Barrel Toroid
Flexure, shear, and torsional defonnation of the proposed barrel toroid induce
significant internal stresses and strains. The transverse plates that are bolted or welded
longitudinally can carry bending, shear and torsional stresses in the plane of the plates
provided flexure of the barrel is small. The deflection limitation and flexure strength of the
barrel is governed by the size and amount of bolts or welds, which are limited by the
magnetic and construction requirements of the barrel toroid.
Refer to Figure 3 where a flexure is shown to be induced in the toroid by slab
deflection of a moving forward toroid load. The limit of toroid deflection is shown to be 2
mm over a 28 meter span. Refer to Figure 4 where a beam flexure is shown to be induced in
the toroid by slab deflection of a moving calorimeter load. The limit of toroid beam
deflection is shown to be 2 mm over the full length of the toroid (28.032 meters).
The flexibility ratio between the toroid and the concrete floor slab is worth
consideration. The toroid will encounter induced stresses according to the inverse proportion
of the flexibility ratio. If the floor slab were to be designed as a flexible system, the toroid
will be forced to carry the majority of the induced load. Conversely, the stiffer the floor
slab, the less the induced load will be in the toroid. Therefore, the flexibility of the toroid is
recommended to be considered in the design of the concrete floor slab foundation system.
CONCLUSION
The SDC detector is very sensitive to foundation floor perfonnance. The relative
stiffness ratio between the detector and the concrete sub-structure dictates that the detector
will experience the majority of its own dead load stresses. A detector structural steel failure
will occur long before the concrete sub-structure experiences any significant internal stresses.
Therefore, the design of the concrete sub-structure must be defonnation-controlled rather
than stress-controlled. The foundation defonnation governs the adequacy of the design and,
as such, controls the design requirements given to the Architect Engineer / Construction
Manager for detailed design of the foundation structure.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the contributions of the following people: R.
Gates, J. Petry and J. Tseng.
REFERENCES
1. J. Bensinger, "SDC Barrel Steel Definition and Specification Proposal," Version 2.6,
January 14, 1992.
2. K.K. Leung and J. L. Western, "An Engineering Design Study of Detector
Deformation Limits in the SDC - Detector," Paper III-E-3, IISSC, New Orleans,
Louisiana, March 4-6, 1992.
3. J. Cherwinka, editor, "SDC Muon Barrel Toroid Conceptual Design Report Using
Long and Short Bolted Blocks," SSCL/SDC Note: SDC-91-105, September 25,
1991.
4. J. Tseng and J. Petry, "SDC Detector Foundation Load Distribution Study,"
SSCL/SDC Report: SDT-00OOI5, January 16, 1992.
1125
ENGINEERING THE sse UNDERGROUND
James E. Monsees and Gordon T. Clark
The PB/MK Team

Design Division
5610 Redbird Center Drive, Suite 400
Building 5
Dallas, Texas 75237
ABSTRACT
Underground construction has begun on the SSe. TIlis paper will discuss that construction
from the underground engineering viewpoint. First, the geology will be reviewed, then the
principle challenges will be discussed. These include the design and construction of underground
structures in weak geologic materials which are overstressed by the construction. The first magnet
delivery shaft which is elliptical in shape, has finish dimensions of 30 ft. by 60 ft. This shaft
extends into the shale and is penetrated off-center by the collider tunnel. TIlis results in a unique
and very large underground opening with one side nearly planar. Tunnels in chalk are expected to
remain generally elastic and need only localized support while tunnels in shale and deep marl are
expected to experience elastic-plastic behavior and require full and continuous support. Finally,
ongoing designs must consider intersections of shafts, tunnels, and adits. This will be an
engineering challenge that will put the designers ingenuity to test.
INTRODUCTION
Two years ago at the IISSC we discussed the application of information gained in shaft and
tunnel construction in Austin to the SSC site in Ellis County. Today we will take a look at the real
thing - actual properties and anticipated behavior for the SSC geology, actual designs for
underground SSC facilities, and actual experience from construction at the SSC site. It is true,
construction is underway for SSC underground facilities!
GEOLOGY
For the actual construction of underground SSC facilities, three main geologies (and the
interfaces between them) are of primary concern: Eagle Ford Shale, Taylor Marl, and Austin
Chalk. Figure 1, the so-called "lamp shade" drawing, shows schematically the alignment of the
collider main ring. Starting at the West Campus (northwest of Waxahachie) note that the collider
ring is in shale, proceeds through a mixed-face zone into chalk, then into marl, back into chalk,
and then returns to the shale. Table 1 contains a summary of the lengths of the collider in those
various geologies.
As an approximate appraisal of the strength of the geologic materials, consider the
following Table 2.

800
700
600
.....
0-
~ 500
::z .00
C>
~
300
200
100
10 15 .0
DISTANCE (!.IlLES)
~ TAYLOR !.IARL ~ AUSTIN CHALK ~ EAGLE FORO SHALE
Figure 1 Geological Cross Section of Collider Main Ring
Table 1
SSC Collider Main Ring
Summary of Geology
Approximate
Geology Length of Tunnel Depth (ft)
Miles Feet Percent
Chalk 30.8 162,400 56.6 30-200
Marl 17.2 90,800 31.6 100-240
Shale 4.1 21,800 7.6 130-220
Mixed Face - Shale to Chalk 1.8 9,700 3.4 60-200
Mixed Face - Chalk to Marl 0.4 2,200 0.8 100-130
Table 2
Unconfined Compressive
Geologic Material Strength Approximate Comparison
Shale 200-300 psi Very stiff soil to very soft rock
Marl 300-450 psi Very stiff soil to soft rock
Chalk 2000-3000 psi Weak concrete
1128
Thus, if we consider an underground structure at a depth of 200 ft. and with a typical stress
concentration factor of two to three, it is seen that the stresses around an underground opening
would be approximately 500 psi. This stress level would indicate overstressing at structures in
shale or marl, but not in the chalk. Thus, it would be expected that the ground around openings in
shale or marl would experience the formation of plastic zones and the production of new fracture
zones as the geology and geometry attempt, with time, to accommodate the induced stresses that
exceed its strength. In contrast, for structures in chalk, the stresses are comfortably below the
strength and the behavior around underground openings is expected to be well within the elastic
zone.
THE CHALLENGE
For the main collider ring of the sse there are several challenges, the most important ones
being at intersections between shafts and tunnels, or shafts and adits, or adits and tunnels. Figures
2, 3, and 4 are representative of such intersections.
Around any opening in the underground, be it vertical (a shaft) or horizontal (an adit or
tunnel), there is a concentration and redistribution of stresses due to the interaction between
opening and geologic host. Such concentrations tend to be greatest near the opening and to
decrease with distance from the opening, usually decreasing to approximate free field stresses
approximately two diameters from the opening. However, when one is faced with intersecting
structures, then these stress concentration fields also intersect, leading to very complicated stress
and structural configurations. In addition, the opening is not wished into place, but excavated in a
series of steps, producing an ever-changing stress field.
When these stresses exceed the elastic strength of the material, as is the case for the sse,
the model must include elastic-plastic or elastic-plastic-fractured behavior. Analyses of such
conditions far exceed the capability of conventional (closed form) methods and, as will be shown,
are actually on the limit of the state of the art of analyses by high speed computer.
The state of the mt analysis techniques include three dimensional finite element analysis
software tools which account for axisymmetric material properties, non-linear material behavior,
time dependency, and material properties coupled to pore water pressure. The PB/MK Team has
been using a high speed mini-computer in conjunction with sophisticated three dimensional
meshing and analysis software to develop new modeling techniques based on applied sequential
finite element analysis.
Figure 2 UtilitylPersonnel Shaft
1129
A typical model being analyzed represents a cube of earth approximately 300 feet on each
side with the shaft liner centrally located. PB/MK has pioneered a technique to automatically step
through the excavation sequence required to construct a shaft and adit. In analyzing the combined
Personnel/Utility shaft configuration, over 20 steps are used to excavate ground and place sections
of unstressed liner as the sequence progresses from the top of the shafts to the far end of the
connecting adit. With over 5,400 elements and 20,000 degrees of freedom this model is one the
largest non-linear sequential analyses being run in the country, and possibly in the world. Just five
years ago, it would have been impossible to analyze such a model on any computer with the
possible exception of the largest Cray super computer offered at the time. Today, Convex
Computer Company, which markets a super computer comparable to Cray, is using one of
PBIMK's models as a bench mark in optimizing their FORTRAN compiler.
Progress on the modeling will be summarized later, but first let us consider some actual
structures to understand the challenges further.
Figure 3 Vent Shaft Figure 4 Magnet Delivery Shaft
MAGNET DELIVERY SHAFT
The magnet delivery shaft at N15 (immediately north of the west campus) will have
finished dimensions of 30 ft. by 60 ft. and will be used to lower magnets from the surface down to
collider level. This shaft is 235 ft. in total depth, starting with 15 ft. in the overburden, then
proceeding 180 ft. in chalk, and bottoming out with 40 ft. in shale.
1130
The shaft sits directly over the collider tunnel so that the magnets can be lowered directly
into that tunnel. However, the combination of operational requirements and geometry required that
the collider not pass exactly through the shaft centerline but rather be offset from that centerline by
10 ft. As shown in Figure 5, the net effect is that the intersection of shaft and tunnel results in an
unsymmetrical opening at the bottom. An opening that approximates a straight wall 70 ft. long
rather than a more stable curved-wall structure.
Figure 5 Magnet Delivery Shaft at Tunnel Elevation
To analyze and design such a structure, engineers have three possible approaches:
empirical methods, closed form solutions, and finite element computer methods. Each approach
has limitations:
Empirical methods rely on previous experience and observations from similar

structures, geometries, and geologies. None existed.
Closed form solutions generally exist only for simpler geometries (usually circular
structures or those that could be approximated as circular ones) and Simpler material
properties (elastic-plastic, but not elastic-plastic-fractured).
Finite element methods require some simplification of material properties but otherwise the
use of large models, complicated geometries, and material behavior all at the limits of
existing technology.
Using applicable elements of all three approaches, it was decided to provide initial support
for the magnet delivery shaft as shown in Figures 4 and 5. Note that in the chalk support consists
of pattern rock bolts and fiber reinforced shotcrete (pneumatically applied concrete); in the shale,
support consists of pattern rock bolts, fiber reinforced shotcrete, and lattice girders. In addition, for
the latter, i.e., the portions of the shaft in shale, the design documents tightly control or limit the
contractor. For example, the contractor cannot expose more than four feet of this shale without
installing initial support elements, plus those elements must be installed within four hours of
exposing new shale faces. These restrictions are supported by observations at the exploratory shaft
where some cracking is observed in shale faces unsupported for 1-112 hours and much more serious
cracking and moving occurs in faces unsupported for 8 hours.
It is believed that the initial support system described above will stabilize the shaft so that
the final lining will not be called upon to carry significant structural loads. However, the shaft will
be instrumented and should continued creep be measured, provisions have been made, as shown in
Figure 6 for a reinforced concrete final lining up to three ft. thick should the need for such a lining
be demonstrated by those instruments or by other observations.
1131
Figure 6 Final Lining of Magnet Delivery Shaft in Shale
TUNNEL IN SHALE
Construction bids for the first tunnel in shale, a 2.7-mile run from N15 to N20 in the
northwest sector of the collider ring were taken in mid-December, 1991. It is a requirement that
this tunnel be excavated by tunnel boring machine (TBM) with the tunnel supported by precast
concrete segments as shown in Figure 7.
Analysis of the probable behavior of this tunnel was simpler than that of the magnet
delivery shaft for the following reasons:
The 14 ft-6 in circular tunnel is a smaller, more stable opening with a desirable circular
shape.
The TBM cuts a more stable (circular) opening than does shaft mining.
The TBM supports the ground while the precast segments are erected within the tail. The
ground is unsupported only to a maximum of one segment width and only for the time
required to drive the machine that far and expand the segments (approximately 30
minutes).
For these reasons, it was believed that structural analyses could be performed either by
elastic-plastic closed form solutions or by elastic-plastic finite element computer solutions. Both
analyses were used and both showed that even at 250 ft. depth the tunnel would be safely supported
by precast concrete segments. Table 3 tabulates some of the parameter studies conducted in these
analyses and the resultant stresses in nine inch thick concrete segments. Note that a working stress
allowable of 0.45 fc was used in the analyses. For the contracts in Shale, a nine inch thick precast
liner segment with concrete strength of 8,000 psi was specified.
1132
Figure 7 Precast Concrete Tunnel Liner Segments used in Shale
Table 3
Load Max Liner Required

Case Stress (psi) fc
Description Axial Bendin~ Combined psi
1 Undrained - 0% elastic relaxation of 2888 438 3326 7400

ground prior to liner placement
2 Drained - 0% elastic relaxation of ground 2924 552 3476 7800

prior to liner placement

4 Drained - 50% elastic relaxation of 2289 517 2806 6300


Weakened zone in crown using FJ4
6 Drained - 50% elastic relaxation of 2400 1538 3070 6900

Weakened zone in crown using FJ4
1133
It should also be noted that these precast concrete segments are both initial support and
final lining, and that similar segments will be used in all shale, marl, and mixed face tunnels. The
joints in the segments are not sealed or gasketed because calculations show that water inflow into
the tunnels will be negligible, except possibly at sheared or broken zones. At these zones some
crown (upper 120· or so) grouting may be performed in segment joints or through the joints so that
water does not drip into the tunnel at the broken zones. The otherwise small water flows, if they
occur at all, will be removed by the usual tunnel drainage system.
TUNNEL IN AUSTIN CHALK
The first tunnel in chalk (N25 to N40) will be 5.4 miles in length and is in the (upper)
north to northeast section of the collider ring. This tunnel also will be driven by tunnel bOring
machine in this weak limestone. However, as shown earlier, the unconfined compressive strength
of this generally massive limestone is 2000 to 3000 psi or two to three times the anticipated stress
concentrations around a circular opening in this medium.
Therefore, when this tunnel is in the massive limestone it is expected to be stable and to
require no continuous support. However, it must be noted that this limestone, while massive, has
shaley lenses in the bedding at approximately four ft. centers. When the TBM is cutting near such
lenses in the crown, the combination of thin limestone, bedding, softer shale lenses, and stress
concentrations may lead to the formation of loose slabs (corbelling) in the arch. When such
behavior is encountered in the field the contractor will be directed to bar down the slabs and install
crown bolts as necessary for safety.
When the tunnel passes into broken zones, additional support may be required. This
support will be decided upon in the field and installed as needed, it being impossible to predict in
advance where such zones may occur. The contract documents, however, are based on the
assumption that such zones may be encountered in 15 percent of the tunnel and that they will be
supported by one of two support types as shown on Figure 8: support type A includes three rock
bolts every five feet and is assumed to be required in 10 percent of the tunnel; support type B has a
four bolt pattern supplemented with three inches of fiber reinforced shotcrete over the top 120
degrees of the crown and is assumed to be required in 5 percent of the tunnel. Adjustments to these
patterns as required by field conditions are encouraged.
A B
Figure 8 Tunnel Reinforcement in Chalk
1134
ONGOING WORK
As mentioned earlier, our modeling of these underground structures for the sse is
"pushing the envelope" for such work and is being followed with interest by others in the structural
analysis field. Figure 9 shows the complexity of the model required for the combination utility and
personnel shaft (referred to as the "hammer head" shaft). Remember that this model includes 5,400
elements, 20,000 degrees of freedom and 20 excavation steps. As point of comparison, just seven
years ago, a model with 1/10 this number of degrees of freedom and only one excavation step
required an overnight run on the largest commercial computer then available in the state of
California.
At the next I1SSC we will report on the detailed results of this ongoing modeling.
Figure 9 Finite Element Mesh for UtilitylPersonnel Shaft Liner
1135
SYSTEMS ENGINEERING AT THE
SUPERCONDUCTING SUPER COLLIDER
John Nonte
Project Management Division

Superconducting Super Collider Laboratory-
Dallas, TX 75237
ABSTRACT
Systems Engineering at the Superconducting Super Collider Laboratory has made

significant advances in the past year. It is heavily influenced by organizational changes,
specifically the changes of the Project Manager. Technical documentation now includes the
project plans necessary to guide the establishment and maintenance of the cost, schedule, and
technical baselines. Specification and interface document preparation are progressing at a
rapid pace to support the pace of construction and procurement. Speciality engineering has
matured as evidenced by the completion of the first safety analysis report for the Accelerator
Surface String Test and by the reallocation of reliability availability requirements to the
individual components per an assumed operating scenario.
INTRODUCTION
Systems Engineering has made significant advances in the past year at the
Superconducting Super Collider Laboratory (SSCL). While these advances took the form of
technical documentation and speciality engineering, they were directly influenced by the
organization, specifically the Project Manager and the Project Management Division. This
paper discusses the impact of the organizational changes in Project Management, the technical
documentation relative to the plans and baselines, and speciality engineering, system safety,
and reliability.
Systems Engineering takes on the character of project management specifically under the
guidance of the Project Manager. Because the attitude and the personality of the Project
·Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

Manager form the mode and the mood in which Systems Engineering takes place, rapid
changes by the Project Manager can sorely test the application of Systems Engineering. Even
under these adverse conditions, Systems Engineering tools of discipline apply and may be
more important because of the stability they provide. The tools are used to document the
requirements, to direct the flow of requirements to lower levels of implementation, to define
the interfaces, to approve and track changes, and to document the reliability and safety
analyses of the system. Since the 1991 IISSC conference, a new Project Manager has been
selected, and the project has taken a course toward "projectizing," with a less formal
atmosphere than existed under the previous Project Manager. There are two additional Deputy
Project Managers: one is responsible for the resistive accelerators, consisting of the Linear
Accelerator (Linac), Low Energy Booster (LEB), and Medium Energy Booster (MEB); the
other is responsible for the superconducting accelerators, consisting of the High Energy
Booster (HEB) and the Collider. The Deputy Project Managers have appointed "machine"
leaders, a group of individuals each of whom is responsible for one of the accelerator
machines. These machine leaders were previously under the Accelerator Design and
Operations Division, which has lost its identity and has been absorbed into Project
Management. The new Project Manager has given more authority and responsibility to the
machine leaders in the form of cost and schedule control in addition to the previously assigned
technical responsibilities. Machine leaders now report to, and have the support of, the Project
Manager when dealing with the technical divisions that supply the components for the
machines. An example of the decreased formality is the Machine Leaders Meeting, formerly
the Task Leaders Meeting, that is conducted as a discussion rather than in the previous
presentation style. The meeting is now held twice rather than once per month. There is more
emphasis on resolving issues than on an informative briefing or a report of past
accomplishments. The emphasis of these meetings is action, from resolution of issues to
follow-up on action items to ensure accomplishment.
Systems Engineering follows the Project Manager's guidance in supporting the machine
leaders in their job of integration. The Systems Engineering group is now under the Project
Manager's staff, directed by the Configuration and Interface Department. In this department,
the focus is on systems engineering activities and configuration management. Systems
Engineering at the SSCL remains decentralized, i.e., most of the Systems Engineering
personnel reside in the technical divisions and directly support those divisions; at the same
time, there is increased emphasis on supporting integration across technical division
boundaries. There is no change in the Systems Engineering discipline; it still provides tools to
get the job done in a disciplined manner with processes to develop the documentation of
requirements, changes, and speciality engineering.
The present Project Management organization has changed the character of Systems
Engineering to a focus of integration and action, employing all elements of the SSCL
organization to get the job done. The tools are still important, but the emphasis has changed.
Possibly, sufficient time has elapsed so that the separate and distinct cultures ofthe SSCL are
beginning to merge into a hybrid set of disciplines and tools utilizing the best of academia and
industry. The organizational influence and cultural distinctions are sharply defined, but the
cultures are blending. The trend toward increased action is meaningful.
The remainder of this paper summarizes the systems engineering activities and
accomplishments, and it hints at the future of the Systems Engineering effort.
DOCUMENTATION
Project documentation takes two forms, Plans and Baselines. Plans consist of the
Project Management Plan, Engineering Management Plan, and Configuration Management
Plan. Baselines consist of cost, schedule, and technical performance.
1138
Plans
The Project Management Plan has existed for a little more than a year, and its
implementation has caused no apparent difficulty. Both the SSCL and the Department of
Energy (DOE) are working on an update. The major changes are organizational changes
within the SSCL and DOE. The SSCL organizational change was discussed earlier, and the
DOE change consists of the consolidation of the DOE office in Germantown with the on-site
DOE office. Other changes are considered minor.
The Engineering Management Plan, formerly called the Systems Engineering
Management Plan (SEMP), discusses the engineering management anticipated at the SSCL.
The name change is an indication of the influence of the organization-i. e., the machine
leaders are responsible for system integration, but the idea is for the entire SSCL to participate
in the integration of a machine. This plan is nearing completion, and signatures are imminent.
The Configuration Management Plan update has been completed with DOE's guidance.
It was resubmitted to DOE in February 1992 for signature.
Baselines
There are three baselines: cost, schedule, and technical. The cost baseline was
established in June 1990, with subsequent changes approved through the Configuration
Control Board. The schedule baseline is emerging as the Integrated Project Schedule (IPS),
which has undergone numerous changes during the past year. The IPS has integrated more
than 15,000 events. The technical baseline was established as the Site-Specific Conceptual
Design Report, but it is now superseded by the technical specifications. The level-l
specification (top level) has been signed by the SSCL and is at DOE for signature. Level 2
specifications are in process. Level-3 machine specifications have all been signed except for
the collider, which is to be final by the end of March 1992, and the Global Accelerator System
(sometimes referred to as the control systems). Level-4 and lower specifications are
developed as needed to support design or procurement of the components.
The update of the specification tree, published in February 1992, depicts the level-l
through level-4 specifications expected to be developed for the technical project. Figure 1, an
example of the specification tree, shows the flow-down of the collider specifications from the
top-level specification, levell, through the level-4 specifications by following the unshaded
portion of the figure. Note that the shaded box "Superconducting Magnets" would be filled in
at level 4, but it does not fit on the page. The machine-level specifications are in the 3A and
3B levels. Machine-level specifications are developed and controlled by the Project
Management Office; the Project Manager controls level 3A, and the machine leaders control
level 3B. The numbers in the boxes represent Work Breakdown Structure (WBS) elements.
With the specification tree aligned with the WBS, the WBS aligned with the cost baseline, the
cost baseline imposed upon the IPS, and the IPS tied to the technical baseline (specification
tree), we have integrated the entire baseline.
Interface control documents are also part of the technical baseline. Work on interfaces
throughout the past year has been done on the Accelerator Surface String Test (ASST) and on
preparing an Interface Control Document Guide. Not all of the required interfaces require
formal interface control documents. Some of the interfaces will be contained in the component
drawings, and others will be included in the specifications. A separate interface control
document will be prepared only when the interface definition is complex enough to warrant
separate treatment. The interface definitions on the ASST are documented, and they set a
precedent for the interface documents to be developed. Development of interface definitions
on the accelerators has just begun, starting with block diagrams of required interfaces in the
level-3 specifications. The level-3 diagrams are followed by the technical divisions in
developing detailed interface block diagrams with division and individual responsibilities.
1139
.j:>
o
f- Z
U 0
f- I SUPERCONDUCTI NG 1
l a«
a:
"-
U
<;:
5UPERCOLllDEf1
>-- 1- e:.; 1
....
"-
If)
'f! 1:
2 ........ ACCELERATOR
I-- ... 1 I~ £X~~MtN~ 11SSC'~~; x
"' ~ .' 'r~ 4
If) If) SYSTEMS ,t ''' 5YSfEl'W> I I' '.$Sf: ,$~S I
,.. >-
If) If) ~
I ::::::!:::: ::::::b
If)
,.. lEST
Z
3~ ,..~ I~~<;~ ~I~ ~~e1 ~. MinHtIoQS'rtR
ENERGY, COLLIDER
+ .
~ «
u
.
svsnM :"' ~~I~t~'l tJ.QWtN~
eo,os. rn I I'~
" aoos't~ JI .12 eeAM$ I
....
on <;: ~ ::::::!::::= ==== I
f-- - 8"- == === 1
If) on
!- ,.. Coliidcr Coilider
Z Z Colllder IR TecMlca l
Arc SectJ ons .12 1 UlIll\y Str
FaC illtl('s ~
..AS$T
36 £~ 321 .121
-' !oJ 324
.... If)
I ~
I Ifupetrondn<:tin,
MligUetS ... H Surface Bldgs
3 1X 3
If) Utilities
1:
Con tro ls Ring Magne t P S 11 1 ..b "
I
1190 1060 11 30 Underground
~
4 If)
>-
I
r- RF
Y FaCllltles
If) Correc t or P S Vacuum 31XI/2
CD Spool I 120
=>
Pieces 1060 11 11 40 ~
I
'" 1100
lQuencn Protection P.5 Abor t Sys tem
r-f-
Correction Magne1 - 1170
1060 1090
Personne l
Scrapersl
ACcess I Ins trumen tation
Safe t y (PASS) Cryogenics Collimators
1180 II I 1 10
1200 1- 1090
Figure 1. Sample Specification Tree.

DESIGN REVIEWS
Perhaps the most evolutionary element of the Systems Engineering process has been the
technical design reviews. The standard nomenclature of the Department of Defense design
reviews as depicted in Mil-Std-1521A is generally used for the SSCL, but it has taken on a
discrete meaning. The level-4 reviews are more traditional: they more closely follow Mil-Std-
1521A, so there is little to elaborate on with respect to these reviews. The level-3 design
reviews are more interesting and deserve further discussion. They consist of the Preliminary
Design Requirements Review (PDRR), Preliminary Design Review (PDR), Critical Design
Review (CDR), and Acceptance Test Plan Review (ATPR).
The PDRR is a review of the level-3 specifications and a discussion of the technical
design concept. The criteria for a successful review is that there is at least one solution to the
technical specification or technical parameters.
The PDR reviews the changes to the level-3 specifications and discusses the technical
design. The criterion for a successful review is an acceptable solution to the technical
specification or technical parameters. The level-3 PDR is conducted when most of the level-4
PDRs are complete, i.e., the level-3 PDR is a summary of the level-4 PDRs.
The CDR reviews the changes to the level-3 specifications and discusses the design,
interfaces, and risks of proceeding with the build of the components. These reviews should
precede the first major level-4 CDR, but this is not always practical. The PDR is conducted
late in the PDR process, and the CDR is conducted early in the CDR process, making the
timing of the two very close. Thus there may be more than the desired level-4 CDRs
completed before the level-3 PDR takes place. The timing of the reviews is a risk assessment
in itself, as determined by the machine leader. The idea is that major, costly components do
not enter production before the total machine risk is assessed. We want to understand the risk
before proceeding with something that might have to be redesigned at major expense.
The ATPR at level 3 is an installation and commissioning plan. In most machines the
installation and commissioning plans are separated because they are at least a year apart.
However, the purpose of the review is to answer two questions: 1.) How do we know that
the specifications are satisfied? and 2.) How do we know when we are finished? In order to
answer these questions, the test approach must be known--e.g., component or integrated
components test, bench or installed test, etc. It is early to speak of commissioning before the
installation, but it must be known how the machine will be commissioned in order to know
what tests must be conducted prior to commissioning. While it is true that commissioning
must be in the minds of the design engineers throughout the design, the outline of the
commissioning plan comes together at the ATPR.
The level-3 review schedules for each of the machines are complete and are being
integrated into the IPS. The Project Manager can manage the project through these level-3
design reviews and can get a good idea of how the project is progressing. The machine
leaders will have to plan a set oflevel-4 design reviews to be able to assess the progress of the
component design and integration.
DESIGN SAFETY
Design safety has been integrated into the SSCL Environmental Safety and Health
organization. The purpose of design safety is to influence the design in order to prevent safety
problems, as opposed to correcting such problems after they occur. Design safety personnel
help prepare the Safety Analysis Reports (SAR) and influence the specifications or
requirements. The SAR contains a definition of a hazard, an analysis, and mitigation action. A
preliminary SAR (PSAR) is prepared for the CDR, and the SAR is completed prior to
commissioning. Commissioning includes commissioning the cryogenics system,
1141
commissioning without beam, and commissioning the system with beam. The first SAR, on
the ASST, is complete and has been forwarded to DOE for approval. This SAR sets the
precedent for SARs to be completed in the future.
RELIABILITY
Reliability allocations have been made to the machines as part of the level-3
specifications. These allocations are very high in order to achieve a machine availability of
0.8, a DOE controlled requirement. Reliability analyses have been conducted to determine
how the initial allocations could be changed to give component designers some relief from the
stringent, unobtainable reliability requirements. The reallocation study is nearly complete; it
shows how the requirements for the machine components can be relaxed, but it also requires a
redefinition of the operating mode of the SSC. The operating mode definition previously
contained a three-month downtime reserved for maintenance, e.g., tune-up, repair, and
replacement of components. If the reliability requirements for the components are relaxed, the
three-month downtime may not be available. This scenario has many implications. For
example, if the three-month downtime is not available, the run time is increased, thereby
increasing the power requirements. Increased power requirements mean increased operating
costs due to use of power during peak power periods as defined by the electric utility.
Increased power usage also means that cooling must be increased. This is only one example
of the reliability impact on the SSCL operations scenario. The example points out the need for
development of and agreement on an operations plan.
Reliability also impacts design. For example, if the reliability allocation for downtime
for the power supplies is one hour, the design must allow for automatically taking a failed
power supply off-line without physical replacement. It remains to be seen how many of these
types of design considerations enter the design process of the SSCL.
After components are produced, a major reliability effort is put forth in the failure
reporting and corrective action system (FRACAS). Craig Arden's paper in the Systems
Engineering section of this book discusses an example of the FRACAS used on the
superconducting magnets. Magnets built by General Dynamics and Westinghouse at the
national laboratories have begun to experience discrepancies. In all cases, corrective action
has enabled the magnets to meet the specification requirements. FRACAS provides a trend
analysis and points to actions required to ensure a good product. Emphasis must be on
follow-up with corrective action. This is a time-consuming effort that yields a major payoff.
SUMMARY
Rather than concluding with a summary, I prefer to point out how the reader can obtain
further information. The technical documentation that forms the baseline resides in the
Document Control Center of the SSCL. Any baseline document, specification, interface
control document, technical drawing, etc., can be located by author, title, or number. All
documents are available upon request. Once on the distribution list for a document, a reader
will automatically receive document updates.
1142
IMPLEMENTING CORRECTIVE ACTIONS FROM THE
FAILURE REPORTING, ANALYSIS AND CORRECTIVE ACTION
SYSTEM (FRACAS) TO ACHIEVE MAGNET RELIABILITY AND
AVAILABILITY GOALS
Craig S. Arden

Dallas, Texas 75237
ABSTRACT
Implementing corrective actions from the Failure Reporting, Analysis and Corrective
Action System (FRACAS) is one of the methods from the reliability integrity approach used
to ensure that magnets can achieve reliability and availability goals.
FRACAS achieves early feedback by monitoring magnets from assembly through
operation, thereby saving time and reducing the number of items that may have a particular
design or manufacturing process defect.
This paper will discuss the FRACAS program as it has developed in the Magnet
Systems Division of the Superconducting Super Collider Laboratory, the type of
infonnation FRACAS has accumulated on the magnets, and corrective actions implemented
to help achieve the reliability and availability goals.
INTRODUCTION
The Collider Dipole Magnet (CDM) reliability goal is to design for failure-free life
during the operation of the collider. This is an enonnous technical challenge, as minimum
reliability data exist for the present design or for the specified operation conditions.
As a result, a conventional reliability approach that uses historical data from
equipment already operating in understood environments would not be applicable. At the
Magnet Division, the reliability program is an Integrity Methodology that emphasizes the
product.
"Operated by the Universities Research Association, Inc., for the U.S. Department of Energy under Contract
No. DE-AC35-89ER40486.

