1
2
3
4 SFP-DD MSA
5
7
8 for
9
10 SFP112 AND SFP DOUBLE DENSITY PLUGGABLE TRANSCEIVER
11
12 Revision 5.1
13
14 March 11, 2022
15
16
Abstract: This specification defines: the electrical and optical connectors, electrical signals and power supplies,
mechanical and thermal requirements of the pluggable SFP112 module, Double Density SFP-DD module, and
Double Density SFP-DD112 connector and cage system. This document provides a common specification for
systems manufacturers, system integrators, and suppliers of modules. 21
22
23
24
25 POINTS OF CONTACT:
Ali Ghiasi (Technical Editor) Ghiasi Quantum 19947 Lindenbrook Lane Cupertino, CA 95014 ali at ghiasiquantum dot com | Scott Sommers (Chair) Molex 2222 Wellington Court Lisle, IL 60532-1682 scott.sommers at molex dot com |
26
27
30
Limitation on use of Information:
This specification is provided "AS IS" with NO WARRANTIES whatsoever and therefore the provision of this
specification does not include any warranty of merchantability, noninfringement, fitness for a particular
purpose, or any other warranty otherwise arising out of any proposal, specification or sample. The authors
further disclaim all liability, including liability for infringement of any proprietary rights, relating to use of
information in this specification. No license, express or implied, by estoppel or otherwise, to any intellectual
property rights is granted herein. 38
39 Permissions:
40 You are authorized to download, reproduce and distribute this document. All other rights are reserved. The
41 provision of this document should not be construed as the granting of any right to practice, make, use or
42 otherwise develop products that are based on the document. Any and all IP rights related to this document
43 and the designs disclosed within, except for the rights expressly mentioned above, are reserved by the
44 respective owners of those IP rights. 45
46
1 The following are Promoter member companies of the SFP-DD MSA.
Alibaba Group | Hewlett Packard Enterprise | Lumentum |
Broadcom | Huawei | Molex |
Cisco | Intel | TE Connectivity |
Dell Technologies | Juniper Networks | |
II-VI | Nvidia |
2
3
4 The following are Contributor member companies of the SFP-DD MSA.
Accelink | Genesis Connected Solutions | Nokia |
Amphenol | Hisense Broadband | Senko |
AOI | Infinera | Source Photonics |
Eoptolink | InnoLight | US Conec |
Foxconn Interconnect Technology | Maxim Integrated | Yamaichi Electronics |
Fourte International | MultiLane | ZTE |
5
6
Revision | Date | Changes |
1.0 | September 14 2017 | 1st public release. |
2.0 | September 17, 2018 | 2nd public release. Added Type 2 module. Changed IntL pin to reserved, Changed TxFault to TxFault/Int Updated drawings to include new key. Added test conditions for insertion removal forces. |
3.0 | April 10 2019 | 3rd public release. Added 5 W power class. Added SN, MDC connectors. Deleted requirement that host shall not change the state of LPMode when module is present. |
4.0 | Withdrawn | |
4.1 | August 10, 2020 | 4th public release. Added ResetL, IntL, ePPS, Fault signals. Added timing tables for low speeds signals, soft control and status. Chapter 7-Management Interface is now part of Chapter 4-Electrical Specification. Port mapping, optical connectors, and module color coding moved out of Mechanical and Board Definition Chapter-5 and into a new Chapter-5. Appendix A- Normative Connector Performance Requirements added. |
4.2 | August 17, 2020 | 5th public release. Added dual functionality IntL/TXFaultDD signal definition for SFP-DD. |
5.0 | October 1, 2021 | 6th public release. Added chapter 5 Electrical Specifications for SFP112, added chapter 8 Mechanical specification of SFP-DD112, added chapter 9 Mechanical Specification of SFP112. Added ePPS/Clock signals definition for SFP-DD/SFP-DD112. |
5.1 | March 11, 2022 | 7th public release defined a new improved power supply test method 4.10, squelch level reduced to 50 mV for 112G operation 4.8.2. TWI bus timing removed from chapter 4 as identical timing diagram already included in CMIS. |
8
9
The development work on this specification was done by the SFP-DD MSA, an industry group. The
membership of the committee since its formation in May 2017 has included a mix of companies which are
leaders across the industry. 14
The members of the SFP-DD MSA would like to acknowledge the contributions of Edmund Poh. He was an
excellent engineer; his technical skills and collaborative attitude will be missed. 17
18
1 TABLE OF CONTENTS
2
4. Electrical Specification and Management Interface Timing for SFP-DD/SFP-DD112 14
4.3 Overview of the Low Speed Electrical Hardware Signals 17
34 4.8.1 RD0+, RD0-, RD1+, RD1 28
35 4.8.2 TD0+, TD0-, TD1+, TD1 29
7.2 SFP-DD/SFP-DD112/SFP112 Datums, Dimensions and Component Alignment 65
7.5 SFP-DD/SFP-DD112/SFP112 Module Flatness and Roughness 71
7.7 SFP-DD/SFP-DD112/SFP112 Module Extraction and Retention Forces 74
5
6
7
1 List of Tables
2
Table 7- Timing for SFP-DD/SFP-DD112 soft control and status functions 27
Table 12- Power Supply specifications, Instantaneous, sustained and steady state current limits 32
Table 13- Power Supply Output Noise and Tolerance Specifications 35
Table 17- Timing for SFP112 soft control and status functions 50
Table 21- Power Supply specifications, Instantaneous, sustained and steady state current limits 55
Table 22- Truncated Filter Response Coefficients for Host Power Supply Noise Output 56
Table 23- Power Supply Output Noise and Tolerance Specifications 57
34
35
1 List of Figures
2
Figure 2: SFP-DD/SFP-DD112/SFP112 Cage, Connector and Module 13
Figure 4: Example SFP-DD/SFP-DD112 Host Board Schematic for Optical Modules 20
Figure 5: Example SFP-DD/SFP-DD112 Host Board Schematic for active copper cables 21
Figure 6: Example SFP-DD/SFP-DD112 Host Board Schematic for passive copper cables 22
Figure 7: SDA/SCL options for pull-up resistor, bus capacitance and rise/fall times 24
Figure 10: Instantaneous and sustained peak currents for Icc Host (see Table 12) 33
Figure 16: Example of SFP112 Host Board Schematic for Optical Modules 41
Figure 17: Example of SFP112 Host Board Schematic for active copper cables 42
Figure 18: Example of SFP112 Host Board Schematic for passive copper cables 43
Figure 19: SDA/SCL options for pull-up resistor, bus capacitance and rise/fall times 45
Figure 24: SFP112 Recommended Host Board Power Supply Filtering 54
Figure 25: Instantaneous and sustained peak currents for Icc Host (see Table 21) 56
Figure 26: Truncated Transfer Response for Host Board Power Supply Noise Output measurement 57
Figure 28: Optical Media Dependent Interface port assignments 59
Figure 29: MPO-12 Single Row optical patch cord and module receptacle 60
Figure 30: Duplex LC optical patchcord and module receptacle 60
Figure 31: LC connector excursion above module nose height 61
Figure 33: SN optical connector plug and two-port module receptacle 61
Figure 34: MDC optical connector plug and two-port module receptacle 62
5
The scope of this specification is the definition of a high density 1-channel and 2-channels modules, cage and
connector system. SFP-DD supports up to 100 Gb/s in aggregate over a 2 x 50Gb/s electrical interface.
SFP112 supports 100 Gb/s over single electrical lane, and SFP-DD112 supports up to 200 Gb/s in aggregate
over a 2 x 100 Gb/s electrical interface. The cage and connector design provides backwards compatibility to
SFP+/SFP28 [30], [28] modules which can be inserted into an SFP-DD/SFP-DD112 cage and connector using
1 of the electrical channels. SFP112 cage and connectors are compatible with SFP+/SFP28 modules.
Furthermore, SFP112 modules maybe inserted into SFP-DD112 cage and connector using 1 of the electrical
lanes.
10
11 1.1 Overview of SFP-DD/SFP-DD112/SFP112 Specifications
12 SFP-DD/SFP-DD112/SFP112 specifications are organized into 10 chapters and one appendix addressing electrical,
13 mechanical, environmental, and management aspects of the module.
14
15 Chapter 1 Scope
16
17 Chapter 2 References and Related Standards and SFF Specifications
18
19 Chapter 3 Introduction
20
21 Chapter 4 Electrical specifications and management interface timing for SFP-DD/SFP-DD112 22
23 Chapter 5 Electrical specifications and management interface timing for SFP112 24
25 Chapter 6 Optical port mapping and optical interfaces
26
27 Chapter 7 SFP-DD Mechanical specifications, printed circuit board recommendations 28
29 Chapter 8 SFP-DD112 Mechanical specifications, printed circuit board recommendations 30
31 Chapter 9 SFP112 Mechanical specifications, printed circuit board recommendations
32
33 Chapter 10 Environmental and thermal considerations 34
35 Appendix A Normative module and connector performance requirements.
36
37
The following interface standards and specifications are relevant to this Specification.
4 | ANSI FC-PI-6 32GFC (INCITS 533) | |
5 | ANSI FC-PI-7 64GFC (INCITS 543) | |
6 | ANSI FC-PI-8 128GFC | |
7 | ASME Y14.5-2009 Dimensioning and Tolerancing | |
8 | Common Management Interface Specification (CMIS) 5.1, | |
9 | https://www.oiforum.com/wp-content/uploads/CMIS5p1_Third_Party_Spec. | |
10 | [6] | CS-01242017 CS optical connector and receptacle, see http://www.qsfp-dd.com/optical-connector/ |
11 | EIA-364-1000 TS-1000B Environmental Test Methodology for Assessing the Performance of Electrical | |
12 | Connectors and Sockets Used in Controlled Environment Applications, revision B 2009 | |
13 | EN6100-4-2 (IEC immunity standard on ESD), criterion B test specification | |
14 | Human Body Model per ANSI/ESDA/JEDEC JS-001 | |
15 | IEC 61754-7-1 (Fibre Optic Interconnecting Devices and Passive Components - Fibre Optic Connector | |
16 | Interfaces - Part 7-1: Type MPO Connector Family - One Fibre Row) | |
17 | IEC 61754-20 (Fibre Optic Interconnecting Devices and Passive Components - Fibre Optic Connector | |
18 | Interfaces - Part 20: Type LC Connector Family) | |
19 | IEEE Std 802.3TM-2018, annex 86A, 83E, and 120E | |
20 | IEEE Std 802.3cd (50 Gb/s, 100 Gb/s, and 200 Gb/s Ethernet), clause 136 and annex 136A | |
21 | IEEE Std 802.3ck (100 Gb/s, 200 Gb/s, and 400 Gb/s Electrical Interfaces) | |
22 | IEEE Std 1588 Precision Clock Synchronization Protocol PTP, 2019 | |
23 | InfiniBand Architecture Specification Volume 2 | |
24 | JEDEC JESD8C.01 Interface Standard for Nominal 3.0/3.3 V Supply Digital Integrated Circuit (LVCMOS) | |
25 | NEBS GR-63 Physical Protection Requirements for Network Telecommunications Equipment | |
26 | NXP UM10204, I2C-bus specification and user manual, Rev 6 – 4 April 2014. | |
27 | OIF CEI 4.0, CL-13 CEI-28G-VSR, CL-16 CEI-56G-VSR PAM4, and CEI-112G-VSR PAM4 specifications | |
28 | SFP-DD Management Interface, Rev. 2.0 | |
29 | SN-60092019 SN optical connector and receptacle, see http://www.qsfp-dd.com/optical-connector/ | |
30 | TIA-604-5 (FOCIS 5 Fiber Optic Connector Intermateability Standard- Type MPO) | |
31 | TIA-604-10 (FOCIS 10 Fiber Optic Connector Intermateability Standard- Type LC) | |
32 | USC-11383001 MDC optical plug and receptacle, see http://www.qsfp-dd.com/optical-connector/ | |
33 | 2.2 | SFF Specifications: |
34 | INF-8074i SFP (Small Formfactor Pluggable) Transceiver, Rev. 1.0 | |
35 | SFF-8071 SFP+ 1x 0.8 mm Card Edge Connector, Rev 1.1 | |
36 | SFF-8402 SFP+ 1x 28 Gb/s Pluggable Transceiver Solution (SFP28), Rev. 1.1 | |
37 | SFF-8419 SFP+ Power and Low Speed Interface, Rev. 1.3 | |
38 | SFF-8431 SFP+ 10 Gb/s and Low Speed Electrical Interface, Rev. 4.1 | |
39 | SFF-8432 SFP+ Module and Cage, Rev. 5.2a | |
40 | SFF-8433 SFP+ Ganged Cage Footprints and Bezel Openings, Rev. 0.7 | |
41 | SFF-8472 Diagnostic Monitoring Interface for Optical Transceivers, Rev. 12.4. | |
42 | ||
43 | Sources |
44 The SFP-DD MSA SFP-DD Hardware Specification for SFP DOUBLE DENSITY 2X PLUGGABLE
45 TRANSCEIVER can be obtained via the www.SFP-DD.com web site. 46
2
3 This Specification covers the following items:
4
Electrical interfaces including pad assignments for data, control, status and power supplies and host
PCB layout requirements.
7
Optical interfaces (including optical receptacles and mating fiber plugs for multimode and single-mode
duplex and parallel fiber applications). Breakout cable applications are also specified.
10
11 Mechanical specifications including dimensions and tolerances for the connector, cage and module
12 system. Includes details of the requirements for correct mating of the module and host sides of the
13 connector.
14
15 Thermal requirements
16
17 Management Interface Timing requirements
18
19 Electrostatic discharge (ESD) requirements by reference to industry standard limits and test methods.
20
21 This Specification does not cover the following items:
22
Electromagnetic interference (EMI) protection. EMI protection is the responsibility of the implementers
of the cages and modules.
25
Optical signaling specifications are not included in this document but are defined in the applicable
industry standards.
28
29 Management Registers. 30
SFP-DD/SFP-DD112 electrical signals, channel assignments, TWI, timing, and power requirements are
defined in Chapter 4. SFP112 electrical signals, TWI, timing, and power requirements are defined in Chapter
5. Optical port mapping and optical interfaces are defined in Chapter 6 to ensure that the pluggable modules
and cable assemblies are functionally interchangeable. Implementations compliant to dimensions, mounting
and insertion requirements defined in Chapter 7 for SFP-DD bezel, optical module, cable plug, cage and
connector system on a circuit board ensure that these products are mechanically interchangeable. Chapter 8
describes an improved SFP-DD form factor called SFP-DD112 with improved signal integrity for 100 Gb/s per
lane operation with an aggregate bandwidth of 200 Gb/s. Chapter 9 describes an improved SFP+ form factor
called SFP112 with improved signal integrity for 100 Gb/s per lane operation and with an aggregate bandwidth
of 100 Gb/s.
42
The SFP-DD form factor system consisting of a transceiver module, cage and connector provides two
channels for high speed signals that can support a two-lane trunked application or two independent single-lane
applications. The cage and socket can also accept SFP+/SFP28 and SFP112 modules in which case a single
lane channel is supported. To support classic SFP+ modules the electrical connector maintains the twenty
contacts row defined for SFP+ modules and adds another twenty contacts row to support a second channel. 49
In addition to contacts for the high speed data signals, the connector provides contacts for module and
channel control and status signals including a pair that form a Two-Wire Interface (TWI) [19] or communication
with the module's memory. Contacts for high speed data signals, channel level control and status indicator
signals and power supply sources for the SFP module are repeated in the row for the second channel.
