Re: [SI-LIST] : Convert an IBIS model to SPICE Model.

Fred Balistreri (fred@apsimtech.com)
Fri, 03 Apr 1998 11:06:19 -0800

John Synesiou wrote:
>
> You bring up a good point, SPICE with IBIS extensions would allow
> standardization of syntax as well as the flexibility to do more detailed
> analysis if needed. I'm not sure if such a product exists, rather some SPICE
> vendors provide an IBIS to SPICE translator.
>
> Regards
>
> John Synesiou jsynesio@us-power.com
> U.S. Power, Inc Phone (612)826-1111
> 6497 City West Parkway Fax (612)826-1003
> Eden Prairie Date: 04/03/98
> MN, 55344 Time: 10:28 AM
>
> -----Original Message-----
> From: owner-si-list@silab.Eng.Sun.COM
> [mailto:owner-si-list@silab.Eng.Sun.COM] On Behalf Of Andrew Ingraham
> Sent: Friday, April 03, 1998 7:54 AM
> To: 'SI_LIST'
> Subject: RE: [SI-LIST] : Convert an IBIS model to SPICE Model.
>
> > Your solution is one way to solve this problem in SPICE, but a far
> easier
> > method would be to use a SPICE simulators that can do behavioral
> modeling
> > using lookup tables. Some examples are SPICE 3 B-elements with
> if-then-else
> > extensions, or XSPICE extensions (many SPICE vendors offer XSPICE
> > extensions).
>
> But many do not.
>
> Rather than choosing a SPICE with XSPICE extensions, wouldn't it be
> better to choose a SPICE with IBIS extensions and avoid having to
> convert anything yourself?
>
> (Assuming, of course that you even have the freedom to choose *any*
> SPICE in the first place!)
>
> Regards,
> Andy
I'm not sure what are IBIS extentions? IBIS is just a data sheet that
gives electrical information useful for SI work. Every vendor Spice or
other must then take that data sheet and convert it into a model. Now
some simulators take in table data like that produced by IBIS. And
believe it or not some of those vendors are Spice vendors. But actually
what your'e looking for is an SI solution. As I mentioned before there
is a proposal from EIA in Japan that essentially attempts to standardize
a table spice format. That format is very compatible with IBIS. In fact
IBIS would do well to adapt it. Many of the "birds" and issues of
flexibility, capability and accuracy would go away with this standard.
The standard describes the transistor in table format. No propietary
data is included. The difference between that spec and IBIS is that
this spec is general since its at the transitor level. For IBIS lovers
the data for the I/V is still retained so in essense IBIS becomes a
subset at least for the technical part. Then everybody would be happy
since Spice guys can still use the topologies allowed by Spice and the
transistor data would be there. The data shows nothing of process so
IC vendors would be happy. Current IBIS users should be happy because
the data they want would still be there and non spice vendors who do
not care for transistor level data could make use of the information
that's already in the spec. As those guys become more sophisticated in
their modeling technique they could start incorporating information
about the Gate for example. Which is a BIG problem in IBIS at the
moment.

Now what's wrong with this. Can anybody please explain to me why this
should not be implimented? Certainly from a technical standpoint this
would be a great standard. Lets here from both the IBIS and Spice fans
on this one.

Best Regards,

-- 
Fred Balistreri
fred@apsimtech.com

http://www.apsimtech.com