Yes but not at the actual event.
Had once a conversation with him and Jim Williams (who hated math, LaPlace analysis)
that Spice would eventually model correctly. Even though manufacturers purposely
releasing incomplete models to mask proprietary innovation. Still a problem today
I think. Good example of that is RR input OpAmps where the crossover behavior is
not modeled. EG distortion numbers will be off.
My discussion with them was coming off an experience where as a production engineer
I wrote a Fortran program to analyze yields and wafer ET data, and later test machine results.
In that activity my boss was against computers, and I had to do this at night. Upon com-
pletion I presented results that led to process changes and fab investigations that ultimately
improved yields. On the test side I found a machine that consistently had issues, that resulted
in a > $ 1M pop in revenue for the month for fallout parts I was stashing under my desk. Recovered
most of them because of repaired tester results. I was dinged for having all thes parts under my
desk. Discouraging so much resistance to change by fellow engineers. Later as a field engineer I
saw that on a regular basis at various accounts, some EEs not able to move with technology.
I was in central applications at the time of spice discussion. My boss there hugely progressive.
Jim Moyer. He had a friend, Dale Mrazek, inventor of Tri State and lead manager on National
2900 bit slice, and Barry Siegal, manager hybrid linear ICs (high performance stuff), and I would
get to go on lunches with them occasionally. Kept my mouth shut and learned tons. Dumb luck
to be surrounded by folks like this. The one person I missed (in my opinion) more contact with was
Tom Frederiksen , great insight (his books reflected that) into semiconductor physics and
IC design.
Regards, Dana.