Minutes of 6/27/2000 IBIS-X teleconference:

We got off to a slow start with the teleconference service not
recognizing our ID. I hope everyone who tried to attend got my
emails so I could call them. Attending:

John Angulo
Al Davis
Mike LaBonte
Arpad Muranyi
Stephen Nolan

We discussed the updated macro language documents sent by Al Davis.
This consumed all of the meeting time, and the component pin mapping
proposal from Mike LaBonte was not discussed. The notes below are
not in chronological order, but is rather organized by topic.

Arpad registered general concern about the emphasis on remaining bound
tothe capabilities of existing simulators. There was a well-received
discussion about having level 1 and level 2 languages defined now,
where level 1 is probably implementable using existing simulation
technology, and level 2 may take a few years. Some of the issues:

- Should we have frequency dependent resistors, etc?
- How about temperature dependent devices?

Al has provided sample models in the new language, which show that
existing IBIS models can be described. Although the language is
designed to describe new technologies more easily than the current
IBIS standard, Arpad is concerned that a more comprehensive (level 2)
language with new simulation technology may be required for newer
models.

Equations vs. tables? The language has table data and four arithmetic
operators, but no functions. Arpad would prefer to have a powerful
expression language. This may add work to the implementation, and
there was concern over 

Should we specify the interpolation algorithms for PWL table data?
Maybe. Stephen Nolan asked if there is an algorithm to reduce a curve
to the minimum number of points. Model makers need this, but the
accuracy of the resulting curve depends on the PWL interpolation
algorithm. One idea is to mark curve data as to which algorithm
was assumed in the production of the data. On the subject of curve
extrapolation, Al Davis proposed that time curves should fall flat
off the ends, others should continue at the final slope.

On the subject of circuit element parameters, Al stated that the diode
model should accept the SPICE 2G6 subset of parameters, to approximate
real diodes. Although there have been requests for a "real" diode
model capability in IBIS, it is not certain that model makers would
have the required parameters anyway. Also, given that IBIS clamps
have the TT parameters, there was no conclusive answer to the question
"what is wrong with IBIS diodes?". There was general agreement that
subckts must accept parameters. To proceed further, details like this
must be filled in for all of the circuit element types.

Action items:

- Al: write more detailed descriptions of the integrator and differentiator
  elements, which we will discuss at the next meeting.

- Next teleconference: Tuesday July 11, 2000 at 10:00a PST.

Mike LaBonte