P1394b Working Group Meeting Crowne Plaza Hotel Natick, Mass., Wednesday, September 24, 1997 Chair: David Wooten Secretary: Richard Churchill Group Reflector: p1394b@fireflyinc.com Group FTP Site: (accessible via ...) http://www.fireflyinc.com/p1394b/ 1. Introductions 2. Procedures 3. Review of Minutes 4. Spec. Development Schedule 5. PMD Liaison Report(s) 5.1 High-speed Long-haul - Colin Whitby-Strevens 5.2 Low-speed Long-haul - ??? 6. Task Group Reports 6.1 PHY-Link Interface - Martin Sodos 6.2 Detection of Beta mode capable node - Steve Bard 6.3 Initialization of Beta mode communications - Colin Whitby-Strevens 6.4 Standard protocol - Alistair Coles 6.5 Accelerations - Mike Teener 6.6 Network simulation - Chen 6.7 Cable/Connector - Max Bassler 6.7.1 Bill Northey, Berg 6.8 Standard electrical - Eric Hannah 7. New Business 8. Meeting Schedule 8.1 October (Maui, 23 & 24, Aston Wailea Hotel) 8.2 December (Florida?) 8.3 January (Houston?) 9. Review of Action Items 10. Adjournment Meeting called to order at 9:00 AM 1. Introductions Introductions around, and the attendance sign-up list was passed 2. Procedures Comments regarding necessity of signing up in order to be counted present were made. 3. Agenda and minutes Agenda of the present meeting was approved, and minutes of the previous meeting as well, without objection. 4. Spec. Development Schedule 5. PMD Liaison Report 5.1 High-speed long-haul report - Colin Whitby-Strevens Colin reported on the latest meeting, held the previous day. The principal order of business was the presentation of proposals for a high-speed, long-haul fiber connector system. Five presentations were made, by Panduit, AMP, ???, 3M and Lucent. A schedule for selection of a single fiber connector was agreed. All materials to be used in consideration of a connector must be available via the reflector and p1394b FTP site by October 13. Demonstrations are planned for all proposed connectors, with associated tools, etc., at the Maui meeting, October 22nd. Generation of a comparison matrix will be generated during the meeting on October 23rd. Final voting on a connector is planned for the December meeting tentatively planned for Florida. All of the proposed connectors are presently in some phase of being standardized by some standards body, and are required to be either standardized or in process as of the date of the final vote. Del Hanson presented data regarding optical transceivers, indicating that 1 and 2 GBd solutions are accomplishable, and that we can, to some extent, "future proof" for eventual support of 4 GBd by using a 50 micron glass fiber. He also made some recommendations regarding provision of support for POF in the future. Issues discussed included S800 interoperability with slower speed devices, concluding that there is no problem interoperating with S400 devices, but no economic incentive for S400 only devices. Also, there would be some technology development required for interoperability with S100 devices, since such technology is not currently on the market. Discussion also concluded that there would be a "factor of 2" cost entailed in specifying an S100, or S100/200 transceiver that is fully interoperable with S800/1600/3200 devices. After some debate, it was decided that the group would develop its final report and specification based upon S800 and higher data rates, and not deal with low-speed implementations. Moved by Max Bassler that we send liaison letters to DAVIC, VESA Home Networking, TIA and other groups regarding our recommendation of and plans to specify 50 micron fiber, and in the process of selecting from the present list of proposals for a connector. Seconded by Steve Bard. Motion carried unanimously. [Discussion of where the low-cost/low-speed folks are, and what they are doing, etc., was held, with comments about where they should present conclusion and recommendations, etc., with no real conclusion.] [David Wooten raised the point of the voting rules, reminding attendees that to vote at a particular meeting, the voter must be present at the meeting, and at one of the previous two meetings. Thus, registering attendance via the sign-in list is essential for purposes of maintain- ing voting privileges.] 5.2 Low-speed Long-haul No representatives of the Low-speed Long-haul group were present and ready to present a liaison report. 6. Task Group Reports 6.1 PHY-Link Interface - Martin Sodos Mr. Sodos apologized for not having slides ... they melted. Mr. Sodos noted the contentious the issue is, since opinions regarding whether the interface should grow wider or faster in order to support the increasingly faster data rates of S800, S1600 and beyond. A general description of clocking, management, etc., in dealing with the problems was given, with a confession that his views have changed over time. A call for volunteers was made, with Eric Hannah offering to send Martin all needed FrameMaker tools. [Suggestions as to possibly taking a SCSI-like fast/wide/fast&wide approach were offered. Also, a suggestion that the interface will not be a problem, due to the eventual integration of PHYs and LINKs obviating the problem. S800 might then be the only speed to address, with the succeeding speeds "solving" the problem via integration, and saving us the pain of defining the interface at higher speeds. The converse case of people wishing to control costs via use of "mis- matched" LINKs and PHYs (S400 LINK with S800 PHY, for one) and a general observation that separate LINKs and PHYs are with us for the foreseeable future were made.] [Default assumption is that the requirements will match all existing requirements, and planned future requirements ...] Martin Sodos' email address is 'msodos@issiusa.com'. [Wooten requested a show-of-hands for those willing to help. About seven indicated they were willing. Comments regarding the need for both PHY and LINK folks to participate were made. A suggestion that discussions be held on the p1394b reflector until such time as the volume of such traffic gets unmanagable. A further request that at least some systems folks be involved was made by Wooten.] 6.2 Detection of Beta mode capable node - Steve Bard Power-On Start-Up - Port powers up in suspend state - Delay one TpBias debounce interval, plus .5 BIAS... .... Beta "DS" to "DS" Start-up - Responds as a "Resume Target" - Reference Suspend/Resume Beta to Beta Start-up - No TpBias after BIAShold time (8,192 SCLKs ~170 us) - port transitions from suspend state to "b" low-on state - apply BS Tone on TPB pair for TONEhold time - "listen" for tone on TPA Pair for TONEmonitor time .... (*** hand-drawn timing diagrams *** _________ _________ port(x) ________| |_________________| |_______________ TONEhold TONEmonitor _________ ________ port(y) ____________________| |_____________________| | _______ _______________________ port(x) _______| |______________________| |__ __________________________________________ port(y) ___________________| |__ <<<<<< additional slides of dense text >>> [A sign-up sheet was passed ... Volunteers were requested to deal with the issues, which Steve does not see as being as difficult as the original suspend/resume effort. Colin observed that this must be done in the context of the normal start-up, and Steve agreed that he views this as a sub-group of the overall start-up team effort.] [There were questions about whether we had somem of the signals/control codes already over-loaded ...] [... If you are in Beta you stay in Beta; in DS you stay in DS ...] 