Contacts for module level control and status signals in the SFP module remain in place and signals for new
module level functions were added to the second row. 3
New global features and associated signals, Low Power Mode, Reset and Interrupt were added and the
memory map for SFP-DD was expanded and reorganized for better alignment with CMIS functionality and
structure.
7
SFP-DD management specifications is based on SFP-DD MIS [21]. SFP-DD112 and SFP112 management
specifications are based on CMIS [5]. 10
Within the MIS and CMIS the SFP112 and SFP-DD/SFP-DD112 classic contacts, see Figure 3, are associated
as Channel 1 and the SFP-DD/SFP-DD112 additional contacts are associated as Channel 2. Within this
document the name, Channel 1, is synonymous with the name SFP “classic” Channel and Channel 2 is
synonymous with the name DD “additional” Channel. 15
Adding a second channel to SFP-DD/SFP-DD112 results in increased module power consumption.
Accommodations for the increased consumption include defining additional power classes, defining a Type 2
module and defining an enlarged heat sink seating area with surface flatness and roughness requirements. A
low power mode was added to provide the host a means for power management. 20
Another accommodation for a second channel was the inclusion of addition optical connectors including MPO
[10] and [23], duplex LC [11] and [24], see Figure 28.
23
24
25
26
2
This specification defines a connector, cage and module for single or double lane applications at up to 112
Gb/s (56 GBd) per lane. Intended applications include but are not limited to Ethernet and/or InfiniBand and/or
Fibre Channel. The SFP-DD/SFP-DD112/SFP112 interface can support pluggable modules or direct attach
cables based on multimode fiber, single mode fiber or copper wires. 7
8 An application reference Model, shown in Figure 1, shows the high-speed data interface between an ASIC and
9 the SFP-DD/SFP-DD112/SFP112 modules.
10
11 Figure 1: Application Reference Model
12
13
14 Figure 2: SFP-DD/SFP-DD112/SFP112 Cage, Connector and Module
1 4. Electrical Specification and Management Interface Timing for SFP-DD/SFP-DD112
2
This chapter contains signal definitions and requirements that are specific to the SFP-DD/SFP-DD112
modules. High-speed signal requirements including compliance points for electrical measurements are defined
in the applicable industry standard. 6
The SFP-DD/SFP-DD112 modules are hot-pluggable. Hot pluggable refers to plugging in or unplugging a
module while the host board is powered. 10
For EMI protection the signals from the host connector should be shut off when the SFP-DD/SFP-DD112
modules are not present. Standard board layout practices such as connections to Vcc and GND with vias, use
of short and equal-length differential signal lines are recommended. The SFP-DD/SFP-DD112 modules signal
ground contacts GND should be isolated from module case. An isolated SFP-DD/SFP-DD112 module case
from signal ground provides equipment designer flexibility regarding connections between external
electromagnetic interference shields and circuit ground “GND” of the module. 17
All electrical specifications shall be met over the entire specified range of power supplies given in section
The SFP-DD/SFP-DD112 module edge connector consists of a single paddle card with 20 pads on the top and
20 pads on the bottom of the paddle card for a total of 40 pads. The pads positions are defined to allow
insertion of either an SFP-DD/SFP-DD112 module or an SFP28 into the SFP-DD/SFP-DD112 receptacle. The
classic signal locations are deeper on the paddlecard, so that classic SFP module pads only connect to the
longer row of connector pins, leaving the short row of connector pins open circuited in an SFP application. 27
The pads are designed for a sequenced mating:
First mate – ground pads
Second mate – power pads
Third mate – signal pads. 32
Because the SFP-DD/SFP-DD112 module has 2 rows of pads, the additional SFP-DD pads will have an
intermittent connection with the classic SFP/SFP+ pins in the connector during the module insertion and
removal. The 'classic' SFP+/SFP28/SFP112 pads have a “B” label shown in Table 1 to designate them as the
first row of module pads to contact the SFP-DD/SFP-DD112 connectors. The additional SFP-DD/SFP-DD112
pads have a “A” label with first, second and third mate to the connector pins for both insertion and removal.
Each of the first, second and third mate connections of the classic SFP+/SFP28/SFP112 pads and the
respective additional SFP-DD/SFP-DD112 pads are simultaneous. 40
Figure 3 shows the signal symbols and pad numbering for the SFP-DD/SFP-DD112 modules edge connector.
The diagram shows the module PCB edge as a top and bottom view. There are 40 pads intended for high
speed signals, low speed signals, power and ground connections. 44
Table 1 provides more information about each of the 40 pads. Figure 40 and Figure 41 show pad
dimensions. The surface mount configuration is shown in Figure 46. 47
For EMI protection the signals from the host connector should be shut off when the SFP-DD/SFP-DD112
module is not present. Standard board layout practices such as connections to Vcc and GND with vias, use of
short and equal-length differential signal lines are recommended. The chassis ground (case common) of the
SFP-DD/SFP-DD112 module should be isolated from the module’s circuit ground, GND, to provide the
equipment designer flexibility regarding connections between external electromagnetic interference shields
and circuit ground, GND, of the module.
(Module Side)
Module Card Edge (Host Side)
Bottom side as viewed from top through the board
30 | GND | 10 | GND | ||||
29 | Speed2DD | 9 | Speed2 | ||||
28 | RxLOSDD | 8 | RxLOS | ||||
27 | Speed1DD | 7 | Speed1 | ||||
26 | ResetL | 6 | Mod_ABS | ||||
25 | LPMode | 5 | SCL | ||||
24 | ePPS/Clock | 4 | SDA | ||||
23 | TxDisableDD | 3 | TxDisable | ||||
22 | IntL/TXFaultDD | 2 | TxFault | ||||
21 | GND | 1 | GND |
Additional SFP-DD/ SFP-DD112 Pads
Classic SFP-DD/SFP-DD112/ SFP112 Pads
GND
RD1-
RD1+
GND
VccR1
VccT1
GND
TD1+
TD1-
GND
GND
RD-
RD+
GND
VccR
VccT
GND
TD+
TD-
GND
(Module Side)
Module Card Edge (Host Side)
Top side viewed from top of board
Additional SFP-DD/ SFP-DD112 Pads
3
Classic SFP-DD/SFP-DD112/ SFP112 Pads
4 Figure 3: SFP-DD/SFP-DD112 module pad layout
5
Because the SFP-DD/SFP-DD112 module has 2 rows of pads, the additional SFP-DD/SFP-DD112 pads will
have an intermittent connection with the classic SFP pins in the connector during the module insertion and
removal. SFP-DD/SFP-DD112 module pads are compatible with SFP+/SFP28 [30], [28] are designated as
“classic” pads in Table 1 to designate them as the second row of module pads to contact the SFP-DD/SFP-
DD112 connector. The additional module pads are designated as ”DD” pads in Table 1 to designate them as
the first row of module pads to contact the SFP-DD/SFP-DD112 connector. The additional SFP-DD pads have
first, second and third mate to the connector pins for both insertion and removal. Each of the first, second and
third mate connections of the classic SFP pads and the respective additional SFP-DD/SFP-DD112 pads are
simultaneous. For a reliable interconnect, a sufficient contact wipe of the connector pins sliding over the
module gold pads is required. In the past, long signal pads have been used to provide the mechanical wipe.
As operating speeds were relatively slow, the electrical stub was not an issue with signal integrity. 17
As operating speeds have increased, signal pad lengths have become shorter and shorter to reduce electrical
stubs, however this caused insufficient mechanical wipe. A solution is to add a small separation of the signal
pad such that there is a passive 'pre-wipe" pad and an active signal pad. In SFP-DD/SFP-DD112, there are
also long pre-wipe pads between the additional SFP-DD/SFP-DD112 pads and the classic SFP pads. This
provides connector pins a gold platted pad surface over which to slide between rows.
1 Table 1- Pad Function Definition
Pad | Logic | Symbol | Module Pad Descriptions | Plug Sequence4 | Notes |
0 | Case | Module case | 0 | ||
1 | GND | Ground | 1B | 1 | |
2 | LVTTL-O | TxFault | Module Fault Indication: optionally configured as classic SFP Module Fault Indication via TWI as described in the SFP-DD MIS | 3B | |
3 | LVTTL-I | TxDisable | Transmitter Disable for classic SFP channel | 3B | |
4 | LVCMOS-I/O | SDA | Management I/F data line | 3B | |
5 | LVCMOS-I/O | SCL | Management I/F clock | 3B | |
6 | LVTTL-O | Mod_ABS | Module Absent | 3B | |
7 | LVTTL-I | Speed1 | Rx Rate Select for classic SFP channel | 3B | |
8 | LVTTL-O | RxLOS | Rx Loss of Signal for classic SFP channel | 3B | |
9 | LVTTL-I | Speed2 | Tx Rate Select for classic SFP channel | 3B | |
10 | GND | Ground | 1B | 1 | |
11 | GND | Ground | 1B | 1 | |
12 | CML-O | RD0- | Inverse Received Data Out for classic SFP+ channel | 3B | |
13 | CML-O | RD0+ | Received Data Out for classic SFP+ channel | 3B | |
14 | GND | Ground | 1B | 1 | |
15 | VccR | Receiver Power | 2B | 2 | |
16 | VccT | Transmitter Power | 2B | 2 | |
17 | GND | Ground | 1B | 1 | |
18 | CML-I | TD0+ | Transmit Data In for classic SFP channel | 3B | |
19 | CML-I | TD0- | Inverse Transmit Data In for classic SFP channel | 3B | |
20 | GND | Ground | 1B | 1 | |
21 | GND | Ground | 1A | 1 | |
22 | LVTTL-O | IntL/ TxFaultDD | Interrupt: optionally configured as TxFaultDD via TWI as described in the SFP-DD MIS | 3A | |
23 | LVTTL-I | TxDisableDD | Transmitter Disable for DD channel | 3A | |
24 | LVTTL-I | ePPS/Clock | Precision Time Protocol (PTP) reference clock input | 3A | 3 |
25 | LVTTL-I | LPMode | Low Power Mode Control | 3A | |
26 | LVTTL-I | ResetL | Module Reset | 3A | |
27 | LVTTL-I | Speed1DD | Rx Rate Select for DD channel | 3A | |
28 | LVTTL-O | RxLOSDD | Loss of Signal for DD channel | 3A | |
29 | LVTTL-I | Speed2DD | Tx Rate Select for DD channel | 3A | |
30 | GND | Ground | 1A | 1 | |
31 | GND | Ground | 1A | 1 | |
32 | CML-O | RD1- | Inverse Received Data Out for DD channel | 3A | |
33 | CML-O | RD1+ | Received Data Out for DD channel | 3A | |
34 | GND | Ground | 1A | 1 | |
35 | VccR1 | Receiver Power for DD channel | 2A | 2 | |
36 | VccT1 | Transmitter Power for DD channel | 2A | 2 | |
37 | GND | Ground | 1A | 1 | |
38 | CML-I | TD1+ | Transmit Data In for DD channel | 3A | |
39 | CML-I | TD1- | Inverse Transmit Data In for DD channel | 3A | |
40 | GND | Ground | 1A | 1 | |
Notes:
|
The SFP-DD/SFP-DD112 connector allocates contacts for a set of low speed signals for control, status and
management by the host. These include dedicated hardware signals and TWI signals. The dedicated
hardware signals are the following:
TxDisable, TxDisableDD
RxLOS, RxLOSDD
Speed1, Speed2, Speed1DD, Speed2DD
TxFault
IntL/TXFaultDD
Mod_ABS
LPMode
ResetL
ePPS/Clock.
14
The TWI signals are the following:
SCL – clock
SDA – data.
18
TxDisable and TxDisableDD are module input signals. When TxDisable or TxDisableDD are asserted high or
left open, the appropriate SFP-DD/SFP-DD112 module transmitter output shall be turned off unless the
module is a passive cable assembly in which case this signal may be ignored. This signal shall be pulled up to
VccT in modules and cable assemblies. When TxDisable or TxDisableDD are asserted low or grounded the
module transmitter is operating normally. 25
RxLOS (Rx Loss of Signal) and RxLOSDD are open drain/collector outputs that require a resistive pull up to
Vcc_Host with a resistor in the range 4.7 k to 10 k, or with an active termination according to Table 5.
When high it indicates an optical signal level below that specified in the relevant standard. 30
LOS may be an optional function depending on the supported standard. If the LOS function is not
implemented, or is reported via the TWI only, the RxLOS contact shall be held low by the module and may be
connected to GND within the module. 34
RxLOS, RxLOSDD assert min and de-assert max are defined in the relevant standard. To avoid spurious
transition of LOS a minimum hysteresis of 0.5 dBo is recommended. 37
Speed1, Speed2, Speed1DD and Speed2DD are module inputs and are pulled low to GND with >30 k
resistors in the module. Speed1 optionally selects the optical receive signaling rate for channel 1. Speed1DD
optionally selects the optical receive signaling rate for channel 2. Speed2 optionally selects the optical transmit
signaling rate for the channel 1. Speed2DD optionally selects the optical transmit signaling rate for channel 2.
For logical definitions of hardware rate selects Speed1, Speed2, Speed1DD, Speed2DD, see 4.9. 44
Note: At 128 GFC the FC LSN no longer require to use Speed1, Speed2, Speed1DD and Speed2DD, it is
under consideration to reclaim these signals for programmable or other functions. 47
TxFault is a module wide (channel 1 and channel 2) output signal that when high, indicates that the module
has detected a fault condition and has entered the Fault state. TxFault signal can optionally be configured as
classic SFP Module (channel 1) Fault Indication via TWI as described in the SFP-DD MIS. If TxFault is not
implemented, the contact signal shall be held low by the module and may be connected to GND within the
module. The TxFault output is open drain/collector and shall be pulled up to the Vcc_Host on the host board
with a resistor in the range 4.7 k to 10 k, or with an active termination according to Table 5. 4
IntL/TxFaultDD is an open collector output that optionally can be configured for either the IntL signal or the
TxFaultDD signal. It shall be pulled to Vcc Host on the host board with a resistor in the range 4.7 k to 10 k,
or with an active termination according to Table 5. At power-up or after ResetL is released to high,
IntL/TxFaultDD is configured as IntL. If supported IntL/TxFaultDD can be optionally programmed as
TxFaultDD using TWI as defined in the SFP-DD Management Interface Specification. 11
When IntL/TxFaultDD is configured as IntL, a Low indicates a change in module state, a possible operational
fault or a status critical to the host system. The host identifies the source of the interrupt using TWI. The IntL
signal is de-asserted High after all set flags are read. 15
When IntL/TxFaultDD is configured as TxFaultDD, a High indicates that the module has detected a fault
condition in lane 1 and has entered the Fault state. (See SFP-DD MIS section 6.3.1.11) and if TxFaultDD is
not implemented, the contact signal shall be held low by the module. 19
Mod_ABS must be pulled up to Vcc Host on the host board and pulled low in the module. The Mod_ABS is
asserted “Low” when the module is inserted. The Mod_ABS is deasserted “High” when the module is
physically absent from the host connector due to the pull up resistor on the host board. 24
LPMode is an input signal from the host operating with active high logic. The LPMode signal must be pulled
up to Vcc in the SFP-DD/SFP-DD112 module. The LPMode signal allows the host to define whether the SFP-
DD/SFP-DD112 module will remain in Low Power Mode until software enables the transition to High Power
Mode as defined in the SFP-DD management specification. In Low Power Mode (LPMode de-asserted Low),
the module may immediately transition to High Power Mode after the management interface is initialized. 31
The ResetL signal shall be pulled to Vcc in the module. A low level on the ResetL signal for longer than the
minimum pulse length (t_Reset_init) (See Table 7) initiates a complete module reset, returning all user module
settings to their default state. 36
Host ePPS/Clock The ePPS/Clock input is a programable timing and clock input, that can support
unmodulated 1PPS (1 pulse per second), modulated (1PPS), and reference clock. The ePPS/clock is a
LVCMOS compatible signal with series termination (TBD) on the host board and a parallel termination of at
least 4.7 k in the module. To improve signal integrity for faster clocks (i.e., 156.25 MHz) the parallel
termination can be reduced to as low as 470 and optionally AC coupled. 43
For high-performance Precision Time Protocol (PTP) applications [15], the ePPS (Enhanced Pulse Per
Second) reference either with 1PPS modulated or unmodulated may be provided from the host to the module
for time synchronization, see Table 2 for advertise capability. This can be used for either offline delay
characterization or real-time delay compensation within the module. The ePPS is used to synchronize tightly
the Host Time-of-Day counter to the module internal Time-of-Day Counter.
