Minutes, IBIS Quality Committee

04 September 2012

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
Signal Integrity Software         * Mike LaBonte
Teraspeed Consulting Group:       * Bob Ross
Texas Instruments:                  Pavani Jella

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

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

Call for opens and IBIS related patent disclosures:

Mike: Lance is unable to discuss his correlation presentation today
- He may be able to join us in two weeks

AR Review:

- None

New items:

Mike showed the 2012 Sorrento summit presentation from Eckhard:
- slide 27:
  - Eckhard: The voltage range here is impossible, but the model works OK
  - Mike: Are these voltages close to the actual swing?
  - Eckhard: No
  - Mike: The output voltage is a multiplier of the algorithmic output times [Voltage Range]
    - Some models have algorithmic output of 1.0 for unequalized signal, the [Voltage Range] gives the actual voltage
    - Others have [Voltage Range] of 1.0V, and the algorithmic output gives the actual voltage
  - Bob: I don't know how this works as an Open_sink
    - There would be no Pullup table
    - An external resistor might be used
  - Mike: I have never seen an Open_sink AMI model
  - Bob: Eckhard. have you tried setting [Voltage Range] higher?
  - Eckhard: No, the model works correctly
  - Mike: Would an Open_sink model be symmetric enough for SerDes operation?
  - Eckhard: The Pullup resistor is non-linear, so it matches
  - Bob: The AMI model runs with -0.5V and +0.5V
  - Mike: That is the input to the TX
  - Bob: Voltage Range should have no effect on AMI output

Mike showed the AMI section of the IBIS 5.1 spec:
- Mike: This says nothing about how the analog model is treated
- Bob: The analog model is used for channel analysis only

Resuming the summit presentation:
- slide 28:
  - Eckhard: An external runtime library is required, but the version is critical
  - Mike: A TX and RX might require different versions
- slide 29:
  - Eckhard: We were told that the package model is in the AMI model
  - Mike: A package coded into the AMI can't participate in reflection simulations
- slide 30 & 31:
  - Eckhard: This model did not implement the multiple Drive values in the datasheet
    - It was easy to fix the AMI file to change Value to List
- slide 32:
  - Eckhard: The TX and RX require different OS versions, so we are stuck
  - Mike: AMI models should be computation intensive, should not have OS problems
- slide 33:
  - Eckhard: We are not told that we have to change Model Selector when the TX_Swing changes
    - This relates to Walter's dependency table proposal
- slide 34 & 35:
  - Eckhard: I was able to solve the Description problem
    - The Format Table problem could not be solved
  - Bob: Some files legal with IBISCHK 5,.0 might not pass 5.1
  - Mike: That is the BUG 133 problem
  - Bob: That will be fixed in 5.1
- slide 36:
  - Eckhard: The AMI description for the EQUAL parameter does not match the datasheet
  - Mike: The IQ specification calls for a some data to be checked against the datasheet

- Bob: Do you have any models that work?
- Eckhard: Some models come in different versions for different vendors
  - Sometimes they run but you don't know if the results are reasonable
- Bob: I suspect major tool incompatibilities

- Mike: Only some of these 7 can become checks
- List of issues in the order shown above:
  - 1 Bob: this needs to be explored, the issue is not conclusive
  - 2 Mike: This might be a check calling for more OS compatibility testing
  - 3 Eckhard: There is no language forbidding algorithmic package modeling
    - Mike: It could be a quality check
  - 4 Eckhard: A check against the datasheet would help
  - 5 Mike: This is another OS compatibility issue, there is no check for this
  - 6 Mike: Without dependency tables this is a documentation issue, there could be a check
    - Bob: Changing Model Selector should work
  - 7 Mike: This is a documentation issue, it could be checked

AR: Eckhard test Open_sink AMI model with different [Voltage Range] values

Next meetings:
- Next meetings Sep 4 and Sep 18

- Meeting ended at 12:17 ET