INF-8077i 10 Gigabit Small Form Factor Pluggable Module: SFF Committee
INF-8077i 10 Gigabit Small Form Factor Pluggable Module: SFF Committee
INF-8077i 10 Gigabit Small Form Factor Pluggable Module: SFF Committee
SFF Committee
INF-8077i
Abstract: This specification defines the electrical, management, and mechanical interfaces of the XFP module. The
module is a hot pluggable small footprint serial-to-serial data-agnostic multirate optical transceiver, intended to
support Telecom (SONET OC-192 and G.709 “OTU-2”) and Datacom applications (10 Gb/s Ethernet and 10 Gb/s
Fibre Channel). Nominal data rates range from 9.95 Gb/s, 10.31 Gb/s, 10.52 Gb/s, 10.70 Gb/s, and the emerging
11.09 Gb/s. The modules support all data encodings for these technologies. The modules may be used to imple-
ment single mode or multi-mode serial optical interfaces at 850 nm, 1310 nm, or 1550 nm. The XFP module design
may use one of several different optical connectors. An adaptable heatsink option allows a single module design to
be compatible with a variety of hosts.
Support: This document was adopted by the XFP MSA and is provided to the SFF for distribution and as a source
document for related SFF projects.
Documentation: This document has been prepared according to the guidelines and agreements of the XFP MSA.
The XFP MSA has the stated intention of encouraging broad and rapid industry adoption of the specification. The
XFP specification and the technologies it uses may be offered to formal standards bodies to further support the
adoption of the specification.
POINTS OF CONTACT:
SFF Chair
I. Dal Allan
14426 Black Walnut Court
Saratoga, CA 95070
Voice: (408) 867-6630
Fax: (408) 867-2115
Email: [email protected]
The SFF Committee does not require expressions of support for documents accepted as informational.
The user's attention is called to the statements of intellectual property rights and copyright in the XFP specification,
page ii. By distribution of this Specification, no position is taken with respect to the validity of these statements or of
any patent rights in connection therewith. The XFP adopter’s agreement, available from XFP Promoters, contains
agreements with respect to intellectual property rights and other rights and responsibilities.
If you are not a member of the SFF Committee, but you are interested in participating, the following principles have
been reprinted here for your information.
The SFF Committee is an ad hoc group formed to address storage industry needs in a prompt manner. When
formed in 1990, the original goals were limited to defining de facto mechanical envelopes within which disk drives
can be developed to fit compact computer and other small products.
In November 1992, the SFF Committee objectives were broadened to encompass other areas which needed simi-
lar attention, such as pinouts for interface applications, and form factor issues on larger disk drives. SFF is a forum
for resolving industry issues that are either not addressed by the standards process or need an immediate solution.
Documents created by the SFF Committee are expected to be submitted to bodies such as EIA (Electronic Indus-
tries Association) or an ANSI Accredited Standards Committee. They may be accepted for separate standards, or
incorporated into other standards activities.
The principles of operation for the SFF Committee are not unlike those of an accredited standards committee.
There are 3 levels of participation:
- Attending the meetings is open to all, but taking part in discussions is limited to member companies, or
those invited by member companies
- The minutes and copies of material which are discussed during meetings are distributed only to those who
sign up to receive documentation.
- The individuals who represent member companies of the SFF Committee receive documentation and vote
on issues that arise. Votes are not taken during meetings, only guidance on directions. All voting is by letter
ballot, which ensures all members an equal opportunity to be heard.
Material presented at SFF Committee meetings becomes public domain. There are no restrictions on the open
mailing of material presented at committee meetings. In order to reduce disagreements and misunderstandings,
copies must be provided for all agenda items that are discussed. Copies of the material presented, or revisions if
completed in time, are included in the documentation mailings.
The sites for SFF Committee meetings rotate based on which member companies volunteer to host the meetings.
Meetings have typically been held during the INCITS T10 weeks. The meetings of the SFF Special Subject Work-
ing Group on Transceivers have typically been held during INCITS T11 weeks.
The funds received from the annual membership fees are placed in escrow, and are used to reimburse ENDL for
the services to manage the SFF Committee.
Each electronic document mailing obsoletes the previous mailing of that year e.g. July replaces May. To build
a complete set of archives of all SFF documentation, retain the last SFF CD_Access mailing of each year.
Company: ______________________________________________________________
Address: _______________________________________________________________
_______________________________________________________________________
Email: _________________________________________________________________
MC/Visa/AmX______________________________________ Expires_________
When 2 1/2" diameter disk drives were introduced, there was no commonality on external dimensions e.g. physical
size, mounting locations, connector type, connector location, between vendors.
The first use of these disk drives was in specific applications such as laptop portable computers in which space
was at a premium and time to market with the latest machine was an important factor. System integrators worked
individually with vendors to develop the packaging. The result was wide diversity, and with space being such a
major consideration in packaging, it was not possible to replace one vendor's drive with a competitive product.
The desire to reduce disk drive sizes to even smaller dimensions such as 1.8" and 1.3" made it likely that devices
would become even more constrained in dimensions because of a possibility that such small devices could be
inserted into a socket, not unlike the method of retaining semiconductor devices.
The problems faced by integrators, device suppliers, and component suppliers led to the formation of an industry
ad hoc group to address the marketing and engineering considerations of the emerging new technology in disk
drives. After two informal gatherings on the subject in the summer of 1990, the SFF Committee held its first meet-
ing in August.
During the development of the form factor definitions, other activities were suggested because participants in the
SFF Committee faced problems other than the physical form factors of disk drives. In November 1992, the mem-
bers approved an expansion in charter to address any issues of general interest and concern to the storage indus-
try. The SFF Committee became a forum for resolving industry issues that are either not addressed by the
standards process or need an immediate solution.
At the same time, the principle was adopted of restricting the scope of an SFF project to a narrow area, so that the
majority of documents would be small and the projects could be completed in a rapid timeframe. If proposals are
made by a number of contributors, the participating members select the best concepts and uses them to develop
specifications which address specific issues in emerging storage markets.
Those companies which have agreed to support a documented specification are identified in the first pages of each
SFF Specification. Industry consensus is not an essential requirement to publish an SFF Specification because it is
recognized that in an emerging product area, there is room for more than one approach. By making the documen-
tation on competing proposals available, an integrator can examine the alternatives available and select the prod-
uct that is felt to be most suitable.
Suggestions for improvement of this document will be welcome. They should be sent to the SFF Committee, 14426
Black Walnut Ct, Saratoga, CA 95070. The suggestions should additionally be sent to the XFP Chair and Technical
Editor.
The development work on this specification was done by the XFP Group, an industry MSA.
SFF Specifications
There are several projects active within the SFF Committee. At the date of printing, document numbers had been
assigned to the following projects. The status of Specifications is dependent on committee activities.
F = Forwarded The document has been approved by the members for forwarding to a formal standards
body.
P = Published The document has been balloted by members and is available as a published SFF Speci-
fication.
A = Approved The document has been approved by ballot of the members and is in preparation as an
SFF Specification.
C = Canceled The project was canceled, and no Specification was Published.
E = Expired The document has been published as an SFF Specification, and the members voted
against re-publishing it when it came up for annual review.
e = electronic Used as a suffix to indicate an SFF Specification which has Expired but is still available in
electronic form from SFF e.g. a specification has been incorporated into a draft or pub-
lished standard which is only available in hard copy.
i = Information The document has no SFF project activity in progress, but it defines features in developing
industry standards. The document was provided by a company, editor of an accredited
standard in development, or an individual. It is provided for broad review (comments to the
author are encouraged).
s = submitted The document is a proposal to the members for consideration to become an SFF Specification.
Document Sources
Copies of ANSI standards or proposed ANSI standards may be purchased from Global Engineering.
Copies of SFF Specifications are available by joining the SFF Committee as an Observer or Member or by down-
load at ftp://ftp.seagate.com/sff
Review Revision
Copyright © 2002-2005 by XFP Promoters
See copyright notice for restrictions.
This specification has been formally adopted by the XFP Promoters. Changes to the document may be
made to correct any errata in the specification, including clarifications and other changes. Verify that you are
using the latest revision of the specification by visiting the XFP web-site at www.xfpmsa.org.
The promoters of the XFP specification, Broadcom Corporation, Brocade Communications Systems, Inc.,
Ciena Corporation, Emulex Corporation, Finisar Corporation, JDS Uniphase Corporation, Innovation Core
SEI, Inc., Maxim Integrated Products, Tyco Electronics Corporation, and Velio Communication (“XFP Pro-
moters”), and many contributors, collaborated to develop a specification for a 10 Gigabit small foot print
pluggable module. The promoters stated a wish to encourage broad and rapid industry adoption of the spec-
ification. This specification is the result of such collaboration. The XFP specification may be offered to formal
standards bodies to further support the adoption of the specification.
This specification may contain, and sometimes even require the use of, intellectual property owned by
others. No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted
herein, except as indicated in the following copyright notice. Rights to certain of such intellectual property
may be provided by the XFP Promoters Agreement, XFP Adopters Agreement and the XFP Contributors
Addendum.
Copyright by XFP Promoters in accordance with XFP Promoters. XFP Promoters are Broadcom Corpora-
tion, Brocade Communications Systems, Inc., Ciena Corporation, Emulex Corporation, Finisar Corporation,
JDS Uniphase Corporation, Innovation Core SEI, Inc., Maxim Integrated Products, Tyco Electronics Corpo-
ration, and Velio Communication. All rights are reserved except that a license is hereby granted to copy and
reproduce this specification, in unmodified form only, for fee-free distribution or for use internal to the orga-
nization copying the specification.
If errors are identified in this specification, please notify the chair and editor identified on the first page.
Please provide a clear identification of the error, the text of any proposed correction, and a justification or
explanation of the correction.
4.0 Adopted Specification. Updated CRC R/W, Updated Designations, and Updated Reg- 4/13/04
isters Nomenclature
4.1 I2C Errata, S-parameters adjustment, Annex E3, Correcting mixed mode equation of 4/20/05
Annex C4, Annex for BER 1E-15, address role over
4.5 Non-EQJ Jitter definition, Annex E3, and Bail Color. 8/31/05
XFP Foreword
The XFP Promoters, Broadcom Corporation, Brocade Communications Systems, Inc., Ciena Cor-
poration, Emulex Corporation, Finisar Corporation, Innovation Core SEI, Inc. (a Sumitomo Electric
Company), JDS Uniphase Corporation, Maxim Integrated Products, Tyco Electronics Corporation,
Velio Communication, and many contributors gathered together to develop a specification for a 10
Gigabit small foot print pluggable module. The Promoters stated a wish to encourage broad and
rapid industry adoption of the specification. The XFP specification may be offered to formal stan-
dards bodies to further support the adoption of the specification. The XFP Promoters and Contrib-
utors solicited technical review and contributions to the XFP specification among many component,
module, and system designers and manufacturers to assure that it meets a broad variety of re-
quirements.
Signed agreements exist among the promoters and between the promoters and contributors that
allowed them to carry forward these activities in a manner that encouraged rapid development,
open sharing of technology, and the timely resolution of disagreements. The agreements further
explained and protected the rights of both promoters and contributors. Similar agreements protect
the rights of adopters of the specification.
Organizations may become adopters of the XFP specification by contacting any XFP Promoter and
requesting the necessary forms and following the required procedures.
XFP Contributors
Fujitsu Primarion
Infineon Technologies Sigma-Links, Inc. (an OKI and Fujikura joint venture)
XFP Acknowledgments
The XFP members would like to acknowledge and thank Dr. Ali Ghiasi for his tremendous effort
and diligence in editing this specification.
The members would like to thank Dr. Lew Aronson and Mr. Ed Bright for writing the Management
and Mechanical chapters. In addition, they would like to thank Mr. Tom Lindsay for the jitter Ap-
pendix, Dr. Piero Bradley for jitter section, Mr. Lawrence Williams for the XFI Channel Simulation
and Modeling and the Optimum Via Appendix, Mr. Henrik Johansen for the compliance test boards
Appendix, Mr. Steve Silverman for the power supply section, Mr. Gary Heitkamp for the Thermal
Appendix, Mr. Randy Clark for the 2-Wire Protocol, Mr. Ron Miller for the TDR Appendix, and Dr.
Stefanos Sidiropoulos for the VPS section.
The members would also like to thank Mr. Robert Snively for skillfully steering the MSA, as without
him we could not have achieved our goal.
Special thanks are due to all those who attended the meetings and made contributions, as this doc-
ument is a collection of their work.
1
2
3
4
5
GND 15 6
16 GND 7
RX_LOS 14 8
17 RD- 9
MOD_NR 13
10
18 RD+ 11
MOD_ABS 12
12
19 GND 13
SDA 11
14
20 VCC2
10 15
SCL
21 P_DOWN/RST 16
VCC3 9 17
22 VCC2 18
VCC3 8 19
TOWARD TOWARD
23 GND 20
ASIC BEZEL
GND 7 21
24 REFCLK+ 22
VCC5 6 23
25 REFCLK- 24
TX_DIS 5 25
26 GND 26
INTERRUPT 4 27
27 GND 28
MOD_DESEL 3 29
28 TD- 30
VEE5 2 31
29 TD+ 32
GND 1
33
30 GND 34
35
36
37
Figure 1 Host PCB XFP Pinout Top View
38
39
40
41
42
1
2
Table 1 Module Electrical Pin Definition 3
4
Pin Logic Symbol Name/Description Note
5
1 GND Module Ground 1 6
2 VEE5 Optional -5.2V Power Supply 7
3 LVTTL-I Mod_DeSel Module De-select; When held low allows module to respond to 2-wire serial interface 8
4 LVTTL-O Interrupt Interrupt; Indicates presence of an important condition which can be read over the 2-wire 2 9
serial interface
10
5 LVTTL-I TX_DIS Transmitter Disable; Turns off transmitter laser output
11
6 VCC5 +5V Power Supply 12
7 GND Module Ground 1 13
8 VCC3 +3.3V Power Supply 14
9 VCC3 +3.3V Power Supply 15
10 LVTTL-I/O SCL 2-Wire Serial Interface Clock 2 16
11 LVTTL-I/O SDA 2-Wire Serial Interface Data Line 2 17
12 LVTTL-O Mod_Abs Indicates Module is not present. Grounded in the Module 2 18
13 LVTTL-O Mod_NR Module Not Ready; Indicating Module Operational Fault 2 19
14 LVTTL-O RX_LOS Receiver Loss Of Signal Indicator 2 20
15 GND Module Ground 1 21
16 GND Module Ground 1 22
17 CML-O RD- Receiver Inverted Data Output
23
24
18 CML-O RD+ Receiver Non-Inverted Data Output
25
19 GND Module Ground 1
26
20 VCC2 +1.8V Power Supply 3
27
21 LVTTL-I P_Down/RST Power down; When high, requires the module to limit power consumption to 1.5W or
below. 2-Wire serial interface must be functional in the low power mode.
