Ieee C57.98-2011

IEEE Guide for Transformer Impulse
Tests
IEEE Power & Energy Society
Sponsored by the
Transformers Committee
IEEE
3 Park Avenue IEEE Std C57.98™-2011
New York, NY 10016-5997 (Revision of
USA IEEE Std C57.98-1993)
9 March 2012
Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on November 09,2017 at 17:51:35 UTC from IEEE Xplore. Restrictions apply.
IEEE Std C57.98™-2011
(Revision of
IEEE Std C57.98-1993)

Tests
Sponsor
Transformers Committee
of the
IEEE Power & Energy Society
Approved 11 December 2011
IEEE-SA Standards Board
Abstract: Transformer connections, test methods, circuit configurations, failure analysis of
lightning impulse, and switching impulse testing of power transformers are addressed. This guide
is also generally applicable to distribution and instrument transformers.
Keywords: digital recordings, IEEE C57.98, non-linear devices, switching impulse, transfer
function, transformer impulse test
The Institute of Electrical and Electronics Engineers, Inc.

3 Park Avenue, New York, NY 10016-5997, USA
Copyright © 2012 by The Institute of Electrical and Electronics Engineers, Inc.

All rights reserved. Published 9 March 2012. Printed in the United States of America.
IEEE is a registered trademark in the U.S. Patent & Trademark Office, owned by The Institute of Electrical and Electronics
Engineers, Incorporated.
PDF: ISBN 978-0-7381-7194-4 STD97201

Print: ISBN 978-0-7381-7215-6 STDPD97201
IEEE prohibits discrimination, harassment, and bullying. For more information, visit http://www.ieee.org/web/aboutus/whatis/policies/p9-26.html.
No part of this publication may be reproduced in any form, in an electronic retrieval system or otherwise, without the prior written permission
of the publisher.
Notice and Disclaimer of Liability Concerning the Use of IEEE Documents: IEEE Standards documents are
developed within the IEEE Societies and the Standards Coordinating Committees of the IEEE Standards Association (IEEE-
SA) Standards Board. IEEE develops its standards through a consensus development process, approved by the American
National Standards Institute, which brings together volunteers representing varied viewpoints and interests to achieve the final
product. Volunteers are not necessarily members of the Institute and serve without compensation. While IEEE administers the
process and establishes rules to promote fairness in the consensus development process, IEEE does not independently
evaluate, test, or verify the accuracy of any of the information or the soundness of any judgments contained in its standards.
Use of an IEEE Standard is wholly voluntary. IEEE disclaims liability for any personal injury, property or other damage, of
any nature whatsoever, whether special, indirect, consequential, or compensatory, directly or indirectly resulting from the
publication, use of, or reliance upon any IEEE Standard document.
IEEE does not warrant or represent the accuracy or content of the material contained in its standards, and expressly disclaims
any express or implied warranty, including any implied warranty of merchantability or fitness for a specific purpose, or that
the use of the material contained in its standards is free from patent infringement. IEEE Standards documents are supplied “AS
IS.”
The existence of an IEEE Standard does not imply that there are no other ways to produce, test, measure, purchase, market, or
provide other goods and services related to the scope of the IEEE standard. Furthermore, the viewpoint expressed at the time a
standard is approved and issued is subject to change brought about through developments in the state of the art and comments
received from users of the standard. Every IEEE standard is subjected to review at least every ten years. When a document is
more than ten years old and has not undergone a revision process, it is reasonable to conclude that its contents, although still of
some value, do not wholly reflect the present state of the art. Users are cautioned to check to determine that they have the
latest edition of any IEEE standard.
In publishing and making its standards available, IEEE is not suggesting or rendering professional or other services for, or on
behalf of, any person or entity. Nor is IEEE undertaking to perform any duty owed by any other person or entity to another.
Any person utilizing any IEEE Standards document, should rely upon his or her own independent judgment in the exercise of
reasonable care in any given circumstances or, as appropriate, seek the advice of a competent professional in determining the
appropriateness of a given IEEE standard.
Translations: The IEEE consensus development process involves the review of documents in English only. In the event that
an IEEE standard is translated, only the English version published by IEEE should be considered the approved IEEE standard.
Official Statements: A statement, written or oral, that is not processed in accordance with the IEEE-SA Standards Board
Operations Manual shall not be considered the official position of IEEE or any of its committees and shall not be considered to
be, nor be relied upon as, a formal position of IEEE. At lectures, symposia, seminars, or educational courses, an individual
presenting information on IEEE standards shall make it clear that his or her views should be considered the personal views of
that individual rather than the formal position of IEEE.
Comments on Standards: Comments for revision of IEEE Standards documents are welcome from any interested party,
regardless of membership affiliation with IEEE. However, IEEE does not provide consulting information or advice pertaining
to IEEE Standards documents. Suggestions for changes in documents should be in the form of a proposed change of text,
together with appropriate supporting comments. Since IEEE standards represent a consensus of concerned interests, it is
important to ensure that any responses to comments and questions also receive the concurrence of a balance of interests. For
this reason, IEEE and the members of its societies and Standards Coordinating Committees are not able to provide an instant
response to comments or questions except in those cases where the matter has previously been addressed. Any person who
would like to participate in evaluating comments or revisions to an IEEE standard is welcome to join the relevant IEEE
working group at http://standards.ieee.org/develop/wg/.
Comments on standards should be submitted to the following address:

Secretary, IEEE-SA Standards Board
445 Hoes Lane
Piscataway, NJ 08854-4141
USA
Photocopies: Authorization to photocopy portions of any individual standard for internal or personal use is granted by The
Institute of Electrical and Electronics Engineers, Inc., provided that the appropriate fee is paid to Copyright Clearance Center.
To arrange for payment of licensing fee, please contact Copyright Clearance Center, Customer Service, 222 Rosewood Drive,
Danvers, MA 01923 USA; +1 978 750 8400. Permission to photocopy portions of any individual standard for educational
classroom use can also be obtained through the Copyright Clearance Center.
Notice to users
Laws and regulations

Users of IEEE Standards documents should consult all applicable laws and regulations. Compliance with
the provisions of any IEEE Standards document does not imply compliance to any applicable regulatory
requirements. Implementers of the standard are responsible for observing or referring to the applicable
regulatory requirements. IEEE does not, by the publication of its standards, intend to urge action that is not
in compliance with applicable laws, and these documents may not be construed as doing so.
Copyrights
This document is copyrighted by the IEEE. It is made available for a wide variety of both public and
private uses. These include both use, by reference, in laws and regulations, and use in private self-
regulation, standardization, and the promotion of engineering practices and methods. By making this
document available for use and adoption by public authorities and private users, the IEEE does not waive
any rights in copyright to this document.
Updating of IEEE documents

Users of IEEE Standards documents should be aware that these documents may be superseded at any time
by the issuance of new editions or may be amended from time to time through the issuance of amendments,
corrigenda, or errata. An official IEEE document at any point in time consists of the current edition of the
document together with any amendments, corrigenda, or errata then in effect. In order to determine whether
a given document is the current edition and whether it has been amended through the issuance of
amendments, corrigenda, or errata, visit the IEEE-SA Website at http://standards.ieee.org/index.html or
contact the IEEE at the address listed previously. For more information about the IEEE Standards
Association or the IEEE standards development process, visit the IEEE-SA Website at
http://standards.ieee.org/index.html.
Errata
Errata, if any, for this and all other standards can be accessed at the following URL:
http://standards.ieee.org/findstds/errata/index.html. Users are encouraged to check this URL for errata
periodically.
Patents
Attention is called to the possibility that implementation of this standard may require use of subject matter
covered by patent rights. By publication of this standard, no position is taken by the IEEE with respect to
the existence or validity of any patent rights in connection therewith. If a patent holder or patent applicant
has filed a statement of assurance via an Accepted Letter of Assurance, then the statement is listed on the
IEEE-SA Website http://standards.ieee.org/about/sasb/patcom/patents.html. Letters of Assurance may
indicate whether the Submitter is willing or unwilling to grant licenses under patent rights without
compensation or under reasonable rates, with reasonable terms and conditions that are demonstrably free of
any unfair discrimination to applicants desiring to obtain such licenses.
iv
Copyright © 2012 IEEE. All rights reserved.
Essential Patent Claims may exist for which a Letter of Assurance has not been received. The IEEE is not
responsible for identifying Essential Patent Claims for which a license may be required, for conducting
inquiries into the legal validity or scope of Patents Claims, or determining whether any licensing terms or
conditions provided in connection with submission of a Letter of Assurance, if any, or in any licensing
agreements are reasonable or non-discriminatory. Users of this standard are expressly advised that
determination of the validity of any patent rights, and the risk of infringement of such rights, is entirely
their own responsibility. Further information may be obtained from the IEEE Standards Association.
v
Participants
At the time this IEEE guide was completed, the Dielectric Test Working Group had the following
membership:
Arthur Molden, Chair

Joseph Melanson, Vice Chair
Reto Fausch Bertrand Poulin Pierre Riffon

Jim McBride Loren Wagenaar
The following members of the individual balloting committee voted on this guide. Balloters may have
voted for approval, disapproval, or abstention.
Satish Aggarwal Michael Haas Bansi Patel

Carlo Arpino David Harris J. Patton
Donald Ayers Jeffrey Hartenberger Brian Penny
Peter Balma Roger Hayes Christopher Petrola
Paul Barnhart Steven Hensley Paul Pillitteri
Barry Beaster Gary Hoffman Alvaro Portillo
W. J. Bill Bergman William Hopf Bertrand Poulin
Steven Bezner Charles Johnson Lewis Powell
Wallace Binder Laszlo Kadar Gustav Preininger
Thomas Bishop Chad Kennedy Iulian Profir
Thomas Blackburn Gael Kennedy Jean-Christophe Riboud
William Bloethe Sheldon Kennedy Johannes Rickmann
W. Boettger James Kinney Pierre Riffon
Steven Brockschink Joseph L. Koepfinger Michael Roberts
Steven Brown Jim Kulchisky Zoltan Roman
Carl Bush Saumen Kundu John Rossetti
Bill Chiu John Lackey Marnie Roussell
Craig Colopy Chung-Yiu Lam Thomas Rozek
Stephen Conrad Aleksandr Levin Dinesh Sankarakurup
Jerry Corkran Hua Liu Bartien Sayogo
John Crouse Thomas Lundquist Devki Sharma
Gary Donner Greg Luri Gil Shultz
Donald Dunn Richard Marek Hyeong Sim
Fred Elliott J. Dennis Marlow Charles Simmons
Gary Engmann John W Matthews James Smith
James Fairris Omar Mazzoni Jerry Smith
Joseph Foldi James McBride Steve Snyder
Bruce Forsyth Susan McNelly Gary Stoedter
Marcel Fortin Nigel McQuin John Vergis
Eduardo Garcia Joseph Melanson Loren Wagenaar
Saurabh Ghosh Arthur Molden David Wallach
Jalal Gohari Daniel Mulkey Joe Watson
Edwin Goodwin Jerry Murphy Ernesto Jorge Wiedenbrug
James Graham Ryan Musgrove Alan Wilks
William Griesacker K. R. M. Nair James Wilson
Randall Groves Michael S. Newman John Wilson
Ajit Gwal Joe Nims William Wimmer
Lorraine Padden
vi
When the IEEE-SA Standards Board approved this guide on 11 December 2011, it had the following
membership:
Richard H. Hulett, Chair

John Kulick, Vice Chair
Robert M. Grow, Past President
Judith Gorman, Secretary
Masayuki Ariyoshi Jim Hughes Gary Robinson

William Bartley Joseph L. Koepfinger* Jon Walter Rosdahl
Ted Burse David J. Law Sam Sciacca
Clint Chaplin Thomas Lee Mike Seavey
Wael Diab Hung Ling Curtis Siller
Jean-Philippe Faure Oleg Logvinov Phil Winston
Alexander Gelman Ted Olsen Howard L. Wolfman
Paul Houzé Don Wright
*Member Emeritus
Also included are the following nonvoting IEEE-SA Standards Board liaisons:
Satish Aggarwal, NRC Representative

Richard DeBlasio, DOE Representative
Michael Janezic, NIST Representative
Francesca Drago
IEEE Standards Program Manager, Document Development
Erin Spiewak
IEEE Standards Program Manager, Document Development
vii
Introduction
This introduction is not part of IEEE Std C57.98-2011, IEEE Guide for Transformer Impulse Tests.
Early in 1955 a Working Group was appointed by the Dielectric Test Subcommittee of the AIEE
Transformers Committee to prepare an Impulse Test Guide for oil-immersed transformers. The present
content of this guide is a consolidation of all the revisions that have occurred since then, revisions that
introduced new developments in testing methods and new developments in impulse recording and fault
detecting methods. In keeping with the continuing development of this Guide, additional sections are
included in this edition on the testing of transformers that include non-linear devices, the use of digital
impulse recording systems, and an Annex on the advanced processing of digital records and the application
of the transfer function algorithm.
viii
Contents
1. Overview .................................................................................................................................................... 1
1.1 Scope ................................................................................................................................................... 1
1.2 Purpose ................................................................................................................................................ 1
2. Normative references.................................................................................................................................. 2
3. Impulse Testing .......................................................................................................................................... 2

3.1 General ................................................................................................................................................ 2
3.2 Impulse waveshapes ............................................................................................................................ 3
4. Lightning impulse test circuits ................................................................................................................... 5

4.1 Waveshape control .............................................................................................................................. 5
4.2 Chopped-wave impulse testing of transformers ................................................................................ 15
4.3 Non-linear devices............................................................................................................................. 17
4.4 Arrangement of lightning impulse test circuits.................................................................................. 18
4.5 Measurement of lightning impulse voltages ...................................................................................... 20
4.6 Digital recording instruments ............................................................................................................ 21
4.7 Failure detection ................................................................................................................................ 23
4.8 Normal test procedure........................................................................................................................ 26
4.9 Troubleshooting................................................................................................................................. 26
4.10 Dry type transformers ...................................................................................................................... 28
4.11 Voltage and current transformers .................................................................................................... 29
4.12 Examples of impulse waveforms..................................................................................................... 32
4.13 Methods of presenting lightning impulse test results....................................................................... 42
5. Switching impulse testing......................................................................................................................... 44

5.1 Switching impulse testing techniques................................................................................................ 44
5.2 Switching impulse waveshapes ......................................................................................................... 45
5.3 Switching impulse test circuit............................................................................................................ 46
5.4 Measurement of switching impulse voltage ...................................................................................... 49
5.5 Switching impulse failure detection .................................................................................................. 52
5.6 Switching impulse and non-linear devices ........................................................................................ 53
5.7 Methods of presenting switching impulse test results ....................................................................... 53
6. Grounding practices.................................................................................................................................. 55
6.1 General .............................................................................................................................................. 55
7. Impulse generator size.............................................................................................................................. 58
Annex A (informative) Advanced processing of digital records .................................................................. 62

A.1 Introduction ...................................................................................................................................... 62
A.2 Transfer function background information ....................................................................................... 62
A.3 Transfer function theory ................................................................................................................... 63
A.4 Application of transfer function........................................................................................................ 65
A.5 Transfer function of chopped-wave records ..................................................................................... 71
A.6 Transfer function of a full wave and chopped wave ......................................................................... 73
A.7 Transfer function example with test equipment problems ................................................................ 75
A.8 Coherence function........................................................................................................................... 76
Annex B (informative) Bibliography............................................................................................................ 78
ix
Tests
IMPORTANT NOTICE: This standard is not intended to ensure safety, security, health, or
environmental protection. Implementers of the standard are responsible for determining appropriate
safety, security, environmental, and health practices or regulatory requirements.
This IEEE document is made available for use subject to important notices and legal disclaimers.
These notices and disclaimers appear in all publications containing this document and may
be found under the heading “Important Notice” or “Important Notices and Disclaimers
Concerning IEEE Documents.” They can also be obtained on request from IEEE or viewed at
http://standards.ieee.org/IPR/disclaimers.html.
1. Overview
1.1 Scope
To aid in the interpretation and application of the impulse testing requirements of the IEEE Standard Test
Codes for Transformers.
1.2 Purpose
This guide is written primarily for power transformers, but it is also generally applicable to distribution and
instrument transformers. Other IEEE standards, plus the purchaser’s specifications determine the specific
requirements for impulse tests. The purpose of this guide is not to change those standards in any way, but to
add background information that will aid in the interpretation and application of those standards. The
information contained in this guide is a compendium of technical information provided by engineers and
technicians well versed in the art of transformer impulse testing. It is hoped that this guide will provide a
basis for a better understanding of impulse test techniques and troubleshooting procedures.
1
IEEE Std C57.98-2011
IEEE Guide for Transformer Impulse Tests
2. Normative references
The following referenced documents are indispensable for the application of this document (i.e., they must
be understood and used, so each referenced document is cited in text and its relationship to this document is
explained). For dated references, only the edition cited applies. For undated references, the latest edition of
the referenced document (including any amendments or corrigenda) applies.
IEEE Std 4TM, IEEE Standard Techniques for High-Voltage Testing. 1, 2
IEEE Std C57.12.00TM, IEEE Standard General Requirements for Liquid-Immersed Distribution, Power,
and Regulating Transformers.
IEEE Std C57.12.01TM, IEEE Standard General Requirements for Dry-Type Distribution and Power
Transformers Including Those With Solid Cast and/or Resin-Encapsulated Windings.
IEEE Std C57.12.90TM, IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating
Transformers.
IEEE Std C57.12.91TM, IEEE Standard Test Code for Dry-Type Distribution and Power Transformers.
IEEE Std 1122TM, IEEE Standard for Digital Recorders for Measurements in High-Voltage Impulse Tests.
3. Impulse Testing
3.1 General
Insulation is recognized as one of the most important constructional elements of a transformer. Its chief function
is to confine the current to useful paths, preventing its flow into harmful channels. Any weakness of insulation
may result in failure of the transformer. A measure of the effectiveness with which insulation performs is the
dielectric strength. It was once accepted that low-frequency tests alone were adequate to demonstrate the
dielectric strength of transformers. As more became known about lightning and switching phenomena, and as
impulse testing apparatus was developed, it became apparent that the distribution of impulse-voltage stress
through the transformer winding was very different from the low-frequency voltage distribution.
Low-frequency voltage distributes itself throughout the winding on a uniform volts-per-turn basis. Impulse
voltages are initially distributed on the basis of winding capacitances. If this initial distribution differs from
the final low-frequency inductance distribution, the impulse energy will oscillate between these two
distributions until the energy is dissipated and the inductance distribution is reached. In severe cases, these
internal oscillations can produce voltages to ground that approach twice the applied voltage.
As circuit voltages became standardized, impulse levels corresponding to the respective voltage classes
were also standardized. Impulse levels, now referred to as basic insulation levels (BIL), were established in
1937 by an AIEE-EEI-NEMA Committee on Insulation Coordination. This committee was formed to
consider laboratory technique and data, to determine the insulation levels in common use, to establish the
insulation strength of all classes of equipment, and to establish insulation levels for various voltage
classifications. Through the use of these BILs, apparatus can be specified on the basis of demonstrating that
the insulation strength of the equipment will be equal to or greater than the selected basic level, and
protective equipment can be selected to provide adequate protection. The BILs and other insulation-test
voltages are listed in IEEE Std C57.12.00 and C57.12.01.
1
The IEEE standards or products referred to in this clause are trademarks of The Institute of Electrical and Electronics Engineers, Inc.
2
IEEE publications are available from The Institute of Electrical and Electronics Engineers, 445 Hoes Lane, Piscataway, NJ 08854,
USA (http://standards.ieee.org/).
2
During the 1950s, it became apparent that the lightning impulse test did not represent all the transient
voltages to which a transformer would be subjected. As transmission voltages increased to the extra high
voltage (EHV) level (i.e., 345 kV and above), transient voltages, caused by various switching operations,
had to be considered in both the internal and external transformer insulation design. The magnitude of
surges resulting from switching operations is dependent upon system characteristics.
As a result, a new switching impulse test was developed initially for the EHV levels. A standard switching
transient waveshape was agreed upon and the crest voltage level to ground was established at 83% of the
lightning impulse crest voltage.
3.2 Impulse waveshapes
Impulse tests are made with waveshapes that simulate those encountered in service. From the data
compiled by the 1937 AIEE-EEI-NEMA Committee on Insulation Coordination about natural lightning, it
was concluded that system disturbances from lightning can be represented by three basic waveshapes: full
waves, chopped waves, and front-of-waves and, as later determined, by the switching impulse wave. In
Figure 1, these waves are represented in their approximate magnitude and time.
Normalized BIL
2,500 microsecond record

