I would agree with Ls7 on the fact that a liquid with a greater thermal expansion would result in improved flow and thus better temps. I've thought such a setup over before, and have a few things to add to the conversation.
First is that the tubing from the CPU block to the Radiator should be as short as possible. In my mind, this will facilitate more "pressure" between the different thermals in the waterblock and radiator.
Second the radiator should be single pass. In my mind this creates more pressure between the thermals of the waterblock and the rad. I'm sorry I can't explain this right now..... I'm distracted.
Third, the block should obviously be one with high surface area, such as the "lumpy channel" block. I'm also fond of the inlet/outlet orientation of such a block. It should be mounted so that the inlet is on the top and the outlet is on the bottom (Cold sinks, heat rises). One would have to tinker with inlet outlet to get the best results.
One of the thing to seriously think on is tubing size. Too small inhibits flow, too large leads to a decrease in pressure and allows for thermal backflow.
I never have been a fan of pelts...... they just suck too much energy for me. In this case they are probably necessary. It might not be a bad idea to have this be a chilled liquid setup. Meaning that at some point in the loop, the pelt is sandwiched between two blocks. This would allow you to have some liquid be VERY hot and some to be VERY cold. It would also allow you to control the direction of flow, and it would probably result in better flow rates.
The key to making this setup work is to generate the largest possible thermal differentials in the fluid to generate as much flow as possible through the CPU block.
|