1
2 The ePPS/Clock module input optionally can be configured to provide reference clock to the CDR/DSP, see
3 Table 2 for advertise capability.
4
5 Table 2- ePPS/Clock Advertising Capabilities
CMIS Byte Location (TBD) | Bit | Mode Supported |
xxxxxx-- | 00 | ePPS/Clock not supported |
xxxxxx-- | 01 | ePPS/Clock module supports either 1PPS mode, modulated 1PPS, or clock input for encoding see Table 3 |
xxxxxx-- | 10 | ePPS/Clock supported TOD (Time of Day) |
xxxxxx-- | 11 | ePPS/Clock - Reserved |
6
7 Table 3- ePPS or Clock Modes
CMIS Byte Location (TBD) | Bit | Mode Supported |
xxxx--xx | 00 | RF clock for frequency see table y |
xxxx--xx | 01 | 1PPS send as unmodulated pulse duration TBD |
xxxx--xx | 10 | 1PPS send as 75%/25% duty cycle on RF modulated clock, for clock frequency see Table 4 |
xxxx--xx | 11 | ePPS/Clock - Reserved |
8
9 Table 4- ePPS or Clock Frequency
CMIS Byte Location (TBD) | Bit | Mode Supported |
----xxxx | 0000 | 10 MHz |
----xxxx | 0001 | 12.5 MHz |
----xxxx | 0010 | 20 MHz |
----xxxx | 0011 | 24.576 MHz |
----xxxx | 0100 | 25 MHz |
----xxxx | 0101 | 156.25 MHz |
----xxxx | 0110-1101 | Reserved |
----xxxx | 1110-1111 | Custom |
10
Editor’s Note: registers to support optional ePPS/Clock will be added in future revisions of CMIS.
12
13
SCL is the TWI clock and SDA is the TWI data line. SCL and SDA are pulled up to Vcc_Host by resistors on
the host board. For TWI electrical specifications see 4.5.1 and for TWI protocol and timing specifications see
Figure 8.
Figure 4, Figure 5 and Figure 6 show examples of SFP-DD/SFP-DD112 host PCB schematics with
connections to CDR and control ICs. Note alternate electrical/optical interfaces are supported using optical
multiplexing (WDM) or electrical multiplexing.
Note: Filter capacitors values are informative and vary depending on applications, 0.1 uF capacitors should be placed in close proximity to power pins and may be duplicated for individual pins to provide additional high frequency filtering.
Note: VccT, VccT1 may be connected to VccR, VccR1 provided the applicable derating of the maximum current limit is used.
1
2
3 Figure 4: Example SFP-DD/SFP-DD112 Host Board Schematic for Optical Modules
1
Note: Filter capacitors values are informative and vary depending on applications, 0.1 uF capacitors should be placed in close proximity to power pins and may be duplicated for individual pins to provide additional high frequency filtering.
Note: VccT, VccT1 may be connected to VccR, VccR1 provided the applicable derating of the maximum current limit is used.
2
3
4
5 Figure 5: Example SFP-DD/SFP-DD112 Host Board Schematic for active copper cables
6
Note: Filter capacitors values are informative and vary depending on applications, 0.1 uF capacitors should be placed in close proximity to power pins and may be duplicated for individual pins to provide additional high frequency filtering.
Note: Recommended filtering is only valid for dedicated passive copper cable ports. For ports supporting both passive and active modules use recommended filtering from Figure 4 and Figure 5.
1
2 Figure 6: Example SFP-DD/SFP-DD112 Host Board Schematic for passive copper cables
Low Electrical requirements for low speed signals TxDisable, TxDisableDD, RxLOS, RxLOSDD, Speed1,
Speed2, Speed1DD, Speed2DD, TxFault, IntL/TXFaultDD, Mod_ABS, LPMode, ResetL and ePPS are based
on Low Voltage TTL (LVTTL) [17] operating at a module supply voltage Vcc of 3.3V +/- 5% and with a host
supply voltage Vcc_Host range of 2.38 to 3.46V. Vcc is used as a generic term for the supply voltages of
VccTx, VccRx, VccPullup or Vcc1. Host biasing requirements (e.g. pullup resistors) are defined in 4.3 and
illustrated in 4.4.
8
Electrical requirements for the TWI signals, SCL and SDA are based on Low Voltage CMOS (LVCMOS) [17]
operating at Vcc and compatible with [19]. Host biasing requirements (e.g. pullup resistors) are defined in 4.3
and illustrated in 4.4. Capacitance loading requirements are defined in Table 5 and tradeoffs are illustrated in
Figure 7.
13
The SFP-DD/SFP-DD112 low speed electrical specifications are given in Table 5. Implementations compliant
to this specification ensures compatibility between host bus initiator and TWI. Tradeoffs among Pull up
resistor values, bus capacitance and rise time are shown in Figure 7. 18
19
20 Table 5- Low Speed Control and Sense Signals
Parameters | Symbol | Min | Max | Unit | Condition |
SCL and SDA | VOL | 0 | 0.4 | V | open-drain or open-collector at 3 mA sink current: VDD > 2 V, IOL=3.0 mA for fast mode, 20 mA for fast mode plus |
SCL and SDA | VIL | -0.3 | Vcc*0.3 | V | |
VIH | Vcc*0.7 | Vcc+0.5 | V | ||
Capacitance for SCL and SDA I/O signal | Ci | 14 | pF | ||
Total bus capacitive load for SCL and SDA | Cb | 100 | pF | For 400 kHz clock rate use 3.0 k Pullup | |
200 | pF | For 400 kHz clock rate use 1.6 k pullup | |||
TxDisable(DD), ResetL, LPMode, Speed(n) and ePPS | VIL | -0.3 | 0.8 | V | For 0V<Vin<Vcc |
VIH | 2 | Vcc+0.3 | V | ||
LPMode, ResetL, TxDisable(DD), Speed(n) | |Iin| | 360 | A | ||
ePPS | |Iin| | 6.5 | mA | ||
Mod_ABS | VOL | 0 | 0.4 | V | IOL=2.0 mA, Mod_ABS can be implemented as a short-circuit to GND on the module |
RxLOS(DD), TxFault(DD) | VOL | -0.3 | 0.40 | V | 4.7 k Pullup resistor to Vcc_Host where Vcc_Host_min<Vcc_Host<Vcc_Host_max |
IOH | -50 | 37.5 | uA | ||
IntL | VOL | 0 | 0.4 | V | IOL = 2.0 mA |
21
1
2 Figure 7: SDA/SCL options for pull-up resistor, bus capacitance and rise/fall times
3
A management memory interface (See SFP-DD MIS [21], CMIS[5]), as already commonly used in other form
factors like QSFP-DD, QSFP, and SFP+ enables module functionality and flexibility beyond that supported by
the dedicated hardware signals. Read/Write functionality and protocols are defined in SFP-8419 [29]. 8
Some timing requirements are critical, especially for a multi-channel device, so the interface speed may
optionally be increased. Byte 00 on the Lower Page or Address 128 Page 00 is used to advertise the use of
the CMIS [5] for SFP-DD112 or MIS [21]. When a classic SFP+/SFP28 module is inserted into an SFP-
DD/SFP-DD112 port the host must use the SFP+ memory map, i.e., SFF-8472 [33]. This case is outside the
scope of this document. 14
In some applications, muxing or demuxing may occur in the module. In this specification, all references to
channel numbers are based on the electrical connector interface channels, unless otherwise indicated. In
cases where a status or control aspect is applicable only to channels after muxing or demuxing has occurred,
the status or control is intended to apply to all channels in the mux group, unless otherwise indicated. 19
Editor Note: It is anticipated that SFP-DD112 management is targeted to use future revisions of CMIS that
supports FC.
24
25
26
27
28
29
30
31
1
The SFP-DD/SFP-DD112 TWI and memory management timing illustrated in in Figure 8 and the parameters
given in Table 6. Implementations compliant to these specifications ensure compatibility between host bus
initiator and the TWI. 6
7
9
10
11
Parameters | Symbol | Min | Max | Min | Max | Unit | Conditions |
Fast Mode (400 KHz) | Fast Mode Plus (1 MHz) | ||||||
Clock Frequency | fSCL | 0 | 400 | 0 | 1000 | KHz | |
Clock Pulse Width Low | tLOW | 1.3 | 0.50 | µs | |||
Clock Pulse Width High | tHIGH | 0.6 | 0.26 | µs | |||
Time bus free before new transmission can start | tBUF | 20 | 20 | µs | Between STOP and START and between ACK and ReStart | ||
START Hold Time | tHD.STA | 0.6 | 0.26 | µs | The delay required between SDA becoming low and SCL starting to go low in a START | ||
START Setup Time | tSU.STA | 0.6 | 0.26 | µs | The delay required between SCL becoming high and SDA starting to go low in a START | ||
Data In Hold Time | tHD.DAT | 0 | 0 | µs | |||
Data In Setup Time | tSU.DAT | 0.1 | 0.1 | µs | |||
Input Rise Time | tR | 300 | 120 | ns | From (VIL, MAX=0.3*Vcc) to (VIH, MIN=0.7*Vcc) | ||
Input Fall Time | tF | 300 | 120 | ns | From (VIH, MIN=0.7*Vcc) to (VIL, MAX=0.3*Vcc) | ||
STOP Setup Time | tSU.STO | 0.6 | 0.26 | µs | |||
STOP Hold Time | tHD.STO | 0.6 | 0.26 | µs | |||
Serial Interface Clock Holdoff “Clock Stretching” | T_clock_hold | 500 | 500 | s | Maximum time the SFP-DD module may hold the SCL line low before continuing with a read or write operation | ||
Complete Single or Sequential Write | tWR | 80 | 80 | ms | Complete (up to) 8 Byte Write | ||
Accept a single or sequential write to volatile memory | tNACK | 10 | 10 | ms | Time to complete a Single or Sequential Write to volatile registers. | ||
Time to complete a memory bank/page | tBPC | 10 | 10 | ms | Time to complete a memory bank and/or page change. | ||
Endurance (Write Cycles) | 50K | 50k | cycles | Module Case Temperature = 70 oC |
2
3
4
Timing for SFP-DD/SFP-DD112 soft control and status functions are described in Table 7. Squelch and
Parameters | Symbol | Min | Max | Unit | Conditions |
MgmtInitDuration | Max MgmtInit Duration | 2000 | ms | Time from power on1 to hot plug or rising edge of reset until completion of the MgmtInit State | |
ResetL Assert Time | t_reset_init | 10 | µs | Minimum pulse time on the ResetL signal to initiate a module reset. | |
IntL Assert Time | ton_IntL | 200 | ms | Time from occurrence of condition triggering IntL until Vout:IntL=Vol | |
IntL Deassert Time | toff_IntL | 500 | µs | Time from clear on read2 operation of associated flag until Vout:IntL=Voh. This includes deassert times for Rx LOS, TXFault and other flag bits. | |
Rx LOS Assert Time | ton_los | 200 | ms | Time from Rx LOS condition present to Rx LOS bit set (value = 1b), LOS signal asserted and IntL asserted. | |
Rx LOS Assert Time (optional fast mode) | ton_losf | 1 | ms | Time from Rx LOS state to Rx LOS bit set (value = 1b) and LOS signal asserted IntL asserted. | |
TXFault Assert Time | ton_TxFault | 200 | ms | Time from TXFault state to TXFault bit set (value=1b) and IntL asserted. | |
Flag Assert Time | ton_flag | 200 | ms | Time from occurrence of condition triggering flag to associated flag bit set (value=1b) and IntL asserted. | |
Mask Assert Time | ton_mask | 100 | ms | Time from mask bit set (value=1b)3 until associated IntL assertion is inhibited | |
Mask Deassert Time | toff_mask | 100 | ms | Time from mask bit cleared (value=0b)3 until associated IntL operation resumes | |
DataPathDeinit Max Duration | DataPathDeinit_MaxDuration | See SFP-DD MIS Table 7-39 or CMIS memory P01h: B144 | |||
DataPathInit Max Duration | DataPathInit_MaxDuration | See SFP-DD MIS Table 7-39 or CMIS memory P01h: B144 | |||
Module Pwr Up Max Duration6 | ModulePwrUp_MaxDuration | See SFP-DD MIS Table 7-48 or CMIS memory P01h: B167 | |||
ModulePwrDn Max Duration | ModulePwrDn_MaxDuration | See SFP-DD MIS Table 7-48 or CMIS memory P01h: B167 | |||
Data Path TX Turn On Max Duration | DataPathTxTurnOn_MaxDuration | See SFP-DD MIS Table 7-48 or CMIS memory P01h: B168 | |||
Data Path TX Turn Off Max Duration | DataPathTxTurnOff_MaxDuration | See SFP-DD MIS Table 7-48 or CMIS memory P01h: B168 | |||
Notes: |
Measured from low to high SDA edge of the Stop condition of the read transaction.
Measured from low to high SDA edge of the Stop condition of the write transaction.
Rx LOS condition is defined at the optical input by the relevant standard.
