BIRD13 Comment

To Bob and IBIS Committee:

Some comments were already made and recorded in the minutes on BIRD13.
Here are two additional minor points.

(1) I feel that the clarification at the end of 2) and before 3) should
be restated or deleted. It concerns allowing a relaxation of the voltage
ranges for measurement derived data vs simulation derived data. I actually
support reasonable ranges for data regardless of the source and let the
simulators take care of any necessary extrapolations. Many simulators
automatically extrapolate data, but those that do not would have to provide
mathematically from the IBIS data the added points to, for example build the
piece-linear sources. However, based on the fact that the conversions to such
simulators currently do not require such extrapolations, there may be
objection to relaxing the data range requirements for only measurement derived
data. Either the data ranges would have to be relaxed for ALL data or
for NO data because the simulator does not know the source of the data.
As it stands, my current expectations are that models produced by measurements
(or simulations) may have to provide mathematically extended data points to
comply with the IBIS ranges. So the note could be rewritten to state this,
or else deleted, or the measurement range requirement be for guidance only.

(2) The proposal to add a note after step 7 that the rise/fall time
typical of actual circuit in operation, I feel, may go too far. For some
configurations, I do not think that a source ramp with slight rounding of
corners should have any effect on the output. Many inputs are in the
0 to 3V or 0 to 5V range, and the actual switching thresholds are only
several hunderd millivolts in the center of the ranges. So, while the
intent of the statement is good, I think it extends into an unneccessary
detail. Furthermore, there are some other technical considerations on how
sensitive the actual response is to the input rise/fall times, the source
impedance, etc,. and whether we are getting data for validation of specs
or whether we are getting MODELING data. So I would prefer that this
recommendation be deleted and expect that the manufacturers do what is
necessary to get the data best suited for modeling the part.

With the discussed corrections including using "load" notation of BIRD12.1,
the other aspects of BIRD13 look good to me.

Bob Ross,
Interconnectix, Inc.
Received on Sat May 7 19:47:35 1994