General Session,
Wednesday, August 6, 1997,
Ala Moana Hotel,
Honolulu, Hawaii

chairman:	Colin Whitby-Strevens
secretary:	Richard Churchill

P1394b reflector: 'p1394b@fireflyinc.com'
P1394b ftp site:  'ftp://ftp.fireflyinc.com/ftp/pub/p1394b'

Agenda
 1. Welcome, Introductions, Objectives
 2. Requirements
         - Whitby-Strevens
 3. Connection scenairios review
    3.1  Need to define connector?
         - Whitby-Strevens
 4. Media
    4.1  POF presentations
         - Nakamura-san
         - Ken Taylor
         - Sugabe-san
         - Morikuda-san
    4.2  MMF presentations
         - Tad Spostak?
         - Dan Ransen?
    4.3  Common Media for all speeds?
    4.4  EMC and eye safety
 5. Provision of power to sub-phys
    5.1  Power consumption by sub-PHYs (3W enough?)
         - Whitby-Strevens
    5.2  Power class (draw power, not repeat power, etc.)
 6. Wall-plate power
    6.1  Conflict with TIA 48.2 building wiring specifications?
    6.2  Leave this to VESA Home Network to solve?
    6.3  Central DC power supply?
         - Whitby-Strevens
 7. Encoding schemes
    7.1  Review of suitability of candicates for S100 and S800 (No decisions,
         information in preparation for P1394b debate on Friday)
         - Watanaba-san
         - Alistair Coles
         - Fujimori-san
         - Whitby-Strevens
 8. Sub-PHY properties
         - Whitby-Strevens
 9. Timing changes required to P1394a to handle long-haul
    9.1  Root contention
         - Dave LaFollette
         - Fujimori-san
         - Alistair Coles
    9.2  PHY delay
    9.3  Beta mode start-up
10. P1394b short haul electrical signal integrity
   10.1  (possibly deferred to Thursday PM if insufficient time)
         Dave Brunker
 1. Welcome, Introductions, Objectives

    Colin Whitby-Strevens, chairman, welcomed all to the meeting.

    General description of meeting objectives, etc.

    Colin described the task of P1394b group, and changes to accomodate the
    long-haul implementations desired by the 1394 TA, VESA, DAVIC, etc.
    Bifurcated (low- and high-speed) effort described, and basis of this
    meeting, to assist in maintaining a consistent solution across the
    two efforts (or possibly to merge the efforts if that makes sense).
    The session is not intended to deal with many formal votes, but if the
    concensus is that such is appropriate, so be it.

    - Meeting Objectives
       - Exchange of information between the S100 and S800 Task Groups
          - find maximum common ground
          - identify areas where different approaches are necessary
          - Topics covered include
             - media,
             - transceiver delay,
             - power consumption,
             - root contention timings,
             - PHY delay,
             - encoding schemes,
             - EMC,
             - scrambling,
             - eye-safety,
             - beta mode start-up, etc.
          - Short haul copper PMD issues (connector and signal integrity issues)

    Further discussion of the general schedule of the P1394b effort (currently
    chaired by David Wooten) followed.  Also, the appropriatness of assorted
    presentations to the current meeting, or of the subsequent sessions, is
    left to the individuals desiring to present materials.

    Taka Fujimori discussed general objectives of the low-cost, low-speed
    meeting for the next day (Thursday), with presentation of the present
    draft, etc.

    Introductions around were made.

    At this point, the agenda was presented, speakers were solicited and
    accepted, and the agenda was approved.  As this was not a formal meeting
    of a single acting body, there were no minutes of a prior meeting to
    approve.  (Previous meeting of same general interest in context of May 5,
    1997 P1394b meeting is treated as a part of that broader meeting.  Its
    minutes appear on the P1394b ftp site as part of the P1394b minutes of
    that date.)

 2. Requirements
    S100/200/400
     1. 50m hop
     2. UTP and Optical fiber at S100, optical for higher speeds
            minimize cost of S100
            common media when possible
            possible dynamic speed determination
                PMDs must be able to identify their max speeds
                PHYs must negotiate speeds
            possibly use PMD intelligence to reduce PMD cost
     3. P1394a (and P1394b?) above the PMD layers
        << Taka desires to get his task done before p1394b may be done. >
            agreed to use a common encoding for all P1394b PHYs/PMDs
            same tree-ID algorithms, Self-ID process
     4. Amateur installable
            Installation guidelines, installation test, ...
     5. Facilitate FCC Class B emissions compliance
     6. Lasers must be class 1 EYE-SAFE (IEC 825-1 and CDRM (r-DA))
        for regulatory purposes
            for amateur installability