6.3 Initialization of Beta mode communications - Colin Whitby-Strevens P1394b Start-up - Colin Whitby-Strevens - Scope: start-up procedure fro Beta-capable links - Arbitation state machine operation (bus reset, tree-id, self-id, and normal operation) will follow AFTER this start-up procedure - Recent developments - concept of a sub-PHY, to take care of long-haul - common coding scheme for all long .... Port Properties - 1 - a P1394b port a. may be capable of operating at any speed range from S100 upwards b. " " " " " " " S800 upwards c. shall be capable of operating in Beta mode d. when capable of operating at S100, shall be capable of DS at S100-S400, may be capable of Beta mode at S100-S400 e. may be brought out to a 1394-1995 connector ONLY when capable of operating in DS modes (S100/S200/S400) f. NB improved connector electrical properties are required for S800 g. Open issue - connector for S1600 and S3200 h. may be connected directly to a suitable transceiver for long-haul connection (thus also providing DC isolation) i. may use DC (e.g. capacitive or galvanic) isolation when operating in Beta mode - This is also the "short haul" port on a P1394b sub-PHY. Port Properties - 2 - The "long haul" port on a P1394b sub-PHY a. may be capable of operating at any single speed (s100-S3200) ??? b. may be capable of operating at any speed range from S100 upwards (i.e. S1 or S1+S2, or S1+S2+S4, etc.) c. may be capable of operating at any speed range from S800 upwards d. operates in Beta mode only e. is not required to interoperate with a 1394-1995 or p1394a port f. shall not be brought out to an interface using a 1394-1995 port g. connects through a suitable transceiver (this may be integrated), galavanic isolation and an RJ-45 connector for S100 UTP h. connects through a suitable tansceiver and, optionally, a suitable optical connector for optical fibre tansmission i. "undertands" the speed limitations of the connected long-haul transceiver by an implementation dependent mechanism SubPHY long-haul port - A sub-PHY long-haul port is essentially a stripped down 1394b port. it's a 1394b port ... .... [Did we ever resolve if a sub-PHY is a node? Yes, though we did not resolve whether a cable has one or two node IDs ...] Port types and properties 1. S100+ DS - a port as defined in 1394-1995/1394a 2. S100+ DS/Beta - as 1., but also capable of Beta mode 3. S100+ Beta - Beta mode only at S100+, i.e. a long-haul only port for slower speeds - Cannot be brought out to a 1394 connector 4. S100+ DS; S800+ Beta - as in 1., but also capable of operation in beta mode only at higher speeds (>=S800) 5. S100+ DS/Beta; S800+ Beta - as in 4., but also capable of beta mode at S100+ - this is the bells and whistles port 6. S800+ Beta - only operates at speeds of S800 and upwards, and does not suppport the 1394-1995 signaling interface (i.e. long-haul for higher speeds) - cannot be brought out to a 1394 connector. [Distinctions were drawn between various types of ports/PHYs, where the respective types may be wholly encapsulated, etc. ...] [Comment made regarding lack of single-speed devices in this list.] Interoperability table *** table *** S100+ S100+ S100+ S100+ DS S100+ DS/Beta S800+ DS DS/Beta Beta S800+ Beta S800+ Beta Beta yes yes no yes yes no DS DS DS DS yes yes yes yes no DS(?) Beta DS DS(?) yes no Beta General goals - Start-up in Beta mode if - both ends are Beta capable at S800+, or - at least one end is not DS capable, or - there's no DC connection - Corollary: start-up in DS mode if - ... [May wish to go to Beta to gain .b arbitration enhancements ...] Power-on state - Ports power on assuming they are disconnected and in suspend mode - every port is in disconnected state - Suspend/Resume low power connection detect mechanism transits the port state machine to Resume state - mechanism relies on a DC connection - need another mechanism for use in AC connections - resume state actions rely on TpBias - need another handsake mechanism for AC connections - Other port state transitions rely on TpBias and LpCon - above mechanisms need to substitute for tehse functions - Connction_status monitor relies on low poer connection detect mechanism and on TpBias output control flip-flop - need to avoid a port looking as if it is ... Start-up phases - determination of connection - Steve Bard presentation on suspend/resume - determination of beta or ds mode - ditto - determination of what speed to start beta mode (S800 or S100) - see later - bit synchronization - ... .... Speed determination - 1 - This is largely still open, and requires further study - requires ... .... Speed determination - 2 - Proceed through bit-sync, codeword sync and scrambler sync - The sequence S1 ... .... Speed determination - 3 - If the PHY is capable of a higher speed, it then transmits S3 repeatedly, otherwise it transmits S4 repeatedly - ... .... Await Connection - The Beta capable PHY ... .... Low Power connect detect - use a periodic burst of signals during suspend - ... .... Loss of Synchronization - This is a topic for further study 6.4 Standard protocol - Alistair Coles Alistair commented upon the nature of the selection of task group chairmen, the proceded with a discussion of how to structure the overall effort for protocols. *** hand-drawn diagram *** Tree ID, Self ID | 1394a Normal Arbitration | ------^---v------------------------------- p data arb states start-up o scrambling, coding | r speed signalling | domain of protocol group t error handling | padding lower data rates | ------^---v------------------------------- P M D .... Speed signaling - proposed on web site this week Error handling - how does a coding error (e.g. invalid 10 bit character is received) get communicated to the link layer? Timings - e.g. max delays DS <<--> Beta and Beta <<--> Beta Input to: Alistair Coles, 'anc@hplb.hpl.hp.com' Interfaces: 1394A? PMDs? [There will be interplay between protocols and the connectors ...] Speed Signaling *** hand-drawn diagram *** \ \ \ \ \ \ \ \ \ \ \ Data pref. speed = S400 \ data ....... \ \ \ \ DS v v v v ---- ------------------------------------------------------------------- Beta \ \ \ \ S800 \ \ \ \ \ data prefix \ \ \ Beta \ \ \<<- >= 140 ns -------->\ S800 v v v v ---- ------------------------------------------------------------------- Beta S400 Beta s400 ---- DS DS ---- - Speed signal length should be constant for a given packet speed, independent of actual port speed. - Need to maintain delta-T from speed signal to data, such that speed signal latency <<= data latency e.g. S200 packet over S800 port: 1 data byte per 4x10-bit characters such that speed signal <<= 4x10-bit characters Proposal *** hand-drawn diagram *** port TX speed S100 S200 S400 S800 S1600 ------------------------------------------------------------------------ S100 - S200 - Sa SRQP ---- Sa = SPEEDa = control symbol 0100 Sb = SPEEDb = contaol symbol 1000 - Use of speedb for S400 allows revision of S200 to S400, with rapid detection [Discusion of the meaning of this proposal ... 6.5 Accelerations - Mike Teener P1394b Arbitration Acceleration Michael D. Johas Teener Firefly, Inc. +1-408-461-4900 mike@fireflyinc.com Assumptions - Link continues to be half duplex - link cannot send packet (or ack) while receiving packet (or ack) - PHY is full duplex, but only for control - p1394b is different from P1394a, because the PHY is always full duplex, even when sending data .... PHY model *** diagram *** Outline Proposal - when link or child makes arb request, PHY passes request immediately up to parent - p1394a and 1394-1995 links make arb requests even while receiving a packet - P1394b PHY will propagate the request even while paket data is present - once a request is sent to the parent, all child/link requests are ignored until ... .... Request propagation - if data is coming from parent, request propagates on unused signal going back to parent - if data is going to parent, request is embedded within that data stream - requires that PHY elasticity buffer has extra capacity to allow insertion of extra symbols. Grant Propagation - if data is coming from child, grant propagates on unused signal going back to child - if data is going to child, request is embedded within that data stream - requires that PHY elasticity buffer have extra capacity to allow insertion of extra symbol - request and grant go in opposite directions, so only need a single extra symbol in elasticity buffer Error Conditions - lost request or grant - request times out, retry - timeout value same as subaction gac. - others?? Attached legacy devices - p1394b should be "root cloud" - child ports attached to p1394a nodes will send idle signals while S800+ - can accept request signal and forward request to root - wait to propate a "grant" to p1394a child until S800+ traffic is over - what is p1394a request timeout? Coding issues - request and grant codes can be identical - direction of signal determines meaning - must be neutral disparity - may be embedded within 8B/10B data traffic [There are tie-ins that await .a completion ... There will be no sign-up sheet, since there are such intimate tie-ins to Alistair's group's work. Much simulation will be required. The presentation will be on the group reflector tomorrow, revised back from 'idle' to 'tone'.] 