Peltier Cooler

[John Connor]

I'm in the process of buying parts for a new computer build and was thinking of using a peltier on an i5 6600k with a Cooler Master Hyper 212. http://www.amazon.com/s?ie=UTF8&field-keywords=Peltier%20&index=blended&link_code=qs&sourceid=Mozilla-search&tag=mozilla-20

Would there be any disadvantages, other than perhaps heating up my room even more?

 

[MrTeal]

I'm in the process of buying parts for a new computer build and was thinking of using a peltier on an i5 6600k with a Cooler Master Hyper 212. http://www.amazon.com/s?ie=UTF8&field-keywords=Peltier%20&index=blended&link_code=qs&sourceid=Mozilla-search&tag=mozilla-20

Would there be any disadvantages, other than perhaps heating up my room even more?

Noise will also be an issue, as you're going to have to run that 212 much harder than you would otherwise. Ideally you would use a fan controller and some temperature sensors to manage the peltier, so its not running full out at idle and possibly causing condensation issues.

 

[JeffMD]

Yea, one big ass disadvantage. You have to seal and protect all the electronics in and outside of the cpu socket %100! A peltier will bring your cpu temp close to 32f, this will cause condensation to form. This includes anything on the cpu and the pins themselves. For me, even using dielectric grease, the source of my cpu fail was a short in the pins of the socket.

If you want overclock cooling, water is much safer.

 

[BonzaiDuck]

Yea, one big ass disadvantage. You have to seal and protect all the electronics in and outside of the cpu socket %100! A peltier will bring your cpu temp close to 32f, this will cause condensation to form. This includes anything on the cpu and the pins themselves. For me, even using dielectric grease, the source of my cpu fail was a short in the pins of the socket.

If you want overclock cooling, water is much safer.

Wouldn't it be possible to run the peltier so the temperature doesn't drop below dew-point, or otherwise somewhere between 60 and 75 Fahrenheit?

Otherwise, you'd need a rubber seal similar to those used on your typical phase-change cooler such as had been sold at FrozenCPU. The trick being -- to eliminate all the air around the CPU from which any water would condense, thick enough so the outer surface of the rubber would never get cold enough to cause condensation there.

 

[DrMrLordX]

It's theoretically possible that you could run the TEC in such a way to keep temps above the dew point. You'd have to have fairly careful control of ambient temps and other factors. There are some weird circumstances where that might be desirable.

I'm seeing scant evidence that there are some better thermocouples than p-type/n-type bismuth telluride out there, but good luck finding anything else in commercially-available TECs that you could use for CPU cooling.

 

[John Connor]

The only thing I'm seeing regarding a rubber seal is a rubber silencer. http://www.frozencpu.com/cat/l1/g42/Window_Kits.html

 

[BonzaiDuck]

The only thing I'm seeing regarding a rubber seal is a rubber silencer. http://www.frozencpu.com/cat/l1/g42/Window_Kits.html

I may not have been clear enough. FrozenCPU still sells these exotic phase-change kits:

http://www.frozencpu.com/products/1..._940_AM2_Xeon_CE-48-D-1C-1G.html?tl=g49#blank

I'm pretty sure those copper "things" that sit on the CPU or a GPU have some sort of foam-rubber sheath that keeps them from coming into contact with air to avoid condensation. Off the top of my head, I'd think you could manufacture something like an insulator using a product like Spire acoustic pads. Whether or not it works? That's a matter of a test with a risk to it, or maybe you could prove it with no risk.

 

[guskline]

Have you had experience in custom Water Cooling?

 

[aigomorla]

The TEC must be paired with CPU perfectly.

Too much power will cause condensation and freeze over you board / cpu when its in idle.
Too little power will cause a run away from equalibirum where you'll end up cooking the cpu because the TEC will not be able to move enough heat away from the cpu.