    Objective is to be done in November ... up to S200

    <<Issues --
        Relative timescale of P1394b, etc.
        S100 includes wall-plate connector
        Service interface- 
        S100 ....
        ....
        ....
    >

    S800/1600/3200
     1. 50m reach per hop
            need P1394a PHY pinging to determine worst-case delay
     2. Optical fiber support for S800 and greater speeds
        - minimize cost
        - Then common media where possible
        - no need for dynamic speed determination
           - PMDs must be able to identify their max speeds
           - PHYs must negotiate speeds
        - possibly use PMD intelligence to reduce PMD cost further
     3. P1394a and P1394b above the PMD layers
            agreed to use a common encoding for all P1394b PHYs/PMDs
            same Tree-ID/Self-ID algorithms
            Fully interoperable with current 1394 - No bus bridging required
     4. Amateur installable
        - Installation guidelines, installation test, ...
     5. Facilitate FCC Class B emissions compliance
     6. Lasers must be class 1 EYE-SAFE (IEC 825-1 and CDRM (r-DA)) for
        regulatory purposes
        - In order to be amateur installable
     7. Timescale: report in November

 3. Connection scenarios
     *** diagram ***
     PHY--[]--subPHY/PMDn--[]---------[]--PMDn/subPHY--[]--PHY
     - Cable-powered dongle provides EO/OE conversion, or P1394b electrical
       to long-haul electrical for S100 UTP
     - sub-PHY provides PHY repeater function
       - see later for full spec.

    << Taka points out that S100 would use a full PHY >
    << Alistair Coles - Are we precluding the PMD being attached to anything
    but a PHY?  Colin - I hope not!>

    Taka --- S100
    Long-haul low cost (S100 models)
    1. Repeater/Wall-plate
       *** diagram ***
                   -95/a/DS    Self-ID? Yes - may have more ports
       link/PHY--[]-------long/full/PHYb----------
                               WALL           1394b-S100TG(POF/UTP/HPCF)
    2. Dongle
       *** diagram ***
            -95/a/DS
       link/PHY[]long/full/PHYb[]------|--------------------
                    self-ID,          WALL
               may have more ports

    3. direct long distance
       *** diagram ***
       link/PHY/longPMD-[]---------------------------

    <<will standardize an optical connector, etc. ....>
    <<comments regarding desirability of common service interface, >
    <<near term scope focusses on S100/200 ... >
    <<There is pressure to use higher speed Ethernet for the home network,
    which means that we need to get moving. >

    S800
    1. Wall plate
       standardized the fiber and optical performance
       *** diagram ***
       PHY--[]-------[]--subPHY/PMDn----------
                    WALL

    2. Long-distance cable incorporating a dongle
       standardized the medium and the optical parameters
       PHY--[]--subPHY/PMDn---------------------
                  Dongle

    <<There is a need to re-engineer the standard connector to accommodate
    higher data rates, which is a general P1394b issue.>

    Other connection scenarios
    3. Long-distance cable incorporating a dongle, with passive wall-plate
       standarize the medium, the optical parameters and the wall-plate
       connector
       *** diagram ***
       PHY--[]--subPHY/PMDn---------------------[]--------------
                                               WALL     wall box

    4. Long-haul interface on the equipment
       standardize the medium, optical parameters, and the connector
       *** diagram ***
       PHY----------[]--------------

    - Prioritise scenarios 1 and 2

    S100 (Taka)
    scenario 2 may be first, but scenario 1 may be preferred by customers
    Scenario 3 will take some time.  There are also questions of power, and
    an active wall-plate may be needed.  VESA should probably deal with the
    question of active versus passive wall-plates.

    Model 1 and 2 equal
    Model 3 later
    cable power or active wall-plate or power brick
    power to be solved by the VESA HN/DAVIC committees, or whoever
        defines the connector

    <<We really need to define what we mean by a "PHY".  This is not a present
    clear, as it is in the the -1995/1394a environment.  Colin - What we are
    trying to do is extend our understanding of the PHY, but the definition
    of a PHY is more than the port definition.  The definition for PMD is what
    is needed to drive the bits down the line.  ... We need to define a service
    interface.  Peter J. - Does VESA include providing power via the power
    bus a part of their problem?  Do we solve this, or leave it to the 
    application?  VESA is for home application, while 1394 is for business
    and home, and we should solve this.  Do you see 1394 networking as an
    important business solution?  Yes, due to various problems that will be
    arising.  An office is in interconnect environment.  I do see it as an
    important business solution. - Jim ?>

