Re: FW: EIAJ home page

From: Fred Balistreri <fred@apsimtech.com>
Date: Tue Apr 14 1998 - 12:22:15 PDT

Charles Hymowitz wrote:
>
> Thanks John. I looked at the EIAJ prED -5302 standards page on I/O
> Interface Model Project Group. The pdf file for the standard doesn't
> have all the fonts embedded so its kind of hard to download but
> attached is the letter I sent the webmaster. What do you (and
> others) think of my conclusion??
>
> Dear prED -5302 Standards Webmaster
> We are about to post a free utility to convert IBIS models into SPICE
> models on our web site; a very important utility that is missing from
> most SPICE users software. So we are very interested in the SPICE
> side of things. We would, in future, like to support the I/O modeling
> standard.
>
> But one big problem, unless I am missing something with regard to the
> proposed syntax, is that you are using non-standard Berkeley SPICE
> syntax for the MOSFET element. I don't know who proposed this but it
> won't work with anyone's SPICE simulator (except the one who proposed
> the syntax I imagine)
>
> Therefore this will slow acceptance of this standard dramatically
> unless one of three things happen:
>
> 1) someone modifies the MOSFET code or B element behavioral model code
> in Berkeley SPICE 3 to accept the syntax (then every SPICE vendor
> could adapt the free Berkeley code into their simulator) or
>
> 2) someone creates an XSPICE model (public domain AHDL available from
> Georgia Tech Institute) using C code (again, then every SPICE vendor
> could adapt the XSPICE code into their simulator)
>
> 3) someone creates a utility to convert the prED-5302 model code into
> a SPICE 2G.6 or SPICE 3 style model.
>
> The problem with 3) is that you can't easily do table models in
> SPICE2G syntax which is the only compatible standard SPICE syntax, and
> SPICE 3 doesn't support a table model. Therefore unless 1) or 2) are
> accomplished it appears that this "standard" will not become one.
>
>
> Charles Hymowitz - V.P. Prod. Dev.
> Intusoft
> email: charles@intusoft.com, CompuServe: intusoft@compuserve.com
> Web: http://www.intusoft.com
> Tel. (310) 833-0710, Fax (310) 833-9658
> P.O. Box 710, San Pedro CA. 90733-0710
>
> Please visit our web site @ http://www.intusoft.com for more information
> and lots of SPICE goodies.

The syntax is a proposal that can unify and null out some of the
objections IC vendors have in giving out Spice models. It is true that
this is a new model and Spice vendors will have to impliment this in
their particuliar brand of Spice. Once done the model is portable in
any Spice that supports it. And there lies the advantage. There is no
standard for Spice. Therefore Berekley models are supported as they
evolve but not in the Berekley form. Hspice routinely modifies both
the model parasitics and syntax for example. They then may "fix" some
bugs and whala a non-standard model that only Hspice can support. There
are certain advantages in everybody's Spice. And I'm sure EDA vendors
want to keep it that way to distinguish and differentiate themselves.
This proposal if adapted can become a standard for Spice vendors. It
will take some work on the part of the Spice vendors just like it took
work to support IBIS models in any Simulator. There are Spice vendors
who do support IBIS by the way. Again each Spice simulator takes
advantage of the features implimented in their Spice to accomplish this.
Bob Ross posted an example IBIS Spice behavioral model that can be used
in most generic Spice programs.

Best Regards,
Received on Tue Apr 14 13:05:58 1998

This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:53:46 PDT