Normalized BIL
250 microsecond record

Figure 1 —Impulse voltage waveshapes
3
It is recognized that lightning disturbances will not always have these basic waveshapes. However, by
defining the amplitude and shape of these waves, it is possible to establish a minimum impulse-dielectric
strength that transformers should meet. A curve can be drawn through the points established by the
amplitude and normal duration of each wave as shown in Figure 1. For the front-of-wave and chopped
wave, the points would be located at the intersection of a vertical line drawn at the time-to-chop and a
horizontal line drawn through the crest, while for the full wave and switching impulse the vertical line
would be located at the time of half value (see IEEE Std 4). This curve is referred to as the volt-time curve
of the composite insulation structure of the transformer. The strength of the insulation to waveshapes other
than those defined by IEEE Std 4 can be approximated from the curve. Impulse tests demonstrate that the
insulation will withstand impulses that lie below the volt-time curve. Applying protective equipment that
has a volt-time curve lower than that of the transformer serves to provide an adequate degree of protection
for the transformer insulation.
If a lightning disturbance travels some distance along the line before it reaches a transformer, its waveshape
approaches that of the full wave as shown in Figure 1 by curve a. This is a wave that rises from zero to
crest value in 1.2 μs and then decays to half of crest value at 50 μs. It is generally referred to as 1.2/50
wave. The part of the wave between zero and the crest is called the front and the part beyond the crest is
called the tail. A wave traveling along the line might flash over an insulator after the crest of the wave has
been reached. This wave is simulated by the chopped wave that is chosen to be of magnitude as defined in
IEEE Std C57.12.00. It is shown by curve b. If a severe lightning strike hits directly at or very close to a
terminal, the surge voltage may rise steeply until it is relieved by a flashover, causing a sudden, very steep
collapse in voltage. This condition is represented by the front-of-wave curve c
The switching impulse, as its name implies, is generally characterized by the transmission systems response
to switching operations on the network. Sudden changes in load or connection or disconnection of sections
of the network can stimulate switching transients. Generally such transients are of lesser magnitude and
have slower rise and fall times than do lightning induced transients. A typical switching wave is indicated
by curve d.
As can be seen in Figure 1, the waves are quite different in duration and in rates of voltage rise and decay,
and consequently produce different reactions within the transformer winding. The full wave, because of its
relatively long duration, causes major oscillations to develop in the winding and consequently stresses not
only the turn-to-turn and section-to-section insulation throughout the winding, but also develops relatively
high voltages, compared to power frequency stresses, across large portions of the winding and between the
winding and ground (core or adjacent windings).
The chopped wave, because of its shorter duration, does not allow the major oscillations to develop fully
and generally does not produce high voltages across large portions of the windings or between the winding
and ground. However, because of its greater amplitude, it produces higher voltages at the line end of the
winding; and because of the rapid change of voltage following flashover of the chopping gap, it produces
higher turn-to-turn and section-to-section stresses.
The front-of-wave is still shorter in duration and produces still lower winding-to-ground voltages deep
within the winding. Near the line end, however, its greater amplitude produces higher voltages from
winding-to-ground. This, combined with the rapid change of voltage on the front and chopped portion of
the wave, produces a high turn-to-turn and section-to-section voltage near the line end of the winding.
The longer duration switching transients generally produce voltage distribution in a transformer that
approaches the linear distribution of the steady state network voltage. However, some exceptions may
occur, particularly if the switching transient includes higher frequency oscillations around the voltage peak.
Note that the voltage distribution associated with switching impulses is discussed in Clause 5.
4
Depending on the size and rating of a transformer and on the applicable test standards, the impulse voltage
tests applied to a transformer may include one or all of the above impulse voltage types. The selection of
the type and number of test waves is determined by the test specification. These various test specifications
may include a number of purchaser specifications in addition to those required by IEEE Standards.
4. Lightning impulse test circuits
4.1 Waveshape control
4.1.1 Simple circuits
Impulse waves are generated by an arrangement that charges a group of capacitors in parallel and then
discharges them in series; see Abetti [B1], Aeschlimann [B2], Aicher [B3], Aicher [B4], Anderson [B7],
and Transformer Reference Book [B52]. The magnitude of the voltage is determined by the initial charging
voltage, the number of capacitors in series at discharge, and the voltage efficiency of the circuit. The
waveshape is determined largely by the constants of the impulse generator or IG and the impedance of the
load. Figure 2 includes a single-stage representation of an impulse generator, its typical circuit parameters,
and load.
Figure 2 —Lightning impulse circuit
To illustrate how the various transformer and impulse-generator parameters affect the generated
waveshape, some simple circuit examples will be given. Assume that for the following examples, a
capacitor, which will be referred to as the generator capacitance C, is charged to a constant value and then
the various circuit parameters are connected to the generator capacitor terminals through a quick-closing
switch (such as a discharge across a sphere gap). The circuit will be assumed to have zero resistance and
inductance except as indicated. Figure 3 shows the circuit parameters and the resulting waveshape for each
example. In all these examples a high-impedance oscilloscope will be connected at points X-X, to indicate
the variation of voltage with respect to time as the various parameters are connected. In each example the
capacitor will be charged initially to the same voltage.
5
Figure 3 —Impulse waves from simple circuits
With only the oscilloscope connected to X-X as in Figure 3(a), an oscillogram similar to A will result when
the switch is closed. By connecting a capacitor across the X-X terminals as in Figure 3(b), a waveshape
similar to B will appear. It will have a shape like A but will have a smaller crest magnitude. The magnitude
will be decreased in accordance with the relationship in Equation (1),
C
V E (1)
C  C2
where
V is the voltage across the load capacitor

E is the applied voltage of the generator capacitor
C is the generator capacitance value
C2 is the load capacitance value
The time to reach crest for both of these traces will be zero since there is no series resistance or inductance
to limit the current in charging the load capacitance. Changing the value of the capacitor will only affect the
magnitude of the wave and not the time to reach crest magnitude.
When a resistor is placed across these terminals as in Figure 3(c), a trace similar to C will result. Again the
crest will be reached instantly but now the tail of the wave will decay exponentially to zero. Decreasing the
resistance will cause the tail of the wave to decrease faster as shown in curve C′. This can be explained by
studying the time constant of a resistor and capacitor circuit that is equal to the product RC. Reducing R or
C decreases the duration of the wave in direct proportion to the decrease in these parameters.
6
Placing an inductance across the generator capacitor as in Figure 3(d), will cause an oscillatory wave
similar to D to appear. Again the time to crest will be reached instantly but the tail now will oscillate about
the zero line rather than reach the zero line exponentially as in the case of the resistor. This is due to the
interchange of energy between the electrostatic field of the capacitor and the electromagnetic field of the
inductance. Decreasing the inductance or capacitance causes the wave to change from D to D′, which has a
shorter period of oscillation. This change in period, as shown in Equation (2), is proportional to the square
root of the LC product since the period itself is equal to
T= 2  LC (2)
where
T is the period of oscillation in seconds

L is the inductance in henries
C is the capacitance in farads
Placing a capacitor in parallel with an inductance across the generator capacitor as in Figure 3(e), causes an
oscillation similar to E. Here again the wave is oscillating but because of the load capacitor, the magnitude
is initially less than D; the period is increased because the total circuit capacitance is increased.
When a circuit consisting of a resistor in series with the capacitor is placed across the generator capacitance
as in Figure 3(f), a trace similar to F will appear on the oscilloscope. Now the time to reach crest is
affected. Closing the switch causes all the voltage to appear across the resistor initially and also causes a
current to flow in the circuit, which is limited by the resistor. This current starts to charge the load
capacitor, which decreases the voltage drop across the resistor. After the capacitor is charged, no further
current will flow in the circuit and thus all the voltage will appear across the capacitor.
The time required to charge the capacitor is proportional to the time constant RsC2. The capacitor will reach
95% voltage in approximately 3RsC2. Increasing the resistance, Rs will lengthen the front from F to F′.
Increasing the capacitor C2 will also increase the time to crest and decrease the magnitude as shown by F′.
If the resistor is initially replaced with an inductance in this example, basically the same result will be
obtained; except that high-frequency oscillations will be superimposed on the crest. The inductance will
initially limit the current available to charge the load capacitor and thus will increase the time to crest.
These few examples provide an insight into the effect of various circuit parameters of a transformer
impulse testing circuit upon the generated waveshape. Combining all of these parameters results in an
equation that is cumbersome and difficult to handle. There have been many technical papers written on
the subject of surge generator characteristics for transformer testing; see Aicher [B3], Aicher [B4],
Anderson [B7], and Transformer Reference Book [B52]. These examples show that the time to crest of an
impulse wave is affected by the series inductance, series resistance, and load capacitance. The tail of the
wave is controlled by the generator capacitance, load resistance, load inductance, and also load capacitance.
Figure 3(g), shows a simplified generator and transformer circuit consisting of the parameters discussed.
The waveshape G results because the transformer circuit is a complicated network instead of the simplified
circuits shown.
7
In general, if the effective impedance of the test circuit is high or if the test circuit parameters are
predominantly resistive and capacitive, the following formulas can be used to calculate the approximate
value of the front time T1 and tail time T2, as in Equation (3) and Equation (4):
C  C2
T1  2.5Rs (3)
C  C2
where
T1 is the lightning impulse front time in seconds

Rs is the total series or front resistance in ohms
C is the output capacitance of the impulse generator in farads
C2 is the total load capacitance in farads
T2  0.7 R (C  C2 ) (4)
where
T2 is the lightning impulse tail time in seconds

R is the total tail resistance in ohms
C is the output capacitance of the impulse generator in farads
C2 is the total load capacitance in farads
Should the first calculated value not yield the required front or tail time, then adjustments should be made
to the test circuit and another front or tail time obtained; a series of such iterations will eventually result in
an optimum test set up.
4.1.2 Waveshape control of transformer test circuits
4.1.2.1 The transformers’ effect on the waveshape
The types of transformers that are commonly impulse tested range from small instrument transformers of a
few tens of VA, to very large power transformers of 1000 MVA. This presents an extremely wide range of
transformer loads on which the required impulse test waveshapes are applied. Further, the actual load
presented to the impulse test equipment by a given transformer is very dependent on the voltage rating of
the terminals to be tested since the impedance of the transformer windings are related by their turns ratio
squared. It is therefore quite common to find that a given impulse system will successfully apply the
required waveshapes to the high-voltage terminals of a transformer but just not have the capacity to apply
the same waveshape to the low-voltage terminals of the same transformer. When such transformers are
being tested, the test equipment operators must resort to various circuit techniques to optimize the test
circuit for the best waveshape available; techniques such as paralleling of the impulse generator stages,
Glaninger circuit connections, see Glaninger [B21], and resistive loading of non-impulsed terminals should
be explored.
Large transformers with high load capacitance values require a reduction in the front resistance which may
result in oscillations on or around the voltage peak. When this is the case, the front resistance should be
adjusted and the oscillations and/or overshoot should be limited to 5% of the peak voltage. In special cases
when the load capacitance is high and when agreed upon between a manufacturer and customer,
oscillations/overshoot of 10% may be acceptable. It should be noted that for oscillatory waves the peak
voltage value shall be measured according to the methods specified in IEEE Std 4.
8
Long connections between the impulse generator and the terminals under test can also be responsible for
front oscillations. In general the best results will be obtained when the connections between the impulse
generator and terminal under test are short, for example, no longer than the height of the terminal under test
above the test floor.
As depicted in Figure 3(g), the transformer impulse test circuit includes both load capacitance and load
inductance. The load capacitance includes that of the impulse measuring circuit and preload capacitor (if
one is being used), the bushing(s), and the effective value of the transformer winding series capacitance and
capacitance to frame or ground. The effective capacitance of the transformer is mainly determined by its
design, and the capacitance value is relatively independent of the connections made to the non-impulsed
terminals of the transformer. On the other hand, the load inductance, the effective inductance of the
transformer, is very dependent on the connections made to the non-impulsed terminals of the transformer.
In general transformer impulse tests are carried out with their non-impulsed terminals connected to the
frame, effectively short circuiting the non-impulsed windings; this has the effect of reducing the effective
inductance of the transformer to its minimum value.
4.1.2.2 High-impedance windings (Lt > 100 mH)
For higher impedance windings, the front and tail parameters can generally be determined according to the
principles mentioned above for the purely resistive and capacitive impulse test circuits, referenced in Figure
3(a) through Figure 3(f), using Equation (3) and Equation (4).
If the effective capacitance of the transformer is not known and therefore not accounted for, the wave forms
obtained could be outside the acceptable range of rise and fall times; in which case circuit adjustments will
have to be made and subsequent impulses applied until a suitable waveshape is obtained. It should be noted
that, as specified in IEEE C57.12.90 and IEEE C57.12.91, all impulses applied to a transformer terminal
greater than 40% of the BIL shall be recorded; set up shots should therefore be limited to less than 40% of
the BIL.
For larger transformers, and transformer windings of effective inductance Lt in the range of 20 mH to 100
mH, the winding impedance considerably reduces the discharge time constant that would normally be
expected based on the impulse generator parameters and Equation (4) above. Experience has shown that for
tests on windings with effective inductance in this range the impulse generator tail resistance value may
need to be from two to ten times greater than that derived from Equation (4).
For tests on the lower impedance windings there will come a point when increasing the tail resistance has
little or no effect on the tail time. This effect will generally become noticeable as the required tail resistance
value tends to ten times that derived from Equation (4). At this point it becomes necessary to adjust the
other parameter included in Equation (4), the generator capacitance.
For a given impulse generator (IG), the generator capacitance parameter is the output capacitance of the
generator. The output capacitance is directly proportional to the stage capacitance of the generator and the
number of stages connected in parallel. The output capacitance is inversely proportional to the number of
stages connected in series. If only a small increment in output capacitance is required, it can be achieved by
reducing the number of series connected stages, providing that the number of stages used is sufficient to
obtain the required output voltage. Alternatively, and if sufficient numbers of additional stages are
available, it will be necessary to parallel connect generator stages to increase the output capacitance.
4.1.2.3 Low-impedance windings (Lt < 20 mH)
As a first approximation, for the adjustment of the impulse front time the same procedure as followed for
high-impedance windings can be applied. However, it will be found that with very low-impedance
windings, adjustments of the front resistance will also produce a considerable change to the tail time of the
9
impulse waveshape. For windings in this inductance range even a moderate value of front resistance has the
effect of seriously damping and attenuating the impulse generator output waveshape. There are a number of
circuit modifications that can be employed to reduce the damping effect of the front resistor. Reducing the
ohmic value of the front resistor will help extend the tail but will very often cause the front time to become
too short and possibly to cause oscillation around the peak. In this case connecting additional loading
capacitors (preload capacitors) in parallel with the impulse generator can help restore the front time. This
circuit modification has limited scope of application due to the fact that the combination of low front
resistance and high front capacitance can cause oscillations to occur on the front of the impulse wave.
As the front resistance value is reduced, the fundamental voltage shape produced at the output of the
impulse generator tends to that of a damped cosine wave rather than an exponential. The voltage wave
oscillates about the origin with the consequence that the transformer and the impulse generator are subject
to oscillating voltage reversals. The magnitude of the reversals and the repetitive oscillations, if not
controlled, may stress the transformer interturn and interwinding insulation more than required by the test
standards, and so some degree of damping has to be arranged. It is therefore recommended that the degree
of damping be such that the magnitude of the opposite polarity peak be limited to 50% of the applied
voltage. This degree of damping does have the effect of reducing the tail time of the cosine wave to 70% of
the value it could have without damping.
Depending on the natural frequency of the oscillations it may be necessary to view the applied voltage
wave over an extended time duration to be sure of recording the reverse voltage peak. Once the peak
magnitude has been determined and appropriate circuit adjustments made, the time duration of the impulse
record can be returned to that recommended in Table 2 for the type of test being performed.
To limit the voltage reversals to 50% it would be necessary to adjust the damping of the circuit (by
adjustments of the front and /or tail resistors) such that the damping factor k of the circuit was 0.25; that is,
the damping resistance used would be 25% of that required for critical damping. Equation (5) relates the
required tail time to the capacitance and inductance of the test circuit is:
T2  0.5  Lt  C (5)
where
T2 is the tail time in seconds

