Re: [SI-LIST] : Modeling Package parasitics

Mattan Kamon (matt@memcad.com)
Tue, 17 Feb 1998 11:55:19 -0500

With some side mention by Greg Fitzgerald of the MIT parasitic extraction tools,
let me take this opportunity to "announce" the MIT inductance tool, FastHenry.
The inductance work was a significant part of my graduate research and the tools
are completely free, so I hope this appropriate for this list.

I sent mention to this list in 1994, but since then the code has changed and may
be more useful to people. The most recent version is dated Nov. 1996.

FastHenry is a 3-D, Partial Element Equivalent Circuit (PEEC) based inductance
code which uses an iterative algorithm and the Fast Multipole Method to rapidly
solve for the terminal (n-port) lumped resistance and inductance. It has a simple
text based interface for defining wire-like structures and reference plane
structures. Some limited ability for mesh planes and arbitrary via holes is
provided.
A postprocessor is available to convert the output to a spice3 equivalent circuit
for either R and L at a specific frequency (easy to use) or over a wider frequency
range (a little harder to use).

It comes with a manual, and compiles on most Unix platforms.
You can get the code from the directory

ftp://rle-vlsi.mit.edu/pub/fasthenry

and the file is fasthenry-3.0-12Nov96.tar.Z (see the README file at the site for
details.)

I spent a good deal of my time taking this from a research project to something
usable by others, so the more people that use it, the better. :-)

Other tools (in which I played no part) at the site include the 3D capacitance
extraction code, FastCap,
available at ftp://rle-vlsi.mit.edu/pub/fastcap. Most recent modification date
there is 1992. This resulted from the graduate work of Keith Nabors.

Hope this is useful,

Mattan Kamon
matt@memcad.com (Now)

Gregory P. Fitzgerald wrote:

> There are several tools on the market that will allow you to model package
> parasitics. Assuming you don't have access to Pacific Numerics (offers
> tools specifically designed for modeling packages) or Ansoft 2D/3D, or any
> of a host of other parameter extraction tools (some are even available on
> the web
> from MIT I'm told), calculation is the next best thing.
>