<<<<<<<< Brief recess called at 10:32, resuming at 11:01 >>>>

 4. Media
 4.1  POF presentations
    Nakamura-san, Mitsubishi-rayon
      - POF for Higher Speed;
        Technology Status
        P1394b Honolulu, HI, Aug 6, '97
        
      - Intro
        Advantages of POF
        Technology status
        Roadmap

      - Advantages of POF
         - low cost
            - fiber itself
            - termination
            - simplicity
         - security
         - EMI immunity

      - Technology Status
         - Low NA SI-POF; ESKAMEGA(tm)
         - Satisfies the ATM Forum 156 Mbps@50m spec.
         - Multi vendor environment
         - S100 compatible @ 50m
         - S200 (@50m) under study
      - GI-POF
         - ESKAGIGA(tm) Sample dstribution '95-
         - new fiber
         - ...

      - Roadmap
         - Target
            - SI_POF S200 challenge -1997/E
            - GI-POF S200-S400 -97/E
            - GI-POF 1GHz -'98/E
         - Collaboration with 650nm Transceiver vendor
         *** diagram ***
         SI-POF                            GI-POF
         S100       S200        S400         S800

        <<What kind of launch conditions, etc., are required?  0.3 na>

    Ken Taylor, Boston Optical Fiber, Inc.
      - Graded Index Fiber, 1st generation
         - core material, etc., for later distribution

      - Graded index fiber, 2nd generation
         - core materical, etc., for later distrubtion

        <<How does this relate to the previously presented material?
        Much the same.  Fibers are tested to higher speeds, but the
        second generation has not in the form of a fiber.  Bandwidth
        measurements for launch conditions, etc., into the fiber have
        been done ... low na ...  What is the mode mixing length?
        unknown. ....>

    Kazuki Sogabe-san, Sumitomo Electric Industries, LTD.
      - High NA Wide-Bandwidth Hard Plastic Clad Silica Optical Fiber
        for IEEE1394b

      - Structure of fiber (H-PCF)
        *** illustration ***
        Core-cladding-buffer-tension member-shell
         - Core: 200 um Pure Silica or Ge doped Silica for high performance
           optical properties
         - Cladding: 225 um Hard Polymer Cladding for strength and fatigue
           resistance
         - Buffer: 500 um ETFE Resin for resistance to abrasion and chemicals
         - Tension Member: Aramid Yarn (optional)
         - Jacket: 2.2 mm PVC

      - Features of H-PCF
         - emi/emc immunity, etc.
         - High NA and large core diameter accomplish high coupling efficiency
           with LED and allow to using potentialy economical data-link devices
           and connectors
         - Easy filed connector assembly works by using "Climp & Cleave"
           optical connectors

      - Product data sheet
        *** table ***
        *** table ***

      <<How does this compare with standard glass fiber in terms of cost?
      Connector is much easier to install, cheaper, etc.  What is the cost
      benefit?  Glass is higher?  (Comment by chair that the question of
      media cost must be asked on an individual basis, outside the confines
      of the meeting.  We should be asking what is the best for our group
      requirements ...)>
      [See reflector for these

    Morikuda-san, Matsushita Electric Industrial Co., Ltd

      - Outline
         - What is the optical transmission using H-PCF?
            - Fundamental Characteristics of H-PCF
            - ....

      - Fundamental characteristics of POF and H-PCF
      *** diagram ***

      - Features of Optical Transmission using H-PCF
      *** diagram ***

      - Waveforms at 200 Mbps
      *** illustration ***
      *** illustration ***
      *** illustration ***

      <<Colin observed that edges on the eye diagrams were amazingly sharp.
      Were these as observed?  Yes.>

      - Bit Error Rate characteristics at 200 Mbps (1)
      *** diagram ***

      - Bit Error Rate characteristics at 200 Mbps (2)
      *** diagram ***

      - Bit Error RAte characteristics at 200 Mbps (3)
      *** diagram ***

      - Waveforms at 400 Mbps
      *** illustration ***
      *** illustration ***

      - Bit Error Rate characteristics at 400 Mbps (1)
      *** diagram ***

      Application Area of Optical transmission using H-PCF at 850 nm
      *** diagram ***

      <<Conclusion is that we must use the GI-PCF if we wish to go to
      400 Mbps at 50m?  Yes.>

 4.2  MMF presentations
    Dan Hansen, HP

      - Perspective on 1394b 4 GBd Multimode fiber

      - Outline
         - Exploring 1, 2 & 4 Gbd Link Design Objectives
         - Link Technology Selection and Analysis Framework
         - Receiver Technology Selection for 4 GBd operation
         - Comparison of Gigabit Ethernet (GbE) and 1394b Link
           worst-case parameters
         - 4GBd Link Penalty and Length Analysis
         - ....

