Re: IBIS models and i2spice

Arpad Muranyi wrote:
>
> Text item:
>
> All interested,
>
> I just came back from a week of vacation and read the EMAIL correspondence on
> the IBIS to SPICE conversion issues with great interest. I agree with the
> statement that there is a great need for IBIS support in the SPICE tools.
> Ideally, I would like to have a SPICE/behavioral combo tool that has the
> advantages and capabilities of both worlds. If needed, this tool could run
> transistor level SPICE models for detailed buffer design work, or higher level
> behavioral simulations for fast interconnect design work. Due to this double
> feature, this tool would also cover the situation when one has both types of
> models which need to be simulated together. Unfortunately, a standard would
> still be required for covering the various flavors and levels of transistor
> descriptions.
>
> I vould like to make a couple of comments on what I have read in the
> correspondence.
>
> 1) IBIS actually started on HSPICE when I wrote behavioral models (for internal
> use) using HSPICE. Just because a specific SPICE tool vendor does not support
> IBIS, it doesn't mean that the user cannot write behavioral models that use IBIS
> data. Bob Ross gave an excellent example for how to do this in one of the most
> recent replies to this subject. Most SPICE flavors have behavioral elements
> (controlled voltage and current sources) which can be used to create a
> behavioral model. However, the support from the SPICE tool vendor is still
> badly needed because of the problems that arise with the usage of these
> elements, such as non convergence, time step too small errors, etc...
> Specialized features (such as event triggered sources to detect a threshold
> crossing, for example) would significantly improve behavioral modeling. Of
> course, the ideal solution would be if the user would not have to develop the
> behavioral model at all, just plug in the IBIS data into something that the
> vendor provides.
>
> 2) Another reason I use IBIS models under a SPICE tool is actually something
> that I have not read in any of the replies. The problem I see with most (if not
> all) of the IBIS simulator tools is that they all lack substantial amounts of
> freedom. In my work I have to specify buffer characteristics as well as
> interconnect guidelines. For this I run sweep simulations on many of the buffer
> as well as interconnect parameters. I like to plot the simulation results in
> many different ways, such as flight time vs. buffer strength, amount of cross
> talk vs. signal pattern, etc... to make the data meaningful. Many tools don't
> have variable sweeping or scaling capabilities, scaling factors for altering the
> I-V curves of the IBIS models for sweeps, or arbitrary measurement definition
> capabilities, and have nothing but a voltage vs. time (waveform) plotting
> capability. Some tools provide some of these features, but in very primitive
> shape and form. Even though it is painful some times, I can do most of these
> tasks fairly well in HSPICE, but I could still use better features and support.
>
> To sum it up, IBIS tool vedors could eliminate part of the need for running IBIS
> simulations in SPICE tools by adding a bunch of much desired felixibility
> features to their tools which can only be done today in SPICE tools. These
> tools are only useful if they provide the necessary user interface, and
> automation features which are required for the work at hand and for being more
> productive. But I would still prefer a tool that can do both worlds together...
>
> Arpad Muranyi
> Intel Corporation
> ================================================================================
>
> Text item: External Message Header
>
> The following mail header is for administrative use
> and may be ignored unless there are problems.
>
> ***IF THERE ARE PROBLEMS SAVE THESE HEADERS***.
>
> Content-type: text/plain; charset=us-ascii
> MIME-Version: 1.0
> X-Incognito-Version: 4.10.130
> X-Incognito-SN: 1467
> Subject: Re: low budget IBIS tools
> Errors-to: <fabrizio=zanella%eng%emchop1@fishbowl02.lss.emc.com>
> Reply-To: <fabrizio=zanella%eng%emchop1@fishbowl02.lss.emc.com>
> From: "fabrizio zanella" <fabrizio=zanella%eng%emchop1@fishbowl02.lss.emc.com>
> To: "'ibis-users@vhdl.org'" <ibis-users@vhdl.org>,
> <owner-ibis@server.vhdl.org>
> X-Priority: 3 (Normal)
> Message-ID: <vines.8VJ8+NT8IoA@fishbowl02.emc.com>
> Date: Fri, 24 Oct 97 9:51:21 -0400
> Received: by fishbowl02.emc.com with VINES-ISMTP; Fri, 24 Oct 97 9:59:15 -0400
> Received: from fishbowl02.emc.com (fishbowl02.lss.emc.com [168.159.48.62]) by em
> cmail.lss.emc.com (8.8.7/8.7.3) with SMTP id KAA06131 for <ibis-users@vhdl.org>;
> Fri, 24 Oct 1997 10:04:52 -0400 (EDT)
> Received: from emcmail.lss.emc.com (emcmail.lss.emc.com [168.159.48.78]) by serv
> er.vhdl.org (8.8.5/8.8.3) with ESMTP id HAA12557 for <ibis-users@vhdl.org>; Fri,
> 24 Oct 1997 07:01:24 -0700 (PDT)
> Received: from server.vhdl.org (server.vhdl.org [198.31.14.3])
> by hebe.or.intel.com (8.8.6/8.8.5) with ESMTP id HAA18933;
> Fri, 24 Oct 1997 07:19:25 -0700 (PDT)
> Received: from hebe.or.intel.com by relay.hf.intel.com with smtp
> (Smail3.1.28.1 #2) id m0xOkbx-000qLxC; Fri, 24 Oct 97 07:21 PDT
We have looked into the enemy's face and the enemy is us. Such tools
alredy exist. The user community needs to look at the various SI
companies and solutions carefully! We are just one of several SI spice
based tools that have had IBIS models working together with Spice
transistors and coupled lossy transmission lines since 1991! WAKE UP!
Where have you guys been! We attend the various converences and
marketing shows for a REASON. In our particular case interfaces exist
into the most popular CAD products. These type of products have been put
to good use for quite some time by hundreds of companies!

-- 
Fred Balistreri
fred@apsimtech.com
408-434-0967 ext. 102