Hehe,
Happened to stumble upon this site while surfing last night:
Oil Cooling A Computer Seems strange that a computer could survive being immersed in any sort of liquid.
Why not, oil is an excellent insulator.
Exactly. There are millions of oil-filled transformers in service, with an excellent record. And most ham radio operators have used a "cantenna", which is a big 50 ohm resistor in a paint bucket full of oil, used as a dummy antenna.
I wouldn't have used vegetable oil for this application, though, because it will quickly go rancid exposed to air like that. Straight mineral oil would be better (though the druggist might give you a funny look when you buy a couple gallons of the stuff!).
I would be interested in the long-term reliability of a setup like this. Most plug-in connectors are NOT designed for oil immersion, so there might be intermittent contact problems in a few months/years. The cooling oil might tend to leach the plasticizers out of plastic parts, perhaps causing the plastic to disintegrate or become brittle, or making the oil itself conductive.
If I was doing this I would get the product made for the purpose, transformer oil.
Back in the olden days IBM did have oil cooled core storage.
I wonder where the hard drive is? That won't work long in oil. They are somewhat "sealed" but they still breathe through a hepa filter. Eventually some oil will get in there and crash the heads.
Nevermind, I just looked again and the drive is outside the oil with the power supply How are they cooling that?
Heat-pipes are used in some laptops for cooling.
These will shift heat at incredible rates away from heat sensitive components - 500 times the conduction rates of a pure copper bar of the same cross section area are possible, with no moving parts except the usual fan on the cold end. They can be built to even outperform the sun in terms of kw/square centimeter.
Back when mainframes were silicon transistor logic (not CMOS) we used water cooling on the big ones. They had a 3hp Gould pump (actually 2, one was a backup) and they ran hoses under the floor to the CPU frames, pumping distilled water. This was heat exchanged to glycol (customer water) that went to a big chiller (again usually 2 in parallel for backup).
The logic itself was on chip that was mounted on a stainless plate, another stainless plate with water flowing through it bolted to that with heat sink compound between them.
Back when mainframes were silicon transistor logic (not CMOS) we used water cooling on the big ones. They had a 3hp Gould pump (actually 2, one was a backup) and they ran hoses under the floor to the CPU frames, pumping distilled water. This was heat exchanged to glycol (customer water) that went to a big chiller (again usually 2 in parallel for backup).
I've always wanted to build a water cooled computer, obviously nothing on the scale mentioned above, but my only real fear in a system like that is a coolant leak.
The logic itself was on chip that was mounted on a stainless plate, another stainless plate with water flowing through it bolted to that with heat sink compound between them.
Stainless??
I would have thought that copper would give better heat transfer properties.
I think they were more worried about corrosion than heat transfer. These things ran pretty cold anyway. 65F or so.
It also looked pretty cool when you opened the gate and saw all those shiny cold plates and hoses.
I water-cooled my old 1.3GHz Athlon. Had a fish-tank pump in a tupperware running water and antifreezer through a water-block on the CPU up to a heater core off an old toyota that was mounted in the top of the case. Even rigged up a relay into a 120V receptacle so the pump would kick on whenever the PC was on. Ran great for years until the tubing started to get brittle and lose its elasticity; every time I jostled the tubing to much with my hard drives or add another stick of RAM, a minute amount of coolant would seep out and drip into my graphics card. There was never a trace, and never a drop any time I looked. Destroyed 3 cards before I discovered what was going on.
The temperatures I got weren't even all that great. But I had a lot of fun tinkering with it, which is was the real reason I did it in the first place.
After the fiasco with the leaks, I switched to passive evaporative phase change (heat pipe) when I build my 3.2GHz P4 box.
Had a fish-tank pump in a tupperware running water and antifreezer through a water-block on the CPU up to a heater core off an old toyota that was mounted in the top of the case.
We may be onto something here. We could sell this "technology" at pet stores, housewares stores AND auto parts stores. If all else fails, we could sell the whole package at Best Buy for $399.99 and Geek Squad could install it for another $199.99. I smell easy money here!
Surpluscenter.com is selling the guts out of one of those counter top soft drink dispensers. item 1-1302 $19.95
It has a 1/12 hp compressor reefer. You could do some serious PC cooling with one of those.
I guess if you like complexity, go for it. Personally I have never seen a PC that couldn't be air cooled. I know IBM got away from water cooled machines as soon as they could. (when they got fast CMOS)
Even with transistor logic we ran everything but the biggest machines on air.
There are only two real reasons for any consumer watercool:
1) Eliminate the fans and make it quieter
2) Agressive overclocking using TECs for cryogenic cooling
I was overclocking my PC about 30% when I was watercooling, which is agressive, but not overly so; I'm doing the same 30% overclock in my PC right now with a simple heat pipe. The big problem is when using TECs, as for every watt they cool, they pump out about 2W for every 1W of cooling, and traditional heat sinks just can't keep up. With sub-zero temperatures, people can get the processors sometimes 50% faster than they're supposed to be capable of.
My first thought is the CPU is seldom the bottleneck in a PC . Are you also water cooling the other chips like the video processor or the drives? You still need to move some air.
The CPU is a veritible space heater. Removing that heat from the case makes everything else that much cooler. I've seen some people watercool the GPU, north bridge and PSU, but I didn't- usually, they do that more to go fanless than for any sort of speed increase.
The performance bottleneck really depends on the application; in my case, the CPU speed helped, but my FSB was starting to become the bottleneck in the benchmarks I cared about, so 30% CPU overclock only really gave about a 25% speed inrease.
Wow, for $40, that's pretty kick-ass cool! I know people who were taking apart small refrigerators and dehumidifiers to do that, and you can't even get a custom plate milled for that.