Because the available reliability data for CDM are minimal, a strong effort went into
the development and execution of the Failure Reporting Analysis and Corrective Action
System (FRACAS).
The subject of problems/failures is a sensitive issue. It is not the intent of this paper to
identify the faults but to give credit for the openness of information from each of the
associate labs. In fact, each lab is considered to be a stronger supporter by the number of
FRACAS reports generated.
FRACAS
FRACAS is a closed-loop problem reporting system that ensures that corrective action
is taken. When a FRACAS program is utilized on a product such as the CDM, it can
optimize the program by addressing product problems early in the schedule and providing
refinement and reproducibility of the production process. The results are a cost-effective
program and a better product.
FRACAS programs are usually initiated during the testing and qualification phase of
the product development. However, due to the stringent reliability goals and minimum
magnet testing presently being performed, it was decided to initiate FRACAS during the
manufacture process of the magnet. Magnet Reliability is highly dependent on the
production process. This effort has resulted in tight process controls and product
improvements of which the entire lab is aware and proud.
FRACAS generates a discrepancy report upon discovery of a problem with
procedures, test equipment, manufacture equipment, or the product. The discrepancy report
must then be addressed by the cognizant engineer or physicist for problem classification,
disposition, and possible corrective action.
Problem classifications are defined as Class 1 and Class 2 discrepancies. Normally,
Class 1 problems are product discrepancies that affect either the function, fit, reliability, or
safety. Class 2 problems are simply defined as not Class 1 discrepancies (e.g., process
documentation).
All discrepancy reports are provided to the Magnet Reliability Group for tracking, data
recording, information dispersion, and Class 1 close-out meetings. A Failure Review Board
(FRB) has been developed to address critical and timely problems requiring Magnet
Division action. The FRB also has the responsibility to review each Class 1 discrepancy
report for corrective action prior to its close-out of the report.
A FRACAS Report is considered closed when the discrepant item's disposition is
successfully completed and the corrective action identified is implemented for prevention
on subsequent units. At this time the FRACAS Report is presented to the FRB for close-out
concurrence. However, some items are not so clear-cut. In these cases all the applicable
information is obtained and reviewed and the best disposition and corrective actions, if any,
are determined. Again, the FRACAS Report is presented to the FRB for close-out
concurrence.
Data
As of January 15, 1992, there had been 538 Discrepancy Reports logged. Table 1,
FRACAS Discrepancy Report Summary, shows the breakdown of the Class 1 and Class 2
reports generated. Approximately 15% of the reports are Class 1. Table 1 also lists the
number of Class 1 closed reports and the number of discrepancy reports suggested for close-
out. Suggested reports are those that will be presented to the FRB at the next close-out
forum.
1144
Table 1. FRACAS Discrepancy Report Summary.
Total DRs Suggested for

Received Closed FRB Closure
Class 1 83 37 21
Oass2 ill NfA NLA
Total 538 37 21
The breakdown of the total number of FRACAS Reports received from each of the
associated labs is shown in Table 2, National Lab Discrepancy Report Summary. It is
important not to misinterpret this table. The most important item to remember is that the
number of reports recorded is a positive value, as more impact can be made when more
problems are identified and corrected. Each of the associated labs does not have the same
number of magnets or the same manpower available, which may affect the number of
discrepancy reports identified.
Table 2. National Lab Discrepancy Report Summary.
SSCLDRs FNALDRs BNL DRs LBL DRs

Received Received Received Received
Class 1 12 59 9 3
Class 2 ..1fJ... ill -ll. -.2Q..
Total 88 385 42 23
After this CDM-specific reliability data has been generated and logged, it must be
analyzed for trends. There are as many ways to look at this data as there are data records.
One way to examine FRACAS Records is to group them by problem causes. This tells us
which area is more susceptible to Class I problems and what type of problems may be
expected. Table 3, FRACAS Class 1 Problem Cause Summary, lists "coil shorts" as a major
heading, then lists specific causes of that problem. Prevention of coil shorts is critical to
optimal functioning of the magnets. The remaining major headings are the result of a "best
fit" categorization effort.
LESSONS LEARNED
Coil shorts are a primary concern during magnet manufacturing and testing. The
primary cause of coil shorts was contamination. This information is significant because it
shows that the feasibility of manufacturing is high and that cleanliness controls are
significant during cable and coil processes. One could go a step further and evaluate the
cost-effectiveness of a semi-clean room for cable insulation and coil winding processes. It
is strongly believed that no magnet assembly should be initiated until the complete
manufacturing area and tooling items are assembled and cleaned.
Coil shorts resulted also from overpressure of the coil, which occurred because of the
breakdown of the insulation scheme due to high stresses. One way to address this issue is an
engineering evaluation of the insulation scheme to minimize the stress or to increase the
insulators' capability to handle high stress. A second way to maintain product integrity is
by assembly process verification. This may include testing coils for tum-to-tum shorts
during the curing process prior to unpackaging. It may also include loading the ends and the
coil straight sections prior to the next higher level of assembly.
1145
Table 3. FRACAS Class 1 Problem Cause Summary.
Problem Cause FNAL BNL LBL SSCL Total
Coil Shorts
Contamination 6 1 0 0 7
Over Pressure 3 0 0 0 3
Dimensional 0 0 0
Craftsmanship 2 0 0 3
Strand Fault 0 0 0 1
TBD J... _1_ .JL .JL ....!..
(Sub-Total) (15) (4) (0) (0) (19)
Bus Hi-Pot 7 0 0 8
Coil Bonding 2 0 0 0 2
Heater Short 4 0 0 0 4
Dimensional 17 0 1 2 20
Workmanship 6 3 2 3 14
Certification 0 0 0 0 0
Cleanliness 0 0 0 1
Documentation 0 0 0 1
Configuration 3 0 0 1 4
OtherrrBD J... _1_ .JL ....6.... ...lil
TOTAL 59 9 3 12 83
1146
The next category of problem causes is dimensional. Again, this infonnation is critical
as it shows that strong emphasis on parts control and component inspection is significant in
building the best product possible. This information also shows that manufacture of the
magnets is very feasible using process controls.
Workmanship is the next most probable cause for a Class 1 problem with a magnet.
This is a serious issue, as the producibility of the magnet is a direct function of the
technician's capabilities and experience level and of the true manufacture of the magnet.
While there have been a few isolated problems due to a lack of awareness by personnel, in
the lab environment it is believed that the technician's skill levels are higher than what may
be available during full production. This would then translate the cause from workmanship
to engineering design for producibility. This effort is best performed by the magnet
contractors, who iterate their present design for producibility and use the FRACAS data to
determine which areas need to be analyzed. The need for experience and practice cannot be
overemphasized at this point, especially in the cable insulating and coil winding processes.
Presently there is a large effort to semi-automate the coil winding process, which can in
large measure address this major identified cause of failure.
The Power Bus Hi-Pot problem is a specific failure, not a cause. The cause for the
excessive current leakage during test was a flux containing Zinc-Chloride, which had a
tendency to absorb moisture, causing an ionization reaction and resulting in current
leakage/insulation breakdown. A process and material change to use a resin-based flux and
a 60/40 tin/lead solder was incorporated quickly and effectively to minimize the schedule
and cost impacts. This is a good example of the positive effects of the FRACAS program.
Similar to the Power Bus Hi-Pot problem, the strip heater is also a specific failure and
not a cause. The primary cause for strip heater problems is due to breakdown of the kapton
insulation as a result of overstress or contamination. However, as a result of heater problems
the issue of destructive testing, such as Hi-Potting versus megger testing strip heaters and
coils, has been identified and is being addressed. This issue cannot be decided here, but it
was FRACAS which has elevated the problem to be addressed by Magnet Systems,
Engineering, Production, and Test.
Configuration is primarily aimed at parts, not to the proper or latest revision. This
problem is a normal parts control and configuration function. Under true production and
statistical controls, this type of issue is expected to be more closely monitored and
controlled. However, once a change occurs, it is always possible for the incorrect part to be
used, thus prolonging the problem.
The documentation heading is primarily for in-process control specifications,
travellers, and/or manufacturing operational procedures. The number of these problems is
due mainly to the incorporation of the need for such documentation during each magnet
assembly step. A lesson well learned is to develop and evaluate these procedures prior to
beginning any formal magnet assembly. The documentation control, such as revisions, is a
serious issue that must be addressed to ensure smooth operation.
A few other items that have been identified during magnet assembly include, but are
not limited to, the following: tooling break-in, proper cryostat pipe alignment, and tom
kapton/B-stage during various processes. It has been learned that new tooling, primarily the
coil winding mandrel and associated hardware, need to go through a few curing processes to
get baked-in prior to producing high-quality, damage-free coils. Another problem still
being analyzed is the cryogenic pipe locations. The present tolerance leaves little margin
for error, while the design was well developed, and only reasonable design efforts were used
to attempt to meet the tolerances. This is one producibility issue that may be best addressed
by the COM contractors. During coil winding, kapton problems were fairly common and
showed signs of increasing when new winding crews/personnel performed the task. Again,
producibility is the issue here, and the strong effort for semi-automated coil winders is a
positive step.
1147
The level of effort is much higher than anticipated because of the challenge of
performing FRACAS during magnet assembly. The workload from maintaining and
tracking all the appropriate documentation has a tendency to minimize the amount of
involvement on the Class 1 problems that occurred. Additional manpower and computer
automation are the principal solutions, both of which have been initiated. All FRACAS
participants should be aware of the manpower needs upfront, or the system can become so
unbearable that it may fail to achieve its full potential.
Summary
The lessons learned are based on the experience and work performed to date, and are
limited to magnet assembly. There is minimal data on cryostating, and even less on
interconnect work, short-term testing, and long-term life testing. New issues and new
lessons learned will develop as the program moves further along the schedule. Using the
FRACAS approach, these problems will be identified and corrected as effectively and
efficiently as feasible.
The FRACAS program has also been a key instrument leading to engineering design
changes, magnet assembly process changes, and magnet assembly test changes. Even in the
short period of time that the FRACAS concept has been utilized, the amount of significant
data for our specific and unique use is overwhelmingly high.
The amount of data accumulated in approximately one year could not have been
possible without the support of each of the associated labs. The management support of
such an effort is the fundamental reason for such success.
FUTURE
We plan to continue as we are through the remainder of the associated lab works.
FRACAS is working, and we will continue a good thing. It is presently foreseen that once
again the level of effort will increase if superconducting wire manufactures are brought into
the FRACAS System. As the magnet contractors begin their model magnet programs, they,
too, will use FRACAS.
In order to better handle the increased number of problem reports an effort has been
initiated to develop an automated database between the Magnet Division and the many
contractors. This system is scheduled to be in controlled use within six months and fully
operational in one year. This effort will allow more time to work on the issues and to
accurately analyze trends.
In the future, more single-magnet testing will occur, the SSCL string test will be
operational, and possibly some reliability extended cycle testing will be in progress-and
FRACAS will be there! It is expected that magnet autopsies will be performed on specific
life-tested magnets to determine long-term wearout modes and corrective action, again
using FRACAS.
ACKNOWLEDGEMENTS
Thanks go to James Franciscovich, SSCL and Lockheed Systems Engineering,

Systems Engineering Group Manager, who provided technical input, support, and document
review.
1148
RISK MANAGEMENT AS APPLIED TO
COLLIDER DIPOLE MAGNET DEVELOPMENT
RekhaPatel
General Dynamics
Systems Engineering(MZ: CI-8770)
P.O. Box 85990
INTRODUCTION
Purpose
The overall purpose of the COM risk management activity is to identify risk warning
thresholds in the areas of Engineering, Test, Production, Facilities,Quality and Management,
and to identify and implement risk mitigation plans in order to reduce adverse impacts to the
CDM program.We use DoD 4245.7-M, Transition From Development to Production and
Best Practices as a guideline.
Scope
Scope of our risk activity covers Technical, Cost and Schedule risk at all levels of the
CDM program. Risks are summarized in a hierarchical fashion and summarized and
presented to SSCL at the program level. Risk assessment is presented at each Program
Management Review(PMR) and also displayed in our Management Information
Center(MIC).
RISK MANAGEMENT BOARD
The CDM risk management activity is tied to program requirements and the goal of low
risk as described in Figure 1.0. Shaded area in this figure represents risk management
activities that are completed. Risk identification and resolution activities will continue
throughout all phases of the program.

01
o
COMPLETED
TASK
• DEVELOP RISK
MANAGEMENT
PROCESS COMPLETED
TASK
• ESTABLISH RMBI
DIRECTIVE • REVISEIVALIDATE
PROCESS
• CANDIDATE
ASSESSMENT RISK IDENTIFICATION 1-------..,
- TECHNICAL • PRODUCTION
PROGRAM
• TECHNICAL. COST - COST INDICATOR
RISK - SCHEDULE
AND SCHEDULE
RISK ASSESS • RISK REDUCTION
RISK ASSESSMENT
MONITORING
• TOP RISK ISSUES RISK MITIGATION
• BACKUP SOLUTIONS • PROOF TEST
PLANS
• RISK MITIGATION FOR HIGH RISK MONITORING
SELECT TRACKING
OF HIGH RISK
SDR PDR CDR LRIP GOAL

Figure 1.0 sse Risk Management Goal
Key to implementation of the General Dynamics risk management methodologies are the
Risk Management Board (RMB) and the closed-loop tracking. monitoring. and reduction
procedure as described in Figure 2.0. The SSC RMB provides communication between
management, program personnel, vendors and the customer to exchange ideas and
professional judgements to reduce the program risk. The RMB is chaired by Program
Manager Bob Baldi, and Co-chaired by VP & Plant Manager Dino Salvador. The Risk:
Management committee consists of Design, Test, Manufacturing. Reliability, Quality.
Systems Engineering and the follower. We are responsible for managing and monitoring
subcontractor risk activities that may have technical, cost and schedule impact to the COM
transition from development to production. Our Risk Management activities are coordinated
with our vendors who are committed to provide us timely delivery of the risk assessment
data. The RMB will meet weekly until PDR and then once a month to status previously
identified risks, implement solutions, establish schedule requirements, and identify any new
risks that may arise. Responsible individuals present their risk issues, cause, concern and
mitigation plan(s) to the Board for review and approval prior to implementation. Action
items resulting from the RMB are tracked and monitored until they are answered
satisfactorily. This process helps the program manager identify top risk issues based on risk:
analysis.
RISK ASSESSMENT METHODOLOGY
As part of the Concurrent Engineering philosophy, the five step (Figure 3.0) Risk
Management methodology allows individuals to express their concern (risk) and solution
(mitigation) toward meeting cost, schedule and technical requirements goals.
STEP 1 The first step for risk analysis identifies candidates that are detrimental to the
program using applicable templates (DoD 4245.7-M) and Best Practices approach
as a guidelines. COM Program tailored transition from development to production
templates show in Figure 4.0. We then quantified the risk by setting three levels
of risk magnitude (high, medium, low) for each template listed under
Engineering, Test, Production, Facility, Management, Material, Quality
Assurance and Information Resource Management (IRM). A total of 43 CDM
related templates are identified.
STEP 2 Using Step 1, risk assessment results are displayed in the SSC MIC Room.
High and medium risks that are identified in Step I are reviewed. and updated by
the Risk Management Board. This process will continue throughout the program
until risk is reduced to acceptable levels.
STEP 3 Based on the magnitude of the risk, a brief risk mitigation report is prepared
which covers technical, cost and schedule risk reduction activities and rationale
for the risk as described in Figure 5.0.
STEP 4 For critical parameters. a schedule is established for risk mitigation activities
showing the activities needed and when they need to be completed. This is used
to provide a monthly risk status (Figure 6.0).
STEP 5 As the system design matures, the critical parameters that have been identified
through this process will be monitored using the Technical Performance
Measurement (fPM)f:Rjsk Report
1151
()'1
I\)
Risk Management Board

Chaired by : Bob Baldi - Prog . Mgr.
Co-Chaired by : Dino Salvador - VP &
Plant Mgr. Risk management/resolution
RMB Committee tracking system
• Input & ranking

• Impact assessment
• Status updates
(Your role aSaRMB member)

