Subject: RE: [IBIS-Users] FW: Extreme currents????
From: Ingraham, Andrew (Andrew.Ingraham@hp.com)
Date: Thu Jul 11 2002 - 14:15:13 PDT
> For what it is worth I loved the model that had 900 AMPS at -3.3V.
900 amps ain't so bad, is it? It's not nearly as unrealistic as 1e25
amps. Do you think that 900 amps is the wrong number?
> How wide would the trace on the PCB have to be to carry this type of
load?
...
> It is obvious that
> model makers could do a better job in QA'ing their data.
...
> Perhaps
> extremely high currents should be changed to an error instead of a
warning.
> This would force correction to this issue at the source.
Folks, don't forget, these values in the models are supposed to
accurately represent the characteristics of the device; they are NOT
supposed to describe how you USE the device. I would hope no normal
user ever actually applies -3.3V with no current limiting to it. If you
did, you deserve the puff of smoke you get.
Just because a short wire is capable of momentarily carrying 20,000 amps
when you put 5 volts across it, doesn't mean that you can or should
actually use it that way.
Or that you can only connect them to enormously wide board traces.
Nor does it mean that you need to model every darn wire as a 10 ohm
resistor just to make sure the simulator can't possibly see that it's
capable of 20,000 amps.
If the SPICE model is correct (no unrealistic Rs=0 diodes, nothing
omitted, etc.) and it says 900 amps at -3.3V, then I claim the data
isn't unrealistic at all. It's just that the part can't support it for
long without vaporizing. But that's another issue entirely.
If measurements show that you can get huge currents before the part
blows up, then that data isn't unrealistic either.
Part of the purpose of having IBIS data that extends from -Vcc to
+2*Vcc, is to make sure the simulator can't possibly "fall off" the end
of the I-V table to where there is no more data, which may give it
nothing to force it back to normal voltages! It's not uncommon for a
circuit simulator to momentarily try outlandish voltages. It just
doesn't converge there. But it needs some data there, to tell it which
way to go. If the I-V data says only 300mA, because the IBIS model
creator doctored it, thinking that anything more than that was abnormal,
then the simulator may not have what it needs to nudge it back to where
it belongs. Worse, it might even falsely converge there.
We've been using SPICE models for ages. The SPICE model for my 1K
resistor says that it would conduct 1000 amps if I put a million volts
across it. Guess what? It's right. (A real resistor just won't do it
for long without blowing up.) The SPICE model for my diode or
transistor says that it would conduct 1000 amps if I put 5V across it.
Guess what? It's right too.
The only thing that makes IBIS models different, is that IBIS lists
these big currents in a table where you can see them. SPICE puts them
in equations, where we can't examine them as easily to discover that, by
golly, there might be some Really Big Currents if you apply unusual
voltages.
> There is something
> to
> be said also about ibis itself here. If the specification allows
unrealistic
> data to be acceptable then model makers will continue to provide it.
So should we condemn SPICE too because it uses equations ... which give
you even bigger currents than the IBIS models do (because you can stress
the SPICE model well beyond -Vcc to +2Vcc)?
Finally, don't forget that those ideal wires we use all the time in our
simulations, would have infinite current if you put a voltage source
across them. The morale is just that you don't go putting voltage
sources across them. But those little 0.1 inch long ideal wires really
don't hurt our simulations, simply by being capable of such enormous
currents.
Regards,
Andy
|------------------------------------------------------------------
|For help or to subscribe/unsubscribe, email majordomo@eda.org
|with just the appropriate command message(s) in the body:
|
| help
| subscribe ibis <optional e-mail address, if different>
| subscribe ibis-users <optional e-mail address, if different>
| unsubscribe ibis <optional e-mail address, if different>
| unsubscribe ibis-users <optional e-mail address, if different>
|
|or email a written request to ibis-request@eda.org.
|
|IBIS reflector archives exist under:
|
| http://www.eda.org/pub/ibis/email_archive/ Recent
| http://www.eda.org/pub/ibis/users_archive/ Recent
| http://www.eda.org/pub/ibis/email/ E-mail since 1993
This archive was generated by hypermail 2b28 : Thu Jul 11 2002 - 14:30:37 PDT