28
Reset; The falling edge initiates a complete reset of the module including the 2-wire 29
serial interface, equivalent to a power cycle. 30
22 VCC2 +1.8V Power Supply 3 31
23 GND Module Ground 1 32
24 PECL-I RefCLK+ Reference Clock Non-Inverted Input, AC coupled on the host board 33
25 PECL-I RefCLK- Reference Clock Inverted Input, AC coupled on the host board 34
26 GND Module Ground 1 35
27 GND Module Ground 1 36
28 CML-I TD- Transmitter Inverted Data Input 37
29 CML-I TD+ Transmitter Non-Inverted Data Input 38
30 GND Module Ground 1 39
1. Module ground pins Gnd are isolated from the module case and chassis ground within the module. 40
2. Shall be pulled up with 4.7K-10Kohms to a voltage between 3.15V and 3.45V on the host board. 41
3. The 1.8 V power supply can be optionally programmed to voltages lower than 1.8 V in modules supporting the variable power supply. 42
2.4.3 INTERRUPT 1
Interrupt is an output pin. When “Low”, indicates possible module opera- 2
tional fault or a status critical to the host system. The Interrupt pin is an 3
open collector output and must be pulled up to Host_Vcc the host board. 4
5
2.4.4 TX_DIS 6
TX_DIS is an input pin. When TX_DIS is asserted High, the XFP module 7
transmitter output must be turned off. The TX_DIS pin must be pulled up 8
to VCC3 in the XFP module.
9
10
2.4.5 MOD_ABS
11
Mod_ABS is pulled up to Host_Vcc on the host board and grounded in the
12
XFP module. Mod_ABS is then asserted “High” when the XFP module is
physically absent from a host slot. 13
14
2.4.6 RX_LOS 15
The RX_LOS when High indicates insufficient optical power for reliable 16
signal reception. The RX_LOS pin is an open collector output and must 17
be pulled up to Host_Vcc on the host board. 18
19
2.4.7 P_DOWN/RST 20
This is a multifunction pin for module Power Down and Reset. The 21
P_Down/RST pin must be pulled up to VCC3 in the XFP module. 22
23
24
2.4.7.1 POWER DOWN FUNCTION 25
The P_Down pin, when held High by the host, places the module in the 26
standby (Low Power) mode with a maximum power dissipation of 1.5W. 27
This protects hosts which are not capable of cooling higher power mod- 28
ules which may be accidentally inserted. 29
30
The module’s 2-wire serial interface and all laser safety functions must be 31
fully functional in this low power mode. During P_Down, the module shall
32
still support the completion of reset Interrupt, as well as maintain function-
ality of the variable power supply as described in section 5.7. 33
34
35
36
2.4.7.2 RESET FUNCTION 37
The negative edge of P_Down/RST signal initiates a complete module 38
reset. 39
40
41
42
The XFP port on a host board is tested with a resistive load in place of the 1
XFP module, each voltage rail at maximum current supported by the host. 2
Voltage is measured at the module side of the XFP connector. The test is 3
performed with all other portions of the host board/system active. Hosts
4
with multiple XFP modules shall test ports one at a time, with active XFPs
in all the remaining ports. 5
6
The XFP module is tested with a high quality power supply connected 7
through the sample filter Figure 3. Voltage is measured at the host side of 8
the XFP connector, between the sample host filter network and the XFP. 9
The module must pass this test in all operating modes. This test ensures 10
the module will not couple excessive noise from inside the module back
11
onto the host board.
12
13
14
2.7.2 POWER NOISE SUSCEPTIBILITY 15
A module will meet all electrical requirements and remain fully operational 16
in the presence of noise on all voltage inputs. The recommended suscep- 17
tibility test is to sweep a sinuosidal waveform on each voltage input, with 18
peak amplitude as described in Figure 2. This test applies at minimum 19
and maximum DC setpoint levels. It is also desirable for a module and
20
host to each tolerate a degree of random or semi-random noise on all
voltage pins simultaneously, but the characteristics of this noise are be- 21
yond the scope of this document. 22
23
24
5 25
Noise as a% of VCC/VEE (peak-peak)
26
4 27
28
29
3 30
31
32
2
33
34
1
35
36
37
1.0 10.0 100.0 1000.0 10000.0 38
39
Frequency (KHz)
40
Figure 2 Power Noise Requirement 41
42
For power supply noise susceptibility testing and methodology see E.7. 1
2
3
2.7.3 HOST FILTERING
4
The example host board power supply filtering shown in Figure 3 will 5
meet the noise filtering requirements in most systems. Other filtering
6
implementations or local regulation may be used to meet the power noise
output requirements described in section 2.7.1. 7
8
Any voltage drop across a filter network on the host is counted against 9
the Host DC setpoint accuracy specification in Table 4. For this reason, 10
the example filter illustrated in Figure 3 may not be appropriate for a host
powering multiple XFPs and/or other host components from a shared 11
voltage supply. 12
13
14
15
16
17
Host Board
18
Host +5V 4.7uH VCC5 19
20
0.1 uf 22uf 0.1uf
21
Host +3.3V 22
XFP Connector
4.7uH VCC3
23
0.1 uf 22uf 0.1uf 24
XFP Module 25
Host +1.8V 4.7uH VCC2 26
22uf 0.1uf 27
0.1 uf
28
Optional Host -5.2V 4.7uH VEE5 29
30
0.1 uf 22uf 0.1uf 31
32
33
Figure 3 Example of Host Board Supply Filtering Network
34
35
36
37
38
39
40
41
42
1
2
Table 4 XFP Power Supply 3
Parameters Symbol Conditions Min Max Units 4
5
VCC5 (5V)/VEE5 (-5.2V) Pin
6
Power Supply Noise including Ripple [peak-to-peak] VCC5/VEE5 see 2.7.1 % 7
Host DC set point accuracy -5 +5 % 8
Module Maximum Current Inrush Note 1 0.5 A
9
10
Module current ramp rate Note 1 100 mA/µs
11
Maximum Power for the +5 and -5.2V Rail 2.5 W 12
13
Maximum Power for the +5 and -5.2V Rail During Power Down Note 2 0.25 W
14
Host Bulk bypass capacitance for the module Note 3 22 uF
15
VCC3 (3.3V) Rail 16
Power Supply Noise including Ripple [peak-to-peak] VCC3 see 2.7.1 % 17
18
Host DC set point accuracy -5 +5 %
19
Module Maximum Current Inrush Note 1 0.75 A 20
Module current ramp rate Note 1 100 mA/µs 21
Maximum Power for the +3.3V Rail 2.5 W
22
23
Maximum Power for the +3.3V Rail During Power Down Note 2 1 W
24
Host Bulk bypass capacitance for the module Note 3 22 uF 25
VCC2 (1.8V) Rail
26
27
Power Supply Noise including Ripple [peak-to-peak] VCC2 see 4, 2.7.1 %
28
Host DC set point accuracy -5 +5 % 29
Module Maximum Current Inrush Note 1 1 A 30
Module current ramp rate Note 1 100 mA/µs
31
32
Maximum Power for the 1.8V Rail 1.8 W
33
Maximum Power for the 1.8V Rail During Power Down Note 2 1 W 34
Host Bulk bypass capacitance for the module Note 3 22 uF 35
36
1. Modules which present a small capacitive load to the host during hotplug ( C ≤ 300nF ) are exempt from the inrush current requirements
since they limit the total in rush charge.
37
2. Maximum module power dissipation shall not exceed 1.5 W. 38
3. Host provides bulk capacitance to suppress low frequency noise inside the XFP module. Host and module must each implement appropri- 39
ate high frequency bypass to independently meet the noise requirement defined in section 2.7.1. 40
4. VCC2 can be optionally programmed to voltages less than 1.8V in modules supporting the Variable Power Supply 5.7. 41
42
2.8 ESD 1
The XFP module and host XFI pins (High Speed Pins) shall withstand 500 2
V electrostatic discharge based on Human Body Model per JEDEC 3
JESD22-A114-B. 4
5
The XFP module and all host pins with exception of the XFI pins (High 6
Speed Pins) shall withstand 2KV electrostatic discharge based on Human
7
Body Model per JEDEC JESD22-A114-B.
8
The XFP module shall meet ESD requirement given in EN61000-4-2, cri- 9
terion B test specification such that units are subjected to 15KV air dis- 10
charges during operation and 8KV direct contact discharges to the case. 11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
Connector 4
5
D C’ C XFP Module 6
*
RX 7
Driver 8
ASIC/SerDes Caps
9
* 10
TX
A B’ B 11
12
* Signal Conditioner
13
Host Board 14
15
Figure 4 Application Reference Model 16
17
XFI Compliance points are defined as the following: 18
19
• A: SerDes transmitter output at ASIC/SerDes package pin on a DUT 20
board 3.6 and A.1 21
• B: Host system SerDes output across the host board and connector 22
at the Host Compliance Test Card 3.7.1 and A.2 23
• B’: XFP transmitter input across the Module Compliance Test Board 24
3.8.1 and A.3. 25
• C: Host system input at the Host Compliance Test Card input 3.7.2 26
and A.2 27
• C’: XFP module output across the Module Compliance Test Board 28
3.8.2 and A.3. 29
30
• D: ASIC/SerDes input package pin on the DUT board 3.6.2 and A.1.
31
ASIC/SerDes compliance points are informative.
32
33
34
3.3 XFI TERMINATION AND DC BLOCKING 35
The XFI link requires nominal 100 Ω differential source and load termina- 36
tions on both the host board and the module. The XFI terminations shall 37
provide both differential and common mode termination to effectively ab- 38
sorb differential and common mode noise and reflections as described: 39
40
41
42
Rt
Rp
D C’ C Rn 24
Rn 25
Capacitor
26
XFI Receiver
XFI Transmitter 27
Capacitor
Rp
28
Rt 29
Rt
Rp
A B’ B 30
Rn
31
Rn
Capacitor 32
Capacitor
33
Reference Clock 34
35
XFI CONNECTOR 36
37
Figure 5 XFI Termination and AC Coupling
38
39
40
41
42
XFI return loss, therefore differential coupling effects may be neglected for 1
insertion loss S21. Please see Appendix C: for differential measurements 2
and conversions. 3
4
5
The above equation is valid from 250MHz to 7 GHz. An approximate dia- 6
gram is shown in Figure 6. 7
8
0 9
10
11
-2
Example of a Compliant Channel 12
13
SDD21 (dB)
-4 14
Min. Allowed
15
Channel Transfer
-6
16
17
18
-8 19
20
-10 21
0 1 2 3 4 5 6 7 8 9 22
Frequency (GHz)
23
Figure 6 Channel Transfer Model 24
25
26
The channel transfer model is given by two port S-parameters SDD21 in 27
Figure 6. Detailed descriptions of two port and 4 port measurements and 28
conversions are provided in Appendix C:.
29
30
Architectural Note 31
For loosely coupled symmetrical structures, differential SDD21 may be 32
approximated by direct measurement of S21. 33
34
35
36
The XFI compliance channel must meet a differential return loss of 10 dB 37
from 1MHz-7500MHz. For frequencies in the range of 7.5 to 15 GHz 38
channel return loss is given by: 39
40
SDD11 ( dB ) = 10 – 25Log -------
f
10 7.5 41
42
1
2
Table 8 XFI ASIC/SerDes Transmitter Output Electrical Specifications at A 3
Parameter - A Symbol Conditions Min Typ Max Units 4
5
Reference Differential Impedance Zd See Figure 5 100 Ω
6
Termination Mismatch ∆RM See E.6, Figure 5 5 % 7
DC Common Mode Voltage Vcm 0 3.6 V 8
9
Output Rise and Fall time (20% to 80%) tRH, tFH 24 ps 10
Output AC Common Mode Voltage See E.5 15 mV (RMS) 11
12
0.05-0.1 GHz 20 dB
Differential Output Return Loss 1 SDD22
13
0.1-7.5 GHz 10 dB 14
15
7.5-15 GHz see 2
16
Common Mode Output Return Loss 2 SCC22 0.1-15 GHz 6 dB 17
1. Return Loss given by equation SDD22(dB)= 10 - 16.6 Log10(f/7.5), with f in GHz.
18
2. Common mode reference impedance is 25 Ω. Common Mode Return Loss helps absorb reflections and noise for EMI. 19
20
Transmitter jitter specifications are listed in Table 9. Figure 7 gives the 21
compliance eye mask. 22
23
24
Table 9 XFI ASIC/SerDes Transmitter Output Jitter Specifications at A
25
Transmitter - A Symbol Conditions Min Typ Max Units 26
Determinstic Jitter DJ See E.1, 1 0.15 UI (p-p) 27
28
Total Jitter TJ See E.1, 1 0.30 UI (p-p)
29
Eye Mask X1 0.15 UI 30
31
Eye Mask X2 0.4 UI
32
Eye Mask Y1 180 mV 33
34
Eye Mask Y2 385 mV
35
Jitter Generation for Telecom Applications 50KHz to 8 MHz 6.5 mUI (RMS) 36
see 2 37
1. In loop timing mode, includes jitter that transfers through the ASIC from the receiver during any valid operational input 38
conditions.
39
2. Measured with a filter with 50 KHz high-pass cutoff desinged frequency and 8 MHz low pass cut off frequency. The filter rolls off
40
at least for one decade on each side of the passband with -20 dB/Dec. Does not apply to a host designed with Optional Synchronous
CMU clock, when used in conjunction with a Synchronous CMU module 3.10.1. 41
42
1
2
3
4
Y2 5
6
Absolute Amplitude
Y1
7
8
0
9
10
-Y1 11
-Y2 12
13
0.0 X1 X2 1-X2 1-X1 1.0 14
Normalized Time (UI) 15
16
Figure 7 XFI ASIC/SerDes Transmitter Differential Output 17
Compliance Mask at A 18
19
20
3.6.2 XFI ASIC/SERDES RECEIVER INPUT SPECIFICATIONS AT D (INFORMATIVE)
21
XFI receiver electrical specifications are given in Table 10 and measured 22
at compliance point D. The return loss at D is measured with the SerDes
23
on a DUT board A.1. Receiver input impedance is 100 Ohms differential.
