Computer Shelves (RF EMI/EMC issue)

Schilling, Richard RSchilling at
Tue Feb 20 11:44:43 PST 2001

During college, I lived in a trailer, and had my CPU next to the water
heater.  Some times it would turn on, others not.  Eventually found out that
when the water heater turned on, the EMR would not let the computer work.
That was really confusing until I figured it out.  Also, the radio, when
plugged into the wall would howl like Robert said.  

So be careful of putting CPUs next to other equipment, including circuit
breakers.  They could disrupt the machines.

Richard Schilling

> -----Original Message-----
> From: Robert G. Brown [mailto:rgb at]
> Sent: Tuesday, February 20, 2001 11:30 AM
> To: Jim Lux
> Cc: Leonardo Magallon; Peter Eriksson; beowulf at
> Subject: Re: Computer Shelves (RF EMI/EMC issue)
> On Tue, 20 Feb 2001, Jim Lux wrote:
> > I doubt that metal shelving would have any beneficial EM 
> properties, and, in
> > fact, might actually make things worse by being a 
> capacitively coupled
> > radiator.  Trust in your FCC Class B cases...
> >
> > Actually, does anyone have any evidence of Electromagnetic 
> Compatibility
> > problems in a cluster?  Lots and lots of computers, all 
> with switching power
> > supplies, running off fairly long extension cords all 
> strewn about (neatly,
> > of course), and the conditions are certainly ripe for 
> EMC/EMI problems.
> > Fortunately, the computers themselves are probably fairly 
> immune, but what
> > about other devices (i.e. does your cell phone still work 
> standing next to
> > your cluster?).
> Back in oh, 1983 or 1984 my 4.77 MHz IBM PC would radiate 
> like all hell
> on FM 100 or thereabouts.  I could even hear it "work" in the static.
> I'm sure that any good phreaker could have told you character by
> character what I was typing from fifty feet away.  Messed up 
> my ability
> to listen to music and work.
> Shortly thereafter, those Class B cases started to be required and
> motherboard designs improved (probably fewer sharp turns and more
> attention paid to conductive path shapes and lengths) -- 
> haven't heard a
> peep for well more than a decade out of anything except the sound card
> of the system itself, which DOES sometimes register a bit of system
> state-dependent noice.  Given that this is an amplifier with all sorts
> of resonant loops sitting INSIDE the box I'd have to say this 
> is pretty
> darn good.
> I honestly don't think that metal shelves or the lack thereof make any
> real difference to cluster stability or performance these days.
>    rgb
> -- 
> Robert G. Brown	              
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at

Beowulf mailing list, Beowulf at
To change your subscription (digest mode or unsubscribe) visit
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Beowulf mailing list