6.6 Network simulation - Chen Simulation Setup, ??? Chen, Symbios Logic *** diagram *** S800 driver ---+--->80-pin PQFP model---->Termination---->pcb traces--- ^ ds xcvr -------+ --->berg connector model---->tensolite 1394 calbe model------berg cnctr-- --->pcb traces---->Termination---->80-pin PQFP model---+---||--S800 rcvr | +----DS xcvr *** wave form diagram *** *** wave form diagram *** *** wave form diagram *** *** wave form diagram *** Summary - System level simulation preformed with realistic input waveforms. - Everything semes to be OK at S1600 - 6.7 Cable/Connector - Max Bassler pIEEE 1394 connector WG - Charter to identify and document the copper interface (connector and cable) needed for 800-3200 Mb/s as defined as pIEEE 1394b. - Goal: develop a plug compatible copper interface to IEEE 1394-1995 (100-400 Mb/s) that will operate at one or all of the data rates of p1394b. Maintain backward compatibility to IEEE 1394-1995. - Need- Most current electrical criteria for pIEEE 1394b including realistic rise time and skew budget for cable. - Plan - Develop selection criteria for copper interface and post on reflector with 2 weeks for review - Solicit test data from all sources for the existing 6 ckt connector to review for higher speeds. - Setup off cycle meetings to develop standard to the planned schedule. - Have WG meetings concurrent with regular 1394b meetings [Volunteers were solicited. Bill Northey has volunteered to be the secretary for the group. Q: When do we pursue a connector differentiation due to cost deltas ... Can we or do we want to have a standard and a premium connector?] [What do we do about manufacturing standards? There are current cables, that are not so good ... How do we solve the problem for the user, such that plugging things together just works ... How do we indicate which cable to use ... These are open issues, and must be worked by the group as a whole and the task group ... Workmanship is a serious problem ... Do we want to have a new cable and connector scheme that is .b only? If we have a high-speed cable, how do we ensure that cable is what is used in that set of environments that it is needed? ... What about use of a high-speed cable versus a low-speed cable for connecting a slow device to a fast device? Fast devices will need to modify the launch slope to accomodate the slow-speed cable and device ...] [For the selection criteria, do you need to start with a problem definition? We know many of the problems ... We must be able to plug any cable into any device so that we can support A-B connectivity ... We may need some standard mechanism to let software know that the cable (or whatever) dictates that the node operate at a lower speed.] 6.7.1 Bill Northey, Berg P1394b TDR measurement *** Wave/signal forms *** Near End Cross Talk (measured) *** wave form *** [This is near end cross talk with a long cable ... How many pins are you signalling on? Within a pair ... Common mode cross talk ... Did you study differential cross talk? It's out for additional testing.] Far End Cross Talk (Measured) *** wave form *** 6.8 Standard electrical - Eric Hannah [The following is largely as presented at previous meetings, and the reader is referred to the group ftp repository, and to the minutes of previous meetings for further information. Changes to this version were limited, but the ftp respository should be checked for differences.] Connector TDR Data *** wave form *** *** wave form *** *** wave form *** *** wave form *** S Parameters Calculation *** wave forms *** S Parameters *** wave forms *** Back Calculation of TDR *** wave form *** Cable Model *** greek and mathematical symbols *** Experimental Cable Analysis *** wave forms *** .... [There are going to be problems with the present cable at higher data rates/frequencies. We also need to a better connector. The shielding is probably going to fail FCC requirements, as the shielding in the present cable is likely not adequate at higher (GHz+) signalling speeds.] 7. New Business No new business ... Or at least no-one willing to confess any ... 8. Meeting Schedule 8.1 October (Maui, 23 & 24, Aston Wailea Hotel) 8.2 December (Florida?) <<<<<<<< Secretaries Note, concerning the December meeting After the adjournment of the P1394b meeting, information finalizing a meeting in Fort Lauderdale. This meeting will be held at the Fort Luaderdale, FL, Marriot North, on Tuesday and Wednesday, December 2 and 3. Further information will be posted to the reflector, and to the 1394 TA's "Upcoming Events" page, as it becomes available. >>>> 8.3 January (Houston?) Avoid the TA meeting dates ... TA is in Menlo Park January 5? May not have much to report on that soon after Holidays and the preceeding meeting in early December. Will try for early 8.4 Others Chuck Brill offered St. Petersburg beach for Sept. and ??? Target is every month, except in November for Fall Comdex.`` 9. Review of Action Items All task group chairs - begin organization of their respective task groups, and start operations in earnest. Wooten - finalize arrangements for upcoming meetings in December and January All presenters - deliver suitable format digital copies of presentations to the ftp manager (Mike Teener) and the reflector. Wooten and Whitby-Strevens - Liaison letters. 10. Adjournment =============================================================================== Attendance List --------------- Khaled Amer Rockwell khaled.amer@nb.rockwell.com 714-221-3610 Tatsuya Arai Hirose tatsuyaa@hiroseusa.com 805-522-7958 Oleg Awsienko Intel oleg_awsienko@ccm.ch.intel.com 602-554-9666 Richard Baker Solectron Texas rtb@artb.itg.ti.com 512-425-6205 David M. Barnum Stewart Connector Systems dbarnum@stwconn.com 717-235-7512 Steve Bard Intel steve_bard@ccm.jf.intel.com 503-264-2923 Max Bassler Molex mbassler@molex.com 630-527-4490 Charles Brill AMP cebrill@amp.com 717-592-6198 Mike Brown Intel mike_brown@ccm.ch.intel.com 602-554-3713 Dave Brunker Molex dbrunker@molex.com 630-527-2622 Jim Busse NEC jimb@ccgate.sj.nec.molex 415-528-3810 Ed Butler Intel cbutler@sedona.intel.com 602-554-0751 Dao-Long Chen Symbios dao-long.chen@symbios.com 970-223-5100 x9461 Richard Churchill Compaq Computer Corp. richard.churchill@compaq.com (281)514-6984 Dan Colegrove IBM colegrov@us.ibm.com 408-256-1978 Alistair Coles HP anc@hplb.hpl.hp.com +44 117 922 8750 Jim Doyle Intel jdoyle@sedona.intel.com 602-554-2051 Jean Jacques Dumont FCI jdumont@iway.fr +33 1 39 49 2071 Stephen Finch Silicon Systems Inc. steve.finch@tus.ssi1.com 714-573-6808 Mike Fogg AMP mike.fogg@amp.com 717-986-5802 Bob Gannon C&M rgannon@cm-corp.com 860-779-4249 James Gay Motorola jimg@oakhill.sps.mot.com 512-891-2218 Eric Hannay Intel ehannah@mipos2.sc.intel.com 408-765-4441 Del Hanson HP del_hanson@hp.com 408-435-6246 Jerry Hauck Intel jerry_hauck@ccm.sc.intel.com 408-765-5528 Keith Heilmann IBM kheilmann@us.ibm.com 914-892-2413 John Hill AMP [no email address given] 717-592-6175 Jack Hollins Adaptec jack_hollins@eng.adaptec.com 408-957-2309 Derek Imschweiler Stewart Connector dimschweiler@stwconn.com 717-235-7512 David Instone Xyratex dinstone@uk.xyratex +44 1705 486 363 David Johnson Texas Instruments dkjohnson@ti.com 972-480-3632 Kevin Johnston Motorola kevinj@oakhill.ops.mot.com 512-891-3065 Al Kelley Tensolite akelley@tensolite.com 904-829-5600 x281 Dave LaFollette Intel dlafolle@mipos2.sc.intel.com 408-765-2587 Chong Lee Lucent chlee@lucent.com 610-712-3002 Fred Leung ALi fred_leung@acer.com 408-467-7449 Paul Levy VLSI Technologies paul.levy@vlsi.com 602-752-6382 Paul Lind Intel plind@sedona.intel.com 602-554-3774 Palanisamy Mohanraj Intel palanisamy_mohanraj@ccm.ch.intel.com 602-554-4243 Doug Moran Rockwell