Lt is the transformer inductance in henries
C is the impulse generator output capacitance in farads
Equation (5) can readily be transposed to yield the value of C for the required tail time T2 for a given
winding inductance Lt but, there is an important point to note here about the circuit parameters. Whereas,
for the case of exponential waves the relationship between the required tail time and the generator
capacitance was linear, for the oscillatory case produced by low-inductance windings the relationship now
follows a square law. Extending the wave tail requires larger increments of generator capacitance.
It should also be noted that, as Equation (5) suggests, for a given impulse generator capacitance the tail
time will increase if the effective inductance of the winding is caused to increase, as it would do if the non-
impulsed terminals were not directly connected to the frame, but connected to the frame via “loading”
resistances.
For any given transformer, with its non-impulsed terminals connected directly to the frame, the effective
inductance presented to the impulse generator will be of some fixed value. Being fixed, the circuit
adjustments available to cause variations in the tail are limited to adjustments of the circuit resistance and
capacitance. The capacitance range is limited by the available generator and the resistance range is limited
by the damping factor requirements. There is another circuit modification that is commonly used to help
extend the tail time of a low-impedance transformer impulse test circuit and this is the Glaninger circuit.
10
4.1.2.4 The Glaninger circuit
In the basic Glaninger circuit an inductance is connected in parallel with the front resistor(s) as indicated in
Figure 4.
Glaninger
Impulse Generator Transformer
Circuit
L
SG
Rf Rf
Cg Internal External
Rt Rt
Internal External
In general Rt Int > Rt Ext & Rf Int << Rf Ext
Figure 4 —The Glaninger circuit
Inductors can be inserted in parallel with the impulse generator (IG) front resistors in the IG stages or, as is
more normally the case, a separate circuit assembly can be interposed between the IG output and the
transformer terminal under test.
The effect of the inductance is to shunt the low-frequency components of the impulse current around the
front resistor. By choosing a suitable value of inductance, the effective impedance of this R/L parallel
network can be made to pass the lower frequency components of the impulse current thereby extending the
tail duration. The damping effect of the front resistance and its corresponding effect on the tail duration is
reduced. The reduced damping of the circuit caused by the introduction of this inductor will now allow the
generator output voltage to oscillate and produce voltage reversals greater than 50%. To compensate for
this, it is necessary to arrange additional circuit damping by connecting a resistor in parallel with the
transformer winding under test. In effect, the Glaninger circuit reduces the damping effect of the front
resistor on the fundamental frequency of the test circuit and allows the circuit damping to be controlled by
a separate circuit component. The choice of front resistor value is not so constrained as it would be without
a Glaninger circuit. This circuit is more commonly employed during tests on lower voltage windings (200
kV BIL and below).
With the aid of the Glaninger circuit the maximum utility can be obtained from the available impulse
generator capacitance. If these circuit measures still do not yield an impulse with the required time
parameters, then alternative means, such as the use of loading resistors or series resistors, need to be
considered, as discussed in 4.1.3 and 4.1.4.
4.1.3 Impulse testing of low-impedance windings by alternative means
There are three alternate methods for testing windings having very low impedance. The concerns related
with each method are discussed herein and should be considered by the manufacturer in recommending the
appropriate method(s). The three methods are as follows:
11
1) Method 1: All terminals of the same windings and BIL are connected together.
2) Method 2: A resistor of not more than 500 ohms could be inserted in the grounded end.
3) Method 3: A normal impulse test is applied and the short length of the wave tail could be accepted.
4.1.3.1 Method 1: terminals tied together
Connecting the terminals together produces a high stress on the winding-to-ground insulation and a rather
low sustained stress to the turn-to-turn, coil-to-coil, and across-the-coil insulation. This is due to the turn-
to-turn and coil- to-coil stress being primarily a function of the capacitance from one end of the winding to
the other and the capacitance-to-ground. These statements can be better visualized with the following
example:
Example: Let the transformer constants be represented by an equivalent circuit as shown in

Figure 5(a). The through capacitances (capacitances from one end of the winding to the other end) of
the transformer are represented by C1 and C2, and the ground capacitance by C3. L1 and L2 represent
the transformer inductances. If the through capacitances, C1 and C2, are large with respect to the
ground capacitance, an initial distribution similar to curve X in Figure 5(b) will result. Since the final
distribution is line Y, the envelope of the winding oscillation will be between curves X and X′. This
example demonstrates a low turn-to-turn and coil-to-coil stress but a high stress to ground throughout
the winding. If the through capacitances are small compared to the ground capacitance an initial
distribution similar to curve Z in Figure 5(b) will result. The same final distribution line Y will occur
and thus the envelope of oscillation will be between curves Z and Z′. This produces a high turn-to-
turn, coil-to-coil, and insulation-to-ground stress.
The objections to using this method of test for a transformer having the parameter relationships assumed is
that part of the winding may theoretically oscillate to 200% of the applied voltage. Testing in this manner is
not recommended since in service a surge is rarely applied to both terminals simultaneously. This method
of testing does not lend itself to ground-current measurements because only the capacitance current of the
winding under test to the tank and other windings can be measured.
Figure 5 —Low-impedance windings-connecting terminals together
4.1.3.2 Method 2: series resistance
Inserting a resistance in the grounded end of the winding will produce different turn-to-turn and coil-to-coil
stresses than Method 1. The change in stress is a function of the winding constants. Figure 6(a) shows the
typical equivalent network of the transformer with one end of the winding grounded through a resistor. If
the through capacitance is extremely large compared to the ground capacitance, a distribution similar to
12
curve P in Figure 6(b) will result. The final distribution would be something similar to line Q, where all or
almost all of the voltage is across the resistor. The envelope of oscillation will then be between curve P and
P′. In this case, the turn-to-turn and coil-to-coil stress is increased compared with the example that has the
same capacitance relationship in Method 1. When the ground capacitance is large compared to the through
capacitance, a distribution similar to curve S in Figure 6(b) will occur. The final distribution can again be
assumed to be line Q. The envelope of oscillation now is between S and S′. Again it is possible to produce
excessively high voltages to ground in parts of the winding. It is general practice to insert only enough
resistance to produce a 50μs tail and the voltage appearing across the resistor is usually limited to not more
than 80% of the BIL of the grounded end of the winding. If, in the last example, the resistance required to
produce a 50μs tail had been smaller, the final distribution line would be lowered to Q′ and the envelope of
oscillation would then be between S and S′. The tail length and the voltage across the resistance should be
measured to determine the value of resistance to be used. A low-voltage impulse generator and oscilloscope
may be used to make these measurements.
Figure 6 —Low-impedance windings: winding grounded through a resistor
This method of testing applies a 50μs wave to the line end insulation and is suitable for ground current
measurements, although it is felt that the resistance may slightly reduce failure detection sensitivity.
Initially, the full-impulse voltage is applied across the winding and resistance in series; therefore, the stress
across the winding will be reduced.
4.1.3.3 Method 3: short tail time
By applying all the voltage across the winding, even though a short tail wave is used, the greatest stress to
the insulation between portions of the winding is generally produced. The stress to ground at the middle of
the winding may not be as great as Method 1 and Method 2 since the short tail will not sustain the voltage
for a long time.
In Figure 7(a), the equivalent transformer is pictured with one end of the winding grounded solidly. If the
through capacitances are large compared to the ground capacitance, then a voltage distribution similar to
curve M of Figure 7(b) will result. The final distribution is presented by line N, which means that the
envelope of oscillation will be between M and M′. When the through capacitances are extremely small
compared to the ground capacitance, then a voltage distribution similar to curve O in Figure 7(b) will
occur, which will result in an envelope of oscillation between O and O′. Again, with this method of tests
there are portions of the winding that may exceed the applied potential to the line terminals, but generally
these windings have long time constants, and the time for point T to oscillate to its maximum is usually
long enough that the voltage applied at the terminals has decreased to 50% of the crest value. This method
of test does not produce a sustained stress to the insulation-to-ground as does either Method 1 or Method 2,
13
but it does stress the insulation of the winding. The low-frequency test will produce sufficient stress to test
the insulation-to-ground.
Figure 7 —Low-impedance windings: effects due to short length of wave tail
This method of testing is very suitable for current measurements since there is no increase in the circuit
resistance and the circuit therefore has good response to high-frequency disturbances. No distribution test is
required to determine the value of the resistor to be used.
4.1.3.4 Low-impedance windings: winding construction
From these examples it can be seen that the transformer construction is a controlling factor in selecting the
method of testing low-impedance windings. Each manufacturer should be familiar with the response of
transformer construction type and should use the applied test method that will stress the winding in a
manner expected in service.
Note that, when the test equipment capabilities do not allow the standard waveshape to be obtained, all
available test circuit and wave-shaping options should be explored so that the best available compromise
waveshape can be used in the test.
4.1.4 Connection of non-impulsed terminals
Neutral terminals shall be solidly grounded except in the case of low-impedance windings. Line terminals,
including those of autotransformers and regulating transformers, shall be either solidly grounded, or
grounded through a resistor with an ohmic value not in excess of the following values shown in Table 1.
14
Table 1 —Maximum value of grounding resistance for system voltage class
Nominal system voltage (kV) Maximum resistance (ohms)

345 and below 450
500 350
765 300
NOTE—The above values represent typical transmission line characteristic impedance. 3
The following factors should be considered in the actual choice of grounding for each terminal:
a) The voltage-to-ground on any terminal that is not being tested should not exceed 80% of the full-
wave impulse voltage level for that terminal (or 50% if a delta connected winding since the voltage
transferred may be additive across the winding).
b) If a terminal has been specified to be directly grounded in service, then that terminal shall be
solidly grounded.
c) If a terminal is to be connected to a low-impedance cable connection in service, then that terminal
should either be directly grounded or grounded through a resistor with an ohmic value not in excess
of the surge impedance of the cable, for example 30 ohms.
d) If a terminal is to be connected to a gas insulated switchgear (GIS) or gas insulated transmission
line (GIL) bus, then the ohmic value should not exceed the impedance of the bus, for example 75
ohms.
e) Grounding through a low-impedance shunt for oscilloscope current measurements may be
considered the equivalent of a solid ground.
For terminals not being tested, see IEEE Std C57.12.90 and IEEE Std C57.12.91.
4.2 Chopped-wave impulse testing of transformers
4.2.1 Types of chopped wave
As mentioned in 3.2, the rate of rise of the impulse front and the rate of voltage collapse and time of
chopping all contribute to the voltage stress impressed at the line end of a transformer winding. Depending
on the construction and arrangement of a particular winding, variation of chopping time Tc will result in
variations of the voltage stress throughout the winding, and so it is not possible to state one particular time
to chop that would produce the most effective test. For this reason there have historically been two types of
chopped-wave tests performed on transformers (in North America), the front-of-wave (FOW) test and the
tail chopped-wave (CW) test. Improvements in the operating characteristics of transmission line protection
devices, such as surge arrestors, have allowed some relaxation in the chopped-wave impulse testing
requirements imposed on manufacturers by their customers. The advent of improved station shielding
practices has permitted removal of the FOW test from IEEE Std C57.12.00 test table, but it is still specified
by some utility companies, and so, some mention of it will be included herein.
3
Notes in text, tables, and figures of a standard are given for information only and do not contain requirements needed to implement
this standard.
15
4.2.2 Length of chopping gap connecting leads
For both types of chopped wave, the wave to be used shall be as specified in IEEE Std 4. The chopping gap
used to facilitate the chop may be a standard rod gap or an electronically controlled chopping gap. In either
case, whether rod gap or electronically controlled gap, it should be located as close as possible to the
transformer terminal under test without disrupting the electrical field around that terminal or the voltage
divider. The lead used to connect the chopping gap to the transformer terminal under test should follow the
shortest and most direct route to that terminal. The return conductor for the chopping gap should also be
connected, by a short copper strap, back to the frame of the transformer under test; the intention here is to
minimize the dimensions of the chopping gap loop (and therefore its inductance), so as to produce a fast
rate of voltage collapse. As a general guide, the chopping gap leads should be limited in length to no longer
than the height above the test floor of the transformer terminal under test.
4.2.3 Chopped-wave overshoot
The impedance of the chopping gap loop should be minimized so as to produce the fastest rate of voltage
collapse. Should the voltage collapse produce an opposite polarity voltage reversal (overshoot) greater than
30% of the applied voltage peak, steps may be taken to reduce this overshoot to the 30% level. This may be
arranged by relocation of the chopping gap or by including a suitable rated resistance in the high-voltage
lead of the chopping gap. However, these circuit adjustments should be limited to control of the overshoot
magnitude and not to limiting the rate of voltage collapse.
It should be noted that with some chopping circuits, particularly those associated with the testing of low
impedance, high input capacitance windings, the rate of voltage collapse can be very slow and naturally
damped. In such circuits the chopped voltage approaches the origin exponentially and does not produce
oscillations or overshoot. For these cases, providing the chopping gap location and connections are as
specified above, the natural voltage collapse produced will have to be accepted.
4.2.4 Front-of-wave chop
The ability to produce accurate voltage and time measurements during a FOW test requires a voltage
divider to have a very well defined scale factor and response time. Without knowledge of these parameters
the ability to obtain FOW voltage and time measurements is questionable. To obtain this information for a
divider that does not have the required published data, it will be necessary to either compare the divider
against a reference impulse divider or to perform a response analysis test. For information on reference
dividers and the determination of a dividers step response refer to IEEE Std 4.
The FOW test requires a particular rate of voltage rise that usually requires a different arrangement of impulse
generator front resistance, and because of the higher test voltage level, possibly a different number of impulse
generator stages than the normal CW and FW impulse test setups. Therefore, it may be necessary to make
changes to the impulse test circuit between the various impulse voltage applications. Care should be exercised
during the circuit changes to limit movement of the voltage and current recording circuit conductors, so as to
limit the impact such movements may have on the repeatability of the acquired voltage and current records.
4.2.5 Tail chopped impulses
The tail chopped impulse is intentionally chopped after the voltage peak so as to reduce the measurement
uncertainties and voltage divider requirements mention for the case of FOW testing above. By this time in
the impulse voltage epoch the divider response time should have allowed the recorded voltage to settle to
within a small percentage of the actual peak value. At this point (after the voltage peak), the relatively slow
rate of change of voltage around the voltage peak does not cause the recorded values of the voltage and
time to be subject to the same degree of uncertainty they had during the rising portion of the impulse.
16
However, even though the tail chopped impulse voltage magnitude and time of chop can be determined
reliably, changes in the time of chop will produce changes in the voltage stress produced in the line end
sections of the transformer, and will also produce changes to the voltage and current records obtained for
those tail chopped impulse applications. Since different transformer designs will produce different voltage
stress distributions for a given time of chop, the time of chop is not regarded as a test parameter, except to
require that it occur within a certain time period. The actual time period varies depending on the type of
transformer being tested (dry type, oil type, distribution, or power) but in general it is required to occur
within the limits of 2 μs to 6 μs after the start of the impulse. Note that for the longer impulse rise times the
2 μs limit may need to be shifted to 3 μs or so to ensure the CW is chopped after the voltage peak.
4.3 Non-linear devices
Depending upon the transformer design, non-linear-protective devices such as ZnO disks, as used in surge
arresters but specially coated for immersion in transformer oil, may be used in transformers. Other types of
non-linear devices may also be used. These devices may be connected across the whole, or sections of the
windings. Their purpose is mainly to limit transient over-voltages, which may be impressed or induced
across the windings, to safe levels. These devices are voltage and temperature sensitive and display non-
linear impedance vs. voltage characteristics. Their impedance up to a certain voltage level is very high. If
voltage across these devices exceeds this level, their impedance decreases in a non-linear manner. The
characteristics of these devices are so chosen that during normal transformer operation they present very
high impedance, thus allowing whole windings or winding sections to perform in a normal manner.
However, when voltage across them exceeds a certain level, their impedance decreases to limit the voltage
and protect the winding sections.
Attention should be paid in the selection of the non-linear device characteristics. Each impulse application
that results in current flowing through the non-linear elements can result in a temperature rise of the non-
linear elements. These devices have a limited energy withstand capability and when they have reached their
nominal temperature capability they should be allowed to cool down before applying additional impulses.
By their very nature, non-linear protective devices connected across the windings may cause differences between
the reduced full-wave and the full-wave impulse oscillograms. That these differences are indeed caused by
operation of these devices should be demonstrated by making several intermediate reduced full-wave impulse
tests at different voltage levels to show that the trend of the changes seen on the impulse oscillograms are caused
uniquely by operation of the protection device. Providing that the temperature of the protection devices remains
relatively constant, impulse waveshapes recorded at the same voltage level should be identical.
In order to distinguish a transformer failure from the normal operation of the non-linear devices it is necessary to
demonstrate the repeatability and reversibility of any changes being caused by the non-linear devices. This can
be achieved by the application of additional impulse applications as in the sequence suggested below.
 One reduced full wave at a level between 50% and 70% of the required full-wave impulse level
 One or more intermediate reduced full waves at a magnitude between 75% and 100% of the
required full-wave impulse level
 One full wave at 100% of the required impulse level
 Two chopped waves at 100% of the required chopped-wave impulse level
 One full wave at 100% of the required impulse level
 One or more intermediate reduced full waves at the same voltage levels that were used before the
first 100% full wave
 One reduced full wave at the same voltage level, between 50% and 70% that was used prior to the
first intermediate reduced full-wave shots
17
NOTE 1—The voltage level to be applied for the intermediate reduced full wave is not specifically given.
Only a range is proposed because the threshold operating level of the non-linear devices is transformer-
design dependent. Generally, a lightning impulse within that specified voltage range would cause the
operation of the non-linear devices. The specific number of intermediate full-wave tests and their voltage
levels cannot be given here. The number of intermediate full-wave tests and their respective voltage level
for a given transformer should be chosen by the manufacturer and agreed to by the user.
NOTE 2—In some design cases, tests at 100% of the required full-wave impulse level with the standardized
lightning impulse waveshape will not cause operation of the non-linear devices. If this is the case,
additional intermediate reduced full-wave wave tests are not necessary and may be waived.
NOTE 3—In some other special cases, the operation of the non-linear devices can be observed only during
the chopped-wave impulse tests. If this is the case, the intermediate reduced full-wave tests are also not
necessary and may be waived. A comparison of the recorded oscillograms may be done by comparing the
two chopped-wave test records up to the time of chopping. For such cases, reduced chopped-wave impulses
at a test level of approximately 75% of the required chopped-wave test level may be a useful tool to assess
that the differences on the recorded oscillograms are solely caused by the operation of the non-linear
devices. If reduced chopped-wave tests are performed, they should, by agreement, be performed before and
after the required chopped-wave tests.
Because of the operation of the non-linear devices, the comparison of the voltage and current oscillograms
shall be made only between two tests performed at the same voltage level; comparing, for example, two
90% reduced full-wave tests. It will therefore be necessary to perform two 100% full-wave impulse tests so
as to demonstrate the successful withstand of this test level. All intermediate reduced full-wave tests
performed after the full-wave tests shall be compared with the corresponding intermediate reduced full-
wave test performed prior to the 100% full-wave tests.
4.4 Arrangement of lightning impulse test circuits
The physical arrangement of the impulse test circuit plays an important part in the successful control and
measurement of the impulse test. Very high rates of change of voltage and current will occur during the test
and the physical size and geometry of the various test circuit loops cause them to produce electromagnetic
interference. Also, high rates of change of current can produce relatively high potential differences in the
return conductors of the test circuit, so it is important that all return conductors and grounding leads be
made as short as possible by interconnecting the various items in the test circuit by as direct a path as
possible. The return and grounding conductors themselves will preferably be made from flat metal strips so
as to minimize their self-inductance.
The test circuit can be divided into three major subcircuits, as follows:
1) The source, including the impulse generator and any additional circuit components used for
waveshape control
2) The voltage measuring circuit
3) The chopping gap circuit
The three subcircuits should be connected to the transformer terminal under test in a “star” type connection
arrangement, as indicated in Figure 8.
18
Figure 8 —Typical transformer impulse test connection circuit
The arrangement of the test equipment, test object, interconnecting high-voltage (HV) leads, return and
grounding conductors shall take into account any limitations in the size of the available test area and safe
electrical clearances to other objects. As a means of planning the arrangement of the test circuit the
following guidelines are offered:
a) The main circuit connection between the impulse generator and the transformer terminal
under test should be as short and direct as conveniently possible. This is particularly
important if the winding under test has high input capacitance. The length and self-inductance
of this lead tends to increase the impulse voltage rise time and produce oscillation on the
impulse front.
b) The length, position, and diameter of the lead connecting the voltage divider to the high-
voltage terminal of the test object is an integral part of the voltage measurement circuit. It
should be of the type and length and in the position for which the divider was calibrated. The
divider itself shall be positioned with adequate electrical clearances so as not to compromise
its scale factor or response time. As a general rule, the distance between the divider and any
other test equipment or metal ground planes should be at least equivalent to the height of the
divider. The high-voltage connection from the test terminal to the divider should extend
directly to the divider and not to the impulse generator or any other test object (refer to IEEE
Std 4).
c) The connection to the chopping gap should be as short as conveniently possible so as to limit
the self-inductance of the chopping gap loop, but not so short that location of the chopping
19
gap compromises the electrical clearance around the transformer terminals or around the
voltage divider.
d) The return conductors that connect the test equipment to the frame of the transformer under
test should follow a direct path, be just long enough, and, should not be allowed to pass too
close to any other return conductor, grounding lead, or metallic object in the test area. This is
particularly important for the chopping gap return conductor (and for electronically triggered
chopping gaps for the control and triggering cables). As mentioned above, high rates of
change of current can produce very high voltage differences in the return conductors. Any
sparking that occurs between the various conductors and between the conductors and other
metallic objects or cables will likely cause discrepancies to occur on the impulse records.
e) All impulse system control cables and coaxial measuring cables should be arranged within
the test area so that they are clear of all return conductors, ground conductors, and metallic
objects.
It should be noted that the above guidelines have not made comment about the connection points for the
ground conductors or grounding leads. Refer to Clause 6 for ground point locations and recommendations.
4.5 Measurement of lightning impulse voltages
4.5.1 General comments on impulse records
Measurement of the amplitude and shape of applied waves that have peak values ranging from 30 kV to
over 2800 kV, and time periods from 0.5 μs to 1000 μs requires special measuring equipment. An
oscilloscope with adequate writing speed and good accuracy, and a voltage divider with response times
suitable for extremely fast transients, are required. Either analog or digital oscilloscope systems may be
used for recording lightning impulse voltage and current records.
Regardless of the type of recording system used, it is preferable that the vertical input of the recorder be
capable of accepting input voltage signals in the range of several hundreds of volts. Communicating the
higher voltage measuring signals across the test floor helps limit their susceptability to the electromagnetic
interference produced by the high-voltage discharges in the test circuit. The measurement signals can be
attenuated to the tens of volts level by attenuators contained within the shielded enclosure of the recording
system.
The attenuation ratios to be used during the test should be chosen to produce displays that occupy at least
one half of the available screen vertical deflection, and the record should be positioned on the display so
that there is no truncation or clipping of any part of the normal record. It is to be expected that during some
fault conditions the record may go off scale.
Table 2 lists the recommended sweep lengths (for the case of analog oscilloscopes) or displayed record
lengths (for the case of digital recorders or oscilloscopes) for the various impulse types.
20
Table 2 —Recommended impulse record lengths
Impulse type Displayed record length (μs) Oscillogram ID