5
6
7
8
9
10
11
Parameters | Symbol | Max | Unit | Conditions |
Rx Squelch Assert Time | ton_Rxsq | 100 | ms | Time from loss of Rx input signal until the squelched output condition is reached, see 4.8.1. |
Rx Squelch Deassert Time | toff_Rxsq | 10 | s | Time from resumption of Rx input signals until normal Rx output condition is reached, see 4.8.1. |
Tx Squelch Assert Time | ton_Txsq | 400 | ms | Time from loss of Tx input signal until the squelched output condition is reached, see 4.8.2. |
Tx Squelch Deassert Time | toff_Txsq | 10 | s | Time from resumption of Tx input signals until normal Tx output condition is reached, see 4.8.2. |
Tx Disable Assert Time | ton_txdis | 100 | ms | Time from the stop condition of the Tx Disable write sequence 1 until optical output falls below 10% of nominal, see note 2 and 3. |
Tx Disable Assert Time (optional fast mode) | ton_txdisf | 3 | ms | Time from Tx Disable bit set (value = 1b)1 until optical output falls below 10% of nominal, see note 2 and 3. |
Tx Disable Deassert Time | toff_txdis | 400 | ms | Time from Tx Disable bit cleared (value = 0b)1 until optical output rises above 90% of nominal, see note 2 and 3. |
Tx Disable Deassert Time (optional fast mode) | toff_txdisf | 10 | ms | Time from Tx Disable bit cleared (value = 0b)1 until optical output rises above 90% of nominal |
Rx Output Disable Assert Time | ton_rxdis | 100 | ms | Time from Rx Output Disable bit set (value = 1b)1 until Rx output falls below 10% of nominal |
Rx Output Disable Deassert Time | toff_rxdis | 100 | ms | Time from Rx Output Disable bit cleared (value = 0b)1 until Rx output rises above 90% of nominal |
Squelch Disable Assert Time | ton_sqdis | 100 | ms | This applies to Rx and Tx Squelch and is the time from bit set (value = 0b)1 until squelch functionality is disabled. |
Squelch Disable Deassert Time | toff_sqdis | 100 | ms | This applies to Rx and Tx Squelch and is the time from bit cleared (value = 0b)1 until squelch functionality is enabled. |
Notes: |
Measured from LOW to HIGH SDA signal transition of the STOP condition of the write transaction.
CMIS 4.0 and beyond the listed values are superseded by the advertised DataPathTxTurnOff_MaxDuration and DataPathTxTurnOn_MaxDuration times in CMIS or SFP-DD MIS P01h.168.
Listed values place a limit on the DataPathTxTurnOff_MaxDuration and DataPathTxTurnOn_MaxDuration times (CMIS or SFP-DD MIS P01h.168) that can be advertised by such modules (for CMIS 4.0 and beyond).
1
For detailed electrical specifications see the appropriate specification, e.g. IEEE Std 802.3-2018 [12]; IEEE Std
802.3cd [13], 802.3ck [14]; FC-PI-6 [1], FC-PI-7[2], FC-PI-8 [3]; OIF-CEI-4.0 [20], or the InfiniBand
specifications [16].
6
Partial or complete squelch specifications may be provided in the appropriate specification. Where squelch is
not fully defined by the appropriate specification, the recommendations given in clause 4.8.1 and 4.8.2 may be
used.
10
11 4.8.1 RD0+, RD0-, RD1+, RD1-
RD(n)+/- are SFP-DD/SFP-DD112 module receiver data outputs. Rx(n)+/- are AC-coupled 100 Ohm
differential lines that should be terminated with 100 Ohm differentially at the Host ASIC(SerDes). The AC
coupling is inside the SFP-DD/SFP-DD112 module and not required on the Host board. When properly
terminated, the differential voltage swing shall be less than or equal to 900 mVpp or the relevant standard,
whichever is less.
Output squelch for loss of optical input signal, hereafter RX Squelch, is required and shall function as follows.
In the event of the Rx input signal on any optical port becoming equal to or less than the level required to
assert LOS, then the receiver output(s) associated with that Rx port shall be squelched. A single Rx optical
port can be associated with more than one Rx output as shown in Table 24. In the squelched state output
impedance levels are maintained while the differential voltage amplitude shall be less than 50 mVpp. 4
In normal operation the default case has RX Squelch active. Rx Squelch can be deactivated using Rx
Squelch Disable through the 2-wire serial interface. 7
8 4.8.2 TD0+, TD0-, TD1+, TD1-
TD(n)+/- are SFP-DD/SFP-DD112 module transmitter data inputs. They are AC-coupled 100 Ohm differential
lines with 100 Ohm differential terminations inside the SFP-DD/SFP-DD112 optical module. The AC coupling
is implemented inside the SFP-DD/SFP-DD112 optical module and not required on the Host board. 12
Output squelch for loss of electrical signal, hereafter Tx Squelch, is an optional function. Where implemented
it shall function as follows. In the event of the differential, peak-to-peak electrical signal amplitude on any
electrical input lane becoming less than the TX Squelch Levels specified in Table 9 when terminated in to 100
differential, then the transmitter optical output associated with that electrical input lane shall be squelched
and the associated TxLOS flag set. If multiple electrical input lanes are associated with the same optical output
lane, the loss of any of the incoming electrical input lanes causes the optical output lane to be squelched. 19
20 Table 9- TX Squelch Levels
Data Rate | Levels | Unit |
OIF 28G-VSR/IEEE 802.3 CL83E | 70 | mV 1 |
OIF 56G-VSR/IEEE 802.3 CL120E | 70 | mV 1 |
OIF 112G-VSR/IEEE 802.3 CL120G | 50 | mV 1 |
1. Differential peak-peak. |
21
For applications, e.g., Ethernet, where the transmitter off condition is defined in terms of average power,
squelching by disabling the transmitter is recommended and for applications, e.g. InfiniBand, where the
transmitter off condition is defined in terms of OMA, squelching the transmitter by setting the OMA to a low
level is recommended. 26
In module operation, where TX Squelch is implemented, the default case has TX Squelch active. TX Squelch
can be deactivated using TX Squelch Disable through the 2-wire serial interface. TX Squelch is an optional
function. If TX squelch is implemented, the disable squelch must be provided. 30
The module provides 4 inputs Speed1, Speed2, Speed1DD and Speed2DD that can optionally be used for
rate selection as defined in Table 10. Speed1, Speed1DD specifies the highest rate advertised by the
channels. Speed2, Speed2DD specifies a lower speed for each channel. Speed1 controls the receive path
signaling rate capability for the channel 1 receive path. Speed1DD controls the receive path signaling rate
capability for the channel 2 receive path. Similarly, Speed2 controls the transmit path signaling rate capability
for channel 1 and Speed2DD controls the transmit path signaling rate capability for the channel 2 transmit
path.
39
The rate select functionality can also be controlled by software as defined by the SFP-DD or SFP-DD112
management specification.
42
For 64 GFC FC-PI-8 [3] operation link speed is determined with FC Link Speed Negotiation (LSN) exchange
while the link is operating at 32 GFC. 45
46 Table 10- Rate Select Hardware Control
Parameters | State | Conditions |
Speed1, Speed1DD | Low | RX signaling rate of 14.025 GBd (16 GFC)1 |
High | RX signaling rate of 28.05 GBd (32 GFC) | |
Speed2, Speed2DD | Low | TX signaling rate of 14.025 GBd (16 GFC) 1 |
High | TX signaling rate of 28.05 GBd (32 GFC) | |
Note 1: For SFP-DD112 operating at 128 GFC there is no requirement for 16 GFC or rate select. |
1
SFP-DD/SFP-DD112 has four designated power supply pins VccT, VccT1 VccR, VccR1 in the connector.
Power is applied concurrently to VccT, VccR in the row for channel 1 and concurrently to VccT1, VccR1 in the
row for channel 2.
6
A host board together with the SFP-DD/SFP-DD112 module(s) forms an integrated power system. The host
supplies stable power to the module. The module limits electrical noise coupled back into the host system and
limits inrush charge/current during hot plug insertion. 10
All power supply requirements in Table 12 shall be met at the maximum power supply current. No power
sequencing of the power supply is required of the host system since the module sequences the contacts in the
order of ground, supply and signals during insertion. 14
15 4.10.1 Power Classes and Maximum Power Consumption
16 SFP-DD/SFP-DD112 power classes are defined in Table 11. Since different classes of modules exist with
17 pre-defined maximum power consumption limits, it is necessary to avoid exceeding the system power supply
18 limits and cooling capacity when a module is inserted into a system designed to only accommodate lower
19 power modules. It is recommended that the host identify the power class of the module before allowing the
20 module to go into high power mode. 21
22 Power levels associated with classifications of modules are shown in Table 11. 23
24 Table 11- Power Classification
Power Class | Max Power (W) |
1 | 1.0 |
2 | 1.5 |
3 | 2.0 |
4 | 3.5 |
5 | 5.0 |
6 | reserved |
7 | reserved |
8 | See management register for maximum power consumption. |
25
In general, the higher power classification levels are associated with higher data rates and longer reaches.
The system designer is responsible for ensuring that the maximum case temperature does not exceed the
case temperature requirements. 29
30
The host board should use the power supply filtering equivalent to that shown in Figure 9. 3
Note: Filter capacitors values are informative and vary depending on applications.
VccT, VccT1 may be connected to VccR, VccR1 provided the applicable derating of the maximum current limit is used.
4
5 Figure 9: Recommended Host Board Power Supply Filtering
6
7
Any voltage drop across a filter network on the host is counted against the host DC set point accuracy
specification. Inductors with DC Resistance of less than 0.1 Ohm should be used in order to maintain the
required voltage at the Host Card Edge Connector. It is recommended that the 22 F capacitors each have an
equivalent series resistance of 0.22 ohms. 12
The specifications for the power supply are shown in Table 12. The limits in Table 12 apply to the combined
current that flows through all inductors in the power supply filter (represents ICC host in Figure 24). An
example test method for measuring inrush current can is Keysight Technologies application brief 5991-
2778EN.pdf.
17
In order to avoid exceeding the host system power capacity, upon hot-plug, power cycle or reset, all SFP-
DD/SFP-DD112 modules shall power up in Low Power Mode if LPMode is asserted. If LPMode is not asserted
the module will proceed to High Power Mode without host intervention. Figure 10 shows waveforms for
maximum instantaneous, sustained and steady state currents for Low Power and High Power modes.
Specification values for maximum instantaneous, sustained and steady state currents at each power class are
given in Table 12.
25
1
2 Table 12- Power Supply specifications, Instantaneous, sustained and steady state current limits
Parameter | Symbol | Min | Nom | Max | Unit |
Power supply voltages VccTx, VccTx1, VccRx, VccRx1 including ripple, droop and noise below 100 kHz1 | 3.135 | 3.3 | 3.465 | V | |
Module inrush - instantaneous peak duration | T_ip | 50 | µs | ||
Module inrush - initialization time | T_init | 500 | ms | ||
Low Power Mode2 | |||||
Power Consumption | P_lp | .5 | W | ||
Instantaneous peak current at hot plug | Icc_ip_lp | - | 200 | mA | |
Sustained peak current at hot plug | Icc_sp_l p | - | 165 | mA | |
Steady state current | Icc_lp | See Note 4 | mA | ||
High Power Class 1 module | |||||
Power Consumption | P_1 | 1.0 | W | ||
Instantaneous peak current | Icc_ip_1 | - | 400 | mA | |
Sustained peak current | Icc_sp_1 | - | 330 | mA | |
Steady state current | Icc_1 | See Note 4 | mA | ||
High Power Class 2 module | |||||
Power Consumption | P_2 | 1.5 | W | ||
Instantaneous peak current | Icc_ip_2 | - | 600 | mA | |
Sustained peak current | Icc_sp_2 | - | 495 | mA | |
Steady state current | Icc_2 | See Note 4 | mA | ||
High Power Class 3 module | |||||
Power Consumption | P_3 | 2.0 | W | ||
Instantaneous peak current | Icc_ip_3 | - | 800 | mA | |
Sustained peak current | Icc_sp_3 | - | 660 | mA | |
Steady state current | Icc_3 | See Note 4 | mA | ||
High Power Class 4 module | |||||
Power Consumption | P_4 | 3.5 | W | ||
Instantaneous peak current | Icc_ip_4 | - | 1400 | mA | |
Sustained peak current | Icc_sp_4 | - | 1155 | mA | |
Steady state current | Icc_4 | See Note 4 | mA | ||
High Power Class 5 module | |||||
Power Consumption | P_5 | 5.0 | W | ||
Instantaneous peak current | Icc_ip_5 | - | 2000 | mA | |
Sustained peak current | Icc_sp_5 | - | 1650 | mA | |
Steady state current | Icc_5 | See Note 4 | mA | ||
High Power Class 8 module | |||||
Power Consumption | P_83 | W | |||
Instantaneous peak current | Icc_ip_8 | - | - | P_8/2.5 | A |
Sustained peak current | Icc_sp_8 | - | - | P_8/3.03 | A |
Steady state current | Icc_8 | - | - | 4 | A |
Notes:
|
3
LOW POWER MODE
Transmitters OFF
HIGH POWER MODE
Transmitters OFF
HIGH POWER MODE
Transmitters ON
T_init
T_ip
T_ip
T_init
T_ip
T_init
TxOff state inrush current
graph is informative. It cannot exceed the High power mode Transmitter On currents. No currents or power consumption are specified
Icc Host (ma)
t (ms)
Hot Plug
Instant
Host Enables High Power Mode
Host Enables Transmitters
3
The host noise output on VccT/VccT1, and VccR/VccR1 supplies are defined with resistive loads that draws
the maximum rated power supported by the host power class, see Figure 11. The resistive loads are
connected in place of the module between VccT/VccT1 and VccR/VccR1 and the Vee. When the noise is
measured on VccT/VccT1, VccR/VccR1 is left open circuit, and vice versa. Host power supply limits are given
in Table 11. The noise power spectrum is measured for each of the 2 rails then integrated from 40 Hz to 10
MHz and converted to a voltage, eN_Host, with limit specified in Table 13. 11
12
13 Figure 11: Host Noise Output Measurement
14
15
16
The SFP-DD/SFP-DD112 modules, when plugged into a reference module compliance board shall generate
noise less than the value in Table 13. The module must pass module power supply noise output test in all
operating modes. This test ensures the module will not couple excessive noise from in- side the module back
onto the host board. A power meter technique, or a spectrum analyzer technique with integration of the
spectrum, may be used. 7
The RMS module noise voltage output is defined in the frequency band from 40 Hz to 10 MHz. Module noise
output shall be measured with an appropriate probing technique at point X, Figure 12 and must meet limits
given in Table 13. The module must pass module power supply noise output test in all operating modes. This
test ensures the module will not couple excessive noise from inside the module back onto the host board.
12
13 Figure 12: Module Noise Output Measurement
14
The SFP-DD/SFP-DD112 modules shall meet all requirements and operate within the design specifications in
the presence of a reference noise waveform described in Table 13 superimposed on the DC voltage. The
reference noise waveform consists of a sinusoidal 40 Hz to 10 MHz noise generated by Osc1 and added to
Vcc PSU, see Figure 13. This emulates the worst-case noise that the module must tolerate and operate
within the design specifications. The reference noise is generated by Osc1 and amplified by the Power
OpAmp then added to Vcc PSU through a Bias-T, see Figure 10. Example of suitable Power OpAmp are
Analog Devices ADA4870 and TI THS3491. With power supply filter components removed, point X measures
the noise voltage applied to the module. To facilitate power supply tolerance testing at frequencies < ~100
kHz due to Power OpAmp interaction with PSU and low frequency response of the Bias-T, it is recommended
to use noise source Osc2 modulating PSU sense line to generate sinusoidal noise directly on the PSU output,
see Figure 14. Osc2 amplitude level is adjusted while observing point X amplitude level as defined in Table 13
for module in low power and high-power modes. To modulate the PSU sense lines, the PSU must have high
speed sense tracking. An example of PSU with high-speed sense tracking are TI TPSM5D1806 and Keysight
N6700 with N6781/N6782 plugins. 30
For modules without or limited input stage power filtering one may measure the applied noise to the module by
measuring point X directly while the module is active and either in low or high-power modes. But to take credit
for any input stage power filtering in the module, the DUT module is replaced with a resistive load drawing
equivalent current of a module configured in low power mode, the DUT module is then replaced with a
resistive load drawing equivalent current of a module configured in high power mode. Osc1 and Osc2 are
adjusted to produce maximum PSNR level as defined in Table 13 at point X with resistive loads drawing the
same power as the module in low and high-power modes. The resistive loads are then replaced with the DUT
module with the same Osc1/Osc2 amplitude settings that produced the max PSNR with the resistive loads. 39
Notes: An appropriate probing technique is required for noise measurement at point X. For modules with
limited or no decoupling directly connected to host PSU, the PSNR can be directly measured at point X with
module plugged into the host for module operating in low power and high power modes. Osc1 or Osc2 are
adjusted to provide maximum PSNR at point X for a given module in low power and full power modes. 5
6 Table 13- Power Supply Output Noise and Tolerance Specifications
Parameter | Symbol | Min | Nom | Max | Unit |
Host RMS noise output 40 Hz-10 MHz (eN_Host) 1, 3 | 25 | mV | |||
Module RMS noise output 40 Hz - 10 MHz 2, 3 | 30 | mV | |||
Module sinusoidal power supply noise tolerance 40 Hz - 10 MHz (p-p) 2, 3 | PSNRmod | 66 | mV | ||
Notes: 100, 200, 300, 400, 500, 600, 700, 800, 900 Hz 1, 2, 3, 4, 5, 6, 7, 8, 9 kHz 10, 20, 30, 40, 50, 60, 70, 80, 90 kHz 100, 200, 300, 400, 500, 600, 700, 800, 900 kHz 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 MHz. |
Host must be tested for all supported power classes
Module must be test at low and high power modes
Recommended test frequencies: 40, 50, 60, 70, 80, 90 Hz
7
8
1
2
4
5
6 Figure 14: Module Low Frequency Noise Tolerance
7
Where ESD performance is not otherwise specified, e.g. in the InfiniBand specification, the SFP-DD/SFP-
DD112 module shall meet ESD requirements given in EN61000-4-2 [8], criterion B test specification when
installed in a properly grounded cage and chassis. The units are subjected to 15 kV air discharges during
operation and 8 kV direct contact discharges to the case. All the SFP-DD/SFP-DD112 module and host pins
including high speed signal pins shall withstand 1000 V electrostatic discharge based on Human Body Model
15
2
This chapter contains signal definitions and requirements that are specific to the SFP112 modules. SFP112
modules are backward compatible with SFP+ [30] and SFP28 [28] modules. High-speed signal requirements
including compliance points for electrical measurements are defined in the applicable industry standard. 6
The SFP112 modules are hot-pluggable. Hot pluggable refers to plugging in or unplugging a module while the
host board is powered. 10
For EMI protection the signals from the host connector should be shut off when the SFP112 module is not
present. Standard board layout practices such as connections to Vcc and GND with vias, use of short and
equal-length differential signal lines are recommended. The SFP112 module signal ground contacts GND
should be isolated from module case. An isolated SFP112 module case from signal ground provides
equipment designer flexibility regarding connections between external electromagnetic interference shields
and circuit ground (GND) of the module. 17
The SFP112 module edge connector consists of a single paddle card with 10 pads on the top and 10 pads on
the bottom of the paddle card for a total of 20 pads compatible with SFF-8071 [27], but improved for 112G
operation.
24
The pads are designed for a sequenced mating:
First mate – ground pads
Second mate – power pads
Third mate – signal pads. 29
Figure 15 shows the signal symbols and pad numbering for the SFP112 modules edge connector. The
diagram shows the module PCB edge as a top and bottom view. There are 20 pads intended for high speed
signals, low speed signals, power and ground connections. 33
Table 14 provides more information about each of the 20 pads. Figure 58 shows SFP112 improved pad
dimensions. The surface mount SFP112 SMT connector configuration is shown in Figure 59. 36
37
10 GND
9 Speed2
8 RxLOS
7 Speed1
6 Mod_ABS
5 SCL
4 SDA
3 TxDisable
2 TxFault/IntL
1 GND
(Module Side)
Module Card Edge (Host Side)
Bottom side as viewed from top through the board
SFP112 Pads
GND
RD-
RD+
GND
VccR
VccT
GND
TD+
TD-
GND
(Module Side)
Module Card Edge (Host Side)
Top side viewed from top of board
SFP112 Pads
1
3
SFP112 module pads are compatible with SFP+/SFP28 [30], [28], for list of pad functions see Table 1. For a
reliable interconnect, a sufficient contact wipe of the connector pins sliding over the module gold pads is
required. In the past, long signal pads have been used to provide the mechanical wipe. As operating speeds
were relatively slow, the electrical stub was not an issue with signal integrity. 8
As operating speeds have increased, signal pad lengths have become shorter and shorter to reduce electrical
stubs, however this caused insufficient mechanical wipe. A solution is to add a small passive 'pre-wipe" pad
prior to active signal pad.
Pad | Logic | Symbol | Module Pad Descriptions | Plug Sequence3 | Notes |
0 | Case | Module case | 0 | ||
1 | GND | Ground | 1 | 1 | |
2 | LVTTL-O | TXFault/IntL | Module Fault Indication optionally can be configured IntL via TWI as described in the CMIS | 3 | 4 |
3 | LVTTL-I | TxDisable | Transmitter Disable | 3 | |
4 | LVCMOS-I/O | SDA | Management I/F data line | 3 | |
5 | LVCMOS-I/O | SCL | Management I/F clock | 3 | |
6 | LVTTL-O | Mod_ABS | Module Absent | 3 | |
7 | LVTTL-I | Speed1 | Rx Rate Select | 3 | |
8 | LVTTL-O | RxLOS | Rx Loss of Signal | 3 | |
9 | LVTTL-I | Speed2 | Tx Rate Select | 3 | |
10 | GND | Ground | 1 | 1 | |
11 | GND | Ground | 1 | 1 | |
12 | CML-O | RD0- | Inverse Received Data Out | 3 | |
13 | CML-O | RD0+ | Received Data Out | 3 | |
14 | GND | Ground | 1 | 1 | |
15 | VccR | Receiver Power | 2 | 2 | |
16 | VccT | Transmitter Power | 2 | 2 | |
17 | GND | Ground | 1 | 1 | |
18 | CML-I | TD0+ | Transmit Data | 3 | |
19 | CML-I | TD0- | Inverse Transmit Data | 3 | |
20 | GND | Ground | 1 | 1 | |
Notes: |
SFP112 uses common ground (GND) for all signals and supply (power). All are common within the SFP112 module and all module voltages are referenced to this potential unless otherwise noted. Connect these directly to the host board signal- common ground plane.
VccR, VccT shall be applied concurrently. Requirements defined for the host side of the Host Card Edge Connector are listed in Table 21Table 12. VccR and VccT may be internally connected within the module in any combination. The connector Vcc pins are each rated for a maximum current of 1000 mA.
Plug Sequence specifies the mating sequence of the host connector and module. The sequence is 0, 1, 2, 3.
Support to configure TXFault/IntL will be provided in future CMIS revisions 5.1+.
2
The SFP112 connector allocates contacts for a set of low speed signals for control, status and management
by the host. These include dedicated hardware signals and TWI signals. The dedicated hardware signals are
the following:
TxDisable
RxLOS
Speed1, Speed2
TxFault/IntL
Mod_ABS.
12
The TWI signals are the following:
SCL – clock
SDA – data.
16
TxDisable is a module input signals. When TxDisable is asserted high or left open, the appropriate SFP112
module transmitter output shall be turned off unless the module is a passive cable assembly in which case this
signal may be ignored. This signal shall be pulled up to VccT in modules and cable assemblies. When
TxDisable is asserted low or grounded the module transmitter is operating normally. 3
RxLOS (Rx Loss of Signal) is an open drain/collector outputs that require a resistive pull up to Vcc_Host with a
resistor in the range 4.7 k to 10 k, or with an active termination according to Table 15. When high it
indicates an optical signal level below that specified in the relevant standard. 8
LOS may be an optional function depending on the supported standard. If the LOS function is not
implemented, or is reported via the TWI only, the RxLOS contact shall be held low by the module and may be
connected to GND within the module. 12
RxLOS assert min and de-assert max are defined in the relevant standard. To avoid spurious transition of LOS
a minimum hysteresis of 0.5 dBo is recommended. 15
Speed1, Speed2 are module inputs and are pulled low to GND with >30 k resistors in the module. Speed1
optionally selects the optical receive signaling rate. Speed2 optionally selects the optical transmit signaling
rate for.
20
TxFault/IntL is an open collector output and shall be pulled to Vcc Host on the host board with a resistor in the
range 4.7 k to 10 k, or with an active termination according Table 15. 24
TxFualt/IntL default setting indicates TxFault, but optionally can be configured as IntL through CMIS. TxFault
indicate the module has detected a fault condition and has entered the Fault state. When TxFault/IntL is
configured as IntL, a Low indicates a change in module state, a possible operational fault or a status critical to
the host system. The host identifies the source of the interrupt using TWI. The IntL signal is de-asserted High
after all set flags are read. 30
At power-up or after ResetL is released to high, TxFault/IntL is configured as TxFault. If supported
TxFault/IntL can be optionally programmed as IntL using TWI as defined in the CMIS. 33
Mod_ABS must be pulled up to Vcc Host on the host board and pulled low in the module. The Mod_ABS is
asserted “Low” when the module is inserted. The Mod_ABS is deasserted “High” when the module is
physically absent from the host connector due to the pull up resistor on the host board. 38
SCL is the TWI clock and SDA is the TWI data line. SCL and SDA are pulled up to Vcc_Host by resistors on
the host board. For TWI electrical specifications see 5.5.1 and for TWI protocol and timing specifications see
Figure 20.
43
44 5.4 Example SFP112 Host Board Schematics
45 Figure 16, Figure 17, and Figure 18 show examples of SFP112 host PCB schematics with connections to CDR
46 and control ICs. Note alternate electrical/optical interfaces are supported using optical multiplexing (WDM) or
47 electrical multiplexing.
Note: Filter capacitors values are informative and vary depending on applications, 0.1 uF capacitors should be placed in close proximity to power pins and may be duplicated for individual pins to provide additional high frequency filtering.
Note: VccT may be connected to VccR provided the applicable derating of the maximum current limit is used.
1
2
1
Note: Filter capacitors values are informative and application dependent, 0.1 mF capacitors should be placed in close proximity to power pads and may be duplicated for individual pads to provide additional high frequency filtering.
Note: VccT may be connected to VccR within the module provided the applicable derating of the maximum current limit is used.
2
3
4
5
6
7
8
9
10
11 Figure 17: Example of SFP112 Host Board Schematic for active copper cables
12
Note: Filter capacitors values are informative and vary depending on applications, 0.1 uF capacitors should be placed in close proximity to power pins and may be duplicated for individual pins to provide additional high frequency filtering.
Note: Recommended filtering is only valid for dedicated passive copper cable ports. For ports supporting both passive and active modules use recommended filtering from Figure 16 or Figure 17.
1
2
3
4
5
6
7
Low Electrical requirements for low speed signals TxDisable, RxLOS, Speed1, Speed2, IntL/TxFault,
IntL/TXFaultDD, and Mod_ABS are based on Low Voltage TTL (LVTTL) [17] operating at a module supply
voltage Vcc of 3.3V +/- 5% and with a host supply voltage Vcc_Host range of 2.38 to 3.46V. Vcc is used as a
generic term for the supply voltages of VccTx, VccRx, or VccPullup. Host biasing requirements (e.g. pullup
1
Electrical requirements for the TWI signals, SCL and SDA are based on Low Voltage CMOS (LVCMOS) [17]
operating at Vcc and compatible with [19]. Host biasing requirements (e.g. pullup resistors) are defined in 5.3
and illustrated in 5.4. Capacitance loading requirements are defined in Table 15 and tradeoffs are illustrated in
Figure 19.
6
The SFP112 low speed electrical specifications are given in Table 15. Implementations compliant to this
specification ensures compatibility between host bus initiator and TWI. Tradeoffs among Pull up resistor
values, bus capacitance and rise time are shown in Figure 19. 11
12
13 Table 15- Low Speed Control and Sense Signals
Parameters | Symbol | Min | Max | Unit | Condition |
SCL and SDA | VOL | 0 | 0.4 | V | open-drain or open-collector at 3 mA sink current: VDD > 2 V, IOL=3.0 mA for fast mode, 20 mA for fast mode plus |
SCL and SDA | VIL | -0.3 | Vcc*0.3 | V | |
VIH | Vcc*0.7 | Vcc+0.5 | V | ||
Capacitance for SCL and SDA I/O signal | Ci | 14 | pF | ||
Total bus capacitive load for SCL and SDA | Cb | 100 | pF | For 400 kHz clock rate use 3.0 k Pullup resistor, max. For 1000 kHz clock rate refer | |
200 | pF | For 400 kHz clock rate use 1.6 k pullup resistor max. For 1000 kHz clock rate refer to | |||
TxDisable, Speed1, Speed2 | VIL | -0.3 | 0.8 | V | For 0V<Vin<Vcc |
VIH | 2 | Vcc+0.3 | V | ||
TxDisable, Speed1, Speed2 | |Iin| | 360 | A | ||
Mod_ABS | VOL | 0 | 0.4 | V | IOL=2.0 mA, Mod_ABS can be implemented as a short-circuit to GND on the module |
RxLOS, TxFault/IntL | VOL | -0.3 | 0.40 | V | 4.7 k Pullup resistor to Vcc_Host where Vcc_Host_min<Vcc_Host<Vcc_Host_max |
IOH | -50 | 37.5 | uA |
14
1
2 Figure 19: SDA/SCL options for pull-up resistor, bus capacitance and rise/fall times
3
A management memory interface (such as CMIS [5], SFP+ MIS [33]) and commonly used in other form factors
like QSFP-DD, SFP-DD, or SFP+ enables module functionality and flexibility beyond that supported by the
dedicated hardware signals. Read/Write functionality and protocols are defined in SFP-8419 [29]. 8
Some timing requirements are critical, especially for a multi-channel device, so the interface speed may
optionally be increased. Byte 00 on the Lower Page or Address 128 Page 00 is used to advertise SFP112
using CMIS memory map [5] vs SFP+/SFP28 using SFF-8472 memory map [33]. When a classic
SFP+/SFP28 module is inserted into an SFP112 port the host must use the SFP+/SFP28 memory map (i.e.,
SFF-8472 [33]).
14
In some applications, muxing or demuxing may occur in the module. In this specification, all references to
channel numbers are based on the electrical connector interface channels, unless otherwise indicated. In
cases where a status or control aspect is applicable only to channels after muxing or demuxing has occurred,
the status or control is intended to apply to all channels in the mux group, unless otherwise indicated. 19
22
23
24
25
26
27
28
29
30
31
1
The SFP112 TWI and memory management timing illustrated in in Figure 20 and the parameters given in
Table 16. Implementations compliant to these specifications ensure compatibility between host bus initiator
and the TWI. 6
7
9
10
11
Parameters | Symbol | Min | Max | Min | Max | Unit | Conditions |
Fast Mode (400 KHz) | Fast Mode Plus (1 MHz) | ||||||
Clock Frequency | fSCL | 0 | 400 | 0 | 1000 | KHz | |
Clock Pulse Width Low | tLOW | 1.3 | 0.50 | µs | |||
Clock Pulse Width High | tHIGH | 0.6 | 0.26 | µs | |||
Time bus free before new transmission can start | tBUF | 20 | 20 | µs | Between STOP and START and between ACK and ReStart | ||
START Hold Time | tHD.STA | 0.6 | 0.26 | µs | The delay required between SDA becoming low and SCL starting to go low in a START | ||
START Setup Time | tSU.STA | 0.6 | 0.26 | µs | The delay required between SCL becoming high and SDA starting to go low in a START | ||
Data In Hold Time | tHD.DAT | 0 | 0 | µs | |||
Data In Setup Time | tSU.DAT | 0.1 | 0.1 | µs | |||
Input Rise Time | tR | 300 | 120 | ns | From (VIL, MAX=0.3*Vcc) to (VIH, MIN=0.7*Vcc) | ||
Input Fall Time | tF | 300 | 120 | ns | From (VIH, MIN=0.7*Vcc) to (VIL, MAX=0.3*Vcc) | ||
STOP Setup Time | tSU.STO | 0.6 | 0.26 | µs | |||
STOP Hold Time | tHD.STO | 0.6 | 0.26 | µs | |||
Serial Interface Clock Holdoff “Clock Stretching” | T_clock_hold | 500 | 500 | s | Maximum time the SFP112 module may hold the SCL line low before continuing with a read or write operation | ||
Complete Single or Sequential Write | tWR | 80 | 80 | ms | Complete (up to) 8 Byte Write | ||
Accept a single or sequential write to volatile memory | tNACK | 10 | 10 | ms | Time to complete a Single or Sequential Write to volatile registers. | ||
Time to complete a memory bank/page | tBPC | 10 | 10 | ms | Time to complete a memory bank and/or page change. | ||
Endurance (Write Cycles) | 50K | 50k | cycles | Module Case Temperature = 70 oC |
2
The timing attribute tBUF is the bus free time between sequential TWI transactions, see Figure 21. It’s
measured from the low to high SDA edge of the Stop condition of the Write transaction to the high to low SDA
edge of the Start condition for the next transaction. 7
The timing attribute tWR is the time required for a module to complete its internally timed write cycle after a
single or sequential write to non-volatile memory before the next basic management operation can be
accepted, see Figure 21. It’s measured from the low to high SDA edge of the Stop condition of the Write
transaction to the high to low SDA edge of the Start condition for the next transaction.
1
23
The timing attribute tNACK is the time required for a module to complete its internally timed write cycle after a
single or sequential write to volatile memory before the next basic management operation can be accepted,
see Figure 22. It’s measured from the low to high SDA edge of the Stop condition of the Write transaction to
the high to low SDA edge of the Start condition for the next transaction.
9
10 Figure 22: Bus timing tNACK
11
The timing attribute tBPC is the time required for a module to complete the change for the requested Bank
and/or Page selection, see Figure 23. It’s measured from the low to high SDA edge of the Stop condition of
the Write transaction to the high to low SDA edge of the Start condition for the next transaction.
5
6 Figure 23: Bus timing tBPC
7
8
9
10
11
12
Timing for SFP112 soft control and status functions are described in Table 17. Squelch and disable timings
4
Parameters | Symbol | Min | Max | Unit | Conditions |
MgmtInitDuration | Max MgmtInit Duration | 2000 | ms | Time from power on1 to hot plug or rising edge of reset until completion of the MgmtInit State | |
ResetL Assert Time | t_reset_init | 10 | µs | Minimum pulse time on the ResetL signal to initiate a module reset. | |
IntL Assert Time | ton_IntL | 200 | ms | Time from occurrence of condition triggering IntL until Vout:IntL=Vol | |
IntL Deassert Time | toff_IntL | 500 | µs | Time from clear on read2 operation of associated flag until Vout:IntL=Voh. This includes deassert times for Rx LOS, TXFault and other flag bits. | |
Rx LOS Assert Time | ton_los | 200 | ms | Time from Rx LOS condition present to Rx LOS bit set (value = 1b), LOS signal asserted and IntL asserted 4. | |
Rx LOS Assert Time (optional fast mode) | ton_losf | 1 | ms | Time from Rx LOS state to Rx LOS bit set (value = 1b) and LOS signal asserted IntL asserted 4. | |
TXFault Assert Time | ton_TxFault | 200 | ms | Time from TXFault state to TXFault bit set (value=1b) and IntL asserted. | |
Flag Assert Time | ton_flag | 200 | ms | Time from occurrence of condition triggering flag to associated flag bit set (value=1b) and IntL asserted. | |
Mask Assert Time | ton_mask | 100 | ms | Time from mask bit set (value=1b)3 until associated IntL assertion is inhibited | |
Mask Deassert Time | toff_mask | 100 | ms | Time from mask bit cleared (value=0b)3 until associated IntL operation resumes | |
DataPathDeinit Max Duration5 | DataPathDeinit_MaxDuration | see CMIS memory P01h: B144 | |||
DataPathInit Max Duration5 | DataPathInit_MaxDuration | see CMIS memory P01h: B144 | |||
Module Pwr Up Max Duration6 | ModulePwrUp_MaxDuration | see CMIS memory P01h: B167 | |||
ModulePwrDn Max Duration6 | ModulePwrDn_MaxDuration | see CMIS memory P01h: B167 | |||
Data Path TX Turn On Max Duration5 | DataPathTxTurnOn_MaxDuration | see CMIS memory P01h: B168 | |||
Data Path TX Turn Off Max Duration5 | DataPathTxTurnOff_MaxDuration | see CMIS memory P01h: B168 | |||
Notes: change Vout:IntL=Vol. |
Measured from low to high SDA edge of the Stop condition of the read transaction.
Measured from low to high SDA edge of the Stop condition of the write transaction.
Rx LOS condition is defined at the optical input by the relevant standard.
Measured from the low to high SDA edge of the Stop condition of the Write transaction until the IntL for the state change Vout:IntL=Vol, unless the module advertises a less than 1 ms duration in which case there is no defined measurement.
Measured from the low to high SDA edge of the Stop condition of the Write transaction until the IntL for the state
5
6
7
Parameters | Symbol | Max | Unit | Conditions |
Rx Squelch Assert Time | ton_Rxsq | 100 | ms | Time from loss of Rx input signal until the squelched output condition is reached, see 5.8.1 |
Rx Squelch Deassert Time | toff_Rxsq | 10 | s | Time from resumption of Rx input signals until normal Rx output condition is reached, see 5.8.1 |
Tx Squelch Assert Time | ton_Txsq | 400 | ms | Time from loss of Tx input signal until the squelched output condition is reached, see 5.8.2 |
Tx Squelch Deassert Time | toff_Txsq | 10 | s | Time from resumption of Tx input signals until normal Tx output condition is reached, see 5.8.2 |
Tx Disable Assert Time | ton_txdis | 100 | ms | Time from the stop condition of the Tx Disable write sequence 1 until optical output falls below 10% of nominal, see note 2 and 3. |
Tx Disable Assert Time (optional fast mode) | ton_txdisf | 3 | ms | Time from Tx Disable bit set (value = 1b)1 until optical output falls below 10% of nominal, see note 2 and 3. |
Tx Disable Deassert Time | toff_txdis | 400 | ms | Time from Tx Disable bit cleared (value = 0b)1 until optical output rises above 90% of nominal, see note 2 and 3. |
Tx Disable Deassert Time (optional fast mode) | toff_txdisf | 10 | ms | Time from Tx Disable bit cleared (value = 0b)1 until optical output rises above 90% of nominal. |
Rx Output Disable Assert Time | ton_rxdis | 100 | ms | Time from Rx Output Disable bit set (value = 1b)1 until Rx output falls below 10% of nominal |
Rx Output Disable Deassert Time | toff_rxdis | 100 | ms | Time from Rx Output Disable bit cleared (value = 0b)1 until Rx output rises above 90% of nominal |
Squelch Disable Assert Time | ton_sqdis | 100 | ms | This applies to Rx and Tx Squelch and is the time from bit set (value = 0b)1 until squelch functionality is disabled. |
Squelch Disable Deassert Time | toff_sqdis | 100 | ms | This applies to Rx and Tx Squelch and is the time from bit cleared (value = 0b)1 until squelch functionality is enabled |
Notes: |
Measured from LOW to HIGH SDA signal transition of the STOP condition of the write transaction.
CMIS 4.0 and beyond the listed values are superseded by the advertised DataPathTxTurnOff_MaxDuration and DataPathTxTurnOn_MaxDuration times in P01h.168.
Listed values place a limit on the DataPathTxTurnOff_MaxDuration and DataPathTxTurnOn_MaxDuration times (P01h.168) that can be advertised by such modules (for CMIS 4.0 and beyond).
1
For detailed electrical specifications see the appropriate specification, e.g. IEEE Std 802.3-2018 [12]; IEEE Std
802.3cd [13], 802.3ck [14]; FC-PI-6 [1], FC-PI-7[2], FC-PI-8 [3]; OIF-CEI-4.0 [20], or the InfiniBand
specifications [16].
6
Partial or complete squelch specifications may be provided in the appropriate specification. Where squelch is
not fully defined by the appropriate specification, the recommendations given in clause 5.8.1 and 5.8.2 may be
used.
10
RD(n)+/- are SFP112 module receiver data outputs. Rx(n)+/- are AC-coupled 100 differential lines that
should be terminated with 100 Ohms differentially at the Host ASIC(SerDes). The AC coupling is inside the
SFP112 module and not required on the Host board. When properly terminated, the differential voltage swing
shall be less than or equal to 900 mVpp or the relevant standard, whichever is less. 16
Output squelch for loss of optical input signal, hereafter RX Squelch, is required and shall function as follows.
In the event of the Rx input signal on optical lane/lanes becomes equal to or less than the level required to
assert LOS, then the electrical receiver output shall be squelched. SFP112 may have one or more Rx optical
lanes associated with the electrical Rx output as shown in Table 24. In the squelched state output impedance
levels are maintained while the differential voltage amplitude shall be less than 50 mVpp. 4
In normal operation the default case has RX Squelch active. Rx Squelch can be deactivated using Rx
Squelch Disable through the 2-wire serial interface.
TD(n)+/- are SFP112 module transmitter data inputs. They are AC-coupled 100 Ohm differential lines with
100 Ohm differential terminations inside the SFP112 optical module. The AC coupling is implemented inside
the SFP112 optical module and not required on the Host board. 11
Output squelch for loss of electrical signal, hereafter TX Squelch, is an optional function. Where implemented
it shall function as follows. In the event of the differential, peak-to-peak electrical signal amplitude on electrical
input channel is less than 70 mVpp, the transmitter optical output/outputs shall be squelched and the
associated TxLOS flag set. The loss of the incoming electrical input channels causes the optical output
channel/channels to be squelched. 17
For applications, e.g., Ethernet, where the transmitter off condition is defined in terms of average power,
squelching by disabling the transmitter is recommended and for applications, e.g. InfiniBand, where the
transmitter off condition is defined in terms of OMA, squelching the transmitter by setting the OMA to a low
level is recommended. 22
In module operation, where TX Squelch is implemented, the default case has TX Squelch active. TX Squelch
can be deactivated using TX Squelch Disable through the 2-wire serial interface. TX Squelch is an optional
function. If TX squelch is implemented, the disable squelch must be provided. 26
The module provides 2 inputs Speed1 and Speed2 that can optionally be used for rate selection as defined in
Table 19. Speed1 controls the receive path signaling rate and Speed2 controls the transmit path signaling rate
capability.
31
The rate select functionality can also be controlled by software as defined by the SFP112 management
specification.
34
For 64 GFC FC-PI-8 [3] operation link speed is determined with FC Link Speed Negotiation (LSN) exchange
while the link is operating at 32 GFC. 37
38 Table 19- Rate Select Hardware Control
Parameters | State | Conditions |
Speed1 | Low | RX signaling rate of 14.025 GBd (16 GFC)1 |
High | RX signaling rate of 28.05 GBd (32 GFC) | |
Speed2 | Low | TX signaling rate of 14.025 GBd (16 GFC) 1 |
High | TX signaling rate of 28.05 GBd (32 GFC) | |
Note 1: For SFP112 operating at 128 GFC there is no requirement for 16 GFC or rate select. |
39
40
41
42
43
2
SFP112 has two designated power supply pins VccT and VccR in the connector. Power is applied
concurrently to VccT and VccR. 5
A host board together with the SFP112 module(s) forms an integrated power system. The host supplies stable
power to the module. The module limits electrical noise coupled back into the host system and limits inrush
charge/current during hot plug insertion. 9
All power supply requirements in Table 21 shall be met at the maximum power supply current. No power
sequencing of the power supply is required of the host system since the module sequences the contacts in the
order of ground, supply and signals during insertion. 13
14 5.10.1 Power Classes and Maximum Power Consumption
15 SFP112 power classes are defined in Table 20. Since different classes of modules exist with pre-defined
16 maximum power consumption limits, it is necessary to avoid exceeding the system power supply limits and
17 cooling capacity when a module is inserted into a system designed to only accommodate lower power
18 modules. It is recommended that the host identify the power class of the module before allowing the module
19 to go into high power mode. 20
21 Power levels associated with classifications of modules are shown in Table 20. 22
23 Table 20- Power Classification
Power Class | Max Power (W) | CMIS Register |
1 | 1.0 | Direct readout of Page 00h Byte 200[000xxxxx] |
2 | 1.5 | Direct readout of Page 00h Byte 200[001xxxxx] |
3 | 2.0 | Direct readout of Page 00h Byte 200[010xxxxx] |
4 | 3.5 | Direct readout of Page 00h Byte 200[011xxxxx] |
5 | 5.0 | Direct readout of Page 00h Byte 200[100xxxxx] |
6 | reserved | NA |
7 | reserved | NA |
81 | > 5 W | Direct readout of Page 00h Byte 200[111xxxxx] |
Note: 1. When a module reports power class 8 the host must read CMIS Page 00h Byte 201 to determine module power dissipation. Please see CMIS Byte 201 register definition for more information. |
24
In general, the higher power classification levels are associated with higher data rates and longer reaches.
The system designer is responsible for ensuring that the maximum case temperature does not exceed the
case temperature requirements. 28
29
Note: Filter capacitors values are informative and vary depending on applications.
VccT may be connected to VccR provided the applicable derating of the maximum current limit is used.
The host board should use the power supply filtering equivalent to that shown in Figure 24. 3
4
6
7
Any voltage drop across a filter network on the host is counted against the host DC set point accuracy
specification. Inductors with DC Resistance of less than 0.1 Ohm should be used in order to maintain the
required voltage at the Host Card Edge Connector. It is recommended that the 22 F capacitors each have an
equivalent series resistance of 0.22 ohms. 12
The specifications for the power supply are shown in Table 21. The limits in Table 21 apply to the combined
current that flows through all inductors in the power supply filter (represents ICC host in Figure 24). An
example test method for measuring inrush current can is Keysight Technologies application brief 5991-
2778EN.pdf.
