Peltier Cooler ?

Blackhawk47 · 15 Dec 2007 at 23:02

Okay so the other day i was pretty stumped. A question came up (at an IT academy i go to called zenos if anyone is interested) discussing what would be the best way to cool a pc which is failing due to heat and therefore under very hot conditions.

The choices were as follows (as far as i can remember) had to pick one :

Air Cooling (by adding more system/case fans)
Water cooling
Peltier Cooler

and there was something else which was completely wrong anyway.

so i picked water cooling. but no in fact it is a peltier cooler. now i have no idea exactly what this is. researching it i think its some kind of heatsink/air conditioner thing.

has anyone heard of this or can explain it in simple terms ? and would it actually be better than water cooling ?

Jokester · 15 Dec 2007 at 23:11

A Peltier cooler (or Thermo Electic Cooler) is a device that, when you provide a power source to it transfers heat from one side of it to the other, maintaining a temperature difference across the plate. If you cool the hot side enough, the cold side will go below ambient temperatures.

You need to use watercooling to make it worthwhile, not only do you have to deal with the heat from the CPU but also the considerable heat from the TEC, there will also be issues with condensation that need to be dealt with. Ultimately, for a CPU you're better off using a phase change unit as it's more reliable and a lot less problematic.

Jokester

platinum87 · 16 Dec 2007 at 03:55

youd think phase change cooling would be far more difficult/ problematic then a tec cooler where as condensation occurs when the saturation point/dew point is reached at 100%RH, whereas a rising temperature ensures it is impossible for condensation to occur, as the airs capacity for water increases with temperature, ie, why high pressure = no cloulds.

in either way maintaining a temperature gradient would be beneficial allowing for more efficient cooling..... although i dont wanna argure against a scottish guy with 15k+ posts, because that would be insane

Zefan · 16 Dec 2007 at 03:58

Why not? Just because he has a lot of posts doesn't mean he's right!

He is right though

, condensation is unavoidable if you want to exploit the benefits of using a TEC system.

Cyanide · 16 Dec 2007 at 10:27

TEC's suck, I had one cooled by water cooling last year and it damn near boiled the water in the loop. It was great at first, got my CPU down to -2c at idle and 4c at low to mid load but as soon as the load increased I couldnt cool the hot side enough so the cold side started to heat up and from then onwards its a downwards spiral to CPU death. You also get the problem of having a CPU cooled to -2c heating back up when its turned off so you have to waterproof the whole board to spot anything shorting out. I'd just stick to water/decent air cooling if I were you, the Thermalright Ultra-120 Im running at the moment is just as good as anything else I've used.

pendulum666 · 16 Dec 2007 at 10:32

Yeah I have seen some great results using TECs but as said, really its all about the hassle and if you can stand it. I prefer'd and only ever experimented with good watercooling, which was fun enough really. Then it got tedious when I realised you get similar results with air and actually the same noise levels saving you 2-300 hundred squid.

Jokester · 16 Dec 2007 at 10:35

platinum87 said:
youd think phase change cooling would be far more difficult/ problematic then a tec cooler where as condensation occurs when the saturation point/dew point is reached at 100%RH

Difference is that phase change units usually have their heads designed with insulation in mind, while TEC waterblocks are usually normal blocks with the TEC and cold plate bolted to it with insulation as an after thought. When I was using a Maze4 I had to use a lot of silicone grease to fill the various crooks and crannies in the blocks to prevent ice forming there (which would then melt and run out when you switched it off).

Jokester

WJA96 · 16 Dec 2007 at 10:42

I think I would argue for the mark, to be honest, as a Pelt isn't actually a cooler, it's a heat transfer method. As everyone else has said, you still need a cooler to take the heat away from the hot plate on the pelt otherwise it will stop working.

WJA96 · 16 Dec 2007 at 10:44

pendulum666 said:
Then it got tedious when I realised you get similar results with air and actually the same noise levels saving you 2-300 hundred squid.

This is so close to the truth it almost hurts.

Dist · 16 Dec 2007 at 21:56

does a peltier generate less heat on the hot side then the CPU? im just wondering of its possible to use 1 peltier to cool another, with the most powerfull cooling the cpu, and progressively less powerfull peltiers to cool the peltier infront of it, and at the end with the low power peltier, a heatsink, to make the system compleatly silent.

If the peltiers generate more heat then it cools, you would need progressively more powerfull peltiers to cool the one in front of it, which would be stupid and not possible to pull off my idea.

Jokester · 16 Dec 2007 at 22:02

Dist said:
If the peltiers generate more heat then it cools, you would need progressively more powerfull peltiers to cool the one in front of it, which would be stupid and not possible to pull off my idea.

That is basically what it boils down to.

You still need to cool the heat from the CPU, but in addition to that you also need to cool the heat generated by the TEC(s). All a TEC does is move heat from one place to the other generating even more heat in the process.

Jokester