Front-of-wave voltage 2 to 5 FOW
Front-of-wave current 2 to 5 FOWC
Chopped-wave voltage 5 to 10 CW
Chopped-wave current 5 to 10 CWC
Reduced full-wave voltage 50 to 100 RFW
Reduced full-wave current 100 to 600 RFWC
Full-wave voltage 50 to 100 FW
Full-wave current 100 to 600 FWC
To facilitate the assessment of the test results, which are primarily based on a comparison of the reduced
and full-voltage records, it is advantageous to adjust the vertical deflection or input attenuator of the
recording system so that the differing voltage levels produce the same screen or record deflection. This
process is sometimes referred to as “normalizing” and should apply to both voltage and current records.
Impulse current is usually the most sensitive parameter in failure detection; therefore the current records
provide the main basis for the pass/fail criteria of the test. Depending on the type of current trace (FW or
CW) and on the use of linear or logarithmic sweeps, it may be necessary to use more than one record to
provide a clear display of the important periods of the total recorded length. The important periods being:
for the initial capacitive distribution of the impulse voltage, the first 30 µs; for the inductive component of
the voltage distribution the full record length, the time scale of which should extend until the inductive
component of the current has reached its peak.
4.6 Digital recording instruments
4.6.1 General
For information on the requirements for digital recorders, refer to IEEE Std 1122. The principle of digital
recording is to repetitively measure the input quantity (voltage or current) periodically, and to time stamp
each quantity measured and store those quantities and time stamps in memory. To obtain a useful record of
an event, it is necessary to obtain a sufficient number of samples and to have sufficient memory in which to
store them. Obviously, to obtain sufficient numbers of samples, it will be necessary to sample an event
before the event terminates or disappears and so the sampling “clock” needs to be fast enough to ensure
capture of sufficient numbers of samples. Once the samples are contained in memory they can be
conveniently recalled and displayed as dots on an oscilloscope or computer monitor. If the dots are
sufficient in number they will be so close together that they will appear as a continuous line and the line
will reproduce a time-based record of the original quantity.
Capturing sufficient numbers of samples is paramount for good quality records. Not only does the clock
have to be fast enough to record sufficient numbers of samples in a given time, but also the vertical
resolution has to be high enough that small discrete levels of the signal can be resolved. Both the vertical
and horizontal scales need to be high in resolution.
21
The minimum recommended vertical resolution for transformer testing was stated as 9 bits, or 29, or a full
screen resolution of 512 discrete levels or data points. The discrete levels may be likened to the bars on the
bargraph display of a multimeter. If the bargraph has only 100 points or bars to represent full scale, the
meter can only indicate magnitudes in 1% steps, it could not display 70.3%, for example.
The minimum digitizer resolution recommended for use in transformer testing was therefore 10 bits. This
would allow for the fact that the available divider and attenuator ratios for use in a test may not yield a full
screen (10 bit) deflection of the record. For those cases, providing at least a half screen deflection could be
arranged, the recommended minimum 9 bit resolution would be obtained. There are now 12 and 14 bit
digitizers available with suitable sampling rates of 100 MHz or more.
The digitizing procedure requires very fast and very precise electronic measurement and control circuits to
implement which, of necessity, operate at very low voltage (5 volts or less). To successfully produce digital
records in the high-voltage testing environment requires that those electronic circuits be contained in well-
shielded enclosures.
4.6.2 Digital recording of impulse waves
To successfully acquire an accurate record of an impulse voltage and current, the digitizer shall be set to the
correct voltage range, the correct sample rate, and have sufficient memory to store those samples. Some
less-sophisticated digitizing oscilloscopes have short fixed memories and automatically change their
sample rate to suit the available memory size as the horizontal time scale is adjusted. Some digitizing
oscilloscopes may also specify their sampling rate as “equivalent time,” time sampling, or may not specify
the sampling method at all. Such oscilloscopes should be avoided for transformer impulse testing purposes.
To accurately display an impulse wave, the impulse should be captured with adequate vertical and
horizontal resolution. That is, the ranges chosen for capturing the wave should be the lowest voltage or
shortest time that will not clip or truncate the record. For the vertical range the impulse will need to be
positioned with a particular offset so that overshoot or reverse polarity oscillations are not clipped. For the
horizontal range, a sufficiently high sampling rate needs to be selected and a memory, or record length, also
needs to be selected to contain all the data points. If, for example, there are expected to be 10 MHz
oscillations on the records and there are to be 10 data samples per period of the oscillations, then 10 MHz ×
10 samples per period requires a sample rate of 100 samples per microsecond. If 100 microseconds of time
are to be recorded the memory required would be 100 samples × 100 microseconds or 10 000 samples of
total record length. The above numbers are provided as an example only, for more information, refer to
IEEE Std 1122.
4.6.3 Digital recorders and impulse processing software
Most digital recorders provide a zoom feature whereby portions of the record can be recalled to the display,
thereby displaying smaller portions of the horizontal or vertical record in greater detail. Still other software
techniques allow records to be added or subtracted from each other and differences displayed with varying
degrees of magnification. Mathematical algorithms may also be provided to perform transformations of the
records between the time and frequency domains and to produce various forms of transfer function (see
Annex A). When making use of these additional software features it is important to realize that no two data
records are exactly identical; even when recorded from a perfect test object there will be noise and
statistical variations in the records that, when subject to high magnification (high zoom factor), may be
interpreted as differences between the records. To limit the “differences” caused by digitizer uncertainties
to a minimum, the records for comparison should be recorded using the same vertical and horizontal
resolution.
22
4.7 Failure detection
4.7.1 General requirements
One of the most important aspects of impulse testing is troubleshooting and fault detection. The ability to
detect and resolve equipment or test object problems that cause discrepancies to occur on the impulse
records requires considerable test experience and a methodical troubleshooting procedure, along with an
intimate knowledge of the test equipment, test circuit, and test object. Of paramount importance in
facilitating the troubleshooting procedure is the use of high-quality impulse voltage dividers and impulse
current transducers (shunts or CTs). To facilitate good fault detection sensitivity, dividers and shunts for
impulse voltage and impulse current measurements are recommended to have response times, as defined in
IEEE Std 4, of the order of 200 ns or less.
Simultaneous recording of both the applied impulse voltage and the impulse current of the winding under
test is preferred. However, when a simultaneous record of the current and voltage cannot be obtained, then
the recommended procedure is to have the voltage record precede the current record.
Compromises in the quality and reliability of the measuring and recording equipment can result in
compromised detection sensitivity, incorrectly diagnosed problems, and extended test and troubleshooting
time periods. Troubleshooting procedures rely on the ability to accurately compare the various impulse
records obtained during a test sequence. The records to be scrutinized are usually the applied voltage signal
obtained from the impulse voltage divider and the winding current signal obtained from an impulse current
shunt or current transformer. However, other signal records are sometimes used as troubleshooting aids
during a test. These “other” signals may be voltage or current signals obtained from non-impulsed windings
or from the test object tank or enclosure. Comparisons are made between one voltage record and another
and, between one current record and another. If the recording and test equipment were perfect and the test
object was also perfect, and linear, repeated impulse voltage applications would produce identical records;
the records could then be overlaid and their traces would match perfectly. However, when there is
interfering noise present on the records or when there is an equipment problem or an insulation failure in
the test object, then, when overlaying traces, some discrepancies between them would probably occur at
some point on the records. It is the association of such discrepancies with particular types of test circuit
problems that is the bases of impulse test troubleshooting.
When comparing two records, the greatest sensitivity will be obtained when the two records have the same
vertical and horizontal resolution and occupy at least 50% of the available full-scale vertical deflection. The
vertical attenuation factors used during the acquisition of records at different test voltage levels should
therefore be selected so as to produce the same record magnitude; that is, the ratio of the attenuation factors
chosen to make records at two different test voltage levels should be the reciprocal of the ratio of the test
voltage levels. For example, the attenuation factor chosen for a 100% test voltage application should be 0.5
times the attenuation factor used to record a 50% test voltage application. If suitable attenuators are not
available, the signal magnitude can be adjusted by making changes to the voltage divider ratio, and to the
resistance of the current shunt or ratio of the current transformer. To maintain horizontal resolution the
records should be recorded using the same horizontal time base or digital sampling frequency.
4.7.2 Ground current records
Impulse current records are obtained by monitoring the signal from a current transducer connected between
the impulse system return conductor and the non-impulsed end of the winding under test. The transducer
may be a shunt or impulse current transformer. The magnitude and waveshape of the ground current signal
is a function of the impulse current magnitude and shape and the sensitivity of the transducer in volts per
amp. Generally, transformer winding faults cause changes to the impedance of the transformer windings
that directly affect the current in the winding under test. Therefore, monitoring the winding current can be a
23
more sensitive method of detecting winding faults than can indication of voltage changes recorded by the
impulse divider.
The signal produced by an appropriate transducer contains the following three main components of current,
the relative magnitude and duration of which are largely dependent on the transformer design:
1) The capacitive component that represents the current charging the distributed series capacitance of
the transformer winding during the rising or front portion of the impulse voltage wave. This
component appears at the very beginning of the impulse current record as a fast pulse that may or
may not be oscillatory. Once the distributed capacitance is charged (within the first 2 to 10 µs,) the
winding current tends to a lower, less oscillatory value. Note that when any later rapid changes in
voltage occur, for example, the chop portion of a chopped wave or a voltage change produced by a
fault, a redstribution of charge occurs that produces a similar winding current response (possibly
larger or smaller or of opposite polarity), as did the impulse front.
2) A period of smaller and maybe slower oscillations produced by the mutual inductance coupling
individual winding sections. This current component generally occurs in the 6 to 20 μs time frame.
3) The main inductive component of winding current. This component very often includes
superimposed large amplitude oscillations due to travelling waves in the winding. It is the slowest
and fundamental frequency of the winding current, and for large transformers may not reach a peak
until some 100 to 200 μs after the start of the record.
Depending on the type of winding tested, the relative prominence of these three components can vary
widely; for example, a multilayer winding of an instrument transformer will have very high self-inductance
and therefore a negligible inductive component of current, while the capacitive component will be
relatively large. On the other hand, a non-interleaved disc winding of a power transformer will have a
relatively small capacitive component, with large amplitude oscillations (travelling wave) being the most
prominent. The time period over which the impulse currents are recorded should be selected so as to
display all three components if possible, but, for the full-wave impulse applications the peak value of the
inductive component should always be included in the record.
4.7.3 Full-wave current records
The relatively low-frequency content of a full-wave impulse stimulates appreciable magnitude inductive
components of current. Impedance changes caused by turn-to-turn or section-to-section failures cause
changes to occur in the inductive current components that provide a reliable indication of transformer
winding problems.
The capacitive component of the ground current can give an early indication of a failure, provided the
failure can produce detectable change in the magnitude of that component. This depends on the extent of
the breakdown and on the value of the series capacitance of the winding. The larger the series capacitance,
the more dependable the fault indication will be.
The period of small oscillations due to mutual inductance coupling is not usually very useful for fault
detection because of their small amplitude relative to the other two components.
Note that the normal connection practice for non-impulsed winding terminals has the effect of short
circuiting those windings. By short circuiting those non-tested windings the normal inductive component of
the impulse current is increased several fold. This has the effect of reducing the relative change in the
inductive component that might be caused by a turn-to-turn fault, that is, the fault detection sensitivity of
the inductive current component is reduced. For a typical (non-interleaved) disc winding as employed in
medium and large power transformers where the capacitive component is small, reduction in the inductive
component sensitivity directly affects the overall system fault detection sensitivity to faults comprising a
small percentage of the winding. However, experience and tests indicate that ground current fault detection
24
sensitivity is adequate for all but the lower kilovoltampere (kVA), high-voltage distribution, and instrument
transformers. Due to the large inductance of these low kVA distribution and instrument transformer
windings, the inductive current may be so small that the normal current transducer detection methods are
incapable of producing reliable fault detection sensitivity when non-impulsed terminals are short circuited.
For these transformer types, alternative testing and fault detection techniques should be used.
Connecting loading resistors across the terminals of non-impulse windings can help increase the inductive
component sensitivity. Care should be taken in the selection of the loading resistance value so that the
voltage induced on the non-impulsed terminals of wye-connected windings does not exceed 80% of the
terminal BIL or, for delta-connected windings, 50% of the terminal BIL.
The type of shunt to use for the ground current recording depends on which current component is
considered the most important for fault detection. For power transformers a pure ohmic shunt is used,
sometimes with an added parallel capacitor to limit the amplitude of the capacitive component of the
record. This type of shunt is unsuitable for testing high-impedance multilayer-type windings for reasons
mentioned above. A pure capacitance shunt (paralleled by a high-value bleeder resistor) gives a more
sensitive indication and is capable of detecting a one-turn fault in several thousand, provided the untested
windings are open circuit or resistance loaded only to the extent necessary to limit the transferred voltage.
Impulse current transformers are sometimes used instead of an ohmic shunt.
4.7.4 Impulse record discrepancies
Any discrepancies between the voltage and current records being compared may be an indication of
transformer failure and so should always be investigated. However, it is understood that certain
inconsistencies may occur, especially during the first 2 μs or so of a record, due to normal test equipment
operational effects, such as variations in IG spark gap commutation time or inconsistencies in trigger
synchronization (trigger jitter). When discrepancies occur and such inconsistencies are suspected as the
cause, a demonstration of this fact needs to be made. Such a demonstration can be performed at reduced
voltage level and small variations in the position of the IG spark gaps or small changes to the triggering
time can be arranged and the outcome observed.
Discrepancies that are the result of actual winding problems will vary depending on the type of fault and
type of winding tested. A fault to ground or the frame will tend to shunt current around the winding and
reduce the magnitude of the winding current. Faults in the minor insulation of the winding (turn-to-turn
faults) will tend to reduce the impedance of the winding and therefore cause an increase in the winding
current. With power-transformer windings, faults will generally produce some initial high-frequency
disturbance followed by a shift in the inductive components of the current. When the inductive component
alone shows an increase, without any other visible discrepancy in the shape or phase of the oscillations, it
may indicate magnetic core saturation rather than a dielectric failure. This is apt to occur with small power
or distribution transformers and calls for careful demagnetization prior to the application of the full wave.
With the resistance type shunt, the part of the trace at which the discrepancy begins to show can give some
indication as to the location of the fault in terms of winding length from the impulsed end. With the
capacitive shunt, such as used for testing high impedance, low kVA multilayer windings, any sustained
minor insulation failure will only cause a gradual increase in the current, which builds up over a fairly long
time. Location of the failure by measuring the time of its occurrence is not usually possible with the
capacitive shunt.
Sometimes the increase in the inductive current may not be apparent, but changes in the shape of the higher
or intermediate frequency components may occur. This may be the case when the fault encompasses only a
small number of turns in a low kVA, high-voltage winding where all the untested windings are short-
circuited.
25
Trivial changes in the ground current, such as a slight change in slope of one of the minor peaks or a
minute ripple or spikes superimposed on the trace, often arise from causes outside the transformer, such as
grounding problems (sparking between ground return conductors) or imperfectly made impulse
connections. They can also indicate such internal problems as partial discharge or incipient breakdown.
As a general rule, whenever discrepancies occur, the normal test procedure will need to be stopped and
investigations made. If the cause of the discrepancies are found to be external to the test object, then
corrections should be made before the test can continue. If there is any doubt as to the cause of the
discrepancies, additional voltage applications need to be applied, including several 100% full waves. If the
deviation increases in magnitude, it indicates progressive dielectric failure in the transformer. If there is no
progressive increase in discrepancy, the chopped-wave test followed by several full waves should be
applied. If there is still no progressive increase in deviation or if the deviation disappears, this may indicate
that the cause of the discrepancy is a non-injurious partial discharge, such as might occur on a sharp spot on
bare metal.
Small ripples or corona spikes superimposed on the ground current trace that otherwise show no changes in
shape can be due to poor grounding of the core. They can also be caused by partial discharge in transformer
bushings. It is not unusual for such faults to only appear on the current trace, producing no discernable
changes on the less sensitive voltage trace.
4.8 Normal test procedure
During a normal impulse test the impulse voltage applications will follow the procedure outlined in the
applicable IEEE Standard Test Code. Each impulse application will produce a voltage and current record.
From the voltage records the impulse voltage level and impulse waveshape will be obtained and recorded.
If the impulse application did in fact produce the required voltage magnitude and waveshape (within the
specified tolerance), the next impulse voltage application can be made. If the voltage or waveshape was out
of tolerance then corrections should be made and that application repeated.
As successive applications are made (RFW, FOW, CW, and FW), the voltage records should be compared
one to another and the current records compared one to another and scrutinized for any differences that
might suggest equipment or test object problems. Note that when comparing a chopped wave to a full wave
record, only the portion of the records up to the time of chop can be expected to be suitable for comparison.
On those occasions when the time of chop for two chopped-wave records is identical (within 0.1 µs), the
complete chop wave records may generally follow the same oscillatory shape; if the oscillations are grossly
different or completely absent in one of the records, it may be an indication of a chopped-wave failure in
the transformer.
During the course of a test sequence, when any unexpected waveshape changes occur the normal test
procedure should be halted and an investigation made into the cause of the discrepancies. Note that when
the test object contains non-linear devices, some waveshape changes are to be expected when the non-
linear devices operate. As mentioned in 4.3, demonstration that the recorded changes were due to operation
of the non-linear devices becomes a part of the test procedure.
4.9 Troubleshooting
4.9.1 Initial investigations
Discrepancies, when they occur, may be large or small, early or late in time; they could be a single or
repetitive event; they may be obvious on both the voltage and current records or they may only be obvious
on one of the records. When a discrepancy does occur the first task is to determine if the discrepancy is due
26
to an equipment problem or a test object problem. If the cause of the problem is not obvious then a repeat
impulse application at a previously applied lower voltage level will provide a check on the test circuit
repeatability at that lower voltage level and also help determine if the problem is sensitive to the applied
voltage level. If the problem goes away at the lower voltage level, additional applications at slightly higher
voltage should be made until either the problem returns or the voltage at which the problem first occurred is
again applied to the test object. In general, if there has been an insulation failure in the test circuit, whether
in the test equipment or test object, then there may be a progressive deterioration of that insulation as
additional voltage applications are made. If the insulation mix at the point of failure is mainly solid
insulation, (not self-restoring), the voltage withstand of that point will be seriously compromised, and the
failure may reoccur at very low voltage. If the insulation at the point of failure is largely liquid or gaseous
insulation (self-restoring), the voltage withstand of that point may not be compromised so much by an
earlier failure and a full voltage withstand may be possible.
Once the voltage sensitivity of the problem has been assessed, if the cause is still not obvious, it may be
worth disconnecting the test object, rearranging the high-voltage leads, and testing the HV circuit without
the test object up to the required test voltage level. The applied waveform will change when the test object
is removed; also, there will be no test object winding current to provide an impulse current record,
however, additional information about possible equipment problem(s) may be determined.
4.9.2 Discrepancies associated with the transformer
In general, when a discrepancy between two records occurs, it may fall into one of the following three
types:
1) Major discrepancies, the voltage may suddenly collapse toward zero

