Minutes, IBIS Quality Committee

07 August 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:

- None

AR Review:

- Mike check on IEEE FSV status
  - No reply yet from Bruce Archambeault

- Mike ask Anders about his level of interest in External Test
  - No reply yet

- Mike prepare a presentation on FOM issues
  - Done

New items:

Mike showed a presentation "A Look at the IBIS FOM"
- slide 2:
  - Bob: The numbers seem high for the differences
  - Mike: The problem is that the denominator is the entire rectangular area
  - Bob: The variable X could be time, not voltage
  - Mike: The ack.c code treats it as voltage

Mike showed the Accuracy Handbook
- Eckhard: This allows for measuring error in the X axis as well as the Y axis
- Mike: The X axis is a good metric for edges, but not for DC errors
- Bob: In the time domain it's meaningless to measure over several cycles
- Mike: It's meaningless to measure any time away from edges
  - David and Roy specified specific features to measure:
    - DC levels
    - switching time
    - slopes
  - We might specify an exact data pattern that must be measured
    - For example 010
    - For SerDes it should not be necessary to run 1 million bits to test model accuracy
- Bob: If there is settling involved it would not catch that

Mike showed the Ack user guide and C code
- Mike: The code uses a passed in value for the denominator, default = 1.0
- Bob: Lance might be interested in this

AR: Mike ask Lance if he can repeat his correlation presentation

- Bob: My view was that 95% correlation is where significant differences show up
  - But this shows much worse differences at 95%
- Mike: A sliding time interval might be a better reference
  - An error at a given time has nothing to do with what happens at other times
- Bob: Also a sample interval might be chosen that shows good correlation for waveforms
  that actually do not correlate well

Next meetings:
- Next meetings August 21 and September 4

- Meeting ended at 12:15 ET