The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Does tnom in netlist options override tnom setting in models files?
https://designers-guide.org/forum/YaBB.pl?num=1264114803

Message started by CircuitNerd on Jan 21^st, 2010, 3:00pm

Title: Does tnom in netlist options override tnom setting in models files?
Post by CircuitNerd on Jan 21^st, 2010, 3:00pm

Which tnom setting takes precedence in the Spectre simulator - the one you set in the .options statement (top level netlist), or the value set inside the models files (called by .include statements in the top level netlist)?

Title: Re: Does tnom in netlist options override tnom setting in models files?
Post by Andrew Beckett on Jan 22^nd, 2010, 8:01am

The tnom on the model has higher precedence (for that model).

If you look at spectre -h bsim4 (say), you'll see:

Code:

326 tnom (C) Parameters measurement temperature. Default set by
options.

and spectre -h options:

Code:

6 tnom=27 C Default component parameter measurement temperature.

So put another way, the temperature at which the model was characterized is specified by tnom in the model, and if that's not set, by tnom in the simulator options, and if that's not set, it will be 27.

Regards,

Andrew.

Title: Re: Does tnom in netlist options override tnom setting in models files?
Post by CircuitNerd on Jan 22^nd, 2010, 12:36pm

Thanks for your reply. Interestingly, it is EXACTLY THE OPPOSITE of the answer from my CAD group.

So, can you prove the validity of your assertion by citing any Cadence documentation (e.g. Spectre Reference Manual, pg. 72) or something similar? I personally believe you are correct, but if I can cite vendor docs to prove it, convincing the CAD armada will certainly be much easier...

Thanks!

Title: Re: Does tnom in netlist options override tnom setting in models files?
Post by Andrew Beckett on Jan 22^nd, 2010, 3:08pm

Given that parts of the documentation are built from the "spectre -h" text, you should find exactly the same definitions in the spectre manuals. I'm at home right now, so can't easily look it up, but I'm sure you can...

If you think about it, having the opposite behaviour would be rather strange - it would mean that your carefully characterized models, each characterized at a defined nominal temperature, would then get messed up by somebody overriding it globally.

It would be rather strange to have a global overrules local approach. Not unheard of though, I guess. HSPICE has (or least had, I'm not sure what the default is these days) a rather counterintuitive global wins over local parameter approach. I remember arguing this with Meta-Software nearly 20 years ago, and they then added the PARHIER parameter to change the parameter inheritance to be the more conventional local wins over global.

Regards,

Andrew

Title: Re: Does tnom in netlist options override tnom setting in models files?
Post by Andrew Beckett on Jan 23^rd, 2010, 8:51pm

Here's the proof (better than just referencing the manual, I got the simulator to tell me). If you use this netlist:

Code:

//
model withTnom bsim4 type=n tnom=50
model withoutTnom bsim4 type=n

MwithTnom (d g s 0) withTnom w=1u l=0.5u
MwithoutTnom (d g s 0) withoutTnom w=1u l=0.5u

opts options tnom=34

models info what=models where=screen

This (as you can see) has two models defined, one which explicitly sets tnom to 50, and one which doesn't have it set. Additionally I have the global tnom set in the options statement. Finally I'm using an info analysis to output all the model parameters.

Here's what it produces (irrelevant parameters chopped out for the sake of brevity) using the latest MMSIM71 ISR (what I had handy):

Code:

********************************
Model parameter values `models':
********************************

Model: withoutTnom
Primitive: bsim4
type = n
level = 14
...
fcsw = 500e-03
tnom = 34 C
trise = 0 C
ute = -1.5
...

Model: withTnom
Primitive: bsim4
type = n
level = 14
...
fcsw = 500e-03
tnom = 50 C
trise = 0 C
ute = -1.5
...

So that proves what I was saying - the global option only affects models which don't have tnom explicitly set themselves. Hope this helps to convince your CAD group!

Regards,

Andrew.

Title: Re: Does tnom in netlist options override tnom setting in models files?
Post by CircuitNerd on Jan 26^th, 2010, 8:15am

Well, that looks pretty conclusive! Thanks for your efforts.

CircuitNerd