The load must provide differential termination and minimize differential to 24
common mode conversion for high quality signal termination and low EMI, 25
as given by Table 10. 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
Table 10 XFI ASIC/SerDes Receiver Electrical Input Specifications at D 3
Parameter - D Symbol Conditions Min Typ Max Units 4
5
Reference Differential Impedance Zd See Figure 5 100 Ω
6
Termination Mismatch ∆ZM See E.6, Figure 5 5 % 7
AC Common Mode Voltage See E.5 25 mV (RMS) 8
9
0.05-0.1 GHz 20 dB
Differential Input Return Loss SDD11
10
0.1-7.5 GHz 10 dB 11
7.5-15 GHz see 1
12
13
Common Mode Input Return Loss2 SCC11 0.1-15 GHz 6 dB 14
Differential to Common Mode Input SCD11 0.1-15 GHz 12 dB 15
Conversion2 16
1. Return Loss given by equation SDD11(dB)= 10 - 16.6 Log10(f/7.5), with f in GHz C.4. 17
2. Common mode reference impedance is 25 Ω. SCD11 relates to conversion of differential to common mode and the associated 18
generation of EMI C.4
19
The XFI jitter specifications at reference D are listed in Table 11 and the 20
compliance mask is shown in Figure 8. Only a compliant transmitter 21
passing through the XFI channel is guaranteed for interoperablity. 22
23
Table 11 XFI ASIC/SerDes Receiver Input Jitter Specifications at D 24
Receiver- D Symbol Conditions Min Typ Max Units
25
26
Total Jitter TJ See E.1, E.2 0.65 UI (p-p)
27
Total non-EQJ Jitter See E.1, E.2 0.45 UI (p-p) 28
29
Sinusoidal Jitter Tolerance SJ E.3 see 1
30
Eye Mask X1 Note 2 0.325 UI 31
Eye Mask Y1 55 mV
32
33
Eye Mask Y2 Note 3 525 mV 34
1. Sinuosidal jitter tolerance for Telecom and Datacom respectively given by Figure 9 and Figure 10. 35
2. Mask coordinate X1=0.225 if total non-EQJ jitter is measured. 36
3. Out of 525 mV, 100 mV is allocated for multiple reflection. 37
38
39
40
41
42
1
2
3
4
Y2 5
Absolute Amplitude
6
Y1 7
8
0
9
-Y1
10
11
-Y2 12
13
0.0 X1 1-X1 1.0 14
Normalized Time (UI) 15
16
Figure 8 XFI ASIC/SerDes Receiver Differential Input Compliance
17
Mask
18
19
An XFI receiver for Telecom (SONET OC-192 and G. 709 “OTU-2”) appli- 20
cations must meet jitter tolerance given by Figure 9 with the addition of 21
input jitter given by Table 11. An XFI receiver for Datacom (Ethernet 22
802.3ae or Fibre Channel 10GFC) must meet the jitter tolerance given by
23
Figure 10 with the addition of random and deterministic jitter given by
Table 11. 24
25
26
27
15.2
Sinuosidal Jitter Tolerance (UIp-p)
28
-20dB/Dec 29
30
1.7 31
32
33
34
0.17 35
36
0.05 37
38
0.01E-3 2E-3 17.9E-3 0.4 4 8 27.2 80
Frequency (MHz) 39
40
41
Figure 9 XFI ASIC/SerDes Receiver Input Telecom Sinuosidal Jitter
42
Tolerance 1
2
3
4
5
Sinuosidal Jitter Tolerance (UIp-p)
1
2
Table 12 XFI Host Transmitter Output Electrical Specifications at B 3
4
Parameter - B Symbol Conditions Min Typ Max Units
5
Reference Differential Impedance Zd 100 Ω
6
Termination Mismatch ∆ZM See E.6, Figure 5 5 % 7
DC Common Mode Voltage Vcm 0 3.6 V 8
9
Output AC Common Mode Voltage See E.5 25 mV (RMS)
10
0.05-0.1 GHz 20 dB 11
Differential Output Return Loss SDD22 12
0.1-5.5 GHz 8 dB
13
5.5-12 GHz see 1
14
Common Mode Output Return Loss2 SCC22 0.1-12 GHz 3 dB 15
1. Return Loss given be equation SDD22(dB)= 8 - 20.66 Log10(f/5.5), with f in GHz. 16
2. Common mode reference impedance is 25 Ω. Common Mode Input return loss helps absorb reflections and noise reducing EMI. 17
18
The XFI jitter specifications at reference point B are listed in Table 13 and 19
the compliance mask is shown in Figure 11. Only a compliant transmitter
20
passing through the XFI channel is guaranteed for interoperablity.
21
Table 13 XFI Host Transmitter Output Jitter Specifications at B 22
23
Receiver- B Symbol Conditions Min Typ Max Units 24
Total Jitter 1,
TJ See E.1 0.61 UI 25
Total non-EQJ Jitter 1 See E.1, E.2 0.41 UI
26
27
Eye Mask X1 see 2 0.305 UI
28
Eye Mask Y1 60 mV 29
30
Eye Mask Y2 see 3 410 mV
31
Jitter Generation at B Telecom Applications 50KHz-8MHz, see 4 7 mUI (RMS) 32
1. In loop timing mode, includes jitter that transfers through ASIC from the receiver during any valid operational input conditions. 33
2. . Mask cordinate X1=0.205 if total non-EQJ jitter is measured. 34
3. Out of 410 mV, 25 mV is allocated for multiple reflection. 35
4. Measured with a filter with 50 KHz high-pass cutoff desinged frequency and 8 MHz low pass cut off frequency. The filter rolls
36
off at least for one decade on each side of the passband with -20 dB/Dec. Does not apply to a host designed with Optional
37
Synchronous CMU clock, when used in conjunction with a Synchronous CMU Signal Conditioner module see 3.10.1.
38
39
40
41
42
1
2
3
4
Y2 5
Absolute Amplitude
6
Y1 7
8
0
9
-Y1
10
11
-Y2 12
13
0.0 X1 1-X1 1.0 14
Normalized Time (UI) 15
16
Figure 11 XFI Host Transmitter Differential Output Compliance Mask
17
at B
18
19
3.7.2 XFI HOST RECEIVER INPUT SPECIFICATIONS AT C 20
The XFI Host receiver electrical specifications at compliance point C are 21
given in Table 14. The load must provide differential termination and min- 22
imize differential to common mode conversion for quality signal termina- 23
tion and low EMI, as given in Table 14. 24
25
Return loss at compliance point C is measured with the Host Test Board
26
plugged into the host as shown in A.2.
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
Table 14 XFI Host Receiver Input Electrical Specifications at C 3
Parameter - C Symbol Conditions Min Typ Max Units 4
5
Reference Differential Impedance Zd See Figure 5 100 Ω
6
Termination Mismatch ∆ZM See E.6, Figure 5 5 % 7
Source to Sink DC Potential Difference Vcm 0 3.6 V 8
9
Input AC Common Mode Voltage See E.5 15 mV (RMS) 10
Input Rise and Fall time (20% to 80%) tRH, tFH 24 ps 11
12
0.05-0.1 GHz 20 dB
Differential Input Return Loss SDD11
13
0.1-5.5 GHz 8 dB 14
15
5.5-12 GHz see 1
16
Common Mode Return Loss 2 SCC11 0.1-12 GHz 3 dB 17
Differential to Common Mode Conversion 2 SCD11 0.1-12 GHz 10 dB
18
19
1. Return Loss given be equation SDD11(dB)= 8 - 20.66 Log10(f/5.5), with f in GHz C.4. 20
2. Common mode reference impedance is 25 Ω. SCD11 and SCC11 help minimize generated EMI C.4.
21
22
Transmitter jitter specifications are listed in Table 15. Figure 12 gives the
host compliance eye mask. 23
24
25
26
Table 15 XFI Host Receiver Input Jitter Specifications at C 27
Transmitter - C Symbol Conditions Min Typ Max Units 28
29
Determinstic Jitter DJ See E.1 0.18 UI (p-p)
30
Total Jitter TJ See E.1 0.34 UI (p-p) 31
32
Sinusoidal Jitter Tolerance SJ E.3 see 1
33
Eye Mask X1 0.17 UI 34
35
Eye Mask X2 0.42 UI
36
Eye Mask Y1 170 mV 37
Eye Mask Y2 see 2 425 mV
38
39
1. Sinuosidal jitter tolerance for Telecom and Datacom respectively given by Figure 13 and Figure 14. 40
2. Out of 425 mV, 25 mV is allocated for multiple reflections.
41
42
1
2
3
4
Y2 5
Absolute Amplitude
6
Y1
7
8
0
9
10
-Y1 11
-Y2 12
13
0.0 X1 X2 1-X2 1-X1 1.0 14
Normalized Time (UI) 15
16
Figure 12 XFI Host Receiver Differential Input Compliance Mask at C 17
18
19
The XFI receiver for Telecom (SONET OC-192 and G. 709 “OTU-2”) ap-
20
plications must meet the jitter tolerance with the addition of deterministic
and random jitter given by Table 15. The XFI receiver for Datacom 21
(Ethernet 802.3ae or Fibre Channel 10GFC) Host must meet jitter toler- 22
ance given by Figure 14 with the addition of deterministic and random 23
jitter given by Table 15. 24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
15.2 4
24
1.13x ------- + 0.1 , f in MHz
0.2 25
f
26
27
-20dB/Dec
28
29
30
0.17
31
32
0.05 33
34
0.04 4 8 27.2 80
Frequency (MHz)
35
36
37
Figure 14 XFI Host Receiver Input Datacom Sinuosidal Jitter
38
Tolerance
39
40
41
42
1
2
Table 16 XFI Module Transmitter Input Electrical Specifications at B’ 3
4
Parameter - B’ Symbol Conditions Min Typ Max Units
5
Reference Differential Input Impedance Zd See Figure 5 100 Ω
6
Termination Mismatch ∆ZM See E.6, Figure 5 5 % 7
8
Source to Sink DC Potential Difference Vcm 0 3.6 V
9
Input AC Common Mode Voltage See E.5 25 mV (RMS) 10
0.05-0.1 GHz 20 dB 11
Differential Input Return Loss SDD11 12
0.1-5.5 GHz 8 dB
13
5.5-12 GHz see 1 14
15
Common Mode Input Return Loss2 SCC11 0.1-12 GHz 3 dB
16
Differential to Common Mode Conversion2 SCD11 0.1-12 GHz 10 dB 17
18
1. Return Loss given be equation SDD11(dB)= 8 - 20.66 Log10(f/5.5), with f in GHz C.4.
2. Common mode reference impedance is 25 Ω. Differential to common mode conversion relates to generation of EMI C.4. 19
20
21
22
Table 17 XFI Module Transmitter Input Jitter Specifications at B’ 23
Receiver- B’ Symbol Conditions Min Typ Max Units 24
25
Total Non-EQJ Jitter 0.41 See E.1, E.2 0.41 UI (p-p)
26
Total Jitter TJ See E.1, E.2 0.61 UI (p-p) 27
28
Sinusoidal Jitter Tolerance SJ E.3 see 1
29
Eye Mask X1 see 2 0.305 UI 30
Eye Mask Y1 60 mV
31
32
Eye Mask Y2 see 3 410 mV 33
1. Sinuosidal jitter tolerance for Telecom and Datacom respectively given by Figure 16 and Figure 17. 34
2. Mask cordinate X1=0.205 if total non-EQJ jitter is measured. 35
3. Out of 410 mV, 50 mV is allocated for multiple reflection. 36
37
38
39
40
41
42
1
2
3
4
Y2 5
Absolute Amplitude
6
Y1 7
8
0
9
-Y1
10
11
-Y2 12
13
0.0 X1 1-X1 1.0 14
Normalized Time (UI) 15
16
Figure 15 XFI Module Transmitter Differential Input Compliance
17
Mask at B’
18
19
XFI Module for Telecom (SONET OC-192 and G. 709 “OTU-2”) applica- 20
tions must meet the jitter tolerance given by Figure 16 with addition of 21
input jitter given by Table 17. The XFI Module for Datacom (Ethernet 22
802.3ae and Fibre Channel 10GFC) must meet the jitter tolerance given
23
by Figure 17 with the addition of random and deterministic jitter given in
Table 17. 24
25
26
15.2 27
Sinuosidal Jitter Tolerance (UIp-p)
28
29
-20dB/Dec 30
1.7 31
32
33
34
0.17 35
36
0.05 37
0.01E-3 2E-3 17.9E-3 0.12 4.08 80 38
Frequency (MHz) 39
40
Figure 16 XFI Module Transmitter Input Telecom Sinuosidal Jitter 41
Tolerance 42
1
2
3
4
1
2
Table 18 XFI Module Receiver Output Electrical Specifications at C’ 3
Parameter - C Symbol Conditions Min Typ Max Units 4
5
Reference Differential Output Impedance Zd See Figure 5 100 Ω
6
Termination Mismatch ∆ZM See E.6, Figure 5 5 % 7
DC Common Mode Potential Vcm 0 3.6 V 8
9
Output AC Common Mode Voltage See E.5 15 mV (RMS) 10
Output Rise and Fall time (20% to 80%) tRH, tFH 24 ps 11
12
0.05-0.1 GHz 20 dB
Differential Output Return Loss SDD22
13
0.1-5.5 GHz 8 dB 14
15
5.5-12 GHz see 1 dB
16
Common Mode Output Return Loss 2 SCC22 0.1-12 GHz 3 dB 17
1. Differential Return Loss given be equation SDD22(dB)= 8 - 20.66 Log10(f/5.5), with f in GHz C.4.
18
2. Common mode reference impedance is 25 Ω. Common Mode Output Return Loss helps absorb reflection and noise improving 19
EMI C.4. 20
21
Transmitter jitter specifications are listed in Table 19. Figure 18 gives the
22
compliance eye mask.