Institute risk management procedure
• Identify, review, and evaluate risk and
Risk mitigation
prioritizing
action assignment
• Manage and monitor risk reduction program
• Integrate risk mitigation plans into overall
development plans as appropriate
• Assign to actionees
RISK ISSUES Formulate approach
I sfc, • Monitor & display • Input to tracking system
• Technical , costs/schedule impact
~~~~ Status review

& resolution
Figure 2.0 SSC Risk Management Operation· Continuous Risk Reductln Process
Our risk management approach provides continuous improvement of quality and
performance at reduced cost and on schedule. sse risk analysis accounts for the impact of
technical uncertainties on project time and costs. Performance risk analyses result in
contingency plans that become an integral part of the overall program plan. This structured
management approach assures that technical risk levels of individual technical items decrease
as the work: progresses. However, if risk management tracking and monitoring indicates that
technical risk levels have not sufficiently decreased, then appropriate measures are taken to
implement related contingency plans
SUMMARY
The risk management procedure described here represent sound engineering and
management practice. By implementing these practices in the eDM program it has allowed
us to maximize the resources for overall productivity. The risk management program is
supported by all personnel working for the GDSS eDM program, which includes
Engineering, Test, Quality Assurance, Production, Facility, Material and Management. Best
of all, it is supported, encouraged, directed and monitored by program management.
Using this structured process we are able to identify and resolve problems and issues
before they have adverse impact to the program. This procedure serves as a long range
"Attention-Getter" before preventable problems tum into crises. Risk management provides
a broad range of potential risk assessment and allows us to utilize management resources and
expertise for effective and timely solutions. This is a useful management tool to use in a
concurrent engineering environment to reduce the barriers between Design, Test, Production,
etc. It provides a creative climate and improves communication. Individuals receive
feedback and support, not criticism.
During Risk Management Board reviews we have placed emphasis on production of

results. As an example, specific benefits to the CDM program are:
Risk Concern Solution Benefit

Drawing Release • 450 Drawings Between • Revise Drawing • Minimizes Schedule
Schedule SDR & PDR (10 mo) Release Procedure Impact On Long Lead
Items
Supplier • Aggressive Schedule • Form Supplier • Establish Procurement
Management Management Team Demand Schedule
• Defmes The
Requirements
Process Control • Magnets Require A Utilization Of Three Good Definition Of
Very DiSCiplined And Primary Loops Of Process Concrol
Responsive Process Systematic Feedback Requirements For The
Concrol - Concurrent Magnet. Integration Of
Engineering. Statistical FNAL Lessons Learned
Concrol And Employee
Involvement
Test Experience • Shortage Of Early Identification Of • Utilize FNAL/BNL
Experienced Test Test Personnel Test Experience
Personnel At
Hammond
Test ;\teasurement Eleccromagnetic • Hire Experienced • A void Redesign Of
Integrity Interlace (EMI) ConrraclOr To Define Test Equipment,
EM! Requirement Tooling And Facility
1153
U1
~
TRANSITION FROM DEVELOPMENT

TO PRODUCTION RISK ASSESSMENT TO EACH TRANSITION
TEMPLATES TEMPLATES
• IDENTIFY CONCERNS AREAS

ISTEP{) • TECHNICAL COST AND SCHEDULE RISK ASSESSMENT
• IDENTIFY CRrnCAL PARAMETERS
SSC RISK MANAGEMENT
• MEET TWICE A MONTH

RISK QUANTIFICATION • IDENTIFY NEW RISKSJRESPONSIBILE
~DIVIDUAL
I . REVIEWIUPDATEIAPPROVE
HAIRMAN: Bob Baldi-Prog. Mgr. MITIGATION PLAN
CO-CHAIRMAN: D. Salvador-VP & Plant Mgr. • ASSIGN RISK REDUCTION
ACTION ITEM(S)
- DESIGN • MONITORITRACK RISK
- TEST REDUCTION ACTIVITY
- MANUFACTURING • REPORT TO CUSTOMER
- RELIABILITY • DISPLA Y RESULTS IN THE SSC
- SYSTEMS ENGINEERING CONFRM
- SUBCONTRACTOR
LOW - QUALITY ASSURANCE
RISK ~ FOLLOWER ~
X ~~~
, -,--& . -p R
ISTEPV I r::: .. ; .. _. PLANNING I
H
~I~~ALI
Figure 3.0 sse Risk Management System(RMIAS)
sse
TRANSITION FROM DEVELOPMENT TO PRODUCTION
T C S ,_.-
DOD DOD' ,
T C S T C S T C S T C S
T C S
Or::I_=~=AC=_=-O
REU
- . . .---,
..
oOlr- o01..---....-=··,-., o o0·'--:1..=:.n='"""1 oOr-"":I=::=-""11
001---· o01 DEVELOPIIENT DOl woau:.. DOl- 001 =.=.1
001.:=. o01 PROTOTYPE. DOl..:... 001 :::= oOl~rrli
TEO DOl == DOl- DOl.:=.. I
001= oOI--::a
001 c":"" o01 QUAUflCA_ 001 ":="'
o01 ~IIELD oOl.=~.
001 :=.
001~-
001-- T C S , _._n
001 =: DOD' ,
001-:='1 T C S T C S Ir---::::=~
T C S
oOr=lsulCOl.....lTl AC=::::;-"I oOI"=NTI 001..------,-.,1
, -
c- lECHtlCAL fiSK
~_
001 ::: I 001:551 oOl.:=. 1
•- ICIIEIIIU_ 001 c _ 1
L-
M-
LOW_t_NI
__ _
001==1 o01c::~1
11- --tftUOW)
001--1
Figure 4.0 sse Risk Management Templates
U1
U1
0'1
en
RISK ITEM TITLE: Quench Behavior DATE: 8/16/91 H-HIGHRISK

RISK LEVEUMAGNITUD~ M - MEDIUM RISK
PREPARED BY: Scott Peck TECHNICAL [EJ COST M SCHEDULE @ L-LOWRISK
(IF ANY): SPECIRC RISK REDUCTION SUGGESTION:

TECHNICAL RISK RATIONALE AND PROPOSED SOLUTION
~ Collider performance is limited by the quench behavior 01 the dipole magnets. Ahhough TRADE/ANALYSIS:
understood in principle, quenches are caused by the microscopic mechanical behavior ollhe coil, which is
dinicuh if not impossible to accurately analyze. Proposed design changes will have unknown enects on TEST:
quench behavior unlil coils can be tesled. OTHER: Critical peer review of design
Proposed Solulion· Pay close atlention to conductor quality, tOOling, winding, collaring, and yoking RESOLUTION DATE:
processes, tolerances on winding components, and lifecycle thermo-mechanical history of winding to
assure sullicient coil pre-load for a lightly packed coil, particularly at the ends. ESOLUTION BY:
CRITICAL TRACKING PARAMETER: Margin, number of training quenches
(IF ANY): ' SPECIFIC RISK REDUCTION SUGGESTION:
COST RISK RATIONALE AND PROPOSED SOLUTION
TRADE/ANALYSIS:
BaIimlaIe.; Reasons for quenches are often subtle and cannot be reliably predicted by analysis. Therefore
testing of actual coils is the only way to know that a design has good quench behavior. II problems do TEST:
show up, the number 01 test coils, design and tooling changes required lor a fix could lead to significant OTHER:
added cost to the program.
Proposed Solution- Verily design changes with short test mag nels, plan testing and institute changes RESOLUTION DATE:
such that cause and effect are well understood, design in adaptability in parts and tooling. ESOLUTION BY:
CRITICAL TRACKING PARAMETER: Number of test coils and changes required to achieve desired quench behavior
SCHEDULE RISK RATIONALE AND PROPOSED SOLUTION (IF ANY): SPECIFIC RISK REDUCTION SUGGESTION:
BaIimlaIe.; The current schedule is success oriented w~h respect to quench behavior. II short magnet and TRADE/ANALYSIS:
prototype testing show that there is a quencMraining problem, redesign and additional testing will have to
be added 10 the program. In addition, required design changes could conceivably require changes to TEST:
tooling or long lead items such as conductor. OTHER:
prooosed Solulion: Pertorm short magnet tesls as early as possible, minimize variables in test magnets, RESOLUTION DATE:
establish fallback positions with contingency plans in place.
ESOLUTION BY:
CRITICAL TRACKING PARAMETER: _.
Time required for redesign, addilionallesls to verily improved !l~rforman.c:lI. __J
Figure 5.0 sse Program Risk Report· Development Phase

H· HIGH RISK
M • MEDIUM RISK
L· LOW RISK
RISK MITIGATION ACTIVITY
P~A LA~
PROGRAM MILESTONE PROTO STAAT V LAIPSTAAT V
RISK PROJECTION HIGH x:X:X ~X ~X:X X:X:X. ~X:X:X:X:X

, ........ ~X :X:X:X X ···· ......: .
(CH£CK APPAOPAIA TE BLOCK I : : :
MED .. xlx XX:X X:X~ : : ~ ': I
LOW :x:x:x:x
: .. :
xx:x:x
.: x:xlx:xlx:x:x:
.... : x
( us, IoITIG.lnON ACnVlnU.
SHOW SCHEDULE WllH BAA ANDJOA ..L£STONEI
TECHNICAL
TRACK KEY DESIGN PARAMETERS
SHORT MAGNET TEST PROGRAM
ESTABLISH FALLBACK POSITION WITH
CONTINGENCY PLAN
~
SAME AS ABOVE
SCHEPULE
SAME AS ABOVE
Figure 6.0 sse Program Risk Reduction Schedule
U1
......
ACKNOWLEDGEMENT
Research Association in support of the Superconduction Super Collider project for the
U.S. Department of Energy.
1158
SUMMARY OF PRODUCTION DEVELOPMENT, DETAILS,
AND RESULTS FOR MODEL DIPOLES AT SSCL
Stephen J. Dwyer, Eric P. Vrsansky, and Ronald G. Wood

Dallas, TX 75237
INTRODUCTION
DSA-101 and DSA-102 are the first two model dipole cold masses to be assembled at
the Superconducting Super Collider Laboratory (SSCL) in Texas. They were "built-to-
print" per the first Fermilab-designed 50-mm short dipole magnet. Because the magnets
were designed in English units, all results are presented in English units. Both magnets were
completed and tested in the fall of 1991.
The SSCL developed this model dipole magnet program in order to initiate magnet
production capability in Texas. The approach used was to expand on the continuing
progress of the national laboratories-Lawrence Berkeley Laboratory (LBL), Brookhaven
National Laboratory (BNL), and Fermi National Accelerator Laboratory (FNAL)-and to
develop internal tooling and processes to build model dipole magnets. By producing two
50-mm Collider Dipole Magnets, the SSCL was able to verify technology and provide an
opportunity to develop in-house technical experience.
Both magnets performed well. DSA-101 exhibited very little quench training; DSA-
102 had several non-plateau quenches. All quenches were well above the nominal SSC
operating current of 6500 A.I
Construction details and the results of the first two model dipoles will be presented.
Procedural considerations and assembly details, along with a discussion of lessons learned
during the start-up phase of magnet construction, will also be presented.
DESCRIPTION OF MAGNETS AND INSTRUMENTATION
The two DSA magnets were built per the first Fermilab 50-mm model magnet design.
They were identical to FNAL's DSA-321 and DSA-323. Features of the magnets included:
No. DE-AC35-89ER40486.

Vertically split yoke
External ramp splice
Stainless steel end can and collet
1.3:1 Cu/Sc for inner, 1.8:1 Cu/Sc for outer.
Magnet instrumentation included one strain gauge collar pack, bullet-type end gauges,
and voltage taps. Spot heaters and quench protection heaters were not installed.
COIL WINDING
The coil winding for DSA was performed on the Model Coil Winder and Tensioner,
which were designed by SSCL based on concepts from FNAL and LBL. The mandrel was
procured by the SSCL using the FNAL 50-mm dipole coil configurations. The mandrel was
designed as solid tooling (vs. laminations or EDM blocks), and peripheral tooling was
fabricated for mounting to the winder. Other design-specific tooling (key inserts, winding
keys, and mold retainers) was received from the tooling designed and procured by FNAL.
All coil parts were duplications of the direct material used by FNAL. The cable and
wedges were insulated by FNAL and shipped to the SSCL. The DSA end parts (G-1O end
keys, end spacers, and saddles) were procured by SSCL, and the additional pieces were
shipped to FNAL in support of their model program. Modifications were made to the
tooling, coil parts, and processes during the practice coils, which were used as
developmental tools.
The solid mandrel was fabricated with a slot on the top surface to provide clearance to
a nut bar that was used to mount the winding keys. Inserts were made to fill the slot at the
ends of the mandrel so that the end turns would be made over a continuous surface. Another
tooling issue that arose involved the hold-down clamps at the coil ends. These clamps are
designed to extend over the ends of the coils to provide alignment and rigidity of the end
parts (see Figure I).
Key Insert
Mandrel
Figure 1. Inner Coil End.
The clamps provided an interference for the insertion of the end parts, which meant
that there was also minimal clearance for the cable. No changes were made for DSA, with
the intention that the problem will be resolved in later model magnet programs.
Slight modifications to the coil parts were made prior to the initial process
development to facilitate the winding. The end pieces were designed to the dimension of a
cured coil, which created an interference fit on the wound coil. Relief slots were cut into the
pieces to ensure conformance of the ends prior to curing. Wedges were received in 12-ft
sections, which provided an opportunity for development of proper wedge length. Based on
the gaps between the wedges and end pieces on the practice coils, the wedges were cut to
0.06 in. over print length.
The process variable of primary concern during the initial development was the cable
tension. Experiments were performed to find the proper average between the cable not
1160
conforming (tension too low) and the cable collapsing (tension too high). The result was a
tension of 70 Ib for both inner and outer coils. Another large factor in the development of
coil winding dealt with the experience of the technicians. A definite lesson learned is that
the practice coils were invaluable for building technical expertise.
COIL CURING
The coil curing was performed in the Model Curing Press, which was procured as a
turn-key press from an outside supplier. The press is similar in concept to the model curing
presses at FNAL and LBL. The curing cavity and pusher bars were procured by the SSCL
using the FNAL 50-mm dipole coil configurations. The curing cavity was also designed as
solid tooling (vs. laminations), and peripheral tooling was fabricated to distribute the
heating oil. The modifications made during curing of the practice coils primarily pertained
to the processes rather than the tooling.
The initial development centered on the end loading, which was applied mechanically
by torquing a screw jack. Several trials were performed in order to examine the effect of
increasing end loads on the coil end compression. The final torque used was 110 ft Ib,
which translated to an estimated load of 3800-4950 lb. Even though coil lengths within 0.06
in. were achieved, the hydraulic end loading appears to have the most consistent results.
The next process variable that was examined was the temperature profile during the
cure cycle. The specification for the epoxy system in the fiberglass prepreg presents a
certain required profile to maximize curing results. The design of the curing cavity and
mandrel incorporated thermocouples for routine monitoring of the cycle, but the proximity
of the thermocouples suggested that there would be a temperature lag and delta for the coil.
To verify the process, additional thermocouples were placed directly on the practice coils,
and the cycle was derived from the monitored temperatures (see Figure 2).
275 RAMP GEL RAM~ - - _C.!LRE - - T 270°F RAMP

I ------ - 250 OF
250 I
I I
225 SPEC: I SPEC:
SPEC: 194°F I SPEC:
5°F/minMAX
90 min +15
-10
2.4° 97 min 4.5°F/min

F/min I
I
25 I
I
0+-~~~~~~-+~~-+~~~4-~~~~1L~~4-~~+-+4-+4
o 2 3 4 5 6
HRS
Figure 2. Temperature Profile.
The majority of the development effort during the practice curing focused on an
adhesion problem that surfaced at the start ofthe program. Basically, there was a breakdown
of mold release at the key inserts and mold retainers due to the high coil loads. This allowed
1161
the epoxy to bond to the tooling and tear during the coil disassembly. Experimentation with
mold releases was performed, and a Frekote Sealer and 44-NC mold release was found to be
the best combination. The mold release solved the problem involving the key inserts, but the
adhesion to the mold retainers still existed. It was concluded that there was still a
breakdown of mold release due to the friction of coil sizing, concentrated at the large wedge
(see Figure 3).
Insulation
Adhesion
Figure 3. Inner Coil Configuration.
It was determined that the majority of the coil movement for sizing occurred during
the initial loading, so the curing cycle was modified to minimize the friction without
sacrificing proper cable positioning. Table 1 describes the final curing cycle.
Table 1. Final Curing Cycle.
Step Description Comments
Load azimuthally to 620 lb/in. Closed press and provided majority of coil
movement.
2 Load radially to 65 lb/in. Positioned cable.
3 Load azimuthally to 2800 lb/in., radial load Further positioning of cable and coil.
increased to approx. 300 lb/in.
4 Release azimuthal and then radial load to line Allowed clearance for end pre-loading.
pressure.
5 Load radially to 35 lb/in. Re-positioned cable.
6 Load azimuthally to 2300 lb/in., radial load Pre-loaded coil in preparation for 1st heating
increased to approximately 150 lb/in. ramp.
7 At gel temperature, release azimuthal load to Allowed clearance for final end loading.
500 lb/in. and then radial load to 50 lb/in.
8 Load azimuthally to 13,670 lb/in., radial load Final positioning and sizing of coil at a coil
maintained below 500 lb/in. stress of approximately 13,400 psi.
With the lessons that were learned and the modifications that were made with the
practice coils, the DSA coils were produced without any major complications.
COIL AZIMUTHAL SIZE
Coil size was measured in the usual "deviation-from-standard" manner. The average
coil sizes for the coils used in the DSA magnets are shown in Table 2. Ideally, the magnet
design calls for no pole shims to compensate for variations in coil azimuthal size. The
Fermilab magnets, DSA-321 and DSA-323, were assembled without pole shims. Generally,
the SSCL coils were smaller than the Fermilab coils, requiring that pole shims be used.
DSA-101 and DSA-I02 were assembled with pole shims in order to obtain coil prestresses
within the target range.
1162
COIL ASSEMBLY AND COLLARING
During the first attempt at collaring (practice collaring), no pole shim was used on the
outer coils. However, outer coil prestresses were too high. The only alternative available
was to remove one 5-mil layer of kapton ground plane insulation from the outer pole
surface. (There are four such layers on the outer pole and three on the inner pole.) This
resulted in a negative value for the outer pole shims. Inner coil G 10 pole shims of 8 mil
worked well for both magnets. As shown in Table 2, the average coil stress was on the high
end of the target range. Average peak coil stress during collaring is also shown in Table 2.
Springback losses were high, which is consistent with the square key method of collaring
that was used on these magnets.
There were no electrical shorts detected in any of the coils at any time during the
assembly of both magnets. There was a problem with excessive current leakage during
routine hypot testing of DSA-102 at 5000 V. The path ofthe current leakage was found to
be across the surface of the G 10 pole piece on the lead end of the magnet. The G 10 material
had become contaminated with a glycerol-based liquid soldering flux that was used for
voltage tap installation. The pole pieces were insulated with kapton tape and re-assembled
into the magnet. There were no further problems with excessive current leakage.
Table 2. Comparison of collaring data for FNAL and SSCL DSA magnets.
Azimuthal Size Shim Average Coil Peak Coil

Average (mil) Size Stress (psi) Stress (psi)
SSCL Magnets
DSA-lOl
Inner Coils -3.4 +8 12,570 18,319
Outer Coils +3.2 -7 9,859 17,553
DSA-102
Inner Coils -3.7 +8 13,447* 20,423
Outer Coils +2.1 -2 9,198* 16,521
SSCL Target Coil Prestresses: Inner: 9,000-13,000 psi Outer: 6,000-10,000 psi
FNAL Magnets 2,3

DSA-321
Inner Coils +9.8 0 8,975 14,442
Outer Coils +6.5 0 13,217 22,165
DSA-323
Inner Coils +5.7 0 6,900 11,507
Outer Coils +6.5 0 11,470 20,275
END COLLET ASSEMBLY
The stainless steel end cans were installed with Fermilab-designed tooling that
hydraulically presses the end cans over the end collets. The design did not allow for added
insulation between the coil ends and the collet pieces. Kapton could have been added to
increase prestress in the coil ends; however, hydraulic pressures used to seat the end cans
indicated that there was sufficient prestress in the coil ends when compared to previous
magnets built at Fermilab (see Table 3). The hydraulic pressure used to seat the end cans,
along with end can deflection data, are the only indicators available to determine whether
1163
sufficient prestress exists in the coil ends. FNAL-built DSA-321 was used as a benchmark
for this data, since it also used a stainless steel end can design. (Later designs used an
aluminum end can.)
Table 3. Hydraulic Pressure Required to Seat End Cans (psi).
Magnet Lead End RetumEnd
DSA-321 7000 5000

DSA-lOl 8500 9000
DSA-102 7500 6500
YOKE ASSEMBLY
The yoke assembly operation was performed on the shell welding press insertion table,
which was procured by the SSCL using FNAL 50-mm dipole shell welding press drawings.
A 10-ft section ofthe press was constructed in the SSCL's Magnet Evaluation Lab for DSA
model dipole fabrication. The FNAL design-specific tooling, consisting of two types of
laminated molds, was modified to account for the difference in length between the 15-m-
long collider dipoles and the 1.8-m-Iong DSA model dipoles.
Along with the tooling, the cold mass components were obtained. The collared coil
assemblies for both DSA magnets were manufactured at the SSCL as described previously.
The yoke packs were fabricated at FNAL because SSCL did not possess yoke pack
assembly tooling at the time of the magnet builds. The half shells and alignment bars were
procured by the SSCL materials group.
Miscellaneous fixtures were designed by the SSCL to aid in yoke assembly. These
fixtures included a yoke pack lifting device used to load yoke packs weighing as much as
350 lbs onto the assembly/insertion table, and a shell rounding fixture that clamped the
upper shell around the yoke pack assembly before the cold mass assembly was installed in
the shell welding press.
Several mechanical measurements were taken during the yoke assembly of the DSA
magnets. The collared coil and yoke pack assembly lengths were compared, and the yoke
pack length was modified by removing yoke laminations to allow enough clearance for the
strain gauge wires of the collared coil to exit the cold mass assembly. Shell-to-alignment bar
gap measurements were recorded, and the gaps were set at the average gap measurement to
keep the cold mass from twisting prior to shell welding.
Once the cold mass assembly was completed, a series of electrical tests were
performed to verify that the magnet was still functioning properly. These electrical tests
included coil resistance, inductance, and "Q" measurements, a coil impulse test, and coil-to-
ground hypot tests.
SHELL WELDING
Shell welding of the cold mass was performed after the cold mass assembly was
hydraulically compressed in the shell welding press. The press hydraulic system, consisting
of 6O-ton hydraulic cylinders spaced at 12-in. centers on both sides of the press, lowered the
upper mold tooling onto the upper shell of the cold mass assembly until a pressure of
5800 psi was exerted on the cold mass. Gas tungsten arc welding (GTAW) was used to
manually apply tack welds between the upper half shell and alignment bar simultaneously
down both sides of the cold mass. The lower half shell was tack welded to the alignment bar
1164
in the same manner. One root pass and one filler pass were applied to each of the four gaps
between the half shells and the alignment bars.
Two practice cold mass assemblies were welded prior to the DSA magnets so that
weld samples could be analyzed for welder qualification and process certification. As a
result of successful weld samples, the DSA magnets simulated welding consistent with
ASME code, Section VIII requirements. This effort was made to prepare for welding to
code requirements on future magnet builds.
During the practice welding, it was found that the shell ends were drawn inward by the
combination of the thinner end yoke packs not supporting the shell against the press force at
the ends of the cold mass, and of the welding itself. This shrinking of the shell openings
prevented the end plates from being instaIled in the cold mass during practice final
assembly. To resolve this problem, expanding mandrel fixtures were designed and
fabricated at the SSCL. These mandrels were installed in the ends of the sheIl to support the
shell ends against the inward force of the press during shell welding of the DSA magnets.
Throughout the sheIl welding of the two DSA magnets, the collar pack strain gauges
were monitored to collect coil stress data using the SSCL strain gauge data acquisition
system. A comparison of the post-collared and post-welded average coil stress of the DSA
magnets is shown in Table 4 (inner coil) and Table 5 (outer coil). For both magnets, the
average inner coil stress increased after sheIl welding, and the average outer coil stress
decreased. These results are consistent with the coil stress data for the FNAL 50-mm model
dipole magnets.
Table 4. Comparison of DSA average inner coil stress (psi).
DSA-lOl DSA-102
Post-collared Post-welded Post-collared Post-welded
12,544 13,327 13,447 15,023
Table 5. Comparison of DSA average outer coil stress (psi).
DSA-lOl DSA-102
Post-collared Post-welded Post -collared Post-welded
9,843 9,625 9,198 8,542
FINAL ASSEMBLY
Final assembly of the DSA magnets included cutting the cold mass shell to length;
installing the pressure plate, temperature compensation gauges, and bullet gauges on the
return end of the magnet; and installing and welding the end plates.
Shell cutting was performed using a split frame circular cutter. A fixture was designed
at the SSCL to set the cut line of the shell at the return end. This dimension was critical in
spacing the gap between the inner surface of the return end plate and the outer surface of the
bullet gauge pressure plate. The DSA-I02 bullet gauge wires and temperature compensation
gauges were damaged because the shell was not cut properly at the return end, resulting in
too small a gap between the end plate and pressure plate. The end plate was removed, the
bullet and temperature compensation gauges were replaced, and spacers were installed and
welded between the shell and return end plate to resolve the gap problem. Bullet gauge wire
routing was also modified to keep the wires away from the tightest gap between the two
plates.
1165
After both end plates were installed, they were manually welded to the cold mass shell
using GT AW. Electrical tests of the bullet and temperature compensation gauges were
included in the process control specification before and after end plate welding to verify that
the gauges functioned properly. The tests were incorporated before end plate welding
because the temperature compensation gauges on DSA-IOI were damaged during end plate
assembly. (This discovery was made when they were tested after welding.) The decision
was made to not replace these gauges for cold testing of the magnet.
TEST RESULTS
Quench histories of both magnets are shown in Figure 4.1 DSA-I01 exhibited very
little training, while DSA-102 had several "non-plateau" quenches. All quenches for both
magnets were significantly above the sse operating current of 6500 A.
DSA101 QUENCH HISTORY
.~n...Turn18""'''''
• Lo.rn-Tum19narHp11c:oo ....
• ~InrwrTl• .,'41ud ...
\
First CooIdown
; ....
!
..' j..........,. j... . .!
RlmpRaIfIStu:les 1.
.... i
I.
i 3.8 K
I UK UK
6500 -------i---------- _______ ~ _____~~~~~~:_:'___ 1 __________________________ _

4.3K! 4.3K! 4.3K
Quench Number
DSA102 QUENCH HISTORY
• nTuml!hpIQ'"
• l~ _ _ Tun", I';
Second Cooldown
• UI-T.... 1. . . . . .
• UI-nDnSpb 7...,,. '1
First Cooldown
:
,,,
:
.
UK 1 ~
RlmpR.t.SIudiM j : •• :
.... •••• -i-·
1
.. . ...• \. • . ... ·1· ·1:,
!
-i
! lie !
\ UK ' .• K :; i '.OK
------------------1---~-c-~~~-:"::~--t-------- ---------r-----i--:------------
QUENCH NUMBER
Figure 4. DSA Quench Histories.
REFERENCES
1. M. J. Lamm, "Test Results of 50mm Aperture Model Dipoles DSAI0l and DSAI02," presented at
MSlM, Lawrence Berkeley Laboratory, January 7, 1992.
2. J. Strait, "DSA-321 Collaring and Yoking: Strain Gauge Data," TS-SSC-91-026, February 5, 1991.
3. S. Delchamps, "DSA-323 Collar Keying Data," TS-SSC-91-036, February 22, 1991.
1166
STATUS OF THE MAGNETIC DESIGN OF THE SSC DIPOLE MAGNETS
H.E Gurol, I Lf V. Nguyen, I D.C. Bliss, I J.~ Elliot, I

D.M. Madura, G.V. Mehle l and M. Krefta
IGeneral Dynamics, Space Systems Division
Energy Programs, P.O. Box 85990, MfZ CI-9540
2Westinghouse Electric Corporation
1310 Beulah Road, Pittsburgh, PA 15235
INTRODUCTION
The purpose of this paper is to discuss the status of the magnetic design of the SSC
collider dipole magnets (CDM's). The majority of the work performed to date is the
optimization of the coil cross-section, with most of the emphasis on the b2 , b4 , and a l
multipoles. The basic approach to cross-section optimization involves the following steps.
First, a detailed 180 degree finite-element model is constructed. This model also includes
material magnetic property data for the Nitronic-40 collars, Fe yoke, steel cryostat, and beam
tube. The resulting transfer functions for all the normal multipole components ~ - blO as
well as the skew quadrupole a l (due to the asymmetric cryostat) are calculateo for the
nominal conductor cross-section. Persistent current effects are incorporated into the model
using recent 5Omm, 1.8m long coil data. The initial cross-section is then modified to include
off-sets in the ~ and b4.. multipoles due to the permeable collar, cryostat, beam tube,
persistent currents, and deflections due to cool-down and Lorentz loads. The resulting cross-
section represents our best estimate of what the dimensions of the warm, collared and yoked
coil are. The off-sets due to the various deflections are a result of a fmite-element stress
analysis; any uncertainties in these calculations will impact the deflection off-sets put into the
magnetic cross-section. However, the calculated off-sets are still an indication of the trend of
the various effects, and are expected to provide greater understanding of the measured
systematic multipoles.
In the following sections we discuss: the detailed numerical model, the resulting cross-
section, tunability of the selected cross-section, 1Bdl allotments, material properties, beam
tube effects, and the selected a l correction approach. It should be emphasized that the work-
to-date is mostly on systematic effects. Manufacturing tolerance build-up effects and impact
on the RMS multipoles will be the focus of future work.
CALCULATION APPROACH
A number of tools are being used in this study. The code C0P7 1 is being used for initial
calculation of multipoles, optimization, and tunability studies. COP7 is an analytical, infinite
iron code that is used in conjunction with an optimizer. Off-sets due to material permeability
as well as persistent currents, and conductor deflections (due to cooldown and Lorentz loads,
for example) can be incorporated into the COP7 optimization. COP7 calculates, for a given
set of multipole off-sets, and iron/conductor dimensions, the wedge sizes and pole angles that
will meet the multipole specifications.

The detailed two-dimensional calculations of the cross-section are performed using
the code PE2D2. Three-dimensional calculations are performed using the code TOSCA2.
The PE2D model showing the detailed cross-section is shown in Figure 1. The
dimensions for the collar, yoke, cryostat and beam tube are consistent with the current
General Dynamics design drawings. A significant level of detail was included in the
model in order to have as much confidence as possible in the multipole calculations. This
is particularly true for the al correction approach where the details of the tooling slots in
the outer yoke have an effect on the calculated al. The model is a 180 degree model in
order to account for the asymmetry of the cryostat. The mesh used is adjusted so that
there is top-bottom (mesh) symmetry, resulting in less source of numerical error in the
calculation of aI, a3, etc. The far boundary is about a factor of ten removed from the
cryostat. Approximately 23,000 elements are used. The mesh density is adjusted for
maximum accuracy. The multipoles an and bn are calculated at a radius of 2 cm for better
numerical accuracy (except when the permeable beam tube region is active). The results
are then scaled back to the 1 cm reference radius. The normal and skew multipoles bn and
an are defined respectively as:
21t
1t 0 r 0
f
bn='!"'(B 1 n) de cos(ne) B(r,e)
21t
an =.:! (B 1 n)
1t 0 r 0
f de sin(ne) B(r,e)
13i
138
Figure 1. A Detailed Finite-Element Model of the Cross-Section Has Been Developed.
The numerical accuracy of all the finite-element calculations were compared with an
independent code WEMAP by Westinghouse. WEMAP was run with up to 6th order
finite-elements; PE2D uses 2nd order elements. The elements of order 4, 5 or 6 were
found not to significantly improve the numerical accuracy. Second and 3rd order elements
were found to be sufficient for calculating magnetic fields with a fine mesh. Comparison
of PE2D with WEMAP gave good agreement for all the systematic multipoles ~ - blO, as
well as the skew quadrupole a l'
1168
SELECTED CROSS-SECfION
In this section, the selected cross-section will be discussed. We start with a cross-
section, very similar to the BNL DSX-201, which includes only the effect of the Fe yoke
in the conductor placement optimization. The detailed finite-element model (FEM) is then
run with the permeability effects of the Nitronic-40 collars, cryostat, and 304 LN beam
tube included. The data used for the material properties is collected in Reference 3.
Persistent currents were included by using data from DSA207 4, which is the first 1.8 m
long, 50 mm bore test magnet. The data used is shown in Figure 2. The persistent
current contributions to b2 and b4 were obtained by subtracting the geometric parts of b2
and b4 from this data and attaching it to the geometric portion of the calculated transfer
functions. The 1.3: 1 Cu to s.c. ratio was included in the model by increasing b2 and b4
by -10%. The effects of cooldown and Lorentz load deflections are then included in the
model by using the calculated displacements from the finite-element stress analysis as off-
sets in the magnetics optimization.
40
30
20
j
.e 10
:
0
1000 2000 3000 4000 5000 6000 7000 8000
-10
-20
1(_ _ _1
Figure 2. Persistent Current Contribution is Obtained by Subtracting the Geometric Part

of b2 and b4 From Above Data.
The resulting transfer functions for ~ and b4, including persistent currents, all permeable
materials and mechanical cooldown and Lorentz load deflections are shown in Figures 3a
and 3b.
The changes in b2 and b4 due to permeable materials is summarized in Table 1. The
cryostat produces about +0.2 units of b 2 at full field, and shifts the b 2 peak to higher
fields, since a higher field is required to affect the cryostat. The Nitronic-40 collars
produce a shift in b 2 by about -0.5 units for all fields (the magnetic behavior of Nitronic-
40 is very linear). The value of b4 is not affected significantly by the presence of the
asymmetric cryostat; however, the collars result in a shift of about 0.06 units at all fields.
The effect of the beam tube is found to be negligible. In addition, the value of the packing
factor for the yoke was varied between 97.5% and 99%. A negligible effect on the
multipoles was found. The specification values of systematic b2 and b4 are 0.8 and 0.08
units at high field, and 2 and 0.08 units at low field, respectively.
In addition to the off-sets due to the magnetic effects of permeable materials, we
have accounted for the effects of deflections in the cross-section on the multipoles. The
deflections were calculated using a Nastran FEM5. The stress model included collaring
(for a given pre-load), yoking, cooldown, and the Lorentz loads. Up to this point, we
have only included the cooldown and Lorentz load deflections in the magnetics analysis.
The mechanical deflections due to collaring and yoking may have too much uncertainty at
this time. The effect of these deflections on the systematic multipoles can be included in
the cross-section optimization during the model magnet program. To investigate the effects
of deflections on the cross-section, a "block" model of the conductors was used. The
changes in the multipoles were then calculated by inputting each of the deflections on the
1169
, .00
·· .. ·· .
'---~-r-~-----'--~--'--~-"'--~-'r_~-----,--~-.,--~---,
-----~ -----
·· .. ··· .... ~~7~
:...--'r:
- .... ~ - . - - - . - - - - ! - - - - -
____ . ! o ____ ____ .1 ____ _
••
,------
..... ; ..... 1 ...
I :
0.50 t-:--i:-'--::i.::::::::;:::;I:ii;::=:;=O;e==j='-:"--j-:------t-:1
: :
················t·····~I"'··-:·~··········'··········:··· ...... , .......... ,., .. .
0.00 ..-.-..-.1-.-.y-.---f:-
+_._-..-.T- ..-. .-.T-;.-..-.-1.r---.-
..--:.•..'F- ..-.. -+;--+.-.-..-.T-..-..-."';.r---.-..-..-+-.-.-..+.-.-.. -."7-.-
.. -..+-..-..-.+".-.-..--;.
1- 0 50
. ~
..... ..... I-.. ~ ..... ..... j.......... j.. --. ----- - - - - - - -- - - ~ - - - - - - ---------- ------:-----
/
~
2 ., .00
, • I ,
+---:----!I--__:-_+-~-+---:--+_--:---I--__:-_+----:--+_-__:______I
..... ..[ [
..... .......... [ ..... , .......... , .......... , .......... ,
-, .50 +--+-+-I------1--+--+--+---+--+--+--I------1--+--+-+--+-------;
..... [..1-- ..... ;.......... ;..................... .
~ ..........
.~ .......... ".... .
+. ..
. ,
... "; •........ [.......... [.......... i····· ..... j......... -i-........ ·iJ··· ....
-2.00 +_-.;.;-t-----.;:--+-~:-+-";'.--+-~'--+--..;.-_+-~-+_-~____I
-2.50 .l....--------t-L-------"--~--'---~-.l....--~-'-----'-----"---t_I---'------'
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00

BO (Tesla)
Figure 3a. Resulting b2 as a Function of Field Includes Effects of Permeable Collar, Cryostat, and
Beam Tube, and Deflections Due to Lorentz Loads and Cooldown.
0.200 ,-----,---~---,-~-__r-~--r--~-,..--~-,r_~---.-~----,
.
_____ 1 _____
.· ..
_____ 1 _ _ _ _ _ _ _ _ _ _ 1 ____ _ .
_____ J _ _ _ _ _ _ _ _ _ _ J _____
.
_____ .'. ___________ , _____ _
··· ... ,
•
,
,
•
,
I
., ,
•
··· ..,
•
,
•
,
,
.
,
,
,
I
..
I
I
•
,
,
•
,
,
0.'00
!.a. ------:------------:------
··· ..
...
~
,
,
,
....
,
0.000
~K---. . "!.--:-c..:-::.-::-:;._----+--_ ....... ......:........... ':' :...- _.

..... ,..... ..... ;..... . .... j..... ..... ;..... t....
··· ,
-0.' 00 .l....-_ _ _ .L..._~__'__~ _ __'__~

··· _
,
,
.
,
,
__'__ _ _.L..._~_____'_~_ __'__~----'
o 2 3 4 5 6 7 8
BO (Te.I_)
Figure 3b. Resulting b4 Includes Same Off-Sets as b2 in Figure 3a.
corners of the conductor blocks. The resulting effects of the cooldown and Lorentz load
deflections and the total change in b n are summarized in Table. 2. The cooldown and
Lorentz loads are significant, and may produce changes that are opposite in sign for certain
multipoles. The resulting total change in the multipciles is shown on the last column of
Table 2. The resulting total L1b n in Table 2 are then combined with the L1bn values that
result from permeable materials (Table 1). The cross-section is then re-optimized using
COP7 to obtain the conductor locations, wedge angles, and pole angles that will offset the
multipoles by these amounts. The new optimized cross-section is then put into the finite-
element PE2D model and run to verify the values of the off-set multipoles. The intent is to
put in enough of an off-set to shift the bn transfer function curves by an amount that will
satisfy both the low field and the high field specifications. The dimensions of the resulting
