IBIS Macromodel Task Group

Meeting date: 29 October 2024

Members (asterisk for those attending):
Achronix Semiconductor:       Hansel Dsilva
Altair:                       Junesang Lee
Amazon:                       John Yan
ANSYS:                      * Curtis Clark
                            * Wei-hsing Huang
Aurora System:                Dian Yang
                              Raj Raghuram
Cadence Design Systems:     * Ambrish Varma
                            * Jared James
Dassault Systemes:            Longfei Bai
Google:                       Hanfeng Wang
                              GaWon Kim
Intel:                      * Michael Mirmak
                              [Kinger Cai]
                            * Chi-te Chen
                              Liwei Zhao
                              Alaeddin Aydiner
                              Sai Zhou
Keysight Technologies:        Fangyi Rao
                              Stephen Slater
                              Ming Yan
                              Rui Yang
Marvell:                      Steve Parker
Mathworks (SiSoft):         * Walter Katz
                              Graham Kus
Micron Technology:            Justin Butterfield
Missouri S&T:               * Chulsoon Hwang
                            * Yifan Ding
                              Zhiping Yang
Rivos:                        Yansheng Wang
SAE ITC:                      Michael McNair
Samsung:                      Jun-Bae Kim
Siemens EDA (Mentor):       * Arpad Muranyi
                            * Randy Wolff
Signal Edge Solutions         Benjamin Dannan
Teraspeed Labs:               [Bob Ross]
Zuken USA:                    Lance Wang

The meeting was led by Arpad Muranyi.  Curtis Clark took the minutes.

--------------------------------------------------------------------------------
Opens:

- None.

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

Michael:  Prepare a new proposed update to the language surrounding Ts4file and
          Ramp and related cross checking that incorporates comments from the
          meeting.
          - Done.

Yifan:  Perform additional investigations into the applicability of BIRD220.1
        for devices with independent pre-driver stages.
        - Done.

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

- None.

-------------------------
Review of Meeting Minutes:

Arpad asked for any comments or corrections to the minutes of the October 22nd
meeting.  Jared moved to approve the minutes.  Michael seconded the motion.
There were no objections.

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

Ts4file and [Ramp]:
Michael reviewed his latest change proposal, which incorporated Ambrish's
feedback from the previous meeting and had been sent to the ATM list shortly
after the last meeting.  Ambrish requested one additional change.  He asked that
"if present" be added to the last sentence of the new Ts4file language for extra
emphasis on the fact that waveforms and I-V tables are optional when Ts4file is
used (slide 4 - "Proposed New Language (2 of 6)").  Michael agreed.  No other
comments were offered.

Michael said he would create a BIRD draft now that the language had been
reviewed.  Arpad suggested one final review cycle in ATM.  Michael agreed.

Arpad commented on the history of the possible confusion about [Ramp] and
Ts4file.  He said some early models that utilized s4p files, prior to the
adoption of the Ts4file AMI parameter, had distributed the s4p files as stand-
alone files that users manually imported into their schematics.  In those early
pre-approval models, the IBIS model ([Ramp]) was trying to drive an ideal step
into the external (to the IBIS model) s4p files.  Those pre-release models may
have led to the confusion.  If they were properly implemented as Ts4file models,
then [Ramp] would not have served that "ideal stimulus" purpose.

Arpad recalled that as part of these discussions we had considered the idea of
extending [Ramp] with a few additional subparams.  He said the idea would be to
allow [Ramp] to be checked against all waveforms.  He said it was reasonable to
do this as a separate BIRD, but he asked whether people wanted it done.  Ambrish
said he didn't think it was necessary, and he asked why we would want to
resurrect [Ramp], which is largely superseded by new keywords.

BIRD220.1 update:
Yifan reviewed the presentation IBIS_validation_1029.pptx.  Previously, Yifan
had found that the following equation for output pad PSIJ:
  PSIJ_total = PSIJ_pre-driver_power_noise + PSIJ_final_driver_power_noise
did not hold in general for devices with independent pre-drivers.  It held for
experiments with the test load connected to GND but had not held when the test
load was connected to Vcc.

At the previous meeting, the group suggested the loading conditions used in the
experiments might explain the unexpected Vcc results.  The group had suggested
redoing the experiments with the test load always connected to Vddq.  That is,
when measuring the pre-driver PSIJ by varying the voltage on the pre-driver
rail, the test load would be connected to the nominal Vddq.  When measuring the
final driver PSIJ by varying the Vddq provided to the final stage, the test load
would be connected to the varying Vddq.

Yifan had redone the experiments under these conditions, and she reported that
for all of her test cases, including single pre-driver stages, independent pre-
driver stages with similar delays, and independent pre-driver stages with
differing delays, the equation holds when the PSIJ values are measured with
the test load tied to Vddq.  In addition, for single pre-driver cases, the
pre-driver PSIJ may be measured at the pre-driver output or the final output
pad.  For devices with independent pre-driver stages, both the pre-driver PSIJ
and the final driver PSIJ should be measured at the output pad.

With the new results, Yifan said that the previous restrictions on the device
types for which the BIRD could be applied were no longer required.  Yifan said
she would prepare a new draft of the BIRD and remove the language regarding any
restrictions on the use of its algorithm.

Per Arpad's request at the previous meeting, Yifan shared several waveforms from
experiments comparing SPICE results to IBIS results with the new PSIJ algorithm
applied.  Yifan noted that her baseline IBIS model was non-power-aware, due to
limitations in the SPICE->IBIS toolset she used.  However, given that starting
model, she was able to demonstrate that the IBIS model with the new PSIJ
algorithm successfully demonstrated the edge shifts due to pre-driver PSIJ that
the SPICE model exhibited.  She showed results for test loads of 50 Ohms to Vcc,
50 Ohms to GND, and 20pF to GND, and the IBIS model with the PSIJ enhancement
matched the SPICE model's predicted edge shifts.

Arpad and Randy thanked Yifan for her tireless investigations.

- Curtis: Motion to adjourn.
- Ambrish: Second.
- Arpad: Thank you all for joining.

New ARs:

Michael:  Create a BIRD draft incorporating the Ts4file and [Ramp] language
          changes.

Yifan:  Create a new draft of BIRD220.1 that removes the previous restrictions
        on its applicability.

-------------
Next meeting: 5 November 2024 12:00pm PT
-------------

IBIS Interconnect SPICE Wish List:

1) Simulator directives
