Re: [SI-LIST] : IBIS Model Verification

Weston Beal (beal@lisbon.eng.hou.compaq.com)
Wed, 9 Sep 98 14:28 CDT

> From silab.Eng.Sun.COM!owner-si-list Wed Sep 9 14:01:02 1998
> From: Abe Riazi <ariazi@anigma.com>
> To: "'si-list@silab.eng.sun.com'" <si-list@silab.Eng.Sun.COM>
> Subject: [SI-LIST] : IBIS Model Verification
> Date: Wed, 9 Sep 1998 11:21:48 -0700

> Hi Everyone,
>
...
CHOP
...
> Before using an IBIS model in a simulation, first I check the model for
> possible syntax /format errors using
> the ibis_chk program, followed by an examination of the I-V data and
> rise/fall times for monotonic behavior (with
> the aid of the Visual IBIS Editor software). However, this level of
> evaluation can prove insufficient in some cases.
>
> Therefore, I am interested to learn about procedures and tools which
> allow a comprehensive verification of an
> IBIS model, thereby minimizing the possibility of unreliable simulation
> results.
>

Abe,

Valid point! I am an IBIS consumer too. And I'm getting tired of
messy, incomplete, and blatantly wrong IBIS files. I do the same as
you. I run ibis_chk and then look at each VI and VT curve with
ibisplot. About 90% of the files I receive require some fixing on my
side. It appears that many people are creating IBIS files with very
little understanding of the specification. Just because s2ibis ran
without errors does not guarantee a good IBIS file! I think the best
we can do right now is to tell the suppliers about the errors that we
find in their files. Most of them want to learn how to do it right,
and will gladly accept suggestions when presented tactfully.

Here are some of the flags I look for:
s2ibis 0.91BETA (!!! WARNING !!!)
no Vmeas, Cref, Rref, or Vref
clamp curves left in pullup and pulldown (s2ibis)
pullup and pulldown curves not crossing 0,0
2 significant digits (s2ibis) causing stairstep curves

I don't want to take away from the massive amount of work done in
developing s2ibis. It can be a great tool if you know how to use it.
Don't just run it without understanding what it does. I changed some
things I didn't like in the code and use it internally when I have
spice models.

I was saying just yesterday that we need an ibis_fix program to look
for the common errors and try to fix them. EDUCATION is the only real
answer.

Regards,
Weston Beal

> Thank you in advance for your response.
>
> Best Regards,
>
> Abe Riazi
> SI Engineer
> Anigma, Inc.