From: Bob Ross [mailto:bob@xxxxxxxxxxxxxxxxx]
Sent: Tuesday, September 1, 2015 2:42 PM
To: wkatz@xxxxxxxxxx; 'IBIS-ATM' <ibis-macro@xxxxxxxxxxxxx>
Subject: RE: [ibis-macro] C_comp, C_comp_*, C_comp*, [C Comp Corner]
All,
There are two issues that must be treated independently:
1. Whether to use the C_comps under [Model] or under [C Comp Corner]:
Do not mix [Model] C_comps with [C Comp Corner] C_comps if both are
present.
If specified, use only [C Comp Corner] if the EDA tool supports [C Comp
Corner].
Use [Model] C_comps if the EDA tool does not support [C Comp Corner]
Even if the EDA tool support [C Comp Corner], it could use only [Model]
C_comps as a tool-specific feature for range sensitivity analysis at each
analysis corner
So only 3. (for [Model] C_comps) or 4. (for [C Comp Corner] C_comps) are
correct are correct.
2. How to split C_comps or to assign C_comps when only partial information
is given
The specification unclear/open about this, and we can really get bogged
down making arbitrary rules.
In fact (for non PI analysis) an expectation is that C_comp can go to a
local ground, not to any buffer terminal; but that is another discussion.
Bob
From: ibis-macro-bounce@xxxxxxxxxxxxx
<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Walter Katz
Sent: Tuesday, September 01, 2015 9:03 AM
To: IBIS-ATM
Subject: [ibis-macro] C_comp, C_comp_*, C_comp*, [C Comp Corner]
All,
Bob, Arpad, and I had a very interesting e-mail thread on the meaning of
life (strike that the meaning of the following two lines in IBIS 6.0).
. "If [C Comp Corner] is present, its value or values override
any other C_comp* representations."
. "It is not illegal to include the C_comp subparameter together
with one or more of the remaining C_comp_* subparameters, but in that case
the simulator will have to make a decision whether to use C_comp or the
C_comp_pullup, C_comp_pulldown, C_comp_power_clamp, and C_comp_gnd_clamp
subparameters."
Based on these two line, what do you think the model maker expects the EDA
tool to do when encountered with the following Model when running the Min
(aka Slow, Weak Corner:
[Model] C_comp
C_comp_pullup 2.0p 1.0p 3.0p
[C Comp Corner]
C_comp_pulldown 4.0p 8.0p 2.0p
There are the following possible expectations:
1. Capacitor to A_puref 1.0pF
2. Capacitor to A_puref 3.0pF
3. Capacitor to A_puref 3.0pF|1.0pF (Note treat
C_comp_pullup as a range and try both extremes)
4. Capacitor to A_pdref 8.0pF
5. Capacitor to A_pdref 8.0pF Plus Capacitor to A_puref
with the values in expectations 1., 2. or 3.
If we can all agree to one of the above 5 possible expectations (it is #
4), then we can write a BIRD to clarify the above text in IBIS 6.0.
We should at least publish to the IBIS model making community that if the
model is to communicate accurately what the model maker expects the EDA
tool to do, there should be no [Model] C_comp* sub-parameters, and the
model maker should only use C_comp_* sub-parameters.
[Model] No_Model_C_comp | Is this illegal?, It is Illegal!
[C Comp Corner]
C_comp 4.0p 8.0p 2.0p
[Model] No_Model_C_comp | Is this illegal? , It is Illegal!
[C Comp Corner]
C_comp_pulldown 4.0p 8.0p 2.0p
[Model] With_Model_C_comp
C_comp 2.0p 1.0p 3.0p
[C Comp Corner]
C_comp_pulldown 4.0p 8.0p 2.0p
Make following change
. "If [C Comp Corner] is present, its value or values override
any other C_comp* representations."
to
. "If [C Comp Corner] is present, its value or values override
all [Model] C_comp, C_comp_* representations, regardless of whether the
sub-parameter sets are the same or not.."
Walter
Walter Katz
<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx
Phone 303.449-2308
Mobile 303.335-6156
