spice,spectre differ why

Sir,
I tried the simulation of PMOS for its current characteristics in
SPICE,spectre using spice Netlist reader and also in cadence format.
I compared the simulation results of all. Each simulator results
doesn’t match with one another. Even the spectre results using
Netlist reader and cadence format.

Why these differences occur?

.How to tune the tool to produce desired results?

How to find what parameter values of pmos devices are taken for
calculation?


Help.

With regards,
V G Ramnath

Remove XXX and YYY to mail.

 

which company wrote SPICE? And version?

What is Netlist reader format?
What do you mean by cadence format?

Are you using the same models for both tools? And what model are you
using? bsim3?

By using proper model parameters for your transistors, by using proper
options for the simulator.

in spectre you can use spectre -h model to find a list of models that
spectre understand. then you can use spectre -h <model> (example:
spectre -h bsim3) and it will list documentation on use of that
particular model.

In spice3 normally a list of all parameters used are listed in the
output file, but you don't say what kind of spice simulator you use.
There are so many out there.

 

How to find what parameter values of pmos devices are taken for

I don t think this is what he meant with this question. Let me
reformulate how I understood it.

If you have a modelfile like this:

nominalwidth=0.21
someunderetcheffect=0.003
finalwidth=nominalwidth-someunderetcheffect
subckt mymosmodelcir (1 2 3 4)
model themosmodel bsim3v3 type=n tnom=22 version=3.2 tox=2.222n
toxm=2.222n
instanceofthemodel (1 2 3 4) w=finalwidth
ends mymosmodelcir


the question would be, how do you print the numerical value that is
taken by "w" in the core bsim3v3 model of the FET.
I would know how to peek in the psf file, but I don t know how to do
that directly with spectre options.

a related question: can you ask spectre to print the value of
"finalwidth" also ?

an a personal request, too: in a spectre modelfile with a process
section (but no worst-cases ), how can you have spectre check that all
model parameters will stay in their valid range when process params are
at + or - 6*sigma ?

 

You'd use spectre's "info" analysis. ADE adds several info analyses to the
outputs, and stores the results in psf format, but you can do things like:

myInfo info what=models where=logfile

Do "spectre -h info" for more details.

If you're getting different results, I can only assume that something is
different between your models - since it's only a difference in the language
syntax on the front - the simulator is the same.

Andrew.