The smaller you want to make a cooling system with this, the more closely you will have to control temperatures. The method used in the OP is basically the "Overkill" model in which there is plenty of fluid to take the place of converted fluid and where the heatsink is large enough to condense the gas at a rate higher than that of evaporation. That in itself is a more involved task. That being said, I think I have a solution.
In my mind the problem with a smaller system would be trying to keep the CPU/GPU/APU supplied with enough liquid so that the evaporation process can continue as fast as possible. I do think that gravity may be an asset here. Something that I think may have been overlooked (may have missed it while dredging through all 9 pages) is that normally gasses equally disperse to fill a volume. Assuming that in a gaseous state this is somewhat lighter than air, the placement of the condenser might make all the difference. Bear with my diagram:
Gravity should keep the fluid level, but if the reaction is quick enough a pump could be added, if the the heatsink was sufficient enough to condense as quickly as it evaporated. Even if the substance condensed before hitting the heatsink, why would it be a problem? it would return back to the cpu anyway and just have another chance to phase change. as long as there was fluid on both sides of the pump, there would not be a foreseeable problem and the pressure differential would not be an issue.
This being said, it still is really just old heatpipe technology expanded upon. Either way, I don't care if the results are a constant 34* temp Edited by dizzy4 - 1/17/12 at 11:13pm