IBIS Macromodel Task Group

Meeting date: 11 Dec 2007

Members (asterisk for those attending):
* Ambrish Varma, Cadence Design Systems
  Arpad Muranyi, Mentor Graphics Corp.
  Barry Katz, SiSoft
* Bob Ross, Teraspeed Consulting Group
  David Banas, Xilinx
  Donald Telian, consultant
  Doug White, Cisco Systems
  Essaid Bensoudane, ST Microelectronics
  Ganesh Narayanaswamy, ST Micro
  Hemant Shah, Cadence Design Systems
* Ian Dodd
  Joe Abler, IBM
* John Angulo, Mentor Graphics
  John Shields, Mentor Graphics
  Ken Willis, Cadence Design Systems
* Kumar, Cadence Design Systems
  Lance Wang, Cadence Design Systems
  Luis Boluna, Cisco
* Michael Mirmak, Intel Corp.
* Mike LaBonte, Cisco Systems
* Mike Steinberger, SiSoft
  Patrick O'Halloran, Tiburon Design Automation
  Paul Fernando, NCSU
* Radek Biernacki, Agilent (EESof)
* Randy Wolff, Micron Technology
  Ray Comeau, Cadence Design Systems
* Richard Mellitz, Intel
  Richard Ward, Texas Instruments
  Sanjeev Gupta, Agilent
  Shangli Wu, Cadence
  Stephen Scearce, Cisco Systems
* Steve Pytel, Ansoft
  Syed Huq, Cisco Systems
  Syed Sadeghi, ST Micro
* Todd Westerhoff, SiSoft
* Terry Jernberg, Cadence Design Systems
  Vikas Gupta, Xilinx
  Vuk Borich, Agilent
* Walter Katz, SiSoft

--------------------------
Call for patent disclosure:

- No one declared a patent.

-----
Opens:

-------------
Review of ARs:

- Michael M. sketch new IBIS interconnect plan
  - Will contact people offline.

- EDA vendors: meet to decide differential BIRD direction
  - TBD

- Walter coordinate completion of true diff BIRD draft
  - In progress

- Michael M. draft Eye keywords for true diff BIRD
  - TBD

- Randy draft derating keywords for true diff BIRD
  - TBD

- Arpad:  Write parameter passing syntax proposal (BIRD draft)
  for *-AMS models in IBIS that is consistent with the
  parameter passing syntax of the AMI models
  - TBD

- TBD:    Propose a parameter passing syntax for the SPICE
  - [External ...] also?
  - TBD

- Arpad:  Review the documentation (annotation) in the macro libraries.
  - Deferred until a demand arises or we have nothing else to do

-------------
New Discussion:

Richard Mellitz presented:
- "Strategies for Coping with Non-linear and Non-time-invariant V0"
- Slide 12:
  - Using Ansoft simulator
  - Circuit uses ideal current sources
- Slide 14:
  - Mike S: Are true and complement legs balanced?
    - Looks like we are sweeping a 4-dimensional design space
- Slide 11:
  - Why is there an equation for Rout?
  - Poly-silicon elements do not behave like ideal resistors.
  - Steve P: Impedance effect is a function of voltage.
- Slide 18:
  - This is the output of the channel.
- Slide 20:
  - Eye is nearly the same for EQ at Tx or Rx
  - Jitter distribution is different though.
- Slide 24:
  - Eye difference is about 1.5mV/6ps
- Slide 25:
  - Difference is due to re-reflection at Tx
  - S-parameters run more slowly
- Slide 30:
  - Single resistor models are insufficient
    - Todd: 10% to 90% is enormous non-linearity
  - Convolution eye is OK if:
    - Equalize taps at Tx or impedance range < 60%
    - Capacitance variation is minimal
    - Don't require more than a few ps or mV of resolution.
  - Adaptive EQ may be impacted by non-linearity
    - Jitter distribution varies

- Mike S:
  - Have seen cases where EQ tap settings not as expected.
  - It is possible to identify correct tap weights for comparison.
  - Does there exist a set of tap weights that gives identical time-domain wave?
    - Rich: Can get same eye but not waveform.
- Todd:
  - Not clear that we understand why impedance varies so much.
  - Operating outside linear region is begging for trouble.
  - Rich: Operating rail to rail, so transistor turns off.
  - Steve P:
    - Large interaction between impedance and Vgd.
    - Also Miller capacitance has an effect.
    - Higher speeds and lowered voltage rails makes this more prominent.
  - Mike S: Transistor matching isn't what it used to be.
- Mike S:
  - How severe if no Tx de-emphasis used?
  - Rich: Tx de-emphasis not needed.
    - Effects are from re-reflection at Tx
- Mike S: Error in tap settings is a function of transmit swing
  - Todd: Not sure if PC-space and the serdes world are that different.
  - Kumar: Can handle this inside the model itself.
    - Can have external parameters to pass in channel information.
  - EDA vendors should provide algorithmic models.

AR: AMI BIRD authors meet to prove AMI BIRD can handle Mellitz cases

AR: Mike S send non-proprietary information on studies he performed

- Michael M: Is this something we need to account for when making models?
  - Rich: yes
  - Kumar: AMI model can model interactions of channel and Tx
  - There is an interaction between the Rx, channel, and Tx
- Todd: presentation should clarify non-linear operation up front
  - Not sure if this will apply to all systems.
  - Rich: Not sure f linear analysis of channels will always be valid.
- Kumar: Models can account for these effects
  - Todd: We should not encourage writing of arbitrarily complex models.

- We will discuss differentials next week

Next meeting: 18 Dec 2007 12:00pm PT