doug.moran@rss.rockwell.com 619-535-3306 Neil Morrow TI nmorrow@ti.com 972-480-6211 James Nave Texas Instruments jim_nave@ti.com 972-480-3468 Al Neves ICS neves@icst.com 408-925-9492 Bill Northey Berg Electronics northewa@bergelect.com 717-938-2119 Takayuki Nyu NEC new@optsys.cl.nec.co.jp +81 44 856 2082 Bill Prouty HP billp@hprnd.rose.hp.com 916-785-4631 Juan Pulido 3M jmpulido@mmm.com 512-984-5188 Brad Saunders Rockwell bradley.saunders@rss.rockwell.com 714-221-6513 Bob Scharf Stewart Connector rscharf@stwconn.com 717-235-7512 Tushar Shah Sun tushar.s.shah@east.sun.com 508-442-0468 Masood Shariff Lucent mshariff@lucent.com 732-957-5479 Robbie Shergill National Semi robbie.shergill@nsc.com 408-721-3002 Ron Soderstrom IBM rons@vnet.ibm.com 507-253-6290 Martin Sodos ISSI msodos@issiusa.com 408-969-4683 Michael Sorna IBM sorna@vnet.ibm.com 914-892-3425 Ju-Ching Tang Cirrus Logic jctang@corp.cirrus.com 510-623-8300 x5189 Ken Taylor Boston Optical Fiber ktaylor@bostonoptical.com 508-342-3309 Michael Johas Teener Firefly mike@fireflyinc.com 408-461-4901 Mike Wang Intel mike_wang@ccm.ch.intel.com 602-544-8555 Yasuhiro Watanabe Fujitsu ywata@flab.fujitsu.co.jp +81 44 754 2535 Donald Wiencek Panduit dsw@panduit.com 708-732-1800 Lee Wilson IBM leewils@us.ibm.com 512-838-6569 Colin Whitby-Strevens SGS Thomson colinws@bristol.st.com +44 1454 611 500 Phil Young NEC youngp@euk.nec.co.uk +44 1908 837 247 Patrick Yu NEC patrick_yu@el.nec.com 408-588-5436 Peng Zhang Texas Instruments pzhang@ti.com 972-480-3109 =============================================================================== The following is the list of those signing up for the "PHY/LINK" task group, during the September 24, 1997, P1394B meeting, in Natick, MA. Michael Johas Teener mike@fireflyinc.com David K. Johnson dkjohnson@ti.com Al Neves neves@lcst.com Colin Whitby-Strevens colinws@bristol.st.com Doug Moran doug.moran@rss.rockwell.com Cheng Lee chlee@lucent.com Eric Hannah ehannah@mipos2.sc.intel.com James Nave jim_nave@ti.com Keith Heilmann kheilman@us.ibm.com Michael Sorna sorna@vnet.ibm.com Dave Brunker dbrunker@molex.com Max Bassler mbassler@molex.com David Barnum dbarnum@stwconn.com Mike Brown mike_brown@ccm.intel.com Oleg Awsienko oleg_awsienko@ccm.ch.intel.com Jack Hollins jack_hollins@eng.adaptec.com James Gay jimg@oakhill.sps.mot.com Marin Sodos msodos@issiusa.com Ganesh Murthy ganesh_murthy@ccm.jf.intel.com =============================================================================== The following is the list of those signing up for the "Port Protocol" task group, during the September 24, 1997, P1394B meeting, in Natick, MA. Lee Wilson IBM leewils@us.ibm.com Michael Johas Teener Firefly mike@fireflyinc.com Dave Instone Xyratex dinstone@uk.xyratex.com Takayuki Nyu NEC new@optsys.cl.nec.co.jp Yasuhiro Watanabe Fujitsu ywata@flab.fujitsu.co.jp Palanisamy Mohanraj Intel palanisamy_mohanraj@ccm.ch.intel.com Oleg Awsienko Intel oleg_awsienko@ccm.ch.intel.com Dave LaFollette Intel dlafolle@mipos2.sc.intel.com James Nave TI Jim_Nave@ti.com Keith Heilman IBM kheilman@us.ibm.com Michael Sorna IBM sorna@vnet.ibm.com Paul Levy VLSI paul.levy@vlsi.com Jerry Hauck Intel jerry_hauck@ccm.sc.intel.com Alistair Coles HP anc@hplb.hpl.hp.com Colin Whitby-Strevens SGS Thomson colinws@bristol.st.com =============================================================================== The following is the list of those signing up for the "P1394b Start Up" task group, during the September 24, 1997 P1394B meeting, in Natick, MA. Colin Whitby-Strevens colinws@bristol.st.com Alistair Coles anc@hplb.hpl.hp.com Jerry Hauck jerry_hauck@ccm.sc.intel.com Dave LaFollette dlafolle@mipos2.sc.intel.com Oleg Awsienko oleg_awsienko@ccm.ch.intel.com Takayuki Nyu new@optsys.cl.nec.co.jp Yasuhiro Watanabe ywata@flab.fujitsu.co.jp David K. Johnson dkjohnson@ti.com Steve Bard steve_bard@ccm.jf.intel.com =============================================================================== The following is the list of those signing up for the Power management (Suspend/Resume) task group, during the September 24, 1997, P1394b meeting, in Natick, MA. Kevin Johnston Motorola 512-891-3065 kevinj@oakhill.sps.mot.com Jim Busse NEC 415-528-3810 jimb@ccbate.sj.nec.com David K. Johnson TI 972-480-3632 dkjohnson@ti.com Takayuki Nyu NEC +88-44-856-2082 new@optsys.cl.nec.co.jp Mike Brown Intel 602-554-3713 mike_brown@ccm.intel.com Dave LaFollette Intel 408-756-2587 dlafolle@mipos2.sc.intel.com James T. Doyle Intel 602-554-2051 jdoyle@sedona.intel.com Jerry Hauck Intel 408-765-2587 jerry_hauck@ccm.sc.intel.com Michael Sorna IBM 914-892-3425 sorna@vnet.ibm.com Cheng Lee Lucent 610-712-3002 chlee@lucent.com Colin Whitby-Strevens SGS-Thomson +44 1454 611500 colinws@bristol.st.com Alistair Coles HP +44 117 922 8750 anc@hplb.hpl.hp.com Bill Prouty HP 916-785-4631 bprouty@hp.com Stephen Finch Silicon Systems 714-573-6808 steve.finch@tus.ssil.com Steve Bard Intel 503-264-2923 steve_bard@ccm.jf.intel.com P1394b S800 Long-haul PMD Task Group meeting, Crowne Plaza Hotel, Natick Massachusetts, Tuesday, September 23, 1997 Chair: Colin Whitby-Strevens Secretary: Richard Churchill Reflector of Parent Working Group: p1394b@fireflyinc.com Web Site/Page of Parent Working Group: http://www.fireflyinc.com/p1394b/ This task group does not maintain a seperate reflector or web site/page. 1. Welcome and Introductions 2. Aproval of Agenda 3. Approval of Minutes of the last meeting (previously distributed) 4. Voting rules 5. Report from S100 PMD Task Group 6. Presetations in response for call for optical connector 6.1 Donald Wiencek - Panduit 6.2 Chuck Brill - AMP 6.3 Bob Scharf - Stewart 6.4 Juan Pulido - 3M 6.5 Masood Shariff - Lucent 6.6 Firooz Farhoomand - Matsushita 7. Optical transceiver parameters 7.1 Del Hanson - MP 8. Any other long-haul PMD issues 8.1 Keying 8.2 Review Criteria 9. Copper (short-haul) PMD issues (physical media -related issues, electrical specification and integrity) 10. future plans and schedule 11. AOB 12. Adjourn <<<<<<<< The meeting was convened at 9:00 AM. >>>> 1. Welcome and introductions were made, along with the status of this group as a study group of a study group, and not a meeting of the parent p1394b study group. 2. The agenda was approved without objection. 3. The minutes of the last meeting were approved without objection. 4. Voting Rules To this point, the rules have been, "If you are here, and you feel that you are competent to vote, then you can vote." This will not be the case in the future, as we must make certain "irreversible" decisions, and so the next meeting should be that you must have attended one of the previous two meetings, and be in attendance, in order to vote. Simple majority rules, with the chair (as per IEEE rules) retaining the right to declare groups as blocks. Mail ballots between meetings to be con- sidered as part of the immediate preceeding meeting. Physical presence precludes Proxies. Moved by Richard Churchill, and seconded by Steve Bard. Motion was carried unanimously. 5. Report from S100 PMD Task Group No parties were present who felt competent to present the minutes of, or report on the latest S100 PMD TG meeting. 6. Presetations in response for call for optical connector Objectives were ... - Low Cost (v important) second slide connector requirements . .... 