I would advise on not using TEC's to cool anything outside a water chiller, and even then it should only be attempted by someone who has some experience and should not be on a system which is going to be unattended.

 

[tcsenter]

Consensus of experience seems to be Peltier is more of a pain in the ass than justified by the benefit. It's why it fell out of favor after being a fad/trend for a couple years. But hey, if you have the spare time and want to find out for yourself, go get em.

 

[John Connor]

Have you had experience in custom Water Cooling?

No, and I don't want to have water near my computer.

I guess I'll scrap this idea. Good thing I asked. I didn't think about condensation at all. No, I think the Cooler Master Evo 212 will do just fine I guess.

 

[DrMrLordX]

If you were considering the trouble and expense of using a TEC, you may as well just buy a bigger HSF. There are plenty of reasonably-priced alternatives to the 212.

 

[aigomorla]

Consensus of experience seems to be Peltier is more of a pain in the ass than justified by the benefit. It's why it fell out of favor after being a fad/trend for a couple years. But hey, if you have the spare time and want to find out for yourself, go get em.

its not just a PITA its a PITFA to get right.

It can be done tho, and it can also be more efficient then a phase change unit if you have the proper setup, like you run overpowered TEC's at its highest CoP, and then use a PWM on them so they pulse instead of constantly gobble electricity.

It also helps to have them cool water, and then have the water pumped to a cpu block as the holding potential in water will prevent less of a loss if insulated right. It also helps to have them a thermometer so they turn off @ certain temp and then back on once temp exceeds X degree's, this way you can avoid condensation. For example i had my TEC's turn off when water temp reached 10C, and then kick back on again when water temp reached 13C. This left my window of 3C in which water temp could be held at, and condensation does not occur @ 10C, unless its ridiculously hot outside, and you have a high humidity index.

The beauty in tec's is that they can be shut off, and then turned on rapidly, unlike a compressor which requires a cooldown period between each run, hence if your really milking the CoP it can be more efficient then a phase system.

However, again... PITFA....
I think my TEC setup honestly only lasted 2 weeks tops before i shelf'd it because it was so terribly inefficient.

RIP:

it had 5 Tec's on that chiller, required 2 loops (hot / cold) side, where the hotside was cooled by 2 massive 480's.

 

[MrTeal]

its not just a PITA its a PITFA to get right.

It can be done tho, and it can also be more efficient then a phase change unit if you have the proper setup, like you run overpowered TEC's at its highest CoP, and then use a PWM on them so they pulse instead of constantly gobble electricity.

It also helps to have them cool water, and then have the water pumped to a cpu block as the holding potential in water will prevent less of a loss if insulated right. It also helps to have them a thermometer so they turn off @ certain temp and then back on once temp exceeds X degree's, this way you can avoid condensation. For example i had my TEC's turn off when water temp reached 10C, and then kick back on again when water temp reached 13C. This left my window of 3C in which water temp could be held at, and condensation does not occur @ 10C, unless its ridiculously hot outside, and you have a high humidity index.

The beauty in tec's is that they can be shut off, and then turned on rapidly, unlike a compressor which requires a cooldown period between each run, hence if your really milking the CoP it can be more efficient then a phase system.

However, again... PITFA....
I think my TEC setup honestly only lasted 2 weeks tops before i shelf'd it because it was so terribly inefficient.

RIP:

it had 5 Tec's on that chiller, required 2 loops (hot / cold) side, where the hotside was cooled by 2 massive 480's.

I was actually contemplating doing something like that. I have a couple dozen Antec Kuhler Shelf coolers sitting in boxes that are just taking up space. Rather than dual loop I was planning on mounting them outside the building and running a water/glycol mix in the primary loop, passings through a big series string of cold plate/tec/CPU cooler units. Of course, then I got to planning control, and automatic bypass valves to onboard rads as a backup, and proper PID loops, and the whole thing died in planning paralysis once I actually thought about how much work it would take for what might be a couple hundred MHz gain.