      - 4 GBd 50MMF Link Penalties and Limits with 850 nm VCSELs
      *** diagram ***

      <<This means we can do S800/1600/3200 on the same fiber with the
      same transceiver?  Yes, potentially.>
      <<It would be remiss of us to do something that some member would
      know did not meet safety requirements, ... Dan - HP has always
      sought to assure safety ...>
      <<Colin - ONE of the desired goals is to have a common medium for
      all solutions.  What would be the pain level for the low-speed
      solutions to use the same solution as the high-speed interconnect?
      This is a complex question, which will be addressed in the fuller
      presentation tomorrow.  The job of the group is to determine where
      things are in the sense of our requirements, and select accordingly.
      I (Dan) can construct an arguement for any of these solutions ...
      There is no reason to assume that 4 Gb parts will consume any more
      power than 1 Gb parts.  Also there are innovations in the pipe that
      might yield even lower power consumption than is presently available
      in the market.>

    Tad Spostak, 3M corp.

      - 3M fiber-to-the-desk-system

      - Network Cost Sensitivity
         - ....

      - Volition(tm) The System
      *** illustration ***

      - Volition, Low Cost, Small format optical interconnect
         - .....

      - 3M Objective
         - Is simple and easy to sue
         - Has supporting cost effective electronics
         - REpresents the lowest total cost solution
         - Is Future proofed -----3GHz+
         - Has the best backing and support
         - Affordable total cost
            - comparable to CAT-5

      Video shown, entitled "3M Volition Premises Cabling System."

<<<<<<<< Recessed for Lunch at 12:13, to resume at 1:30.  (actually 1:43) >>>>

 4.3  Common Media for all speeds?

    Colin - additional slide

      - S100 looking to use UTP, POF and H-PCF
      - Last meeting, the S800 PMD group identified 850nm multi-mode
        glass fiber (with 50um VCSEL technology) as suitable for S800,
        S1600 and S3200 (1, 2 and 4 GBd)
         - different launch/sensitivity parameters for each speed
      - Issues
         - Emphasize single medium for future-proofness?
         - Will POF meet all speed requirements on sutiable timescale?
         - Are the benefits of POF for slow speed sufficient to justify
           two media?
         - Do the costs of MMF justify seeking an alternative for slow
           speed?
         - How easy is connectorization/installation for the various media?
      - Presentations, please ....

    Straw polls ---
    Question: Should we standardize a connector?
        Half the attendees indicated yes, one no, and the remainder abstained.
    Question: Should we standardize on a single media?
        <<Should we attempt such a poll now, given that we don't really have
        much information from which to work in making a decision?>
        Question: Should we take this poll?
            five in favor, but 15 or more against, so poll skipped.

    <<Discussion of desirability of having a high-speed medium support the
    lower speeds expressed, with comments regarding whether it is desirable
    or advisable to allow use of a lower grade medium for lower speeds.  A
    standards committee may either bless any solution that meets its
    peculiar requirements OR specify one solutions, or a small set.  Chair
    and others expressed the view that a standards committee's function IS
    to limit choices, in order to make the solution truly standardized.
    There is an inherent risk that any failure to adequately specify will
    result in interoperability problems.  There are issues of whether we
    can use a common connector with multiple media types, with some connectors
    supporting such implementations, and others not.>

 5. Provision of power to sub-phys
 5.1  Power consumption by sub-PHYs (3W enough?)
 5.2  Power class (draw power, not repeat power, etc.)

    Colin Whitby-Strevens, SGS-Thomson Microelectronics
      - Power Issues
         - Provision of power to sub-PHYs
            - Power consumption by sub-PHYs (Is 3W enough?)
            - Power class (draw power, not repeat power, etc.)
            - How reported?
         - Wall-plate Power
            - Conflict with TIA 48.2 building wiring specifications?
            - Leave this to VESA Home Networking Committee to solve?
            - Central DC power supply?