6.1 Donald Wiencek - Panduit Fiber Jack Connector IEEE 1394b Fiber Jack TIA FOCIS-6 Connector for POF and GOF Fiber Jack and IEEE 1394b - Fiber Jack is the only TIA (FOCIS-6 adn 'Detail' specification) ..... Successful standards employ de facto standard products Fiber Jack = SC/ST + RJ45 Fiber Jack Meets 1394b Criteria Class Optical Fiber - ... TIA Specification Draft Fiber Jack Performance *** table *** [required for EIA/TIA ...] Fiber Jack Insertion Loss Multimode (FOTP-34 *** diagram *** ???? Highlights of GOF Fiber Jack - Safe (fragile fibers are protected by using proven ferrule technologies) - Mechanically robust for 20 year lifespan in the home - Small size, superior ergonomics, adapterless, flush mounted jacks (same as RJ45) - Field and factory terminable for both jacks and plugs with standard tools and cable - Fits ... - .... Fiber Jack Meets 1394b Criteria Plastic Optical Fiber - Cost is equal to or lower than RJ-45 - Termination time can be faster than RJ-45 - Already accepted by the ATM forum - 100 meter transmissions at > 100 Mbps [How is this terminated? STripping, cut to precise with hand tool, etc. For a loss of less than 2 dB you need to polish. No polishing means a loss of about 2 dB. Cost of plastic ferrules for smf connectors as opposed to ceramic MMF ferrules is about 1/10th. .... Are you claiming this for 100 meters? Yes, but still speced for 50 m ... There are challenges. Not claiming 100 m, but the technology is present ...] The ATM Forum Technical Committee .... IEC 1754-AA (NWIP) ... TIA/EIA ... *** diagram *** POF Fiber Jack Insertion Loss *** diagram *** Connector Requirements P1394b - Single Interface for All Media - Fiber Jack Specified for Both POF and GOF - Low Cost - POF Yes - GOF not yet - 20 Year Life - POF - FOTP Test - GOF - 20 plus years of history Connector Requirements P1394b - Suitable for Equipment, Wall Plates, etc. - e.g. Pan-Net - Suitable for Multi-mode Fiber - OPTI-JACK(tm) Fiber Optic Connector - Low Loss << 1.5 dB - GOF less than .75 dB - POF ... Connector Requirements P1394b - Pre-asssembled Patch Cords - Currently Available - Both Plug and Jack are Field Terminable - Assist Physical Safety - POF inherently safe - GOF safe due to standard SC/ST Ferrules [What core size for POF? ... 650 to 680 u in this case ...] Connector Specification 1394b - Easy Termination - POF Yes - GOF In development - Standards in Progress - TIA, IEC, ATM, IEEE, VESA HN - Small Size, Duplex, Asymmetric - RJ-45 Design [Are these crossed over? That is a requirement ...] [Is this open to be keyed? Yes. Targetted for home use? Yes ....] [Problem is when a building is wired during construction, the builder may have one objective and future owner different ones ...] Fiber Jack Connector *** diagram *** [Questions raised about PM connector, and whether this connector needed to presented again today, since it was presented previously. The answer is yes, it should be presented again today for formal consideration.] [Patents application[s] has/have been filed, with suitable letter regarding fare and reasonable terms, etc., has been filed with TIA(?)] [What about contamination? No shutter is present, but decided not to use one. Panduit offers shutters on the face-plate ... What about eye safety? Transceiver is responsible for the eye safety problem. How many insertion cycles? (500) Don't have the data handy. How does it degrade over time and insertion cycles?] [Light output from transceiver is supposed to be eye-safe. The burden should not be placed on the connector. Transceivers are Class 1, Eye- Safe to prevent accidental exposure to hazardous levels. The shutters are used to prevent contamination, not for safety purposes.] [Has anyone considered color coding? Yes, originally, but people end up wanting colors just for aesthetics ...] Colin offered the following additional requirements for consideration - Provide Letter on Patents to IEEE, copying P1394b - Provide .PDF on FTP site (send to Mike@fireflyinc.com) of the presentation. - Tooling estimated costs for GOF and POF, professional and Joe six-pack. - Matrix of what is available " " under development (with timeline) GOF and POF - Insertion cycles/Durability information - Do you meet 500 cycles? - Do you meet 1500 cycles? - Independent testing information (round robin) - Number of vendors providing connector? [Protest lodged that the presentation was not previously provided via the ftp site. There was a strong request that such presentations be previously posted, but this is a meeting to provide information, and not to make a final decision. Presentations should be posted to the ftp site within seven (7) days of this meeting, but at least sent to Mike Teener within seven (7) days.] [Questions raised regarding keying, with opinions raised on both sides.] [What would a polishing tool cost? Don't know. What about the tool costs relative to GOF and POF. Colin - We are not permitted to quote prices that are not out of a catalogue for a shipping product ...] [Do we wish to specify two levels of connector durability? 500 and 1500 insertion/removal cycles? Yes ...] 6.2 Chuck Brill - AMP IEEE 1394b Meeting in Natick, MA September 23, 1997 Charles E. Brill Global Product Standards, ... Plastic Optical Fiber Fiber Optic Features - EMI immunity - Wide bandwidth: BOF specifies stepped index POF as 300 Mb/s - ... Fiber Optic Features - msall size: <<= 2.2mm - light weight: 3.7 Kg/km (2.2m PMMA ...) - Easy light coupling - .... Plastic Fiber Applications Plastic Optical Fiber Inerfaces - FO7 - PN - modified FO7 - ... F07 and PN Interfaces *** picture *** F07 connector - product of AMP Harrisburg - Spring loaded metal ferrules - Floating ferrule design for vibration ... - .... PN - FO7 modified - Product of AMP Japan - Fixed Ferrule - ... ... DNP (Dry non Polish) - An AMP product for 20 years - 1000 micron plastic fiber - low cost - simple field assembly using internal retention sleeve - Connectors, couplins, active devices available - Cusom cable-assemblies Audio-DNP (FO5) - Used for audio application in Japan - 1000 micron plastic fiber - Compliance to EIAJ RC-5720 for digital audio equipment - Also complies to JIS FO5 and IEC ... Mini DNP *** image *** Mini DNP - 500, 750 and 1000 micron pOF (more size versatility than any in industry) - .... .... D2B - Product of AMP Germany - Audio applications - 1000 micron Plastic fiber - .... For more information: call the AMP Product Information Center at: 1(800)522-6752 .... [Is this hot-plate termination? Yes there is hot-plate termination available ...] [Colin - Your presentation did not address the requirements on a point by point basis ... Brill - There are no patents outstanding on this (FO7?) connector ... Graded Index POF will reach 3 Gb/s Colin added the following requests of the presenter ... - Usable with Glass? - Confirm no patents outstanding. - Would like to see the previous seven requests of Panduit addressed by AMP on the PN, and for the FO7 ... 6.3 Bob Scharf - Stewart The direct Optical Intercnnection system A proposal for long-haul p1394b applications .... [No letter to IEEE, as this is the first presentation of this product.] Presentation agenda .... P1394b connector requriements per 1394-sig/BOF Cost of ownership - ... .... Performance/Function - Duplex, Assymetric ... .... What is Direct? *** diagram *** DIRECT := Drop-In Replacement of Electromagnetic Component Technology Performance targets - Drop-in Optoelectronic component replacements for standard twisted pair interfaces - .... DIRECT Components *** diagrams *** - Direct PCM mounted Optoelectronic Jack activates optical power only if plug is inserted - Direct PCB mounted Transceiver modules interface optoelectronic drivers and receivers to +PECL - Direct Duplex Optical Cable is compatible with MMF (GOf, PCF or POF) Direct RJ-45 product implementation - Direct Optical Plug Latching and strain relief is common to RJ-45 cable plug - Direct PCB mounted .... - .... Advanced Performance Options - Support for battery powered devices *** diagram *** Integral Modular Plug Electrical Contacts Enable Sumultaneous Optical and Electrical Power and Electronic Signal Transmission [Demonstration of laser power switching ...] [Class 1 Eye-Safe requires safety for 10^4 seconds ... Hard to imagine anyone sitting still long enough to be a problem.] [Could, by various means, make scheme disable on disconnection at either end.] Field installation process -POF cables *** four images *** 1. strip ... 2. Insert ... 3. squeeze ... 