2) Minor discrepancies, distortion of portions of the record(s)
3) Minor discrepancies, the record(s) may differ but only in the magnitude of some small, high-
frequency oscillations.
When discussing the above types of discrepancies, it will be assumed that the test equipment, including
measuring circuits, has been checked and functions satisfactorily:
The first type of discrepancy, type (a), for example when the voltage record appears to be chopped,
indicates a major insulation failure, possibly to ground. If the chop or rate of collapse of voltage is fast, that
is, the voltage collapse occurs within 1 μs or so, the failure involves almost all of the winding under test. If,
simultaneous with the voltage collapse, the impulse current record increases dramatically or even goes off
scale completely, the fault probably occurred directly across a large portion of the winding turns. If, at the
time of voltage collapse, the impulse current record includes a burst of oscillation that may go off scale but
settles back eventually toward zero, the fault was probably from the line end of the winding to ground, or
from the bushing lead to ground, or may even be an indication of a bushing failure. When the rate of
collapse is slower, it may indicate that some portion of the winding is included in the fault path. Slower,
damped rates of collapse may indicate that the failure is occurring by creepage along the surface of some
solid insulation.
The second type of discrepancy, type (b) is the most common (when the record(s) are distorted in shape)
and indicates failure in the minor insulation of winding(s), or sometimes a partial failure of the insulation of
the bushing. The time of the appearance of the discrepancy relative to the start of the record can serve as a
guide as to the possible location of the failure. For this second type of discrepancy the sensitivity of fault
indication by the voltage record depends to some extent on the location of the fault in the transformer and
the source impedance of the IG and connecting leads. The higher the source impedance of the IG and the
nearer to the line end is the fault, the greater will be the effect of the fault on the voltage shape at the
terminal under test. A failure remote from the line end of the winding (e.g., in a tapping region), may
produce no indication whatsoever on the voltage trace.
27
Momentary discrepancies occurring at the crest of the voltage record are often caused by partial discharge
in the bushing or sometimes at the line end of the winding. A discrepancy that persists for a longer time is
an indication of a sustained winding failure. A dielectric failure that results in short-circuiting a part of the
winding usually (but not always) reduces the impedance of the winding sufficiently to cause a reduction in
the duration of the wave tail. Failure in a non-impulsed winding, through a transferred surge, can also
produce discrepancies but generally these are much later in time.
The third type of discrepancy, type (c) where the amplitudes (but not the frequency) of small high-
frequency oscillations differ, is likely to be the result of partial discharges (PD) between highly stressed
portions of the winding. Partial discharge has the effect of increasing the test circuit losses at the point of
the PD, causing additional resistive components of current that tend to damp the local circuit oscillations.
Discrepancies at the beginning of the records can also be caused by changes in the IG spark gap
commutation time.
For liquid-cooled transformers, if a winding fault is suspected and if inspection covers are included on the
tank header, removing a cover and checking for smoke and bubbles may often confirm those suspicions and
also provide some indication of the fault location.
4.10 Dry type transformers
With dry type transformers the accessibility of the core and coils at test time provides the opportunity to
perform a test on the sensitivity of the impulse recording system. A loop of wire may be temporarily placed
around the outside surface of one of the transformer coils and several relatively low impulse voltage shots
applied, with and without the loop of wire or simulated turn, short circuited on itself. While a hard short, as
is made use of here, does not provide the same type of indication that an actual turn failure might. For
example, an actual turn failure may suddenly occur at some particular voltage in which case, up to that
point in time on the records there would be no failure; at the time of failure there would be a sudden
transition or transient produced, possibly producing a small burst of oscillations followed by a slow change
in the inductive component of the current. Whereas a hard short causes a change to the transformer
inductance over the whole of recorded time. Nevertheless, the inductance change produced by the hard
short will cause a change in the inductive component of current. The degree of change, if discernable, will
provide an indication of the sensitivity of the recording system to a single turn fault.
Figure 9 provides an indication of the type of discrepancy that can be produced by a simulated shorted turn.
Dry-type transformers may exhibit partial discharge at impulse voltage levels. This discharge increases
the apparent resistance loss of the winding resulting in a damping of ground current. As indicated in
Figure 9(c), the resulting full-wave current trace shows a reduced swing of the oscillations when
compared with the reduced full wave, but no change occurs in the phase of the peaks and valleys. This is
not necessarily an indication of failure.
28
Figure 9 —Lightning impulse current waves
Figure 9(c) is an expanded view of typical impulse current waves for one 50% full wave and one 100% full
wave. When impulse voltage partial discharge is suspected, the following procedure is recommended after
the normal series of impulse tests has been applied. Apply a series of impulse waves at 80%, 90%, and
100% levels. The changes between the waves at each level and the original 50% and 100% waves are then
assessed. A judgment is then made as to whether the changes indicate a progressive damping of oscillations
as voltage is increased, or, if there are also changes in oscillation frequency, that may suggest a failure in
the coil.
4.11 Voltage and current transformers
Transformer windings composed of many turns of fine wire have the highest inductance and lowest series
capacitance. For such windings, typical of potential transformers, the capacitance current is small and the
inductive current does not build up to a large value, even after many microseconds. Figure 10 shows the
current traces of the primary of three small distribution transformers that are typical of this class. The low-
frequency components are well illustrated in Figure 10(a), to a lesser degree in Figure 10(b), and are not
evident in Figure 10(c).
29
Figure 10 —Full-wave current records on high-inductance windings
Figure 11, the inductive component of current, shows indication of failure in a 25 kVA distribution
transformer. A large change in the low-frequency component is indicative of a failure that might result if a
large section of winding was short-circuited by the fault.
Figure 11 —Fault indication by inductive component of impulse current
In the case of small-distribution transformers and voltage transformers, grounding the winding through a
capacitor instead of a resistor for the grounding shunt produces sensitive failure detection. For such
windings a turn-to-turn fault causes very little change in the current. Further, because the inductance is so
large, whatever change does occur is of long duration and has the appearance of a slight shift of the whole
wave. There are no distinctive short-duration waveshape changes and the small overall waveshape
magnitude makes fault detection difficult. In such cases the resistance shunt is replaced with a capacitor of
such size that over the duration of the impulse the capacitor charges to a voltage sufficient to produce a
satisfactory deflection on the trace. This deflection should be equivalent to that used for recording the
voltage wave. Sweep durations of the order of 500 μs are used. When a turn-to-turn failure occurs, the
slight long duration change in the current wave is integrated by the capacitor so that at later times a failure
wave deviates more and more from the reduced wave. A bleeder resistance of the order of 100 000 ohms is
connected across the capacitor to bleed off the charges between successive impulses.
30
Figure 12 —Comparison of resistive and capacitive shunt impulse current records
In Figure 12a, the current is measured with a resistance shunt. In Figure 12b, the current waves are
integrated through a capacitor. Curves shown for unfaulted and two faulted conditions demonstrate the
extreme sensitivity of capacitance grounding when used on distribution and voltage transformers. While
fault detection could be obtained by superimposing the upper waves, especially if a sizeable deflection
were used, the lower set of waves show the fault in an obvious way. To obtain such large deviations, the
value of the capacitance through which the winding is grounded should be carefully selected. Such a
method is especially useful for production testing of a large number of duplicate units.
Another method of failure detection used for voltage transformers makes use of the voltage waveshapes
induced in the low-voltage winding.
Current transformers pose a problem in that there is no winding from line to ground, and consequently
there is no neutral current to record. On the other hand, there is little likelihood of small failures, such as
turn-to-turn; consequently, the high sensitivity of the current method is not so necessary. Often only the
voltage wave is used for failure detection. However, detection sensitivity can generally be improved by
recording the voltage produced across a resistor that connects the tank and the short-circuited secondary
winding to ground.
31
4.12 Examples of impulse waveforms
Figure 13 —Voltage and currents
Figure 13 illustrates a coil-to-coil failure near the line end of the winding. The high-voltage winding is a
continuous disc “pancake” type and the transformer is rated 20/26.6/33 MVA, 138 kV delta-13.8 kV wye.
The coil-to-coil failure is indicated when deviations occur between the reduced and full-wave traces. When
the deviations first appear near the crest of the voltage trace, a location near the line end can be suspected.
32
Figure 14 shows the same type of coil-to-coil failure as in Figure 13, but on a different transformer in a
different testing facility. The winding in this case is also a disc type and the transformer is 25 MVA with a
650 kV BIL high-voltage rating.
33
Figure 15 shows examples of a coil-to-coil failure near the taps at the middle of a winding. The unit is an
autotransformer, rated 78/104/130 MVA, 138 kV wye–69 kV wye–13.9 kV delta. The taps for de-
energized operation are near the middle of the series winding, which is a continuous disc type. The high-
voltage terminals were impulsed with a 400  resistor between the low-voltage terminal and ground.
Figure 16 —Currents (resistance)
Figure 16 is the current oscillogram recorded during investigative procedures and measured at a low-
voltage terminal. A comparison of these oscillograms shows a significant inductive current change.
34
Figure 17 illustrates partial discharge near the line end of the winding. The voltage wave forms show
appreciable change near the crest, while the current wave forms show very little inductive change.
Examination showed tracking on a barrier tube under the high-voltage winding, but there was no turn-to-
turn failure.
Figure 18 —Currents
Figure 18 is another illustration of partial discharge during the impulse test. In this case, the evidence of
partial discharge was found in a winding oil duct. In this particular case the voltage wave had not changed.
Therefore, a voltage wave is not illustrated here.
35
Figure 19 is the result of tests that show the failure from the high-voltage winding to a static shield in a 138
kV unit.
Figure 20 shows wave forms resulting from a failure of a no-load tap changer switch on a 650 kV BIL, 30
MVA unit. Examination showed tracking to have occurred across open contacts of the switch. Deviations
late in time on the current wave may indicate a source remote from the line end.
36
Figure 21 —Voltages and current
Figure 21 shows the failure of a capacitor type bushing on the high side of a 12 MVA transformer with a
450 kV BIL.
Figure 22 —Voltages and currents
Figure 22 illustrates a partial breakdown within a capacitor bushing, which provides the terminal for a 110
kV BIL winding.
37
Figure 23 —Currents
Figure 23 shows a characteristic of current wave traces termed autotransformer action. These oscillograms
show the minor oscillatory mismatch from a test on a single-phase shell type 525 kV, 333 MVA, 1425 kV
BIL autotransformer.
Figure 24 —Front-of-wave
Figure 24 shows low-voltage bushing failures during front-of-wave tests on a 220 kV, 80 MVA, 750 kV
BIL transformer.
38
Figure 25 shows a static plate problem on the tertiary winding of a 230 kV, 210 MVA, 900 kV BIL
transformer with tertiary ratings of 13.8 kV delta, 110 kV BIL, and 41 MVA.
Figure 26 indicates the normal operation of non-linear surge protection devices in the tested winding as
discussed in 4.3. In this case RFW, 75% FW, and 100% FW traces are shown.
39
Figure 27 shows a failure through considerable oil and paper between the high-voltage and low-voltage
windings of a series-connected transformer. The junction between series and main transformers
experienced overvoltages due to wave reflections.
40
Figure 28 depicts the effects of inadequate core grounding. Note that the mismatch is much more apparent
in the highly oscillatory current waves than in the voltage traces.
Figure 29 —Voltages and current
Figure 29 shows the effects of discharge from an ungrounded core shield of a shunt reactor.
41
Figure 30 indicates wave mismatches that are caused by considerations that are external to the windings. In
this instance, a flashover occurred between the transformer tank and an insufficiently grounded cooler.
4.13 Methods of presenting lightning impulse test results
A report of the impulse tests conducted on equipment can be very useful to the purchaser. It provides the
purchaser with a permanent record of the tests performed. If the purchaser does not witness the factory
tests, the report provides the source of information regarding the tests performed. Well-prepared reports can
be useful to the purchaser in educating inspectors or others who witness factory tests. To be useful to the
purchaser, the test results should include the following minimum data:
a) General information, that is, type and rating of equipment tested, serial number, date of test,
witnesses to the test, etc. See the suggested form in Figure 31.
b) A tabulation showing impulse tests conducted on each terminal including type and magnitude of
test waves. The connection of untested terminals of all windings should be described as outlined in
Figure 31.
c) Reproductions of the pertinent recordings taken during the tests are an important part of the test
report. When specified, these recordings should be properly identified and arranged so that the
necessary comparisons between full waves, chopped waves, and front-of-wave can be easily made.
Copies of recordings taken on 35 mm film should be enlarged to a size that permits direct visual
inspection.
d) It is preferred that the recordings of waves to be compared be overlaid. A less-desirable alternative
is to reproduce the recorded films of waves to be compared as closely together as possible in a
vertical array.
e) Timing waves or timing pips may be on each recording so placed that the test wave is not obscured.
Acceptable substitutes would be to mark the time scale on the recordings.
42
f) Recordings to demonstrate that the transformer has successfully withstood all the required impulse
tests generally are included in the final test report. Where the manufacturer has conducted additional
tests to explain discrepancies, etc., the pertinent recordings also should be included in the report.
Lightning Impulse Test Report
Manufacturer’s Name (Logo) Purchaser’s Name
_________________________________ ___________________________________
_________________________________ ___________________________________
Date of Test_______________________ PO No(s)____________________________
Mfr’s No(s)________________________ Serial No(s)__________________________