    <<It seems that 3W will be adequate, and in fact 1.5W might be sufficient
    according to the numbers available from some vendors at this time.>
    <<Questions regarding how power consumption is to be reported were raised,
    in the context of how consumption is reported via self-ID.  Suggestion
    was made that the sub-PHY should be power class 4.  How are these classes
    to be interpreted with regard to passing power, etc.?  Class 4 is the
    closest fit we have ... class 4 if powered from cable, class 0 if powered
    seperately.>
    <<Will a cable report a node ID for each end of the cable, or a single to
    represent the cable as a whole?  At least ONE node ID reported.>
    <<Issue of active components has been raised with the TIA, and the idea
    was voted down.  Specific votes called for any active component being
    located outside the wall-plate, so that the "owner" can know what's
    present in the home.  There should be no problem with a "pimple" outside
    the wall, even if this is attached to the wall.  There was discussion of
    how "pimples" are managed, with Max Bassler offering references to the
    appropriate EIA and Smart House standards.

    EIA Interim Standard, Book 1 (volumes 1 through 4) EIA Home Automation
    System (CEBus), EIA/IS-60, October 1992; by Electrical Industries
    Association (EIA), Engineering Department.  Contact ---

        EIA Engineering Publications Office
        2001 Pennsylvania Ave., N.W.
        Washington, DC, 20006

	phone: (202)457-4963

    Smart House for Code Officials: Electrical.
    

    <<Point made that these documents cannot be obtained by individuals, but
    that we as an organization may obtain them.>
    << .... >
    <<A solution wherein we run parallel copper and fiber, with the copper
    carrying low voltage AC or DC for use withing the "pimple".>
    <<Suggestion made that the chair examine the available documents during
    an upcoming break so as to clarify the issues as dealt with by these
    standards.>
    <<We will NOT leave this to the VESA committee ...>
    <<Likely that the power would be AC, and NOT DC.  The possibility exists
    that the "pimple" would provide 1394 power bus power, at a minimum of
    15W, at 8V minimum, 20V appropriate to the current power distribution
    and managment proposals.>
    <<The general impression is that we will need to provide the power solution
    to the VESA HN committee, etc.>

    <<Observation made that the P1394b committee voted overwhelmingly that we
    shall share a common coding scheme, etc., across all long-haul and p1394b
    short-haul solutions.  The low-speed group has been studying the use of
    4B/5B, while the committee as a whole, and the high-speed committee in
    particular have been working towards an 8B/10B solution.  We must examine
    what the issues are, and make clear the liabilities of each coding scheme
    relative to the intended solutions.>

<<<<<<<< Recess called at 2:43, with resumption at 3:05. >>>>

 7. Encoding schemes
 7.1  Review of suitability of candicates for S100 and S800 (No decisions,
      information in preparation for P1394b debate on Friday)

    Fujimori-san, Sony

      - POF/UTP 1394 PHY Layering
      *** diagram ***
      LINK Layer --------------------------------------------------------
                Media Access  |                1394                     |
                  Control     |              Link/MAC                   |
           ==============================================================
             Physical Coding
       ....

      - 4B/5B Code Mapping to 1394
      *** table ***
      CODE            SYMBOLS              1394 LINE STATE
      00000_11111     Q/I                  Bus Reset
      ....

      - How Packets Look Like
         - S100 Long PHY.  A symbol takes 40 ns to transmit
                 DATA_PREFIX        DATA                   DATA_END
                 JKJKJK  ....   1 2 3 4 5 6 ..............  9TTTTTT
         - S200 Long PHY.  A symbol takes 20 ns to transmit
                 DATA_PREFIX        DATA                   DATA_END
           S100: JKJKJKJK ...   1 0 2 0 3 0 ..............  90TTTTT
           S200: JKJKJKJK ...   1 2 3 4 5 6 ..............  9TTTTTT
         - S400 LOng PHY.  A symbol takes 10 ns to transmit
                 DATA_PREFIX        DATA                   DATA_END
           S100: JKJKJKJK ...   1 000 2 000 3 000 ........  9000TTT
           S200: JKJKJKJK ...   1 0 2 0 3 0 4 0 5 ........  90TTTTT
           S400: JKJKJKJK ...   1 2 3 4 5 6 7 8 9 ........  9TTTTTT

      <<Since you only have 120 ns for the data prefix, you can only
      send a three symbols during data prefix ... There are other ways
      to signal DATA_PREFIX than to send back-to-back JKs, which are
      in themselves problematic.>
      <<Use symbol pairs to simplify decoding ...>

      - Scrambling 8B/10B
      ....

