IEEE P1394b Working Group Meeting,
Marriott Hotel,
Santa Clara, CA,
Monday, April 7, 1997,
8:30 - food
9:00 - Start of Session

Chair:           Michael D. Johas-Teener
Editor:
        Eric Cabot Hannah
Secretary:
  Richard Lloyd Churchill

Agenda
  Administrivia
      Introductions
      Next Meeting, 5/5/97, Santa Clara
      Par Status
  Draft Standard version 0.01 (Eric Hannah)
  Simulation Results
      IBM
      Berg
  Protocol Group Volunteers
      Schedule, Rev A - June 97
  P1394a requirements

  1. Administrivia

  Introductions around ...

  Next meeting will be May 5, in Santa Clara, since a site couldn't be found
  in Mass. We will attempt to colocate with 1394a group. There is a meeting
  planned for Honolulu in August.

  PAR is in process, and should see action in the next couple of months.

  Lack of progress on protocol noted, with a need to correct the situation.
  A small sub-group needs to be established to start work on this, which
  involves start-up, initialization, etc. ... Mike Sorna, Eric Hannah,
  Colin Whitby-Strevens, David Wooten, Mine Teener, Steve Bard are the
  "volunteers."   This will cover a wide range of duties regarding interplay
  between b and older versions of 1394, arbitration, 8B/10B encoding and
  behavior, etc. Need to delineate, and parcel out tasks.

  Patrick Yu will seek closure on IBM's licensing of the proposed 8B/10B
  encoding scheme.

  Fiber connectivity is a requirement, and we must address it. This will
  require pushing some additional burden on the P1394a group, with "ping"
  being highly desired and in need of more work. There are questions about
  jitter budgets, etc. .... What is the model of the connecitivity?

      (How can TA drive a fiber standard? -- Can serve as a discussion group
      for getting things started. Must ensure that there is only one model
      of fiber connectivity when all is said and done.)

  Connection Model -- How do people this working? What will it be used for?

  1. Draft Standard, version 0.01

  Pages up to 31 are boiler-plate, with much left to be filled in. Content
  starts at page 31.


  Editor's opinion is that the tolerances on impedence must be more tightly
  constrained.

  Signal attenuation is also critical, according to editor.

  Propagation skews, relative between pairs, needs to be examined. Fiber
  Channel calls for < 5% (50 ps) based upon limit attainable by cable vendors
  at the time. Cables may be as low as 5 ps -- at a high price. This is
  linked to signal rise times. (This is differential skew.)   Karl Nakamura
  will assist ... (action item -- he needs to write down some numbers so we
  are at the same starting point.)

  (Karl has additional action item to talk to cable folks about skews, etc.)

  Questions raised regarding the need for a 2 V differential swing, as we are
  not seeking cable lengths comparable to Fiber Channel. We can likely use
  far less. (FC may be going to lower swings at present lengths ...)   Can
  likely live with 1.1 V, perhaps lower (800 mV). Also, what voltage, etc.,
  are required to directly drive a VCEL. We are only trying to solve the
  problem for 5m copper and shorter.

  Deterministic and random jitters are "tentative", and may be looser. The
  trade-offs will need to be examined.

  There may be problems in 1999/2000 time frame producing S1600-bilingual
  PHYs. There may be several means to accomplish "comparable" functionality,
  but the quantity of work that has gone into this so far is not great. There
  was some tentative discussion of how such could be accomplished.

  Receiver sensitivity of 300 mV recommended by Karl Nakamura.

  Table 5-3, through connection values, S800 through S3200, need to change,
  also the operating distance limited to 5m, not 25m. All charts need to be
  made consistent.

  Clause 6 is based upon IBM 8B/10B scheme, which is patented. Bit ordering
  is an issue. (Lines 61 and 62, page 41, are internally inconsistent.)   We
  may just use the same ordering as FC and Gbit Ethernet. (There was a lot
  of discussion regarding bit transmission ordering, with much confusion as
  to who was asserting what. We need to ensure that we end up consistent on
  transmission ordering. P1394.2 mapping is probably the same as FC, so if
  we are consistent with them we should have well-tested cells already.)

  Jerry Hauck has action item regarding 8B/10B coding, CRCs and bit/burst
  error detection.

  In clause 6.1.6, information relative to FC and Gb Ethernet ...

  Clause 7 is very tentative, and will be extensively revised by the protocol
  sub-group. Present clause is based upon the Fiber Channel mechanisms, and
  was originally written by Dave LaFollette.

  1. Simulation results
  2. Berg - Bill Northey

      Task -- redesign -1995 I/O system to --
          allow data rates of 4 Gb/s,
          maintain backward compatibility.

      Produced complex simulation schematic.

      Believe existing cable good to 3 Gb/s (7.5 GHz).

      Geometry changes to connector in cable, receptable, gets existing cable
      to 4.5 Gb/s (approximate bandwidth = 12 GHz).