Winding Rating Voltage BIL Connection Tap Position for Non-linear
(kVA) (kV) (kV) Impulse Test Device
H
X
Y
Terminal Crest Wave Sweep Connection

lightning Test a Voltage shape or Time to Oscillogram time or of non-
impulse (kV) rate of rise flashover number record impulsed
applied to (us) length terminals b
a
Reduced full-wave voltage (RFW), reduced full-wave current (RFWC), full-wave voltage (FW), full-wave
current (FWC), chopped-wave voltage (CW) and front-of-wave voltage (FOW).
b
Terminals grounded (GRD), terminals connected to arresters (ARR) and terminals connected to ground through
linear resistance (RES).
NOTE_ Additional sheets need only repeat this table plus purchaser’s order number and manufacturer’s reference
number
Test witnessed by _______________________________ Date _________________
I herby certify that this is a true report based on factory test made in accordance with the latest revisions of IEEE Std
C57.12.00 & C57.12.90 OR C57.12.01 & C57.12.91
Signed ____________________ Date _________________ Approved ____________________
Figure 31 —Transformer lightning impulse test report
43
5. Switching impulse testing
5.1 Switching impulse testing techniques
When higher transmission voltages (345 kV and above) were introduced, it became apparent that switching
impulses play an increasingly important role in the design of the higher voltage power system than they do
at lower system voltages. High-frequency and high-amplitude voltages may result when transmission lines
at these high voltages are switched. The higher the system voltage and the longer the line, the higher the
switching impulse amplitude will be. The waveshapes vary considerably from one system to another. The
time to reach the crest amplitude and total time duration of these switching impulses are much longer than
those of lightning impulses.
Also, the switching impulse amplitude may vary depending on the location of the switching device with
respect to transformers and the design of the switching device. Many high-voltage power circuit breakers
have impulse-suppressing closing resistors to limit the impulse amplitude.
Since their time to crest and the total time duration are much longer than those of the lightning impulses,
the voltage distribution of these switching impulses within the winding of the transformer will be more
uniform. The distribution will be essentially on a volts per turn basis, approaching the uniform distribution
of low-frequency steady-state voltages. Because there may be non-linearity in some windings, it cannot be
generalized that the distribution will be uniform in all situations.
Since the switching impulse waveshape is somewhere between the low-frequency and lightning impulse
waveshapes, the assumption is usually made that a transformer that withstands both the low-frequency and
lightning impulse tests will also withstand the switching impulses if the magnitude of the switching impulse
crest is in the order of 80% to 85% of the lightning impulse crest value.
However, industry experience shows that this assumption does not hold true in all cases, and the switching
impulse withstand capability of a transformer cannot be merely interpolated from other tests. For this
reason, switching impulse testing of high-voltage power transformers is recommended.
The generally accepted crest value, also defined as the basic switching impulse level (BSL), for the
switching impulse is 83% of the BIL as outlined in IEEE Std C57.12.00.
Switching impulse tests on the highest voltage line terminals of a transformer may over-test or under-test
lower voltage line terminals depending upon the relative BSLs, the turns ratios, and test connections.
Switching impulse voltages are generally transferred between windings by approximately the turns ratio.
However, there are situations where untested windings of a three-phase transformer may show heavy
oscillations with considerably higher crest voltages than those calculated using turns ratio. These
oscillations have to be damped out; otherwise, they may lead to external phase-to-phase clearance
problems. Such situations should be resolved by the manufacturer and the user. Regardless of this fact, the
specified test voltage on the highest voltage terminal shall be the controlling test level. The insulation of the
other windings shall be capable of withstanding the induced voltages resulting from such tests even though
such voltages may exceed their specified BSL. In cases where the switching impulse test on the highest
voltage terminal results in an induced voltage on the other winding less than the required BSL for that
winding, no additional test is required to demonstrate switching impulse insulation withstand capability.
Switching impulse tests are performed by applying or inducing the required BSL voltage, line to ground, on
each high-voltage line terminal. For unique applications, phase-to-phase switching impulse tests can be
performed. Since the phase-to-ground switching impulse testing has been accepted for most applications,
phase-to-phase switching impulse tests are special tests. If the need arises, the user should include such
tests in the transformer specifications. Testing techniques and related requirements (voltage level, duration,
connections, etc.) for these tests should be agreed upon by the user and the transformer manufacturer prior
44
to designing the transformer. This should be done to ensure that it is practical to perform the specific phase-
to-phase switching impulse test.
The scope of this switching impulse section is limited to Class II power transformers (high-voltage rating
of 115 kV and above). At lower voltage ratings, switching impulse tests are not normally of concern, nor
specified.
5.2 Switching impulse waveshapes
Switching impulses to which transformers are subjected in service vary considerably in waveshape and
amplitude. Among the factors that influence the waveshape and amplitude are the system’s characteristics,
grounding, and configuration, as well as the source and location of the switching event. It is not practical to
test the transformer with all possible waveshapes; therefore, a representative waveshape, as described
below, is established to provide a consistent basis for testing. Such a test also demonstrates insulation
design and manufacturing integrity in the general range of switching transient waveshapes.
Switching impulse test procedures are outlined in IEEE Std C57.12.90. The required test waveshape is
shown in Figure 32. This wave rises from zero to crest in not less than 100 s, the voltage shall exceed
90% of the required BSL crest value for a minimum uninterrupted period of 200 s, and the time to the first
voltage zero shall be at least 1000 μs.
t c - t 0  100  s
t 2 - t i  200 s
t 3 - t 0  1000  s
E = REQUIRED BSL
Figure 32 —Switching impulse voltage waveshape
IEEE Std 4 describes a general switching impulse wave that rises to the crest value in 250μs and a time to
half value of 2500 μs. However, the 250/2500 μs wave applies to equipment that does not have a magnetic
circuit that can saturate.
The time duration of high-voltage switching impulse waves can cause the core to saturate such that air core
conditions then exist. When this occurs, the tail of the wave will decay rather rapidly to zero making it
difficult to control the tail duration. So as to extend the time to saturation the magnetic circuit is usually
magnetized in the opposite polarity prior to the start of the test such that the magnetic circuit will not
saturate as quickly when the test waves are applied. This can be accomplished by passing a small direct
current through the winding prior to the full level test. Alternatively, reduced level impulses of opposite
polarity may be applied before each full switching impulse application. The preferred method of pre-
45
biasing the core is by application of opposite (i.e., positive) polarity impulses of approximately 50% test
level. To achieve the most consistent tail duration results, the core should be driven to the same remnant
state (preferably saturation), prior to each full voltage shot. Saturation is achieved when the time to the first
voltage zero of successive applications of the opposite polarity reduced impulse remains constant (does not
move forward in time) on consecutive impulse applications. The number of shots required to achieve this
will depend on the transformer design and the voltage level chosen for the reduced level shots. To avoid
any problems with external flashovers during this procedure, the level of such positive polarity pre-
magnetizing impulses should not exceed 50% to 60% of the rated test voltage.
5.3 Switching impulse test circuit
5.3.1 Basic circuit
The basic test circuit parameters are the same as shown in Figure 2. An impulse generator is used to apply
the voltage to the transformer. The waveshape and amplitude are obtained with the impulse generator
adjustments, wave-shaping resistor, and possibly an external load capacitance.
The required switching impulse voltage amplitude may be obtained by either applying the impulse directly
to the high-voltage winding or by inducing it from a lower voltage winding into the high-voltage winding.
The direct application of switching impulse to high-voltage terminal is preferred. However, if a lower
voltage winding is used for inducing the impulse into the high-voltage winding, the applied impulse wave
should be monitored on an expanded scale to ensure that higher-than-intended-level voltage spikes are not
applied to the windings. In either case, the test measurements shall be made on the highest voltage winding.
The switching impulse voltage applied to the lower voltage winding is stepped up into the high-voltage
winding approximately by turns ratio.
Negative polarity is recommended for switching impulse tests on transformers.
Tap connections can significantly influence voltages developed within windings and from winding turns to
ground during the switching impulse test. Unless the user specifies otherwise, the choice of tap connections
for all windings shall be made by the manufacturer. Regardless of tap connection, a switching impulse test
on the highest voltage terminal results in an induced voltage on all turns.
5.3.2 Connection of non-impulsed terminals
Since switching impulse voltages are induced into other windings approximately by turns ratio, the
connection of terminals in the induced windings is important. It should be noted that due to the complex
nature of interwinding capacitive and inductive coupling in transformers, some transformer designs may
cause there to be oscillations superimposed on the voltages transferred to the non-impulsed terminals. This
can result in interphase voltage stress higher than required by the test standard. In such cases high ohmic
value resistors of a sufficient voltage rating may be connected to the non-impulsed terminals to achieve an
appropriate degree of damping. However, the degree of damping applied should not cause the transformer
to load the test equipment to a degree that compromises its ability to produce the required switching
impulse wave on the terminal under test.
5.3.2.1 Connection of non-impulsed terminals (single-phase transformers)
The tester should ground the non-impulsed terminals having the same instantaneous polarity as the
grounded terminal in the winding being tested. Though not required, a good general practice is to monitor
non-impulsed terminals for verification of voltage magnitude.
46
Some examples for single-phase transformer test connections are shown in Figure 33.
Figure 33 —Single-phase transformer test connections
5.3.2.2 Connection of non-impulsed terminals (three-phase transformers)
Since switching impulse voltages are induced in other windings approximately by turns ratio, core
geometry and internal connections have a significant effect on how non-impulsed terminals are connected
on three-phase transformers. Test connections will also be determined by whether phase-to-phase testing or
line-to-ground testing is being performed.
47
Figure 34 outlines most of the common methods of connections used for switching impulse for three-phase,
three-legged, core-type transformers.
Figure 34 —Three-phase transformer test connections (three limb core)
Figure 35 outlines the most common methods used for connection of five-legged, core-type, and shell-type
transformers. This guide describes commonly used test connections and does not preclude the use of other
suitable test connections as determined by the manufacturer.
48
Figure 35 —Three-phase transformer test connections (five-limb core)
5.4 Measurement of switching impulse voltage
The measurement of switching impulse voltages involves methods and equipment similar to those used in
lightning impulse tests. Due to the longer front and tail time on the switching impulse wave, some
differences in methodology and instrumentation are required. Measurement of a switching impulse is
usually less demanding than of a lightning impulse. Less bandwidth is required, grounding requirements
are less stringent, and high-voltage traveling wave effects are not usually of concern.
Figure 36 shows a typical test circuit used for switching impulse tests. The measurement devices in Figure
36 include the voltage divider and the oscilloscope or transient voltage recorder.
49
Figure 36 —Switching impulse test circuit
5.4.1 Switching impulse voltage dividers
The following two basic types of voltage dividers are used for switching impulse testing:
1) Capacitive
2) Damped capacitive
The output of the voltage divider is connected to the oscilloscope or transient recorder by either a coaxial
cable or some other means such as a fiber-optic transmission link for transmitting the waveform. With any
system it is necessary to design the overall circuit including the voltage divider, the cable, and the
oscilloscope or recorder such that the signal into the recorder has the same waveshape as the voltage across
the voltage divider without excessive distortion or oscillations due to the voltage divider circuitry.
To ensure that the switching impulse waveshape measured at the oscilloscope is correct, the capacitive
voltage divider circuit (including the voltage divider, the interconnecting cables, and the oscilloscope)
should have a flat frequency response in the range of 50 Hz to 20 kHz. Damping resistance should be used
with the interconnecting cables to limit oscillations due to traveling waves. Before using a capacitive
voltage divider circuit for the first time, frequency response tests and low-voltage impulse wave
verification tests should be performed to ensure that the switching impulse will be correctly reproduced.
For convenience in switching impulse testing, capacitance tap type bushings of the unit being tested are
commonly employed as the high-voltage capacitor for the voltage divider circuit. Such voltage dividers
should correctly reproduce the test waveshape as outlined previously.
50
Figure 37 —Typical capacitance voltage divider circuit
Figure 37 shows a typical capacitive voltage divider. The divider ratio is obtained from Equation (6):
C1  C2  C3  C s
R (6)
C1
where
R is the divider ratio

C1 is the high-voltage arm of the voltage divider
C2 is the low-voltage arm of the divider
C3 is the cable capacitance
Cs is the input capacitance of the oscilloscope
RT is made equal to the characteristic impedance of the coaxial cable and effectively terminates the cable at
high frequencies. The time constant of Rs, the oscilloscope input impedance, and Cs, the low-voltage arm
capacitance, shall be large enough to minimize measurement errors. Ordinary laboratory type oscilloscopes
may be used for switching impulse test recording, providing that they are adequately shielded.
Figure 38 shows a typical damped capacitive voltage divider circuit. The damped capacitive voltage divider
has a wide frequency response (Hz to MHz). The damped capacitive voltage divider circuit contains
resistance at the termination of the coaxial cable to eliminate traveling wave reflections and distortion of
the waveform. The damped capacitive voltage divider has the advantage that high-frequency transient
oscillations in the MHz range are damped by the resistance in the voltage divider.
51
Figure 38 —Typical damped capacitive voltage divider
5.4.2 Recording devices
The measurement device for recording the impulse waveshape may be either an analog oscilloscope or a
digital transient recording device. The recording equipment used for switching impulse tests is essentially
identical to that used for lightning impulse tests with lower required bandwidth and sampling rate. Since the
time associated with switching impulse waves is longer than that associated with lightning impulse waves,
the sweep time for the oscilloscope or transient recorder should be increased accordingly. The
recommended sweep times are 2000 to 5000 μs.
5.5 Switching impulse failure detection
Switching impulse test failures are primarily detected by analysis of test voltage oscillograms of all applied
or induced transients, including reduced voltage bias transients. Specifically, each voltage oscillogram is
scrutinized for recognizable indications of failure. The underlying principle for this approach is based on
the dependency of the test voltage waveshape on the high-impedance open circuit state of the transformer
under test. Turn-to-turn failures, partial winding failures to ground, and other problems, even on windings
other than the high-voltage winding, caused recognizable indications on the voltage oscillograms. Figure
39, Figure 40, and Figure 41 illustrate typical traces of reduced waves or full waves for successful tests
(normal) and failure indications.
Figure 39 —Typical normal reduced or full wave (SI)
52
Figure 40 —Core saturation (not a failure)
Figure 41 —Typical switching impulse failure
Acoustic detection by use of acoustic sensors or just careful listening for unusual sounds during the test
may be used as a supplemental failure detection or diagnostic method.
Although superimposing reduced and full-voltage oscillograms in total is not practical, the waveshapes
should match until the point of saturation. In most cases, saturation will be obvious in that it will occur at a
relatively long time as shown in Figure 40. The voltage wave will decay rather quickly to zero, yet it will
not chop directly to ground as would occur for most switching surge failures. Successive oscillograms may
differ slightly because of the influence of magnetic saturation on impulse duration and test circuit used.
Ground current oscillograms are generally not necessary for failure detection, but can provide useful
information to confirm suspected failures, and aid in diagnosing the problem.
5.6 Switching impulse and non-linear devices
Depending upon the transformer design, non-linear protective devices may be built into the transformers.
These devices may be connected across the whole, or sections of, windings. Their purpose is mainly to
limit transient over voltages, which may be impressed or induced across the windings during lightning
impulse surges, to safe levels as described in 4.3. Note that during switching impulse tests the non-linear
devices shall not limit the specified switching surge level at the highest voltage terminal.
5.7 Methods of presenting switching impulse test results
A report of the switching impulse tests conducted on the transformer can be very useful to the purchaser. It
provides the purchaser with a permanent record of the test performed. If the purchaser does not witness the
factory tests, the report is a source of information regarding the tests performed. Well-prepared reports can
be useful to the purchaser in educating inspectors or others who witness factory tests. To be useful, the test
results should include the following minimum data:
53
a) General information—type and rating of equipment tested, serial number, date of test, witnesses to
the test, etc.
b) A tabulation showing switching impulse test conducted on each terminal including type and magnitude
of test waves. Actual test voltages (not nameplate BSL) should be recorded. Actual waveshape should
be recorded rather than a statement of: the standard wave (e.g., 100/200/1000 microsecond wave
applied). The connection of non-impulsed terminals of all windings should be described.
c) Reproductions of the pertinent recordings taken during the tests are an important part of the test report.
When specified, these recordings should be properly identified and arranged so that the necessary
comparisons between the full waves and the reduced waves can be made easily. Copies of the
oscillogram recorded on 35 mm film should be enlarged to a size that permits direct visual inspection.
Switching Impulse Test Report
Manufacturer’s Name (Logo) Purchaser’s Name
_________________________________ ___________________________________
_________________________________ ___________________________________
Date of Test_______________________ PO No(s)____________________________
Mfr’s No(s)________________________ Serial No(s)__________________________

