Subject: Signaling ad-hoc conference call (September 02, 2004)

Attendance:

Joe Abler [IBM]
Michael Altmann [Intel]
Majid Barazande-Pour [Vitesse]
Harmeet Bhugra [IDT]
Brian Brunn [Xilinx]
Joe Caroselli [LSI Logic]
John D'Ambrosia [Tyco]
Justin Gaither [Xilinx]
Joel Goergen [Force10]
Stewart Goudie [Acuid Ltd]
Tom Gray [Artisan]
Adam Healey [Agere Systems]
Glen Koziuk [Vitesse]
Mike Lerer [Rapid Prototypes]
Mary Mandich [Lucent]
Richard Mellitz [Intel]
Charles Moore [Agilent]
Tom Palkert [Xilinx]
Petre Popescu [Quake]
Shannon Sawyer [Agilent]
Brian Seemann [Xilinx]
Jimmy Sheffield [Tyco]
Jeffrey Sinsky [Lucent]
Fulvio Spagna [Intel]
John Stonick [Synopsys]
Brian Von Herzen [Rapid Prototypes, Inc.]
Chris Wittensoldner [Agere Systems]


- Meeting called to order 
- Mike A. reviewed the agenda (http://ieee802.org/3/ap/public/signal_adhoc/altmann_01_0904.pdf) and the 
  outline of the channel model to be used in simulation. 
    - discussion follows regarding the methodology to be followed in converting s-parameters data to time domain impulse response (windowing techniques, etc.) 
    - discussion continued on the suite of models that will be needed to effectively simulate the channel. This allowed mike to introduce John D'Ambrosia presentation. 
- John reviewed his presentation (http://ieee802.org/3/ap/public/signal_adhoc/dambrosia_01_0904.pdf) 
    - A typo was detected in the description of the second configuration listed on slide 6. The description should read: 6" line card > 20" backplane > 6" line card 
    - Richard questions whether the list includes resonant cases 
    - As slide 7 is reviewed, the list of proposed data is presented. Case 1,2,3 focus on long channels while the other cases focus on medium and short interconnect. All data is focused on ATCA pinouts (tx, rx are interleaved as per ATCA spec). 
    - Tyco is willing to make the data available to non-competitors members under NDA. Tyco is willing to let members share simulation results based on that data. Tyco will make a reference presentation of raw data which should be used in reporting results in light of presenting raw data. 
    - discussion follows: 
        - J. Abler suggests that John writes the terms of the NDA 
          - John D'Ambrosia agreed to do so 
        - J. Abler indicates that the proviso of sharing data to "non-competitors" is too restrictive and not inline with IEEE guidelines. 
          - Adam indicates that the issue is not quite clear and he will try to consult with Bob Grow
        - Justin states that it's necessary to get out of the mode of treating other parties as competitors 
        - As the discussion continued, Joel made the point that the idea of accepting the definition of an improved FR4 based on dk, df is to create a template that all can accept. In  particular this should give other manufacturers the capability of duplicate John's work as long as they stick to the accepted definition of FR4 and best design practice. 

    - At this point Mike A. tried to poll the participants to find out if there was agreement that the suite of cases was comprehensive (necessary and sufficient). 
        - Charles commented that he needed to look at the raw data before being able to pass a judgment 
        - Richard commented that impedance variations may not have been included 
    - ... and John moved on to the last slide which listed other considerations that will affect channel characteristics 
    - Once again, Mike tried to get the group to comment on whether the configuration listed in John's presentation were comprehensive 
    - Lots of concerns were stated regarding particular issues 
    - As a result of the discussion, John requested that the signaling ad hoc would state the number of simulations deemed feasible to characterize a certain design/solution
    - A straw poll asked for objections to Johns presented list of channels being used as a starting point
      - No objections 
    - A straw poll asked the group to comment on whether they thought more or less than 20 sims were required 
      - less than 20: 17 votes 
      - greater than 20: 2 votes 
      - abstained: 3 votes 
- John Stonick walked through his presentation (http://ieee802.org/3/ap/public/signal_adhoc/stonick_01_0904.pdf). 
- John indicates that s-parameters data can be remapped to emulate short/long channels 
      - Brian expresses the concern that this can cause unrealizable channels 
        - John agreed indicating that the technique should be used more as a perturbation 
      - John presented a statistical approach to simulating crosstalk 
      - John will clarify how this works 
- Meeting ended with the agreement to schedule an out of program conference call for next Thursday (9 Sept'04) to keep the work going. Also the participants are encouraged to continue the discussion on the reflector. 
- Meeting closd