Is there anything in the current BIRD 189 specification that prevents Sam
Chitwood from building the models he thinks are required?
You say that "And I haven't seen any IBIS package models yet, which had
ideal shorts for the GND pins and "double" values on the POWER pins.". But
if you read Sam's [Model Data] Extraction Methodology he picks one GND
net and excludes if in the [Model Data] pin list, and makes the pulldown
node the "ref" node. This is shorting GND pins and making them the "ref"
node. This is "ground based" power aware simulation. This paper was
written in 2008 (presented by Brad Brim), so I do not know if their
position has changed, or if they are saying they need to do "ground based"
simulation because of other limitations in [Model Data].
978.461-0449 x 133
<ibis-interconn-bounce@xxxxxxxxxxxxx> On Behalf Of Muranyi, Arpad
Sent: Wednesday, March 21, 2018 7:25 PM
To: 'IBIS-Interconnect' <ibis-interconn@xxxxxxxxxxxxx>;
Subject: [ibis-interconn] Re: Putting to Bed A_gnd
Regarding "Almost universally, package interconnect, board interconnect
and connector interconnect SPICE
models use Node 0, and almost universally power aware simulations are
If the first half of this statement is true, why don't we just say in
BIRD189 that the N+1th reference terminal
is connection to A_gnd, and give no other options? That could eliminate a
lot of the "problem cases" we
have discussed in this thread.
I tend to disagree with the second half. If we can make power aware IBIS
[Model]s, and if the IBIS [Pin]
and various package keywords allow for package RLC on POPWER and GND pins,
we are talking about
non-Ground Base modeling. And I haven't seen any IBIS package models yet,
which had ideal shorts for
the GND pins and "double" values on the POWER pins. This "problem" in
IBIS was actually pointed out
and heavily critiqued in several IBIS summit presentations by the then
Sigrity folks a long time ago.
Other than that, I am not totally opposed at making A_gnd a global
reference (like Node 0), but I would
prefer something less restrictive. I still don't see what the problem is
with making it a [Component]
level local reference, and let the EDA tool decide whether to connect it
to their global reference in
simulation or not.
[mailto:ibis-interconn-bounce@xxxxxxxxxxxxx] On Behalf Of Walter Katz
Sent: Wednesday, March 21, 2018 3:01 PM
To: 'IBIS-Interconnect' <ibis-interconn@xxxxxxxxxxxxx
<mailto:ibis-interconn@xxxxxxxxxxxxx> >; ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-interconn] Putting to Bed A_gnd
This discussion of A_gnd has gotten way out of control, unnecessary and a
huge waste of time.
A_gnd should be defined as a Global Simulator Reference Node (e.g. Node
* If a simulator supports IBIS-ISS then it must support Node 0.
* BIRD 189 is based on IBIS-ISS.
It is very important to note that the model maker who creates BIRD 189
models can choose to
* Not use A_gnd in his Interconnect model terminals and not use Node
0 in any of his IBIS-ISS files.
* Use A_gnd in his Interconnect model terminals or use Node 0 in any
of his IBIS-ISS files.
* The currents that go to the component "ground pins" (as defined in
the Component Data Book)" may not account for the currents that go to Node
Almost universally, package interconnect, board interconnect and connector
interconnect SPICE models use Node 0, and almost universally power aware
simulations are "ground base".
I would like to make the following motion:
1. IBIS 189 should state that A_gnd is the Simulator Reference Node
(Node 0 in IBIS-ISS)
2. We add the following Note:
a. Using A_gnd as Interconnect Model Terminals or Node 0 in IBIS-ISS
subckts may not account for all currents going to "ground" pins, and
therefore potentially introduce simulation results when doing power aware
simulations that are not ground based.
Anyone can write a supporting document on why it is better to make models
that do not use A_gnd or Node 0. This does not belong in the BIRD, but can
be submitted as a supporting document to the BIRD.
978.461-0449 x 133