cross-section are shown in Figure 4. The minimum wedge angles are 0.9 and 1.1 degrees
for the inner and outer coils. The inter-coil spacing is taken to be made up of 3 layers of
127 11m Kapton sheets, resulting in copper-to-copper distance of 617 11m.
1170
TABLE 1. Multipole Changes Due to Permeable Materials. Multipoles are
Expressed in Units.
Cryostat Collar Beam Tube

(Full-field) (All Fields)
~Bo (Gauss) 100 43 1.4

~b2 0.2 -0.5 -0
~b4 -0 0.06 -0
TABLE 2. Multipole Changes Due to Coil Deformation. Multipoles are

Expressed in Units.
Cooldown Lorentz Loads Sum
~BO (Gauss) 117.20 -12.20 105

~b2 0.28 -0.58 -0.3
~b4 -0.0342 -0.0164 -0.05
~b6 -0.0015 -0.0011 -0.0026
~b8 0.0012 -0.0035 -0.0023
~blO 0.0005 0.0001 0.0006
46.93·
FACE AN GLES
19.23'
Figure 4. The Resulting Cross-Section Includes Off-sets Due to Permeable

Materials, Cooldown, and Lorentz Loads. Dimensions are in em.
One final note is worth making about the effect of the deflections shown in Table 2.
These values are for the coil pack as a whole. The actual data put into the PE2D block
model was for individual blocks. We have also investigated the multipole contribution of
individual blocks. The contribution of individual conductor blocks to a given multipole
are not only different in magnitude, but also in sign. This means for example that one
block may contribute positively to b2, but another block contributes negatively. The sum
of all the deflections (due to all the blocks) produces values of b n that are the result of
differences of the contributions from the individual blocks. One finds that individual
block deflections may contribute significantly more to multipo1es than the coil pack as a
whole when all the deflections are included. This may be a clue to understanding
1171
achievable RMS values of the multipoles. Because manufacturing errors will affect
deflections on the coil pack as a whole, the resulting effect on b n may not be as great as
one might predict by assuming that error is taken up in a localized section of the coil pack.
Even if the error is localized (for example a wedge having a slightly different dimension),
the coil pack as a whole will re-adjust itself, in effect reducing the final impact on the
multipole values.
TUNABILITY
The "tunability" of the cross-section refers to the range of multipole values that can
be accommodated by making changes in the cross-section during the pre-prototype and
prototype phases of the CDM program. The allowable changes must be those that can be
accommodated with no impact on schedule. Furthermore, the farther along we are in the
program, the smaller the allowable changes can be.
The goal of the tunability study was to determine the range of 1>2 and b4 off-sets that
can be used and still create cross-sections that meet the magnetics requirements. The
details are given in Reference 6. A harmonic off-set of some value means that the
optimization program (COP7 was used) generates a cross-section that produces a
harmonic of that value. For example, if -1 units of b2 was found to be generated when a
magnet is tested (due to an effect that was not foreseen in the design), then an off-set of + 1
units would be chosen to counteract it. A matrix of 81 optimizations with 1>2 ranging from
-4 to +4 units, and b4 ranging from -0.4 to +0.4 units was performed. The optimization
was limited to varying wedge shapes and pole angles for both the inner and outer coils.
The minimum wedge size constraints were: 0.6 minimum for the inner coils, and 0.4 for
0 0
the outer coils.. These were the smallest wedge dimensions that GDSS manufacturing
engineers felt could be used in the cross-section (similar to a dull knife edge).
Looking at the matrix of b2 and b4 values, it was found that the entire region (-4 !:>
b2 !:> + 4, and -0.4 !:> b4!:> + 0.4) was tunable, except for a small region where both 1>2
and b4 were large and negative (for example b2 = -3 units, and b4 = 0.4 units). The inner
pole angle variation was 0.56 degrees; the outer pole angle variation was 1.74 degrees. In
summary, this appears to be a tunable cross-section; it should be easy to tune as much a 3
units of b2 and 0.3 units of b4.
EFFECTS OF PERMEABLE COLLAR
The effects of the permeable collar on the field quality were determined7. The
permeability of the Nitronic-40 for a wide range of field values is relatively constant at
about Il = 1.00215 3. Non-linear calculations were performed for currents between 325
and 7,800 amps. The changes in the multipoles due to the Nitronic-40 collar were
evaluated. The collars produce significant changes in the multipoles: 1>2 changes by -0.52
units at 6.5 kA (spec. is 0.8 units), b4 changes by + 0.052 units (spec. is 0.08 units), and
b6 changes by -0.005 units (spec. is 0.013 units). These off-sets should be included in
the magnetic design of the coil. These changes are virtually constant across all current
levels. This is due to the almost linear B-H characteristics of Nitronic-40. The effect of
changes in the permeability were also analyzed. The changes in the multipoles are found
to vary almost linearly with changes in the collar permeability. For example, if the
permeability were to increase from 1.00215 to 1.00430 (~Il is doubled), the value of ~1>2
would go from -0.52 units to -1.04 units. This relationship between ~bn and ~Il allows
the evaluation of the allowable values of permeability variability of the collars.
IRON YOKE PROPERTIES
Analysis was performed to evaluate the effect of the variation in the iron yoke
properties on the field qUality8. A similar analysis was performed by Kahn and Morgan of
BNL9. The allowable tolerances in the yoke properties were set based on the tolerance
requirement of the central field. This requirement comes from the allowed variation in
fBdl, which is discussed in one of the following sections. The required Fe yoke
properties were based on 50% of the allowable RMS in ~Bdl being assigned to the yoke.
At low field, the variation in the central field is dominated by the variation in the
coercivity He. The variation in He was simulated by subtracting ~Hc from the standard
BH curve. The standard BH data used is summarized in Reference 3. The maximum
1172
magnetic intensity H seen by the yoke is about 10 Oe at 650A. The variation in the field
harmonics were calculated for llHe of 1 and 2 Oe. A factor aHe was calculated to show
the sensitivity of the central field in variations in He. aRc is defined as
where Bo is the central field. The results show that llHe has a maximum effect on the
central field at 650A. Field harmonics higher than the dipole field are found to be
negligible (this was found to be true at all field levels). By using one half of the fBdl
allowed variation for the Fe yoke, results in llBolBo = 0.0005. This value of llBoIBo
when used with the PE2D calculated value of aHe = 1.98 x 10-4 , gives an allowable
variation in He of 2.5 Oe. This value of llHe is quite large, and in fact accelerator steel
data (either used in or proposed to be used in a number of accelerators) indicates that llHe
of less than 0.25 Oe should be attainable.
The desired value of He can be determined by the correlation between He and the
permeability 1.1. Data shows that permeability 11 increases with decreasing coercively.
However, there is little gain in 11 for He values below about 1.8 Oe. Therefore, the
selected value for He is He ~ 1.8 Oe.
At high fields, the variation in the central field is mainly due to the variation in the
saturation magnetization Ms of the yoke. To evaluate the effect of the variation in M s,
different BH curves were obtained corresponding to changes in Ms using the following
approach. BH data can be characterized at intermediate and high field levels by the
Frohlich-Kennelly relationship:
where a is a function of H, and is basically an empirically determined parameter. This

relationship was used to construct new BH tables for a given variability in the saturation
magnetization llM s. Using this approach, the value of llMsfMs was determined for an
allowable variation llBo in the central field, which was calculated with the finite-element
model. The allowable llBo at high field is about ±35 Gauss. The allowable RMS value of
llMsfMs is found to be about ± 0.4%.
The approach used for intermediate field levels is similar to that used for high
fields. At intermediate field levels, the variation in the central field is sensitive not only to
the variation in M s, but also to the variation of a(H) (In the Frohlich-Kennelly relation).
The allowable variation in a(H) was determined by assuming that the variations in Ms and
a(H) are independent of each other and can be added in quadrature as:
where
llBoIBo llBdBo
<Xa == !la ,and <XMs == llMsfMs
The factor <Xa has its peak at 5,200A. Using values corresponding to 5,200 A results in an
allowable RMS variation in a(H) ofless than 6 x 10 -5.
BEAM TUBE EFFECTS
The multipoles (that result from the beam tube) were evaluated to determine the
effects of beam tube material, thickness, and y-axis off-set lO . Nitronic-40 and 304 LN
stainless steels were evaluated for the seamless beam tubes. The increase in central field
due to the presence of the beam tube was found to be low for both materials: 0.5 Gauss
for Nitronic-40 and 1.4 Gauss for 304 LN beam tubes (@ 6.6T). The beam tube
thickness was taken to be 2 mm, with an outer diameter of 36.3 mm. Y-axis off-sets of 0,
0.5 mm, and 1.0 mm were analyzed. The change in the induced multipoles is very low
1173
for both materials. For example, the induced a 1 due to a y-axis shift of the beam tube of
1 mm is about 0.0015 units for 304 LN, and 0.005 units for Nitronic-40. Either material
can therefore be used for the beam tube. The baseline choice is 304 LN.
al CORRECTION PLAN
The magnetization of the asymmetrically placed cryostat at high field (greater
then -4.5 T) produces a significant skew quadrupole al. At full central field of 6.67 T,
the value of al is about 0.1 units; the specification value of al is 0.04 units. A number of
methods were considered for correcting a Ill. These included putting extra iron outside
the bottom of the He vessel, reducing the size of the former "heater holes" in the lower
yoke half, and increasing the outer radius of the yokes by about 1 cm. The selected
method for correcting al is to place square bars made of high permeability steel in the
lower tooling slots, as shown in Figure 5. The permeability of the correction pieces was
the same as that of the Fe yoke. Filling the slots completely over-corrected at. which is
desirable since the correction approach needs to have some margin for tuning during the
model magnet and prototype magnet phases. Circular tubes were also analyzed; they did
not provide sufficient correction, indicating a fair amount of sensitivity to the geometry of
the correction region. The size of the square bar that is sufficient for correcting al was
found to be 6 x 6 mm2. The resulting al transfer function has a peak value of -0.03 units
at 6 T, and -0 units at 6.67 T. If needed, a total correction capability of +0.08 units is
available with this approach. The intent is that the partially filled lower tooling slots will
become a permanent feature during the production phase.
-2.0
.... 0
-6.0
-8.0
-10.0
-1 2.0
-1 4.0
Figure 5. Square Bars in Tooling Slots are Used to Correct al.
1174
fBdl VARIABILITY ALLOTMENTS
The requirement on fBdl that the CDM magnets are being designed to is:
fBdl = 100.1 ± 0.1 T - m.

The nominal value of 100.1 T-m determines the required central field. When coupled with
the Fe yoke material properties, the nominal coil geometry is determined along with the
transfer function and field margin. However, the allowed variability of fBdl of ± 0.1 T-m
is satisfied by considering the variability of those items that affect either the central field or
the length of the coil. The items that affect the central field include: iron yoke, steel
collars, beam tube, and cryostat. The variability in the length of the coil is determined
mainly by the curing process.
The total central field contributions of the cryostat, collar and beam tube are small:
100,43, and 1.4 Gauss respectively at full current. Therefore, their effect on the central
field variability will be negligible. The dominant ilB term comes from the Fe yoke. We
have taken the Fe yoke variability to use up 1/2 of the total ± 0.1 T-m. This gives an
allowable LlB due to the Fe yoke of ± 34 Gauss at full current. It is this requirement that
determines the specifications of the Fe yoke material. The Nitronic-40 variability in the
susceptibility was calculated to be ± 6.3% based on requiring the collar contribution to LlB
to be 5%. These specifications in the collar and yoke materials appear to be achievable.
The greatest variability in total coil length is expected to be due to the curing process
and results from variability in conductor properties. The winding and curing processes
can be controlled quite accurately and will be consistent from coil to coil in a production
environment. The variability in cured coil lengths was measured at FNAL for 5 inner and
5 outer coilsl2. The coils were made from one vendor's cable. The results show a
maximum difference between measured coil lengths of 2.8 mm, or a variability of ± 1.4
mm. Since the sample size is small, for purposes of allotting an allowance on ill, this
number was doubled to ill = ± 3 mm. This choice of Al will probably have contingency
in case other processes such as collaring, and inserting the end pieces do have some effect.
This Al produces a value of AfBdl = ± 0.02 T -m.
The major contributions to the variability of fBdl have been determined to be the
field due to the Fe yoke and the length of cured coils. Since these contributions are RMS
values, they must be added in quadrature. Hence, there will be a significant amount of
contingency. In addition, BNL and FNAL results to date on long 40 mm magnets is
encouraging. NMR measurements taken at 1900 A on the group of magnets DD0028 -
DC0206 (measured at BNL), and DD0017 - DD0028 (measured at FNAL)13, show that
the standard deviation of fBdl is 60% of the allowed value.
SUMMARY
The selected magnetic cross-section for the collider dipole magnets includes
systematic off-sets in the multipoles due to collars, cryostat, beam tube, persistent
currents, as well as deflections due to cooldown, and Lorentz loads. The cross-section is
tunable across a wide range of values of b2 and b4. Preliminary ilfBdl allotments have
been set, with a considerable amount of contingency. These allotments result in
specifications for the yokes and collars, which can be relaxed somewhat if they become
difficult to meet. The beam tube is seamless 304 LN; its effect on the central field,
multipoles, and placement accuracy in the vertical direction are negligible. The selected
approach for correcting the skew quadrupole harmonic al is to fill the lower tooling slots
with 6 x 6 mm2 iron pieces. The tooling slots provide more than enough al correction
capability. Confirmatory analyses by Westinghouse have been performed using an
independent code WEMAP. The calculations of all multipoles agree very well with each
other. Future work on the CDM magnetic design will concentrate on manufacturing
tolerance effects on the field quality, and the three-dimensional analysis of the coil ends.
REFERENCES
1. D. Orrell, "Coil Optimization Program COPT', SSCL Instruction Manual, December 1990.
2. PE2D and TOSCA, 2 and 3-Dimensional Magnetics Software by Vector Fields.
3. "Material Data Used in CDM Magnetics Analysis", GDSS Report MAG-92-005, February 1992.
1175
4. Brookhaven National Laboratory Report TMG-607, December 1990; Test of DSA107-First SOmm,
1.8m R& D Magnet.
5. SSC CDM Cold Mass Mechanical Design Mini Review, General Dynamics Memo # SSC8443S
A-023, January 1992.
6. D. C. Bliss, "Optimization Sensitivity to b2 and b4 Offset Variation", GDSS Report MAG-91-030,
December 1991.
7. L. V. Nguyen, "Effects of Permeable Collar on the Field Quality of the SSC Dipole Magnets",
General Dynamics Space Systems Report MAG-91-031, December 1991.
8. L. V. Nguyen and H. F. Gurol, "Effects of Variations on the Iron Yoke Properties on the Field
Quality of the SSC Dipole Magnets", GDSS Report MAG-91-01S, December 1991.
9. S. A. Kahn and G. H. Morgan, "Magnetic Properties of Iron Yoke Laminations for SSC Dipole
Magnets", 1991 Particle Accelerator Conference, San Francisco, CA., May 1991.
10. J. L. Elliott, "Effects of the Beam Tube on the Magnetic Field", GDSS Report MAG-91-029,
December 1991.
11. H. F. Gurol and J. L. Elliott, "a1 Correction Plan", GDSS Report MAG-91-028, December 1991.
12. J. Brandt, Fermi National Accelerator Laboratory, private communication, December 1991.
13. J. DiMarco, P. Radusewicz, "Summary ofNMR/Hal1 Axial Measurements on 4 cm Aperture,
17m-long Dipole Magnets, October 1991.
ACKNOWLEDGEMENT
The work described herein is being accomplished under contract to the

Universities Research Association in support of the Superconducting Super Collider
project for the U.S. Department of Energy.
1176
ALTERNATE MANUFACTURING PROCESSES
AND MATERIALS FOR THE SSC DIPOLE
MAGNET COIL END PARTS
Arie Lipski, Rodger Bossert, Jeffrey Brandt,

Jay Hoffinan, Gregory Kobliska,
and John Zweibohmer

Box 500
Batavia,IL 60510
William Higinbotham, Robert Shields,

and Richard Sims

Dallas, Texas 75237
ABSTRACT
Modern magnet designs such as the SSC dipole utilize smaller bore
diameter and wider superconducting cable. Challenging winding techniques
place greater emphasis on the role of the coil end parts. Their complex
configuration as derived from the function of confining the conductors to a
consistent given shape and location.
Present end parts, made of G-I0 composite, are manufactured utilizing
complex and expensive 5-axis machining techniques. Several alternate
manufacturing processes and materials described in this paper will result in a
substantial cost reduction for mass producing the end parts.
The alternate processes are divided into two major groups. The composite
group consists of Resin Transfer Molding (RTM), Compound Transfer Mold
(CTM), Injection Molded Composite (IMC) and Compression Molded Composite
(CMC). The base metal coated group consists of Chemical Vapor Deposition
(CVD) dip coating and hard coatings/anodizing. The paper will provide an
overview of the various processes and compare test performance and cost to
that of the process currently used.
* Operated by the Universities Research Association, Inc., for the U.S. Department of Energy.

INTRODUCTION
The magnetic field pattern in a superconducting coil is governed by the
arrangement of the current conductors, thus making the precise coil geometry
and location of utmost importance. This precision must be maintained in spite
of the large Lorentz forces acting on the current conductors.
In modern magnet designs such as the SSC dipole, which utilizes small
bore diameter and wider superconductors, cables are subjected to high internal
stresses while being wound around the end part. End parts are designed to
confine the conductors to a predetermined position in the cold mass coil
assembly (Figure 1). The unique shape of the end parts was developed to
create lower stress paths for the cables, making them easier to wind. This
shape must be well defined to both the part manufacturers and those analyzing
the magnetic field.
End parts used in SSC prototype dipole magnets are made of G-10 CR
composite and manufactured using complex and expensive 5-axis machining
techniques.
This paper will describe and summarize a two year program at Fermilab to
develop processes and materials to replace those currently used.
The objective of the program was to achieve a substantial cost reduction
(50% or better) without compromising quality or performance.
The development program was divided into two main groups:
• Molding of organic materials

a. Resin transfer molding (RTM)
b. Compound transfer molding (CTM)
c. Injection molding composite (lMC)
d. Compression molded composite (CMC)
• Coating of cast metals

a. Hard coating/anodizing
b. Dip coating
c. Chemical vapor deposition (CVD)
While satisfying the mechanical requirements is an important criteria,
withstanding the operating and environmental conditions became the limiting
factor in the material selection process.
Compressive and flexural strength are crucial for sustaining the loads
exerted on the end parts during the curing and collaring processes as well as
the Lorentz forces during operation. Flexural modulus, however, should be
kept low to allow some flexibility during coil winding.
While strength and modulus may not be the governing factors in selecting
the proper coating and metal base material, good electrical insulator properties
and the mismatch between the base material and the coating are the limiting
criteria.
It is desirable that the coefficient of thermal expansion of the chosen
material for end parts will be close to that of the coils so that the assembly as a
whole will respond to changes in temperature in a uniform manner.
Since data on radiation-resistant polymers is limited, and performing
radiation testing is expensive, it was 4ecided to initially target materials with
published radiation resistance data. While it is understood that the base metal
used for the coated end parts should be non-magnetic, the coating should resist
abrasion and wear.
1178
The qualified material for the end parts will have to meet the following
physical properties:
• Compressive strength - 172 MPa
• Flexural strength - 206 MPa
• Coefficient of thermal expansion (CTE) 10.0 to 20.0 x 10-6 mlmIK

approximately .004 mlm between 300 and 4K.
• Sustain an electrical potential as high as 4000 volts.
The above physical properties are expected to be maintained through a
temperature range of 4 to 523K and radiation level of 1.0 x 10 9 rads (10
MGray)
The short duration, high temperature requirement is mainly to
accommodate the curing of the high temperature Kapton insulation being
promoted by Brookhaven National Laboratory. Fermilab coils are cured at
135°C. Since it is not clear that high temperature curing is essential, the high
end temperature requirement could possibly be relaxed to the FNAL level.
END KEY
14 M 681.20 MM
578.000
LEAD END
SLOTTED SPACER 3
W/SHELF
I,~ "--_ _ 14 M 864.08 MM INNER COIL - INSIDE LENGTH

585.200 (MEASURED AT OUTER RAO I US OF ,,-"""'"'-----,
SLOTTED SPACER 2
W/SHELF
LEAD END
SLOTTED SPACER 1
~_ _ _ _ _ _ _ _ 15 M 57.63 MM INNER COIL - OUTSIDE LENGTH
592.820 (MEASURED AT OUTER RADIUS OF---=SA"'"'D"'D"'"'LE:-,S"")-----
~------------- 15 M 125.70 MM O.A. LENGTH----------------I

595.500 IN
Figure 1. sse 50mm cold mass inner coil assembly
1179
RESIN TRANSFER MOLDING (RTM) AND COMPOUND TRANSFER
MOLDING (CTM)
Resin Transfer Molding (RTM) is a closed mold process wherein dry

reinforcements within the mold are injected with a filled catalyzed low
viscosity resin. Vacuum is applied to minimize the amount of trapped air and
cure occurs at room temperature. The dry reinforcement (preform) is composed
of long (continuous) fibers held together in some textile fashion such as
weaving, braiding or plies of glass stitched and glued. The fiber orientation is
dictated by the design requirements of the part and can vary from one end of
the part to the other.
Compound Transfer Molding (CTM) is a modified form of conventional
filled molding compound injection. Using the same mold as for the RTM, a
compound containing chopped glass (1.5 mm long E-glass fibers) is injected into
the cavity. Thin cross sectioned areas where the molding compound would not
provide satisfactory strength are reinforced using continuous fiber plies.
Though structurally superior, parts made by RTM process are more labor
intensive and thus more costly than those made by CTM process.
A two part development program intended to explore the feasibility of
applying RTM and CTM processes to produce end parts was completed by
Spaulding Composites Company in August 1991. Two parts, the inner coil
assembly return end key and saddle (Figure 2 and 3), were produced in both
processes using three different resin systems.
After successfully using Dow Tactix 123 epoxy resin with Tactix curing
agents in the first part of the program, resins which satisfy most of the
performance criteria were used in the second part. Those resins were:
• CTD-101 - Anhydride cured epoxy DGEBA based (400 cp; 60 hours pot life
at 40°C processing temperature)
• CTD-102 - Anhydride cured epoxy NOVOLAC based (450 cp: 50 hours pot
life at 40°C processing temperature)
These two resin systems developed by Composite Technology Development

(CTD) have been tested for performance in cryogenic and radioactive
environments with good results.!
The reinforcement used for the RTM parts was E-glass identical in content
and weave to the glass used in G-10 CR.a) A webbed binder was applied
between the plies during the preforming process to give the preform sufficient
stability for handling and cutting.
Figure 2. sse 50mm dipole cold mass Figure 3. sse 50mm dipole cold mass
inner coil return and saddle inner coil return end key
a) 28-T glass from BGF Industries
1180
Thin cross-sections such as the saddle "legs" which taper to approximately
.13 rom thickness require additional reinforcement. Since this gap is too small
for the CTM molding compound to reach, a strip of E-glass cloth was laid in
along the saddle's "base" to provide the reinforcement. This prepositioning of
reinforcement process can be applied to both the RTM or the CTM methods.
(See figure 4)
There is a difference in the method of filling the mold cavity between
Tactix and the CTD resins for both the RTM and the CTM processes. While
pressure was applied when using the Tactix resin, vacuum only was used to
draw the CTD resins into the mold cavity. The extended fill time helped any
residual air trapped in the resin to escape through the vacuum thus, improving
the surface quality of the parts.
INJECTION MOLDING COMPOSITES (IMC)

Three different thermoplastics which satisfy most of the material
performance criteria were selected.
• Amoco TorIon 5030 - 30% glass fiber poly(amide-imide)
• Green Tweed - ArIonb ) 1160 - 30% glass fiber - polyetheretherketone

(PEEK)
• Green Tweed - Polyetherketone (PEK) - 30% glass fiber
Both PEEK and PEK have the same chemical structure with the difference
being in the ratio between ether and ketone. PEK has better resistance to high
temperatures due to its higher glass transition temperature. As evidenced by
Figure 5, some of the end part configurations can get rather challenging for
molding, in particular for high temperature high pressure polymers like the
PEEK and TorIon. At this stage of the development it was decided to test
machined parts out of injection molded tubes and avoid the high expense of
molds. Five different inner coil return end parts of each material were
machined.
COMPRESSION MOLDED COMPOSITES (CMC)
As opposed to RTM which is a closed mold process, in compression molding

the mold is open when the material is introduced. The material is shaped by
the closing pressure and by heat.
Three sets of compression molds for inner coil return end saddle, key and
spacer (Figures 2, 3 and 5) have been produced to study the process using
various materials. Two different composite compounds were molded:
• Ciba-Geigy - "Green putty"c)
• Emerson Coming - "Blue" stycasted d )
Both materials are highly filled epoxies that have been used in accelerators but
are too brittle to be used as end parts. These materials were selected
b) Trade name with similar properties to those of lei Victrex GL30

c) AV 1580 resin amine + HV 1580 hardener
d) 2850 FT filled resin and hardener
1181
Figure 4. sse 50mm dipole cold mass Figure 5. Prepositioned reinforcement in
inner coil return end spacer #3 RTM and CTM saddle (shaded areas)
for mold tryout only without regard for the end parts performance criteria.
General Dynamics has proposed using Spaulrad, a Spaulding Composites
compound which satisfies the performance criteria as well as the processing
requirements. This chopped epoxy/glass fiber compound will be preformed in a
fabricated premold. Final forming will utilize the existing compression molds
under approximately 275 MPa psi and 422 K. This forming process will
provide a random fiber orientation with fiber lengths of approximately 9.5 mm.
COATINGS OF CAST METALS
Preliminary work was done to test the feasibility of coating aluminum or

stainless steel casted end parts with an electrically insulating coating. In
addition to being a good insulator (500 to 2000 volts depending on the part),
the coating must adhere to the metal surface through high levels of radiation
and cryogenic temperatures. The first coating attempt was using "hardcoat"
anodizing process of aluminum. Though this process did not provide the
required electrical insulation, it was found that applying various sealers to the
"hard coat" can improve the dielectric rating. Various sealers such as
polyimide varnish, polyphenylene sulfide, and an epoxy ester were tested on
non-hard coated machin-ed aluminum lead end keys. In addition, an inquiry
was underway to examine the feasibility of using Chemical Vapor Deposition
(CVD) coating process. This process would apply uniform aluminum oxide
(A.e203) coating onto a stainless steel substrate.
The surface finish of the casted parts can be of major importance to the
quality in some of these coating processes.
TESTING AND RESULTS
Testing objectives were twofold: a) to test material properties of the

various processes; and b) to test the performance of the parts produced in the
different processes.
Molded polymer composites were tested for flexural strength and
coefficient of thermal expansion (CTE) while coated metal parts were tested for
electrical insulation breakdown.
Flexural test values were compared between machined G-IOCR, RTM,
CTM and injection molded processes at room temperature. The CTE test,
however, was measured only for.parts produced in RTM and CTM processes
and was compared to vendors data for the injection molding process (Table 1).
1182
Table 1. Flexural Strength and CTE Test Results
Machined RTM Injection Mold
G-10 CTD-101 Torlon PEEK
Flexural Strength* 482.0 413.2 338 233

(MPa)
Coefficient of 1.15** 1.21** 1.62*** 2.20***

thermal contraction
10-5 mlmIK
* At room temperature
** Between room temperature to 77K
*** Between room temperature to 422K
For the insulation breakdown test 2 ,3 a Hi-Pot tester was connected

between a coil lead and the coated key. Breakdown was determined when the
leakage current rose rapidly above a set value. Voltage was then turned off
and reapplied for the second breakdown reading.
To evaluate performance of the end part material and process in
production and operation, parts are incorporated into coil assemblies and later
placed in a test magnet. After a visual and dimensional inspection parts are
wound and cured (138 MPa and 135°C) into a coil assembly. The ends get
separated from the coil assembly, potted and sectioned (Figure 6). The
sectioned surfaces are surveyed and results are compared to the nominal
design dimensions. The parts get inspected also for material failure or
deterioration.
Coil assemblies containing acceptable parts which warrant further
evaluation may be placed in test magnets. These magnets are subjected to
complete cryogenic and electrical testing which simulate operating conditions
with the absence of beam.
Eleven coil assemblies have been completed to date containing various
combinations of end parts produced by the different processes.
Specifically, RTM return end saddles and keys using CTD-IOl, Torlon and
PEEK return end parts machined out of injection molded tubes and coated
Cut #1 at 15 degrees from
Cut
~ longi tudinal centerl ine in X and
45 degrees rotation about Z axis
Cut #4 transverse
Cente -
line
146.05 MM I
----I
i-oa---
(5.750)
Figure 6. Sectioning of return end inner coil end
1183
aluminum return end keys have all survived the winding and curing process.
However, it was determined that an adhesive other than the one presently
used for the G-10CR end parts should be used with the Torlon and PEEK
parts. Only RTM return end saddles and keys which were produced in the first
part of our development program (using Tactix 123 resin) have been tested in a
test magnet. Those parts tested successfully with similar tests scheduled in
the near future for the RTM parts produced in the second part of the program
(using CTD-101 resin).
CONCLUSION
At the start of this development program the base line performance

criteria was that of the machined G-10 CR end parts. At the present time, due
to changes made in the requirements, the baseline design was changed by
General Dynamics to machined Spaulrad. Issues of higher radiation levels,
elevated curing temperatures, and requirement for stronger keys had to be
addressed in addition to the main objective of reducing cost.
Performance of RTM parts in coil assemblies and in test magnet as well as
results of the flexural test, indicates that this process can produce the strength
required from end parts. Using the CTD-101 resin system gives assurances to
its ability to perform in cryogenic temperatures as well as high levels or
radiation. Temperatures above Fermilab curing temperature (135CO),
however, could present a problem using this resin system. Further
development of the preform production in addition to improving of parts
quality and process reliability are still necessary.
Injection molding on the other hand offers a less expensive alternative
when compared with the RTM process. However, the strength of molded end
parts may prove to be insufficient. It is crucial to produce a set of molds and
parts which can be tested. This will also aid in understanding the challenges
involved when molding with materials like Torlon and PEEK (or PEK).
Though it may offer the required strength and moderately priced parts, the
compression molding process needs to be further studied and investigated. The
use of crushed Spaulrad is presently being studied by Spaulding Composites
and General Dynamics.
Metal coating can offer the added strength required in particular from the
return end keys during curing. Though finding a coating which will satisfy all
the requirements may require some research, it should be further explored.
As can be seen from the cost comparison in Table 2, the methods discussed
in this paper can offer a cost reduction when compared with machined end
parts.
Table 2. Cost Comparison for Return End Saddles - Machined Versus Other
Processes (for 10,000)
Machined RTM CTM Injection Injection

(Material CTD·101 CryoRad CTD-101 Torlon PEEK
& Labor)
$150 $51.0 $205.0 $31.0 $11.0* $37.0*
Tooling: $58,500 $54,000 $45,000 $23,000* $12,000*
* These prices are for 40 mm end parts and are about 1 year old.
1184
ACKNOWLEDGMENTS
The authors gratefully acknowledge the sincere contribution of A.
Knauf as well as K. Swanson.
REFERENCES
1. Munshi, N.A., "Superconducting Magnet Insulators: Radiation-induced
Damage and Effects," presented at CECIICMC, Huntsville, AL,
June 11 - 14, 1991.
2. Sims, R.E., "Insulation Breakdown Test of 50 mm Dipole Aluminum Lead

End Winding Keys," Fermi National Accelerator Laboratory,
Batavia, IL, Technical Support Technical Memo, TS-SSC-91-200,
October 1991.
3. Sims, R.E., "Voltage Breakdown of Various Types of Sealed Aluminum

Hardcoat Anodizing," Fermi National Accelerator Laboratory,
Batavia, IL, Technical Support Technical Memo, TS-SSC-91-006,
October 1990.
1185
MANUFACTURING CONCEPTS FOR THE COLLIDER
QUADRUPOLE COLLARED COILS
J. B6er 1 )i H. Fechteler 1 ){ R. Konschak 1 ), F. ostwald 1 ),

F. Sommer ), D. Krischel 2 }
Siemens, Power Generation Group (KWU)
1) Millheim Works, D-4330 Millheim/Ruhr
2) Accelerator and Magnet Technology, D-5060 Bergisch
Gladbach1
Abstract
On the basis of the existing experience from the HERA
project and the development work at LBL on the CQMs the
interdependance between design and major manufacturing
concepts is discussed, giving the example of splice free
winding.
With a modified coil end configuration this allows for
continuous collaring in vertical position. Other
manufacturing features are shortly described.
Introduction
The Babock & Wilcox/Siemens team in 1991 had been awarded
the contract for development and fabrication of the SSC
collider quadrupole magnets (CQM) up to the low rate initial
production. The prototype, preser ies and LRIP phases wi 11
lead to the final design and manufacturing concepts ready for
full rate production. Technically the work is based on
extensive and successful development efforts including
manufacture of several full length quadrupoles at LBL. [1] In
addition experience is available from the successful
manufacturing of HERA quadrupoles, which used some different
approaches. Evaluation of such technical background as a
projection to the specific CQM requirements are described as
well as the actual status of planning for prototype
manufacture.
LBL-CQM /HERA comparison

Table 1 lists the dominant features of the two approaches
to develop and build collider type of quadrupoles within an
industrial environment. Main parameters of the magnets are
also presented for comparison.

Table l. Design Comparison: LBL versus HERA.
LBL-COM HERA-COM
Cable
Dimensions 9.73 x 1.166 9.S0 x 1.S0
Insulation 1 x Kapton 100 HN 2 x Kapton SO HN
Coli
Diameter 40mm 7Smm
Length -Sm ~2m
Windings, Inner 8 11
Windings, Outer 13 16
Configuration
Layer Crossover splices splice free
Collar Sheet 4 piece,AI 2 piece, SS
Insulation Scheme Kapton, Prepreg Kapton, Prepreg
Quadrupole Collared Coil Design Approach
Design reflecting special manufacturing aspects and

industrialization
Beside providing a safe and reliable design industry must
include the aspect of industrial mass production in their
design and conceptual planning. Therefore sometimes
manufacturing imposes additional constraints and
requirements.
In no way, however, this must lead to any additional
risks for the magnet operation.
Table 2 Pro's and con's of splice free winding

Baseline Concept
• Splice-free winding
Advantages Affects
· Loss-free interconnect Cryogenics, Risk
• High form stability of joint · Field, Structural
two-layer coil
· Allows for full length collaring · Manufacturing,
Structural
• Facilitated current feedthrough • Manufacturing
· Fewer parts · Manufacturing
• Cost reduction • Manufacturing
Disadvantage
• Handling operation of inner · Manufacturing
layer
Neutral
· Double curing of inner layer
1188
Comparing the summary in table 1 we had to check whether
to follow one of the two approaches or find new solutions.
Some major decisions were made as trade-offs derived from our
experience with the HERA project, e. g. splice free winding
of inner and outer layer, fig. 1
A fishbone type GFRP spacer, as shown in fig. 1 is used
between the layers. It provides a well defined winding
surface for the second layer as well as some helium
transparency of the coil layer stack being favorable for
recovery after quench.
In a separated two layer coil current lead ends crossing
over the coil end region demand for a special support
structure in that coil section. An example is given in fig 2
for the LBL design using a set of conical collets. Intercoil
splicing would be made during the final assembly within the
yoke and helium vessel.
Of special importance to manufacturing is the fact that
splice free winding opens the option for collaring with pole
free but otherwise unchanged collars. As can be seen from
fig. 3 an end configuration was developed which allows for
complete assembly of the collared coil including intercoil
