Walter,
This was a semi-private distribution, but now sent to ATM. (Original
proposal attached.)
I have not had time to really study this, but I do have some issues. Sorry
about the lack of detail
1. The terminals for DC I-V extraction that excludes package effects
are different
2. Ext_ref is always related to the [External Reference] terminal (or
node). It is not related to [External Model]. The connection terminals for
[External Model] are already associated with the [Model] terminals
3. I still have problems with a default search for 0.0 V rail rule for
C_comp. I would suggest stating that different EDA tools may have different
ways of handling C_comp including these possibilities:
a. Default to global ground "node 0", but possibly risk not doing
power-aware analysis properly
b. Connect to a node whose Rail is 0.0V
c. Split C_comp into equal components connected to all unique buffer
rail terminals
d. Connect to the Pullup_ref for ECL Outut and I/O ECL models
e. Handle I/O Open devices with special rules since that may not be
significant capacitance to a 0.0 V terminal
f. Etc.
Mandating one solution may not be technically correct. C_comp is still
EXTRACTED as the total capacitance for the buffer to ALL ideal rail
terminals, and it is equivalent to having its value connected to GND.
Simulation is different.
We already have a solution for explicating splitting C_comp with C_comp*
components in IBIS.
I think a reference terminal can be different than any default C_comp
terminal or terminals.
However, while Radek and I discussed "[Local Reference]", I do like the
sub-parameter approach of Reference_terminal.
Bob
From: Walter Katz [mailto:wkatz@xxxxxxxxxx] ;
Sent: Sunday, March 19, 2017 11:37 AM
To: mlabonte@xxxxxxxxxx; Mirmak, Michael; Bob Ross; Muranyi, Arpad
Subject: RE: Reference_terminal BIRD
All,
Just upgraded to Office 2016, and have trouble using my old contact list.
The proposed BIRD this time.
Walter Katz
<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx
Phone 303.449-2308
Mobile 303.335-6156
From: Walter Katz [mailto:wkatz@xxxxxxxxxx] ;
Sent: Sunday, March 19, 2017 2:27 PM
To: mlabonte@xxxxxxxxxx; Mirmak, Michael <michael.mirmak@xxxxxxxxx>; Bob
Ross (bob@xxxxxxxxxxxxxxxxx) <bob@xxxxxxxxxxxxxxxxx>; Muranyi, Arpad
<Arpad_Muranyi@xxxxxxxxxx>
Subject: RE: Reference_terminal BIRD
Walter Katz
<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx
Phone 303.449-2308
Mobile 303.335-6156
From: Walter Katz [mailto:wkatz@xxxxxxxxxx] ;
Sent: Sunday, March 19, 2017 2:24 PM
To: 'mlabonte@xxxxxxxxxx' <mlabonte@xxxxxxxxxx>
Subject: Reference_terminal BIRD
All,
I think we need to address the reference terminal with a BIRD, not just
Editorial changes. I think this deals with legacy IBIS models (what happens
if you change an existing 6.1 file to 7.0, and deals with the special cases
of ECL, MECL, and PECL (Bob might want to introduce some rules for ECL,
MECL, and PECL when Reference_terminal is not specified).
We can put this on the Tuesday agenda.
Walter
Walter Katz
<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx
Phone 303.449-2308
Mobile 303.335-6156
Attachment:
Reference_Terminal.docx
Description: application/vnd.openxmlformats-officedocument.wordprocessingml.document