23
24
25
Table 19 XFI Module Receiver Output Jitter Specifications at C’ 26
27
Transmitter - C Symbol Conditions Min Typ Max Units
28
Determinstic Jitter DJ See E.1, 1 0.18 UI (p-p) 29
Total Jitter TJ See E.1, 1 0.34 UI (p-p) 30
31
Eye Mask X1 0.17 UI
32
Eye Mask X2 0.42 UI 33
34
Eye Mask Y1 170 mV
35
Eye Mask Y2 425 mV 36
37
1. Includes jitter transfered from the optical receiver during any valid operational input condition.
38
39
40
41
42
. 1
2
3
Y2 4
5
Absolute Amplitude
Y1
6
7
0
8
9
-Y1 10
-Y2 11
12
0.0 X1 X2 1-X2 1-X1 1.0 13
Normalized Time (UI) 14
15
Figure 18 XFI Module Receiver Differential Output Compliance Mask 16
at C’ 17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
The XFP Telecom modules receiver shall meet the requirements of both 1
SONET GR-253 and Table 21. 2
3
4
5
Table 21 XFP Telecom Module Receiver Requirement
6
Module Transmitter C’ Symbol Conditions Min Typ Max Units 7
Jitter Transfer Bandwidth BW PRBS 2^31-1 Data, see 1 8 MHz 8
9
Jitter Peaking Frequency <120 KHz 0.03 dB
see 2 10
11
Frequency >120 KHz 1 dB
12
1. OC-192/SDH-64 Sinusoidal Jitter Tolerance Mask 13
2. Only required for loop timing, see E.4. 14
15
16
17
3.9.2 DATACOM MODULE 18
XFP Datacom module shall meet IEEE 802.3ae and 10 GFC requirement, 19
in addition to the requirements given in Table 20 and Table 21. 20
21
Table 22 XFP Datacom Module Transmitter Requirement
22
Module Transmitter B’ Symbol Conditions Min Typ Max Units 23
Jitter Transfer Bandwidth BW PRBS 2^31-1 Data or Scram- 8 MHz
24
bled 64B/66B, see 1 25
Jitter Peaking Frequency >50 KHz 1 dB
26
27
1. Based on IEEE 802.3ae Clause 52 Sinuosidal Jitter Tolerance Mask Figure 52-4.
28
29
30
Table 23 XFP Datacom Module Receiver Requirement 31
32
Module Transmitter C’ Symbol Conditions Min Typ Max Units
33
Jitter Transfer Bandwidth BW PRBS 2^31-1 Data 8 MHz 34
or Scrambled 64B/66B, see 1 35
Jitter Peaking Frequency >50 KHz 1 dB 36
1. Based on IEEE 802.3ae Clause 52 Sinuosidal Jitter Tolerance Mask Figure 52-4. 37
38
39
40
41
42
1
2
3
Table 25 Optional Reference Clock Specifications 4
5
Parameter Symbol Conditions Min Typ Max Units
6
Clock Differential Input impedance Zd 80 100 120 Ω 7
8
Differential Input Clock Amplitude (p-p) AC Coupled PECL 640 1600 mV
9
Reference Clock Duty Cycle 40 60 % 10
Reference Clock Rise/Fall Time Tr/Tf 20%-80% 200 1250 ps
11
12
Reference Clock Frequency f0 see 1 Baud1/64 MHz
13
CMU Reference Clock Skew TD Baud2 -10 +10 UI 14
Single Side Band Phase Noise @1KHz -85 dBc/
15
Hz 16
@10KHz -108
17
@100KHz -128 18
@1MHz -138 19
20
@10MHz -138
21
1. Reference clock frequency is exactly 1/64 of the Baudrate.
22
2. A UI defined by the serial Baudrate.
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
of any module until the 2-wire serial bus communication is complete and 1
the hold time requirement (Host_select_hold - Table 26) is satisfied. 2
3
The 2-wire serial interface address of the XFP module is 1010000X (A0h).
4
In order to allow access to multiple XFP modules on the same 2-wire se-
rial bus, the XFP pinout includes a Mod_DeSel or module deselect pin. 5
This pin (which is pulled high or deselected in the module) must be held 6
low by the host to select the module of interest and allow communication 7
over the 2-wire serial interface. The module must not respond to or accept 8
2-wire serial bus instructions unless it is selected. 9
10
11
Table 26 XFP 2-Wire Timing Specifications 12
13
Parameter Symbol Min. Max. Unit Conditions 14
Clock Frequency fSCL 0 400 kHz
15
16
Clock Pulse Width Low tLOW 1.3 µs
17
Clock Pulse Width High tHIGH 0.6 µs 18
Time bus free before new trans- tBUF 20 µs Between STOP and START 19
mission can start 20
START Hold Time tHD,STA 0.6 µs 21
22
START Set-up Time tSU,STA 0.6 µs
23
Data In Hold Time tHD,DAT 0 µs
24
Data In Set-up Time tSU,DAT 0.1 µs 25
Input Rise Time (100kHz) tR,100 1000 ns From (VIL,MAX - 0.15) to (VIH,MIN + 0.15) 26
27
Input Rise Time (400kHz) tR,400 300 ns From (VIL,MAX - 0.15) to (VIH,MIN + 0.15)
28
Input Fall Time (100kHz) tF,100 300 ns From (VIH,MIN + 0.15) to (VIL,MAX - 0.15) 29
Input Fall Time (400kHz) tF,400 300 ns From (VIH,MIN + 0.15) to (VIL,MAX - 0.15) 30
STOP Set-up Time tSU,STO 0.6 µs 31
32
Host Supplied Module DeSe- Host_select_setup 2 ms Setup time on the select lines before start of a
lect Setup Time host initiated serial bus sequence 33
34
Host Supplied Module DeSe- Host_select_hold 10 µs Delay from completion of a serial bus
lect Hold Time sequence to changes of transceiver select sta- 35
tus 36
Aborted sequence - bus release Deselect_Abort 2 ms Delay from a host asserting Mod_DeSel (at 37
any point in a bus sequence), to the XFP mod- 38
ule releasing SCL and SDA 39
40
41
42
word is received or sent by the transceiver. This address stays valid be- 1
tween operations as long as XFP power is maintained. The address “roll 2
over” during read and write operations is from the last byte of the 128 byte 3
memory page to the first byte of the same page.
4
In order to read both the lower page and currently selected upper page, 5
at least two read operations must be performed, one with start address of 6
ZERO and another with start address of 128. A single 256-byte read op- 7
eration with start address of ZERO would contain the 128 bytes of the 8
lower page twice, instead of the lower page followed by the currently se- 9
lected upper page. 10
11
12
4.5.3 READ OPERATIONS (CURRENT ADDRESS READ) 13
A current address read operation requires only the device address read 14
word (10100001) be sent, Figure 20. Once acknowledged by the XFP, the 15
current address data word is serially clocked out. The host does not re- 16
spond with an acknowledge, but does generate a STOP condition once 17
the data word is read. 18
19
< ---- XFP ADDRESS -
--> 20
21
HOST S M L R N S 22
T S S E A T
A B B A C O 23
R D K P
T 24
1 0 1 0 0 0 0 1 0 x x x x x x x x 1 25
XFP A M L
26
C S
K B
S
B
27
28
<------ DATA WORD ------>
29
30
Figure 20 XFP Current Address Read Operation
31
32
4.5.4 READ OPERATIONS (RANDOM READ) 33
A random read operation requires a “dummy” write operation to load in the 34
target byte address Figure 21. This is accomplished by the following se- 35
quence: The target 8-bit data word address is sent following the device 36
address write word (10100000) and acknowledged by the XFP. The host 37
then generates another START condition (aborting the dummy write 38
without incrementing the counter) and a current address read by sending
39
a device read address (10100001). The XFP acknowledges the device
address and serially clocks out the requested data word. The host does 40
41
42
1
2
< ---- XFP ADDRESS - <--- MEMORY ADDRESS < ---- XFP ADDRESS - 3
--> --> -->
4
HOST S M L W M L S M L R N S
5
T S
A B
S R
B
S
B
S
B
T S
A B
S E
B A
A
C
T
O
6
R I R D K P 7
T T
T 8
E
9
1 0 1 0 0 0 0 0 0 x x x x x x x x 0 1 0 1 0 0 0 0 1 0 x x x x x x x x 1
10
XFP A
C
A
C
A M
C S
L
S
11
K K K B B 12
<------ DATA WORD ------> 13
14
Figure 21 XFP Random Read 15
16
4.5.5 READ OPERATIONS (SEQUENTIAL READ) 17
18
Sequential reads are initiated by either a current address read Figure 22
or a random address read Figure 23. To specify a sequential read, the 19
host responds with an acknowledge (instead of a STOP) after each data 20
word. As long as the XFP receives an acknowledge, it shall serially clock 21
out sequential data words. The sequence is terminated when the host re- 22
sponds with a NACK and a STOP instead of an acknowledge. 23
24
25
26
< ---- XFP ADDRESS -
--> 27
HOST S M L R A A N S 28
T S
A B
S E
B A
C
K
C
K
A
C
T
O
29
R
T
D K P 30
31
1 0 1 0 0 0 0 1 0 x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 1
32
XFP A M L M L M L
C S S S S S S 33
K B B B B B B
34
<---- DATA WORD n ----> <---- DATA WORD n+1 ---
->
<---- DATA WORD n+x ---
->
35
36
Figure 22 Sequential Address Read Starting at XFP Current Address 37
38
39
40
41
42
1
2
< ---- XFP <--- MEMORY < ---- XFP 3
ADDRESS ---> ADDRESS --> ADDRESS --->
4
S M L W M L S M L R A A N S
H
O T S S R S S T S S E C C A T 5
A B B I B B A B B A K K C O
S
T
R T R D K P 6
T E T
7
1 0 1 0 0 0 0 0 0 x x x x x x x x 0 1 0 1 0 0 0 0 1 0 x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 1
8
X
F
A
C
A
C
A M
C S
L
S
M
S
L
S
M
S
L
S
9
K K K B B B B B B
P 10
<---- DATA WORD n <---- DATA WORD <---- DATA WORD 11
----> n+1 ----> n+x ---->
12
Figure 23 Sequential Address Read Starting with Random XFP Read 13
14
15
4.5.6 READ OPERATION PACKET ERROR CHECKING 16
For packet error checking, the host shall begin with the XFP device Ad- 17
dress and Starting Memory Address, followed by the number of bytes to 18
be read. The XFP CRC-8 calculation requires an explicit starting memory 19
address be incorporated as the first byte processed by the algorithm, fol-
20
lowed by the number of bytes to be read, and finally all data bytes. After
the host reads the required number of bytes, the CRC-8 value will be sent 21
by the XFP module. The host will follow with NACK and STOP to complete 22
the READ operation. XFP read operation with packet error checking is il- 23
lustrated in Figure 24. The CRC-8 value is calculated on the starting 24
memory address (MSB first) followed by the number of bytes and the 25
Read Bytes in the order they are sent. The host may then verify the CRC-
26
8 value and reject the read if the CRC-8 value does not correspond to the
received data and desired Read address. Note the XFP address itself is 27
not included in the CRC-8 calculation. 28
29
Only reads of 1-128 bytes are supported. Any transaction that does not 30
follow the protocol described in this section will result in an unpredictable 31
response from the module when PEC is enabled. 32
33
34
35
36
37
38
39
40
41
42
1
2
< ---- XFP <----- MEMORY < ---- Number of < ---- XFP 3
ADDRESS ---> ADDRESS ----> BYTES ----> ADDRESS --->
4
H
O
S M
T S
L W
S R
M
S
L
S
M
S
L
S
S M
T S
L R
S E
5
S
T
A B
R
B I
T
B B B B A B
R
B A
D
6
T E T 7
1 0 1 0 0 0 0 0 0 x x x x x x x x 0 x x x x x x x x 0 1 0 1 0 0 0 0 1 0 x x x x x x x x 8
9
X A A A A M L
F C C C C S S 10
P K K K K B B
11
<---- DATA WORD 1 ---
-> 12
13
14
15
H A A N S
O C C A T 16
S K K C O
T K P 17
x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 1
18
19
X M L M L M L
F S S S S S S 20
P B B B B B B
21
<---- DATA WORD n-1 <----- DATA WORD n --- <----- CRC-8 ---->
----> -> 22
23
Figure 24 XFP READ OPERATION PACKET ERROR CHECKING 24
25
26
4.5.7 WRITE OPERATIONS (BYTE WRITE)
27
A write operation requires an 8-bit data word address following the device 28
address write word (10100000) and acknowledgement Figure 25. Upon
29
receipt of this address, the XFP shall again respond with a zero (ACK) to
acknowledge and then clock in the first 8 bit data word. Following the re- 30
ceipt of the 8 bit data word, the XFP shall output a zero (ACK) and the host 31
master must terminate the write sequence with a STOP condition for the 32
write cycle to begin. If a START condition is sent in place of a STOP con- 33
dition (i.e. a repeated START per the I2C specification) the write is aborted 34
and the data received during that operation is discarded. Upon receipt of 35
the proper STOP condition, the XFP enters an internally timed write cycle, 36
tWR, to internal memory. The XFP disables it’s management interface
37
input during this write cycle and shall not respond or acknowledge subse-
38
quent commands until the write is complete. Note that I2C “Combined 39
Format” using repeated START conditions is not supported on XFP write
40
commands.
41
42
1
2
< ---- XFP ADDRESS - <--- MEMORY ADDRESS <-------- DATA WORD ----- 3
--> --> ->
4
HOST S M L W M L M L S
T S S R S S S S T 5
A B B I B B B B O
R T P 6
T E
7
1 0 1 0 0 0 0 0 0 x x x x x x x x 0 x x x x x x x x 0 8
XFP A A A 9
C C C
K K K 10
11
12
Figure 25 XFP Write Byte Operation 13
14
15
4.5.8 WRITE OPERATIONS (SEQUENTIAL WRITE) 16
XFP’s shall support up to a 4 sequential byte write without repeatedly 17
sending XFP address and memory address information. A “sequential” 18
write is initiated the same way as a single byte write, but the host master 19
does not send a stop condition after the first word is clocked in. Instead,
20
after the XFP acknowledges receipt of the first data word, the host can
transmit up to three more data words. The XFP shall send an acknowl- 21
edge after each data word received. The host must terminate the sequen- 22
tial write sequence with a STOP condition or the write operation shall be 23
aborted and data discarded. Note that I2C “combined format” using re- 24
peated START conditions is not supported on XFP write commands.” 25
26
.
