>>Just here a few rooms away we have a problematic machine with Intel PIII CPU. It has many problems we can't reproduce whenever we take the machine out of that room. We even tried completely testing with all the equipment it's connected including speakers. It doesn't error if tested here in my room. We changed all outer cables in that room and still it randomly produces unexpected things. Problems occur only if the machine is in that room. Electrification is also tested to no avail :(
>>Cetin

>Does this problem machine have it's own UPS? We've run into problems at customer sites where the power coming from the wall was inconsistant (103 volts to 114 volts). When low voltage occured, there would be many seemingly random errors. (Their power was so bad at this site, that the lights flickered when the phones rang!) A UPS (on each machine) cured all of these problems.

One thing to look for is bad grounding, or machines across the network grounded to different groundings. The easiest way to detect this is to measure if there's any voltage between the zero wire and the ground wire; if there is, the voltage between the zero and phase, and ground and phase, should drop by approximately same amount. Over the years I have observed this at several places, and getting the grounding (or just the extension cords) straight fixed the seemingly random problems. Now this applied to coaxial cables, since they connected the groundings of the computers - and I have got hit by a spark on the BNC connectors several times. I'm not sure the wiring goes the same way with UTP - we did UTP in a few really old buildings which had rather problematic power (with UPSes beeping when coffee is made) and had no data problems there... or was it my automatic recovery procedure which hid the problems :)