[SI-LIST] : Paper on IBIS simulation algorithm

Raymond Chen (sigrity@yahoo.com)
Thu, 02 Sep 1999 12:07:19 -0700

Andy and Larry,
We did some studies on IBIS simulation techniques
and algorithms back to 1996, and we did publish a
pretty detailed paper on that topic at 1996 ECTC
conference. You can download that file from our web
page at http://www.sigrity.com/support.htm, which is
the 11th paper titled as "Extraction of Transient
Behavioral Model of Digital I/O Buffers from IBIS".
That paper pointed out how to implement a simulation
algorithm to extract the transient behavior of ibis
model from the static I-V curve and V-T curve so you
can later simulate different load situations without
knowing the transistor level details. Hope that will
help.

Raymond Chen

--- "Ingraham, Andrew" <Andrew.Ingraham@compaq.com>
wrote:
> Larry Forsythe wrote:
>
> >I'm confused about how IBIS is used in actual
> simulations. I haven't seen
> >any articles about this (please indicate if there
> are some).
>
> And you might not find any that go into much detail.
> The actual algorithm
> used by each simulator, especially as to how it
> handles the V-T data, might
> be a guarded secret.
>
> > What I'm
> >confused about is how can a few static I-V curves
> and some V-T curves done
> >at specific loads be used to simulate other loads.
> The I-V curves involve
> >fully on transistors. The V-T curves involve
> partially turned-on
> transistors
> >driving a specific load.
>
> Typically, an output buffer interacts with its load
> over a period of several
> nanoseconds while the signal rattles around and
> eventually settles. Yet the
> time it spends actually changing -- the time for the
> drive to its output
> transistors to switch -- is much less, perhaps a few
> hundred picoseconds.
> It is a fair approximation to treat the output
> buffer as if it were static
> (defined by one I-V curve or the other) almost all
> of the time. In between
> the two, it makes some sort of transition from one
> I-V curve to another, in
> a relatively short interval, the effects of which
> might be swamped by the
> interaction with the load.
>
> > Is it valid to scale the I-V curves using the V-T
> >curves in order to derive the response to driving
> other loads?
>
> Here's where you get into the proprietary details of
> each simulator's
> vendor.
>
> If you're lucky, the load for the V-T curves is
> close to the trace
> characteristic impedance.
>
> > Can the I-V
> >curves and V-T curves be used to determine the
> characteristics of a
> >partially turned on transistor driving a different
> load-I guess that is the
> >source of my confusion.
>
> To varying degrees, apparently yes.
>
> Regards,
> Andy Ingraham
>
>
>

===
-------- Currently on leave for Intel projects --------
Office Phone: 253.371.4261 Mobile Phone: 408.718.3080
-------------------------------------------------------
Sigrity, Inc. Tel: 408.377.2180
2105 Hamilton Ave. Suite 310 Fax: 408.377.2565
Santa Clara, CA 95125 Web: www.sigrity.com
______________________________________

**** 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 ****