      - Scambled 4B/5B vs modified 8B/10B
         - All Aspects prefer scrambled 4B/5B:
         *** table ***
                           Scrambled 4B/5B        modified IBM 8B/10B
         --------------------------------------------------------------
         emission          ---                    +6 or +7 dB more
         --------------------------------------------------------------
         Grant             twice robust           ---
         --------------------------------------------------------------
         gate count        smaller                larger
         --------------------------------------------------------------
         PHY_DELAY         smaller                120ns<<longer at S100
         --------------------------------------------------------------
         scrambler         no beta mode           beta mode & bus reset
         synchronization   conseq. Q, I or H      /complex beta mode
         --------------------------------------------------------------
         symbol duration   40ns at S100           80ns at S100
         intimacy to 1394  Good                   not good
         --------------------------------------------------------------
         UTP/CAT-5(100m)   MLT-3                  impossible

    <<Is it your position that MLT-3 makes this possible over UTP, and that
    it is really impossible to do likewise with 8B/10B?  Yes ...  The item
    stating that 8B/10B over UTP being impossible should be changed to say
    that both are possible using MLT-3.>
    <<????>
    <<It is necessary that you KNOW you are receiving a Q, I or H in order
    to use these to resynch the scrambler.  Alistair will present material
    on Friday showing how 8B/10B scrambler can be synchronized much more
    easily that Fujimori-san indicates.>
 
      - Copies of pages from FDDI working draft shown to illustrate the
        scrambler and related problems ...

    Watanaba-san, Sony Corp.

      - PMD UTP Cat.5 Emission

      - 4B/5B spectrum with scrambler
      *** illustration ***
      *** illustration ***

      - 4B/5B spectrum with scrambler and MLT-3
      *** illustration ***
      *** illustration ***

      - Modified 8B/10B spectrum with scrambler
      *** illustration ***
      *** illustration ***

    <<What have you assumed peak-to-peak voltages?  You need to compare the
    information correctly, and compensate for the use of a three-level code
    as compared to a two-level code.  The additional circuitry for decoding
    a three-level scheme must also be accounted for.  MLT-3 is state to be
    required only for S100 (over UTP) at 100m, and not for higher speeds.
    Where is peak of the lowest frequency FCC doesn't care about? (???)  It
    is well below the FCC threshold frequency.  (back to other topics) There
    is no advantage obtained to 4B/5B by MLT-3, using the same conditions,
    rather than the poorly matched comparative cases presented ... 8B/10B
    should obtain similar results when accurately compared with 4B/5B ...>

    Alistair Coles, HP

      - Overview of the modified 8B/10B code scheme

      - Objectives of coding scheme
         - DC balance
            - essential for all fiber media and UTP
         - Low maximum run length, sufficient transition density for
           clock recovery
         - Spectral properties conducive to FCC class B emissions
         - Sufficient control signals for 1394 (+ spares)
         - Low complexity

      - Data coding
         - the proposed scheme uses exactly the same 8B/10B data coding
           as Fibre Channel and Gigabit Ethernet,
         - DC balanced
            - 10 bit characters have disparity of zero or +/- 2

      - Control Coding
         - A new set of 10 charactgers is used for control signaling.
           These replace the ordered sets used in Fibre Channel, and in
           Gigabit Ethernet.
            - Shorter than ordered sets
            - greater Hamming distance from data codewords
            -

      - New control codeword set

      - Control State Mapping

      - Scrambling

      - (slide from previous meeting showing spectral characteristics of
        the control code set presented.)

      - (slide from previous meeting showing spectral character of sending
        idle using Sony control mapping presented.)

      - Block diagram
      *** diagram ***

      - Spectral properties
      *** diagram ***
         - Differences between scrambled data and control spectra should not
           be significant given redutcion provided by scrambler.  (Clearly,
           this should be checked by EMC data.)

      - Spectrum at S100
      *** diagram ***
         - UTP is only feasible at S100 (unless complex coding is used).
         - Major differences in spectra are below 30 MHz: no problem.
         - Modified 8B/10B is suitable for UTP (and POF) at S100.

