Hello,
Kellee makes a good point:
> Why create yet another method of describing the package when we have a
> method ala
> Steven that extends what we alredy have and performs the job very nicely.
> Am I missing something here?
The problem as I see it is that the Egg proposes one way to model
an MCM which may do for some particular part, but may be inadequate for
some other part. We do not want to add to Ibis a plethora of syntaxes
to support many different package model options.
The matrix package models for pin parasitics are sufficient for traditional
packages. For anything more complicated a different approach is needed.
In general an MCM can be as complicated as a board and requires its
own modeling tool. If such is available then it should be used along
with the models for traces on the board (Cadence SigNoise for example
is currently used in such cases).
Lacking the actual MCM models or the capability of mixing them with
the board models then some other technique is needed. But, that technique
should be as general as possible so that no further extensions in this
direction are required.
One possibility is to use the pin parasitics type package models
(the matrix approach) for the actual pins of a package and use spice
subcircuits with a naming convention to define what pins they attach to.
Such an approach would be perfectly generic, would NOT force unneeded
pins into the circuit, and would never need extension. The Spice elements
used could be restricted to passive elements.
The advantages to this approach are:
1) Everyone understands the SPICE syntax,
2) With only passive SPICE elements there are no questions about models,
3) Extensions would NEVER be needed for any package since every possible
package could be described in this syntax,
4) Automatic software would prune the matrix of pin parasitics to
model only those pins that are needed,
5) SPICE subcircuits that are not needed are simply ignored by the
simulator.
Chris Reid
Received on Tue May 9 06:27:09 1995
This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:52:29 PDT