I never noticed this article before! Funnily enough I made a very similar test rig myself in early 2003 for my dissertation!
Biggest difference is that I tested using a test rig rather than an actual CPU. This incorperated a machined alu block (shaped like an athlon XP processor) with a resistor so that i could generate a set amount heat consistently.
Unfortunately I only have a couple of pics online atm
http://homepage.ntlworld.com/marc.merce ... p_expl.jpg
http://homepage.ntlworld.com/marc.merce ... CF0001.JPG
http://homepage.ntlworld.com/marc.merce ... CF0004.JPG
Back then it was quite hard to get hold of heatpipes - and they were pretty expensive. I was forced to use 2 x 8mm rather than 3 x 6mm pipes because of the cost.
2 x 8mm x 400mm (from memory) cost approx £150! I believe i bought those from thermacore. I still have 4 of them sat in my loft doing nothing :O
I have often wondered if they are a lot cheaper now 2 years on .... but I imagine in small quantities they may still be expensive though. No harm in giving them a call - there are probably other companies cheaper than thermacore too - there was at least one other in 2003, but they took 6 weeks to make the pipes to order.
My original plan was to actually build my own heatpipes. It's not as hard as you'd think at all. I only went with premade ones in the end as time was running short and I needed results!
You don't even need a wick or a fancy working fluid. Water works in just about the ideal range for this kind of application. The hardest thing is filling the tube with the correct amount of working fluid.
I'm sure i found an article where someone did just this about a year ago. I'll try and dig it out.
The biggest problem with building your own system like this is thermal contact. The machining tolerances were pretty poor on my rig and the thermal interface between heatpipe/heatblock and heatpipe/heatsink were one of the biggest areas of thermal resistance.