4. remove and insert ... [2 dB insertion loss without polishing, 1.5 with ...] Additional DIRECT Products - Direct Passive couplers .. - .... *** two images *** Customer implementation options Cable structure - Plastic Optical Fiber (POF) - Plastic Clad Fiber (PCF) - Glass Optical Fiber (GOF) Eye Safety - Optomechanical shutoff - Electronic shutoff if no signal detect Data rates - cable structures - 10 MBd to 200 MBd (POF) up to 100 meters - 10 MBd to 2.5 GBd (GRINPOF) up to 100 meters - 10 MBd to 4.0 GBd (GOF) up to 300 meters [Additional information requested] Product Availability *** table *** Item Schedule ... Features - rj-45 footprint and plug interface - Excludes UTP, STP, ST, SC, FC and ATM - One tool Installation Procedure (POF) - Optical power shutoff if plug is removed - Low Profile Jack and Interface Modules Benefits - Total installed cost equal to CAT-5! - Eliminates Improper Mating Options - Suitable for amateur installations - Integral Eye Safety in Normal Use - PCB Design and Layout Flexibility Direct Summary - Low acquisition, Maintenance and Installation Costs - ... ... [What is the termination process for glass? Same as for any other glass fiber ... SC ...] [Can eliminate the PCB ... Must move to 9-pin foot-print to deal with connection detection ...] Colin requested futher information - Optical tranceiver Parameters, POF and GOF - Glass Termination - File Patent Letter - plus 3-7 of the standard seven previously discussed - Standardization status [Do you plan to take this to TIA, EIA, etc.? Yes, at the next meeting.] 6.4 Juan Pulido - 3M VF-45 Duplex Fiber Optic Interconnect Proposal IEEE P1394b Meeting, Natick, MA September 23, 24, 1997 3M Fiber-To-The-Desk System 3M Volition(tm) Cabling System Applications Driving Fiber Demand - Growth of powerful computer applications - Continued increase in computer processing power - bandwidth for future applications - EMI, RFI adn Cross talk on copper ... Benefits of Fiber Cabling Systems - Proven technology in back bones - Greater flexibility - future-proofed for any bandwidth - Marketable real estate ... - ... ... IT Trends - Computing power doubles approximately every 2 years - Projected ... - ... ... What path for the future? *** diagram *** One Cabling Solution!! Future Proofing Copper *** image *** Current Concerns Issues with current cabling products - Copper - lifetime costs - installation quality - bandwidth limitations - crosstalk, EMI - - ... ... 3M FTTD System Chanllenges - minimize system costs - minimize installation and required skill level - Meet Industry ... ... Volition(tm) Low cost, small form-factor optical interconnect proposal system solution *** diagram *** VF-45 Low Cost Duplex Fiber Optic connector - 1394b - fits 1394b criteria - low cost, simple installation, low loss (TIA compliant) - Suitable for Home wiring with 20 year lifetime - Standards Activity - ... ... VF-45 Low Cost Duplex Fiber Optic Connector - 1394b *** four (4) images *** [VF-45 is the name of the product ... Volition is now the name of the system. Samples are in hand today (at the meeting), with a hands-on demonstration at a subsequent meeting. Will do this at the Maui meeting next month. (action item)] VF-45 Low Cost Duplex Fiber Optic Connector - 1394b *** four (4) images *** VF-45 Low cost duplex fiber optic Connector - 1394b - advantages over standard SC: - significant impovement over SC - low cost (50% of SC) - Sub-assembly components - Installation labor - small size (50% of duplex SC) - High density footprint - duplex/polarized VF-45 Low Cost Duplex Fiber Optic Connector - 1394b - Performance - ISO 11801; TIA-568A - All source technologies - Dust and eye protection - Licensing per ANSI requirements VF-45 ... "performance requirements" *** table *** requirement ISO/IEC 86B TIA/EIA FO-6.3 test method ------------ -------------- ---------------- ----------------- [Tests were performed to pass the more difficult test ...] Current Connector Standard - .... Ceramic Ferrule X-Y-Z Fiber Alignment *** diagram *** Connector components - Core ... ... Single Ferrule / Duplex X-Y-Z Fiber Alignment *** diagram *** Concentric components - core to cladding - fiber to capillary - capillary to capillary - capillary to Ferrule OD - Ferrule OD to sleeve - Pin alignment - Length Components - end face polish - spring pressure [Tests to 1000 insertions have been performed ... Will present furthr material in Maui. VF-45 "Ferrule-less" V-Groove Technology *** diagram *** - Concentric Components - core to cladding - fiber OD to fiber OD - Length Components - minimal VF-45 Low cost ... "V-Groove Fiber Alignment" *** diagram *** VF-45 duplex fiber connector vs. duplex SC - reduced cost - A few ... - ... ... *** two (2) diagrams *** VF-45 Low cost ... - Proposed cost: - Elimination of ALL precision components (Ferrules/sleeves) - ... VF-45 ... - Ease of installation - duplex fiber installtion - perp adn terminate two fibers at one time - minimal tools required - ... - ... ... VF-45 Low Cost Duplex ... - Ease of installation and testing - glass fibe optic cable is robust - pull-strength rating: 160 pounds - TIA 568-A building cabling stnadard for UTP recommended at 25 pounds - Labor to install fiber optic cable and UTP ... ... Transceiver Package V-Groove fiber alignment - Cost reduced E/O device into single package - Improves density *** diagram *** VF-45 Low cost duplex fiber optic ... .... VF-45 low cost duplex fiber optic connector - 1394b - supports VCSEL Technology - higher transmission speeds ... ... VF-45 Low cost duplex fiber optic connector ... .... VF-45 Low cost ... - Availability - Over 14 alpha and beta trials TODAY - Production release January 1998 - Standards activity ongoing - networking - ATM ... ... ... Proposal for hte low cost, small form factor optical interconnect - insertion Loss and durability - ... *** diagram *** [Tooling cost is comparatively low ... Would like to see demonstration such as we have been promised for a long time ... Need to see that this is a reality. "Contractor" kit will be available for Maui, but not the "Home" installation kit.] Colin indicated additional need for futher test data results and PDF file for the group ftp site. Also, a new "standard item 8" was added, being "demonstration." 6.5 Masood Shariff - Lucent LC Connector A High Density Fiber Optic Network Solution LC Connector A High Density Fiber Optic Network Solution Yvonne Reeves Lucent technologies TIA 41.8.1 Working Group August, 1997 LC Connector Background - ... LC Connector Design Goals - Develop one connector/optical interface for all applications in the net - High volume = low overhead - Fewer equipment designs - lower complexity - less R&D expense for xcvr. manuf. and OEMs - Specialty connectors traditionally not successful - makes network more complex - low volume, not accepted by total market - ... (Patch-panel view) *** image *** LC Product Family Multimode and singlemode Initial product offering - patch cords and pigtails - duplex and simplex - MiniCord(tm) (1.6 mm) and 900 um - Hybrid Patch cords - Field-mountable on buffered-fiber and cordage - duplex adapters - standard and reduced height - Panels for outlets and cabinets - Mounting Collar for RJ-45 cutout LC connector Solution The best choice for all segments Multimode and Singlemode *** illustration *** - Campus backbone - ... LC Connector Features - RJ-45 Style - Small size - lower system overhead - polarized - color coded - pull-proof - cable compatible - excellent performance - high reliability LC Connector Feature advantages - RJ-45 Style - User friendly (intuitively obvious operation) - familiarity with oper modular plugs - no tools - easy insertion and extraction - audible click assures full insertion - low-complexity plastic housing - ... LC Connector Latch Enhancements - Patch cords - anti-snag feature on latch - easy ... ... LC Connector Feature advantages - small size - smaller than RJ-45 - ... ... (PHOTO of patch panel) LC connector Feature advantages - Polarized - ... LC Connector Feature advantages - Color Coded .... LC Connector Feature Advantages - Cable compatible - Compatible with _existing_ fiber and cable - can easily be used with the embedded base - 900 micron buffered fiber cables - 250 micron coated (ribbon, loose tube OSP) - standard ... LC Connector Feature Advantages - Excellent performance - Meets industry opt., mech. and env. standards - TIA 568-A - ISO 11801 - Belcore .... ... LC Connector Feature Advantages - High Reliability - .... ... LC Adapter Features and ... LC Connector MiniCordage - small diameter ... LC Standards, Licensing and availability - Standards - Actively working standards - TIA FO6.3 FOCIS work item