The thing I used is a pickup coil designed for RPM measurement etc, I hooked it up with shielded 4 conductor cable, 2 wires per pin, with the shield of the cable connected to the shield of the pickup coil.
I hooked up the pickup coil with a bandwidth of 50KHz to two preamp/filter boxes (wavetek) with I believe 80dB gain total and you can defiantly pick alot of shit up near computers, especially if you sniff near the ports on the metallic shield areas. You pick up audio frequency growling and stuff, and when hooked into a SA there is stuff above audio too. Filter boxes are nice because if you have two, you can easily put a 4th order highpass filter at 150Hz to eliminate power line noise from the audio.
May have been 120dB gain (I think I might have hooked up my 100KHz bandwidth LT1028 based filter first? it has 2 20dB gain stages IIRC.
If you want to find stuff, I found the e-field probes are very good at it, to high bandwidths. I made one by drilling a hole in a steel ball (removed from a fence post) and soldering it to the inner pin of a BNC connector, then epoxying a support around it made of old PVC pipe cut in half. Then coating the ball with epoxy. Much easier to get a signal near stuff like cables etc then with a magnetic loop probe.
Then your next best bet is to make the square magentic loop probe, by folding a square out of a piece of copper wire that's approximately 1x1 inches, and soldering one end to the shield of a BNC and one end to the inner pin.
From there you can make things like hardline loop antenna probes, shielded coils (like the one jim williams uses, figure 8 pickup coil in brass tube, or normal coil in brass tube, though these are VERY directional). Honestly I do not find the hardline probes or jim williams probes that useful in comparison to the square magnetic loop or the e-field ball probe. I think they are more useful for work on a PCB, where you can put em right near a trace, then working on equipment boxes, especially in a field setting. If you can put the object your studying on a work bench then thats a bit better.
also another interesting one is to take a torroidal inductor (small) and cut it in half with a dremel diamond wheel, then wrap wires around it to make a C shaped coil. I have not experimented with this one yet, as I have lost what i built some where.
The thing is, you defiantly want a pre-amp for these things, especially if you are using with a oscilloscope and not a spectrum analyzer.
I have not experimented with measuring stuff coming out of computers at higher frequencies between 50MHz-2GHz though. A wideband dual ridge horn (2-10GHz design range approx) picks up some stuff if you place it face down on a keyboard of a laptop, or press it into the front panel of a monitor. High GHz lens horn antennas, I tested 2, 18-21 GHz and 21-24GHz, do not pick up anything from computers etc, but you can get directional signals inside of a room depending on how you aim em. It would be more interesting if I had another wideband preamp for this region though, this was only a direct connection to my HP70000SA. Indoors the pickup at these frequencies is tiny even when the SA is set to most sensitive settings with 20 (well ~15 or so at this frequency dB preamp). I don't like the little preamp modules from ebay using MMIC because their S21? (right?) parameters are kinda all over the place. I would like to get a nice one meant for wideband use with a SA that has a nice linearish slope on it).
Perhaps I will retest the high frequencies between 18-24GHz with a open computer running a 3.5?GHz processor. maybe you can get some emissions from the side if its harmonics are not dealt with properly.
Unfortunately I did not have any antennas for the frequencies people are interested in, between 100MHz-1GHz. The dual ridged horn for this frequency is large and I ran out of money for metal at the moment. If someone wants to donate me 2000$ to finish my welding table and buy aluminum for a large horn I will happily slave away at its manufacture.