[SI-LIST] Re: SI models at MGH speeds

Sunil,

There could be various reasons for not being able to
generate "accurate" IBIS models for higher frequencies
but it is not the IV curves.  IV curves basically contain
the DC characteristics of the buffer, which are easy to
model accurately.

1)   One group of inaccuracies could be the so-called
second (or higher) order effects in the buffer, which
may be non linear, frequency dependent, voltage dependent,
etc...  One such example would be the C_comp.  In IBIS
it is a constant for all voltages and frequencies, but
in reality it is not.

2)  The other group of reasons revolves around the circuit
design (structure) of the high speed buffers.  We have
pre-emphasis and similar fancy buffer designs now, which
can get tricky to model in IBIS simply because of the
rigid assumptions IBIS keywords make about the buffer's
internal circuitry.  There are keywords which allow some
room for working with such buffers, for example you can
use the [Driver Schedule] keyword for a pre-emphasis=20
buffer, but you may not be able to do everything well.
The *-AMS language extensions were added to the IBIS
specification to remove this rigidity from the specification.

Cadence's macro modeling suggestions can help this second
situation, because with some additional controlled sources
and behavioral elements you can overcome some of these
rigid assumptions built into the IBIS specification without
having to go to SPICE models.  But since their macro
models are built around the basic IBIS IV/Vt engine and/or
predefined controlled sources and delays, there are limits
on how far you can go with it when you compare this technique
with the *-AMS capabilities.

On the other hand, the reason people turn to encrypted SPICE
models is simply because they do not have the time (or don't
want to take the time) to figure out how to make behavioral
models.  If IBIS models were hard enough to make, writing
macro models and *-AMS models are even harder.  However,
SPICE may not be the solution either.  For one, it is slow,
but I also hear issues with its accuracy (contrary to the
popular belief that it is accurate).

You just need to know what you need, and what you are getting
in these models.  Unfortunately there are no black and white
answers, and no free lunches either...

Arpad Muranyi
Intel Corporation
---------------------------------------------------------------

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] =
On Behalf Of sunil.mekad@xxxxxxxxx
Sent: Wednesday, May 25, 2005 9:06 PM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] SI models at MGH speeds


Hi,

Can anyone tell me why it is difficult to generate IBIS models for
mutli-giga hz devices. Most of the vendors give us HSPICE models. Is it
something to do with difficulty in accurate characterisation of V/I
curve at such high frequencies (>5GHz)?

In any case are HSPICE models also the correct solution? Presuming such
interfaces also have pre-emphasis and de-emphasis. Cadence suggests the
use of macromodels for MGH applications however, is it possible to
convert HSPICE models to accurate macromodels for simulation in
SPECCTRAQUEST for >5GHz application?

Thanks
Sunil
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List FAQ wiki page is located at:
                http://si-list.org/wiki/wiki.pl?Si-List_FAQ

List technical documents are available at:
                http://www.si-list.org

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: