Minutes, IBIS Quality Committee

26 Feb 2008

11-12 AM EST (8-9 AM PST)

ROLL CALL
  Adam Tambone
  Barry Katz, SiSoft
  Benny Lazer
  Benjamin P Silva
  Bob Cox, Micron
* Bob Ross, Teraspeed Consulting Group
  Brian Arsenault
* David Banas, Xilinx
* Eckhard Lenski, Nokia Siemens Networks
  Eric Brock
* Guan Tao, Huawei Technologies
  Gregory R Edlund
  Hazem Hegazy
  John Figueroa
  John Angulo, Mentor Graphics
  Katja Koller, Nokia Siemens Networks
  Kevin Fisher
* Kim Helliwell, LSI Logic
  Lance Wang, IOMethodology
  Lynne Green
* Mike LaBonte, Cisco Systems
  Mike Mayer, SiSoft
  Moshiul Haque, Micron Technology
  Peter LaFlamme
  Randy Wolff, Micron Technology
  Radovan Vuletic, Qimonda
  Robert Haller, Enterasys
* Roy Leventhal, Leventhal Design & Communications
  Sherif Hammad, Mentor Graphics
  Todd Westerhoff, SiSoft
  Tom Dagostino, Teraspeed Consulting Group
  Kazuyoshi Shoji, Hitachi
  Sadahiro Nonoyama

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

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

Call for patent disclosure:

- No one declared a patent.

AR Review:

AR: Mike check ibischk source for 5.3.1
- Done

AR: Mike send IQ draft 1.1ac with latest changes.
- Done

AR: Mike invite Walter Katz to talk about measurements BIRDs
- No reply yet, probably busy until after DesignCon.
- Mike ask Walter Katz again

AR: David draft a BIRD to clarify [Model] Vinl/Vinh
  - This probably should be suspended after IBIS open forum discussion.
  - David: Specification language is clear
    - [Model Spec] overrides [Model]
  - Kim: Typical values are made from "whole cloth"

New items:

Eckhard's slides:
Slide 1:
- David: min VinH and max VinL used
  - This is correct
  - Some put in typical values, incorrect
Slides 2:
- Black values are from data sheet
- Red values are WORSE than data sheet (incorrect)
- David: This IBIS makes a liar out of the data sheet.
- Kim: [Model Spec] typical values not usually found in data sheet
- Roy: Data sheet numbers are guaranteed, but not the IBIS model
- David: How well do data sheet numbers reflect actual characterization?
- David: Does the parser check [Model Spec] range against [Model]?
  - Bob: No
  - The test would not be valid for hysteresis models
- David: Maybe [Model Spec] Vinl/Vinh should not be used
  - Eckhard: We need min/typ/max values
Slide 3:
- Eckhard: Voltage thresholds are assumptions
- David: IBIS does not convey the assumptions behind the numbers
  - Taking on risk if the more optimistic numbers
  - Need to know the basis of the characterization
- Mike: [Receiver Thresholds] can convey Vinl/Vinh that change with VDD
  - Bob: [Model] Vinh/Vinl OK for TTL, where voltages are set by diodes
- David: The issue here is chip designers sharing risk with system designers.
- Roy: We are litigation-happy
- Mike: margin on table
- David: data sheet values are for a population of parts
  - Could give Gaussian distribution curve
Slide 10:
- Eckhard: limits are based on 99% confidence
- Roy: risk tolerance depends on hazard levels for failure

We looked at the implications of this discussion on 5.2.2 and 5.2.5:
- 5.2.5 does not indicate that Vinlh and Vinh are needed only for Input
  and I/O buffers.

AR: Mike propose addition to 5.2.5 to clarify that [Model Spec] Vinl/Vinh are not always needed

Next meeting:

04 Mar 2008
11-12 AM EST (8-9 AM PST)

Meeting ended at 12:xx PM Eastern Time.