At 04:45 AM 08/23/2001 -0700, David Honig wrote:
Faustine, look up Faraday cages, TEMPEST, and search the archives. As if you didn't know. Succinctly, the electron gas in metals shields you from the electromagnetic antics of distant, radiating electrons, by shorting the ripples in the aether they make -and this shielding makes it harder to listen to your emissions, too. The problem is that cables and ventilation vents are antennae, for sending and receiving both.
Testing is key. If you don't measure, you don't know.
This stuff was a *lot* easier when computers were slower. I used to test my TEMPEST room at 450MHz, since that was high enough frequency to cover any realistic level of emissions from the upper harmonics from the VAX, and it was also a short enough wavelength that leaks were pretty detectable. It doesn't take much to get a leak - copper foil on a joint wearing out, or the copper mesh we'd stuff inside gaskets getting set unevenly. The waveguides we used for fiber or air vents were typically 1/8 inch wide and an inch or two deep - and if you pushed a paperclip halfway through you'd twang the leak meter. Well, that was fine for computers that were around 10MHz. These days, when 1GHz is slow, there's tons of stray energy above that, and that stuff is much more penetrating, plus you've got all the 100 and 133MHz memory and disk bus stuff. Fortunately, the equipment runs at much lower power levels; you can run on batteries instead of 208-volt 3-phase (:-), but I'm still glad I don't have to design a room or even a box for that level of tightness. That room was still in active use with a VAX 8650; we retired it about when we put in the Sparcstation 1 or 1+ - were those 25MHz?