splicing and also the option to apply axial prestress of the
coil end region, if found necessary the structural analyses.
Thus clear interfaces are provided to further manufacturing
steps and ease of separate testing. No new type of pieces
need to be handled and collaring can be done more or less as
a continuous process in vertical position of the preassembled
coils.
Fig. 1 Splice free two layer winding, layer cross over
1189
Manufacturing concept and status
The winding machine is of the race track type, i. e. the

cable is guided around a stationary winding table. No bar is
allowed when winding the lower coil layer due to the storage
cable for the upper layer, see fig. 4 showing part of the LBL
winding machine.
Fig. 2 Coil end configuration /1/
____ ...:... ____ 1-
1-I--+l- · ...:.s:..j+-t·-9-~oF=· -~-·--- · --·-
Fig. 3 Coil end configuration
1190
Fig. 4 Race track winding machine /1/
The winding mandrel is directly taken in its winding

position as an insert to the curing mould. Different from a
dipole no pusher bars are necessary, fig. 5 showing the
arrangement using modified tooling from the HERA project.
The vertical collaring position facilitates the
application of full length stiffbars smoothing out the local
application of forces from the collar press, which is only
450 mm long.
Fig. 5 curing mould with coil in winding position

(modified HERA tooling)
1191
The press contains eight hydraulic systems arranged
around the coil/collar assembly, four to push on the collars
and another four to press the keys into their groove. An
expandable mandrel will be used working allong the coils axis
preventing local overpressing of the collars on the coils
assembly. The two piece collar lamination must be stacked by
turning azimuthally the position of the pole nose by 90
degree from layer to layer making up a 4 pole configuration.
With the number of 1664 quads to be built actually an
automated stacking procedure is under development.
still in the vertical position the leads from the four
coils are soldered in grooves of the GFRP splice plates.
The CQM hardware program will be started in March 1992
with the manufacturing of 1 m long collared coils being part
of a model magnet program. This will be done on CQM modified
HERA tooling as indicated in fig 5. The intent is to find
confirmation of the concepts e. g. the splice free winding of
a quadrupole wi th only 40 mm inner diameter. Some
restrictions on the final evaluation are still given by this
program due to the reduced magnet length and the tooling.
Therefore the program will proceed at the end of 1992
with a full length model magnet program to provide the basis
for the 5 m length prototpye program.
References
[1] Clyde Taylor et al., Quadrupole Technology Briefing for
the Superconducting Super collider, LBL, Berkely (1990)
1192
NEURAL NETWORKS AS FIRST LEVEL TRIGGERS
IN SCINTILLATING FffiER DETECTORS
Joseph A. Orgeron
High Energy Particle Physics Department

University of Texas at Dallas
MIS Fo.23
P.O. Box 830688
Richardson, TX 75083-0688
E-Mail: [email protected] (Bitnet)
INTRODUCTION
Theuseoftrained feed-forward neural networks has the promi se ofprovi ding a novel approach
to performing extremely fastpattem recognition which could be incorporated into a first-level trigger
in particle detectors such as the one proposed by the Solenoidal Detector Collaboration (SDC) at the
Superconducting Super collider (SSC). Extensive Monte Carlo simulations specifically written for
high energy particle physics, in addition to neural network simulator programs, are being used at
UTD and at the SSC Laboratory to train and test the accuracy, efficiency, and speed of these
networks. The final goal of this research is to determine the optimum architecture, training process
and most cost efficient method of hardware implementation.
At the designed luminosity (10 33 cm-2 sec· l ), there will be about 60 million bunch crossings!
sec; and with an average of 1. 6 interactions per bunch crossing, this brings the interaction rate to about
100 MHz. A typical event at the SSC, as generated large simulation programs, contain on the order
of 500 collision products. In order to find new and interesting physics in an environment such as
this, designers and engineers begin to realize that one of the most ambi tious and non -tri vial tasks will
be to filter out interesting events form the million or more 'trashy' events that separate them.
THE SDC'S SCINTILLATING FffiER DETECTOR
The SDC detector is a general purpose 4n detector that is 24 m (-75 feet) tall, 40 m (-130 feet)
long, and will weigh close to 30,000 tons [Tr90]. The main reason for this immense size is to be
able to have enough material to track, stop and accurately measure the energy of the high energy
collision products. The component that gives the detector its name, the superconducting solenoidal
magnet, will produce a massive magnetic field of 2 Tesla parallel to the beam pipe that will cause

the 'trashy', low energy collision products to be bent into ever tightening spiral trajectories near the
beam pipe and allow very energetic collision products from interesting, hard scatter events to flee
from the vertex area with little deviation from straight radial trajectories. This method ofmomentum
spectroscopy is very useful in determining interesting events for a triggering system. One of the
proposed methods of tracking to be used in the SDC is scintillating fibers with diameters of the order
of a millimeter. As planned now, the fibers are placed in ribbons which are then stacked into
cylindrical superlayers. The superlayers are positioned at the radial distances of75, 105, 135, and
165 em and accommodate either the front or rear half of the tracker [Figure 1]. The structure within
a superlayer, though still dynamic in design, is basically comprised of 4 layers of fibers that run
parallel to the Z-axis or beam pipe and dual layers, or doublets, of fibers that spiral around the Z-
axis with a 3-5 degree offset. These stereo fiber layers, denoted U and V, provide depth information
along the Z-axis by determining the point where the hit stereo fiber layers overlap. Scintillation light
generated inside of the fibers, caused by the passing of a particle, is piped out of the detector through
clear readout fibers, detected and converted to an electrical digital output signal by 85% quantum
effecient VLPC' s (Visible Light Photon Counters). As proposed by this research, it would be at this
point that a copy of this signal would be sent to a neural network which would help to make the
decision of whether or not this particular event should be saved for further evaluation.
Scintillating Fiber Superlayers
+ k-------H,---:Ior---4mm------------~
165m
J.95m
Interaction Point Beam Pipe

Figure 1. Schematic quadrant view of the scintillating fiber detector.
TRIGGER REQUIREMENTS AT THE SSC
The critical process of separating the proverbial 'needle in a haystack' is planned to be

accomplished by performing different levels ofevent triggering. The most significant of these levels
is the one that is directly related to this research, first-level triggering or L 1. It is here that the largest
reduction (or discarding) of events is planned. Currently, the L 1 trigger rejection factor is 10-3 to
10-4, and it is expected for the trigger to return a decision every 16 nsec with a 2-4llsec latency. This
latency time is provided by large FIFO (First In First Out) memory buffers that will temporarily hold
an event's data while the trigger makes its decision. The next level of triggering is the second-level
triggering or L2. At this level, there will be a rejection factor of only 10-1 to 10-2 and 10 msec of
decision time_ This is a reduction in events that is 100 times smaller than L 1, and it has about 1000
times more time to make its decision. These conditions allow for the L2 trigger to use a more precise
algorithm and apply larger amounts of scrutiny on the events that are passed to it. Beyond L2 is the
third-level trigger farm or L3 Farm which is actually a parallel array of computers all running the
same triggering algorithm. This trigger will have approximately the same reduction rates as L2 but
1194
will apply even further scrutiny to the less than 1000 events/sec passed to it and try to reduce the
number of events that are actually written to tape to about 10 to 100 events per second.
Now focusing attention to the L 1 trigger in the central tracking volume, the basic requirements
that are stipulated by the sse can be examined. First and foremost, the L1 trigger has to 'flag' events
when and only when a programmable number (or more) of charged particles of transverse
momentum, p J. = 10 GeV/c, pass through the tracking chamber. Secondly, it must be able to tell the
position of a track with at least the same resolution as the calorimetry system and return an accurate
J
estimate of the energy (or p of the particle. Lasdy, it must be very fast in order to keep up with
the decision rate as mentioned above. In any triggering algorithm, the efficiency is usually degraded
by false and missed triggers. Combinations oflow p J. tracks that coincidentally mimic the track of
a high p J. particle are the usual culprit and a good example of a false trigger.
NEURAL NETWORKS
Neural networks (neural nets) are a massively parrallel and intricately connected collection of
relatively simple processors. Based on the same basic design upon which the human brain operates,
they are usually started in a random state and trained into responding correcdy for problems which
no one good solution algorithm is present. The processors (or neurons) contained in layers above
the input layer begin by summing all of the products of the input values and the synaptic weights
which the input values travel on to reach the neuron. Then a threshold value, unique to that neuron,
is then sumtracted from this sum. The remainder then becomes the argument for a heaviside
stepfunction. If the remainder is larger than zero (0), then that neuron will pass on the value of one
(1) to all neurons above it and pass on zero (0) otherwise. For the purposes of triggering in the SDC
detector one particular type of network, the feed-forward neural net, is preferred due to it's fast
parrallel processing capability. Once the general architecture is determined for the application, a set
of preferably non-redundant input pattems, or vectors, that cover the entire range and scope of the
input phase space that is to be learned is compiled. These input examples must also be accompanied
by the desired output vector or value. The training process is primarily comprised of repetitive
presentations of input patterns and a Comparative analysis of the networks output to the desired
output. For each incorrect response, the synaptic weights and thresholds are slighty adjusted to better
facilitate the correct output for that input pattern on the next presentation. This process continues
until convergence, error = 0, or some maximum number of attempts has been reached. A non-
convergence situation is usually the result ofone or both of the following problems: 1.) The capacity
of the network is too small. In other words, there are not enough neurons and synapses to properly
'map' all of the input vectors to the desired output vectors. 2.) The training set has two or more
conflicting patterns. An example of this would be if one pattern were to inadvertently put into the
training set twice with different output vectors. It is quite easy to see how this second case would
be a non-converging network. Once training is completed, the network can operate in an extremely
fast recall mode as would be needed for an L 1 trigger.
GOALS AND APPROACHES
The main goals of this research work are to: 1.) perform an unbiased investigation into finding
the best pattern recognition neural network for detecting stifftracks in a cylindrical scintillating fiber
tracking chamber and be able to make quantitative classifications, such as transverse momentum of
those found, and 2.) find the most accurate and cost-effective method for hardware implementation
of this neural network.
The planned method for attaining these goals is to use a repetitive 'train and test' process of the
neural networks until the optimum performance is obtained. These networks will be trained with
1195
data that will be generated by SDCSIM (a.k.a. the SHELL). This program was created (and is
constantly being modified) by simulation experts within the SDC to most realistically simulate the
performance ofthe detector as a whole. The majority ofthe intensive computations will be performed
on the sse's Physics Detector Simulation Facilities (PDSF).
GENERAL DESCRJPTION
In order to meet the planned reduction factor of 10.3 , the first-level trigger for the scintillating
fiber tracker has to quickly identify the small fraction events that contain highly energetic decay
products. These decay products pass through the 2 Tesla magnetic field with very little deviation
from straight radial lines when viewed end on. This prominent feature intuitively makes the division
of each half of the barrel into a wedge-type geometry, the best candidate for parallelization. As an
initial configuration, contiguous wedges (211/235 angular displacement in cp) will have no overlap,
but at a later date an optimization study will be performed in order to determine the overlap, if any.
A copy of the digital, binary signals from every Z-axis scintillating fiber in a given wedge is sent
through a 2-stage triggering process [Figure 2].
In the first stage, a 4:1 data compression, referred to as clustering, performs an 8 bit boolean
function on contiguous fibers and returns a TRUE (or +1) output, if the hit combination could have
been caused by a particle with a Pi> lO GeV/c and a FALSE (or 0) otherwise. The preprocessor
outputs now become the inputs to a much larger pattern recognition network. It will basically be
End-On Viow 01 soc Central Blow-Up Viow 01 Superl"yar Construction
Tr"cker wI Wedge with Clustering Bin P"lIem end Schem"tic
01 ' nal PToce
Figure 2. Complete schematic diagram of the proposed neural network triggering system.
1196
a pyramid-style of architecture with about 4-8 outputs at the top and a yet undetennined nwnber of
hidden layers and neurons per hidden layer. The exact number of outputs will be determined once
an optimization is done on the number ofmomentum 'bins' (N)D- = 2°"'18) that will be needed to best
describe the momentum of the triggered particle. Once the triggered momentum from each wedge
is converted from the binary output, all of the non-zero transverse momenta and corresponding
wedge locations will be sent to a programmable 'total tracking volwne' trigger processor that will
use this information to determine if this event should be passed onto the global (all component) Ll
trigger. At this point it is important that it be noted that the scope of this research does not include
the design of the above mentioned programmable 'total tracking volwne' trigger processor nor the
global L1 trigger.
TRAINING, TESTING AND OPTIMIZATION
Being one of the most important aspects of this research project, integration into the framework
of the SHELL allows for the most realistic, raw data to be generated for training and will also serve
as a standardized platform from which the algorithm can be benchmarked for comparisons in
accuracy and efficiency to all other competing triggering algorithms. The SHELL can basically be
described as a compilable and runnable collection of large programs. PYTHIA, ISAJET and
HERWIG, which are event generators; and GEANT, a detector simulator, make up the majority of
the SHELL with the remainder being subroutines for SDC-specific initialization and code
management.
Testing of the full triggering algorithm will be done in two phases. The first and most recently
completed phase consists of sending single particles of varying P.L through wedges in order to
measure the efficiency ofthe clustering network as a function ofP.L [Figure 3]. The second and larger
1.00
..e-'
.R. Ir ~= ~
~
0.90
~
0.80 N
)~
~
0.70
)
/
0.60
~
c
.~
......
0.50
'7
X
!E
~
,£
.((8
.;
Xl [7.1
71 ZXfi X1
0.40
/
::I
U
r .xl ~OO.l• )
0.30
0.20
0.10
f
~
4i
0.00
o 2 4 6 8 10 12 14 16
Transverse Momentum (GaV/c)
Figure 3. Clustering effeciency vs. transvers momentum. Each data point based on approximately 750 cluster
canidates generated by SDCSIM and 8-bit cluster pattern (inset)
1197
phase consists of using full detailed events to see how well a trained neural net can distinguish
between wedges containing high P.l. tracks and those that do not in order to provide a benchmark
quantity which will take both efficiency and false trigger rates into account and give an overall
performance rating to the trigger algorithm.
In order to get a crude approximation of the performance of the second (linking) stage, a very
course simulation was constructed which utilized 1 cm, rather than 1 mm, fibers. The result was a
wedge that had only 16 possible clusters in the four superlayers. Single particles were then shot
through the wedge and the hits with thier corresponding transverse momentum were trained on a 16
input - 1 output neural network. After training, randomly generated single track events were
triggered yielding-95% accuracy. When a second, lowerp.l. 'noise' particle was added, the accuracy
level dropped to only -86% [Figure 4]. These very course tests showed a definite capacity forpattem
recognition in this type of detector geometry.
At the present status of this research, there are three optimization studies that need to be
performed. The first will be to determine the number of output neurons that will be needed to best
describe the P.l. of the triggered particle. The second optimization study deals with determining the
1.00
~
..... ft
I
0.90 1:1
-A...A.
0.80 ........
hI/ / ~
-n
!
0.70
J
II
0.60
u
c
.g
Q.
II
u 0.50
U
~~
~
-<
...
II
'"'"
·c
I-
0.40
[
0.30
$
!~
0.20
-I-
0.10 ...
0.00 i~
o
~~
5 10 15 20 25 30 35 40 45
Transverse ~omentum (GeV/c)
Figure 4. Trigger acceptance vs. transverse momentum. Each data point is based on 500 events passed through
a triggering wedge with tern fibers (16 fibers, see inset). Squares represent single particle events and triangles
represent two particle events with the first at the designated transverse momenta and the second at some
randomly choosen transverse momenta lower than the first.
overlap, if any, needed to have the best probability of catching any given high P.l. track in a wedge.
The third optimization, which will take place after training and testing are complete, involves the
minimization of the network's size in order to reduce the number of neurons and synapses and
facilitate easier hardware implementation.
1198
CONCLUSION - ADVANTAGES OF NEURAL NETWORKS
By far, the most predominant advantage to using neural networks is their very fast parallel
processing speed over any type of serial processor. Another advantage ofneural networks operating
in an SDC environment is their high degree of error resistivity. Unlike serial computers that may
completely fail if one bit or a single instruction is incorrect, it usually takes a sizable fraction of the
network to fail before a degradation in performance is noticeable [Nn90]. Lastly, it is hopeful that
results from this proposed research will be able to show that neural networks are also an economical
advantage that may not necessarily come only in the fonn of low cost neural network hardware
implementation but more possibly in the fonn of peripheral savings such as smaller FIFO buffer
space and no specialized fiber offset and holders.
Hopefully this work will be the first step of a best case scenario in which research would lead
to further testing. prototyping, production, SDC implementation and triggering by the year 2000.
REFERENCES
[Nn90] B. Muller and J. Reinhardt, Neural Networks -- An Introduction, Springer-Verlag, Berlin

Heidelberg, (1990), p.l9.
[Tr90] George Trilling, et al.. Letter ofIntent by the Solenoidal Detector Collaberation, November
30,1990.
1199
PERFORMANCE OF A PROTOTYPE mGH PRESSURE GAS
SAMPLING CALORIMETER
L. Demortier\ D. F. Anderson 2 , J. Budagov3, S. Cihangir2,

G. Fanourakis4, N. D. GiokarisI, K. Goulianos1, D. Khazins1,
M. MorganS, A. Para2 , A. D. Volkov3, and J. Zimmerman2
lExperimental Physics Department, The Rockefeller University

New York, NY 10021
2Fermi National Accelerator Laboratory
P.O. Box 500, Batavia, IL 60510
3Joint Institute for Nuclear Research (Dubna)
P.O. Box 79, 101000 Moscow, Russia
4Department of Physics and Astronomy, University of Rochester
Rochester, NY 14627
5 Ability Engineering Technology, Inc., South Holland, IL 60473
INTRODUCTION
Calorimeters covering the forward region of SSC experiments must be extremely

radiation hard, while at the same time providing adequate energy resolution and signal
speed. The SDC for instance, specifies that its forward calorimeter be able to sustain
a dose of tens of Mrad per year, that the energy resolution be better than 10% at 1
TeV, and that the calorimeter signal be shorter than 50 ns [1]. One technology which
appears capable of satisfying these requirements is based on the use of a gas such· as
argon or xenon, at high pressure, as sampling medium [2-10]. The pressure should
be high enough that the density of the gas is only a few times smaller than that of its
liquid phase. Such a device can then be operated in the ionization mode and has. unity
gain.
When compared with other types of unity gain calorimetry, high pressure gas
technology offers several advantages. In contrast with liquid argon devices, it does not
require cryogenic equipment, which poses well known safety, access, and dead space
problems. High pressure gases are more radiation hard [8], and less sensitive to impu-
rities, than warm liquids such as TMP. For a given electric field, electron drift is faster
in a gas than in a liquid (4 JLm/ns in liquid argon, versus 50 JLm/ns in a mixture of
95%Ar+ 5%CH4 at 100 atm and in a 750 V /mm electric field).

Plenum Press, New York. 1992 1201
Conventional 1 atm gas calorimeters on the other hand, present a different set of
difficulties. Their gas gain is a sensitive function of temperature, pressure, high voltage,
and gas composition, and requires constant monitoring. Because this gain is so high
(typically of order 10 4 ), they do not tolerate high particle fluxes. Finally, their energy
sampling fraction is very low, so that if hydrogen is present in the gas or in the detector
walls, protons scattered by slow neutrons from a hadronic shower or from the beam
halo cause very significant fluctuations in the energy measurement [11].
In this paper we describe the design and test of a prototype sampling calorimeter
made up of parallel steel plates with small gaps containing gas at high pressure [9].
The test was conducted in an electron beam at Fermilab. Its main objectives were to
prove that a calorimeter can be built and safely operated at up to 100 atm, to measure
the signal size and speed, to select an optimal gas mixture, and to measure the energy
resolution and compare it with Monte Carlo calculations.
The penultimate section presents a different high pressure gas calorimeter design,
which we are proposing for the forward region of SDC.
PARALLEL PLATE PROTOTYPE
Our parallel plate calorimeter consists of ten high pressure vessels, each of which is
composed of two steel disks bolted together (see figure 1). The disks have a diameter of
Figure 1. Prototype parallel plate electromagnetic calorimeter.
28.6 cm and are approximately 3.0 cm thick, making the assembled calorimeter almost
30 radiation lengths long. A 21.2 cm diameter, 5 mm deep recess is milled in one of
the two disks that constitute a vessel. Halfway inside this recess, a 1.5 mm thick G10
readout board is supported with ceramic spacers. Two copper pads are etched on each
side of the board: a 17.8 cm diameter electron pad and a 3.2 cm diameter muon pad.
Small holes through the board serve to establish electrical contact between the two
electron pads and between the two muon pads. High pressure gas occupies a 2 mm gap
on each side of the G 10 board. When the device is operated, the pads are brought to
a positive high voltage and collect the charge deposited in the cavity they are facing.
1202
The gas is supplied through a valve mounted in a special port drilled on the side of one
disk. A second valve provides a 101 atm safety relief. Three out of the ten vessels have
an 241Am a-source mounted in front of the muon pad. The spectra from these sources
can be used to monitor the gas. The signal produced by ionization of the gas is carried
from the collection pads to external preamplifiers through high voltage, high pressure
feedthroughs. To minimize the noise due to source capacitance, these preamplifiers are
mounted as close to the feedthroughs as possible.
TESTBEAM SETUP
Test of the prototype calorimeter took place in the Fermilab NT electron testbeam
toward! the end of the 1991 fixed target run. The bulk of our data was taken with a
fast, low noise preamplifier developed by the Shuvalov group [12]. This preamplifier
has an output signal rise time of 5 ns, an effective noise charge of approximately 3000
electrons for zero source capacitance, and an input impedance of 125 n. Preamplifier
output signals were fed into a LeCroy 2280 CAMAC ADC system, which was read out
by a MAC II computer through a GPIB interface. The data acquisition program was
based on the LABVIEW II software system.
Our trigger was BEAM· T1 . TRD . T2, where BEAM is a beam defining, up-
stream set of scintillation counters, T1 is a 2.54 X 2.54 cm 2 counter right in front of the
calorimeter, TRD is a transition radiation detector used to identify electrons, and T2 is
a 10 X 10 cm 2 counter placed behind the calorimeter and used to reject punch-through
hadrons that triggered the TRD. This trigger resulted in very clean electron samples at
20 and 50 GeV. A small fraction of pions however, contaminated our samples at higher
energy.
Pedestal events were collected during each data taking run between consecutive
beam spills. The trigger was provided by a pulser.
Electronic calibration was performed by injecting a pulse of known charge, and with
a width similar to that observed from beam particles, at the input of each preamplifier.
These data provide the absolute charge calibration and the relative channel to channel
calibration.
We took data with a variety of gas mixtures, pressures, and high voltages. For this
reason, it was important to monitor the stability of the calorimeter under a standard set
of run conditions. These standard run conditions were 50 GeV electrons, a gas mixtnre
of 95%Ar+ 5%CH 4 at 100 atm, and a cell high voltage of 1500 V. Figure 2 shows the
calorimeter response under these conditions, as a function of time. The straight line is
a fit to the data points. In the next section, we will describe the results obtained by
varying the incident electron energy from 20 to 125 Ge V, the gas pressure from 20 to
100 atm, and the cell high voltage from 150 to 2500 V.
DATA ANALYSIS AND RESULTS
Data analysis proceeds as follows. For a given run, channel pedestals are computed
by averaging the channel pulseheights over all the pedestal events in the run. For each
beam event in the same run, these pedestals are subtracted from the corresponding
channel pulseheights. The results are then multiplied by individual channel calibration
factors which convert ADC counts into charges. Finally, a small correction factor is
applied to compensate for the loss of pressure due to gas leakage. This correction is
typically of the order of a few percent. It is computed by linear interpolation between
gas pressure measurements performed before and after each run. Since we recorded the
1203
220 300
50 GeV e -
I I I I I
95%Ar+5%CH 4 250 r- -
g 210
1500 PSI
1500 V
g
..., ..., -
,
200 I-
Q/) Q/) 0 0 0
0
'"
.c
u '"
.c
r-
0
-
..,
U
...,:s
200 150
"C "C
0
:s
rn
'.," .,
rn
r- -
::s'"
100
T
::s 190
f 50' -
0 50 GeV e-
95%Ar+5%CH 4 -
1500 PSI
lBO 0 I I I I I
200 250 300 350 400 0 500 1000 1500 2000 2500 3000
Time (hours) High Voltage (V)
Figure 2. Calorimeter response under Figure 3. Charge deposited in the

standard run conditions, as a function calorimeter, versus high voltage.
of time.
time of occurrence of each event, it is possible to apply the pressure correction on an

event by event basis. The sum of the ten dectron channds is subsequently formed,
and the mean and width of the corresponding distribution are extracted by fitting with
a gaussian. A double gaussian fit is used whenever the pion contamination becomes
significant, especially at high energies. Since we did not measure the incident electron
momentum for each event, we correct the width of the electron peak by subtraCting
the beam momentum bite in quadrature. Our current best estimate for this bite is
(2.5 ± 0.5)% [13]. We similarly correct the width of the electron peak for electronic
noise, by subtracting in quadrature the pedestal width.
Figure 3 shows the charge deposited in the calorimeter as a function of the high
voltage across the 2 mm gas gaps, for 100 atm and 50 GeV electrons. The signal
saturates at the comfortable electric field of about 700 V /mm (our standard high voltage
setting was 1.5 kV). The calorimeter response versus pressure is shown in figure 4. It is
g
200
."
00
0:1
.c
150
t..l
"".:s"
.,0:1 100
2"
50
500 1500 2000

Gas Pr.lssure (PSI)
Figure 4. Charge deposited in the calorimeter, versus gas pressure.
1204
linear up to the highest point of 100 atm. The straight line is a fit through the origin
and all the data points. Its slope is (0.1278 ± 0.0001) fC/PSI. The dependence of the
calorimeter signal amplitude on beam energy is presented in figure 5. It is essentially
linear, except perhaps for the 20 GeV point. We are presently still investigating our
understanding of beam momentum for this point. A fit through the origin and the
three highest energy points yields a slope of (3.8 ± 0.4) fC/GeV, where the uncertainty
reflects our lack of knowledge about the absolute calibration of the detector. The energy
resolution is plotted versus 1/v'Ebeam in figure 6. Again, except for the 20 GeV point,
500
12.5
g
400
Ec: 10.0
...<t!
Q)
tID
300 o
:;:;
..c :::I
u "0 7.5
'"...
rn
Q) Q)
200 Ct:
:::I >,
rn
<t! ~ 5.0
Q)
:::e 95%Ar+5%CH.
Q)
c: 95%Ar+5%CH.
100 r.l
1500 PSI 2.5 1500 PSI
• 1500 V 1500 V
25 50 75 100 150
Beam Momentum (GeVjc)
Figure 5. Charge deposited in the Figure 6. Energy resolution versus in-

calorimeter, versus incident beam mo- cident beam energy.
mentum.
a linear relationship is observed. A fit through the origin and the three highest energy
points yields an energy resolution function of (43.2 ± 0.8)%/v'E, in agreement with
EGS4 calculations.
SDC FORWARD CALORIMETER DESIGN PROPOSAL
Dead space and safety concerns arise when one tries to design a parallel plate
calorimeter of the size needed to fill the forward region of SDC. We have therefore
developed a very different design, which is modular and lends itself easily to a tower
segmentation. Figure 7 shows one module ofthe proposed SDC forward calorimeter [10j.
Each module is 3 m long and has a 10 x 10 cm 2 cross section. It consists of 81 steel
tubes inserted through a stack of punched steel laminations, similar to the ones used
for magnet yokes. Charge collection rods run through the centers of the tubes. Tube
and rod sizes are chosen so as to optimize the energy resolution. Based on Monte
Carlo studies, our current best estimates are 6.4 mm and 9.5 mm for the rod and tube
diameters respectively. Ceramic spacers along the rods help minimize sagging, and
perhaps reduce the electromagnetic constant term in the calorimeter energy resolution
function. The tubes are closed at one end (the end closest to the interaction point),
and are welded to a 2.5 em thick steel manifold at the other end. The manifold itself is
welded to a steel plate which carries all the necessary gas valves and high voltage/signal
feedthroughs. All the rods are electrically connected, with a wire or a copper clad GI0
board, at the manifold end of the module. The tubes will be filled with a mixture of
argon gas with 5 or 10% methane, at a pressure of 100 atm.
1205
gas In/out va~"
Figure 7. One module of the high pressure gas calorimeter proposed for the forward region
ofSDC.
A forward calorimeter can be built from about 600 modules disposed almost par-
allelly to the proton beams. Its energy resolution can be parameterized as AI vE $ B.
The stochastic coefficient A is about 60% for both electrons and protons. The constant
term B depends on the angle ex between the tubes and the incident particle. For ex
between 10 and 50, B is approximately 6% for protons. Provided a suitable passive
preshower is placed in front of the calorimeter, the same value can be obtained for the
constant term for electrons.
CONCLUSIONS
We have successfully constructed and tested a prototype parallel plate electromag-

netic calorimeter with 95%Ar+ 5%CH. at 100 atm. This is a new record for high
pressure calorimetry. Preliminary results show that the signal saturates at the com-
fortable electric field of about 700 V Imm, that the calorimeter response is linear in the
pressure range 20 to 100 atm, and in the energy range 20 to 125 GeV, and that the
energy resolution agrees well with the EGS4 Monte Carlo prediction. Signals with a
rise time shorter than 15 ns and a base width of about 60 nsec were achieved.
A modular, high pressure gas calorimeter, constructed from parallel tubes inserted
through punched steel laminations, has been proposed for the forward region of SDC.
1206
ACKNOWLEDGMENTS
This material is based in part upon work supported by the Texas National Research
Laboratory Commission under Grant No. RGFY9160.
REFERENCES
[lJ Solenoidal Detector Collaboration, "Technical Design Report", SDC Note· SDC-
92-201 (1992).
[2J V.I. Baskakov et al., "Multiple-plate total-absorption ionization spectrometer

based on compressed gas", Nucl. Instr. and Meth. 159, 83 (1979).
[3J Ts.A. Amatuni et al., "A thin detector with ionization tubes for high energy elec-
trons and photons", Nucl. Instr. and Meth. 203, 179 (1982).
[4J M. Barranco-Luque et al., "HARP: high pressure argon readout for calorimeters",
Brookhaven National Laboratory Preprint BNL-31011 (1982).
[5J N.D. Giokaris et al., "Proposal for the development of high pressure gas calorimetry
for the SSC", submitted to SSC-DOE, June 1989 (unpublished).
[6J N.D. Giokaris et al., "High pressure ionization chambers for calorimetry in high
energy physics", Nucl. Instr. and Meth. A291, 552 (1990).
[7J D.M. Khazins et al., "Sampling gas ionization electromagnetic calorimeter", Nucl.
Instr. and Meth. A300, 281 (1991).
[8J L. Demortier et al., "Radiation damage studies on high pressure gasses", SDC note
SDC-92-182 (1992).
[9J L. Demortier et al., "Test beam results of a prototype parallel plate electromagnetic
high pressure gas calorimeter", SDC note SDC-92-183 (1992).
[10J L. Demortier et al., "A high pressure gas ionization tube calorimeter for the forward
region", SDC note SDC-92-184 (1992).
[l1J S. Cihangir et al., "Neutron Induced Pulses in CDF Forward Hadron Calorimeter",
IEEE Trans. Nucl. Sci., 36, 347 (1989).
[12J V.M. Golovin et al., "Hybrid low noise preamplifier on bipolar transistors" (in
russian), Preprint IFVE-89-231 (1989).
[13J H. Kobrak and X. Shi, T849, private communication.
1207
SIMULATED PERFORMANCE OF A HYBRID TRACKING SYSTEM
FOR THE SDC DETECTOR AT THE SSC
Martyn Corden, Mike Mermikides and Dong Xiao

Presented by Dong Xiao
Supercomputer Computations Research Institute

Florida State University
Tallahassee, Florida 31306-4052
INTRODUCTION
A hybrid tracking system has been proposed for the SDC outer tracking volume.
It combines two tracking technologies, straw tubes and scintillating fibers, exploiting
the strengths of both. The straw tube system provides high precision azimuthal mea-
surements of tracks in a solenoidal magnetic field. The stereo scintillating fiber layers
are mechanically straightforward to build. The proposed SDC tracking system also
contains a silicon inner tracker. The simulation presented here covers the barrel region
only, and includes 8 layers of silicon detectors. We will briefly describe the simulation
and track/event reconstruction procedures, then present the performance of the system
for high Pt tracking, including the tracking resolutions and efficiencies, and the results
of event reconstruction. We will also make comparisons of the system performance for
different background luminosities.
THE HYBRID TRACKING DESIGN
The major simulated parameters of the hybrid system are listed in Table 1. The
first four superlayers are composed of straw tube layers providing pattern recognition
and Pt measurement in the r-¢ plane. The fifth is a stereo scintillating fiber superlayer
providing z measurements of tracks.
Table 1. Parameters of the Hybrid superlayers.

superlayer radius length TJ structure elements stereo rad. length
1 85cm 500cm ±1.80 6z straws 2678x6 00 0.7%xo
2 1l0cm 600cm ±1.73 6z straws 3464x6 00 0.7%xo
3 135cm 700cm ±1.68 8z straws 4248x8 00 0.8%xo
4 160cm 800cm ±1.65 8z straws 5034x8 00 0.8%Xo
5 165cm 800cm ±1.62 2u2v fibers 13776x4 +5 0, _50 1.5%Xo

The tracking system is divided into two cylindrical modules at z=O. The straw
superlayers are composed of 6 or 8 layers of 4mm diameter straw tubes with axial sense
wires. Each superlayer lies on a carbon fiber mandrel. The total radiation length is
4.5%Xo.
Figure 1. Hybrid tracking system.
There is one stereo fiber superlayer as the outer-most tracking layer to determine
the z coordinate of a track. One or more inner fiber superlayers could be added in the
future to upgrade the system for higher luminosities. The fiber layers use the same
support system as the straw superlayers.
The fiber superlayer consists of two layers of scintillating fibers with stereo angles
of +5° and two layers of -5°. For each stereo angle, the two fiber layers are staggered by
half a fiber diameter. The diameter of each fiber is 1.5 mm. An outer stereo superlayer
will also provide precise matching to the calorimeter and the muon chambers. Figure
1 shows the hybrid tracking system under study.
MONTE CARLO SIMULATIONS
One of the major goals for the SSC ISDC experiment is to look for Higgs particles.
We have simulated the decay of a Higgs (m=300 GeV Ic 2 ) into a pair of ZOs, which,
in turn, decay into leptons. These events were then tracked and digitized within the
hybrid geometry. The Monte Carlo simulation was carried out within the framework
of the developing SDC simulation package called the SDCSIM shell.
Event Simulation
To study single track reconstruction capabilities, single isolated muons with trans-
verse momentum of 200 GeV Ic were generated. They were superposed on various min-
imum bias backgrounds to study the performance at high luminosities. Primary Higgs
1210
events were generated using ISAJET The Higgs mass studied was chosen to be
300GeV /c 2 •
The minimum bias background events were generated by the PYTHIA algo-
rithms 2. An average number of 1.6 background events per bunch crossing corresponds
to a luminosity of 1033 [cm- 2] [S-1]. Four previous and two subsequent beam crossings
to the one containing the triggering event were fully simulated.
Detector Simulation
All interactions and tracks of energetic charged and neutral particles, as well as
the detector geometry and response, were simulated by GEANT 4. The silicon tracker 3
was also simulated so as to study the overall track reconstruction capability.
The simulation took account of drift time and dispersion, (for straws only), prop-
