Minutes from the May 29 IBIS ATM task group meeting are attached.
Regards,
Justin
IBIS Advanced Technology Modeling (ATM) Task Group
Meeting date: 29 May 2018
Members (asterisk for those attending):
ANSYS: Dan Dvorscak
Curtis Clark
Cadence Design Systems: Ambrish Varma
Brad Brim
Kumar Keshavan
Ken Willis
eASIC: David Banas
GlobalFoundries: Steve Parker
IBM Luis Armenta
Trevor Timpane
Intel: * Michael Mirmak
Keysight Technologies: Fangyi Rao
* Radek Biernacki
Ming Yan
Mentor, A Siemens Business: John Angulo
* Arpad Muranyi
Micron Technology: * Randy Wolff
* Justin Butterfield
SiSoft: * Walter Katz
Todd Westerhoff
Mike LaBonte
SPISim: * Wei-hsing Huang
Synopsys: Rita Horner
Kevin Li
Teraspeed Consulting Group: Scott McMorrow
Teraspeed Labs: * Bob Ross
The meeting was led by Arpad Muranyi. Justin Butterfield took the minutes.
--------------------------------------------------------------------------------
Opens:
- None.
-------------
Review of ARs:
- None.
--------------------------
Call for patent disclosure:
- None.
-------------------------
Review of Meeting Minutes:
Arpad asked for any comments or corrections to the minutes of the May 22
meeting. Justin moved to approve the minutes. Bob seconded the motion. There
were no objections.
-------------
New Discussion:
- Discussion: Allow Terminator as a Model_type for IBIS-AMI receivers
Michael Mirmak stated he had heard that the group was leaning toward his
solution #1, which was to simply remove the AMI restriction on the Terminator
Model_type. He commented that this would be a very easy BIRD to draft. Arpad
commented that Figure 11 in IBIS 6.1 shows the Rac and Cac connecting to the
node 0 symbol. If someone wants to use these in a power aware simulation, they
would not be very useful. Arpad commented that if we only use these features
for AMI simulations with ideal power, we could go with solution #1. But, we
would need more edits to the IBIS Specification to make them suitable for power
aware simulations. He has heard of some EDA tools discussing support for power
aware simulation with AMI.
Walter stated he would like to review the three options Michael had presented.
Michael shared his slides with the 3 options:
1. Remove the prohibition on Model_type Terminator for IBIS-AMI.
2. Permit [Rac], [Cac], [Rgnd] and [Rpower] in the other Model_types.
3. Create New Model_type that only uses [Rac], etc.
Arpad noted that Walter had also suggested to put these parameters in the AMI
file. Walter stated he would limit these to Input models, allowing those
keywords for Input models only.
Walter commented on the Figure 11, what we want is to not have the clamp
curves, and these are not required for Inputs. He thought we could just add
the Rgnd and Rpower to the Input model. Radek stated that it would be okay to
put in these keywords for the Input model, but the option #1 is easier. He
suggested that the ground in Figure 11 is a separate issue that we do not need
to resolve now.
Walter stated on page 31 of IBIS 6.1 there is a description of the Model_types
including Terminator. He thought that it would be easier and more logical to
change the Input, then we would not need to scrub the IBIS Specification to
clarify Terminator. Arpad suggested that we should add Rac_pullup,
Rac_power_clamp, Cac_pullup, Cac_power_clamp, etc. keywords. Walter clarified
that we do not have the Pullup and Pulldown versions for Input Model_types.
Arpad agreed. Michael commented that this could also be applied to
single-ended models and asked if this would make things more complicated for
the new C_comp model. Randy stated you could also do this inside the IBIS-ISS
subcircuit for C_comp. Arpad agreed and stated we can also do this with
External Model.
Walter asked if these would override or replace the Ground Clamp and Power
Clamp or if they should be in parallel. Bob suggested that they should be in
parallel. Michael stated the difficulty is that we want the true diode clamps
to be there, but it is more difficult to measure the curves independently. Bob
stated this is an extraction problem, and there are multiple solutions that
work. Arpad noted that you cannot represent the diode clamps with a resistor.
Michael asked if we should write up a BIRD for a modified version of option
#2. Arpad suggested to also add the Rac_powerclamp, Rac_gndclamp,
Cac_powerclamp, and Cac_gndclamp. Michael asked if we should make new keywords
or subparameters. Bob replied we should make them new keywords.
Michael asked about support for the I/O Model_type. Walter suggested to
restrict it to Input only.
Walter asked if Rgnd and Rpower can be represented in the clamp curves.
Michael stated, yes, that the purpose is to simplify the models. Walter asked
if we need Rac and Cac support for Michael's purposes. Michael replied that he
has not seen these in differential algorithmic models. Walter stated these
could cause problems for drivers and C_comp compensation. Arpad asked if Rac
and Cac are not a high importance and if we can wait for Randy's C_comp BIRD
for these. Michael stated he would be okay with this.
Bob stated that we should still have the clamp tables in parallel, and for I/Os
things are more complicated. Arpad asked for outputs how the Rgnd and Rpower
would toggle with the buffer state or if they would be there all the time. He
asked if Walter was intending to include these on the driver. Walter stated if
you have Rgnd or Rpower would only replace the clamps.
Bob commented for Submodels there could some complications with the different
modes. Arpad asked if these keywords are available for Submodels. Michael
noted that the are currently not. But, he agreed that it could be a problem if
we allow support for I/Os. Bob suggested to limit these keywords to Inputs and
Terminators.
Arpad asked if we could add Rgnd and Rpower to Terminators and Inputs and if
this would work for Michael's needs. Michael replied that this would work.
Bob asked why not Rac and Cac. Michael noted that these might not be that
useful for differential drivers and Randy's C_comp should address this.
Radek noted that he still prefers to allow the Terminator for AMI, as the
thresholds are not that useful. Arpad stated that he prefers to allow for Rgnd
and Rpower the Input. Radek thought we could do both. Bob was concerned we
should not allow thresholds for Terminators. Michael noted that this is
illegal.
Michael took an AR to prepare a draft BIRD allowing Rgnd and Rpower for the
Input Model_type.
Bob commented he does still have concern with not including split versions of
the Rac and Cac.
- Discussion: BIRD194 Comments
Arpad asked about Michael's BIRD194 comments and if we need to keep it on the
Agenda. Michael replied that if people have comments to please send them out
to the reflector, but he does not see a need to keep this topic on the agenda.
AR: Michael to prepare a draft BIRD allowing Rgnd and Rpower for the Input
Model_type.
--------------------------
Motion to Adjourn:
Randy moved to adjourn. Michael seconded the motion. Arpad thanked everyone
for joining.
-------------
Next meeting: 5 June 2018 12:00pm PT
-------------
IBIS Interconnect SPICE Wish List:
1) Simulator directives
