IEEE P1355 Working Group Fourth working group meeting 19th October 1993 Minutes The fourth meeting of the IEEE P1355 Working Group was held at Danvers, Boston, MA, in conjunction with IEEE BASW, 08.30 - 17.00 on 19th October 1993 Present Andrew Cofler Bull SA (Editor) Harrison Beasley Texas Instruments Richard Carlson Harting ElektroniK Ed Jacques John Hopkins University Ernst Kristiansen SINTEF Roland Marbot Bull SA Ann O'Neill IEEE Sylvain Paineau Thomson-CSF/LCR Elwood Parsons AMP Inc Colin Whitby-Strevens INMOS Limited (Chairman) 1 Welcome, introductions, approval of agenda CWS welcomed everyone to the meeting. The agenda was proposed and approved. 2 Approval of minutes of 1st September meeting The minutes, previously distributed, were approved as a correct record. Ann O'Neil was invited to join the meeting as a result of preliminary discussions on item 3.1, and provided valuable advice which enabled discussions on 3.1 and 4 to be completed. 3 Chairman's report 3.1 Promotion as an ISO standard CWS circulated a letter he had received from Bob Pritchard, chair of ISO/IEC JTC1/SC26, enquiring whether the group is interested in promoting the standard as an ISO standard, and suggesting two routes for achieving this. Ann O'Neil stated that the IEEE is strongly in favour of standards being promoted in this way. It was noted that combining the ISO NP and IEEE WG activities might result in a slower balloting process (ISO balloting operates on a six month cycle). After discussion of a variety of issues, it was agreed to explore with ISO the amount of effort involved (ACTION CWS to contact Bob Pritchard), and for the various members to contact their national representatives, with a view to proposing a NP (New Work Item Proposal) to SC26 at their November meeting. ACTION ALL. 3.2 Promotion of the WG and standard Discussion of ISO promotion led to wider discussion of the promotion issues. It was agreed that we should seek to maximise the interest and awareness of the activity, and to this end, WG members are encouraged to distribute copies of the current draft to any people and organisations they felt would be interested. They should also pass contact details of interested parties to CWS, so that contact may be maintained. 4 Review of action items from previous meeting ACTION ALL: send to CWS copies of Press Articles, particularly those appearing in national press. ACTION CWS: distribute the collated press articles with the proposed distribution of the next draft. Actions complete ACTION CWS: Check whether it is agreeable from the IEEE point of view to put product information on the ftp server. This was discussed in the meeting, and advice taken from Ann O'Neil. She advised that this is a matter for the working group. Some groups felt it important to separate commercial developments from standards issues, in order to prevent meetings from turning into trade shows. She advised that a disclaimer should be included that no endorsement or statement of quality was given or implied. It was agreed that data sheets etc could be posted on the ftp server with such a disclaimer. WG participants should send any such information to CWS for posting. Action closed ACTION CWS: Explore whether references to trade-marked names may be included within the standard document. This was discussed in the meeting, and advice taken from Ann O'Neil. She stated that in general, trade mark names should not be included in a standards document (there are exceptions, such as directories/dictionaries). It was agreed that the new terminology for the various link technologies made it possible to avoid the trade-marked term `DS-Link'. Andrew Cofler would make the necessary editorial changes. Action closed ACTION ALL. Provide working papers on the relative difficulties of engineering DS-SE links using 50 or 100 Ohms connections. CWS reported that no comments had been received. The draft standard has been written to specify 100 Ohms. Comments on this issue may now be made in the normal way. Action closed 5 Review of draft standard CWS thanked Andrew Cofler, together with all the various contributors, for the hard work in producing an excellent first draft. The document was reviewed in its entirety, and numerous minor changes proposed, which will be incorporated into the next draft. Document organisation RM noted that the copies of the draft which were made available in the registration office were disappearing rapidly. He proposed that the document would be more friendly to the casual reader by including the rationale in the introduction, rather than as Annex I. It was agreed that this should be done, at least until the `final' draft. With this change, it was agreed that the document organisation is OK. Glossary AC has listed many terms which need defining. It was proposed that reference should be made to the IEEE Standard Dictionary to assist in completing these definitions. For more specialised terms, reference should be made to the Fiber Channel, Serial Bus and SCI standards. WG members are requested to provide definitions of the specialised terms which are in their specific areas of expertise. Protocol stack It was agreed that this section should be expanded, particularly at the exchange level, where the generic concepts of flow control would be described (alleviating the need for repetition later in the document). Further diagrams would also be added to assist the introduction of the protocol stack and the role of the various layers (chip to chip, or end to end). A table should be added to give the main properties of each implementation (eg maximum distance, baud rate, byte rate). An informative annex should address implementation recommendations for each technology. The maximum