agation time along the wire or fiber, particle time-of-flight, electronic dead time, and
background from both in- and out-of-time events. An effective drift velocity of 67 mi-
crons/ns in a uniform 2 Tesla magnetic field was assumed. Sensitive times were 40 ns
for straws and 15 ns for fibers. For the geometric calculation of hits, both fibers and
straws were assumed to have circular cross sections and to be close-packed. Geometric
inefficiencies might result from straw tube walls, fiber cladding, and the gaps between
straws or fibers. - For the calculation of particle interactions, the straw material was
assumed to be collapsed into concentric cylinders of the correct density near the rear
of each individual straw layer.
PERFORMANCE
The occupancies for the straw and fiber superlayers as a function of superlayer
radius and background luminosities are shown in figure 2. At the design luminosity, the
st~aw tube occupancies vary from 3% to 8% from the outside to the inside superlayer
for an average beam crossing, and are approximately twice as great for a Higgs -+ 4
lepton trigger. _
The occupancy of the outer fiber superlayer for an average beam crossing at design
luminosity is approximately 0.5%. Tests have been made to include the material of the
• Higgs+minbias r=85cm (straw)

• minbios
0.30
/
c
o
g.0.20
u
g
r=160cm (straw)
~"~')
0.10
0.00 L.1...-II::::=l::=:c:=r:::::;:::r::::::::L......J
o 1 2 345 6
minbias luminosity x 10" (cm-'s-')
Figure 2. Occupancies of the straw tube and scintillating fiber layers.
1211
magnet coil in the simulation. No significant change in the occupancy of the outer
superlayers was observed.
Pattern Recognition
A local on-road track finding method is used for pattern recognition. Since it
only depends on the spatial proximity of candidate hits to the track, the method does
not depend on knowledge of the event vertex, and is relatively insensitive to non-
uniformities in the magnetic field.
As a first step, an independent search is made in each axial straw superlayer for
segments of tracks, which are approximately radial to within about 20°. This is based
initially on the pattern of hit tubes only, and requires 4 (or 5) hit layers out of a 6 (or
8) layer superlayer respectively, with a possible inner gap of no more than one layer.
Then a first fit is made to the drift time, which permits a preliminary resolution of
the left-right ambiguity, and from the degree of stagger determines the to correction
needed for the propagation time along the wire. A final fit gives the segment position
and direction for use in segment linking. At this stage, only axial information is used.
Track segments, from both the hybrid and the silicon trackers, are then linked into
chains, starting from the outside of the detector, where the segment density is lowest.
No assumption is made about tracks originating from the primary vertex. All nearby
pairs of segments are tested to see if they are consistent with a circle on the r-rfJ plane.
This is possible, since each segment has both a position and a direction measurement.
Once a chain of four segments or more has been found, it is re-extrapolated to look for
more segments which might have been missed on the first pass.
The chains of segments are fit to determine the track parameters by a circle fit
on the r-rfJ plane, followed by a separate s-z fit. Where a beam constraint was used,
the uncertainties were ±10 microns in the transverse directions, and ±lmm in the
longitudinal direction.
Single Isolated Track Resolution
The ideal resolution of the hybrid tracking system has been studied using iso-
lated, pt=200 GeV Ic muon tracks with no background. The transverse momentum
0.050,.-----,--,--,----,---,--...,-----,
c:
o
~ 0.040
(5
III
~
~ 0.030
C.,
E
o
E 0.020
.,
.,>
~
~ 0.010
~
0.000 '--_-'--_.1...-_-'--_-'-_....1...._-'-_-'
o 2 4 6
minbios luminosity x 10'" (cm-'s-')
Figure 3. Pt resolution of 200 Ge V / c muons on minimum bias background.
1212
resolution of 200 GeV Ic isolated muons superposed to minimum bias background of
various luminosities are presented in Figure 3.
The track fit imposes a vertex constraint of 10 microns in each transverse direc-
tion, and 1 mm longitudinally. The straw superlayer resolution shows the expected
value of about 60 microns expected in the absence of systematic alignment errors. All
tracks were successfully reconstructed. A momentum resolution of 3% is obtained,
which does not change when a minimum bias background corresponding to the design
luminosity is added.
Track Finding Efficiencies
Track finding efficiencies in the barrel region (45° :S () :S 135°) have been esti-
mated, both for isolated, pt=200 GeV Ic muons, and for muons from the decay of a
Higgs boson. At the design luminosity, the tracks are found with an efficiency of ap-
proximately 99% in each case, as shown in Figure 4. The high momentum tracking
efficiency remains high as the luminosity increases.
1.2
1.0 • em sm
>.
.,c
0
0.8
'u
~., 0.6
C7'
c
3<
0
0.4
~ • ),£ from Higgs on minbios
• isolated jJ. on min bios
0.2
0.0
o 2 4 6
minbios luminosity x 1033 (em-'s-')
Figure 4. Tracking efficiencies for high Pt muons.
Stereo Matching Efficiencies
Finally, the tracks are extrapolated a short distance (5mm) to the stereo fiber
superlayer and are matched to pairs of u and v scintillating fiber clusters. The close
proximity of the u and v stereo plane fiber layers to an axial straw tube layer greatly
reduces the matching ambiguities. If the stereo hits are found, the z coordinates of the
track are then (re)calculated. Figure 5(a) shows the efficiency of the stereo matching,
which is close to 100% for transverse momenta of 10 GeV Ic and greater. There is no
significant variation in the efficiency for high Pt tracks, although it goes down slightly
when Pt is close to 10 GeV Ic. The efficiency falls rather quickly when the track
momentum is below 1 GeV Ic .
Figure 5(b) is the efficiency as a function of the background luminosity. It shows
that the background luminosity does not affect the stereo matching efficiency signifi-
cantly for high Pt tracks.
1213
1.2 120
1.0
+ •••••• g 100
• • • •
+ +
>.
....... 0
c
a< 0.8 .!! 80
8 0
:E.,
""
:c'"
~ 0.6 c 60
c
.!! .....c
0
E
., 0.4 E 40
0
0.2
e!
.!!
OJ
20 • P,>1 OGeV/c tracks
0.0 0
10 40 70 100 0 4 8
PT (GeV/e) Luminosity x 10" (em-'s-')
Figure 5. Stereo matching efficiencies for muons from Higgs decays.
Reconstruction of Fully Simulated Higgs Events
A sample of Higgs events was generated with all four decay leptons lying within
the silicon barrel region. A preselection of tracks with 1771 :5 1.5 is made according
to the hybrid tracking region coverage. The Monte Carlo generator information was
used to identify the reconstructed leptons. The invariant mass distribution of e+ e- and
p.+p.- is shown in figure 6(a), where the width (0- = 1 GeV/c 2 ) is due to the tracking
resolution, since the generated mass of the ZOs is a delta function in ISAJET. The
tail of the mass peak is caused by electron bremsstrahlung. The generated mass is
reconstructed.
100
15
BO
40
5
20
0 .....
60 70
.........~...A
80 90
~-
100 110 ~D~0~~~2~0~0~~~30~0~~~4~0~0~wm~5~00
I+r invariant moss (GeV/e') Z°z!' invariant mass (GeV/c')
Figure 6. 1+ 1- and ZO ZO invariant mass distributions.
The ZO ZO mass distribution showing the reconstructed Higgs peak is seen in figure
6(b). The reconstructed width includes contributions of the momentum smearing and
the width of the Higgs mass. It is 13 GeV /c 2 , compared to the natural width of
8 GeV/c 2 •
1214
SUMMARY
A 3% transverse momentum resolution is achieved. The track finding and stereo

matching efficiencies are close to 100% for high Pt tracks. The efficiency remains high
even beyond the design luminosity. A Higgs event at L=10 33 has a similar occupancy
to a L=2 x 1033 minimum bias background.
References
[1] F. Paige, S.D. Protopopescu, ISAJET Monte Carlo, BNL 38034, 1986.
[2] Hans-Uno Bengtsson, Torbjorn Sjostrand, A Manual to the Lund Monte Carlo for
Hadronic Process, June 1991.
[3] Letter ofIntent by the Solenoidal Detector Collaboration, SDC-90-00151, Nov. 30,
1990.
[4] GEANT, R. Brun et al. , CERN DD/EE/84-1, 1986.
1215
A NOVEL PHOTOMULTIPLIER TUBE
FOR CALORIMETRY AT THE SSC
P. Cushman!, M.Iosue 2, B.Johnson2, M.Madden3 , R.Rusack\ and

M.Szawlowski 3
lYale University, New Haven, CT

2Litton Electron Devices, Tempe, AZ
3 Advanced Photonix Inc., Los Angeles, CA
4Rockefeller University, New York, NY
INTRODUCTION
A silicon target photomultiplier tube with an anode consisting of a large area
Avalanche Photodiode (APD) has been developed. The tube combines the advan-
tages of an avalanche photodiode, excellent linearity and large dynamic range, with
the properties of a phototube. The tube structure is proximity focussing, which is
not affected by high magnetic fields approximately aligned with the axis. The high
linearity makes this device attractive for calorimetry at the SSC. The extension of this
concept to an APD array will provide a cost effective fiber readout for both digital
(tracking) and analogue applications. A tube equipped with an APD array is being
developed currently as a readout option for the SDC Shower Maximum Detector.
HYBRID PHOTOTUBE DESIGN

A 16 mm diameter APD is located at the anode of a proximity focussed pho-
totube, 6.8 mm' from a photocathode. Fig 1 is a schematic of the tube design. The
photocathode is maintained at a negative high voltage (8-12 kV), the front surface of
the APD is held to ground through a 1 k11 resistor, and the back surface or anode of
the APD is biased with a positive voltage of 1-2 kV. Charge is supplied to the APD
anode from a 10 nF storage capacitor to ground and the signal is picked off the APD
cathode. No divider chain is required.
The APD is operated in electron bombardment mode: photoelectrons are ac-
celerated by the electrostatic field and strike the APD with energies in the keV range.
The gain of this first stage is therefore approximated by the photocathode voltage
divided by 3.57 eV, once losses at the APD's surface are accounted for. An additional
gain of 200-1000 is provided by the APD itself, yielding an overall tube gain of 5 x 10 5 •
Because of the first stage gain, the signal to noise of the hybrid tube is greatly im-
proved over that of a bare APD operating in the optical mode. The second stage of
amplification is not present in hybrid tubes equipped with a PIN diode target, where
gains beyond 5000 are difficult to obtain.
Supercollider 4, Edited by J, Nonte

INPUT
jl ~ OPTICAL
, !! SIGNAL
~r:::1",
iONIZATION AND
AVALANCHING
ELECTRONS
rOCAl PLANE
~. I
01. 750
! .~ l-!··~
J
rRONT SURrACE
-.173 or INPUT WINDOW
.077
.765
Figure 1. Schematic of tube design and actual prototype dimensions

I ...... . . . - - - - - - - - - - - - - - - , . . ,
APD Bias = 2 kV
8.00e+5
....... +-"~-_r_~-_r_~-_._~-_._~---j
4000 eooo 8000 10000 12000 14000
Photocathode Voltage (Neg. Volts)
Figure 2. Gain as a function of photocathode voltage for an APD bias voltage of 2 kV.
The characteristic gain curves have been measured in both pulsed and DC
mode. In pulsed mode a scintillator with a spectral peak at 430 nm. was activated
by a N2 ultraviolet laser. These curves as a function of both APD reverse bias and
gap voltage are shown in fig 2 and 3. The gain is extremely linear with applied
gap voltage (Vpc) above 1000 V, a threshold which depends on the surface treatment
of the APD. The gain as a function of reverse bias voltage has a sharp upturn as it
approaches the breakdown voltage. The signal to noise also degrades in this region
and optimum performance is found at reverse biases < 2000 volts.
MAGNETIC FIELD RESISTANCE

Since photoelectrons are accelerated by an electrostatic field, it is easy to
calculate the effect of a magnetic field on the tube. To first order, an axial B-field
will have no effect on the gain and serve mainly to constrain the electrons along the
field lines. This would be an advantage in segmented anode configurations such as
fiber readout since it would minimize optical crosstalk between neighboring cells. As
the angle (J between the B-field axis and the tube (or electrostatic) axis is increased,
the image is shifted to the side by A and the gain goes down as the area A of overlap
1218
$.000+6 , - - - - - - - - - - - - - - - - - ,
Photocathode - -8 tV
."
3.00e.8
2.000+6
.
Ii
..000+6
m
O.OOe+O +-_-..----l"iL---i";..-'"~m_-r-_ _..---~---I
1400 1100 1100 2000 2200 2400
APD bias voltage
Figure 3. Gain as a function of APD bias voltage for a photocathode voltage of -8 kV.
11,-----------------,
L=lmm. V=5kV
...
Ci
C
09
L=4mm V=8kll
-"
l=6mm. V=10kV
,..
~
.. 08
a; 07
D::
06
B=O 8 Tesla
10 20 ) 0 40 50
Angle (Deg) between Band E field
Figure 4. The relative loss in current gain due to an external B-field as a function of the
angle between tube and B-field axis. Three different tube geometries are considered.
between displaced image and APD active area:

i =A
-;- - = -2 ( a - stnacosa
. )
(1)
to 'll"r2 'II"
where a = cos- 1 [Ll/2r] and Ll ~ L/tan(J. This is plotted in fig 4 for three different
gap dimensions. Gain would be reduced by roughly 50% for a B-field at 45° to the
tube axis for the prototype tube design, but this gap can be reduced and depends
on which operating voltage (and thus gain) is desired. The 1 mm gap, however, is
probably only practical for photocathodes with large work functions, such as CsTe
(r/J = 4eV) for which dark currents due to field emission are reduced.
PHOTOCATHODE
The prototype tube has an S20 photocathode with a radiant response of only
10 mA/W at 430 nm (this was a mistake during processing and is not typical). Other
photocathodes are also possible. Tubes with a Gen III, GaAsAI, photocathode are
being manufactured in order to improve the sensitivity to scintillators in the longer
wavelength, more radiation resistant regime and to study the materials compatability
with III-IV materials. By extending the blue sensitivity of the GaAs-based photo-
cathode, we expect a quantum efficiency of 30% for green fibers. The response curves
for this and other available photocathodes are shown in fig 5.
1219
100
50
Hi Blu~
G.~
g
3D
Bialkali
( , ~
.,
G....
~
'\.. ,.-"'- ~ ,
~
~
C.T
~ "-
." I) 1\
10
\ . :~
I
e S2f
'lInG....
E I ~ ;"
,,,,
a" If ~
1'\ ,
~ ,
I
I
l I,
200 400 600 800 1000
Wavelength (nm)
Figure 5. Photocathodes readily available from Litton and their spectral response curves.
Another interesting application of the hybrid tube is ultra-fast calorimetry

using BaF2 which requires sensitivity in the ultraviolet. In addition to conventional
CsTe and CsTeRb, Litton is also developing a solid state photocathode, Aluminum
Gallium Nitride, which is sensitive to the UV and has 5 orders of magnitude falloff
over 80 nm above 290 nm. This would provide sensitivity to the BaF 2 fast components
at 190 nm and 220 nm, while suppressing the slow component at 310 nm by 3xl04 •
APD DESIGN
These are large area APD's with a "bevelled edge" design. The breakdown
voltage is very high with a wide depletion zone. Physically bevelling the edges pre-
vents breakdown at the junction periphery. These APD's therefore have very different
characteristics than the smaller ones (50-500 /Lm diam) typically associated with fiber
tracking readout which are produced using the "reach-through" structure.
The ratio of hole to electron ionization rates is
(2)
where the ionization coefficients of electrons and holes respectively are a and (3. This
is an important quantity since the lower ko, the larger the applied voltage (and thus
gain) before the multiplication curve bends to infinity. Fig 3 shows this happening at
about 2200 volts for the APD in our prototype tube. Typical k-values for the large
area bevelled edge APD's are k ....... 0.0025, a factor of 10 smaller than typical reach-
through designs, due to their lower peak electric field at breakdown l (remember that
f3 / Ot increases with the internal electric field).
Other consequences of a lower k-value are lower excess noise factor and less
variation in gain due to nonuniformities in the doping level of the depletion layer. The
dark currents measured on the set of 8 APD's delivered to Litton for potting ranged
from 160-350 nA for 16 mm diameter devices at room temperature. After potting,
the prototypes had dark currents of 370 nA and 700 nA, representing a factor of 2
increase after bake-out over the bare APD dark current.
TIMING CHARACTERISTICS
A large area also implies large capacitance; 120 pF is typical for the APD
in question, which results in a large RC constant and thus slow risetime. The 8
APD's had measured risetimes of 15-40 ns. However, for the application in mind:
fiber readout of the SDC Shower Maximum Detector, these large area APD's will be
1220
....
I/)
. ••
~o 3
:;
o
~ 2
U m
'Q.. m
.....
g'
...I
-8 -0 ... -2 0
log (Transmission)
Figure 6. The output charge in picoCoulombs as a function of relative light intensity as

defined by the transmission of calibrated neutral density filters.
subdivided into 8 X 8 arrays, reducing the capacitance by a factor of 64 such that

it no longer influences the risetime, which should then be in the neighborhood of 1-2
ns.
In either case, the transit time jitter is very small (tens of picoseconds), making
these excellent time-of-flight counters.2 This is due to 2 factors: (1) since a proximity
focussed tube is simply a planar accelerator, the photoelectrons will not travel dif-
ferent path lengths to the collection sudace, and (2) the electron-hole pairs are all
produced at the sudace of the APD by photoelectrons, rather than throughout the
device by impinging photons.
TEMPERATURE DEPENDENCE AND UNIFORMITY

Both the Temperature characteristics and uniformity depend on the APD. The
gain of an APD decreases with increased temperature. A 40% reduction in gain was
observed for a 20°C change in temperature from O°C to room temperature. However,
since the tube does not need a powered divider chain, it does not heat up during use,
making temperature regulation relatively easy. It also introduces the possibility of
running the device at liquid nitrogen temperatures to increase gain and reduce noise.
Nonuniformities in the APD are measured at the 5% level. Since most nonuni-
formities in conventional phototubes are due to varying collection efficiencies on the
multiplying structures, and since the photocathode itself is very uniform, it is ex-
pected that the tube will have less than 10% variation over the active area. This is a
very important property for multi-anode versions of the tube, since it means channel-
to-channel variations will be negligible. Compare this with the 5:1 average variation
in channel gain from the best multi-anode phototubes, which puts severe constraints
on the digitizing and trigger electronics which follow.
DYNAMIC RANGE AND NOISE

Measurements of the linearity of this device over 6 orders of magnitude have
been performed in both DC and pulsed mode. In the DC mode, a constant light
source illuminated the photocathode and the output current was measured directly
using an electrometer. No departure from linearity was observed over 6 orders of
magnitude. In the pulsed mode (see fig 6), blue light from a scintillator activated by
a N2 laser was directed at a 2 mm spot on the photocathode. A series of calibrated
neutral density filters reduced the amount of primary light reaching the tube. At
1221
4500
4000
3500
3000
2500
2000
1500
1000
500
PMT output ,;DC 4
Figure 7. Single Photoelectron Peak and pedestal
the low light end, the signal was amplified by a high speed preamplifier (x 10.1) and
a PS-777 amplifier (x50). At the high light end, the amplifiers were replaced by a
series of calibrated attenuators. The signal was recorded by an LRS 2249W ADC
with a sensitivity of 0.25 pC/channel. The departure from linearity which is observed
at the high end of fig 6, is due to current limitation caused by the size of the storage
capacitor. At the low end, there is considerable electronic pickup from the laser,
which makes background subtraction very difficult.
By turning off the laser and covering the tube entrance hole, the single pho-
toelectron peak was observed. At Vpc = -8 kV, V APD = 2 kV, it is well separated
from the noise (see fig 7). The signal corresponds to 0.103 pC/photoelectron with
(T = 48%. At Vpc = 7 kV, (T is only 33%. Sigma is expected to be even smaller with
lower reverse bias voltages, but we have yet to improve our laboratory setup to be
able to see the single photoelectron peak at these lower gains.
The noise figure is defined as the ratio of output signal width ((T / <signal»
to input signal width. The noise figure of the hybrid tube has been measured to be
1.4, which is lower than the value of 1.6 typical of conventional phototubes.
A scope trace of the single photoelectron pulse after preamplification is shown
in fig 8. This should be compared to fig 9 which shows a scope trace from the high end
of the linearity studies. The pulse (no amplification, X 5 attenuation) corresponds to
160 volts into 50 O. It shows no evidence of saturation, still exhibiting a characteristic
20 ns risetime.
1222
Figure 8. Scope trace of the single photoelectron signal with a X 10 inverting preamplifier.
Figure 9. Scope trace of the signal at the high end of the linearity study summarized in fig
6. The laser is operating without filters and a factor of 5 attenuator is on the scope input.
The signal corresponds to 160 volts into 50 n, or 3.2 Amps peak output.
1223
CONCLUSIONS AND FUTURE DIRECTIONS
We have designed and built a hybrid phototube which exhibits the following
features:
(1) Gain of 5x10 5
(2) Linear over 6 decades
(3) Sensitive to single photoelectrons
(4) Quantum efficiency of 20-30%, depending on photocathode choice
(4) Resistance to near-axial magnetic fields
(5) 10% variation in gain over active area
(6) < .1 ns transit time jitter
(7) 20 ns risetime limited by area capacitance
(8) Simple, low current operation at room temperature
The lifetime of this tube has still to be quantitatively assessed, and we are in
the process of doing such tests now. Electron bombardment could cause damage to
the silicon itself, and attendant outgassing can poison the photocathode over time. All
that can be said at this time is that after 4 months of testing, averaging 20 hours/week
of "on" time over the whole range of input conditions, we have not observed any major
damage.
We are interested in exploring its use as a readout for scintillating fibers,
specifically for the Shower Maximum Detector which is a finely segmented layer of
the SDC tile/fiber calorimeter. To this end we will be subdividing the APD into
an 8x8 array. The risetime for each channel will thus be several ns. In addition,
the good surface uniformity implies that the channel-to-channel variation will be
negligible. Crosstallc- between channels is also expected to be < 1 % since the tube is a
simple planar accelerator. Axial magnetic fields would improve any vestigal crosstalk.
The multi-anode version would also be very cost effective, an important consideration
when designing readout for 40,000 channels. Initial estimates for 64-channel devices
in quantity, put the cost at $16.00/channel.
REFERENCES
1. P.P.Webb, R.J.McIntyre, and J. Conradi, "Properties of Avalanche Photodiodes",

RCA Review, Vo1.35, p234 (1974).
2. R. Scharenberg, Private Communication.
1224
Attendees
James G. Abbee Bruce Anderson Dan Balcer

u.s. Department of Energy H&J Tool & Die Co. CVI Incorporated
Deborah Abrahamson L. E. Anderson R. W. Baldi
U.S. Department of Energy PDM Inc. General Dynamics Space
Shaul Abramovich David E. Andrews Systems Division
SSC Laboratory Oxford Instruments M. J. (Penny) Ball
Frank S. Adams Thomas Ankennann SSC Laboratory
TSA Process Systems Robert W. Balliett
Joanne Adams International Inc. NRC Inc.
TSA Tamirisa R. Apparao Oscar Barbalat
Jim Adamson SSC Laboratory CERN
Larcan Communications William Appleton Ron Barber
J. M. Adovasio CVI Incorporated Los Alamos National Laboratory
Mercyhurst ColIege/SMU Craig Arden Victoria A. Bardos
Ken Agarwal SSC Laboratory SSC Laboratory
General Dynamics Space Sumikatsu J. Arima Barry Barish
Systems Division U.S. General Accounting Office CalTech
Richard S. Ahlman Julie Armstrong Dan Baritchi
SSC Laboratory SSC Laboratory SSC Laboratory
H. Akimoto J. Grant Armstrong Doug Barnes
Hitachi Ltd. MKS Instruments TUElectric
Dave Alderson Richard Atkinson Paul Barnett
IGC Advanced Superconductors Silvex Energy and Natural Resources
Charles Alesi Stan Augustynowicz VinodBarot
Walker Magnetics Group Inc. SSC Laboratory Du Pont Company
Curt Allen Al Axt R. C. Barquin
SSC Laboratory Westinghouse Electric Washington Consulting Group
Ed Altgilbers Francois Ayer Klaus J. Bartels
Inland Steel Draper Laboratory Spar Aerospace Ltd.
Chris Anderson P. S. Ayres Honorable Joe Barton
DRD Corp. Babcock & Wilcox U.S. House of Representatives
Kathy Anderson Neil Baggett Henry R. Barton Je.
SSC Laboratory SSC Laboratory SSC Laboratory
Charles E. Anderson David B. Bailey Shans Basiti
Air Products and Chemicals SSC Laboratory SSC Laboratory
Richard W. Anderson Richard Bailey K. E. Bassett
Analogic Corp. General Dynamics Space Martin Marietta
Systems Division Astronautics Group
1225
Rick Baumann Regina Borchard H. G. Campbell
Union Carbide Industrial Gases Martin Marietta Babcock & Wilcox
Donald S. Beard Astronautics Group John Campbell
William N. Boroski Cryogas International
Gerad F. Beatty Fermi Laboratory B. I. Cantor
LTV Steel Company Lieuwe Boskma Babcock & Wilcox
David Beechy DEP R. Michael Capers
SSC Laboratory Rodger Bossert Scientific Instruments Inc.
David A. Bein Fermi Laboratory Don Capone
SSC Laboratory Bruce Boswell SSC Laboratory
Thomas J. Bell Westinghouse Electric Mike Caracciolo
Bell Machine Company Inc. Karl Brackhaus SSC Laboratory
Jewel C. Bell Dynapro Systems Inc. Ruben Carcagno
Bell Machine Company Inc. Frank Brian SSC Laboratory
Aldo E. Belloni Katie Group J. W. Carey
Lotepro Corporation Richard J. Briggs TRIUMF
Barron R. Benroth SSC Laboratory Nelson Carlo
U.S. Department of Energy John E. Brogan Abbott Products Inc.
William J. Berry Danco Precision Inc. Lee Carlson
EG&G Idaho Inc. D. A. Brouilette Superconductor Industry
Daryl Bever General Dynamics Space L. A. Carlstrom
General Dynamics Space . Systems Division CVI Incorporated
Systems Division Nat Brown Ron Carlton
Honorable Tom Bevill U.S. Department of Energy Office of Congressman Frost
U.S. House of Representatives Donald P. Brown Jim Carney
Armand Bianchi Brookhaven National Laboratory U.S. Department of Energy
Fermi Laboratory Rick Brown Richard A. Carrigan, Jr.
William B. Bigge U.S. Department of Energy Fermi Laboratory
Noell Inc. William T. Browne Jr. John A. Carson
Ed Bingler Marotta Scientific Controls Inc. Fermi Laboratory
TNRLC H. D. (Buzz) Bruner Harry Carter
JohnW.Bird Bechtel SSC Laboratory
SSC Laboratory John C. Bruno John Carusiello
Tony Bishop J&L Specialty Products Meyer Tool & Manufacturing
Allen-Bradley Co. Charles W. Bruny Shlomo Caspi
Bill Bishop Babcock & Wilcox Lawrence Berkeley Laboratory
Agra IndIVadeko Warren Buckles Annette Caudiano
Charles D. Bizilj Superconductivity Inc. Mitsui & Co. (USA) Inc.
Howden Compressors Inc. Lawrence Buhl Bill Cesarek
Ewart W. Blackmore Copper & Brass Sales Inc. Stellar Industries
TRIUMF Joseph Bularzik James Chafe
George B1anar Inland Steel Naval Surface Warfare Center
Le Croy Corporation Catherine Burns C. David Chaffee
R. J. Blanken SSC Laboratory AISISuper Collider News
IBMlFederal Sector Division Thomas O. Bush Natalie D. Chambers
Miriam Bleadon SSC Laboratory Furukawa Electric
Sci Tech Michael M. Busley Technologies Inc.
Locke Bogart General Dynamics Space Honorable Jim Chapman
General Dynamics Space Systems Division U.S. House of Representatives
Systems Division Martin ButaHa Anthony Chargin
Michael Boivin SSC Laboratory Lawrence livermore
New England Electric Wire Corp. John Butler National Laboratory
Millard Bolt Westinghouse Electric Coby C. Chase
Lockheed M&S Co. WalterBux SSC Laboratory
JohnW.Bonn LABENSPA Gang Chen
CVI Incorporated Regent Cabana Beijing Glass Research Institute
Quebec Government Office
1226
Dennis Cheney John Cryer Moshy Y. Dreksler
EG&G CRSS Architects Inc. Mycom Corporation
Thomas J. Chester Frank Cunningham Gilbert Drouet
SSC Laboratory Swagelok Company CERN
William Chinowsky Richard L. Curl Rens L. Dubbeldam
Lawrence Berkeley Laboratory The PBIMK Team HOLEC
Denis Christopherson Priscilla Cushman John J. Dubbs
SSC Laboratory Yale University Air Products and Chemicals
Adam Chrzanowski Edward Daly Gerald Dugan
University of New Brunswick SSC Laboratory SSC Laboratory
Joseph R. Cipriano Charles A. Dan Jonn C. Dykstra
U.S. Department of Energy SSC Laboratory MVE
Gordon T. Clark Jerry Daniel J. Dykta
The PBIMK Team Westinghouse Electric Leybold Vacuum Products Inc.
Susan Claude Leroy Daniel Dave Earthy
Sheldahl Inc. CERN SSC Laboratory
W. R. L. Clements LtDarby Beth Easley-Toth
MPB Technologies Inc. Babcock & Wilcox Inland Steel
Tony Cochrane Ted Dargan TonyElam
DEP Digital Equipment Corp. IBMlFederal Sector Division
Michael Coffey Yuri Daschko Bob Engel
Cryomagnetics Inc. Industry, Science & Technology Westinghouse Electric
Lindsay Coffman Vladimir Datskov John Engelsted
U.S. Department of Energy SSC Laboratory Walker Magnetics Group Inc.
Cort Coghill Mark Davidson Charlie Ericson
Babcock & Wilcox Fermi Laboratory Westinghouse Electric
Mark W. Coles John Davies Timo Erkolahti
SSC Laboratory Marconi Defense Systems Outokumpu Superconductors Inc.
James W. Collins John R. Davis Behrooz Ershagi
Texas Education Agency Irving Burton Associates Rotoflow Corporation
Eugene Colton Fred De Antonis Len Eschweiler
U.S. Department of Energy Allied Signal Inc. Jade Simulations
Mark Confer Dean De Benedet David B. Everson Jr.
Westinghouse Electric Martin Marietta Everson Electric Co.
J. A. Connor Earl Dedman Y. Eyssa
Babcock & Wilcox Aguirre Associates Inc. Babcock & Wilcox
Milciades Contreras Stephen W. Delchamps Shawn J. Fagen
University of Chicago Fermi Laboratory Advanced Photonix
Roger Coombes Joe Deley EliotFalk
SSC Laboratory Westinghouse Electric EDS Unigraphics Division
Larry Cormell Jonathan A. Demko Roger A. Farrell
SSC Laboratory SSC Laboratory IGC
BiffComing Luc Demortier Chris Farver
EG&G Wasc (SSCL) The Rockefeller University Allen-Bradley Co.
Larry V. Coulson Pete Dent Anthony Favale
SSC Laboratory Lake Shore Cryotronics Inc. Grumman
Kathleen Couzens Martine Deseraud Helmut Fechteler
SSC Laboratory C.E.A. Siemens AG
Scott Covington Arnaud Devred Jan Fickling
Cryenco SSC Laboratory Westinghouse Electric
Dennis Cox Nick Di Giacomo Yuri Filippov
SSC Laboratory Martin Marietta JINR
Phil Craig James Dillard Don Finan
Westinghouse Electric Barber-Nichols SSC Laboratory
R. Frank Cremeens Paul A. Draper Rod Fink
Teledyne Pittsburgh Tool Steel University of Texas at Arlington Acme Cryogenics
1227
Bob Fischer JamesE. Fye Vincent J. Grillo
General Physics Services Corp. mM Federal Division Cryofab Inc.
Gareth D. Aora Tom Gadsden Douglas Grones
Martin Marietta SSC Laboratory SSC Laboratory
Strategic Systems Philip O. Gagnon Kenneth M. Groover
Carlos Aores General Physics Corp. Union Carbide Industrial Gases
Barcel Wire & Cable Mike Gamble Gerard Grunblatt
Adrian L. Auhmann Los Alamos National Laboratory Alsthom Intermagnetics
Auhmann Inc. Henry M. Gandy John A. Gruver
Stephen A. Foerster lNRLC Westinghouse
Tempel Steel Company RaoGanni Tim Guitner
David Forber SSC Laboratory Ebara Technologies
AFA Industries Jeffery C. Gannon R. Gupta
Paul T. Forde SSC Laboratory Brookhaven National Laboratory
NRC Inc. William E. Garber Fred Gurney
Martin Fortner Air Products and Chemicals Westinghouse Electric
U.S. General Accounting Office C. J. Garvey SamGurol
Dave Fortunato Rogers Corp. General Dynamics Space
SSC Laboratory Alicia A Gaydos Systems Division
Gilbert Fox Jr. Amax Specialty Coppers Corp. AhmetGursoy
Plainfield Stamping-Illinois A Ghosh The PBIMK Team
Jim Fraivillig Brookhaven National Laboratory Nathalie Guyol
E. I. duPont G. O. Gigg SSC Laboratory
Janet Francis General Dynamics Space JoeGyan
JPO Equipment Co. Systems Division Baton Rouge Advocate
Jim Franciscovich Paul H. Gilbert Gregory M. Haas
SSC Laboratory The PBIMK Team U.S. Department of Energy
Donald E. Franks Rene Gillier Bernhard Haberthuer
SSC Laboratory Helicoflex Sulzer USA Inc.
Larry Fraser Frederick J. Gilman Thomas Hacker
MPR Teltech Ltd. SSC Laboratory Abbott Products Inc.
Thomas K. Frasier Nikos Giokaris Johnny Hagen
Gardner Cryogenics The Rockefeller University Edwards High Vacuum
Michael Freeman Guy Gistau International
Process Systems L'airLiquide Eric Haggard
International Inc. Sheldon L. Glashow Fermi Laboratory
Jim Freim Harvard University S.Hahn
Furukawa Electric Technologies Leonard M. Goldman Los Alamos National Laboratory
Raymond Fricken Bechtel Corporation Anthony R. Hald
U.S. Department of Energy Stephen R. Goodman ICF Kaiser Engineers Inc.
Bill Frisken USS Phyllis Hale
York UniversitylLBL Carl Goodzeit SSC Laboratory
Melissa I. Fromm SSC Laboratory Robert W. Hamm
Air Products and Chemicals E. T. Gossler Acc Sys Technology Inc.
R. Fromont Babcock & Wilcox Jack Hanlon
Superconductor Cabling James A. Gray Los Alamos National Laboratory
& Engineering SSC Laboratory Mark Hanna
David Frost Thomas H. Gray Techware Systems
Supercon Inc. Allegheny Ludlum Corp. Greg Hanowski
Honorable Martin Frost David Green SSC Laboratory
U.S. House of Representatives Saes Getters/U.S.A. Inc. Patricia A. Hansen
Bob Fuller, Jr. S. Steve Greenfield Hitachi America Ltd.
Westinghouse Electric Parsons Brinckerhoff Hisashi Harada
L. Warren Funk Eric Gregory Mitsubishi Electric Corp.
SSC Laboratory IGC Advanced Superconductors Samuel D. Harkness
James A. Fusco Westinghouse Electric
Iten Industries
1228
William E. Harrison Kenneth T. Hoffman Dave Johnson
Brookhaven National Laboratory Precision Defense Services Inc. Standard Mfg. Co. Inc.
Frank Hart Denslay Hoffman Coleman Johnson
Lydall Manning Nonwovens General Dynamics Space Lawrence Livermore
Don Harwell Systems Division National Laboratory
Cryenco Matt Hogan David E. Johnson
Keiji Hasegawa Office of Congressman Geren SSC Laboratory
Shimizu Corp. E. O. Hooker R. A. (Bob) Johnson
Ronnie J. Hastings Babcock & Wilcox JBC Associates
Waxahachie LS.D. Jerry Hopkins Gwynne Johnston
Toshio Hayashi General Dynamics Space Tempel Steel Company
Kawasaki Steel Communications Systems Division J. Bennett Johnston
Michael Hayes DickHora Senator D-LA
Kinetic Systems Corp. General Dynamics Space Carleton H. Jones
Systems Division Bechtel Corporation
Honorable James Hayes
U.S. House of Representatives Joseph Horn Proctor Jones
BOC/AlRCO EnergylWater Appropriations
Robert Hedderick
SSC Laboratory George Horvat Donald Jordan
Liquid Air Engineering Spaulding Composites Company
Tricia Heger
Fermi Laboratory Kenji Hosoyarna N. B. "Buck" Jordan
KEK Waxahachie Chamber
Warner Heilbrunn
SSC Laboratory Robert E. Hughes of Commerce
Associated Universities Inc. Nicholas Kallas
Ezra D. Heitowit
Universities Research Assn. Robert Hummel SSC Laboratory
Irving Burton Associates Jill F. Kallsen
Keith Helart
Ball Aerospace Systems Group William Hunter TeledyneSC
Laser Techniques Inc. HemKanithi
Roland Hellmer
Balzers Andy Hutnikoff IGC Advanced Superconductors
Allied Apical Co. Alvin Kanofsky
John B. Hendricks
Alabama Cryogenic Engr. Inc. Peter Hwang Lehigh University
Gardner Cryogenics Raphael Kasper
Steve Hensley
CVI Incorporated M. Ikeda SSC Laboratory
Furukawa Electric Julian A. Katz
Wilmot N. Hess
U.S. Department of Energy H. Ishida TRW/SIG
Kawasaki Steel Corp. Sam Kawarnata
Carl Heyne
Westinghouse Electric Tsutom lsobe Hitachi Cable America Inc.
Furukawa Electric E. Kelly
A. D. Hibbs
Quantum Magnetics H. J. Israel Brookhaven National Laboratory
HOLEC K. Kikuchi
Susan Hickey
Lockheed Austin Division Jon Ives Sumitomo Heavy Ind.
SSC Laboratory Ed Killian
Jeannie Hicks
Office of Congressman Chapman James Ivey Brookhaven National Laboratory
SSC Laboratory Mike Kimmy
liro Hietanen
Helsinki University, SEFT Robert A. Jake General Dynamics Space
American Magnetics Systems Division
Gale E. Hill
The PBIMK Team Abdul Jalloh Glenn E. Kinard
SSC Laboratory Air Products and Chemicals
Norman F. Hill
Argonne National Laboratory Michael A. Janoski Alastair J. King
Irving Burton Associates Nexus Amplifier Solutions Corp.
M. Hiller
Babcock & Wilcox Andrew J. Jarabak Ian King
Westinghouse Electric Westinghouse Electric
Ron Hite
Babcock & Wilcox Kenneth Jennings Thomas B. W. Kirk
Modicon Argonne National Laboratory
Arvle F. Hix
Irving Burton Associates D. C. Johanson T. Kishi
Babcock & Wilcox Mycom Corporation
P. A. Hlasnicek
Babcock & Wilcox David H. Johnson Hiroyuki Kitagawa
Howden Compressors Inc. Mitsui & Company Ltd.
1229
Sherrie Kivlighn David Lambert Albert Luna
SSC Laboratory SSC Laboratory Luna Defense Systems Inc.
Mary Klorer Stefano Lami Ted Luskowitz
Allen-Bradley Company State University of New York Contemporary Physics
John Klusarits Michael J. Lamm Education Project
Acme Cryogenics Fermi Laboratory Rick Luther
Quintin W. Kneen Horace Lander SSC Laboratory
CBI Na-Con, Inc. Nippon Steel USA Inc. HylanLyon
Ted Kobel John J. Lapicola Facility Development Associates
Process Systems Avo Multi AmplIDM Charles Lyraud
International Inc. Eric Larson CEN Saclay
Horst Koetting Fermi Laboratory Colin J. Mac Farlane
Babcock & Wilcox Amy Lautenschlager Ontario International Corp.
M. Kohama Fermi Laboratory Jeff Mac Kinnon
Mycom Corporation Charles Laverick Office of Congressman Barton
Ted Koizumi A. Gordon Mac Lennon
Furukawa Electric JoLay Canadian Consulate General
Edward W. Kolb Judge Fite Co. Inc. Realtors John Mackay
UC Santa Barbara Michael Lee SSC Laboratory
David Koopman B.C. Trade Development Corp. R.MichaelMadden
Metallized Products Daniel R. Lehman Advanced Photonix
Serguei Kopylov U.S. Department of Energy James Maddocks
University of Wisconsin Jarkko Lehto SSC Laboratory
Wayne Koska Outokumpu Superconductors Oy James F. Maguire
Fermi Laboratory Thomas A. Leiser Applied Engineering
TedKozman Trammell Crow Company Technologies Ltd.
SSC Laboratory Louis J. Lestochi Ted Mallick
Max Kozyrczak Air Products and Chemicals Westinghouse Electric
Process Systems Eddie Leung Robert J. Malnar
International Inc. General Dynamics Space SSC Laboratory
David Kramer Systems Division Wendell Mansel
Inside Energy Kent K. Leung U.S. Department of Energy
MarkKrefta SSC Laboratory D.Mapes
Westinghouse Electric Michael Levin Babcock & Wilcox
T. Scott Kreilick SSC Laboratory Gerhard Mara
Hudson International Conductors Pejung Li Elin Energieanwendung
Ken Kreinbrink Institute of Ceramics for Bill Marancik
CVI Incorporated SSC Laboratory Oxford Superconducting
Detlef Krischel Joe Lichtenwalner Joe Marcus
SiemensAG Oxford Superconducting Tech Martin Marietta Astronautics
Dave Krudwig Dennis Lieurance Finley W. Markley
Westinghouse Electric General Dynamics Space Fermi Laboratory
George A. Kuhr Systems Division Bob Marsh
Hughes Aircraft Company Rauno Likamaa Teledyne Wah Chang Albany
H. Kunitomo Outokumpu Superconductors Oy Thomas Matuszak
Miti Steve Lindmayer Trent Tube
Ryuichi Kurumisawa B.C. Trade Development Corp. Michael May
C. Itoh Pipe & Tube Melvin Lindner Phelps Dodge Magnet
J. Kuykendall Brookhaven National Laboratory Wire Company
Nissei Sangyo America Ltd. Arie Lipski Lewis May
Hoang GiaKy Fermi Laboratory CRSS Architects Inc.
AISA Said Lounis Peter O. Mazur
Andrew Kytasty C.E.A. Fermi Laboratory
General Dynamics Space Diana (Dino) Lowrey Charles Mazzacato
Systems Division EG&G Idaho Inc. Markham Electric Ltd.
Elizabeth J. La Rosa Jill Ludwigsen James McAdams
Digital Equipment Corp. Nichols Research Corp. Fermi Laboratory
