Mainstream Watercooling, Micron-scale Channels, and Electrokinetic Pumps

[A-Speck]

Hi, everyone, new to the forums here. I've been watercooling for a couple months now, and I figure I finally have enough experience to begin to consort with the Pro/Cooling gurus. Hope I can help add to the forums. You can never have too many thermodynamics majors, eh?

Anyway...

With the news of Apple's watercooled G5, I went and did some Googling, and one company that consistently came up was a start-up called Cooligy. The description of their technology was pretty interesting. They claim to have micron-scale channels in their waterblock-analogue (can't remember, but I think I read that they were made out of silicon...a lithography process, perhaps?) and an electrokinetic pump with no moving parts.

Sounds like pretty heady stuff. The micro-channel concept seems like a logical extension of the technology in today's waterblocks, and come to think of it, creating massive surface areas with photolithography isn't that bad of an idea. The electrokinetic pump has me thrown for a loop, though. I've heard of them before, but I don't have any direct experience with them. The one's I've read about had flowrates measured in microliters per second, though. They were actually used in experimental fluidic logic circuits. I have no idea how well the technology scales, or what kind of P-Q performance you could expect out of one, much less the heat that it would dump into the water. The promise of a truly silent pump is pretty attractive, though.

Anyone out there know any more? About the pumps, about Cooligy, about micro-channels, anything?

[nicozeg]

please do a search within this forum with the word "cooligy"

[HAL-9000]

I have been thinking about that Cooligy pump, it seems to be a real neat little gadget for OEM's. But for enthusiast coolers, I think it has issues.

For one, it is purely electrical in nature. As we have seen in a contest between solid state peltier coolers and phase change coolers, using mechanics offers much better efficiencies for doing mechanical work like moving heat. A Vapochill cools 150W or so, for about a 80 watt hit on the power bill. For a pelt to do the same, to just move the heat, much less lower the temperature, takes at least three times that...at its max operating efficeincy.

For an electrokinetic pump to be able to move the same hydro flowrate as say, an Eheim 1048, I would bet you would need between three to five times the power at least. That just seems to be the way solid-state devices work when moving mechanical loads. That presents two problems to the overclocker:

1. When you start talking about dumping an extra fifty watts in your cooling loop, its going to have an adverse effect on the performance obviously.

2. Since the pump works using ionization and static potential, that means big transient magnetic fields as it operates (the Bad kind of magnetic field). This could be a problem around hard drives. You can perhaps magnetically shield the pump, but it would be an expensive process, as in speakers.

3. The pump's optimum voltage for efficieny would probably not be 12V, or 5V either. There would need to be some potentially expensive transformer in the circuit to keep the pump going with the appropriate voltage. Could be wrong, but Murphy's Law I think applies here, and its rarely wrong.

[killernoodle]

This would also not be a good solution for enthusiasts because there is likely to be little room for upgrading and such. You cannot change anything in the loop because you would most likely need a clean room to fill the thing, it would probably be permanantely attached to the CPU, It is not likely that a radiator will be interchangeable, and it will be too expensive to consider. I dont see anything like this happening soon.

[jlrii]

Originally Posted by killernoodle

This would also not be a good solution for enthusiasts because there is likely to be little room for upgrading and such. You cannot change anything in the loop because you would most likely need a clean room to fill the thing, it would probably be permanantely attached to the CPU, It is not likely that a radiator will be interchangeable, and it will be too expensive to consider. I dont see anything like this happening soon.

Thats what oem is all about...cost will probably be expensive untill their r&d is paid for but with the # of transistors increasing the way it is how will the manufacturers present new products if the don't have a viable way to cool them??
Something will come down the pipeline sooner than you think...mac is already doing it on a very limited scale...I think there's more to it than you think.