Winding Rating Voltage BIL Connection Tap Position for Non-linear
(kVA) (kV) (kV) Impulse Test Device
H
X
Y
Terminal Crest Waveshape Sweep Connection

switching Test a Voltage (µs) Oscillogram time or of non-
impulse (kV) tp td t0 number record impulsed
applied to: length terminals b
a
Reduced switching voltage (RSI), full switching voltage (FSI),
b
Terminals grounded (GRD), terminals connected to arresters (ARR) and terminals connected to ground through
linear resistance (RES).
NOTE_ Additional sheets need only repeat this table plus purchaser’s order number and manufacturer’s reference
number
Test witnessed by _______________________________ Date _________________
I herby certify that this is a true report based on factory test made in accordance with the latest revisions of IEEE Std
C57.12.00 & C57.12.90
Signed ____________________ Date _________________ Approved ____________________
Figure 42 —Transformer switching impulse test report
54
6. Grounding practices
6.1 General
It is the intent of this clause to provide some insight into the conditions that affect the measurement of
impulse waves. A complete analysis of all ground conditions cannot be given because each test setup is
different. Once the philosophy of grounding is understood, compromises can be made to assure the most
accurate and safest measurement.
The currents flowing in the impulse circuit generally are fairly large and have high rates of change (di/dt).
Consequently, a voltage drop exists between points connected by a conductor through which an impulse
current flows. Because of this, it is difficult to hold two different points at the same potential or, stated
another way, to have two different points at ground potential.
The difference in voltage between two points will depend upon the length of the interconnecting lead and
the rate of change of the current flowing in the lead. The voltage difference can be substantial. For
example, if a current changing at a rate of 1000A/μs flows through a wire 10 ft (3 m) long, there will exist a
potential difference along the wire of some 3000 V. This rate of change is not at all unusual for the ordinary
impulse circuit. Because of this, impulse circuits are carefully arranged. This is particularly true of the
circuits used for front-of-wave testing.
The following are two prime considerations when grounding practices are established:
1) Safety to personnel
2) Accuracy of measurements
All the devices in the vicinity of the operator should be at the same potential. If potential differences exist
there is the danger of the operator coming in contact with two pieces of equipment at different voltages. For
accurate measurement, the measuring system should be connected directly across the two points to be
measured such as the leads of a voltmeter. In some cases this would electrically elevate the chassis of the
oscilloscope with respect to other apparatus in the vicinity since the transformer under test might be located
some distance from the oscilloscope. Fulfilling these two considerations is sometimes difficult to achieve.
This is illustrated by considering several circuits.
In Figure 43 the voltage measured by the divider is between points A and B. The main current paths are
indicated by the heavy lines. On the fronts of full and chopped waves, the voltage drop between B and C is
usually negligible, and the capacitive current to the control room shield is also small. On the fronts of front-
of-waves, the drop across BC is dependent on the capacitive current that flows through the transformer and
the inductance of the lead BC. The capacitive current for large kVA, low-voltage windings may produce a
voltage drop across the lead inductance that will be almost 25% of the total voltage measured by the
voltage divider. To eliminate the voltage drop BC, the divider should be connected to point C as shown in
Figure 44, and the return lead from the transformer should be run directly to the bottom end of the impulse
generator.
55
Figure 43 —A grounding method
The method of Figure 44 is especially useful when the generator is some distance from the transformer
providing the bottom end of the generator has sufficient insulation to the ground plane and the voltage drop
between point C and the oscilloscope can be kept small.
Figure 44 —Preferred grounding method
For generators that are not insulated sufficiently at the bottom end and therefore must be grounded to the
building at their bases, the method in Figure 45 is used. In this method, it is particularly helpful to run
several connections from B to the tank (as indicated by BC and BD) and have many leads or a wide foil
from B back to the bottom end of the generator to reduce the voltage drops between these points. However,
the measured voltage will be in error by the magnitude of the voltage drop between the tank and point B.
56
Figure 45 —Grounding with grounded generator
The preferred grounding method shown in Figure 44 eliminates the measurement of the voltage drop in
lead BD.
The location of the resistance shunt for current measurement is also selected with consideration of the
ground problem. In Figure 43 with the resistance shunt located at the transformer tank, the cable sheath is
raised above ground by the voltage drop in lead BC. Current will flow from C through the cable sheath and
back to B causing disturbance on the current wave. If the shield is allowed to float at point C, sufficient
clearance from shield to inner conductor is to be provided.
Small disturbances may appear on current oscillograms, which are due to the voltage drops in the ground
leads discharging to nonconnected or “floating” metal objects. For instance, if a piece of metal was floating
electrically near the ground lead it would be possible for the lead to flash to the metal. The disturbance
indicated on the scope would be a function of the capacitance-to-ground of the metal. A large capacitance
would cause a greater disturbance since more energy would be required to charge up the capacitance. If the
floating metal was located near the measuring cables, the disturbance on the oscillogram would be even
greater.
When front-of-waves are applied, the best procedure is to locate the chopping gap directly on the bushing
of the transformer under test because of the voltage drops that develop due to the high rates of change of
current. For the low-voltage large capacitance windings, the voltage determined by the gap spacing can be
more accurate than the oscillogram record when it is not possible to obtain a well-grounded circuit. As
pointed out in the discussion, the voltage drop across lead BC in Figure 43 might be 25% of the total
voltage measured. If the gap were connected between A and B, full front-of-wave voltage would not be
applied to the transformer.
Another method employed is shown in Figure 46. In this method the shunt can be located close to the
ground mat or floor of the test area and the exposed lead from the transformer to the shunt can be kept short
thus minimizing electrostatic pickup. On the other hand, this method collects all the current flowing out of
the transformer and generally will result in a higher initial inrush current than the previously described
methods.
57
Figure 46 —Grounding with line current method
7. Impulse generator size

The factors affecting the length of the wave tail during factory tests have been discussed previously and are
to follow:
a) The output capacitance of the impulse generator

b) The rating and impedance of the transformer being tested, the particular winding being tested and
whether the non-impulsed terminals of the transformer are open-circuited, short-circuited, or
terminated with resistors, capacitors, or lightning arresters
In general, for a given tail duration (50 s), the higher the rating of a transformer and the lower the
impedance of the winding being tested, the higher must be the output capacitance of the impulse generator.
In general, the higher voltage windings of transformers have sufficiently high impedance not to burden the
impulse generator, whereas the lower voltage windings do burden the impulse generator. The impedance of
low-voltage transformer windings, particularly when coupled to short-circuited windings of the same
phase, are predominantly inductive, low in value, and therefore require high values of impulse generator
output capacitance to produce the required tail duration.
As noted earlier, to produce the required time to half value on low-impedance windings, it is necessary to
allow the test circuit to oscillate at the fundamental frequency determined by the impulse generator output
capacitance and the transformer leakage inductance. The waveform produced by this oscillation is a cosine.
Over damping of this circuit tends to reduce the time to half value thereby requiring larger values of
impulse generator capacitance to achieve the required tail time. Under damping allows the circuit to
oscillate more freely and produce the maximum available time to half value for the damped cosine wave,
(that being the point on the wave corresponding to 60 degrees), and, very large voltage reversals.
To limit the voltage reversals to a maximum of 50% it is necessary to include resistance with a value
equivalent to 25% of the critical damping resistance for the test circuit. Critical damping resistance is
determined from Equation (7):
58
R  2 Lt (7)
Cg
where
R is the critical damping resistance, Ohms

Lt is the transformer inductance, henries
Cg is the output capacitance of the impulse generator, farads
For the damped cosine wave where the damping factor limits voltage reversals to 50%, the time to half
value T2 may be determined for the R, L, C circuit from Equation (8) below:
T2  0.5  Lt  C g (8)
where
T2 is the tail time, seconds

Lt is the transformer leakage inductance, henries
Cg is the impulse generator output capacitance, farads
For the transformer test circuit the IG wave-shaping resistors utilized to obtain the rise and fall time also
serve to damp the fundamental oscillatory cosine wave.
Rearranging Equation (7) with respect to C produces
2(T2 ) 2
Cg 
Lt
Based on the assumption that the impedance of the winding under test is predominantly inductive, the value
of that inductance Lt can be obtained from the transformer nameplate parameters using Equation (9):
Z V 2
Lt  (9)
100    VA
where:
Z is the numerical value of percent impedance

V is the rated voltage class of the transformer winding under test in volts
 is 2f
VA is the rating of the transformer under test in voltamperes
Therefore, rearranging the above formulas and substituting for Lt the impulse generator output capacitance
required to test a given winding can be obtained from:
400    f  (T2 ) 2  VA
C
Z V 2
As mentioned above, this expression yields the value of the required IG output capacitance for the case
where the transformer winding under test is predominantly inductive and where the circuit damping factor
would limit the voltage reversal of the resulting damped cosine wave to 50%.
59
To illustrate the use of the above formulas we can consider the requirements to test a 33 MVA 220/3 to
115/3 kV single-phase 60 Hz autotransformer that has 13.8 kV 10 MVA tertiary winding. The impedance
of the HV windings we will arbitrarily set to 5% and the tertiary winding to 3%.
Note that, this being a single-phase transformer, the applicable voltage rating to be used in the calculation
of the required IG output capacitance will be the rated winding voltage which, for the auto-connected
windings will be the line voltage divided by the square root of three.
Using a simple spreadsheet program and the above formulas the values of Lt and C can readily be
calculated. To illustrate the relationship between the IG output capacitance and the tail time the capacitance
for a 50 s tail (C50) and a 40 s tail (C40) will be calculated. Referring to the transformer windings as
series, common and ty, the values obtained will be:
Lt (series) = 64.8 mH, C50 = 0.077 F and C40 = 0.049 F. Required BIL 900 kV
Lt (common) = 17.7 mH, C50 = 0.282 F and C40 = 0.181 F. Required BIL 550 kV
Lt (ty) = 1.51 mH, C50 = 3.29 F and C40 = 2.11 F. Required BIL 110 kV
Considering two possible IG systems that may be used in a typical manufacturers test department:
1) A 2MV system with 10 200 kV stages of 15 kJ (0.75 F) capacitors per stage
2) A 1.6MV system with 16 100 kV stages of 5 kJ (1 F) capacitors per stage
For these impulse tests the IG will be required to produce the 110% chopped-wave voltage level and may
have a voltage efficiency of say 85%, that is, the voltage output from the generator may only be some 85%
of the total charging voltage (charging voltage per stage multiplied by the number of series stages). The
minimum number of series connected impulse generator stages required for the test should therefore be
based on the required BIL and both the above factors (1.3 total). Table 3 indicates the series connection
parameters available using the two IG systems
Table 3—Output capacitance of two typical IG configurations
BIL Required Series stages Output capacitance Series stages Output capacitance
(kV) Total charge required of stages IG(1) required of stages IG(2) C
1.3*BIL (kV) IG(1) (n) C stage/n (F) IG(2) (n) stage/n (F)
900 1170 6 0.125 F 12 0.083 F
550 715 4 0.188 F 8 0.125 F
110 143 1 0.75 F 2 0.5 F
As can be seen from Table 3, both IG systems in series configuration have enough output capacitance
without paralleling stages when configured for the impulse tests on the series winding.
For the common winding impulse test, neither of the IGs have the output capacitance in series
configuration to produce a 50 s tail. IG (1) could meet the minimum 40 s tail requirement. To obtain the
minimum 40 s tail IG (2) would have to be configured with eight series stages of two stages in parallel.
For the test on the tertiary winding, IG (1) could be configured as one single stage of five stages in parallel
to produce an output capacitance of 3.75 F. This should allow the test circuit to produce a 50 s tail.
60
For IG (2), the tertiary BIL would require two IG stages in series so two stages of seven stages in parallel
would be required to obtain 3.5 F for the 50 s tail.
It should be noted that when paralleling impulse generator stages, if the IG wave-shaping resistors are of
limited thermal capacity or are of unknown thermal capacity, parallel combinations of resistors should be
used in each IG resistor location. A rule of thumb when paralleling IG capacitors would be to use as many
resistors in parallel as there are stages in parallel, for each resistor location.
61
Annex A
(informative)
Advanced processing of digital records
A.1 Introduction
With the advent of high-quality digital high-voltage impulse recorders in the eighties, a new era begun. The
use of digital processing of impulse records further enhanced the interpretation of captured records. Many
new tools were dedicated to better visibility of impulse records and better comparison of oscillograms.
Such tools include expansion (zoom) of traces in both directions, scaling and superposition of voltage or
current traces, calculation of numerical differences between two records, etc. Nowadays, any small glitch in
a trace or any small difference between two curves can be clearly exposed. These new possibilities
increased the sensitivity of the measurements to a point where it soon became clear to test engineers that
new tools would be needed in the area of interpretation of impulse test records. Two of these tools are
described herein: the transfer function and the coherence function.
A.2 Transfer function background information
An impulse test on a power transformer is made by applying a standard voltage wave to one terminal and
measuring the response of the transformer, a current wave, captured at another terminal. In this context, the
transformer under test can be thought of as a linear quadrupole or a four-terminal network (Figure A.1).
Transformer
Zs
V Zt
Voltage Shunt
Divider
Figure A.1—Transfer function circuit diagram
A reduced voltage wave is applied first. It is assumed that at reduced level, no failure occurs in the
transformer. This reduced level impulse is kept as reference. Then, a full level impulse is applied. If that
full level applied voltage wave is identical in shape to the reduced one, then, by linearity, the measured full
level response must be identical to the response measured at reduced level. If not, the system is not linear.
In other words, an impulse test is technically a linearity test. The test is successful if the transformer is
found linear, that is the waves measured at full level match the ones obtained at reduced level. Of course,
the voltage divider, the shunt and any other component in between the two are included in this linearity
measurement.
62
A.3 Transfer function theory
Let Vt be the voltage at the input terminals of the quadrupole and Io the current measured at the output of
the quadrupole. The theory states that:
In the time domain:
Vt(t) * H(t) = Io(t) (A.1)
where
Vt(t) is the input voltage record

* means convolution
H(t) is the quadrupole’s impulse response
Io(t) is the output current record
In the frequency domain, this translates to:
Vt(ω) · H(ω) = Io(ω) (A.2)
where
Vt(ω) is the Fourier transform of Vt(t)

Io(ω) is the Fourier transform of Io(t)
H(ω) is the Fourier transform of H(t) = Quadrupole’s Transfer Function
The quadrupole’s transfer function can be calculated as
H(ω) = Io(ω)/Vt(ω) (A.3)
That is, a transfer function is the quotient of the Fourier transform of the current and the Fourier transform
of the voltage. This transfer function depends only on the quadrupole being measured and is independent of
the individual shape of the input function applied (the voltage wave).
With digital records, the Fourier transform can be approximated by the use of a fast Fourier transform
algorithm known as FFT. It is based on the Fourier series, which assumes that the signal analyzed is
repetitive in time (the wave repeats itself infinitely). For the FFT the duration of the analyzed signal
becomes the period. FFT also assumes that the signal has been quantified (sampled) at equally time spaced
intervals. One interval is referred to as the sampling period. FFT transforms an array of numbers
representing the time samples of a signal into a new array of numbers representing the Fourier coefficients
associated with this signal. These coefficients are complex numbers, which can take two forms: Cartesian
(real and imaginary components) or polar (magnitude and phase). To most engineers, it is easier to
understand the polar form. In particular, the magnitude of a Fourier component means the content of a
given frequency component in the signal. Calculating the transfer function of a system based on impulse
records can be somewhat viewed as applying a series of discrete frequencies at its input and measuring the
response of the system at its output. One of the major difficulties of this calculation comes from the fact
that not all frequency components in the input signal are of equal magnitude. In fact, many of the Fourier
coefficients of the input signal have a very small magnitude. It becomes very difficult if not impossible to
calculate the transfer function of the system at those frequencies which are very small in magnitude.
Numerically, we can say that this leads to a 0/0 mathematical indetermination and the result is totally
unpredictable. In order to correctly assess the reliability of the calculation, one must have an idea of how
reliable the input coefficients are. One way is to have an indicator of the magnitude of the input signal as a
function of the frequency.
63
Figure A.2 shows the magnitude of the transfer function as a function of frequency expressed in MHz.
Figure A.3 shows the indicator of reliability of the input signal as a function of frequency. As long as this
indicator remains at 1, the input signals are assumed reliable. If the indicator functions fall below 1, then
the transfer function calculations should be considered less reliable at those frequencies. Here, it can be
seen that the transfer function graphs start diverging when the indicator falls below 1. Without such an
indicator, it becomes difficult to evaluate the significance of transfer functions.
Note that the word “impulsion” printed across the top of the time domain and frequency domain records
may be translated to mean “impulse.”
In this particular implementation, the indicator function is simply an input signal level indicator, which is
equal to 1 if the input is above or equal to a pre-selected threshold at a given frequency, and is proportional
to the input signal if it falls below the threshold. This threshold has been selected at the level where noise
becomes a significant part of the input signal and would invalidate the calculations. It depends on the
sensitivity of the measuring equipment and its immunity to electromagnetic noise. Each lab should
establish its own threshold based on a careful evaluation. How to determine this threshold is beyond the
scope of this document.
3 IMPULSION STANDARD 2 IMPULSION STANDARD Transfer Function (MHz)

30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.2—Example of an overlay of FW and RFW transfer functions
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.3—Reliability indication for the transfer function
Special software algorithms can be used to make the transfer function less sensitive to noisy signals by
using the indicator function. Figure A.4 is the transfer functions calculated for the same set of impulse
records as used in Figure A.2 with such a smart algorithm.
64
Up to 900 kHz, Figure A.4 is identical to Figure A.2 but, in the area where the indicator falls below 1, the
transfer function magnitude has been reduced significantly. Unfortunately, most impulse recording systems
currently available on the market do not provide such an indicator function. Some of them use one
internally without the user’s knowledge. Hopefully in future revisions of impulse analyzing software, we
will find such an indicator.
3 IMPULSION STANDARD 2 IMPULSION STANDARD Transfer Function (MHz)

30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.4—Transfer function with reduced noise sensitivity
A.4 Application of transfer function
Three-phase transformer transfer function records obtained from an impulse test on the high-voltage
terminals are shown in Figure A.5, Figure A.6, Figure A.7, Figure A.8, Figure A.9, and Figure A.10.
65
83 IMPULSION STANDARD 79 IMPULSION STANDARD Voltage

500
-500
-1000
-1500
-2000
-2500
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.5—Phase 1 RFW and FW voltage records
IMPULSION STANDARD IMPULSION STANDARD Current

1000
800
600
400
200
-200
-400
-600
-800
-1000
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.6—Phase 1 RFW and FW current records
66

500
-500
-1000
-1500
-2000
-2500
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05

1000
800
600
400
200
-200
-400
-600
-800
-1000
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
67

500
-500
-1000
-1500
-2000
-2500
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05

800
600
400
200
-200
-400
-600
-800
-1000
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
68
Phase 1 and Phase 3 current waves overlay very well, however, Phase 2 current waves show small
deviations. Below (Figure A.11 through Figure A.16) are the transfer functions extracted from the previous
records. It can be clearly seen that phase 1 and 3 show a much better match than phase 2, which shows
significant differences at frequencies where the indicator functions show reliable signal levels. Phase 2 was
retested with acoustic sensors placed on the tank and the presence of a failure was confirmed. It should be
noted that the transfer function was used as a tool to investigate further the questionable time-based records
of the test. Also, the transformer was not rejected based on transfer function alone, but rather with the
signals from the acoustic sensors. Transfer function was the tool which put the test engineer on the right
track.
IMPULSION STANDARD IMPULSION STANDARD Transfer Function (MHz)

8
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.11—Phase 1 transfer function
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.12—Phase 1 reliability indicator
69

6
0
0 0.2 0.4 0.6 0.8 1 1.2
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2

5
4.5
3.5
2.5
1.5
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
70
A.5 Transfer function of chopped-wave records
Examples of the application of transfer function using chopped-wave voltage and current records (Figure
A.17 and Figure A.18) from an impulse test on a single-phase transformer are shown.