27
28
< ---- XFP <--- MEMORY <------ DATA <------ DATA WORD 2 <------ DATA <------ DATA
ADDRESS ---> ADDRESS --> WORD 1 ------> ------> WORD 3 ------> WORD 4 ------> 29
H S M L W M L M L M L M L M L S 30
O T S
S A B
S R
B I
S
B
S
B
S
B
S
B
S
B
S
B
S
B
S
B
S
B
S
B
T
O
31
T R T P 32
T E
33
1 0 1 0 0 0 0 0 0 x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 0 x x x x x x x x 0
34
X A A A A A A
F C C C C C C 35
P K K K K K K
36
37
38
Figure 26 XFP Sequential Write Operation
39
40
41
42
The details of each memory space are described in the sections that 1
follow. All 2-Wire registers are read with bit 7 the MSB first. Nomenclature 2
for all registers more than 1 bit long are MSB-LSB. 3
4
5
6
2-Wire Serial Address
1010000X (A0H) 7
0- 8
Digital 9
Diagnostic 10
Functions 11
12
118
13
119-122 4 Byte Password Change
14
126 4 Byte Password Entry 15
Page Select Byte Entry 16
127 17
18
19
20
21
128- 128- 128- 128- 128-
22
23
Reserved XFP MSA
User Vendor 24
for Future Serial ID
EEPROM Specific Reserved 25
Diagnostic Data
Data Functions 26
Functions 223
224- 27
Vendor Specific
ID Data 28
255 255 255 255 255 29
Table 00h Table 01h Table 02h Table 03h-7Fh Table 80h-FFh
30
31
Figure 28 2-wire Serial Digital Diagnostic Memory Map
32
33
The memory structure also provides for an optional password entry loca- 34
tion in the lower memory space that may be used to protect vendor in- 35
ternal functions or user writable memory. Passwords shall not be required 36
to read any serial ID or diagnostics information in the lower memory ad- 37
dress space or in Tables 00h – 02h. Nor shall passwords be required to
38
write any controls defined in the digital diagnostic functions described in
this document. Passwords may be used by vendors to control write ac- 39
cess to MSA defined read only data for factory setup, or to OEMs to limit 40
write access in the User EEPROM Table (02h). Finally, passwords may be 41
42
used to control read or write access to the vendor specific tables 03h – 1
7Fh. 2
3
Separate Passwords value ranges will be defined for host vendor pass-
4
words and module vendor passwords to prevent accidental writing into
critical module control areas by the host vendor. Details are defined in 5.5. 5
6
Note: Unless specifically noted, all informative ID fields must be filled out. 7
Using a value of 0 to indicate a field is unspecified (as is common in the 8
SFP definition) is not permitted. 9
10
11
5.0.1 APPLICABLE DOCUMENTS 12
13
Digital Diagnostic Monitoring Interface for Optical Transceivers SFF doc-
ument number: SFF-8472, rev. 9.3 August 4, 2002. 14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
5.1 DESCRIPTION OF LOWER MEMORY MAP: CONTROL FUNCTIONS, DIAGNOSTICS, TABLE ACCESS 1
The lower 128 Bytes of the 2-wire serial bus address space is used to ac- 2
cess a variety of measurements and diagnostic functions, to implement a 3
set of control functions, and finally to select which of the various upper 4
memory map tables are accessed on subsequent reads. This portion of 5
the address space is always directly addressable and thus is chosen for 6
diagnostics and control functions that may need to be repeatedly ac-
7
cessed.
8
Table 31 shows the general memory map of the lower 128 Byte address 9
space. 10
11
12
Table 31 Lower Memory Map 13
14
Byte Address Description 15
0 Identifier (1 Byte)
16
17
1 Signal Conditioner Control
18
2-57 Threshold Values used for Alarm and Warning Flags 19
(56 Bytes)
20
58-59 Optional VPS Control Registers(2 Byte) 21
60-69 Reserved (10 Bytes) 22
23
70-71 BER Reporting
24
72-75 Wavelength Control Registers (4 Bytes)
25
76-79 FEC control Registers (4 Bytes) 26
80-95 Flags and Interrupt Control (16 Bytes) 27
28
96-109 A/D readout (14 Bytes)
29
110-111 General Control/Status bits (2 Bytes) 30
112-117 Reserved (6 Bytes) 31
118 Serial Interface Read/Write Error Checking (1 Bytes) 32
33
119-122 Password Change Entry Area (Optional) (4 Bytes)
34
123-126 Password Entry Area (optional) (4 Bytes) 35
127 Page Select Byte 36
37
38
39
40
41
42
5.2 IDENTIFIER 1
Byte 0 of the lower memory map contains the module identifier value. The 2
identifier value specifies the physical device described by the serial ID in- 3
formation. This value is also included in the serial ID data table (01h). The 4
defined identifier values are shown in Table 32. The XFP transceiver 5
should use identifier 06h. 6
7
8
Table 32 Identifier values Byte 128 9
10
Value Description of Physical Device
11
00h Unknown or unspecified 12
01h GBIC 13
14
02h Module/connector soldered to motherboard
15
03h SFP 16
04h 300 pin XBI 17
05h XENPAK
18
19
06h XFP
20
07h XFF 21
08h XFP-E 22
23
09h XPAK
24
0Ah X2
25
0Ah-7Fh Reserved 26
80h-FFh Vendor specific 27
28
29
30
5.3 SIGNAL CONDITIONER CONTROL 31
The XFP MSA defines two modes of transmitter signal conditioner opera- 32
tion. The default mode, which must be included in all XFP implementa- 33
tions, uses a REFCLK which need not be synchronous with the data and 34
with relatively loose jitter requirements. This mode is generally intended 35
for Clock and Data Recovery techniques for retiming the transmitter data.
36
An optional mode is defined whereby the host system provides a synchro-
nous REFCLK with relatively tight frequency and jitter requirements. This 37
mode is intended to allow the use of a Clock Multiplication (CMU) based 38
retiming scheme. The electrical requirements for these modes are defined 39
in Section 3.10. 40
41
42
module is supplied and can operate normally with a 1.8-V voltage on its 1
VCC2 pins. 2
3
An XFP module might include a discrete or integrated regulator, which
4
supplies the CMOS ICs with a voltage lower than 1.8-V by stepping down
the VCC2 supply. The module powers up in the default mode of operation 5
capable of operating normally with 1.8-V on the VCC2 pins. 6
7
In the first optional mode (“operational LV regulator mode”) the regulator 8
is still operational but the host system supplies the module with a VCC2 9
voltage lower than +1.8-V. In the second optional mode (“bypassed regu- 10
lator mode”) the regulator is bypassed and the CMOS ICs are connected
11
directly to the VCC2 pins of the module. Serial ID Byte 221 (Table 01h),
bit 7 indicates if optional VPS is implemented. VPS control registers are 12
located in Bytes 58-59 as defined in the Table 33. 13
14
15
16
Table 33 VPS Fields
17
Byte Bit Name Description 18
19
58 7-4 Lowest Voltage Supported with an Operational LV Read Only
Regulator See Note 1, 2
20
21
58 3-0 Voltage Supplied on the VCC2 Pins. Read Write. Powers up to 0000b.
22
See Note 1.
23
59 7-4 Voltage Supported with a Bypassed Regulator. Read Only. 24
See Note 1.
25
59 3-1 Reserved 26
59 0 Regulator bypass mode 0b = Mode disabled. Powers up to 0b. 27
1b = Mode enabled 28
1. The values in these fields are unsigned 4-bit binary integers (INT[3:0]. To translate to absolute voltage use: V[absolute] = (1.8 V)- 29
(INT[3:0])*(0.1-V). 30
2. Any XFP compliant module supporting the optional LV regulator mode should be operational with any power supply voltage between 31
1.8 V and the value signified on Byte58, Bits 7-4.
32
33
34
35
Modules which implement an optional voltage mode must monitor the 36
VCC2 rail as part of the Auxiliary Monitoring described in sections 5.42. 37
38
When a host selects one of the optional voltage modes, the module will
39
automatically adjust the appropriate threshold and alarm register values
described in Table 35. The module will issue an Interrupt if the Vcc2 40
voltage is ever outside the acceptable range. 41
42
When a host selects one of the optional voltage modes, power and current 1
requirements of the module should scale at least linearly with voltage (this 2
assumption is made to simplify the specification - in most cases power 3
savings will be larger than linear). Implementing these optional modes en-
4
ables a module to fall into two classes of power dissipation according to
the mode of operation. However, when a module is switched to one of the 5
optional modes it will not change the values in Table 54. The host can use 6
the default values in Table 54 to calculate the minimum power savings of 7
the optional modes. 8
9
All voltage setting changes should take place when the module is in the 10
low power, stand-by mode (i.e. P_Down pin is held high). The 2-wire serial
11
bus and XFP interface signals must remain fully functional during voltage
setting changes, and the falling edge of P_Down should not reset these 12
changes. To achieve the lowest power operation under any condition a 13
host supporting an optional voltage mode should hold the P_Down pin 14
high on empty module sockets. 15
16
To enable the “operational LV regulator mode” the following sequence
17
must occur:
18
1. Host sets the module into the stand-by mode by holding the 19
P_Down/RST pin asserted. During stand-by mode the module trans- 20
mitter should be disabled. 21
2. Host reads the content of Byte 58 and 59 to determine the most de- 22
sirable setting for VPS and changes the voltage on VCC2 pins of the 23
XFP connector. 24
3. Host then writes Bits 3-0 of Byte58 to indicate the new voltage and 25
module updates settings in Table 35. 26
4. When P_Down is de-asserted module verifies the voltage on VCC2 27
is within range. If voltage is within range, module enables all circuitry 28
within the module and resumes with normal operation. If voltage is 29
not within range, module sets the appropriate voltage monitoring In- 30
terrupt bit to inform the host and asserts Mod_NR. 31
To enable the “bypassed regulator mode” the host should follow the se- 32
quence described above, with the only difference that bit 0 of Byte 59 is 33
written to 1b in step 3. 34
35
To return the module to the default mode (VCC2=1.8-V) the host must as-
36
sert P_Down and return the R/W fields of Bytes 58 and 59 to their default
values. 37
38
In modules that implement both optional modes the value written in Byte 39
58 takes precedence over the value written in Byte 59, i.e. if both of Byte 40
58, bits 3-0 and Byte 59, bit 0 are non-zero the module will be in “opera- 41
tional LV regulator mode”.
42
Two identifier bits indicating the availability of these optional modes are 1
defined in the Serial ID Page in Byte 221, Bits 3-4 Page 01h. Bit 4 corre- 2
sponds to the availability of the “operational LV regulator mode”. Bit 3 cor- 3
responds to the availability of the “bypassed regulator mode”. XFP
4
modules not implementing these optional modes will return 0 from these
locations as well as from all fields of bytes 58-59. 5
6
7
8
5.8 SYSTEM BER REPORTING 9
A host system implementing a forward error correction device can report 10
receive BER. The 2-Wire host management interface can transfer FEC re- 11
ported BER data to the XFP module. The module may use this information 12
to improve system BER. Implementation of BER Reporting functions is in- 13
dicated in Serial ID Byte 220 (Table 01h), bit 4. Control registers are lo-
14
cated in Bytes 70 and 71 (see Table 36).
15
Byte 70 requests the desired level of BER from the XFP module. Byte 71 16
reports the FEC actual BER to the XFP module. The data format for Byte 17
70 it is given by: 18
19
20
21
Byte ( 70 ( MSB – 4bits ) ) – [ Byte ( 70 ( LSB – 4bits ) ) ]
BER = ----------------------------------------------------------- ×10 22
16 23
24
Similarly for byte 71 is given by: 25
26
27
28
Byte ( 71 ( MSB – 4bits ) ) – [ Byte ( 71 ( LSB – 4bits ) ) ] 29
BER = ----------------------------------------------------------- ×10
16 30
31
MSB-Octet is the decimal value of the 4 MSB bits. LSB-Octet is the dec- 32
imal value of the 4 LSB bits. If byte 71 reports Hexadecimal value of 4C 33
then the BER would be 0.25x10E-12. 34
35
36
5.9 WAVELENGTH CONTROL 37
38
This section of memory includes registers (Table 37) that allow the imple-
mentation of a wavelength tunable XFP module. Implementation of 39
wavelength tunability is indicated in Serial ID Byte 221 (Table 01h), bit 1. 40
Bytes 72-73 are used to input the desired wavelength setpoint in the same 41
42
format described for XFP center wavelength in the Serial ID Byte 186-187 1
(units of 0.05 nm). For a wavelength tunable module, the serial ID Bytes 2
for wavelength and tolerance will define the allowed tuning range. Inputs 3
to tuning control bytes 72-73 which are outside this range will be clipped
4
to the allowed tuning range as defined in the serial ID tuning range.