      - Error detection properties (1)
         - Two single bit errors within an 8B/10B coded packet:
            - When decoded, these result in two error bytes
            - with the IBM 8B/10B code, ALL COMBINATIONS OF TWO ERRORS ARE
              DETECTED BY CRC32.
         - Three single bit errors within an 8B/10B coded packet:
            - Three error events will always cause a disparity error
            - Disparity is checked using packet delimiters (similar to Fibre
              Channel).  Disparity at start of packet is indicated by sending
              either DATA_PREFIX+ or DATA_PREFIX-.  Disparity at end of
              packet is check by sending either DATA_END+ or DATA_END-.
            - ALL COMBINATIONS OF THREE ERRORS ARE DETECTED BY THE DISPARITY
              CHECK.
         - Burst Errors
            - CRC32 will detect bursts up to 32 bits long in uncoded data.
              This translates to bursts up to 35 bits long in coded data.

      - Error detection properties (2)
         - Error propagation in forwarded packets
            - Many error combinations that CRC32 may not detect will be
              detected by code violations (invalid codeword received).
            - These errors can be masked when a packet is decoded, forwarded
              and then coded (correctly).
            - 

    <<One of the differences between this and FC is the .... Is this an intended
    trade off?  Yes.  FC and GbE have only tested assuming control code sets,
    not with the scheme presented ...>
    <<How does this relate to the CRC32 error detection scheme, with its
    Hamming distance (equivalent) of three?  Quite well, and this preserves
    the integrity of CRC32 in our application.>

      - S100 1394b and UTP

      - Background
         - During the last five years ...
         - ....

      - Twisted Pair Transmission Issues
      *** illustration ***

      - Radiated Emissions
      *** diagram ***

      - AC Coupling

      - 100BaseTX/FDDI Approach: MLT3

      - MLT3 Power Spectral Density

      - MLT3 Drawbacks

      - RMI-3: An improved 3 level code

      - Drawbacks of MLT3 and RMI-3

      - Another way to reduce emissions ...

      - S100 spectra v. cable length

      - Coding options for S100 1394b

      - (slide from previous meeting entitled "Comparison of coding
        schemes (3)" presented.)

      - Summary

    <<MLT-3 is itself complex, and represents a serious liability in terms
    of gate-count for data reconstruction. ....>
    <<Where we may support DC vs. AC coupling commented upon ... for short-haul
    we might well choose to be DC coupled, but can still use AC coupling in
    the long-haul portion of the standard.>
    <<Additional discussion of relative merits ...>

    Whitby-Strevens

    Encoding Schemes - 1
    Review of suitability of candiates for S100 and S800 (No decision.  Info
    in preparation for P1394b meeting on Friday.)

                       4B/5B                      Modified 8B/10B

    Advantages         2 errors in 40ns for an
                         erroneous grant

    Disadvantages      Long Run-length            8 bit time serial/parallel
                       DC balanced (within 10%)     conversion necessary
                       Short code not a benefit   6 dB worse at lower freqs.
                         (need 8-bit link inter-    (but within FCC limits)
                         face support)            2 errors in 80ns for an
                                                    erroneous grant

    Encoding Schemes - 2

                       4B/5B                      8B/10B

    Advantages for     In market?                 In market @ Gb (not modified)
    S100 optical       

    Advantages for     In market?
    S100 UTP           100m

    Disadvantages      Spectral Peak - May or     Extra 40ns(?) latency
    for S100 optical     may not be a problem       (~7.5m equivalent)
                       Not DC balanced - cost
                         1.25 dB

    Disadvantages      Long run lengths           Extra 40ns(?) latency
    for S100 UTP       not DC balanced              (~7.5m equivalent)
                       Complex analog Electron-   Only 70m, not 100m
                         ics (MLT3) needed

    Advantages for                                In market
    S800               

    Disadvantages      As for S100
    for S800           Spectral peak big issue

    Meeting adjourned due to end of suitable work period.

===============================================================================
Attendees:

