From: karn@qualcomm.com (Phil Karn)
Date: Sat, 3 Apr 93 16:56:53 PST
To: Cypherpunks@toad.com
Subject: Re: TEMPEST in a teapot
Message-ID: <9304040056.AA07411@servo>
MIME-Version: 1.0
Content-Type: text/plain


A few minor comments on a pretty comprehensive and well written article.

Although TEMPEST is closely related to EMI shielding, remember that
TEMPEST is concerned only with *information bearing* radiations, not
interference. In particular, the switching power supply, a potentially
prodigious source of EMI, is not a significant TEMPEST issue because
power supply emissions carry little if any information. (Varying loads
might cause minor modulations of switching frequencies, etc, but this is
probably something that only the paranoid "covert channel" types worry
about. Maybe you coull tell when the floppy drive motor starts and stops,
but I doubt you could do much else.)

Who knows, cutting down on power supply radiation might make it easier
to extract information from the emissions that remain, because of the
jamming effect of power supply noise. But don't let that stop you. It's
your duty to your neighbors to emit as little RF noise as possible. When
I lived in New Jersey, I learned to my chagrin that my two PC clones
made my next door neighbor's AM radio useless. Only 25 miles from New
York, he was unable to listen to WABC, a 50KW clear channel AM station!
The problem disappeared completely when I installed some inline AC RFI
filters in the power supplies. Since we shared a pole transformer, I
theorize that the noise was conducted from my computer to his radio
directly over the power lines.

Which brings me to my next point. I have not seen *any* clone-grade PC
power supplies with adequate power line filtering. They have a minimal
LC lowpass network on the power supply board itself, but this is usually
inadequate. Whenever I buy a new power supply, the first thing I do when
I get it home is to replace the IEC power connector with an integrated,
shielded power connector/RFI line filter. These devices are widely
available for several dollars from electronics surplus houses and
amateur radio "hamfests". I also use power cords with built-in ferrite
"lumps" but these are probably harder to find (one particular hamfest
vendor had a lot of them a few years ago, but I haven't seen them
since.)

Modern monitors are *much* better shielded than the early PC monitors,
especially those no-name Korean or Taiwanese copies of the original IBM
PC monochrome monitor. If you have the misfortune of owning one of those
old monitors, as I do on one of my systems, chances are the lion's share
of its emissions are coming from the +70V power lead that runs from the
main circuit board to the video output stage on the base of the CRT.
(Note! Do NOT confuse this with the high voltage lead going to the anode
of the CRT!)

The +70V power line to the video output driver acts as an antenna for
radiated video signals that can be *quite* strong. I suspect that the
reports we've seen of successfully picking up the image on a computer
display were taking advantage of this. To fix the problem, just replace
the plain wire with a piece of shielded coax, bypass the ends with .01
or .1 uF capacitors of sufficient working voltage, and ground the
shields to circuit board ground on both ends.

Phil