5
Bytes 74 and 75 are optional monitor values that can be used to report the 6
measured error between the actual wavelength and the entered setpoint. 7
It is a 16 bit signed 2’s complement value in units of 0.005. Thus: 8
9
Reported Error [Bytes 74,75] =(Meas. Wavelength – Setpoint [Bytes 10
72,73])*(0.005nm)
11
Finally, the latched flag in Byte 85 bit 5 may be used to report a wave- 12
length error exceeding the manufacturer’s tolerance, and this condition 13
may be configured by the user as an interrupt trigger as described in the 14
Interrupt control section below. 15
16
17
18
5.10 FEC CONTROL
19
In some forward error correction schemes, the host system may optimize
20
the sensitivity of an optical link or minimize the error rate by adjusting the
phase and amplitude threshold for data quantization in the limiting ampli- 21
fier section of the module receive path. Implementation of FEC control 22
functions is indicated in Serial ID Byte 221 (Table 01h), bit 2. Control reg- 23
isters are located in Bytes 76 and 77 (see Table 38). 24
25
Byte 76 is used to set the amplitude threshold of receive path quantiza- 26
tion. It is a 2's complement 7 bit value (-128 - +127), where the threshold
27
is given by:
28
29
Amplitude Threshold = 50% + [Byte(76)/256] *100%
30
31
Similarly, Byte 77 is used to set the phase point in the eye diagram for
32
quantization, and is set in terms of unit interval relative to the eye center:
33
Phase setpoint = 0.5 UI + [Byte(77)/256] UI 34
35
36
37
38
39
40
41
42
sponding Byte. Thus a read of all Bytes from 80 – 87 can be used to reset 1
all latched flags and deassert the hardware Interrupt output pin. 2
3
Because of the bytewise nature of the 2-wire serial bus reads, it is not pos-
4
sible to clear individual flag bits. It is recommended that all latched flag
bits be read and cleared in the same operation. 5
6
The host system may control which flags result in an Interrupt by setting 7
high individual bits from a set of masking bits in Bytes 88-95, and de- 8
scribed in Table 40. A 1 value in a masking bit prevents the assertion of 9
the hardware Interrupt pin by the corresponding latched flag bit. 10
11
Masking bits should be volatile and startup with all unmasked (masking
bits 0). 12
13
Hardware Interrupt Pin = Logical NOR of all (Latched Flag Bit AND 14
NOT Masking Bit) 15
16
The mask bits may be used to prevent continued interruption from on-
17
going conditions, which would otherwise continually reassert the hard-
ware Interrupt pin. 18
19
5.11.1 GENERAL CONTROL AND STATUS BITS 20
In order to allow virtually complete operation of the XFP module via the 2- 21
wire serial interface, a set of control and status bits are provided to mirror 22
the operation of key status and control pins. These are provided in Bytes 23
110-111 and are detailed in Table 42. All status bits except soft bits in this 24
set are dynamic (not latched) 25
26
The bits fall into three categories:
27
• Bits mirroring hardware status lines: TX-DIS, MOD_NR, 28
P_Down/RST, Interrupt, RX_LOS 29
• Bits providing more specific information: TX_NR, RX_NR, TX and RX 30
CDR Loss of Lock, and TX Fault. 31
• A set of read/write bits which can be used to implement the optional 32
“soft” control of TX-Disable and the Power-Down functions. (De- 33
scribed fully in Table 42). Availability of this option is defined in Byte 34
221 (Table 01h). 35
36
37
38
39
40
41
42
. 1
2
Table 35 Alarm and Warning Thresholds 3
4
Address # Bytes Name Description
5
02-03 2 Temp High Alarm MSB at low address 6
04-05 2 Temp Low Alarm MSB at low address 7
06-07 2 Temp High Warning MSB at low address
8
9
08-09 2 Temp Low Warning MSB at low address
10
10-17 8 Reserved A/D Flag Thresholds Reserved A/D Flag Thresholds 11
18-19 2 Bias High Alarm MSB at low address 12
13
20-21 2 Bias Low Alarm MSB at low address
14
22-23 2 Bias High Warning MSB at low address
15
24-25 2 Bias Low Warning MSB at low address 16
26-27 2 TX Power High Alarm MSB at low address 17
18
28-29 2 TX Power Low Alarm MSB at low address
19
30-31 2 TX Power High Warning MSB at low address 20
32-33 2 TX Power Low Warning MSB at low address 21
34-35 2 RX Power High Alarm MSB at low address 22
23
36-37 2 RX Power Low Alarm MSB at low address
24
38-39 2 RX Power High Warning MSB at low address 25
40-41 2 RX Power Low Warning MSB at low address 26
42-43 2 AUX 1 High Alarm MSB at low address
27
28
44-45 2 AUX 1 Low Alarm MSB at low address
29
46-47 2 AUX 1 High Warning MSB at low address 30
48-49 2 AUX 1 Low Warning MSB at low address 31
32
50-51 2 AUX 2 High Alarm MSB at low address
33
52-53 2 AUX 2 Low Alarm MSB at low address
34
54-55 2 AUX 2 High Warning MSB at low address 35
56-57 2 AUX 2 Low Warning MSB at low address 36
37
38
39
40
41
42
1
2
3
Table 36 BER Control Fields 4
5
Byte Bit Name Description
6
70 All Acceptable BER Acceptable BER Reported by the FEC to the Module 7
71 All Actual BER Actual BER Reported by the FEC to the Module 8
9
10
11
12
Table 37 Wavelength Control Fields
13
Byte Bit Name Description 14
15
72 - 73 All Wavelength Set MSB User input of Wavelength setpoint. (Units of 0.05 nm)
16
74 - 75 All Wavelength Error LSB Monitor of Current Wavelength Error (Units of 17
0.005nm) [Signed 2’s complement value]
18
19
20
21
22
Table 38 FEC Control Fields 23
24
Byte Bit Name Description
25
76 All Amplitude Adjustment Relative amplitude of receive quantization threshold 26
77 All Phase Adjustment Phase of receive quantization relative to 0.5 UI. 27
78-79 All Reserved
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
Table 39 Latched Interrupt Flag Fields
3
Address # Bit Name Description 4
5
80 7 L- Temp High Alarm Latched high Temperature alarm.
6
80 6 L- Temp Low Alarm Latched low Temperature alarm. 7
80 5 Reserved 8
80 4 Reserved 9
10
80 3 L- TX Bias High Alarm Latched high TX Bias alarm.
11
80 2 L- TX Bias Low Alarm Latched low TX Bias alarm.
12
80 1 L- TX Power High Alarm Latched high TX Power alarm. 13
80 0 L- TX Power Low Alarm Latched low TX Power alarm. 14
81 7 L- RX Power High Alarm Latched high RX Power alarm. 15
16
81 6 L- RX Power Low Alarm Latched low RX Power alarm.
17
81 5 L- AUX 1 High Alarm Latched high AUX1 monitor alarm.
18
81 4 L- AUX 1 Low Alarm Latched low AUX1 monitor alarm. 19
81 3 L- AUX 2 High Alarm Latched high AUX2 monitor alarm. 20
81 2 L- AUX 2 Low Alarm Latched low AUX2 monitor alarm. 21
22
81 1 Reserved
23
81 0 Reserved
24
82 7 L- Temp High Warning Latched high Temperature warning. 25
82 6 L- Temp Low Warning Latched low Temperature warning. 26
82 5 Reserved 27
28
82 4 Reserved
29
82 3 L- TX Bias High Warning Latched high TX Bias warning.
30
82 2 L- TX Bias Low Warning Latched low TX Bias warning. 31
82 1 L- TX Pow. High Warning Latched high TX Power warning. 32
82 0 L- TX Pow. Low Warning Latched low TX Power warning. 33
34
83 7 L- RX Pow. High Warning Latched high RX Power warning.
35
83 6 L- RX Pow. Low Warning Latched low RX Power warning.
36
83 5 L- AUX 1 High Warning Latched high AUX1 monitor warning. 37
83 4 L- AUX 1 Low Warning Latched low AUX1 monitor warning. 38
83 3 L- AUX 2 High Warning Latched high AUX2 monitor warning. 39
40
83 2 L- AUX 2 Low Warning Latched low AUX2 monitor warning.
41
83 1 Reserved
42
. 1
2
Table 40 Interrupt Masking Bits 3
4
Address # Bit Name Description
5
88 7 M- Temp High Alarm Masking bit for high Temperature alarm. 6
88 6 M- Temp Low Alarm Masking bit for low Temperature alarm. 7
88 5 Reserved 8
9
88 4 Reserved
10
88 3 M- TX Bias High Alarm Masking bit for high TX Bias alarm.
11
88 2 M- TX Bias Low Alarm Masking bit for low TX Bias alarm. 12
88 1 M- TX Power High Alarm Masking bit for high TX Power alarm. 13
88 0 M- TX Power Low Alarm Masking bit for low TX Power alarm. 14
15
89 7 M- RX Power High Alarm Masking bit for high RX Power alarm.
16
89 6 M- RX Power Low Alarm Masking bit for low RX Power alarm.
17
89 5 M- AUX 1 High Alarm Masking bit for high AUX1 monitor alarm. 18
89 4 M- AUX 1 Low Alarm Masking bit for low AUX1 monitor alarm. 19
89 3 M- AUX 2 High Alarm Masking bit for high AUX2 monitor alarm. 20
21
89 2 M- AUX 2 Low Alarm Masking bit for low AUX2 monitor alarm.
22
89 1 Reserved
23
89 0 Reserved 24
90 7 M- Temp High Warning Masking bit for high Temperature warning. 25
90 6 M- Temp Low Warning Masking bit for low Temperature warning. 26
27
90 5 Reserved
28
90 4 Reserved
29
90 3 M- TX Bias High Warning Masking bit for high TX Bias warning. 30
90 2 M- TX Bias Low Warning Masking bit for low TX Bias warning. 31
90 1 M- TX Pow. Hi Warning Masking bit for high TX Power warning. 32
33
90 0 M- TX Pow. Low Warning Masking bit for low TX Power warning.
34
91 7 M- RX Pow. Hi Warning Masking bit for high RX Power warning.
35
91 6 M- RX Pow. Low Warning Masking bit for low RX Power warning. 36
91 5 M- AUX 1 High Warning Masking bit for high AUX1 monitor warning. 37
91 4 M- AUX 1 Low Warning Masking bit for low AUX1 monitor warning. 38
39
91 3 M- AUX 2 High Warning Masking bit for high AUX2 monitor warning.
40
91 2 M- AUX 2 Low Warning Masking bit for low AUX2 monitor warning.
41
42
1
2
Table 41 A/D Values 3
4
Byte Bit Name Description 5
6
Converted analog values. Calibrated 16 bit data.
7
96 All Temperature MSB Internally measured module temperature. 8
97 All Temperature LSB 9
10
98-99 All Reserved
11
100 All TX Bias MSB Internally measured TX Bias Current.
12
101 All TX Bias LSB 13
102 All TX Power MSB Measured TX output power. 14
15
103 All TX Power LSB
16
104 All RX Power MSB Measured RX input power. 17
105 All RX Power LSB 18
106 All AUX 1 MSB Auxiliary measurement 1 defined in Byte 222 Page 01h 19
20
107 All AUX 1 LSB
21
108 All AUX 2 MSB Auxiliary measurement 2 defined in Byte 222 Page 01h 22
109 All AUX 2 LSB 23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
Table 42 General Control/Status Bits
3
Byte Bit Name Description 4
5
110 7 TX Disable State Digital state of the TX Disable Input Pin. Updated within 100msec of change
on pin.
6
7
110 6 Soft TX Disable Optional read/write bit that allows software disable of laser. Writing '1' dis-
ables laser. Turn on/off time is 100 msec max from acknowledgement of
8
serial byte transmission. This bit is “OR”d with the hard TX_DISABLE pin 9
value. Note, per SFP MSA TX_DISABLE pin is default enabled unless 10
pulled low by hardware. If Soft TX Disable is not implemented, the trans-
ceiver ignores the value of this bit. Default power up value is 0. 11
12
110 5 MOD_NR State Digital state of the MOD_NR Pin. Updated within 100msec of change on
pin. 13
14
110 4 P_Down State Digital state of the P_Down Pin. Updated within 100msec of change on pin.
15
110 3 Soft P_Down Optional read/write bit that allows the module to be placed in the power 16
down mode. This is identical to the P_Down hardware pin function except
that it does not initiate a system reset. 17
18
110 2 Interrupt Digital state of the Interrupt output pin.
19
110 1 RX_LOS Indicates Optical Loss of Signal (per relevant optical link standard). Updated 20
within 100msec of change on pin.
21
110 0 Data_Not_Ready Indicates transceiver has achieved power up and A/D data is ready. Bit 22
remains high until data is ready to be read at which time the device sets the
bit low. 23
24
111 7 TX_NR State Identifies Not Ready condition as specific to the TX path
25
111 6 TX_Fault State Identifies Laser fault condition (Generated by laser safety system) 26
111 5 TX_CDR not Locked Identifies Loss of Lock in TX path CDR 27
111 4 RX_NR State Identifies Not Ready condition as specific to the TX path
28
29
111 3 RX_CDR not Locked Identifies Loss of Lock in RX path CDR
30
111 2-0 Reserved 31
32
33
The Data_Not_Ready bit is high during module power up and prior to the 34
first valid A/D reading. Once the first valid A/D reading occurs, the bit is 35
set low until the device is powered down. The bit must be set low within 1 36
second of power up. 37
38
39
40
41
42
1
2
Table 43 Packet Error Checking
3
Byte Bit Name Description 4
5
118 0 Error Checking 0 b = Disable Packet Error Checking
6
1b = Enable Packet Error Checking (see Section 4.5.6 and 4.5.9)
7
118 7-1 Reserved 8
9
.
10
Table 44 Password and Table Select Entry Bytes 11
12
Byte Bit Name Description 13
119-122 All New Password Entry Location of Entry of New Optional Password 14
15
123-126 All Password Entry Location for Entry of Optional Password
16
127 All Table Select Entry Location for Table Select Byte 17
18
Table 45 I/O Timing for Soft Control & Status Functions
19
Parameter Symbol Min Max Unit Conditions 20
21
TX_DISABLE assert time t_off 100 ms Time from TX_DIS bit set 1 until optical out-
put falls below 10% of nominal
22
23
TX_DISABLE deassert time t_on 100 ms Time from TX_DISABLE bit cleared1 until
optical output rises above 90% of nominal
24
25
P_Down assert time T_Pdown 100 ms Time from P_Down bit set1 until module dissi-
pation falls below 1.5W
26
27
P_Down deassert time T_Pup 300 ms Time from P_Down bit cleared1 until module
returns to normal operation.
28
29
RX_LOS assert time t_los_on 100 ms Time from RX_LOS state to RX_LOS bit set
30
RX_LOS deassert time t_los_off 100 ms Time from non-LOS state to RX_LOS bit 31
cleared
32
MOD_NR assert time t_NR_on 100 ms Time from TX_NR or RX_NR state to corre- 33
sponding bit set
34
MOD_NR deassert time t_NR_off 100 ms Time from non-TX_NR or RX_NR state to 35
corresponding bit set
36
Analog parameter data ready t_data 1000 ms From power on to data ready, bit 0 of byte 110 37
set
38
1. Measured from falling clock edge after stop bit of write transaction. 39
40
41
42
5.13 DESCRIPTION OF UPPER MEMORY MAP TABLE 00H – FUTURE DIAGNOSTICS FUNCTIONS 1
Table 00h is reserved for future diagnostics functions. 2
3
4
5
5.14 DESCRIPTION OF UPPER MEMORY MAP TABLE 01H – SERIAL ID MEMORY MAP 6
The serial ID memory map located in Table 01h in the upper address 7
space is used for read only identification information. 8
9
10
Table 46 Serial ID: Data Fields - Page 01h 11
12
Address
Size
Name Description 13
(Bytes)
14
Base ID Fields 15
16
128 1 Identifier Type of serial transceiver (see Table 32)
17
129 1 Ext. Identifier Extended identifier of type of serial transceiver (see Table 47) 18
130 1 Connector Code for connector type (see Table 48) 19
20
138-131 8 Transceiver Code for electronic compatibility or optical compatibility
(see Table 49) 21
22
139 1 Encoding Code for serial encoding algorithm (see Table 50)
23
140 1 BR-Min Minimum bit rate, units of 100 MBits/s. 24
141 1 BR-Max Maximum bit rate, units of 100 MBits/s. 25
142 1 Length(SMF)-km Link length supported for SMF fiber in km
26
27
143 1 Length (E-50µm) Link length supported for EBW 50/125 µm fiber, units of 2 m
28
144 1 Length (50 µm) Link length supported for 50/125 µm fiber, units of 1 m 29
145 1 Length (62.5 µm) Link length supported for 62.5/125 µm fiber, units of 1 m 30
31
146 1 Length (Copper) Link length supported for copper, units of 1m
32
147 1 Device Tech Device technology (see Table 51, Table 52)
33
163-148 16 Vendor name XFP vendor name (ASCII) 34
164 1 CDR Support CDR Rate Support (see Table 53) 35
36
167-165 3 Vendor OUI XFP vendor IEEE company ID
37
183-168 16 Vendor PN Part number provided by XFP vendor (ASCII) 38
185-184 2 Vendor rev Revision level for part number provided by vendor (ASCII) 39
187-186 2 Wavelength Nominal laser wavelength (Wavelength = value / 20 in nm) 40
41
42
1
2
Table 48 Connector values Byte 130
3
Register Description of connector 4
5
00h Unknown or unspecified
6
01h SC 7
02h Fibre Channel Style 1 copper connector 8
9
03h Fibre Channel Style 2 copper connector
10
04h BNC/TNC
11
05h Fibre Channel coaxial headers 12
06h FiberJack 13
14
07h LC
15
08h MT-RJ 16
09h MU 17
0Ah SG 18
19
0Bh Optical pigtail
20
0C-1Fh Reserved 21
20h HSSDC II 22
21h Copper Pigtail
23
24
22h-7Fh Reserved
25
80-FFh Vendor specific 26
27
28
29
5.18 INTERFACE SPECIFICATION 30
The following bit significant indicators define the electronic or optical inter- 31
faces that are supported by the transceiver Table 49. At least one bit shall 32
be set in this field. This table is not backward compatible with the
33
GBIC/SFP serial ID definition.
