Question peltier device

[Roddie.the.gamer]

i saw a video by electroboom where he tried to use peltier devices to make electricity
those peltier devices can also be used to cool / heat up things

so i was wondering if i would be able to use one of those device to cool down my cpu, than to throw a heathsync and a fan on it ( so just put a peltier device between fan and cpu)
im posting this here because i wanted to ask YOU if this would be a smart idea and if it would have any effect on my cpu's temperature (either positive or negative)

p.s. pls dont judge me for my bad english

 

[Ralston18]

English is fine.

The device is very inefficient and I would not risk anything other than an expendable computer do any testing or direct application of the Peltier.

Reference:

https://rimstar.org/science_electronics_projects/peltier_effect_module_cooling_efficiency_test.htm

You can easily find other similar links about the Peltier Effect and its application for cooling.

May be fun to experiment with for a while and learn more about it all. Then move on to cooling something such as a CPU.

Again with the understanding that if there is a problem the CPU (or whatever device is being cooled) could be damaged or destroyed.

 

[InvalidError]

The idea of using thermoelectric cooling on CPUs has been around for ~20 years. There is a very good reason why it never caught on: as Raslton pointed out, TECs have horrible efficiency.

To get peak efficiency out of a TEC, you have to operate it at about 10% of rated power and 10C temperature difference between cold and hot side, not particularly useful for cooling CPUs. LTT, deBauer and a few others did TEC-based vanity projects this year, should give you an idea of how impractical it is to achieve meaningful CPU cooling using TECs.

Unless you intend to go stupid with TECs, don't mind the cost of the TECs themselves, extra PSU(s), custom hardware to put it all together (you'll probably want to liquid-cool the TECs to keep their hot-side temperature even) and the power bill, you are better off with the stock HSF.

 

[Carl2]

I've been working on a Peltier water cooling for a 6 core CPU, a spare computer set to run with a heatsink right now. I like the fact that the CPU will be cooled with water a little over freezing rather than air temps that can get to about 80 F. The simplest way is to mount the Peltier right on top of the CPU but because I'm still in the experimenting mode I'll be using a water cooled water block on the CPU. I've already got the Peltier water system working but just got newer Peltiers with more power to get the water cooler faster. With Peltiers you have the heat of the CPU plus the heat the Peltiers generate to do the cooling so you have even more heat to remove. So for right now I have two water loops, one pumping cool water to the CPU water block and one loop cooling the heated water coming from the Peltier.

 

[InvalidError]

The simplest way is to mount the Peltier right on top of the CPU but because I'm still in the experimenting mode

It may be the simplest but no single TEC can handle the temperature difference and heat output of 60+W CPUs on its own. If you want to run the cold side at 30C below ambient, you would ideally split the drop in two or even three TEC stages so each one only has a 10-15C delta to deal with. If you try doing 30C in a single stage, most of the TEC's power gets wasted fighting the amount of heat getting conducted back from the hot side to cold side within the TEC itself.

 

[4Ryan6]

Peltier CPU cooling is a very viable cooling solution but if you want to learn how, you need to talk to those that know how to do it, not discredit it.

I am OCAddict at overclock.net and this is version 3.0, I am soon to have version 4.0 operational, and FYI, you guys this works!

https://www.overclock.net/forum/62-peltiers-tec/1633988-chilled-water-cooling-vs-3-0-build-log.html

 

[InvalidError]

Peltier CPU cooling is a very viable cooling solution

Few people would call 300+W, a whole case worth of heatsinks, fans, pumps, tubes, PSUs, etc. totaling the better part of $1000 for cooling a 100-150W $300 CPU anywhere near viable.

Feasible, sure. That's pretty much it.

Since nothing fundamentally new has happened with TECs over the past 20 years, I'm not expecting anything game-changing to happen until scientists figure out how to mass-manufacture reliable quantum tunneling solid-state coolers.

 

[tennis2]

Linus Tech Tips has done a few different vids on TECs, even a recent 500W TEC.