Peltier / TEC?

Zarathustra[H]

Extremely [H]
Joined
Oct 29, 2000
Messages
38,744
Peltiers used to be all the rage among extreme cooling enthusiasts in the late 90's through the early 2000's.

1.) Does anyone still use them?

2.) What are the advantages/disadvantages?

3.) Are there any good ones that can fit between a standard cooler and the CPU?
 
2.) What are the advantages/disadvantages?

Efficiency. They used to require around 1W of electrical power to move 1W of heat. So then the heatsink would have to dissipate 2 times the load.
 
Efficiency. They used to require around 1W of electrical power to move 1W of heat. So then the heatsink would have to dissipate 2 times the load.

Ahh.. That's interesting...


Wouldn't you automatically have a more efficient cooler though, if you raise your delta T?

As I recall from Thermodynamics I 10 years ago:

Q = Cp * dT

Correct?

So, as long as you can keep your ambient down, it would seem the cooler would compensate for the need for higher temperature dissipation by getting hotter.

Since the TEC maintains the temperature delta between the hot and the cold side, this shouldn't impact the CPU much, right?

So, in other words, I guess what I am trying to say is, as long as you have a large enough power supply, and are not too concerned about your power bill, who cares if your radiators now are 20C hotter than they were without a Peltier, if it results in a CPU core that's 4C cooler?

Or am I misunderstanding the Peltier concept?
 
I think it's because TECs become less efficient the hotter they run, meaning the hotter the hot side is, the hotter the cold side will be. AFAIK it's possible to actually INCREASE your temps using a TEC if you don't cool the hot side properly.
 
I stopped using them when CPU's created more heat than the TEC's on the market could dissipate. Given the amount of heat you'd have to get rid of now, you'd need something like a 300W TEC and a separate power supply to run it.
 
I stopped using them when CPU's created more heat than the TEC's on the market could dissipate. Given the amount of heat you'd have to get rid of now, you'd need something like a 300W TEC and a separate power supply to run it.

I've seen 245W, and 437W units...

How do you predict how much power a CPU will use/dissipate when overclocked?

I have a 1200W PSU, and for what I'm doing I probably only need about 700W, so I have a little room to spare. They seem fairly low cost, so I may pick one up just to experiment with...

Do you need a copper block with them, or is it OK to just sandwich a peltier between your block and CPU?
 
Ahh,

I thought the peltier merely had to match the CPU dissipation in order to be effective. Turns out it needs to exceed it.

I'd probably need the 437W model, the 245W would likely be too small.

With that I'd generate a lot more heat than I'd know what to do with... I'd probably wind up raising the hot side temp more than the dT I'd be able to attain.

Oh well, it was worth the research to find out how it works though.
 
Yep, according to my calculations based on the dT/W chart for my setup, I'd wind up with a temp increase of 8-17C with the 245W Peltier and a temp increase of 5 - 18C with the 437W unit.

Not worth it :p
 
Last edited:
Yep, you're in phase change territory at that point.
Or maybe a water chiller setup. That would be fun too.
 
Using a TEC as a water chiller is really the only "good" way to use one in any case.

With it in between the cooler and CPU the temps fluctuate quite a bit more than normal.
 
Their performance tends to degrade very rapidly over time. Not worth it if you're looking for a reliable, long term solution.
 
go whit a water chiller or DD. i have run a water chiller for 5 months now at -5f which allows me to run my 1090t at 4.5g 24/7 wheres on regular water i had a hard time getting over 4. just insulate really good and you shouldn't have any problems
 
Back
Top