In order to avoid exceeding the host system power capacity, upon hot-plug, power cycle or reset, all SFP112
modules shall power up in Low Power Mode if TxDisable is asserted. If TxDisable is not asserted the module
will proceed to High Power Mode without host intervention. Figure 25 shows waveforms for maximum
instantaneous, sustained and steady state currents for Low Power and High Power modes. Specification
values for maximum instantaneous, sustained and steady state currents at each power class are given in
Table 21.
24
1
2 Table 21- Power Supply specifications, Instantaneous, sustained and steady state current limits
Parameter | Symbol | Min | Nom | Max | Unit |
Power supply voltages VccTx, VccTx1, VccRx, VccRx1 including ripple, droop and noise below 100 kHz1 | 3.135 | 3.3 | 3.465 | V | |
Module inrush - instantaneous peak duration | T_ip | 50 | µs | ||
Module inrush - initialization time | T_init | 500 | ms | ||
Low Power Mode2 | |||||
Power Consumption | P_lp | .5 | W | ||
Instantaneous peak current at hot plug | Icc_ip_lp | - | 200 | mA | |
Sustained peak current at hot plug | Icc_sp_l p | - | 165 | mA | |
Steady state current | Icc_lp | See Note 4 | mA | ||
High Power Class 1 module | |||||
Power Consumption | P_1 | 1.0 | W | ||
Instantaneous peak current | Icc_ip_1 | - | 400 | mA | |
Sustained peak current | Icc_sp_1 | - | 330 | mA | |
Steady state current | Icc_1 | See Note 4 | mA | ||
High Power Class 2 module | |||||
Power Consumption | P_2 | 1.5 | W | ||
Instantaneous peak current | Icc_ip_2 | - | 600 | mA | |
Sustained peak current | Icc_sp_2 | - | 495 | mA | |
Steady state current | Icc_2 | See Note 4 | mA | ||
High Power Class 3 module | |||||
Power Consumption | P_3 | 2.0 | W | ||
Instantaneous peak current | Icc_ip_3 | - | 800 | mA | |
Sustained peak current | Icc_sp_3 | - | 660 | mA | |
Steady state current | Icc_3 | See Note 4 | mA | ||
High Power Class 4 module | |||||
Power Consumption | P_4 | 3.5 | W | ||
Instantaneous peak current | Icc_ip_4 | - | 1400 | mA | |
Sustained peak current | Icc_sp_4 | - | 1155 | mA | |
Steady state current | Icc_4 | See Note 4 | mA | ||
High Power Class 5 module | |||||
Power Consumption | P_5 | 5.0 | W | ||
Instantaneous peak current | Icc_ip_5 | - | 2000 | mA | |
Sustained peak current | Icc_sp_5 | - | 1650 | mA | |
Steady state current | Icc_5 | See Note 4 | mA | ||
High Power Class 8 module | |||||
Power Consumption | P_83 | W | |||
Instantaneous peak current | Icc_ip_8 | - | - | P_8/2.5 | A |
Sustained peak current | Icc_sp_8 | - | - | P_8/3.03 | A |
Steady state current | Icc_8 | - | - | 4 | A |
Notes: |
Measured at VccT, VccR, VccT1, and VccR1.
Host designers are responsible for handling 1.5W Low Power Mode SFP(INF-8074i)/SFP+(SFF-8431) classic modules as appropriate in their system.
User must read management register for maximum power consumption.
The module must stay within its declared power class.
3
1
LOW POWER MODE
Transmitters OFF
HIGH POWER MODE
Transmitters OFF
HIGH POWER MODE
Transmitters ON
T_init
T_ip
T_ip
T_init
T_ip
T_init
TxOff state inrush current
graph is informative. It cannot exceed the High power mode Transmitter On currents. No currents or power consumption are specified
Icc Host (ma)
t (ms)
Hot Plug
Instant
Host Enables High Power Mode
Host Enables Transmitters
4
The host shall generate an effective weighted integrated spectrum RMS noise less than the eN_Host value in
Table 21 when tested by the methods of SFF-8431 [30] D.17.1 with the following exceptions:
a. The truncated function equation with coefficients from Table 22.
b. The resistor shall draw the maximum rated current per contact.
c. The frequency response of the truncated function is illustrated in Figure 26.
11
12 Table 22- Truncated Filter Response Coefficients for Host Power Supply Noise Output
13
Frequency | a | b | c | d | e |
10 Hz < f < 240.2 Hz | 0 | 0 | 0 | 0 | -0.1 |
240.2 Hz < f < 24.03 kHz | 0.3784 | -3.6045 | 12.694 | -19.556 | 11.002 |
24.03 kHz < f < 360.4 kHz | -22.67038 | 430.392 | -3053.779 | 9574.26 | -11175.98 |
360.4 kHz < f < 12.6 MHz | 3.692166 | -91.467 | 838.80 | -3400.38 | 5139.285 |
12.6 MHz < f < 24 MHz | 0 | 0 | 0 | 0 | -60 |
14
1
2 Figure 26: Truncated Transfer Response for Host Board Power Supply Noise Output measurement
3
The SFP112 modules when plugged into a reference module compliance board shall generate noise less than
the value in Table 21 when tested by the methods of SFF-8431[30], D.17.2. The module must pass module
power supply noise output test in all operating modes. This test ensures the module will not couple excessive
noise from inside the module back onto the host board. 9
The RMS module noise voltage output is defined in the frequency band from 10 Hz to 10 MHz. Module noise
shall be measured with an appropriate probing technique at the input stage of the Vcc PSU filtering network,
see Figure 27.
13
The SFP112 modules shall meet all requirements and remain fully operational in the presence of a reference
noise waveform described in Table 23 superimposed on the DC voltage. The reference noise waveform
consists of triangular sweep of supply voltage Vcc PSU from 3.135 to 3.465 while a sinusoidal 1 kHz to 1 MHz
noise is added to Vcc PSU. This emulates the worst case noise the module must tolerate and be fully
operational. The reference noise shall be injected at the input stage of the Vcc PSU filtering network, see
Figure 27. An appropriate probing technique is required for noise characterization. 21
22 Table 23- Power Supply Output Noise and Tolerance Specifications
Parameter | Symbol | Min | Nom | Max | Unit |
Host RMS noise output 40 Hz-10 MHz (eN_Host) | 25 | mV | |||
Module RMS noise output 10 Hz - 10 MHz | 30 | mV | |||
Module sinusoidal power supply noise tolerance 1 kHz - 1 MHz (p-p) 1 | PSNRmod | 66 | mV | ||
Vcc PSU triangular tolerance waveform amplitude (p-p) 2 | 3.135 | 3.465 | V | ||
Vcc PSU triangular tolerance waveform frequency (p-p) | 0.001 | 1000 | Hz | ||
Notes:
|
23
1
2 Figure 27: Module Noise Injection Point
3
4
Where ESD performance is not otherwise specified, e.g., in the InfiniBand specification, the SFP112 module
shall meet ESD requirements given in EN61000-4-2 [8], criterion B test specification when installed in a
properly grounded cage and chassis. The units are subjected to 15 kV air discharges during operation and 8
kV direct contact discharges to the case. All the SFP112 module and host pins including high speed signal
pins shall withstand 1000 V electrostatic discharge based on Human Body Model per ANSI/ESDA/JEDEC JS-
001 [9].
12
13
1
2 6. Optical Port Mapping and Optical Interfaces
3
Table 24 defines the mapping of electrical TX data inputs to optical ports. The mapping of the RX optical ports
to electrical RX outputs is symmetric. Note that there is no defined mapping of electrical input/output to optical
wavelengths for WDM applications. 8
Electrical Data Input Reference | LC, SN, MDC | SN, MDC, MPO-12 |
1 TX fiber | 2 TX fibers | |
TD0+/- | TX-1 | TX-1 |
TD1+/- | TX-2 |
10
Four examples of the SFP-DD/SFP-DD112 optical interface port are a male MPO receptacle (see Figure 24),
a dual LC (see Figure 30), a SN receptacle (see Figure 33), or a MDC receptacle (see Figure 34). The
recommended location and numbering of the optical ports for each of the Media Dependent Interfaces is
shown in Figure 28.
16
TX
1 2
RX
2 1
TX RX
Duplex LC MPO-12
1
2
TX
TX
RX
RX
SN-Dual
MDC-Dual
1
2
TX
TX
RX
RX
Note: The transmit and receive optical lanes shall occupy the positions depicted here when looking into the MDI receptacle with the connector keyway feature on top.
17
18
19 Figure 28: Optical Media Dependent Interface port assignments
The optical plug and receptacle for the MPO-12 connector is specified in TIA-604-5 [10], IEC 61754-7 [23] and
shown in Figure 29. Note: Two alignment pins are present in each receptacle. 4
Aligned keys are used to ensure alignment between the modules and the patchcords. The optical connector is
orientated such that the keying feature of the MPO receptacle is on the top. 7
8
9
10
11 Figure 29: MPO-12 Single Row optical patch cord and module receptacle
12
The Duplex LC optical patchcord and module receptacle is specified in TIA-604-10 [11], IEC 61754-20 [24]
and shown in Figure 30.
16
17
18
19
20 Figure 30: Duplex LC optical patchcord and module receptacle
21
LC connector latch extends up to 2.15 mm above the SFP-DD/SFP-DD112/SFP112 as shown in Figure 31.
To avoid interference in stacked SFP-DD/SFP-DD112/SFP112 cages configurations the minimum vertical port
pitch is 14.9 mm, see Figure 32.
4
5 Figure 31: LC connector excursion above module nose height
6
7
9
The SN optical connector and receptacle for SFP-DD/SFP-DD112 module is specified in SN-60092019 [22]
and shown in Figure 26. The top key and offset bottom key are used to ensure alignment between the
modules and the patch cords. 14
15
16
17
18 Figure 33: SN optical connector plug and two-port module receptacle
19
The MDC optical plug and receptacle for an SFP-DD/SFP-DD112 module is specified in USC-11383001 [25]
and shown in Figure 27. The optical connector is orientated such that the keying feature of the MDC
receptacle is on the top.
5 | ||
6 | ||
7 | ||
8 | ||
9 | Figure 34: MDC optical connector plug and two-port module receptacle | |
10 | ||
11 | Module Color Coding and Labeling |
12 If provided, color coding shall be on an exposed feature of the SFP-DD/SFP-DD112 module (a feature or
13 surface extending outside of the bezel). Color code are outside the scope of this specification. 14
Each SFP-DD/SFP-DD112 module shall be clearly labeled. The complete labeling need not be visible when
the SFP-DD/SFP-DD112 module is installed, and the bottom of the device is the recommended location for the
label. Labeling shall include:
18
Appropriate manufacturing and part number identification
Appropriate regulatory compliance labeling
A manufacturing traceability code.
22
23 The label should also include clear specification of the external port characteristics such as: 24
Optical wavelength
Required fiber characteristics
Operating data rate
Interface standards supported
Link length supported.
30
31 The labeling shall not interfere with the mechanical, thermal or EMI features. 32
33
The cages and modules specifications defined in this chapter are illustrated in Figure 35 (press fit cage) and
Figure 36 (pluggable module). All Pluggable modules and direct attach cable plugs must mate to the
connectors and cages defined in this specification. (See SFF-8432 [31]) Heat sink/clip thermal designs are
application specific and not specifically defined by this specification. The SFP-DD and SFP-DD112 modules
and cage support both a pull tab and a bail latch solution. Details on bail latch retention and extraction
specifications can be found in SFF-8432. 9
10
11
12
13
14 Figure 35: Press fit cage
1
2
3
4
6
7 Type 2 module
8
9
10 Figure 36: SFP-DD and SFP-DD112 Modules
11
12
A listing of the SFP-DD/SFP-DD112/SFP112 datums for the various components are contained in Table 25.
To reduce the complexity of the drawings, all dimensions are considered centered unless otherwise specified.
Dimensions and tolerancing conform to ASME Y14.5-2009 [4]. All dimensions are in millimeters. 6
7
8
Datum | Description |
A | Width Of module |
B | Bottom surface of module |
C | Leading edge of signal contact pads on module paddle card |
D | Hard stop on module |
E | Host board thru hole to accept primary cage press fit pin |
F | Hard stop on cage |
G | Bottom surface of bezel cutout |
K | Host board thru hole #1 to accept connector guidepost |
L | Host board thru hole #2 to accept connector guidepost |
P | Vertical center line of internal surface of cage |
S | Seating plane of cage on host board |
Z | Top surface of host board |
AA | Center line of module paddle card width |
BB | Top surface of module paddle card |
CC | Center line of connector slot width |
DD | Seating plane of connector on host board |
10
11
12
13
14
Published Specification SFP-DD/SFP-DD112/SFP112 Rev. 5.1
The alignment of the cage, connector and module are shown in Figure 37. 3
4
5
Published Specification SFP-DD/SFP-DD112/SFP112 Rev. 5.1
The mechanical outline for the SFP-DD/SFP-DD112 modules and direct attach cables are shown in Figure 38.
The module shall provide a means to self-lock with the cage upon insertion. The module package dimensions
are defined in Figure 39. The dimensions that control the size of the module that extends outside of the cage
are listed as maximum dimensions per Note 4 in Figure 39. 6
7
8
9
10
11
12 Type 1 module
13
14 Type 2 module
15
16 Figure 38: Type 1 and Type 2 Modules
17
18
19
20
1
9
2
3
1
2
3
1
2
3
4
5 Figure 39: Detailed dimensions of module
6
7
1 7.5 SFP-DD/SFP-DD112/SFP112 Module Flatness and Roughness
2
3 Module flatness and roughness are specified to improve module thermal characteristics when used with a
4 riding heat sink. Relaxed specifications are used for lower power modules to reduce cost. The module flatness
5 and roughness specifications apply to the specified heat sink contact area as specified in Figure 39.
6 Specifications for Module flatness and surface roughness are shown in Table 26. 7
8 Table 26- Module flatness specifications
Power Class | Module Flatness (mm) | Surface Roughness (Ra,µm) |
1 | 0.075 | 1.6 |
2 | 0.075 | 1.6 |
3 | 0.075 | 1.6 |
4 | 0.075 | 1.6 |
5 | 0.050 | 0.8 |
6 | reserved | reserved |
7 | reserved | reserved |
8 | 0.050 | 0.8 |
9
10 7.6 SFP-DD Module paddle card dimensions
11
12
13
1
2
3 Figure 40: Module paddle card dimensions
4
1
2
3
4 Figure 41: Module pad dimensions
5
The requirements for insertion forces, extraction forces and retention forces are specified Appendix A. The
SFP-DD/SFP-DD112/SFP112 cage and modules are designed to ensure that excessive force applied to a
cable does not damage the SFP-DD cage or host connector. If any part is damaged by excessive force, it
should be the cable or media module and not the cage or host connector which is part of the host system.
Examples of module retention mechanisms are found in SFF-8432 [31] Figures 4-4 through 4-7. The contact
9 7.8 Press fit Cage Mechanical
10 The SFP-DD Cage is shown in Figure 42 with detailed drawings in Figure 43. Recommendations for the cage
11 bezel opening are shown in Figure 45. 12
13
14 Figure 42: Press Fit 1x1 Cage
15
1
2
3
4
5
1
1
2
3
4 Figure 43: Press Fit 1x1 Cage Design
5
6
1
2
3 Figure 44: Press Fit Cage Detail
1
2
3 Figure 45: 1 x n bezel opening
4
5
1 7.9 SMT Electrical Connector Mechanical
2 The SFP-DD Connector is a 40-contact, right angle connector. The SMT connector is shown in Figure 46 with
3 detailed drawings in Figure 47.
