MCMs, SIMMs, and such

I agree heartily with Arpad. I see no advantage, and plenty of disadvan-
tages to extending IBIS to include PCB-like packages. Since the existing
tools handle such assemblies very nicely, why re-invent the wheel while
making the IBIS model hopelessly complicated? I like Stephen's concept
of allowing the model provider to use either lumped or matrix form as he/
she deems appropriate, but would much prefer to see the IBIS model used
as the lowest level of a hierarchical model, with what we refer to as the
"first level package" above it. In the case of a SIMM or MCM, the chip
level model is used with a different higher level package model that is
provided by the designer of the MCM, SIMM card, etc. with all the appro-
priate trace lengths, etc.

Now, as for Stephen's proposal itself, I may have missed something. If
I understand it correctly, Stephen proposes that self-inductance would
be in the inductance matrix, but he described "zeroing out" the mutual
terms. Was this just for a specific example, I hope, or is this to say
that all coupling terms are always to be capacitive? If the latter, I
would suggest that the model is inadequate. I do like other aspects of
the proposal, especially the physical correspondence of the framework.

Jay Diepenbrock
Received on Fri May 12 10:07:54 1995