could you "stack" pelts?

thinking about setting up a pelt for my videocard and while i was thinking about some diffrent setups i started to wonder if you could put a cold plate on your cpu a pelt on that then another coldplate on the hot side of the first pelt and then put the cold side of another pelt on that hot side then water cool the 2nd pelts hotside?

i realize that if this would work it would take a ton of power and redefine inefficency but i was curious to whether or not this would work at all and if it did if you would be able to see colder temps on the first pelt and ultimatly on the cpu?

that would take one hefty PSU.... maybe dual PSU... 1 for pelts... the other for everything else... LOL.

if it would even work id think one psu for each pelt and one for rest of pc but i guess it would depend on the power of the 2 pelts youd use.

i just thought this up as a possible idea to get a insane temporary temp for 3dmark runs and things of that nature. i couldnt recall every hearing of anyone with a setup like this and im new to tec cooling but i thought it might be a great idea for a project if it was doable.

it's been done, there's a thread on it around here somewhere, look for it.

Aphex_Tom_9 said:

it's been done, there's a thread on it around here somewhere, look for it.

Click to expand...

k i found a couple thanks for the heads up i only see 2 pages on the extreme cooling and didnt see anything like that so i didnt even think to search. bg me

http://www.ocforums.com/showthread.php?t=110225&highlight=multiple+pelts
http://www.ocforums.com/showthread.php?t=305498&highlight=stacked+pelts

there's a little more-
http://www.ocforums.com/search.php?searchid=518987

If the two pelts were the same, let us say they can shift a q[/] joules in time t.

So pelt 1 transfers q joules to the hot side, leaving q joules to be removed. Now you use peltier 2 to transfer this heat to its hot side.

At the end of the day you still have q Joules to remove, plus whatever the peltiers add as they are not 100% efficient.

Even if the peltiers are different, you would be just as well using one, better, peltier, no?

yeah thats true i was focusing more on the cpu side temp and wasnt thinking it through to the massive amount of heat that would be transfered to the other end.

well, also, if they were the same wattage, it would be completely useless (i think). the one on top would have to be more powerful than the one on the cpu for you to see any effects.

I've seen some already stacked TECs. Basically idea is to make low power one touch heat source and high power - cooler. There were even arrays of 4 or 5 TECs, but all were designed for heat sources that don't produce much power, due to narrow range of temperature at which TEC can operate.
This isn't going to work in my opinion with CPU, but perhaps would work with GPU. TEC touching GPU would be "normal", while the second one should be the real monster. Of course it will add tons of heat to loop.

More interesting idea in my opinion would be simple phase change or TEC + powerful phase change.

Aphex_Tom_9 said:

well, also, if they were the same wattage, it would be completely useless (i think). the one on top would have to be more powerful than the one on the cpu for you to see any effects.

Click to expand...

yeah you would need a more powerful tec on top of the weaker one david summed it up best when he said "at the end of the day you still have q joules to remove." the more powerful tec would be needed to take the heat going to the hotside of the first tec then youd end up with a runaway effect on the last pelt in the series where you would get the total heat from the cpu and the 2 pelts all combined for a staggering amount of heat to deal with.

just another one of my ideas i only thought halfway through before getting excited.

Theres a ratio by which you need to stack them, but overall its not a good approach to our applications. (The top pelt needs to be stronger than the bottom pelt)

I can see about hooking you up with more details when I'm not at work and don't mind google-whacking. I have read some fine articles on this.

The simplest answer is that there is no pelt powerfull enough to serve as your second pelt. If you are using it on your 3.2e then you would probably use the 226 watt pelt, which combined with the heat of the cpu, would require 350w (in the most optimistic scenario) to be removed by the second pelt. Nothing on the market comes close.

Yeah, I have done som excel calc on stacked pelts.
Most of the problems is highlighted by others already.

For a stacked config to beat a single stage you'd need 3 times the pumping effect on stage two.
As far as I remeber two single tecs side-by-side IN THE END beat 1 + 3 TECs dual stage!

Why? two reasons:

Well, the heat output from a 1 + 3 TEC study went really(repeat 5 times) high(we're talking kWs). That heat increased the watertemp so much that the hotside of TECs got far above ambient making cold-side follw up too.

And... TECs are made to operate with hotside temp at +40 to +65C. Going down in temps on hot-side makes the TECs (more) inefficient.

Well, for all purposes the idea was rejected by me, I am think more "parallel" nowdays.

I saw on frozencpu.com that theres a new TEC out "430-something" W pumping. Requires 24V 25A or something... still, doesnt cange much, except I can sell my dual 226W and buy that one instead.....

LabRat23 said:

The simplest answer is that there is no pelt powerfull enough to serve as your second pelt. If you are using it on your 3.2e then you would probably use the 226 watt pelt, which combined with the heat of the cpu, would require 350w (in the most optimistic scenario) to be removed by the second pelt. Nothing on the market comes close.

Click to expand...

i was thinking more gpu than cpu. i already have cpu taken care of and was pondering ways to get my gpu as cold as i possibly could.

So maybe some phase change instead?

Jfettig had great results with pelt cooling his GPU, have you looked at his thread at all?

I think pelt cooling is far more appropriate rather than phase change on the GPU.

yeah id most likely go with a pelt over a phase solution for my videocard phase can be such a hassle on videocards it seems and as much as im always switching things out i wouldnt want to have to mess with the evap block on the gpu eachtime i changed my mobo/videcard.

I.M.O.G

Jfettig had great results with pelt cooling his GPU, have you looked at his thread at all?

I think pelt cooling is far more appropriate rather than phase change on the GPU.

Click to expand...

It certainly is.

Peltier cooling a GPU is pretty easy, and can lend some amazing results with some voltage, a well planned watercooling loop, and a bit of work. Sub-ambient load temperatures make a huge difference when it comes to overclocking, peltier cooling is as effective as it is fun. I'll never run a non-peltier cooled video card ever again in my life, after having tryed peltiers out with my old 9800 Pro .

Ven0m

More interesting idea in my opinion would be simple phase change or TEC + powerful phase change.

Click to expand...

Peltiers get more efficient, as the hot side temperature increases. With a very low hotside temperature, they start to act wonky.

Hellsing was a superb series by the way, Alucard is pretty cool

I am unsure of the scientific explanation for why they do not work at low temperatures, but I am certain that they don't work under phase because of this


I do know, that to get some amazing temperatures with peltiers, for a benchmark session, one can simply hook up a peltiered waterblock, to a pump and a 10 gallon bucket of ice water (no radiator required - the ice acts as a heatsink). You replace Ice as you go, and can apparantly get temperatures on a GPU that are comparable to single-phase setups.

The trick is courtesy of Charlie on XS, I didn't think it up personally .