approved - IEC 86b WG 6: submission approved by USTAG - Broadly .... LC Performance Optical performance, Multimode - insertion loss *** bar graph *** MM Insertion Loss Avg. = 0.10 dB Std. Dev. = 0.07 dB N = 100 Specification: ANSI/TIA/EIA-568-A Max. <<= ... ... Optical performance, multimode - return loss *** bar graph *** Optical performance, singlemode *** graph *** Environmental Performance, Multimode *** graph *** Mechanical performance, mating durability *** graph *** Mechanical Performance, Tensile Load - Singlemode *** graph *** Performance Specification, Multimode - Insertion loss - 0.10 dB .... LC Connector Summary - lower system complexity - RJ-45 housing - the most user friendly - 1/2 size of duplex MM SC - Proven, reliable tecnology - Compatible ... LC Connector Conclusion - bell labs invented the ST connector - Standard for 10base-F - Grandfathered in TIA 568-A and ISO 11801 - most popular optical connector in the world - bell labs also invented the modular plug - most prevalent copper connector worldwide - The LC combines the best of both - Proven technology of the ST - Miniaturized and embodied in a modular plug - Examine all facts from a total system perspective - the LC is the best choice [Questions raised regarding use of a smaller connector, and the acceptance of this across industries. Also, the question of where the electronics is to be place if the electronics need to be moved closer to the connector was raised.] Colin added the items below for further consideration ... - Clarify POF support - Opportunity(?) for tranceiver integration? - Standard eight additional items previous discussed. 6.6 Firooz Farhoomand - Matsushita Presentation was withdrawn 6.7 Colin's additions Requested information to the reflector and to the FTP site via P1394B@fireflyinc.com (reflector) mike@fireflyinc.com (for posting to the ftp site) Reflector material may be in text format, but the ftp material in .pdf format. Material must be available on Friday, October 10, to Mike and the reflector, so that Mike Teener can post to the FTP site by Monday, October 13. Demonstrations should be ready on Wednesday, October 22, at 6pm. Matrix and voting on Thursday, October 23rd. Colin will endeaver to prepare a preliminary matrix. Suggestion was made to defer the vote to the next meeting, but Colin indicated that the material has, for the most part, been available for some time. We should therefore be able to come to some conclusion, once the requested material has been digested. Still, it was observed that we are making a critical decision, which we may be forced to live or die by. Wooten suggested that one of the candidate solutions be "none of the above," so that if that solution wins we know we have more work to do. Colin - How long does this process go on for? This is the first meeting where we have the first representation of all the candidates, and it may be a mistake to base a decision upon such a small base. Contact information for the presenters: Fiber Jack Donald Wiencek DCW@Panduit.com FO7/PN Chuck Brill cebrill@amp.com Direct David Barnum dbarnum@stwconn.com - and - Karl Bergstrom mpeg007@aol.com VF-45 Juan Pulido jmpulido@mmm.com LC Masood Shariff mshariff@lucent.com Compromise proposal on decision is to perform an email vote approximately one month later. This places it in the time-scale of the December meeting, but allows us to use the December meeting to begin assembly of the a proto-standard document. There were questions raised regarding whether we can consider the licensing options as a part of the decision matrix. Observations pro and con were offered. IBM has in the past provided such data for standards groups, down to such details as "royalty free," and for a small one-time charge. We may not base requirements upon a specific set of terms, but we may be allowed to individually make decisions based upon our understanding of what constitutes a fare and reasonable set of terms. It was observed that Hyundai made a requirement that they could license an at-the-time-of-license un- known patent back from licensees of the 4B/5B patent. This was a clearly unacceptable condition in the eyes of many. This must be further clarified ... Is the time allotted sufficient for collecting necessary data for a good decision? The information for all these (save one) was in the hands of other standards bodies for some time, so the data is available publically, and its just a matter of gathering it. The initial understanding has always been that if the long-haul parts can be completed in the p1394b time-scale, it will be included in the 'b' standard. If it cannot, it may be split off in an alter- nate working group. How is this going to affect other standards groups? If we need a decision for other groups, we may need to make such a decision on a very timely basis to avoid being dropped as a portion of other, fixed time-scale, programs and standards. Further discussion revolved around these points ... Colin asked to have a show of hands of those who believed we should vote in Maui. No hands were raised. Hands raised for an email vote prior to the December meeting numbered 19. Hands raised for a vote at the December meeting numbered 25. 7. Optical transceiver parameters 7.1 Del Hanson - MP Proposed 1, 2 & 4 GBd 1394b Multimode Fiber Link Specifications & Issues IEEE 1394b meeting September 23-24, 1997 Outline - 1, 2 & 4 GBd MMF LINK Design Objectives - Analysis of Proposed 1 & 2 GBd Link Specifications - 4 GBd Extension Analysis & Issues - Conclusions 1394b Multimode Fiber (MMF) Link Specification Objectives - Lowest Possible Link Cost By Leveraging & Relaxing Gigabit Ethernet (GbE) specifications for shorter Link Lengths - Focus on 2 GBd Specifications Intially to Create A Single Link Specification Covering 1 & 2 GBd Operation - Define a future 4 GBd Link Specification Extension Which Interoperates with the 1 & 2 GBd Specifications - Achieve a minimum 50 m link length with both 62MMF and 50 MMF 1394b MMF Link Assumptions & Analysis - Lowest Cost Source Technology to achieve moderate Link length for the 1-4 GBd Range Utilizes 850 nm Vertical Cavity Surface Emitting Lasers (VCSEL) - At 850 nm (Based on GbE & TIA FO 2.2 developments) 50/125 um core/cladding diameter MMF (50 MMF) has a 500 MHz*km worst case Modal bandwidth (WCMB) with laser launch vs. 62MMF having WCMB=160 MHz*km - The spread sheet link analysis model adopted by GbE provides the basis for defining optical power penalties and link lengths 2 GBd MMF Link Specification Process - Leverage GbE MMF link specifications in which the laswer 9dB Extinction Ratio (ER) Penalty of 1.1 dB is built into the -17 dBm Receiver sensitivity specification for Operation at 1.25 GBd - Maintain laser output power range of -4 (0 future?) to -10 dBm to ease future eye safty & laser driver inplementation - Reduce Receiver sensitivity by 2 dB @ 1.6x larger bandwidth - Reduce Budget to 6 dB vs. GbE's 8 dB (7 dB + ER of 1.1 dB) - Relax ER from 9 dB to 7.5 dB (Increases Link Penalty = 1.56 dB) - Show ER as an Explicit link penalty to facilitate trade-offs 2 gBd 1394b 62 MMF Link with 850 nm VCSEL leveraged from GbE *** graph *** 1394b 62MMF Link Operated at 1 GbD *** graph *** 2 GBd 1394b 50MMF Link With 850 nm VCSEL Leveraged from GbE *** graph *** 1304b 50MMF link operated at 1 GBd *** graph *** For 50 mmf only, 50 m @ 2 GBd with relaxed specs (Uw=3.0nm, ER=7 dB) *** graph *** 1 & 2 GBd MMF Link Length Summary - Power budget Assumptions - Transmitter Range: -4 (0 Future?) to -10 dBm - Receiver Range: 0 to -16 dBm (center-of-eye) - 62MMF Limitation of 50 m @ 2 GBd (ER=7.5 dB, Uw=0.85 nm) - 62MMF: 50 m @ 2 GBd, 245 m @ 1 GBd - 50MMF: 90 m @ 2 GBd, 460 m @ 1 GBd - For 50MMF Only: Limitation 50 m @ 2 GBd with Relaxed Specs - ER=7 dB, Uw=3.0 nm 155 MBd, 50 m POF (PMMA), worst case 650 nm, ATM Forum Specification summary ----------- - 50 m POF PMMA, attenuation @650 nm = 8.0 dB - Temp (-20/+70C), 95% RW, Bends = 1.1 dB - 640 nm to 660 nm spectral range = 3.4 dB Total POF Cable Atten = 13.5 dB - Loss of 2 connectors = 3.5 dB Power Budget = 17.0 dB ----------- Source range (@T.,. - 4.5 us) : -2 to -7 dBm receiver range : -2 to -25 dBm - - - - - - Worst case rec. dyn. range: 23 dB Actual Rec. Mfg. Dyn. Range: 26 dB ----------- Rec. Sensitivity drops 3 to 4 dB for doubled data rate @ 1 GBd, detector capacitance must be = 0.75 pf (~100 um D??) 4 GBd Specifications Targeted to be compatible with 1 & 2 GBd Links - Reduce laser response times (Tr,f) by 50%, but maintain transmitter output range: -4 (0 Future?) to -10 dBm - Relax ER to support reduced Tr,f; must Increase Power Budget to Compensate - In RIN Penalty Calculation, remove laser response time filtering (Utilize fiber + receiver bandwidth filtering only) - must improve RIN specification from -117 dB/Hz to -120 dB/Hz to achieve 50 m link with 50MMF (Due to reduced RIN filtering) 4 GBd 1394b 50 MMF link with power budget = 6 dB, ER = 7.5 dB, Tr,f=0.13ns *** graph *** 4 GBd 1394b 50 MMF link with power budget = 7 dB, ER=5.5 dB, Tr,f=0.13ns *** graph *** 4 GBd 50 m Link specification issues (relative to 1 & 2 GBd proposal) - a 50 m link length is achievable only with 50 mmf - with the same technology, increasing receiver bandwidth by 2X reduces receiver sensitivity by 4 dB - If ER=7.5 dB, a 6 dB power budget requires a 4 dB improvement in receiver sensitivity relative to 2 GBd; Requires a GaAs detector and preamplifier - If ER must be relaxed to 5.5 dB, the power budget must be increased to 7 dB; requiring a 5 dB improvement in receiver sensitivity which may not be achievable with a single 3.3 V Supply Conclusions - 1 & 2 GBd 1394b mmf link specifications - based on GbE, 50 m link length specifications are readily achievable with either 50 MMF or 62 MMF - some open issues: mmf fiber type(s), optical connector, electrical I/O interface definition & transceiver footprint - Some Open Issues for Compatible 4 GBd Link Specification: - How much must the laser ER be relaxed? - Can the necessary receiver sensitivity and dynamic range be achieved in a cost effective manner? - Can a single 3.3V power supply be utilized? (numbers used in analysis) *** table *** (numbers used in analysis) *** table *** [Most manufacturers would have to redo receivers to deal with the lower speeds ... ISI declines at lower speeds ... Full range from S100 to S1600 can be done on 50 MMF glass ... ] [Should we go forward with these numbers for a draft of the spec.? Del - If I were doing the spec., I would do it that way. Colin - I propose that we go forward with your numbers. (Agreed without objection.) For the slower speeds, largely the same numbers would apply. Do we go forward with two specs. - one fast and one slow - or with a unified one for fast and slow? Fuller - Need more info. since we don't know what the cost advantages/liabilities are. Do we add a tax to higher speeds in order to support slower speeds that might be unreasonable? Might it be cheaper to just go with a high-speed solution? Must target spec. at higher speeds, and ... How low does it make sense for us to specify the fiber oper- ation? Del - it would be an advantage to start at 400 because of noise, but I'd want to take a look at it ... Is there a cost knee at which we are better off stopping? Silicon is good to 1.25 GBd, but that's about it. Faster will require GaAs receivers, etc. ... Dispute over knee due to switching from LED to VCSEL at 125 or 250 Mb/s ... Solution must be interoperable at all speeds supported. Making fast parts support slower speeds entails cost ... We may be better off specifying that we start at S800, and not worry about speeds lower than that. This may be a factor of two receiver cost. factor narrows as you move the lower bound upward. ... Power is a hidden cost for solutions ... We need to avoid making portable solutions unacceptable ... How far down can you go for a part designed to support 2 Gb standard? Del - GbE may be able to go as low as 500 Mb/s ...] [Del Hanson accepted action item to look at what is requied to move down to slower speeds ...] [Need an array of relative costs, top versus bottom, for full range of options ...] Concensus that we will stick to our established range of S800+, and not worry about slower speeds ... 8. Any other long-haul PMD issues 8.1 Keying Make this an item to be listed among features of the connector contenders. <<<<<<<< The meeting was adjourned before completion of all items. >>>> <<<<<<<< The meeting adjourned at approximately 5:50 PM. >>>> ============================================================================= Attendance List --------------- Khaled Amer Rockwell 714-221-3610 khaled.amer@nb.rockwell.com Tatsuya Arai Hirose 805-522-7958 tatsuyaa@hiroseusa.com Oleg Awsienko Intel 602-554-9666 oleg_Awsienko@ccm.ch.intel.com Richard Baker Solectron Texas 512-425-6205 rtb@artb.itg.ti.com Steve Bard Intel 503-264-2923 steve_bard@ccm.jf.intel.com David M. Barnum Stewart Connector 717-235-7512 dbarnum@stwconn.com Max Bassler Molex 630-527-4490 mbassler@molex.com Charles Brill AMP 717-592-6198 cebrill@amp.com Mike Brown Intel 602-554-3713 mike_brown@ccm.intel.com Karl Burgstrom Thomas & Betts 800-888-0211 x4563 mpeg007@aol.com Jim Busse NEC 415-528-3810 jimb@ccgate.sj.nec.com Ed Butler Intel 602-554-0751 ebutler@sedona.intel.com Robert Campbell Lucent 732-957-2669 rrcampbell@lucent.com Dao-Long Chen Symbios 970-223-5100 x9461 dao-long.chen@symbios.com Dan Colegrove IBM 408-256-1978 colegrov@us.ibm.com Alistair Coles HP +44 117 922 8750 anc@hplb.hpl.hp.com James T. Doyle Intel 602-554-2051 jdoyle@sedona.intel.com Jean-Jacques Dumont FCI +011 331 3949 2071 jdumont@iway.fr Firooz Farhoomand Netcom 408-653-4059 firoozf@ix.netcom.com Michael W. Fogg AMP 717-986-5802 mike.fogg@amp.com Nobuo Furuya NEC 408-588-5414 nfuruya@el.nec.com Bob Gannon C&M Corp 860-779-4249 rgannon@cm-corp.com James Gay Motorola 512-891-2218 jimg@oakhill.sps.mot.com Eric Hannah Intel 408-765-4441 ehannah@mipos2.sc.intel.com Del Hanson HP 408-435-6246 del_hanson@hp.com Jerry Hauck Intel 408-765-5528 jerry_hauck@ccm.sc.intel.com Jack Hollins Adaptec 408-957-2309 jack_hollins@eng.adaptec.com Derek Imschweiler Stewart Connector 717-235-7512 dimschweiler@stwconn.com David Instone Xyratex +44 1705 486 363 dinstone@uk.xyratex.com Kevin Johnston Motorola 512-891-3065 kevinj@oakhill.sps.mot.com Al Kelley Tensolite 904-829-5600 x281 akelley@tensolite.com Diana Klashman Sun 9788-442-0575 klashman@east.sun.com Dave LaFollette Intel 408-765-2587 dlafolle@mipos2.sc.intel.com Chong Lee Lucent 610-712-3002 chlee@lucent.com Rick Lehtinen ICS 602-944-1491 rick@insightlearning.com Fred Leung ALI 408-467-7449 fred_leung@acer.com Paul S. Levy VLSI Technology 602-752-6382 paul.levy@vlsi.com Paul Lind Intel 602-554-3774 plind@sedona.intel.com Kiyoshi Miura Sony +81 462 30 6209 miura@lsis.crl.sony.co.jp Palanisamy Mohanraj Intel 602-554-4243 palanisamy_mohanraj@ccm.intel.com James Nave Texas Instruments 972-480-3468 jim_nave@ti.com Al Neves ICS 408-925-9492 neves@icst.com Bill Northey Berg Electronics 717-938-2119 northewa@bergelect.com Takayuki Nyu NEC +1 81 44 856 2082 new@optsys.cl.nec.co.jp Bill Prouty HP 916-785-4631 bprouty@hp.com Juan Pulido 3M 512-984-5188 jmpulido@mmm.com Bob Scharf Stewart Connector Sys. 17-235-7512 dbarnum@stwconn.com Masood Shariff Lucent 732-908-5479 mshariff@lucent.com Robbie Shergill National 408-721-7959 robbie.shergill@nsc.com Michael Shinkarovsky Lucent 610-712-2938 mshinkarovsky@lucent.com Jim Skidmore TI 972-480-2094 j-skidmore@ti.com Ron Soderstrom IBM 507-253-6290 rons@vnet.ibm.com Martin Sodos ISSI 408-461-4901 mike@fireflyinc.com Ju-Ching Tang Cirrus Logic 510-628-8300 jctang@corp.cirrus.com Ken Taylor BOF 508-347-3309 ktaylor@bostonoptical.com Michael D. Johas Teener Firefly 408-461-4901 mike@fireflyinc.com Peter Teng NEC 408-588-5555 pteng@mail.com Mike Wang Intel 602-554-8555 mike_wang@ccm.intel.com Yasuhiro Watanabe Fujitsu +81 44 754 2535 ywata@flab.fujitsu.co.jp Colin Whitby-Strevens SGS Thomson +44 1454 611 500 colinws@bristol.st.com Don Wiencek Panduit 708-532-1800 ocw@panduit.com Lee Wilson IBM 512-838-6569 leesils@us.ibm.com David Wooten Compaq 281-518-7231 david.wooten@compaq.com Phil Young NEC +44 1908 847 247 youngp@euk.nec.co.uk Patrick Yu NEC 408-588-5436 patrick_yu@el.nec.com Peng Zhang TI 972-480-3109 pzhang@ti.com