RE: [SI-LIST] : Best type of models, edge rates & load

Roy Leventhal (Roy_Leventhal@mw.3com.com)
Mon, 26 Jul 1999 13:37:27 -0500

Tom,

I believe you are correct in how the supply voltage is set up - whether measured
or simulated -. It is set at Vcc. But, the V-I curves are measured from -Vcc to
+2Vcc because that is the full range of incident + reflected voltage that
devices can see on a transmission line where reflection coefficient can run from
-1 (short) to +1 (open). I recollect that a voltage source of -Vcc to +2Vcc is
used to drive the I/O and the current into or out of the device is measured or
simulated.

The range of -Vcc to +2Vcc is much greater than the measurement conditions that
historically pervailed when measuring (say) saturated switching times into a
lumped, discrete, non-transmission line load. The accepted convention I used at
Fairchild and National in the early 1970's was 10% to 90% for
rise/fall/delay/storage time for any JEDEC registered 2N data sheet.

If you are correct about the 20% - 80% convention coming from Motorola it may be
an example of specsmanship supplanting accepted convention. Remember that JEDEC
became roundly ignored by the semiconductor manufactuers (in that same time
period) because they found it "too constraining." It's only my opinion, blurred
by time, but I recollect Motorola being in the lead in subverting the JEDEC
conventions. Semiconductor companies started issuing their own "registered"
(yeh, right!) data sheets that allowed them to, among other things, source
completely different die/mask set product lines (with VERY different starting
resitivities, etc., and VERY different Beta-Vs- Ic rolloff curves) so long as
they met a few, fairly wide DC parameter windows. Talk about die shrink! Guess
what, such substitutions sure didn't work in the actual designs.

So, what's your point about the 30% - 80% "transition time" convention where you
correctly point out that it's less than half the output swing? I can dig out the
data sheet and manufactuer's name for you if you wish. And yes, I think it's
pure specsmanship and quite misleading. If you just skim over the data sheet
numbers well, Caveat Emptor.

But, I would guess that the 20% to 80% IBIS convention was chosen to fit in a
more linear range of the I/O characteristics. I'm not sure that the choice has
been a good one.

Best Regards,

Roy

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****