4
5
6 Figure 46: SMT connector
7
1
2
3
1
2
3
4
5
6
7
8
9
10 Figure 47: 1x1 Connector Design and Host PCB Pin Numbers
11
12
A typical host board mechanical layout for attaching the SFP-DD SMT Connector and press fit Cage System is
shown in Figure 48. Location of the pattern on the host board is application specific. 4
To achieve 25-50 Gbps performance pad dimensions and associated tolerances must be adhered, and
attention paid to the host board layout. 7
8
9
10
1
2
3
5
SFP-DD112 module mechanical specifications are compatible with SFP-DD module mechanical specifications
in chapter 7, below are the list of relevant SFP-DD sections applicable to SFP-DD112:
- 7.2 SFP-DD/SFP-DD112/SFP112 Datums, Dimensions and Component Alignment
- 7.7 SFP-DD/SFP-DD112/SFP112 Module Extraction and Retention Forces.
-
The module paddle card dimensions of the SFP-DD112 have been improved to support 100 Gb/s PAM4 (up to
13
SFP-DD112 supports multiple connector/cage form factors. SFP-DD112 cages/connectors/modules are
compatible with SFP-DD cages/connectors/modules, SFP-DD112 cages/connectors also accepts
SFP+/SFP28 [30]/[28] family of modules. Examples of SFP-DD112 cages are:
1x1 surface mount connector/cage. 18
The mechanical outline for the SFP-DD112 module and direct attach cables are identical to Figure 22. The
module shall provide a means to self-lock within the cage upon insertion. The module package dimensions are
identical to Figure 23 with exception of bottom view shown in Figure 49. For SFP-DD112 modules the bottom
surface of the module within the cage shall be flat without a pocket. The options for the position of the label
could include the bottom surface of the module that protrudes outside the bezel of the cage or etched into the
metal surface.
26
27 Caution should be exercised that any etchings do not affect thermal performance. Flatness and roughness
28 specs are defined in 7.5 apply to both top and bottom surfaces of SFP-DD112 module. 29
30
31 Figure 49: SFP-DD112 Module Bottom View Details
32
1 8.3 SFP-DD112 Improved Module paddle card dimensions
2
3
4
1
2
3 Figure 50: Improved module paddle card dimensions
4
5
6
1
2 Figure 51: Improved module pad dimensions
3
1
2 8.4 Press fit Cage Mechanical
3
4 SFP-DD112 press fit cage is shown in Figure 52 with detailed drawings in Figure 43. SFP-DD112 cage bezel
5 opening is identical to SFP-DD cage bezel as shown in Figure 45. 6
7
8
9
10 Figure 52: Press Fit 1x1 Cage
11
12
13
14
15
16
17
18
19
20
21
5
6
7 Figure 53: SMT connector
8
9
1
2
3
4
5
6
7
8 Figure 54: 1x1 Connector Design and Host PCB Pin Numbers
9
10
11
12
A typical host board mechanical layout for attaching the SFP-DD112 SMT Connector and press fit Cage
System is shown in Figure 48. Location of the pattern on the host board is application specific. 4
To achieve 112 Gbps (56 GBd) performance pad dimensions and associated tolerances must be adhered, and
attention paid to the host board layout. 7
8
9
10
1
2
3
4
5
6
SFP112 module mechanical specifications are compatible with SFP+/SFP28 [30]/[28]module mechanical
specifications. Below are the list of relevant SFP+/SFP28 and SFP-DD sections applicable to SFP112:
- IPF General descriptions, [31][31] Chapter 3.0
- IPF Modules Dimensions, Retention/Extraction, and Durability, [31] Chapter 4
- IPF Cage Requirements, [31], Chapter 5
- SFP Single Cage Host Board Mechanical Layout, [26] Chapter 4A
- SFP+ Canged Cage Host Board Mechanical Layout, [32] Chapter 4
- SFP-DD/SFP-DD112/SFP112 Module Flatness and Roughness, 7.5
- SFP-DD/SFP-DD112/SFP112 Module Extraction and Retention Forces, 7.7.
11
The module paddle card dimensions of the SFP112 have been improved to support 100 Gb/s PAM4 (up to 56
SFP112 supports multiple connector/cage form factors. All combinations of cages/connectors defined in the
specification are backwards compatible to accept classic SFP28 and SFP+ modules. In addition, SFP112
modules are compatible with SFP28/SFP+ hosts for operation at lower speed. 19
20 9.2 SFP112 Datum for Module, Cage, Connector and Host
21 A listing of the SFP112 datums for the various components are contained in Table 27. To reduce the
22 complexity of the drawings, all dimensions are considered centered unless otherwise specified. Dimensions
23 and tolerancing conform to ASME Y14.5-2009 [4]. All dimensions are in millimeters. 24
25 Table 27: SFP112 Datum
Datum | Description |
A | Center line of connector slot |
B | Bottom surface of connector |
C | Round guide-post of connector |
D | Leading edge of signal contact pads on module paddle card |
E | Leading edge of low speed contact pads on module paddle card |
F | Front edge of module paddle card |
G | Top surface of module paddle card |
H | Center line of module paddle card width |
K | Host board thru hole #1 to accept connector guide-post |
L | Host board thru hole #2 to accept connector guide-post |
M | Vertical center line of Datum L and Datum K |
N | Top surface of host board |
P | Hard stop on module |
R | Hard stop on cage |
S | Host board thru hole to accept primary cage press fit pin |
T | Center line of module width |
U | Bottom surface of module |
26
27
1
2 9.3 SFP112 Cage, Connector, Module Alignment
3 The alignment of the cage, connector and module are shown in Figure 56.
4
5 Figure 56: SFP112 1x1 Cage and Host PCB Layout
SFP112 modules mechanical dimension are identical to SFP+/SFP28 [30]/[28] modules with exception of
viewing windows to inspect high speed pad as shown Figure 57. A SFP28/56 [28] cage can be used with the
SFP112 connector. For SFP112 modules the bottom surface of the module within the cage shall be flat
without a pocket. The options for the position and the bottom view of the label could include the bottom
surface of the module that protrudes outside the bezel of the cage or etched into the metal surface. Caution
should be exercised that any etchings do not affect thermal performance. 8
Flatness and roughness specs are defined in 7.5 to both top and bottom surfaces of SFP112 module.
NOTES APPLY TO MODULE DRAWINGS:
11
1
2
3 Figure 57: SFP112 High Speed Pads Viewing Windows
The SFP112 module paddle card pad dimensions have been modified to support 112 Gb/s (56 GBd) serial
data rates. SFP112 module paddle card pad dimensions optimized for 56 GBd operation shown in Figure 58.
All other module dimensions, except for the pads are the same as the SFP+/SFP28 [30]/[28] specifications.
5
6 Figure 58: SFP112 improved module paddle card dimensions
The SFP112 Connector is a 20-contact improved right-angle connector compatible with SFP+ modules [31].
The SFP112 SMT connector is shown in Figure 59. 4
5
6 Figure 59: SFP112 SMT Connector
7
60. The detailed optimized SFP112 host pad layout is shown in Figure 60 and the host PCB pad numbers
shown in Figure 62. Note: the cage footprint for SFP112 is the same the same as SFP cage footprint, [26] see
Figure 4A.
6
7 To achieve 112 Gbps (56 GBd) operation the SFP112 pad dimensions and associated tolerances have
8 improved compared to SFP+/SFP28 [30]/[28] as shown in Figure 61. One must adhere and pay attention to
9 the host board layout for 56 GBd operation. 10
11
12 Figure 60: SFP112 Host Pad Layout
13
|
1
2
3 Figure 61: SFP112 Detailed Host Pad Layout
4
5 Figure 62: 1x1 Connector Design and Host PCB Pad Numbers
6
The SFP-DD/SFP-DD112/SFP112 module shall operate within one or more of the case temperatures ranges
defined in Table 28. The temperature ranges are applicable between 60 m below sea level and 1800 m above
sea level, NEBS GR-63 [18], utilizing the host systems designed airflow. 6
7 Table 28- Temperature Range Class of operation
Class | Case Temperature Range |
Standard | 0°C through 70°C |
Extended | -5°C through 85°C |
Industrial | -40°C through 85°C |
8
SFP-DD/SFP-DD112/SFP112 are designed to allow for up to 48 modules; stacked, ganged and/or belly-to-
belly in a 1U 19" rack, with the appropriate thermal design for cooling/airflow. 11
12
13
EIA-364-1000 [7] shall be used to define the test sequences and procedures for evaluating the connector
system described in this document. Where multiple test options are available, the manufacturer shall select
the appropriate option where not previously specified. The selected procedure should be noted when
reporting data. If there are conflicting requirements or test procedures between EIA-364 procedures and those
contained within this document, this document shall be considered the prevailing authority. Unless otherwise
specified, procedures for sample size, data, and collection to be followed as specified in EIA-364-1000. See
EIA-364-1000 Annex B for objectives of tests and test groups. 10
This document represents the minimum requirements for the defined product. Additional test conditions and
evaluations may be conducted within the defined EIA-364-1000 sequences. More extreme test conditions and
failure criteria may be imposed and still meet the requirements of this document. 14
Table 29 summarizes the performance criteria that are to be satisfied by the connector described in this
document. Most performance criteria are validated by EIA-364-1000 testing, but this test suite leaves some
test details to be determined. To ensure that testing is repeatable, these details are identified in Table 30.
Finally, testing procedures used to validate any performance criteria not included in EIA-364-1000 are
provided in Table 31.
20
21 Table 29- Form Factor Performance Requirements
Performance Parameters | Description/ Details | Requirements |
Mechanical/ Physical Tests | ||
Plating Type | Plating type on connector contacts | Precious (refer to 7.6 for plating details) |
Surface Treatment | Surface treatment on connector contacts; if surface treatment is applied, Test Group 6 is required | Manufacturer to specify |
Wipe length | Designed distance a contact traverses over a mating contact surface during mating and resting at a final position. If less than 0.127 mm, test group 6 is required | Manufacturer to specify |
Rated Durability Cycles | The expected number of durability cycles a component is expected to encounter over the course of its life | Connector/ cage: 100 cycles Module: 50 cycles |
Mating Force1 | Amount of force needed to mate a module with a connector when latches are deactivated | SFP112 module: 40 N MAX SFP-DD/SFP-DD112 module: 40 N MAX |
Unmating Force1 | Amount of forced needed to separate a module from a connector when latches are deactivated | SFP112 module: 12.5 N MAX SFP-DD/SFP-DD112 module: 30 N MAX |
Latch Retention1 | Amount of force the latching mechanism can withstand without unmating | SFP112 module: 90 N MIN SFP-DD/ SFP-DD112 module: 90 N MIN |
Cage Latch Strength1 | The amount of force that the cage latches can hold without being damaged. | 100 N MIN |
Cage Retention to Host Board1 | Amount of force a cage can withstand without separating from the host board | SFP112 module: 100 N MIN SFP-DD/ SFP-DD112 module: 100 N MIN |
Environmental Requirements | ||
Field Life | The expected service life for a component | 10 years |
Field Temperature | The expected service temperature for a component | 65°C |
Electrical Requirements | ||
Current | Maximum current to which a contact is exposed in use | 0.5 A per signal contact MAX 1.5 A per power contact MAX |
Operating Rating Voltage | Maximum voltage to which a contact is exposed in use | 30 V DC per contact MAX |
Note: 1. These performance criteria are not validated by EIA-364-1000 testing, see Table 31 for test procedures and pass/fail criteria. |
1
2
Table 30 describes the details necessary to perform the tests described in the EIA-364-1000 test sequences.
Testing shall be done in accordance with EIA-364-1000 and the test procedures it identifies in such a way that
the parameters/ requirements defined in Table 29 are met. Any information in this table supersedes EIA-364-
1000.
Performance Parameters | Description/ Details | Requirements |
Mechanical/ Physical Tests | ||
Durability (preconditioning) | EIA-364-09 To be tested with connector, cage, and module. Latches may be locked out to aid in automated cycling. | No evidence of physical damage |
Durability1 | EIA-364-09 To be tested with connector, cage, and module. Latches may be locked out to aid in automated cycling. | No visual damage to mating interface or latching mechanism |
Environmental Tests | ||
Cyclic Temperature and Humidity | EIA-364-31 Method IV omitting step 7a Test Duration B | No intermediate test criteria |
Vibration | EIA-364-28 Test Condition V Test Condition Letter D Test set-up: Connectors may be restrained by a plate that replicates the system panel opening as defined in this specification. External cables may be constrained to a non- vibrating fixture a minimum of 8 inches from the module. For cabled connector solutions: Wires may be attached to PCB or fixed to a non-vibrating fixture. | No evidence of physical damage -AND- No discontinuities longer than 1 μs allowed |
Mixed Flowing Gas | EIA-364-65 Class II See Table 4.1 in EIA-364-1000 for exposure times Test option Per EIA-364-1000 option 3 | No intermediate test criteria |
Electrical Tests | ||
Low Level Contact | EIA-364-23 | 20 mΩ Max change from |
Resistance2 | 20 mV DC Max, 100 mA Max | baseline |
To include wire termination or connector-to-board | ||
termination | ||
Dielectric | EIA-364-20 | No defect or breakdown |
Withstanding | Method B | between adjacent contacts |
Voltage | 300 VDC minimum for 1 minute | -AND- |
Applied voltage may be product / application specific | 1 mA Max Leakage Current | |
Notes:
|
8
9
10
11
12
13
14
15
16
17
18
1
Table 31 describes the testing procedures necessary to validate performance criteria not validated by EIA-
364-1000 testing. The tests are to be performed in such a way that the parameters/ requirements defined in
Table 29 are met.
5
6 Table 31- Additional Test Procedures
Tests | Test Descriptions and Details | Pass/ Fail Criteria’s |
Mechanical/ Physical Tests | ||
Mating Force1 | EIA-364-13 | -AND- No physical damage to any components |
Unmating Force1 | Mating/ unmating rate 12.7 mm/min To be tested with cage, connector, and module. | |
Latching mechanism deactivated (locked out). | ||
Latch Retention1 | EIA-364-13 | |
Mating/ unmating rate 12.7 mm/min | ||
To be tested with cage, connector, and module. | ||
Latching mechanism engaged (not locked out). | ||
Latch Strength | An axial load applied using a static load or ramped | |
loading to the specified load. | ||
To be tested with cage, connector, and module or | ||
module representative tool without heat sinks | ||
Latching mechanism engaged (not locked out). | ||
Cage Retention to | Tested with module, module analog, or fixtures mated | No physical damage to any |
Host Board | to cage. | components |
Pull cage in a direction perpendicular to the board at a | -AND- | |
rate of 25.4 mm/min to the specified force. | Cage shall not separate from | |
board | ||
Electrical Tests | ||
Current | EIA-364-70 Method 3, 30-degree temperature rise Contacts energized: All signal and power contacts energized simultaneously | Refer to Table 29 for current magnitude |
Note: 1. Values listed in Table 29 apply with or without the presence of a riding heat sink. |
7
8
1
End of Document
2