speed for DS-XX links was discussed. Whilst the links can autobaud, there are issues for connectors and cables. In addition, HS links can autobaud (albeit in a much smaller speed range). It was proposed that the meaning of the digits should be the maximum recommended operating speed, and that, for this purpose, DS-SE and DS-DE links should be designated with speed 02. DS-SE links Several details are still to be provided (CWS to seek proposals from Dave Cormie). Two diagrams need updating. DS-DE Figure 6 needs correcting and moving to the Informative annex. Rich Carlson showed the latest Harting connectors. TS-FO SP presented the range of options for TS-FO optical connectors. These include multiple fibre technologies such as MAC from AT&T, and various duplex technologies under development. CWS reported a conversation with Gene Schramm from AT&T, in which he indicated that he would be surprised if a reduced form of MAC proved the best option for duplex connectors. Both AMP and Harting indicated that they would wish to make proposals at the December meeting. It was agreed that the emphasis should be on specifying a duplex connector. We should await the proposals at the December meeting, meanwhile this is left as an open issue. It was also agreed that the standard should specify a panel connector and a matching cable connector. The issues of connection from the diodes/lasers on board to the panel (ie on- board engineering of TS links) should not be standardised, but recommendations given in an informative annex. SP agreed to draft such an annex in due course. The cable requirements should be retained in the main document, but detailed recommendations relegated to an informative annex (in a manner analogous to the division of information for DS-DE cable). It was agreed that there should be an overall power budget, will allocations to on-board, connectors and fibre. The power budget allocated to the fibre would then determine the maximum fibre length (which would be different for the two types of fibre available). It was noted that similar issues have been addressed in depth in Fibre Channel, and reference should be made to that standard. It was agreed that the signal identities given in table seven should be re-allocated to the connections provided in the fibre connectors, in which case the TS_FO_in would correspond to DS_SE_data_in and DS_SE_strobe_in, and similarly for the `_out' signals. HS-SE Harting made a presentation on a proposal for a connector. This uses the same form factor as the DS-DE connector, but provides two co-axial connections, and uses thumb-screw fixing (rather than latching) for additional security. These proposals are being evaluated by Bull, who will test the part, and feed back, particularly addressing issues of robustness and reliability. It was agreed that these proposals formed a good basis for the standard specification. The issues of cables (the current text implies multiway cables) would be revised by AC. It was also discussed whether there should be a HS-DE specification. RM pointed out that currently there are a number of both technical and commercial difficulties (the lack of suitable integrated differential amplifiers, lack of suitable differential cables, and doubt as to whether a sufficiently significant additional benefit could be gained). It was agreed to include a small rationale for the decision to concentrate on HS-SE at the start of the HS-SE section. Various technical details will be completed after evaluation of the Bullit chip (currently in manufacturing). The issue of BER was discussed. It is an open issue as to what statements (if any) should be made about BER. It was noted that other standards include eye-diagrams, from which BERs can be deduced. However, a basic premise for the HIC technology is that the signal/noise ratio is very good, and that the BERs are consequently very low. It is not at all clear whether eye- diagrams would therefore be useful. What use is made of eye diagrams by implementers? CWS agreed to incorporate something in the rationale on the reliability issues. However, it remains an open issue of what should be specified. The exchange level has been simulated extensively, and implemented in the Bullit chip. This section will be revised (if necessary) in the light of experience with the Bullit chip. HS-FO The issue of connectors remains open, and will be addressed in the December meeting. Issues of power budget will also be addresssed, in a manner similar to TS-FO, following decisions on connectors. Common packet level The issues of which packet formats should be specified, mandated or be optional was discussed, and is identified as an open issue. It was agreed that expert advice on the support which should be incorporated in the standard for multicast should be solicited from experts in ATM, Fibre Channel, SCI, and OMI (for support for combining networks). 6 Next meeting and draft It was agreed that the next meeting should concentrate on the open issues identified above. The current draft should be circulated widely for comment and feed-back, and the next draft prepared after the December meeting. 7 Future meetings It was agreed that future meetings would be December 1st - Bristol March 1st - Paris Week of 11th April - Munich (with BASW) Depending on interest, feedback, and possible participation of members in ATM Platform, a mini-meeting and/or tutorial may be held at Salt Lake City in the week of January 10th with BASW. (Note - this has been scheduled for the evening of Tuesday January 11th). CWS noted that the OMI/HIC participants are required to prepare a plan for participation in future US meetings for approval by the CEC. C Whitby-Strevens - Chairman 20th October 1993