Package Model Arguments

To IBIS Committee:

The three approaches to the package model extension problem are being
discussed:

  (1) physical extension based on mechanical layout,
  (2) electrical extension based on Spice Subcircuit methodology, and
  (3) electrical extension based on IBIS format extension.

(1) The physical extension seems to be for future consideration since it
requires development and agreement on a whole new approach within IBIS
or adoption of another standardized approach.

(2) I believe that the Subcircuit extension provides sufficient functionality
and I would be interested in it being developed more formally into a BIRD. In
this way some additional format detail issues and constraints may then surface.

(3) Good work and refinement has already gone into the IBIS format extension
proposal, and it appears ready for a BIRD.

Here are some arguments in support of the IBIS format extension approach:

(a) It provides a very robust, neutrally formatted feature set compatible
with all simulator communities. It is probably easier to develop translation
utilities to convert this format to variations of Spice than it is to
convert general nodal formatted data into formats used by several simulator
companies which use IBIS directly.

(b) I believe the matrix data formats are already more compact than Spice
syntax structures.

(c) The structure lends itself to validating correct topologies via
syntax checking extensions to IBIS_CHK. The advantage of flexability and
extensibility of the nodal appoach is actually a disadvantage for checking.

(d) It avoids the issue of which maximally compatible subset of Spice
syntax to use, or what (incompatible) extensions to propose.

I still remain open on this issue. I would like to see some completely filled
out numerical examples for small sized PGA and flatpack packages with
coupled matrix sections using the proposed IBIS format extension. I would
also like to see proposed Spice formatted structures to represent the same
data.

Bob Ross
Interconnectix, Inc.
Received on Wed May 10 20:27:49 1995