Colin Whitby-Strevens
no phone #
colinws@bristol.st.com

Brad Saunders
no phone #
bradley.saunders@rss.rockwell.com

David Wooten
no phone #
david.wooten@compaq.com

Jack A. Serafin
no phone #
jack_a_serafin@ccm.ch.intel.com

Dan Rausch
408-435-6689
dan_rausch@hp.com

Alistair Coles
+44 117 922 8750
anc@hplb.hpl.hp.com

Harlan Andrews
408-974-6430
hea@apple.com

Peter Johansson
510-531-5472
pjohansson@aol.com

Tad Szostak
512-984-3847
tszostak1@mmm.com

Juan Pulido
512-984-5188
jmpulido@mmm.com

Jim Tatum
972-470-4572
jtatum@micro.honeywell.com

Peter Teng
408-588-5555
pteng@mail.com

Steve Bard
503-264-2923
steve_bard@ccm.jf.intel.com

Richard Churchill
281-514-6984
richardc@bangate.compaq.com

Michael Fogg
717-986-5802
mike.fogg@amp.com

Nelson Arata
408-721-4979
nelson.arata@nsc.com

James Kuo
408-721-7856
jamesk@galaxy.nsc.com

Geert Knapen
+32 16 390 734
geert.knapen@innet.bc

Daniel Meirsman
+32 16 390 733
daniel.meirsman@leu.ce.philips.com

Jack Merrow
425-486-2222
jmerrow@leviton-telecom.com

Karl Bergstrom
800-888-0211x4563
mpegoot@aol.com

Ken Taylor
508-347-3309
ktaylor@bostonoptical.com

Dick Scheel
408-955-4295
dicks@lsi.sel.sony.com

Kenji Watanabe
+81 3 5448 5362
nabeken@sslab.sony.co.jp

Yukio Otobe
+81 44 754 2532
otobe@flab.fujitsu.co.jp

Farrokh Mottahedin
408-324-7934
fmottahe@qntm.com

Tek Wei
408-432-7026
tcw@cypress.com

David Brief
408-721-4485
david.brief@nsc.com

Jim Busse
415-528-3810
jimb@ccgate.sj.nec.com

Zack Ciccone
602-752-6233
zack.ciccnoe@tempe.vlsi.com

Charles Brill
717-592-6198
cebrill@amp.com

Paul S. Levy
602-752-6382
paul.levy@vlsi.com

Takahiro Seki
+81 462 30 6209
seki@lsis.crl.sony.co.jp

Shuntaro Yamazaki
+81-44-856-2082
yamazaki@ccm.cl.nec.co.jp

Kazuki Nakamura
+81 8275 3 5320
kazuki-n@dtinet.or.jp

Palanisamy Mohanraj
602-554-4243
palanisamy_mohanraj@ccm.ch.intel.com

Kazuki Sogabe
+81 6 466 5539
dw750134@jnet.sei.co.jp

Roy Yasoshima
415-858-1000
yasa@masca.com

Dave Brunker
630-527-2622
dbrunker@molex.com

Susumu Morikura
+81 6 906 4896
morikura@isl.mei.co.jp

Paul Lind
602-554-3774
plind@sedona.intel.com

Jerry Hauck
408-765-5528
jerry_hauck@ccm.sc.intel.com

Dave LaFollette
408-765-2587
dlafolle@mipos2.sc.intel.com

Doug Moran
619-535-3306
doug@brooktree.com

James T Doyle
602-554-2051
jdoyle@sedona.intel.com

Ali Ghiosi
650-786-3310
ghiasi@eng.sun.com

Bill Pherigo
970-229-3586
wlp@fc.hp.com

Bill Prouty
916-785-4631
billp@hprnd.rose.hp.com

Max Bassler
630-527-4490
mbassler@molex.com

Daniel Schwartz
602-413-5320
a186aa@mot.com

Dan Brown
717-986-7812
dan.brown@amp.com

Farrukh Latif
610-712-7546
flatif@lucent.com

Bill Northey
717-938-2119
northewa@bergelect.com

Claude Huss
650-237-2366
claude@mew.com

Michael Nguyen
408-894-3542
michael.nguyen@fpca.fujitsu.com

Edward Butler
602-554-9751
ebutler@sedona.intel.com

Rahoul Puri
408-974-8598
rahoul@apple.com

Karsten Droegemueller
+49 30 386 22635
karsten.droegemueller@p11.moh1.siemens.net

Stan McHann
403-326-1360
stan_mchann@3com.com

Michael Wang
602-554-8555
mike_wang@ccm.ch.intel.com

James Gay
512-891-2218
jimg@oakhill.sps.mot.com

Yasumasa Hasegawa
+81 22 347 1128
yasumasa@ffm.fujifilm.co.jp

Hiyokuzi Mamezaki
+81 22 347 1128
mamezaki@ffm.fujifilm.co.jp

Tatsuya Arai
805-522-7958
tatsuyaa@hiroseusa.com

Kazuyuki Abe
415-528-5904
kabe@sj-pceg.ccgw.nec.com

Vilas Bhade
408-777-4723
vilas@wipro.com

Mike Brown
602-554-3713
mike_brown@ccm.ch.intel.com

Eric Hannah
408-765-4441
ehannah@mipos2.sc.intel.com

Steven Midford
408-765-8370
steve_midford@ccm.sc.intel.com

Jay Neer
561-447-2907
jneer@molex.com