34
35
36
37
38
39
40
41
42
1
2
Table 49 Transceiver codes 3
4
Addr Bit Description of transceiver Addr Bit Description of transceiver
5
10 Gigabit Ethernet Compliance SONET/SDH Codes - Interconnect 6
131 7 10GBASE-SR 135 7 I-64.1r 7
131 6 10GBASE-LR 135 6 I-64.1 8
9
131 5 10GBASE-ER 135 5 I-64.2r
10
131 4 10GBASE-LRM 135 4 I-64.2
11
131 3 10GBASE-SW 135 3 I-64.3 12
131 2 10GBASE-LW 135 2 I-64.5 13
14
131 1 10GBASE-EW 135 1 Reserved
15
131 0 Reserved 135 0 Reserved
16
10 Gigabit Fibre Channel Compliance SONET/SDH Codes – Short Haul 17
132 7 1200-MX-SN-I 136 7 S-64.1 18
19
132 6 1200-SM-LL-L 136 6 S-64.2a
20
132 5 Extended Reach 1550 nm 136 5 S-64.2b
21
132 4 Intermediate Reach 1300 nm FP 136 4 S-64.3a 22
132 3 Reserved 136 3 S-64.3b 23
24
132 2 Reserved 136 2 S-64.5a
25
132 1 Reserved 136 1 S-64.5b
26
132 0 Reserved 136 0 Reserved 27
10 Gigabit Copper Links SONET/SDH Codes – Long Haul 28
29
133 7 Reserved 137 7 L-64.1
30
133 6 Reserved 137 6 L-64.2a
31
133 5 Reserved 137 5 L-64.2b 32
133 4 Reserved 137 4 L-64.2c 33
34
133 3 Reserved 137 3 L-64.3
35
133 2 Reserved 137 2 G.959.1 P1L1-2D2
36
133 1 Reserved 137 1 Reserved 37
133 0 Reserved 137 0 Reserved 38
39
Lower Speed Links SONET/SDH Codes – Very Long Haul
40
134 7 1000BASE-SX / 1xFC MMF 138 7 V-64.2a
41
42
1
2
5.31 VENDOR PN 3
The vendor part number (vendor PN) is a 16-byte field that contains ASCII 4
characters, left-aligned and padded on the right with ASCII spaces (20h), 5
defining the vendor part number or product name. A value of all zero in 6
the 16-byte field indicates that the vendor PN is unspecified.
7
8
9
5.32 VENDOR REV 10
The vendor revision number (vendor rev) is a 2-byte field that contains 11
ASCII characters, left-aligned and padded on the right with ASCII spaces 12
(20h), defining the vendor’s product revision number. A value of all zero in 13
the 2-byte field indicates that the vendor PN is unspecified. 14
15
16
17
5.33 LASER WAVELENGTH
18
Nominal transmitter output wavelength at room temperature. 16 bit value 19
with byte 186 as high order byte and byte 187 as low order byte. The laser
20
wavelength is equal to the 16 bit integer value divided by 20 in nm (units
of 0.05nm). This resolution should be adequate to cover all relevant wave- 21
lengths yet provide enough resolution for all expected DWDM applica- 22
tions. For accurate representation of controlled wavelength applications, 23
this value should represent the center of the guaranteed wavelength 24
range. 25
26
27
5.34 LASER WAVELENGTH TOLERANCE 28
29
The guaranteed +/- range of transmitter output wavelength under all
normal operating conditions. 16 bit value with byte 188 as high order byte 30
and byte 189 as low order byte. The laser wavelength is equal to the 16 31
bit integer value divided by 200 in nm (units of 0.005nm). Thus, the fol- 32
lowing two examples: 33
34
35
36
Example 1:
10GBASE-LR Wavelength Range = 1260 to 1355 nm 37
38
Nominal Wavelength in Bytes 186 - 187 = 1307.5 nm. 39
Represented as INT(1307.5 nm * 20) = 26150 = 6626h 40
41
42
1
2
Table 56 Diagnostic Monitoring Type
3
Data Address Bits Description 4
5
220 7-5 Reserved
6
220 4 Module Respond to FEC BER 7
0= No BER Support, 1=BER Support 8
220 3 Received power measurement type 9
0 = OMA, 1 = Average Power 10
220 2 Reserved 11
12
220 1- 0 Reserved
13
14
15
5.41 ENHANCED OPTIONS 16
“Enhanced Options” is a 1 byte field with 8 single bit indicators which de- 17
scribe the optional digital control and diagnostic features implemented in 18
the transceiver, as well as optional operating modes (see Table 57). 19
20
The optional digital controls are the Soft TX Disable and Power down func-
21
tions which allow the functions to be actuated via the 2-wire serial control
bus. 22
23
The optional operating modes indicate the ability of the module to perform 24
active FEC and wavelength tunability functions, as well as the ability to 25
support the optional Synchronous REFCLK mode. 26
27
.
28
Table 57 Enhanced Options 29
30
Data Address Bits Description 31
221 7 Module Supports Optional VPS
32
33
221 6 Optional Soft TX_DISABLE implemented
34
221 5 Optional Soft P_down implemented 35
221 4 Supports VPS LV regulator mode 36
37
221 3 Supports VPS bypassed regulator Mode
38
221 2 Active FEC control functions implemented
39
221 1 Wavelength tunability implemented 40
221 0 Optional CMU Support Mode 41
42
1
2
5.43 CC_EXT 3
The check code is a one byte code that can be used to verify that the first 4
31 bytes of extended serial information in the XFP is valid. The check 5
code shall be the low order 8 bits of the sum of the contents of all the bytes 6
from byte 192 to byte 222, inclusive.
7
8
9
5.44 VENDOR SPECIFIC ID FIELD 10
Bytes 224-255h of Table 01h may be used for Vendor Specific ID func- 11
tions. 12
13
5.45 DESCRIPTION OF UPPER MEMORY MAP TABLE 02H – USER EEPROM DATA 14
Table 02h is provided as user writable EEPROM. The host system may 15
read or write this memory for any purpose. If bit 3 of Table 01h Byte 129 16
is set, however, the first 10 bytes of Table 02h [128-137h] will be used to 17
store the CLEI code for the module. 18
19
20
21
5.46 DESCRIPTION OF UPPER MEMORY MAP TABLES 03H – 7FH VENDOR SPECIFIC FUNCTIONS
22
Tables 03h-7Fh are reserved for Vendor Specific functions.
23
24
25
5.47 DESCRIPTION OF UPPER MEMORY MAP TABLES 80H – FFH RESERVED 26
Tables 80h-FFh are reserved. 27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Figure 30 XFP Datum Alignment, Depth
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Figure 32 Drawing of XFP Transceiver, Part 2
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Figure 33 Recommended Pattern Layout for XFP Printed Circuit
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Figure 34 XFP Transceiver Electrical Pad Layout 40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Figure 36 XFP Detail Host Board Mechanical Layout, Detail Z 26
27
28
6.6 INSERTION, EXTRACTION AND RETENTION FORCES FOR XFP TRANSCEIVERS 29
The requirement for insertion forces, extraction forces and retention 30
forces are specified in Table 61. 31
32
33
34
Architecture Note 35
The XFP cage and module design combination must ensure excessive 36
force applied to a cable does not damage the XFP cage. If any part is 37
damaged by excessive force, it should be the cable or the media 38
module and not the cage which is part of the host system. 39
40
41
42
1
2
Table 61 Insertion, Extraction, and Retention Forces 3
Measurement Minimum Maximum Units Comments 4
5
XFP transceiver insertion 0 40 Newtons
6
XFP transceiver extraction 0 30 Newtons 7
XFP transceiver retention 90 N/A Newtons No damage to transceiver below 8
90N 9
Cage retention (Latch strength) 180 N/A Newtons No damage to the latch below 180N 10
Cage retention in Host Board 133 N/A Newtons Force to be applied in a vertical 11
direction with no damage to the 12
cage. 13
Insertion / removal cycles, connec- 100 N/A cycles 14
tor/cage 15
Insertion / removal cycles, XFP 50 N/A cycles 16
Transceiver 17
18
19
6.7 COLOR CODING AND LABELING OF XFP TRANSCEIVERS 20
21
An exposed feature of the XFP transceiver (a feature or surface extending
outside of the bezel) shall be color coded as follows: 22
23
• Beige for 850nm 24
• Blue for 1310nm 25
26
• White for 1550nm APD Receiver
27
• Red for 1550nm non-APD Receiver
28
Each XFP transceiver shall be clearly labeled. The complete labeling 29
need not be visible when the XFP transceiver is installed and the bottom 30
of the device is the recommended location of the label. Labeling shall in-
31
clude:
32
• Appropriate manufacturing and part number identification 33
• Appropriate regulatory compliance labeling 34
35
• A manufacturing traceability code
36
Also the label should include clear specification of the external port char- 37
acteristics such as:
38
39
40
41
42
• Optical wavelength 1
• Required fiber characteristics 2
3
• Operating data rate
4
• Interface standards supported
5
• Link length supported. 6
The labeling shall not interfere with the mechanical, thermal and EMI fea- 7
tures. 8
9
10
11
6.8 BEZEL AND EMI GASKET DESIGN FOR SYSTEMS USING XFP TRANSCEIVERS
12
Host enclosures that use XFP devices should provide appropriate clear-
13
ances between the XFP transceivers to allow insertion and extraction
without the use of special tools and a bezel enclosure with sufficient me- 14
chanical strength. For most systems a nominal centerline to centerline 15
spacing of 23.5 mm (0.925”) is sufficient. See Figure 37 and Figure 38 for 16
the recommended bezel designs. Figure 37 illustrates the single sided 17
mounting and Figure 38 illustrates the double sided mounting method. 18
The minimum recommended host board thickness for double sided 19
mounting of the assemblies is 3.0 mm minimum.
20
There are many options for a bezel EMI gasket that functions as a seal 21
between the bezel and the front of the cage. The design of the bezel EMI 22
gasket and the materials used for the gasket are application specific. The 23
preferred method is to fasten the gasket to the back of the bezel with a 24
pressure sensitive adhesive. Assembly of the host board to the bezel will 25
compress the gasket to the recommended range specified by the bezel
26
EMI gasket manufacturer. The surface in the back of the bezel that is in
contact with the bezel EMI gasket must be low resistance and connected 27
to ground. 28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Figure 38 Recommended Double Side Mounting Bezel Design
20
21
22
6.9 XFP CONNECTOR MECHANICAL SPECIFICATIONS 23
The XFP Connector is a 30-contact, right angle surface mount connector 24
and available from several manufacturers1. An example connector such 25
as 788862-2 manufactured by Tyco is shown in Figure 39. Newer versions 26
of this connector are available from Tyco, Molex, and Harting with im- 27
proved electrical performance.
28
29
30
31
32
33
34
35
36
37
38
1. 788862C (Standard PT) and 1367500-1 (New Improved PT Version), Mfg. by 39
Tyco Electronics, www.tycoelectronics.com. 40
Parecon 30 Mfg. by Harting-Electro-Optics, www.Harting-Electro-Optics.com. 41
74441, Mfg. by Molex Inc., www.molex.com.
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Figure 39 XFP Transceiver Connector Illustration 29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
6.10 XFP CAGE ASSEMBLY DIMENSIONS 3
The Cage Assembly requires EMI shielding capability for both front and 4
back portions of the cage along with providing guidance for the connector, 5
retention of the transceiver and features for heat sink attachment. The lo-
6
cation of the EMI gaskets for a reference design is illustrated in Figure 40
and a description of each EMI gasket is described in the sections below. 7
The dimensional requirements for the cage are illustrated in Figure 41. 8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Figure 40 XFP Cage Components
29
30
31
32
6.10.1 XFP CAGE HOUSING 33
The metal cage has compliant leads for assembly to the host board. The 34
cage material is copper alloy and the recommended plating options are: 35
36
• Tin-lead plate 2.54 micrometers minimum over copper flash
37
• Tin plate 2.54 micrometers minimum over 0.76 micrometers mini- 38
mum nickel
39
or the equivalent materials. 40
41
42
1
2
6.10.4 XFP FRONT FLANGE 3
4
The purpose of the front flange is to provide a flat surface to contact the 5
Bezel EMI Gasket (customer supplied) that is mounted on the back side
6
of the bezel. It also serves as a collar to retain the Front EMI Gasket and
to provide strength to the cage housing. The front flange material is zinc 7
alloy and the recommended plating options are: 8
9
• Tin-lead plate 2.54 micrometers minimum over copper flash 10
• Tin plate 2.54 micrometers minimum over 0.76 micrometers mini- 11
mum nickel 12
or equivalent materials. 13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Figure 41 XFP Cage Assembly 42
1
2
6.12 AN EXAMPLE XFP HEAT SINK 3
The heat sink is illustrated in Figure 43. Critical dimensions to insure that 4
the heat sink will be compatible with the Heat Sink Clip are defined. The 5
configuration of the fins or posts is application specific along with the out- 6
side envelope. The heat sink includes a beveled edge which “rides up”
the leading edge of the transceiver as the transceiver is inserted into the 7
cage assembly. The recommended material for the heat sink is aluminum 8
and the surface treatment for the transceiver contacting surface can be 9
anodizing or nickel plating. 10
.
