IBIS war stories Re: [SI-LIST] : response to semiconductor I/O edge

Laurence Michaels (lmichael@techmail.gdc.com)
Fri, 16 Jul 1999 14:12:25 -0400

"D. C. Sessions" wrote:
>
> Roy Leventhal wrote:
> >
> > Good points about etch length and edge rates.
> >
> > Somewhere else in the discussions the issue of programmable drive was also
> > brought up. We are using progarmmable drive strength and a fixed series source
> > terminator (selected by daughter card loading) in at least one instance.
> >
> > Does anyone know a signal integrity simulator that is up to snuff on IBIS 3.2
> > especially with regard to the conventions on programmable devices and has
> > automated the handling of such parts?
>
> The EDA members of the IBIS committee report that their managements have
> assigned IBIS 3.x support a low priority due to lack of customer demand.
>
> --
> D. C. Sessions
> dc.sessions@vlsi.com
[rant type="teaching IC support about IBIS and SI" alt="preaching to the
choir"]
Argh... I'm having a hard enough time getting IBIS models from
manufacturers NOW.
If I hear one more "What diode characteristics are you talking about?
Our output drivers don't have any diodes in them." Err... Yeah? No
Substrate-bias diodes? Then what are these flat areas in the over- and
under-shoot regions of the signal?
Mice?

Now they have yet another excuse. "We don't have enough demand for IBIS
models, and the EDA tool vendors don't support 3.x."

Do I give up and just make major simplifying assumptions, without
knowing exactly how that would affect accuracy? Is it time for DFWTMTP?
(Design For Way Too Many Termination Parts)

It would be nice if manufacturers included all the data necessary to
generate IBIS models inside the data book. Several come close, but not
quite all the way, by including driver and pullup VI curves. Almost
none of them spec resistive rise time, so I have to pick something that
seems reasonable until I get a good answer from the vendor (possibly
after teaching them what I'm asking for). I've seen vendors spec "Slew
rate limitted" drivers without saying what they limitted it to. Those
are a little difficult to model.
[end rant]

I'm assuming most people who've attempted to use and acquire IBIS models
have had similar experiences. True?

-- 
Laurence Michaels
For more help with your frame of reference,
please contact a certified quantum mechanic.

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****