1230
Michael McAshan Chiaki Mizuno Greg Naumovich
SSC Laboratory Mitsui & Company (USA) Inc. Everson Electric Co.
Randy McConeghy Martin Moffat Ron Naventi
SSC Laboratory Teledyne Metal Forming Bechtel
John B. McDaniel Dieter Moller Dick Neal
Siemens Industrial Automation Leybold Ag, Cologne General Dynamics Space
Henry A. Mollicone Systems Division
John R. McDonald
The PBIMK Team Stellar Industries Leonard Neiman
James E. Monsees General Dynamics Space
Kim A. McDonald
Chronicle of Higher Education The PBIMK Team Systems Division
Glen E. McIntosh Les Montford Robert P. Nero

Kaiser Aerotech Howden Compressors Inc.
Cryogenic Tech. Servo
A. D. McInturff Suhas MookeIjee Louis Newsome
Fermi Laboratory ABB Technology Co. Memory Engineering
Debra Law McKeown John S. Moore Renee A. Nichols

1NRLC Illinois Dept Energy & Ceramaseal
R. D. McKinney Natural Resources Wolfgang Nick
Teledyne Metal Forming William Moore SiemensAG
Mike McKinney Office of Congressman Chapman Thomas H. Nicol
TUElectric John J. Morena Fermi Laboratory
Lawrence M. Megan ACMLE Ralph C. Niemann
Teledyne Pittsburgh Tool Steel Gerry H. Morgan Argonne National Laboratory
Thomas Mego Brookhaven National Laboratory MikaNiemi

Keithley Instruments Ira Lon Morgan Helsinki University, Seft
Manuel Mendez IDM Corp. Masaru Nishikawa

Cinvestav Hiroaki Morita University of Tokyo
Andrew Mendo Mitsubishi Heavy Industries Ltd. Y oshio Nishioka
Hamamatsu Corporation Vicki Lynn Morris Mitsubishi Heavy Industries
Thomas Mensah SCI Fred Nobrega
Supercond Technology Paul Morris SSC Laboratory
Z. M. Meratla Teledyne Sc Carl Noggle
CDS Research Ltd. Herbert Moryson Fiore Industries Inc.
John Metzler Siemens AG K. Nohara
U.S. Department of Energy David Moss Kawasaki Steel Corp.
Bill M. Miera Pacific Levitation Systems Inc. EddNoien
Fiore Industries Inc. Bill Mote Office of Congressman Bevill
W. O. Miller Mote & Associates Jim Nonnie

Los Alamos National Laboratory Pascal Mougenot Flexonics Inc.
Mark Miller Alsthom Intermagnetics John Nonte
Keller PSI Inc. G. T. Mulholland SSC Laboratory
John Miller SSC Laboratory Leonard Norman
National High Magnetic Thiru Muliya SSC Laboratory
Field Laboratory SSC Laboratory Joyce North
Joseph Miller Tim Muller Lancaster Chamber of Commerce
U. S. Department of Energy Westinghouse Electric Mary Ann Novak
Lisa Mills Pamela J. Mundo Parsons Brinckerhoff Quade
Hi Tech Mfg. Midlothian Chamber Sean 0' Bar
Masaharu Minami of Commerce Westinghouse Electric
Mitsubishi Heavy Industries Ltd. Joseph F. Muratore Patrick 0' Keefe
William Mischler Brookhaven National Laboratory Bath Iron Works Corp.
Intermagnetics General David Nahmias PhilO' Larey
John Mitchell Air Products and Chemicals Teledyne Wah Chang Albany
Martin Marietta Koichi Naruse Joseph 0' Loughlin
Alex Miyatake Mitsui & Co. (USA) Inc. Rector Associates International
Furukawa Electric Uriel Nauenberg Jim Ochsner
University of Colorado E. I. duPont
1231
Richard H. Oeler Lee Patterson Joseph Raple
Air Products and Chemicals General Dynamics Space U.S. General Accounting Office
Jeff Ogawa Systems Division John R. Rees
ICF Kaiser Engineers Inc. Philippe Paulmier SSC Laboratory
lGnyaOgawa Air Liquide Canada Ltee. Margareta Rehak:
The Furukawa Electric Tim Pawlak: Brookhaven National Laboratory
Torn Ogitsu Swanson Analysis Systems Inc. Larry A. Rehn
SSC Laboratory William W. Pearson Interfet Corp.
Tsuguo Ohkuma EG&GWasc John B. Reid
IHl Scott D. Peck United Magnet Technologies
IsamuOhno General Dynamics Space Ann Reifenberg
IHl Systems Division Dallas Morning News
Lori Okay James B. Peeples Karl Rencher
SSC Laboratory CVI Incorporated Lawrence Berkeley Laboratory
Scott Olexa Arlin M. Pennington Keith Rensberger
Westinghouse Electric SSC Laboratory Cryenco
Joseph A. Orgeron Yvette M. Perez C. M. Rey
University of Texas at Dallas CERN Courier Babcock: & Wilcox
Darryl Orris Lynn Peugh Richard L. Rhodenizer
Fermi Laboratory North Texas Commission Intermagnetics General Corp.
Charles Orsak Aihud Pevsner James K. Rice
Navarro College The Johns Hopkins University Sandia National Laboratories
David Osborne James Pfefferle James H. Richardson
Boston University General Physics Services Corp. SSC Laboratory
M. Oshima Peter C. Piazza Jack: Richman
Hitachi Ltd. Lafayette Advertiser Spectrum Eng. Corp. Ltd.
Aubie Oslin James G. Pierce ReidRihel
The PBIMK Team CVI Incorporated Fermi Laboratory
Gene Oster Armand Pinet Chris Riley
Advanced Technology Inc. Thevenet-Oerjounie Vector Fields Inc.
Johan A. Overweg Dan Pohlabel Edward C. Rinck:
Siemens AG Process Equipment Co. Weldaloy Products Company
Chuck: Owen Douglas A. Pollock: Rick Riney
Westinghouse Electric SSC Laboratory ,
Martin Marietta Astronautics
Satoshi Ozaki John Pope Michael Riordan
Brookhaven National Laboratory The Times-Picayune URA
N.Ozaki Stan Popp Chris Rista
Hitachi Ltd. InveIpOwer Saes Pure Gas Inc.
Tony L. Ozuna Nick Porschek Dan Rivenbark
New Bedford Panoramex Corp. P.E.C. Magnet System Division SSC Laboratory
Peter Papakostantinu Robert L. Powell George Robertson
Papa & Associates Process Systems SSC Laboratory
International Inc.
H. Parish Charles Robinson
CVI Incorporated Robert J. Powers Open Systems Resources
Powers Assoc. Inc.
Joe Parish William F. Robotham
E. I. duPont Ronald P. Pratt Fermi Laboratory
Weldaloy Products Co.
RekhaPatel Peter Roe
General Dynamics Space J. C. Puippe
Watlan Inc.
Systems Division W. Fluhmann Ag
Jimmy Rogers
Pam Patrick Phillip Radusewicz SSC Laboratory
Alcoa SSC Laboratory
Daniel Rogers
Ray Patrylak: Myrna Ramos
Fermi Laboratory
Rogers Corp. Hewlett Packard
E. Parke Rohrer
John Patterson Leida Randall
OR Tech SSC Laboratory
Livio Rosai
Scott Rankin Saes Getters S.P.A.
Andyne Computing
1232
Jeff Rosendahl Steve Schnurer Charles W. Simpson
Energy Research/Science, Space, G.E. Medical Systems Morrison Knudsen Corp.
& Technology Susan Schoenung Richard E. Sims
Carl H. Rosner W. J. Schafer Associates SSC Laboratory
Intermagnetics General Corp. Bill Schumacher Feliciano Sanchez Sinencio
Jim Rossi Armco Cinvestav
Kyser Company H. Donald Schwartz Sharad K. Singh
H. Rougeot Cabot Performance Materials Westinghouse Electric
G.E. Research & Development Bill Schwenterly G. Sintchak
William G. Rueb SSC Laboratory Brookhaven National Laboratory
Morrison Knudsen Corp. Roy F. Schwitters Mark J. Sisinyak
Camille Rush SSC Laboratory Brown & Root Inc.
SSC Laboratory VicScopack Edward J. Siskin
Patrick Russell Cryogenic Ind. SSC Laboratory
Trent Tube Frank Scrima BaIbara Sizemore
R. Ryan Katie Group Fermi Laboratory
Babcock & Wilcox John Scudiere Nils Sjulander
Nobuyuki Sadakata Oxford Superconducting Silvex
Fujikura Ltd. Technology John Skaritka
Shigeo Saito Armin Senn SSC Laboratory
Sumitomo Electric Industries Sulzer Brothers Ltd. Robert F. Smellie
Shuji Sakai Jeffrey M. Seuntjens SSC Laboratory
Hitachi Cable Ltd. SSC Laboratory Christopher C. Smith
G. Salvador Howard Shaffer, Jr. Dynapower Corporation
General Dynamics Space Westinghouse Electric Dave Smith
Systems Division Sol Shapiro Westinghouse Electric
Nicholas P. Samios SSC Laboratory John R. Smith
Brookhaven National Laboratory Douglas Sharp Memory Engineering
W. Sampson General Physics Services Corp. D. K. Smith
Brookhaven National Laboratory Rick Sharp Armco Adv. Mati. Co.
James R. Sanford EIma Greg Snitchler
SSC Laboratory Jerry D. Shay SSC Laboratory
Phillip Sanger Km - Kabelmetal America Volker Soergel
SSC Laboratory Phil Shelley DESY
S. Sato EO&G Philip W. Spaeth
Kawasaki Steel Corp. Ryan Sherman Air Products and Chemicals
Yoshihisa Sato SSC Laboratory Kyle Spann
Toshiba Corp. Peter Shields Stellar Industries
Guy J. Scango Janis Research L. Clay Spencer
U.S. Department of Energy Takao Shiota Brown & Root Inc.
Terry Scanlan FTA David Stankes
AircoGases Dan Shipman Lydall Manuing
Giuseppe Scarfi Contract Mfg. Mike Stanko
Ansaldo Componenti Nanda K. Shivapuja Westinghouse Electric
Richard Schafer SSC Laboratory Jerry Staub
Process Equipment Co. David Shoaf General Dynamics
Sam Scheer Westinghouse Electric Elizabeth Stefanski
Westinghouse Electric Mike H. Shrubsole SSC Laboratory
Jack Scheider Murdock Inc.
C. A. Steinman
Lockheed Austin Division Quansheng Shu General Dynamics Space
Robert I. Schermer SSC Laboratory Systems Division
SSC Laboratory Bruce G. Shuster A. John Stekly
Leonard Schieber SSC Laboratory Intermagnetics General Corp.
Brookhaven National Laboratory Gottfried Simon Kirk Stifle
B ill Schiesser Elin Energieanwendung SSC Laboratory
SSC Laboratory
1233
William Stokes Jukka Teras Voitto Vanhatalo
Brookhaven National Laboratory Cerntech Ltd. Outokumpu Superconductors Oy
Jay Stone Juhani Teuho Gregorio Vargas
U.S. Department of Energy Outokumpu Superconductors Oy Corporation Mexicana de Inv.
Philip C. Stover Roberto Y. Than en Materiales
Nichols Research Corp. SSC Laboratory Alex P. Varghese
Robert L. Thews Gardner Cryogenics
Elford A. Stover Jr.
Bath Iron Works Corp. University of Arizona Mikhail Vereschagin
D. L. Thoman JV"SKIF"
Jim Strait
Fermi Laboratory Burle Industries Inc. Michael Vitale
Bruce P. Strauss Randell Thompson Surnitomo Corporation
Powers Associates Inc. The PBIMK Team of America
Alan J. Street Tim Thompson R. J. Vopelak

Oxford Instruments Los Alamos National Laboratory Babcock & Wilcox
Robin Stringfellow John C. Thome Eric Vrsansky

TNRLC SSC Laboratory SSC Laboratory
Kenneth W. Struve Dennis Threatt UdoWagner
Mission Research Corp. SSC Laboratory Sulzer Brothers Ltd.
Linda Stuntz Will Thur Masayoshi Walee

U.S. Department of Energy Lawrence Berlceley Laboratory Fermi Laboratory
Takayasu Sugiyama Tim Thurston Yukichi Wakita
Nippon Steel USA SSC Laboratory Mitsui & Co. (USA) Inc.
Hilary Sullivan John E. Tigner Ian J. Walker
U.S. General Accounting Office Science Application GMW Associates
Melinda Sumurdy International Corp. Ron Wallace

Ontario Goverument - Dallas John S. Toll Flexonics Inc.
Wally SundeIman Universities Research Assn. William Wallenmeyer
Westinghouse Electric Jean Christian Toussaint SURA
John Supancic C.E.A. Gully Walter
Von Roll Isola Inc. George Trilling Tempel Steel Company
Toru Suzuki Lawrence Berlceley Laboratory P. Wanderer
Nippon Steel Corp. Alvin W. Trivelpiece Brookhaven National Laboratory
Shinichi Suzuki Oak Ridge National Laboratory Don Warner
Shimizu Corp. Hank Trujillo SSC Laboratory
Roger Swensrud Fiore Industries Inc. R. Watanabe
Westinghouse Electric Gary Tuttle Toho Titanium
John R. Swentzel SSC Laboratory Jerry M. Watson
Phelps Dodge Specialty Cu Norihiko Ujiie SSC Laboratory
Vladimir Sytchev Nat'l Laboratory for High J. Waynert
IHEP Energy Physics (KEK) Babcock & Wilcox
K. Tahara K. Ukena C. Weber
Hitachi Ltd. Nissei Sangyo America Ltd. Babcock & Wilcox
K. Takagi Robert C. Ulrich William Weber
Hitachi Ltd. The PBIMK Team MPI International Inc.
Greg Tanimoto John B. Urbin J. G. Weisend
C. !toh Pipe & Tube SSC Laboratory SSC Laboratory
Claude D. Tapley George J. Urich Norman Wells
Km-Kabelmetal America Everson Electric Co. SSC Laboratory
Laurent Tavian Anthony N. Vageonis David Wells
CERN Precision Defense Services Inc. Rotoflow Corporation
Clyde Taylor Hugo Van Den Bergh John Welte
Lawrence Berkeley Laboratory General Dynamics Space Analogic Corporation
Systems Division Milt Werkema
David Taylor
AnncoInc. W. E. Van Eman Sheldahl Inc.
U. S. Steel Division USX Brent West
Lawrence Teel
Kinetic Systems Corp. Rob Van Weelderen Cryogenic Consultants Inc.
CERN
1234
James E. West Kathryn Wirth John B. Yasinsky
Cryogenic ConsultuIts Inc. The PBIMK Team Westinghouse Power Systems
MalkWest Neil E. Wiseman James H. Yeck
Lawrence Belkeley Laboratory General Dynamics Space U.S. Department of Energy
Jim West Systems Division Richard C. York
ffiM Zachary Wolf SSC Laboratory
JeffWestem SSC Laboratory Y. Yoshida
SSC Laboratory JohnWollan Hitachi Ltd.
Andrew White Martin Marietta George P. Yost
University of Texas at Arlington Donald E. Wolman SSC Laboratory
Mike Whiteside SSC Laboratory AdnanYucel
Telesis High Vacuum Inc. Jack Woltz SSC Laboratory
Jon Wikstrom SSC Laboratory Frank J. Zahradnik
MUE James V. Worth Ebara Technologies
Colin D. Wilburn Oxford Superconducting Joseph A. Zawistowski
Micron Semiconductor Inc. Technology Ability Engineering Technology
E. H. Willen Steve Wright Bruce Zeitlin
Brookhaven National Laboratory Kurt J. Lesker Co. IGC Advanced Superconductors
Lloyd Williams Jiing-Liang Wu Arnulfo Zepeda
CERN Westinghouse Electric Cinvestav
Lola Williamson K.C.Wu Burt Zhang
SSC Laboratory Brookhaven National Laboratory SSC Laboratory
Bob Williamson Russell Wylie Ren-Yuan Zhu
SSC Laboratory SSC Laboratory CalTech
Ron Willig DongXiao Bruno O. Ziegler
EML Research Florida State University Sulzer Brothers Ltd.
Kenneth F. Wilson XunXu George O. Zimmerman
CVI Incorporated SSC Laboratory Supersolder Technologies Inc.
Larry Wilson Rynji Yamada Art Zinszer
Douglas Engineering Co. Fermi Laboratory General Dynamics Space
Philip W. Winkler Dan Yamamura Systems Division
Air Products and Chemicals Teledyne Japan K.K.
Phil Winnie K. Yamanishi
Westinghouse Electric Kawasho Corp.
1235
Author Index
Abramovich, S ................................. .207 Capone, D., II ........ 25,491, 661, 669, 685
Adovasio, J........................................ 85 Carcagno, R ................................635,897
Afanasiev, O ..................................... 867 Carr, W., Jr. ............................. 695, 1025
Agarwal, K. ..................................... .219 Carson, 1.. ........... 113, 137, 251, 289, 365,
Akimov, Y ..................................... 1097 395,435,517,527,757
Albee, M ..........................................591 Caspi, S ........................................... 243
Anderson, D .................................... 1201 Chen, G ...........................................809
Anerella, M .......... .113, 137, 535, 551, 559 Chervonekis, A ................................ 1103
Arden, C ......................................... 1143 Chester, T .......................................... 99
Arrendale, H .................................... 1001 Christianson, 0 ................................. 739
Assell, D .......................................... 395 Christoph, M ....................................591
Ayai, N ........................................... .499 Christopherson, D .......... 25, 491, 669, 685
Ayer, F. ........................................... 151 Chromec, B ....................................... 199
Baker, E ...........................................643 Chuboy, H ....................................... .327
Ball, P ........................................... 1015 Cihangir, S ..................................... 1201
Baranko, G ....................................... .591 Clark, F .............................661, 669, 685
Bardos, V ...................................661,669 Clark, G ......................................... 1127
Baritchi, D ........................................ 321 Clerc, R. .......................................... 199
Barkell, J.......................................... 159 Coghill, C., III ..................................275
Bastian, S ....................................... 1039 Coleman, E .......................................661
Bein, D .................................... .225, 677 Coles, M .................................. .467, 905
Bell, W ............................................ 619 Collings, E ....................................... 923
Belz, J ............................................. .591 Connolly, D ...................................... 395
Bensiek, W ................................981, 985 Coombes, R ........................ 1l3, l37, 551
Beoddy, J..........................................967 Cooper, H ....................................... 1025
Berk, E ............................................. 151 Cooper, W ......................................... 13
Berry, W ............................................ 99 Corden, M ...................................... 1209
Billingsly, A. .................................... 967 Coming, B ......................................... 79
Bleadon, M ......... 13,113,137, 251,365,435 Cottingham, J ....... 113, 137, 535, 551, 559
Blecher, L. ........................................ 259 Cowan, P......................................... 365
Blessing, M ......................................517 Craker, P........................................ 1007
Bliss, D .......................................... 1167 Croll, B ............................................219
Boer, J.................................... 981, 1187 Curbow, J ......................................... 841
Bolotin, I. .........................................881 Cushman, P.................................... 1217
Boroski, W .......................................235 Daly, E. .................................... 313,913
Bossert, R ............ 113, 137, 251, 289, 365, Daniel, G .......................................... 591
435,461,517,527,1177 Daoudi, M ........................................591
Brandt, 1.. ................................ 517, 1177 Davidson, M ........................... .461, 1039
Broomer, B ....................................... 591 Dederer, 1.. ........................................739
Brouillette, D .................................... 889 Dejong, T .......................................... l3
Brown, G .................................. .475, 491 Delagi, N ..........................................873
Budagov, J...................................... 1201 Delchamps, S .............. I13, l37, 251,289,
Bush, T. ............................. 113, 137, 551 365,435,517,527
Butalla, M ...................................... 1119 Demko, J ..........................................635
Buyce, R. .......................................... 85 Demortier, L. .................................. 1201
Cahill, J........................................... 395 Devred, A. ......................... 113, l37,251,
Cantor, B ..........................................967 365,435,551
1237
Dickey, C ......................................... 859 Gtisewell, D ...................................... 199
DiMarco, J ........................ .113, 137,251, Guss, C ............................................ 591
365, 435, 551 Haasjoki, P ........................................ 49
Dransfield, G .....................................603 Haggard, J .................................. 517, 757
Dreksler, M .......................................627 Hanft, R ............... 113, 137,251,365,435
Dryer, J ........................................... .467 Hanlon, J ..........................................603
Dubasov, A ....................................... 867 Hannaford, c. .............................. 25, 669
Dugan, G .......................................... 825 Hanson, G ........................................ 159
Dwyer, S ........................................ 1159 Hatfield, D ...................................... 1089
Eartly, D .......................................... 175 Hedderick, R ...................................... 711
Elliott, 1.. ....................................... 1167 Heger, P...........................................289
Enbaev, A .........................................881 Higinbotham, W ................................517
Erdmann, M ...................................... 661 Hill, N ............................................. 183
Erdos, E ...........................................591 Hiller, M .................................. .281, 297
Erdt, W ............................................ 199 Hlasnicek, P ......................................967
Erkolahti, T ....................................... 49 Hoecker, D ................................. 159, 183
Erochin, A. ................................867, 881 Hoffman, D ....................................... 517
Ewald, K ................................... 289, 517 Hoffman, J ...................................... 1177
Eyssa, Y ...........................................305 Hordubay, T. ..................................... 183
Fales, P ............................................. 13 Horton, B .........................................873
Fanourakis, G .................................. 1201 Huang, X ..........................................305
Farrell, R ..........................................259 Ii, H ...............................................939
Fechteler, H ............................. 981, 1187 Ikeda, H ............................................ 583
Fernandez, J ....................................... 57 Ikeda, M ....................................939, 943
Ferrell, J ........................................... 841 Inaba, J ............................................931
Flora, R ...........................................859 Inoue, 1. ....................................939, 943
Franks, D ......................................... 259 Iosue, M ......................................... 1217
Freeman, M ...................................... 191 Ivey, J ............................ 873, 1007, 1015
Frost, D ............................................ 57 Iwaki, G ......................................... 1081
Fulton, H .................................. 289,517 Jaffery, T ..........................................527
Funk, L ............................................ 833 Jain, A. ................ 113, 137, 535, 551, 559
Fuse, M .......................................... .499 Jalloh, A. ........................... 313,321,957
Fuzesy, R .........................................467 J arabak, A ............................................ 1
Gadsden, T ......................................... 21 Jayakumar, R. ...................................695
Gamble, M ................................603, 611 Jenkens, L. .......................................... 5
Ganetis, G ............ 113, 137, 535, 551, 559 Jensen, . R .......................................... 517
Ganni, V ................................... 207, 453 Johanson, D ...................................... 513
Garbarini, J., Jr. ...................... 1007, 1015 Johnson, B ...................................... 1217
Garber, W ............. 113, 137, 535, 551, 559 Johnson, P ........................................ 175
Gathright, T ..................................... .467 Jonas, P ...........................................259
Gertsev, K. .................................... 1103 Jones, J ............................................ 757
Ghosh, A ........................... 113, 137,535, Kahn, S ................ 113, 137,535,551,559
551, 559, 765 Kallas, N ..........................................957
Gilbertson, A. .................................. .461 Kamata, K ...................................... 1081
Gilchrist, E ...................................... .219 Kanithi, H ........................................ .41
Giokaris, N ..................................... 1201 Kazmin, B ........................................ 867
Gonczy, E ........................................ .461 Kelley, J ...........................................967
Goodzeit, C .................. l13, 137, 535, 551 Kelly, E ............... 113, 137, 535, 551, 559
Gordon, M ........................................517 Kerby, J ...........................................289
Gorman, J ......................................... 151 Khazins, D ..................................... 1201
Gossler, E ........................................ .275 Kihlken, D ....................................... .461
Goulianos, K. .................................. 1201 King, D ........................................... .219
Gourlay, S ............ 113, 137, 251, 365,435 Kinney, W ........................................ 527
Govignon, J ...................................... 151 Kishi, T ...........................................627
Graham, S ........................... 25, 669, 677 Kivlighn, S........................................ 21
Greene, A. ............ 113, 137, 535, 551, 559 Kobel, T .................................... 191, 619
Gregory, E ........................................923 Kobliska, G .................................... 1177
Gribkov, V ..................................... 1103 Koganeya, M ............................... 33, 499
Gridasov, V ................................ 867,881 Kondo, T. .........................................943
Grtineberg, H .....................................981 Konschak, R. .................................. 1187
Grut, K ........................................... .281 Kopylov, S.......................................975
Guarino, V ........................................ 183 Koska, W ............ 113, 137, 251, 289, 365,
Gupta, R ........................... 113, 137,535, 419,435,461,517,527
551, 559, 773 Kovachev, V .....................................695
Gurevich, A. .....................................975 Kozyrczak, M ....................................619
Gurol, H .................................. 651, 1167 Krefta, M ................... 327, 335, 575, 1167
1238
Kreutz, R ..........................................981 Myznikov, K ..................................... 867
Krischel, D .............................. 981, 1187 Nagata, M .................................... 33,499
Kuchnir, M ........... 113, 137, 251, 365, 435 Nab, W ....................... 113,137,365,435
Kupiszewski, T ..................................993 Nakazawa, H ................................... 1081
Kumaev, O .......................................867 Nasiatka, J ........................................ 183
Kurtcuoglu, K ................................... 199 Nauenberg, U. .................................. .591
Kuzminski, J ...................... 113, 137,251, Neiman, L. .......................................505
365, 435, 551 Nelson L. .........................................757
Kytasty, A ...................................... 1001 Nguyen, L. .............................. 651, 1167
Lambert, J ...................................... 1007 Nicol, T .................................... 235, 747
Lamm, M .......................... 113, 137,251, Nishiike, U ..................................... 1065
365,435,527 Nohara, K. ............................. 1055, 1065
Larson, W .......................................... 57 Nonte, J ......................................... 1137
Lautenschlager, A ............................. 1031 Norman, L. ..................................... 1111
Leedy, R ........................................... 873 O'Connell, C ...................................... 13
Lehto, J ............................................. 49 O'Connor, C ...................................... 13
Leung, K ................................... 159, 343 Ogawa, K. ........................................ 939
Li, P ................................................ 801 Ogitsu, T ............................ I13, 137,251,
Li, W ..................................... .467, 1007 365, 435, 551
Li, X ...................................... 873, 1015 Ogle, C .......................................... 1001
Liikamaa, R ....................................... 49 Ogren, H .......................................... 159
Lindner, M ...................................... I021 Ohmatsu, M ....................................... 33
Lindsay, R ......................................... 13 Okumurra, K. ......................... 1055, 1065
Lipski, A ........................................ 1177 Orgeron, J.. ..................................... 1193
Liscov, V ....................................... 1103 Orris, D ................ 113, 137, 251, 365, 435
Liu, H ..............................................923 Orsak, C .............................................. 5
Lowery, D ......................................... 93 Oslin, A ............................................ 69
Liihlein, K. ....................................... 199 Ostwald, F. ..................................... 1187
Madden, M ...................................... 1217 Owen, C ...........................................723
Madura, D ....................................... 1167 Ozelis, J ............................. 113, 365, 527
Malamud, E ....................................... 13 Ozeryansky, G ...................................923
Mallick, G ............................... 695, 1025 Packer, M ......................................... 517
Malone, K ........................................723 Palounek, A. ..............................603, 611
Man, S .............................................809 Para, A ........................................... 1201
Mantsch, P ................................ 113, 137 Pare, G ............................................. 967
Marken, K, Jr ....................................923 Parker, J., Jr ..................................... .335
Markley, F ............... 351, 359, 1031, 1039 Pasquinelli, R ..................................... 13
Marler, C ........................................... 93 Patel, R .......................................... 1149
Marschik, D ...................................... 159 Pavlik, D .................................. .327, 849
Marshall, B .......................................873 Pawlak, S ........................................ .395
Matsumoto, K ...................................703 Pearson, W ........................................ 99
Mayhall, J ........................................ 159 Pedler, D ........................................... 85
Mazur, P .............. 113,137,251,365,435 Pellissier, P ......................................467
McAdams, J ............................. 359, 1031 Pena, D ............................................475
McAshan, M ..............................207, 635 Peshkin, M ........................................ 13
McGlohen, P ........................................ 5 Peterson, T ................................ 113, 137
McInturff, A ...................................... 859 Pidcoe, S ........................................ 1075
Meguro, S ........................................939 Pletzer, R .........................................259
Mehle, G ........................................ 1167 Pollock, D .................................. 25,491
Mermikides, M .............. , ................. 1209 Potter, J ........................................... 551
Michaud, F .......................................603 Prima, M ....................................867,881
Mikumo, A ........................................ 33 Prodell, A. ............ 113, 137, 535, 551, 559
Miller, K ......................................... .365 Puglisi, M ......................... 113, 137, 251,
Miller, W ..................................603,611 365,435, 551
Mischler, W ......................................259 Puippe, J ......................................... .381
Mokhtarani, A .................................. .251 Radusewicz, P .............. 113, 137,535,551
Monsees, J ...................................... 1127 Randall, R ......................................... 57
Morena, J ....................................... 1047 Reddy, P ......................................... I089
Morgan, G ............ 113, 137, 535, 551, 559 Reed, T., Jr .......................................669
Morgan, M ..................................... 1201 Rehak, M .......................... 113, 137,391,
Mori, T .......................................... 1065 535, 551, 559
Moriai, H ....................................... 1081 Rehn, L ............................................ 817
Moryson, H ......................................981 Reid, R ............................................603
Mulhall, S...................................... 1021 Ren, S ............................................. 809
Muratore, J ........... 113, 137, 535, 551, 559 Rey, C .............................................967
Murray, F. ........................................259 Reynolds, M ..................................... 757
1239
Rihel, R ................................... .395, 517 Swensrud, R ............................... 159, 183
Rippinger, D .....................................757 Syromyatnikov, 1.. .................. 1007, 1089
Rista, C ...........................................643 Sytchev, V ................................. 867, 881
Roach, J ................................... .403, 731 Sytin, A ...........................................867
Robatzek, W ..................................... 395 Szawlowski, M ................................ 1217
Roberts, D ........................................ 817 Takagi, R .........................................939
Robins, K .........................................765 Takai, K ...........................................499
Robinson, S. L. .................................873 Tanaka, Y .........................................703
Robotham, W ..................... .395, 517, 757 Tassotto, G...................................... .517
Rogers, D ........................................ .351 Taylor, C ................................... 137, 243
Rogers, J ..........................................841 Terami, H ....................................... I065
Rohrer, E .............. I13, 137, 535, 551, 559 Teuho, J ............................................ 49
Royet, J ........................................... 137 Thirumaleshwar, M ............................207
Rusack, R ....................................... 1217 Thomas, R ............ 113, 137,535, 551, 559
Rust, D ............................................ 159 Thompson, P ........ 113, 137,535, 551, 559
Ryan, B .....................................981, 985 Thompson, T .............................603, 611
Saito, S ...................................... 33,499 Threatt, D .........................................711
Sakai, S ......................................... 1081 Tichin, G ..........................................867
SaltInarsh, C ................................... 1007 Tinsley, A ........................................ 517
Sampson, W ...................... 113, 137, 535, Tompkins, J ................ 113, 137,251, 365,
551, 559, 765 435, 551, 873
Sanger, P ............................ 113, 137,551 Toms, J ................................... 695, 1025
Sasaki, H ............................... 1055, 1065 Trekell, R. .............................. .475, 1089
Sashida, T ................................... 33, 499 Trombly-Freytag, S ............................. 13
Sato, S ........................................... 1055 Troupe, B .........................................661
Saxer, W ......................................... .381 Turuer, J .................................... 113, 137
Scanlan, R ........................................ 137 Ujiie, N ............................................583
Scheidemantle, A ... ;...........................207 Ullmark, R ...................................... .395
Scherbarth, D .................................... 183 Unno, Y ...........................................583
Schermer, R ........................ 113, 137, 551 Valaris, P ...................................... .... 41
Schieber, L. ......................................411 van den Bergh, H .........................427, 651
Schiesser, W ..............................635, 897 VandeBogart, J ................................... 513
Schmitz, E ........................................395 Vanhatalo, V ...................................... 49
Schoo, C ..........................................235 Vasiliev, L.. ...............................867, 881
Schultz, G........................................591 Veselov, 0 ........................................ 867
Schwenterly, S ..................................619 Viola, R .......................................... .313
Seuntjens, J ........... 25, 491, 661, 669, 685 Volkov, A. ..................................... 1201
Shiga, S ...........................................939 Vrsansky, E .................................... 1159
Shih, H ............................................897 Wagner, U ........................................ 199
Shu, Q........................................... I089 Wake, M .................... 113, 137, 251, 289,
Shutt, R. .............. 113, 137, 535, 551, 559 365,435,517, 527
Sims, R. ............. .395, 419, 461, 517, 527 Wallis, D .......................................... 859
Singh, S ...........................................739 Wanderer, P.................. 1l3, 137,443,451,
Sizemore, B .................................... 1031 535, 551, 559
Skweres, T ........................................395 Wandesforde, A. .................................243
Smathers, J .......................................591 Warner,D ......................................... 711
Smith, B ..........................................517 Waynert, J ......................... .281, 297, 305
Smith, D ..........................................517 Weisend, J., 11, ..................................859
Snitchler, G .............................. .343, 695 Wente, C .......................................... 591
Snow, G ............................................ 13 Western, 1.. .............................. 159, 1119
Solheim, N .......................................199 Whitson, M ..................................... .461
Sommer, F .............................. 981, 1187 Willen, E .......................... .113, 137,443,
Spann, K ........................................ I075 535, 551, 559
Spigo, G ..........................................113 Winkler, G ........................................199
Stanley, M., ....................................... 13 Winters, M .........................461, 517, 527
Stiening, R ....................................... 113 Wisne, J ............................................ 13
Strait, J ................ 113, 137,251, 289, 365 Wolf, Z ........................................... .467
435, 517, 527, 757 Wolman, D .......................................475
Stutzki, D .........................................281 Woloshun, K .....................................603
Sullivan, D ....................................... 151 Woltz, J ....................................475, 491
Sumption, M ....................................923 Wong, J ............................................ 57
Susai, K ...........................................939 Wood, R ......................................... 1159
Suzuki, K ....................................... 1065 Wu, J ........................................731, 739
Suzuki, T ..................................939, 943 Wu, K ..............................................483
Suzuki, Y ....................................... 1081 Xiao, D .......................................... 1209
Swenson, G ....................................... 57 Xiao, H ............................................809
1240
Xie, Y .............................................. 801 Zhang, B ......................................... .453
Xu, X .............................................. 897 Zhang, J ........................................... 809
Yamada, Y ......................................... 33 Zhao, Y ...................... 113, 137, 251, 365,
Yamamoto, A ....................................943 435, 551, 801
Yarygin, N ........................................867 Zheng, H ........................... l13, 137,251,
Yin, Z .............................................. 801 365, 435, 551
Yokota, M ......................................... 33 Zhu, R .............................................781
Yu, D ............................................... 57 Ziegler, B ......................................... 199
Yu, Y ............................... 113, 137,251, Zimmennan, J ................................. 1201