      Plan preliminary design by end of April '97


              Spice files available in May '97
              Initial Samples in July '97
              Production in October '97

      -- All this is expected to be plug-compatible with the existing standard
      -- and devices.

      [simulated eye diagrams were requested and tentatively promised.]

  1. IBM presenter was a no-show

[Comment -- Colin Whitby-Strevens -- We need to be aware of the pending
closure of the P1394a standard to new proposals. We need to make what
suggestions and requests need to be dealt with very soon.]

[Recess for Lunch]
[Resume at 1:00PM]

  1. 1394a Requirements

      -- Phy delay: 11 given meaning -- indicates delay greater than covered
          by spec. (144 us + 5m line delay), use pinging to determine delay

      -- Pinging is needed
          -- better description of first-bit delay (jitter vs. propagation)
          -- bound min/max of response time
                (should be very tight. Jerry Hauck prefers to have timeer in link.)

      -- Power management proposals: Please consider fiber cases.
          -- Fiber issues? 1394a doesn't need to deal with this, but should be
              extensible.
          -- prefer to have power mgt consider a polled mode
          -- issue of whether a PHY is required to operate at all speeds slower
              than it's highest.

  1. Protocols to be defined

The Group
(Peter Johansson will have to be involved to some extent.)
IBM Keith H.
Intel   Eric Hannah
SGS Colin Whitby-Strevens
CPQ David Wooten
ff Mike D. Johas-Teener - facilitator
Intel   Steve Bard
NEC N. Furuya (volunteered by Patrick Yu)
LSI Karl Nakamura

  Connection Initialization (Start in DS then move to "beta", or start in
            beta)
      Speed resolution
      Shutdown protocol, with power-up procedure
  Mapping of 1394a arbitration onto 1394b (codes, etc.)
      maybe better use of -- 1394a improved over b
          (token arbitration improvements on 1394b Beta )
      phy-link I/F for higher speeds
  Data encoding and Signalling for various speeds (mapping old speeds, etc.)
  Self-ID changes for additional speed codes, maps, etc.
  topology updates
      gap calculations
  testability
      loop-back enable
      self-checking

  Assignments:
      Testability - Patrick Yu (NEC)
      Connection Init - Steve Bard
      Mapping 1394a arbitration - Karl Nakamura


      PHY/LINK I/F - Robbie Shergill
      Data Encoding - Keith H.
      topology updates - Eric Hannah and Jerry Hauck

  1. Next meeting expectations -- Future Meetings, etc.

  Berg connector update
  progress reports from all action items

  June 9 meeting in Seattle
  (July 7 in Maui? -- possibly in place of the meeting at TA)
  Summer TA in Fremont(?), in July -- will meet there on July 28.
  August meeting ???
  Sept. ??? (TI may volunteer -- in Dallas)
  October TA meeting in Phoenix -- co-locate (Oct. 6, X3T10 in Tucson)
  November ???? (Las Vegas?)
  December ???? (Dec. 8 is X3T10 in Orlando)

[Adjourned at ~3:00 PM]

------------------------------------------------------------------------------
List of Attendees:

David R. Wooten
Compaq
davidw@bangate.compaq.com
Robbie Shergill
National Semiconductor
rss@berlioz.nsc.com
Eric Hannah
Intel
eric_hannah@ccm.sc.intel.com
Dave LaFollette
Intel
dlafolle@mipos2.sc.intel.com
Gene Matter
Intel
gene_p_matter@ccm.fm.intel.com
Ganesh Murthy
Intel
ganesh_murthy@ccm.jf.intel.com
Richard Churchill
Compaq
richardc@bangate.compaq.com
Steve Bard
Intel
steve_bard@ccm.jf.intel.com
Cyrus Momeni
Cirrus Logic
Cyrus@corp.cirrus.com
Brad Saunders
Rockwell
bradley.saunders@nb.rockwell.com
Bill Northey
Berg
northewa@bergelect.com
Richard Prentice
Texas Instruments
rprentice@ti.com
Mike Gardner
Molex
mgardner@molex.com
Tatsuya Arai
Hirose Electric
tatsuyaa@hiroseusa
Michael Cheong
Molex
mcheong@molex.com
Jonathan Buck
AMP
jon.buck@amp.com
Dave Brunker
Molex
dbrunker@molex.com
Karl Nakamura
LSI Logic
karln@lsil.com
Patrick Yu
NEC
patick_yu@el.nec.com
Rich Hovey
LSI Logic
rhovey@lsil.com
Jerry Hauck
Intel
jerry_hauck@ccm.sc.intel.com
Daniel Meirsman
Philips
daniel.meirsman@leu.ce.philips.com
Jerry Kachlic
Molex
jkachlic@molex.com
Mark Evans
Quantum Corp.
mevans@qutm.com
Firooz Farhoomand
Matsushita/Panasonic
firoozf@ix.netcom.com
Colin Whitby-Strevens
SGS-Thomson Microelectronics
colinws@bristol.st.com