400
200
-200
-400
-600
-800
-1000
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05 2.50E-05
Figure A.17—Two chopped-wave voltage records

1000
800
600
400
200
-200
-400
-600
-800
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05 2.50E-05
Figure A.18—Two chopped wave current records
71
The fact that the times to chop for the two records are different prevents a direct overlay and comparison of
the waves after the chopping. Transfer function theory states that the transfer function should not be
dependent on the individual shape of the input function. The transfer function graphs of those chopped
waves are shown in Figure A.19 and Figure A.20.

7
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.19—Transfer function of two chopped waves
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.20—Reliability Indicator for the two chopped waves
Up to the point where the indicator function falls below 1, the two curves match perfectly as expected. If
the transfer function is truly independent of the shape of the exciting function, it should be even possible to
compare chopped waves to full waves.
72
A.6 Transfer function of a full wave and chopped wave

As depicted in Figure A.21 through Figure A.24, from 100 kHz up to where the indicator functions remain
above 1, the transfer functions are very close as the theory predicted. The differences in the very low-
frequency range are due to other limitations of FFT associated with the fact that the full-wave signals have
been truncated and not measured until they reach a zero steady state. This prevents a proper calculation of
the low-frequency part of the transfer function. A detailed explanation would be beyond the scope of this
document.

400
200
-200
-400
-600
-800
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.21—Full wave and chopped wave voltage records

800
600
400
200
-200
-400
-600
-800
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.22—Full wave and chopped wave current records
73

7
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.23—Transfer function of the full wave and chopped wave records
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.24—Reliability indicator for the transfer function
74
A.7 Transfer function example with test equipment problems
Differences are clearly visible between the reduced and full waves both in voltage (Figure A.25) and
current (Figure A.26). It was initially concluded that the transformer had failed. But a look at the transfer
functions (Figure A.27 and Figure A.28) raised a doubt.

20
-20
-40
-60
-80
-100
-120
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.25—Reduced full wave and full wave voltage records

100
-100
-200
-300
-400
-500
-600
-700
-800
-900
-1000
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
Figure A.26—Reduced full wave and full wave current records
75

16
14
12
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.27—Transfer function with equipment problems
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure A.28—Reliability indicator for the transfer function
The match was surprisingly good up to the frequency where the signal indicator fell below 1. It was
decided then to look closely at the test circuit. A bad contact was found in the impulse generator. It was
repaired and the transformer was retested successfully. Again, the transfer function led the people
conducting the tests in the right direction.
A.8 Coherence function
The coherence function was developed to act as a smart indicator of the reliability of the transfer function
estimates. This function is a software algorithm that calculates the coherence of two or more signals as a
function of frequency. The result is a curve that relates the degree of coherence between 1 and 0. A
coherence of 1 indicates that the two signals demonstrate a linear relationship at the specified frequency.
When noise is superimposed on the measured signals the coherence function drops. The coherence function
can be very useful to determine the non-linearity in a given test. The drop in coherence can be caused by
improper digitization, injected noise in the measurement, or true circuit non-linearity. If reduced impulses
of different levels are applied to a transformer, their coherence can be calculated and the resulting function
will show at which frequencies the signals can be reliably compared.
76
Coherence function should be used to locate the areas of the transfer function that are valid for comparison.
A slight difference in the coherence function between a set of waveforms taken prior to the transformer test
and a set taken after the test does not indicate a failure of the transformer. However, changes in the transfer
function at frequencies where the coherence function is near 1 should be investigated. If the transformer has
failed causing a true non-linearity in the unit, the coherence function will display a drop. This is because
the failure inherently causes a non-linearity. It is therefore not correct to ignore frequencies where the
coherence function displays a decrease. If these frequencies displayed a high coherence between reduced
wave impulses then they should have a high coherence for the full wave unless there is a failure. The
waveforms would be suspect if the frequencies displayed high coherence between reduced wave shots and
the coherence at those frequencies drops when the full waveform is included in the analysis. In order to
utilize the features of the coherence function, the application of a couple of reduced waves prior to the full
is required. Any reduced shots taken during the setup of the generator can be used to develop a coherence
waveform for the particular test setup.
77
Annex B
(informative)
Bibliography
Bibliographical references are resources that provide additional or helpful material but do not need to be
understood or used to implement this standard. Reference to these resources is made for informational use
only.
[B1] Abetti, P. A., “Bibliography of the Surge Performance of Transformers and Rotating Machines,”
AIEE Transactions, vol. 77, pt. 3, pp. 1150–1168, 1958.
[B2] Aeschlimann, H., “Insulation Stresses on Transformer Winding Coils Due to Chopped Waves,”
CIGRE Report no 126, 1954.
[B3] Aicher, L. C., “Experience with Transformer Impulse Failure Detection Methods,” AIEE
Transactions, vol. 67, pp. 1621–1627, 1948.
[B4] Aicher, L. C., “Some Aspects of Ground Current Measurements During Transformer Impulse
Tests,” AIEE Transactions, vol. 79, pt. 3, pp. 1101–1107, 1961.
[B5] AIEE, “Report on Lightning Arrester Applications for Stations and Substations,” AIEE Committee
Report, AIEE Transactions, vol. 76, pt. 3, pp. 614–627, 1957.
[B6] Allis-Chalmers Manufacturing, Transformer Reference Book. Milwaukee, WI, Allis-Chalmers
Manufacturing Co., 1961. [33 important technical transformer articles reprinted from Allis-Chalmers
Electrical Review.]
[B7] Anderson, J. G., “Ultrasonic Detection and Location of Dielectric Discharges in Insulating
Structures,” AIEE Transactions, vol. 75, pt. 3, pp. 1193–1198, 1956.
[B8] Bean, R. L., Chackan, Jr., N., Moore, H. R., and Wentz, E. C., Transformers for the Electric Power
Industry. New York: McGraw-Hill, 1959.
[B9] Beavers, M. F., Holcomb, J. E., and Leoni, L. C., “Magnetization of Transformer Cores During
Impulse Testing,” AIEE Transactions, vol. 74, pp. 118–123, 1955.
[B10] Beldi, F., “Impulse Testing of Transformers, Measuring Procedures and Test Circuits,” CIGRE,
report no. 112, 1952.
[B11] Bellaschi, P. L., “Characteristics of Surge Generators for Transformer Testing,” AIEE Transactions,
vol. 51, pp. 936–951, 1932.
[B12] Bellaschi, P. L., and Vogel, F. J., “Factors Influencing the Impulse Coordination of Transformers
II,” AIEE Transactions, vol. 53, pp. 870–876, 1934.
[B13] Blume, L. F., Boyajian, A., Camilli, G., and Montsinger, V. M., Transformer Engineering, 2nd
edition, chapters 27 and 28. New York: John Wiley and Sons, 1951.
[B14] Craggs, J. D., and Meek, M. J., High Voltage Laboratory Techniques. London: Butterworths
Scientific Publications, 1954.
[B15] Creed, F. C., The Generation and Measurement of High Voltage Impulses. Princeton, NJ: Centre
Book Publishers, 1989.
[B16] Digital Recorders for Measurements in High-Voltage Impulse Tests, IEEE Panel Session, San
Diego, CA, July 30, 1991.
[B17] Digital Techniques in High-Voltage Measurements, IEEE/CIGRE International Symposium,
Toronto, Ontario, Canada, October 28–30, 1991.
78
[B18] Elsner, R., “Detection of Insulation Failures During Impulse Testing of Transformers,” CIGRE,
report no. 101, 1954.
[B19] Focust, C. M., Kuehni, H. P., and Rohats, N., “Impulse Testing Techniques,” General Electric
Review, vol. 35, pp. 358–366, 1932.
[B20] Ferguson, J. M., “Discussion of Rippon and Hickling’s, The Detection of Oscillographic Methods of
Winding Failures During Impulse Tests on Transformers,” IEE Proceedings, vol. 96, pt. 2, p. 779, 1949.
[B21] Glaninger, P., “Stoßspannungsprüfung an Elektrischen Betriebsmitteln Kleiner Induktivität,” ISH
Zürich Beitrag, 2.1-OS, 1975.
[B22] Hagenguth, J. H., and Meador, J. R., “Impulse Testing of Power Transformers,” AIEE Transactions,
vol. 71, pt. 3, pp. 697–704, 1952.
[B23] Hagenguth, J. H., “Impulse Testing of Transformers According to American Practice,”
Elektrotechnische Zeitschrift, Ausgabe A, vol. 76, no. 23, pp. 828–831, 1955.
[B24] Hagengurth, J. H., “Progress in Impulse Testing of Transformers,” AIEE Transactions, vol. 63, pp.
999–1005, 1944.
[B25] Hawely, W. G., Impulse-Voltage Test, London: Chapman and Hall, 1959.
[B26] Hickling, G. H., “Impulse Testing of Transformers with Special Reference to Failure Detection,”
National Physical Laboratory Symposium on Precision Electrical Measurements, Toddington, Middlesex,
England, paper 23, 1954.
[B27] Hylten-Cavallius, N., “The Technique and Requirements of Impulse Generator Circuits,” National
Physical Laboratory Symposium on Precision Electrical Measurements. Toddington, Middlesex, England,
paper 22, 1954.
[B28] Karady, G., “The Stresses Produced by Chopped Wave Tests in Different Transformer Connections
and Their Variation with Wave Duration,” CIGRE, report no. 114, 1958.
[B29] Langlois-Berthelot, R., Monnet, M., Derippe, J., and Favie, R., “Chopped Wave Tests of
Transformers with a Wave of Reduced Steepness,” CIGRE, report no. 138, 1956.
[B30] Lengnick, G. W., and Foster, S. L., “The Use of Neutral Current Measurements During Chopped-
Wave Impulse Tests on Transformers,” AIEE Transactions, vol. 76, pt. 3, 1957, pp. 977–980.
[B31] Liao, T. W., Nye, J. R., Brustle, H. H., and Anderson, J. G., “Corona Studies, In Relation to
Insulation,” AIEE Transactions, vol. 74, pt. 3, pp. 1046–1050, 1955.
[B32] Liao, T. W., and Kresge, J. S., “Detection of Corona in Oil at Very High Voltages,” AIEE
Transactions, vol. 73, pt. 3, pp. 1389–1395, 1954.
[B33] Miller, C. J., and Wittibschlager, J. F., “Management of Steep Front Impulse Waves with an Isolated
Screen Room Installation,” AIEE Transaction, vol. 77, pt. 1, pp. 262–271, 1958.
[B34] Montsinger, V. M., “Coordination of Power Transformers for Steep-Front Impulse Waves,” AIEE
[B35] NEMA, National Electrical Manufacturers Association’s Standards for Transformers, NEMA
Publication No. TR1-1980, 1980.
[B36] Nueve-Eglise, J., “Transformer Impulse Tests with Special Reference to Fault Detection Requiring
the Interpretation of Small Irregularities in Oscillograms,” National Physical Laboratory Symposium on
Precision Electrical Measurements, Toddington, Middlesex, England, paper 24, 1954.
[B37] Preston, L. L., “Chopped Wave Impulse Testing of Transformers,” CIGRE, report no. 131, 1956.
[B38] Provoost, P. G., “Impulse Testing of Transformers,” CIGRE, report no. 115, 1954.
[B39] Provoost, P. G., “Testing of Transformers with Special Reference to the Assessment of the Results
of Such Tests,” National Physical Laboratory Symposium on Precision Electrical Measurements,
Toddington, Middlesex, England, paper 25, 1954.
79
[B40] Purvis, W. J., “Impulse Failure Detection Methods in Transformer Testing Proceedings,” High
Voltage Symposium, Ottawa, Ontario, Canada, paper no. 14, 1956.
[B41] Rabus, W., “The Impulse Voltage Difference Method for the Detection and Location of Faults
During Full Wave Impulse Tests on Transformers,” CIGRE, report no. 139, 1954.
[B42] Rippon, E. C., and Hickling, G. H., “The Detection by Oscillographic Methods of Winding Failures
During Impulse Tests on Transformers,” JIEE, vol. 96, pt. 2, pp. 769–790, 1949.
[B43] Rohlfs, A. F., and Uhlig, E. R., “A Discussion of Aicher’s Paper,” [ibid.] AIEE Transactions, vol.
79, pp. 1103–1107, 1961.
[B44] Ross, C. W., and Curdts, E. B., “Considerations of Specifying Corona Tests,” AIEE Transactions,
vol. 75, pt. 3, pp. 63–67, 1956.
[B45] Stenkvist, K. E., “Chopped Wave Impulse Testing,” CIGRE, report no. 143, 1956.
[B46] Stenkvist, K. E., “Study of Fault Detection and Failure Location During Surge Testing of
Transformers,” CIGRE, report no. 129, 1952.
[B47] Stewart, H. C., and Holcomb, J. E., “Impulse Failure Detection Methods as Applied to Distribution
Transformers,” AIEE Transactions, vol. 64, pp. 640–644, 1945.
[B48] Thomason, J. L., “Impulse Generator Circuit Formulas,” AIEE Transactions, vol. 53, pp. 169–176,
1934.
[B49] Vogel, F. J., “Corona Voltages of Typical Transformer Insulation Under Oil, Parts I and II,” AIEE
Transactions, vol. 57, pp. 34–36, 194–195, 1938.
[B50] Vogel, F. J., and Montsigner, V. M., Impulse Testing of Commercial Transformers, AIEE
[B51] Westinghouse Electric Manufacturing Corp. “Surge Testing,” in Electrical Transmission and
Distribution Reference Book, Monroeville, PA, 1944.
[B52] Westinghouse Electric Manufacturing Corp., Electrical Transmission and Distribution Reference
Book, 4th edition, Monroeville, PA, 1950.
80

Ieee C57.98-2011

Uploaded by

Copyright:

Available Formats

Ieee C57.98-2011

Uploaded by

Document Information

Original Title

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

Ieee C57.98-2011

Uploaded by

Copyright:

Available Formats

IEEE Guide for Transformer Impulse

IEEE Power & Energy Society

IEEE Guide for Transformer Impulse

Approved 11 December 2011

IEEE-SA Standards Board

The Institute of Electrical and Electronics Engineers, Inc.

Copyright © 2012 by The Institute of Electrical and Electronics Engineers, Inc.

PDF: ISBN 978-0-7381-7194-4 STD97201

Comments on standards should be submitted to the following address:

Laws and regulations

Updating of IEEE documents

Arthur Molden, Chair

Reto Fausch Bertrand Poulin Pierre Riffon

Satish Aggarwal Michael Haas Bansi Patel

Richard H. Hulett, Chair

Masayuki Ariyoshi Jim Hughes Gary Robinson

Satish Aggarwal, NRC Representative

3. Impulse Testing .......................................................................................................................................... 2

4. Lightning impulse test circuits ................................................................................................................... 5

5. Switching impulse testing......................................................................................................................... 44

7. Impulse generator size.............................................................................................................................. 58

Annex A (informative) Advanced processing of digital records .................................................................. 62

Annex B (informative) Bibliography............................................................................................................ 78

IEEE Std 4TM, IEEE Standard Techniques for High-Voltage Testing. 1, 2

3.2 Impulse waveshapes

2,500 microsecond record

250 microsecond record

4. Lightning impulse test circuits

4.1 Waveshape control

4.1.1 Simple circuits

Figure 2 —Lightning impulse circuit

Figure 3 —Impulse waves from simple circuits

V is the voltage across the load capacitor

T is the period of oscillation in seconds

T1 is the lightning impulse front time in seconds

T2 is the lightning impulse tail time in seconds

4.1.2 Waveshape control of transformer test circuits

4.1.2.1 The transformers’ effect on the waveshape

4.1.2.2 High-impedance windings (Lt > 100 mH)

4.1.2.3 Low-impedance windings (Lt < 20 mH)

T2 is the tail time in seconds

4.1.2.4 The Glaninger circuit

In general Rt Int > Rt Ext & Rf Int << Rf Ext

Figure 4 —The Glaninger circuit

4.1.3 Impulse testing of low-impedance windings by alternative means

4.1.3.1 Method 1: terminals tied together

Example: Let the transformer constants be represented by an equivalent circuit as shown in

Figure 5 —Low-impedance windings-connecting terminals together

4.1.3.2 Method 2: series resistance

Figure 6 —Low-impedance windings: winding grounded through a resistor

4.1.3.3 Method 3: short tail time

Figure 7 —Low-impedance windings: effects due to short length of wave tail

4.1.3.4 Low-impedance windings: winding construction

4.1.4 Connection of non-impulsed terminals

Table 1 —Maximum value of grounding resistance for system voltage class

Nominal system voltage (kV) Maximum resistance (ohms)

NOTE—The above values represent typical transmission line characteristic impedance. 3

4.2 Chopped-wave impulse testing of transformers

4.2.1 Types of chopped wave

Date of Test_ PO No(s)______

Mfr’s No(s) Serial No(s)__

Test witnessed by _______________ Date _

Signed ____________ Date _ Approved ____________