Bird 28

Although I can see the reasoning behind trying to make simplifying
assumptions, I agree with Jon that it is too restrictive to require
that any coupling sections must be of equal length. I would also
caution Stephen against designing a solution around specific packages
(e. g., QFPs) that may not apply to other packages (yes, there are
still people out there using DIPs). I would prefer to have a more
complicated model that's accurate than a simple one that isn't. I
suspect that someone who's done a detailed analysis of a real package
will also disagree with the notion that pins never couple to other
pins from which they are separated by a single Ground pin. I have
not done extensive package analysis, and it may be a "noise" effect,
so perhaps that's a reasonable compromise for the sake of complexity
and compute time.

In my opinion, whatever method the group chooses to adopt should not
make any assumptions about the pin assignment as well (e. g., three
signals for each Ground, all the way around), because real packages
are seldom that orderly (and someone "out there" will come up with
another one as soon as the model is done anyway, if not sooner).

Jay Diepenbrock
Received on Thu Jul 20 05:37:49 1995