Minutes, IBIS Quality Committee

08 February 2011

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

ROLL CALL

Cisco Systems:                * Mike LaBonte
Ericsson:                     * Anders Ekholm
Green Streak Programs:          Lynne Green
Huawei Technologies:            Guan Tao
IBM:                            Bruce Archambeault
IOMethodology:                  Lance Wang
Mentor Graphics:                John Angulo
Micron Technology:              Moshiul Haque,
                                Randy Wolff
Nokia Siemens Networks:         Eckhard Lenski
Signal Consulting Group:        Tim Coyle
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:

- No one declared a patent.

AR Review:

- Mike send update PPT to Anders
  - Done
- Anders send PPT and PDF to Mike and Bob
  - Done
- Anders print presentation handouts
  - Done

New items:

Mike showed a feedback document form Anders:
- Extension to IBIS-AMI use?
  - Anders: Todd had suggested this
  - Mike: For time domain only?
  - Anders: [Test Data] could cover other forms
- Eye masks
  - Bob: MM had suggested a format for eye masks
  - Mike: Envelope metric, not overlay
- Test data in the form of features
- PRBS stimuli
  - Mike: Moshiul had brought this up
  - Anders: Need to be sure it is repeatable
  - Mike: There are standards
- Zip archives
  - Anders: There was a long discussion
    - If the zip is big it may be hard to distribute
  - Mike: They only need offer the IBIS file separately
  - Bob: there is a subdirectory issue
  - Mike: Does [External Test Data} give the subdir?
  - Anders: That has platform issues
  - Bob: The tools can add this
  - Anders: Can [External Model] have paths?
  - Bob: No
    - IBISCHK might need a path option
  - Mike: The IBIS spec should not address this
- Automated validation is possible
  - Lynne and Walter gave positive feedback
  - Bob: The comparison is tricky
- Analog stimuli
  - Bob: An s2ibis author was interested in ESD pulse shapes
  - Mike: This would not be a source model
  - Anders: Maybe this is just to test clamps
  - Mike: This should not be for internal development testing
  - Bob: PWL will serve
- Anders comment: How does this correlate to JEITA?
  - They might have ideas
  - Bob: Their test loads are simple SPICE with ideal T lines
  - Anders: They might store this online
    - They have this for a number of simulators
  - Bob: That may be just a plan
    - It's not clear which are the reference results
  - Anders: I had asked why they did not use [Test Data]

AR: Mike research PRBS standards

Mike: Each simulator should need to be verified only once
- Anders: More extensive loads will find more problems
- Bob: Lance has gotten different results in different tools
- Mike: It would be nice for creating test cases
- Bob: [Test Data] should be from silicon SPICE
- Mike: [Test Data] from IBIS simulations would be cheating
- Bob: Except other simulators might differ
  - There would be timestep issues
  - C_comp handling may differ
- Mike: They use waveform sets differently too
- Bob: Users may set up simulations wrong too
  - They may set up options incorrectly

Mike: Why can't IBIS have a .include function?
- Anders: The external keyword would have an include
- Bob: Somebody asked if SPICE could be embedded in IBIS
- Anders: Then many types of includes would be needed
- Mike: It could be a generic include like C and SPICE have
- Anders: Then it would not apply to just test data
  - File delivery could become a problem if they are broken

Mike: Maybe the keyword should be just [External Test]
- Anders: Someone had asked if load and data should be in one file
- Bob: We would have to define keywords for tabular data to combine
- Mike: One [External Test] could point to load, stimulus, and data files
- Bob: The buffer could point to its tests
- Anders: We would need min/typ/max
  - That might be separate
- Mike: Maybe [Test Data] should be part of each [Model]
- Anders: There are many possible combinations of min/typ/max
- Bob: The typ/min/max termination voltages have to be specified
- Anders: We will want the IBIS to pass values into the test circuit

Next meetings:
- Next meetings Feb 15 and Feb 22

Feb 15 agenda:
- Review external test BIRD draft

Meeting ended at 12:15 Eastern Time.