From: Ray Anderson (firstname.lastname@example.org)
Date: Wed Feb 16 2000 - 14:52:24 PST
As Larry if off teaching a class this afternoon I'll give you a quick
answer to your question.
Yes we have observed failures that we've been able to attribute to
inadequately design power distribution systems ('just sprinkle a
liberal layer of caps and hope for the best').
One particular case that I remember at the moment manifested itself
as a system that seemed to work properly in all our usual system tests.
Problem came when the customer ran some particularly noxious application
code. Seems that the application did some massive number crunching then
paused, crunched numbers then paused, crunched numbers then paused....
These pauses came at about a 40Khz-50KHz rate.
As it turns out the power distribution sytem had a very high impedance
at around 40-50 KHz. The repetitive current demands at that frequency imposed by
the application code caused the power planes to exhibit a sort of sinusoidal
ringing at the 40-50 KHz freq. when observed with a scope.
Analysis of the power distribution system with our home-grown tool revealed
the problematic impedance peak. We then selected decoupling capacitors
to provide a flat impedance profile over a wide range (about 10KHz to 500MHz).
Removing the existing decaps and replacing them with the newly selected
parts cured the problem we were seeing. Never was able to cause a power distribution
system induced failure again in that type of system when they were built with the
proper parts. Observation with a scope showed that the ringing had been effectively
This particular example I've just related the details of happened to be a
problem child at a relatively low frequency, but since applications code
used in a customer environment is unpredictable you need to provide a
flat power dist. sytem impedance over a wide range as well as paying particular
attention to hardware dependent freqs (buss frequencies, clock frequencies, etc.)
So much for war (lab) stories. Hope that answers your question.
> To: email@example.com
> Subject: RE: [SI-LIST] : ESR and bypass caps
> Date: Wed, 16 Feb 2000 13:57:42 -0800
> MIME-Version: 1.0
> Regarding capacitors and just sprinkling a bunch down on the board -- have
> you see any product failures caused by this approach? If so, is it then
> readily obvious by looking at the power planes with a high speed scope that
> power distribution is the power?
> I ask because with some managements you might find yourself playing the
> "well, hey, all of our PCBs have worked so far, so why should we invest
> money in all this SI black magic?" game.
> ---Joel Kolstad
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
firstname.lastname@example.org. In the BODY of message put: UNSUBSCRIBE si-list or
UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
si-list archives are accessible at http://www.qsl.net/wb6tpu
This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:35:04 PDT