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Figure 43 XFP Heat Sink 37
38
39
40
41
42
Figure 53 depicts the stack up for the Broadcom reference board. Total 1
board thickness is 36mils and the board material is Nelco FR4-13 with εr 2
= 4.0 and a loss tangent of 0.016. All traces are 1/2 oz. copper. 3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Figure 53 PCB Board Stack up 21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Figure 54 System model for the XFI Channel.
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Figure 55 Comparison of simulated and measured differential s11 21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Figure 56 Comparison of simulated and measured differential s21 42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Figure 57 Measured and simulated eye diagram for the XFP channel 22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Figure 60 Single-ended via with electrical path between adjacent 17
layers leaves an open-circuit stub causing poor impedance match.
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Figure 61 Preferred method for trace routing with vias. Routing the 34
electrical path from the top layer to a layer near or on the opposite side of 35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
(a) 100 mil (b) 62 mil
21
22
Figure 64 S-parameters for optimized via structures with
dimensions given in Table 1. (a) 100 mil thick PCB. (B.) 62 mil thick 23
board. 24
25
26
D.4 GSSG DIFFERENTIAL VIAS 27
An improvement to the traditional differential via is to utilize a ground- 28
signal-signal-ground (GSSG) geometry. The GSSG differential via con- 29
sists of four single-ended vias as shown in Figure 65. The two inner vias 30
are the differential signal lines and the outer two vias are ground return
31
lines. Although differential signaling generally provides for all return path
currents, the GSSG geometry can support common-mode signals with a 32
well-controlled return current path. Any common-mode signals that get 33
coupled or generated can propagate through the via (rather than being 34
scattered) and along the transmission line to the receiver where it is ter- 35
minated. 36
37
Critical dimensions for the GSSG geometry are via diameter (drill size),
pad diameter, gap between the pad and the ground/power plane cutout, 38
via pitch, and via-to-ground via. With the additional conductors of the 39
GSSG structure, there are more tunable parameters that can be used to 40
provide optimal performance. Fullwave 3D electromagnetic simulations 41
were performed to optimize the geometry on a 62-mil stack up. Table 65 42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
(a)S11 (B)S21 21
Figure 66 Extracted s-parameters for the four alternative GSSG via 22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
+Vpk 4
5
Data Eye 0 6
7
8
-Vpk
9
Zero Crossing 10
Histogram
11
12
Template Alignment 13
0 UI 1 UI 14
15
Figure 67 Eye Template Alignment 16
17
18
19
20
Oscilloscope 21
+Data
22
DUT - Data 23
24
Trigger 25
CRU 26
27
28
Figure 68 Eye mask measurement setup - block diagram. 29
30
A clock recovery unit (CRU) should be used to trigger the scope for mask 31
measurements as shown in Figure 68. It should have a high frequency 32
corner bandwidth of 4 MHz and a slope of –20 dB/decade with maximum 33
peaking of 0.1 dB.
34
35
36
37
38
39
40
41
42
stream with exception of the DCD and so cannot be equalized by the re- 1
ceiver under test. It can exhibit a wide spectrum. 2
3
Noticeable amounts of DCD are rare at compliance points B’, C, and D. If
4
DCD is used for compliance testing, it should not exceed 0.05 UI pk-pk.
DCD can be viewed when transmitting a repeating 1010 pattern. 5
6
Moderate amounts of RJ are possible at compliance points B’, C, and D, 7
typically attributable to the clock source and/or a CMU. Large amounts of 8
RJ are not common, though, and so if RJ is used for compliance testing, 9
the following advice is given: 10
11
• the RJ magnitude should not exceed 0.2 UI pk-pk (at 1E-12 probabil-
ity); 12
13
• the RJ should be filtered with a 10 MHz high-pass filter to reduce
sensitivity to variations in CDR tracking; 14
15
• the upper cutoff frequency for the RJ should be at least 80 MHz.
16
Single frequency PJ is sinuosidal jitter, and can be considered as another 17
form of non-EQJ jitter. However, single frequency PJ is not common in
18
normal operation and is harsh as a form of non-EQJ jitter for compliance
testing at B’, C, and D. If single frequency PJ is used, its frequency(s) 19
should be generally above the tracking bandwidth of the CDR of the port 20
being tested. Note - this use of sinuosidal jitter is in addition and should 21
not to be confused with the SJ tolerance mask templates specified in the 22
document. 23
24
A versatile option for PJ is to employ multi-tone jitter created by a PRBS
25
generator. PRBS jitter has a distributed spectrum and can be low-pass fil-
tered to produce a more normal probability density function (pdf) for non- 26
EQJ jitter. MJSQ and CEI have more information on this approach. 27
28
To create non-EQJ & ISI jitter, a section of PCB is recommended to emu- 29
late the behaviors expected in an application. The length and construction 30
details are left to the user, but when completed, the required values for
31
Total Jitter and non-EQJ jitter should be satisfied. Coaxial cable typically
exhibits a different response, both in magnitude and phase. 32
33
34
35
E.3.4 CALIBRATION 36
The signal should be calibrated differentially through a standard compli- 37
ance test fixture (see Appendix A:) and into standard instrumentation 38
loads. If complementary single-ended signals are used, be sure they are 39
carefully matched in both amplitude and phase.
40
41
42
Except for sinuosidal jitter, all calibration should use a single pole high- 1
pass jitter filter with -3 dB frequency at 4 MHz. 2
3
The jitter is defined at a probability level of 1E-12, consistent with the re-
4
quirements of Annex E.1. To calibrate the jitter, methods given in MSJQ
and CEI are recommended. Given random jitter and the nature of the long 5
test patterns, low probability jitter events may be present. A technique that 6
can accurately measure low probability events is recommended to avoid 7
an overly stressful test condition. 8
9
The vertical eye opening is defined at a probability level of 1E-12, consis- 10
tent with the requirements of Annex E.1. Like jitter, low probability vertical
11
closure may be present, and if not accurately measured and accounted
for in calibration, could lead to an overly stressful test condition. 12
13
14
15
E.3.5 TESTING 16
Operate the system with the appropriate data pattern for the application - 17
SONET, 10G Ethernet, 10GFC, or G.709 then test applicable compliance 18
points B’, C, or D.
19
All signals and reference clocks that operate during normal operation shall 20
be active during the test including the other signal path in the duplex pair. 21
The other signal path shall be asynchronous and up to the maximum al- 22
lowable differential PPM value given in the appropriate protocol specifica- 23
tions. 24
25
Apply the SJ tolerance masks, at each SJ tolerance frequency verify that
26
the DUT provides a BER of at least 1E-12 at the applicable compliance
points B’, C, or D. 27
28
29
30
E.4 JITTER PEAKING SPECIFICATIONS AND MEASUREMENTS 31
In Section 3.9 of the XFP MSA, module jitter peaking specifications are 32
given for datacom and telecom modules for both the receiver and trans- 33
mitter paths. In the case of telecom modules to be used for looptiming 34
based regenerator applications, the jitter peaking below 120 KHz is limited
35
to 0.03 dB for both receiver and transmitter. This last specification is de-
rived from the need for the overall system jitter peaking to be kept below 36
0.1 dB for regenerator applications. While these requirements are be- 37
lieved to be practical for existing signal conditioner and module optics 38
technology, verification of this performance presents considerable difficul- 39
ties due to limitations of current measurement equipment. This annex de- 40
scribes these difficulties and suggests alternative characterization
41
42
1
2
3
4
Function Generator
Sinewave 5
6
7
8
9
DC
Block 10
11
12
Ground 13
XFP 14
Power Supply Test 15
DC Voltage Board 16
17
18
Toriodal Inductor Test Point at 19
Connector
20
Figure 71 Power Supply Ripple Setup 21
22
23
24
Table 66 Noise Amplitude for XFP power supplies (Nominal) 25
26
Power Supply 0-1 MHz (2%) 1-10 MHz (3%)
27
5.0V 100mV 150mV 28
3.3V 66mV 99mV 29
30
1.8V 36mV 54mV
31
-5.2V 104mV 156mV 32
33
34
35
36
37
38
39
40
41
42
F.1.1 XFI ASIC/SERDES TRANSMITTER OUTPUT JITTER SPECIFICATIONS AT A FOR BER 1E-15 1
Table 68 is replacement forTable 9 XFI ASIC/SerDes Transmitter Output 2
Jitter Specifications at A. 3
4
Table 68 XFI ASIC/SerDes Transmitter Output Jitter Specifications at A for BER 1E-15 5
Transmitter - A Symbol Conditions Min Typ Max Units 6
7
Determinstic Jitter for Q of 16 DJ See E.1, 1 0.15 UI (p-p)
8
Total Jitter for Q of 16 TJ See E.1, 1 0.321 UI (p-p) 9
10
Eye Mask X1 0.161 UI
11
Eye Mask X2 0.4 UI 12
13
Eye Mask Y1 180 mV
14
Eye Mask Y2 385 mV 15
Jitter Generation for Telecom Applications 50KHz to 8 MHz 6.5 mUI (RMS)
16
see 2 17
1. In loop timing mode, includes jitter that transfers through the ASIC from the receiver during any valid operational input 18
conditions. 19
2. Measured with a filter with 50 KHz high-pass cutoff desinged frequency and 8 MHz low pass cut off frequency. The filter rolls off 20
at least for one decade on each side of the passband with -20 dB/Dec Does not apply to a host designed with Optional Synchronous 21
CMU clock, when used in conjunction with a Synchronous CMU module 3.10.1.
22
23
F.1.2 XFI ASIC/SERDES RECEIVER INPUT JITTER SPECIFICATIONS AT D FOR BER 1E-15
24
Table 69 is replacement forTable 11 XFI ASIC/SerDes Receiver Input
25
Jitter Specifications at D.
26
Table 69 XFI ASIC/SerDes Receiver Input Jitter Specifications at D for BER 1E-15 27
28
Receiver- D Symbol Conditions Min Typ Max Units 29
Total Jitter for Q of 16 TJ See E.1, E.2 0.65 UI (p-p) 30
31
Total non-EQJ Jitter for Q of 161 See E.1, E.2 0.45 UI (p-p)
32
Sinusoidal Jitter Tolerance SJ E.3 see 2 33
Eye Mask X1 Note 3 0.325 UI
34
35
Eye Mask Y1 55 mV 36
Eye Mask Y2 Note 4 525 mV 37
38
1. Total Jitter less ISI.
39
2. Sinuosidal jitter tolerance for Telecom and Datacom respectively given by Figure 9 and Figure 10.
40
3. Mask coordinate X1=0.225 if total non-EQJ jitter is measured.
41
4. Out of 525 mV, 100 mV is allocated for multiple reflection.
42
F.1.3 XFI HOST TRANSMITTER OUTPUT JITTER SPECIFICATIONS AT B FOR BER 1E-15 1
Table 70 is replacement forTable 13 XFI Host Transmitter Output Jitter 2
Specifications at B. 3
4
Table 70 XFI Host Transmitter Output Jitter Specifications at B for BER 1E-15 5
Receiver- B Symbol Conditions Min Typ Max Units
6
7
Total Jitter for Q of 161, TJ See E.1 0.61 UI
8
1
Total non-EQJ Jitter for Q of 16 See E.1, E.2 0.41 UI 9
Eye Mask X1 see 2 0.305 UI 10
11
Eye Mask Y1 60 mV
12
Eye Mask Y2 see 3 410 mV 13
Jitter Generation at B Telecom Applications 50KHz-8MHz, see 4 7 mUI (RMS) 14
15
1. In loop timing mode, includes jitter that transfers through ASIC from the receiver during any valid operational input conditions.
16
2. Mask cordinate X1=0.205 if total non-EQJ jitter is measured.
17
3. Out of 410 mV, 25 mV is allocated for multiple reflection.
4.Measured with a filter with 50 KHz high-pass cutoff desinged frequency and 8 MHz low pass cut off frequency. The filter rolls off 18
at least for one decade on each side of the passband with -20 dB/Dec. Does not apply to a host designed with Optional Synchronous 19
CMU clock, when used in conjunction with a Synchronous CMU Signal Conditioner module see 3.10.1. 20
21
22
23
F.1.4 XFI Host Receiver Input JITTER SPECIFICATIONS AT C FOR BER 1E-15
24
Table 71 is replacement forTable 15 XFI Host Receiver Input Jitter Speci-
25
fications at C.
26
Table 71 XFI Host Receiver Input Jitter Specifications at C 27
28
Transmitter - C Symbol Conditions Min Typ Max Units 29
Determinstic Jitter DJ See E.1 0.18 UI (p-p) 30
31
Total Jitter for Q of 16 TJ See E.1 0.363 UI (p-p)
32
Sinusoidal Jitter Tolerance SJ E.3 see 1 33
34
Eye Mask X1 0.181 UI
35
Eye Mask X2 0.42 UI 36
Eye Mask Y1 170 mV
37
38
Eye Mask Y2 see 2 425 mV 39
1. Sinuosidal jitter tolerance for Telecom and Datacom respectively given by Figure 13 and Figure 14. 40
2. Out of 425 mV, 25 mV is allocated for multiple reflections. 41
42
1
2
3
Max Allowable Conditions for Operation 4
Example of Module Temperature vs Air Flow 5
(Shall conform to IEC 60950-1)
65 6
Inlet Air Temp (C)
7
60
8
55 9
50
Module 1
10
45 Module 2 11
Module 4 12
40
13
35
14
30 15
25 16
0.5 1 1.5 2 2.5 3 17
Airflow (m/s) 18
19
Figure 72 Airflow vs Inlet Air Temperature 20
21
22
• Identical module types will be expected to be characterized I.E. 23
no mix of 850nm,1310nm, or 1550nm is mandated. 24
• When undergoing thermal evaluation, transceivers should output 25
bi-directional data using a PRBS 231-1 data pattern. 26
• Other measurement data provided is at the discretion of the ven- 27
dor. 28
• The system should provide uniform airflow across the vent open- 29
ing and be of constant volume airflow. 30
The test conditions for system A and system B are defined Table 74. 31
32
33
34
Table 74 Environmental Test Conditions
35
Parameter System A (PCI application) System B (Switch application) 36
37
Number of modules 1 or 2 1,2,4
38
Inlet air temperature 5˚C to 50˚C 5˚C to 40˚C 39
Altitude Sea Level 3000 m 40
Air Humidity 40% to 60% 40% to 60%
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Intentionally Left Blank
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42