365,435,551 Zimmerman, W ................................ .517
Yiice!, A ................................: .......... 897 Zinszer, A. ...................................... 1075
Zalikhanov, B .................................. 1097 Zioek, H ....................................603, 611
Zbasnik, J ..................................225, 677 Zolotov, A ...................................... 1089
Zeitlin, B ........................................... 41 Zweibohmer, J ................................. ll77
1241
Subject Index
Accelerator, linear (LINAC) .................... 825-857 Anticryostat design ............................... 1094-1095

beam transport, low energy................. 833-835 Antiquark, see Quark
booster, low energy ............. 826-827,841-857 Archeological Institute of Amer-
block diagram of ....................................... 842 ica ............................................................. 94
design .......................... .... .................. 849-857 ASST, see Accelerator system
evaluation ....................... .................. 852-854 string test
power supply.................................... 849-857 Atom ................................................................. 16-17
cavity ................... 837-838, 843-846, 850-851 Avalanche photodiode ............... 1217, 1220-1221
design ....................................... 843, 849-857 Axis, magnetic, localization ................ 1103-1109
computer program for .......................... 327-334 by magneto-optic film ..................... 1103-1109
current regulator for ............................. 844-846 by stretched wire ......... ................................ 1103
drift tube specifications ....................... 837-838
energy booster Barium fluoride crystal ............................ 781-815
high .. .... ........ .... .... .... .... .... ........ .......... 828-829 absorption, ultraviolet and ......... 803, 806, 807
low .................................... 826-827,841-857 calorimeter ............................................. 781-799
medium ............................................. 827-828 performance ..... .... .... ......................... 783-788
energy storage inductors ..................... 849-857 elements and .......................................... 810-812
field, magnetic ............................. 419,827-834 ions of rare earths effects .................... 809-815
ion source .............................................. 833-835 radiation and ........................ 795-798, 801-815
magnet, see Dipole magnet as scintillator, fastest known ....................... 801
multipole calculations .......................... 327-334 Barrel calorimeter ..................................... 186-190
parameters ............................................. 825-831 element analysis, finite, and .............. 186-190
requirements ................ 825-831, 834, 841-842 Barrel toroid ...... 175,575-578, 1129, 1124-1125
system string test (ASST) .............. 25-32, 191, Beam-gas collision interaction ................ 898-899
468,475,479 energy deposition model.................... 898-899
program .................................................. 25-32 Beam tube .............................. 381-389,1171-1174
tuner bias power supply...................... 844-846 vacuum ................................................... 897-904
Accelerometer ................................... 314,914-915 Billet .................................................. 226-229,924
Acts of the U.S.A. Bolt, connecting stress ............................. 168-173
Antiquities Act of 1906 ..................... 85,93,94 Brahe, Tycho ......................................................... 14
Archeological and Historical Bureau of American Ethnology of
Conservation Act of 1974 ..................... 85 1879 ......................................................... 94
Archeological Resources Pro-
tection Act of 1979 ......................... 85,95
Historic Sites, Buildings and Cable, superconducting ............. 25-32, 115-117,
Antiquities Act of 1939 ........................ 85 447,499-504,517,536,
National Environmental Policy 552,562,663-666,669-
Act of 1966 ...................................... 85, 95 701,939-942,1081-1087
National Historical Presereva- hysteresis and ........................................ 695-701
tion Act of 1966 ........................ 85,93-97 insulation ..................... 222,405,427,429,518
Adsorption model...................................... 899-900 manufacture ........................................... 677-684
Alignment ................................. 175-181,467-468, performance ............................................... 25-32
471-481,505-512 properties .................................... 672-674, 1085
Aluminum, high-strength ......................... 943-955 vendors, eight ........................................ 661-668
Anthropological Society of Wash- Calorimeter ......................... 183-190, 1025-1029,
ington, D.C. ............................................ 94 1097-1101, 1201-1207
1243
Calorimeter (cont' d), see also Barium Computer
fluoride calorimeter clock ................................................................ 176
Calorimetry.......................... ........ .... ...... 1217-1224 data acquisition systems ...................... 461-466
photomultiplier tube for .................. 1217-1224 program for high-precision
Carbon dating .. .... .... .... ............................ ............. 20 magnetic field ............................. 327-334
Carbon dioxide .......................................... 270-271 Concrete slab ............................................. 168-169
Carnot work ............................................... 635, 640 Conductor .............................. 33-47 see also NbTi
Central tracker detector (CTD) ............... 159-166 superconductor
Cerenkov counter ...................................... 591-602 Contraction, thermal ................................. 281-288
design ....... .... .... .... ............ ...................... 591-602 Cooldown ................................. 343-350, 553-554,
Circuit model, thermal, equivalent ......... 994-996 1021-1024,1171
Cliff palace (U.S.Indian) Copernicus ............................................................ 14
discovered in 1888 ................................ 94 Copper
Coefficient of variation .......................... 26,29, 30 electrodeposition .................................. 381-389
Coil, superconducting ............. 343-345,351-364, to superconductor ratio ........................ 451-452
395-402,435-437,443- wedge ..................................................... 518-519
449,451-452,471-474, Corrector magnet wire ............................ 58,61-66
518-522,528-536,539, COSI (Ohio Center of Science
552,561,562655,981, and Industry) .................................... 13-14
1001-1005, 1031-1036, Cradle design ............................................. 282-285
1047-1053,1160-1164, Creep
1171,1177-1192 model ...................................................... 361-364
cross-section .................................................. 277 stress relaxation and ..................................... 364
failure ............................................ 407-408, 422 studies .................................................... 359-364
geometry....................................... 352, 652-653 tensile ................................................. 1032-1036
insulation ............ 522, 1001-1005, 1031-1045 equation ......................................... 1039-1040
jellyroll technique ................................ 412-415 Cryogen .................................................. 1090-1092
-key structure ................................................. 293 transfer line ........................................... 453-460
mandrel ......................................... 397,400,411 Cryostat ..................................................... 207,209,
manufacturing .......................... 539, 1187-1192 235,286-287,321-
models .................................................... 293-294 325,453-460,747-
preloading .............................................. 437-438 750994-1000, 1171
pressure .................................................. 437-439 Curing mold for coil ................................. 519-520
RHIC (relativistic heavy ion Current
collider) and ................................ 412-416 alternating (AC) ................................ 1025-1029
ringer test ............................................... 419-425 direct (DC) ............................................ 731-738
scan for flaws ........................................ 229-232 superconducting .................. 739-746,926-928,
stress ................... 290,294, 305, 348-349, 530, 975-980, 1084-1085
540-546,555-556,569- Cyanate ester ..................................................... 1049
571, 1165
structure .......................................................... 654 Data acquisition, automated .................... 461-466
tomography ........................................... 229-231 Database system for magnet ................ 1015-1020
winding 411-412,417,518,981-983, Decapole variation with current ..... 141, 145, 146
1160-1161,1188-1192 Decay, radioactive .......................................... 19-20
Cooldown ............... 553-554 see also Cool-down Detector ................................ 943-955, 1209-1215
Cold mass ........................ 223,225-233,281-288, collaboration, solenoidal (SOC) ....... 151-181,
297,305-312,322,345, 611-618
427,429,433,506,538, solenoidal .......................................... 1119-1125
560,747,757,898,1159- Device, optical
1166, 1180, 1182 state of the art ....................................... 587-589
cross-section ........................ 306,429,538,560 Dewar facility, vertical............................. 967-973
Cold weld, see Weld Dipole magnet,
Collar ...................... 343-344,423-433, 437-438, superconducting .................. 33, 121-130,
440,528,1171,1172 133-134, 137-149, 167-
Collaring .................................... 522-523, 539-540 175,207,208,211,215,
Collet style clamp assembly................... 289-296, 219-233,251-258,267-
523-524 271,275,279,281-284,
Computer-aided engineering ................... 957-966 289-296,313-350,327-
Computer-aided pictorial process 350,365-380,403-409,
planning (APPP) system ............ 513-516 427-452,461-469,505-
Computer-aided quality assurance .................. 503, 512,517-573,747-755,
515-516 765-772,859-866,898,
1244
Dipole magnet, superconducting (cont'tf) Field
913-921,993-1000,1039, angle measurements ...................................... 525
1075-1080, 1143-1185 harmonics ..................................... ......... 773-780
components, see separate components magnetic ...................... 138,251-258,327-342,
Drift chamber, proportional 572-573,657,765-772,
supermodules ........................................ 175-181 1218-1219
Filament ............................................. 924, 939-942
Eddy current .................................... 141-144,369, area histogram .............................. .................. 926
677-684 Film, magneto-optic, for localiza-
Education program, vocational, in North tion of magnetic axis .............. 1103-1109
Central Texas .................................... 11-12 Fine filament superconductor pro-
Einstein, Albert ..................................................... 15 ject in Finland ........................................ 49
Electron ................................................... 16, 18, 199 Finite element, see Element, finite
Electron positron collider (LEP) ...................... 199 Fire ant, red, imported ....................................... 100
Element, finite FIS, see Facility Information System
analysis (FEA) ..................... 167-169, 186-190 FMEA, see Failure mode and effects
approximations ..................................... 328-331 analysis
computer program for .......................... 327-334 Force, magnetic, coercive .................... 1067-1070
mesh ...................................... 333, 334, 576, 578 FRACAS (failure reporting analysis
method ............................................................ 315 and corrective action
Ellis County, Texas .......................... 85-92, 95-98 system) ..................................... 1143-1148
Employment in North Central Texas ............. 5-12 listing of 16 problem causes ...................... 1146
codes listed
ind us trial..... .... .... .... ........ .... .... ........ ..... ........ 11
occupational ................................................ 11 Gage, mechanical ...................................... 475-481
End of magnet for alignment measurement of
assembly, lead end ......................................... 290 ASST magnet .............................. 475-481
can ...................................................... 1095, 1164 error analysis ......................................... 477-478
cap calorimeter and ..................... 185, 189-190 recommendations ........................................... 480
clamps ................................... 289-296, 523-524 Gammas, electrons and muons
collar ............................................................... 538 experiment ................................... 781-800
collet end clamp .................................... 289-296 barium fluoride crystal calo-
assembly ....................................... 1163-1164 rimeter and ................................... 781-800
cross-section ................................ 291, 337, 340 Gamet film technology........................ 1103-1109
deflection measurements ..................... 291-294 Gas
design ............................................ 290, 335-342 conduction, residual...................................... 997
dipole magnet and ................................ 335-342 high-pressure technology ................ 1201-1202
dome weld ...................................................... 526 species for vacuum ............... 268-271 see also
field magnetic and ......................................... 336 separate gases
force and ..................... 440, 540-546, 556-557, -tungsten arc welding camera
570-571 vision system ............................... 757-763
geometry, three-dimensional....................... 337 GEM experiment, see Gammas, electrons
harmonic, spatial........................................... 336 and muons experiment
mechanics .............................................. 289-296 Geographic Information Systems
model ............................................................... 345 (GIS) .................................................. 69-77
plate ....................................... 439-440, 538, 542 Geology of supercollider main ring
pressure sensitive film and ................. 291-292 in North Central Texas ........... 1127-1129
quench .................................................... 289, 293 GIS, see Geographic Information
spacers, optimization of ...................... 338-342 Systems
stress ....................................................... 290, 294 GIycidyl amine resin ........................................ 1049
thermal ............................ ........................... 347 GIycidyl ether resin ......................................... 1049
Energy level, quantitized .................................... 16 Graphite fiber resin matrix composite ............ 161
Environmental Affairs Office ............................. 80 Ground motion studies ...................................... 314
Extensometer ......................................... 1041-1042
Facility Information System (FIS) ............... 69-77 Hadron

Failure mode and effects analysis calorimeter ......................................... 1097-1101
(FMEA) ........................................ 500, 502 collider ..................................................... 41,591
Failure reporting, see FRACAS Hall effect probe ................................................. 766
Feed can ............................................................. 1095 Harmonics ................................................ 532-534,
Fiber, scintillating ...................... 1193-1199, 1211 773-780
detector .............................................. 1193-1199 spatial ..................................................... 329-334
1245
Health, occupational .................................. 107-111 Lorentz force ................... 121,128,289,290,346,
preventiom hierarchy, listed ........................ 108 366,1171,1178
Heat
flow to cold mass ............................... 998-1000 MAGCOOL subcooler cryogenic
leak ......................................................... 457, 459 system ........................................... 483-490
shield ............................ 321,324,453,455,458 Magnet see separate magnets
Heat Leak Test Facility (HLTF) ............. 236-241, availability goals .............................. 1143-1148
267-274 cryogenic system .............................. 1089-1096
Heavy ion collider, relativistic ......................... 773 database system ....... .... .... ..................... 873-880
HEB dipole magnet ................................... 335-342 manufacture, quality, total.................. 723-729
ring ................................................ 208, 211, 215 parts, see separate parts
Helium .................... 270-271,305,307-312,454, reliability.......... 1143-1148 see also FRACAS
458, 528, 902, 1025 1026, steel .................................................... 1066, 1068
1067, 1093-1094 superconducting .................................... 773-780
flow............................................... 307, 486--487 test stand ............................................ 1089-1096
liquid ...................................................... 731-738 Magnetics connectivity ............................ 961-962
containment vessel................. 747, 757, 898 Magnetization ............................................ 141-144
passage ................................................... 391-394 Magnetohydrodynamics for sub-
pressure ........................................................... 394 marine propulsion .. .............................. 651
refrigeration and .......................... 199-206, 484 bis-Maleimide ................................................... 1049
screw compressor, oil-injected ........... 627-634 Mandrel ............................. 397,400,411,519,520
power loads and ............................... 635-642 Manganese steel .................................... 1055-1064
Higgs events ...................................................... 1214 Mars lander approach in archeology................. 89
Higgs decay ....................................................... 1214 Measurement instrumentation,
HIPing .................................................................... 36 magnetic, for alignment ............. 467-474
Hybrid dipole magnet and ................................ 467-474
performance of ............................................. 1211 multipole magnet and ................. 467,471-474
supedayers ........................................ 1209-1210 quadrupole magnet and ........................ 467-474
tracking system ................................. 1209-1215 Mesh ............................................................ 576, 578
Hydrogen .............................................. 16,170-171 density studies ...................................... 328-334
Hysteresis ................................................... 697-699 Meson scoreboard ................................................ 18
loop ................................. 929, 1025, 1086, 1087 Minor edge packing factor for cable ........... 27-29
calculation ........................................................ 27
ICRAND (simulation computer MLI system ......................................................... 458
program for quality Molding .................................................. 1180-1182
control) ......................................... 295-296 Monofilament ............................................ 661-668
Injection molding composite .......................... 1181 Monte Carlo simulations ...................... 1210-1211
Insulation Motion, sinusoidal, amplification .................... 314
electrical ................................................ 403-409 Multimeter ........................................................... 462
thermal ....................... 235-242,321,324,405- Multiplexed optical transmission
406,453,997-998 readout .......................................... 583-590
Interconnection .......................................... 525-526 MultiplexerlDemultiplexer ............................... 588
Interface inspection methods ................... 480-481 Multipole ......................... 138-142, 146, 148,472,
INTERLINK database ......................................... 12 777,780
IRIS system for scanning ......................... 225-233 calculations ........................................... 327-334
Iron yoke, see Yoke magnet ................................................ 1103-1109
Isotopes .................................................................. 17 Multiwire head .................................................... 276
Jc values ...................... 926-928 see also Current Muon ................................. 175-181,575,582,591
density, critical chamber .................................................. 151-158
Jellyroll technique for coil .............. 412,413,415 magnet .................................................... 168-173
tracking .............................................. 1212-1214
Kapton for coil insulation .................... 1031-1045
Land information system .............................. 71-75 National Environmental Policy

Laser Act (NEPA) ............................................. 79
cutting ................................................ 1070-1074 SSCL compliance with ............................. 79-83
diode characteristics, tabulated ................... 587 NbTi superconductor ..................................... 33-67
tracking for alignment ......................... 509-511 N-channel junction field effect
LHC, see Large hadron collider transistor (JFET) ......................... 817-824
Lines of sight method for alignment ...... 507-510 NFl>A, see National Environmental Policy
LIS, see Land Information System Act
LMI software ........................................................ 12 Network, neural ................................................ 1195
1246
Network, neural (cont'd) Quality
triggering system .............................. 1196-1198 assurance ............................................... 499-504
Neutron ................................................... 17-19,591 control.................................................... 492-503
Nitrogen ................... 207-217,270-271,454,458 method, statistical........................... 492--498
SSC refrigeration and .......................... 207-217 FMEA (failure mode and effects
Nucleon .................................................................. 17 analysis) ....................................... 500, 502
ICRAND simulation by computer ..... 495--498
Ohio Center of Science and Industry, see Quark ...................................................................... 18
COS I Quench, magnetic ........... 131-133,365-372,443-
Optimization of end design 452,483--490,529-530,
magnetic ................................................. 335-342 546-548,552-555,559,
mechanical ............................................. 335-342 563-569,867-871,881-
Oscillation 887,972, 1166
amplitude ......................... .... .................. 769-770
of current ............................................... 975-980 Radioactivity ................................................... 19-20
time dependence ............. .... .................. 770-772 Radiation, infrared .................................... 996-997
Ozone and welding ............................................ 760 Ramp rate ............................................................ 530
dependence ............................................ 369-372
Pareto analysis for failure ................................. 502 study....................................................... 567-569
Particle smasher ................................................... 19 Rare earth electron configuration .................... 813
Periodic table .................................................. 17-18 Refrigeration ............................. 200-202, 207-217
Pest Requirements control system .................. 889-896
definition ............................. ............................. 99 myths listed .................................................... 889
domestic .......................................................... 100 specifications ............................... 889, 895-896
Pesticide ....................................................... 99-105 Resin ............................................ 1049,1180-1181
definition ................................................. 99-100 RHIC (relativistic heavy ion
regulations ............................................. 100-102 collider) ....................... 411--412,416--417
responsibility of applicators ............... 104-105 Ring test ............................................. 419,422--424
SSC and ................................................... 99-105 Risk management .................................. 1149-1158
standards ................................................ 100-102 dipole magnet development and ..... 1149-1158
bis-Phenol A ...................................................... 1049 assessment methodology for ...... 1151-1153
Photocathode .......................................... 1219-1220 Rutherford, Ernest ....................................... 17, 767
Photoelectron ............................................. 592, 596 Safety, occupational ................................... 107-111
Photomultiplier tube for calorimetry .... 1217-1224 prevention hierarchy listed .......................... 108
Photon transmission ........................................... 225 Sample specification tree ................................ 1140
IRIS system and ............ 225-233 see also IRIS Saturation ................................................... 141-144
Phototube design ................................... 1217-1218 Science and imagination ............................... 21-24
Pit slab ........................................................ 931-933 Science and Technology Interactive
Polymer Center at Aurora, Illinois ................ 13-14
fiber-reinforced .............................................. 283 SDC, see Detector collaboration, solenoidal
radiation resistance .......................... 1048-1049 Sensor system ............................................ 175-180
temperature, cryogenic, and ............ 1048-1049 computer clock and ....................................... 176
thermoset resin systems ................... 1049-1050 Sextupole magnet ..................... 141-147,275-279
"universal", theory of ................................. 1048 history............................................................. 558
Power lead ............................ 635--642, 1093-1094 Shell strain
helium-cooled ................. 635--642, 1093-1094 axial ........................................................ 441,442
Propulsion system design, concep- azimuthal ......................... 440--442, 1021-1024
tual ................................................ 651--659 thermal ........ 344,347, 1021-1024,1164-1165
Preservation law, history of ... ....................... 93-94 Shewhart XBAR R control chart ..................... 492
Press pressure ..................................................... 461 control limit calculation,
Pressure preliminary .................................. 494--495
azimuthal ............................................... 124, 126 Shield refrigeration for collider
of press ............................................................ 461 ring ................................................ 207-217
-sensitive film ....................................... 291-292 Shim
Projection microscope ......................................... 15 ring ......................................................... 162-165
Proton ............................................................... 17-19 location ........................................................... 990
beam in dipole magnet ........... ....................... 220 Shuttle system (transfer shuttle)
for dipole magnet .................... 1075-1080
Quadrupole magnet ................. 243-250, 259-266, Silicon detector module
281-288,297-312,513- diode thermometry, cryogenic and ..... 619--626
516,669--676,967-973, specifications ................................................. 604
1187-1192 strip detector ......................................... 583-590
1247
Silicon tracking system ............................ 611-618 Superconductor (cont'd)
detector, solenoidal and ....................... 616-618 multifilament hysteresis loss,
cross-section ..................................... 611,612 theory of .. .... .... ........ ..................... 699-700
Skew quadrupole ................... 509,774,775, 1174 production trial..................................... 950-954
Skin welding operation ............................ 757-759 testing ..................................................... 739-740
Solenoid ...................................................... 943-955 Support post ...................................... 297-303, 996
detector collaboration ................................... 611 Surveillance medical at SSCL .................. 107-111
Solenopsis invicta (fire ant) .............................. 100 Suspension, internal .................................. 455-457
Spruce Tree House (U.S.Indian) discov- System, cryogenic, at SSCL .................... 191-198
ered in 1888, ........................................... 94
SSC, see Superconducting Supercollider Technology transfer .................................. 985-991
Statistical quality control, see Quality Temperature, see also Silicon diode,
control Thermometry
Steel ............................................................. 167-174 dependence ............................................ 703-709
high-manganese ................................ 1055-1064 sensor ................ .... ........ ......................... 619-626
ultra low-carbon ............................... 1065-1074 Thermal equivalent circuit model........... 994-998
Strain gauge ("bullets") ................... 461-466,552 Thermometer test set-up ........................... 620-624
Strand .... .................................... ............ ...... 685-693 Thermometry, cryogenic .......................... 619-626
Straw tube system ................................. 1209, 1211 Thermoplastics ............. ........ ........ ........ .............. 298
Stress Thermoset resin ..................................... 1049-1050
azimuthal ........................................... 1021-1024 "universal" ................................................... 1050
relaxation, compressive .......... 351-364,1031- Thruster, magnetohydrodynamic ............ 651-658
1032,1036 in submarine ................................................... 651
String test magnet ........................ 25-32,475-481 Toroid .......................................................... 575-582
Subcooler .................................................... 483-490 barrel - .................................................... 575-578
Submarine intermediate - ........................................ 579-580
propulsion .... ................ .... ............................ ... 651 muon - .................................................... 575-582
stealth, acoustic ............................................. 651 Tracker ........................................................ 159-165
Superconducting Supercollider Tracking
(SSC) ......... 113-149,219-224,923-930, silicon-based ......................................... 603-610
957-966, 1052, 1055-1064 technology ......................................... 1209-1215
computer systems ............................................ 12 Transfer
cryogen systems ................................... 897-904 line, cryogenic ...................................... 453-460
design ............................ 1112-1115, 1141-1142 molding compound ........................... 1180-1181
by BNL (Brookhaven National Transistor .................................................... 817-824
Laboratory) ................................... 115-119 Transmission, high-voltage .................. 1111-1117
by FNL (Fermi National Transportation ............................................ 313-320
Accelerator Laboratory) ............ 119-120 of magnets ............................................. 313-320
education and ............................................... 5-12 thermal ............................................................ 269
in Ellis County, Texas Triggering ............................................... 1194-1198
archeology ............................................. 97-98 Tungsten drift wire, gold-plated ...................... 152
architecture, historic .................................. 96 Tunnel
paleontology .......................................... 95-97 in Austin (Texas) chalk ............................... 1134
engineering ........................................ 1137-1142 in shale ............................................... 1132-1134
FIS (facility information spool piece ............................................. 643-649
systems) ............................................ 69-77
GIS (geographic information "Universal" polymer theory ........................... 1048
systems) ............................................ 69-77 Universe Project, educational ....................... 13-20
laboratory (CCSL) UNK ................................ 867-872,881-887,1105
computer systems ........................ 1007-1014 Uranium scintillator calorimeter ...................... 183
cryogen systems ............................... 191-198 U.S. Education Office .................................... 21-24
environment and ................ ............ ....... 79-83 programs, nineteen described .................. 22-24
health and .......................................... 107-111
magnet testing ................. 717-722,905-911 Vacuum ..................................... 236-241,267-274,
quality programs .............................. 711-722 455,897-904
layout of ring (54 miles) ..................... 112, 114 Valve, cryogenic ........................................ 643-649
location, geographic ............. 5-12,86-87,208,
1127-1135 Water vapor ................................................ 269-271
subcontractors, major .......................... 719-721 Wavelength Optical Division
underground engineering ................ 1127-1135 multiplexing transmission system ..... 583-590
Superconductor .......................................... 943-955 Wedge
aluminum-stabilized ............................. 943-954 copper - .................................................. 518-519
1248
Wedge (cont'd) Wire, superconducting ................ 52-56,275-278,
symmetrical .................................................... 519 468-471,491-498,703-
Weld 709,939-942,1081-1087
cold ............................................... 528, 685--693
groove configuration .......................... 758-759 Yoke ....... 332-333, 343-350,429-433,437-440,
Welding ....................................................... 757-763 447,524-525,537,538,
ozone and ........................................................ 760 552,561,757-763,1066,
stress, azimuthal and ........................ 1021-1024 1070-1074, 1164, 1172
Windings ..................................................... 975-980
1249

Untitled

Uploaded by

Copyright:

Available Formats

Untitled

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Untitled

Uploaded by

Copyright:

Available Formats

SUPERCOLLIDER 4

Springer Science+Business Media, LLC

Supercnlllder 4 I ed'ted by ~ohn Nonte.

Proceedings of the Fourth International Industrial Symposium on Tbe Super Collider,

© 1992 Springer Scienoe+Business Media New York

All rights reserved

Andrew J. Jarabak, Westinghouse Electric Corporation

Mr. Charles E. Anderson

Charles E. Anderson Air Products and Chemicals Inc.

President and Conference Chairman Charles Anderson

I am delighted to extend warm greetings to all

This impressive gathering of researchers,

The year ahead poses great opportunities and

You have my best wishes for a productive

A Partnership (Program Status)

Opening Remarks, International Industrial Symposium on the

SSC Education Initiatives

An Assessment of the SSC Impact on the Training and Employment

Summary of the Performance of Superconducting Cable Produced for

Environment, Safety, and Health

GISIFIS Development for the SSC

Review of SSC Dipole Magnet Mechanics and Quench Performance

The Engineering Development of an Actively Controlled Precise

Mechanical Considerations for a SDC Central Straw Tracker for the

Design and Construction of the SSCL Magnet Test Laboratory

Technical Poster Session I

Materials and Processes Selection for SSC Collider Dipole Magnets

Structural Dynamic Analysis of the SSC 40 mm Collider Dipole

Stress Relaxation in SSC 50mm Dipole Coils

Theoretical Methods for Creep and Stress Relaxation Studies of SSC

Quench Performance of Six Similar 17 m-Long, 40 mm-Aperture

A Statistical Rationale for Establishing Process Quality Control

Construction Experience with Fermilab-Built Full-Length

Two-Dimensional Magnetic Analysis of the SDC Muon Toroid

Effects of Mounting Methods on Temperature Sensor Accuracy

Conceptual Design of a Superconducting MHD Propulsion System

Raw Materials and Early Monofilament Analysis from the Vendor

SSC Quality Program Overview

DC Dielectric Breakdown Tests of Liquid Helium

Axial Variations in the Magnetic Field of

Barium Fluoride Crystal Calorimeter for GEM

Improved Process for Manufacture of Radiation Hard N-Channe1

Overview of SSC Accelerator Requirements

The SSC Linac: Status of Design and Procurement Activities

The Protection System for the Superconducting Electromagnet Ring

A Dynamic Model for Beam Tube Vacuum Effects on the SSC

Transportation Studies: 40 mm Collider Dipole Magnets

Parallel Technical Sessions IV

Concept Design of the High-Voltage Transmission System for the

Systems Engineering at the Superconducting Super Collider

Summary of Production Development, Details, and Results for Model

Alternate Manufacturing Processes and Materials for the SSC Dipole

Neural Networks as First Level Triggers in Scintillating Fiber

Author Index ..................................................................................................................... 1237

Subject Index ..................................................................................................................... 1243

ON THE SUPERCONDUCTING SUPER COLLIDER

Andrew J. Jarabak, Symposium Chairman

Westinghouse Electric Corporation

Welcome to the Fourth Annual International Industrial Symposium on the

Supercollider 4, Edited by J. Nonte