Minutes, IBIS Quality Committee

06 September 2011

11:00-12:00 EST (08:00-09:00 PST)

ROLL CALL

Cisco Systems:                  Tony Penaloza
Ericsson:                       Anders Ekholm
Green Streak Programs:          Lynne Green
Huawei Technologies:            Guan Tao
IBM:                            Bruce Archambeault
                              * Greg Edlund
IOMethodology:                  Lance Wang
Mentor Graphics:                John Angulo
Micron Technology:              Moshiul Haque,
                                Randy Wolff
Nokia Siemens Networks:       * Eckhard Lenski
QLogic Corp.:                 * James Zhou
Signal Consulting Group:        Tim Coyle
Teraspeed Consulting Group:   * Bob Ross
Texas Instruments:              Pavani Jella
unaffiliated:                 * Mike LaBonte

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

-----------------------MINUTES ---------------------------
Mike LaBonte conducted the meeting.

Call for opens and IBIS related patent disclosures:
- None

AR Review:

- Mike add descriptions to AMI checklist
  - Some progress, not done

- Mike add QUAILs to Mantis
  - Mike showed the new IBIS Mantis page
  - This is separate from the EDA.org Mantis page

- Bob conduct tests of parser behavior for test keywords
  - TBD

New items:

Mike showed Eckhard's edits to the AMI checklist:
- Mike: Syntax items should be things IBISCHK can check
- Greg: What happens if Usage or Type are not present

James showed his updated IBIS-AMI Time-Domain Reference Flow:
- James: Discussed with Todd and Walter
  - There was about a year of work on BIRD 120
- Slide 4:
  - James: This is combined from several slides
- Slide 6:
  - James: Extraction formulas shown in flow chart
    - hAC(t) will be double counted
  - Mike: That is not apparent in the flow diagram
  - James: The channel response is embedded into h3(t)
    - The RX does know know what the TX has done to the signal
  - Mike: For quality the user/tool needs to know this
  - Anders: This has been discussed quite a bit
    - Use_Init_Output was there to solve this
  - Mike: That is now deprecated
- Slide 7:
  - James: Use_Init_Output would only work for linear systems
    - That defeats the purpose of Getwave
- Slide 9:
  - James: This is mapped from the BIRD 120 wording
- Slide 10:
  - James: This shows the flow ion 4 columns

James showed BIRD 120:
- James: There is quite a bit about double counting in step 3
  - It is confusing
  - Walter recommended not changing it

- Slide 12:
  - James: The user has to get TX model info to know if Getwave should be used
  - Anders: What does "re-assign" mean here?
- James: Change the value of Getwave_Exists
- Anders: There was a proposal to have two different step responses
  - One with and one without EQ
- James: The user has to know what is in the TX model
- Anders: A solution is to have separate models for TD and stat
- James: BIRD 120 effectively solved these problems
  - Non-linear TX or RX is not a problem now
- Slide 13:
  - James: Steps 1 - 3 have to be used regardless of flow
- Slide 14:
  - James: Flow can be mapped from block I/Os, but not in 5.0
    - It can be decided whether to use Getwave
  - Anders: Getwave has more accurate info than Init does
  - James: It is more accurate than double counting
    - It is said that most TX models are linear
    - So why make Getwave and Init different?
- Slide 17:
  - James: The Init_Returns_Impulse issue is unresolved
    - It is info only
    - Have not been able to agree with Todd on this

Mike: IBISCHK should be expanded to check TX and RX AMI together
- Bob: That would be a new feature
- Eckhard: Both models would have to be available

Anders: If Init_Returns_Impulse is true there is no point in returning IR again
- James: The flow is the same in either case
- Anders: We agree that the 5.0 spec is insufficient
- James: Init_Returns_Impulse is limited to h2 & h3

James showed a spreadsheet from Todd:
- Dual means both Init_Returns_Impulse and Getwave_Exists are true
- There are 4 identical outputs, so Init_Returns_Impulse has no effect
- Bob: It is interesting but not easy to see that there are 4 cases
- James: RX has a problem if TX Getwave is better than Init
  - But it would not know, it has a blind spot
- Anders: The tool knows and can adapt
- James: My understanding is the EDA tool can not automatically know

AR: James send presentation to Mike for posting

Mike: Have a meeting conflict for Sept 20
- We can skip that week or someone else can run the meeting

Next meetings:
- Next meetings Sep 13 and Sep 20

Sep 06 agenda:
- AMI quality

Meeting ended at 12:20 Eastern Time.