• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Is a peltier water chiller a dead idea?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
wait...did it honestly say 32*C with a peltier chiller....my SLK-800 does around that (alebiet idle) - - and i was planning on doing a peltier chiller...heh...guess i'm not...
 
yes, yes it is.

Its not that it wont work, its just that you can do much better with a phase change based unit for no more money than a pelt based.
 
I've already invested in a watercooling setup but I'm looking for more another way to reduce temps even more. So if it will work then I'm interested! Would an 80w pelt be sufficient? I have an Antec 400w psu.
 
Doc had a intresting setup. He had a huge well insulated watercontainer, I don't remember how big but it was considerable like 30-60 liter well about 10-20 gallons.

Then he had a couple of small TECs I don't remember how much, but it wasn't even 100W. The aircooled TEC's slowly froze the water. When he used the PC the water slowly heated up, when he didn't use it it the water cooled down. During the PC on hours the water only gained a couple of C.

That way he could enjoy cool cpu temps about 5-7 hrs a day and during the non-use time the water got back to freezing.

It wasn't exactly portable and couldn't run all day but it was easy to build, use and maintain.
 
Cool, someone remembered. Thanks Paxmax. That system had a couple of 40 watt TECs on it. The main idea was to run them at less then half the rated voltage so I wouldn't need a large PS and the TECs are more efficient at voltages lower than Vmax. This also makes easy to air cool and the voltage could be switched up from 5 to 12 when needed.
 
80w really isnt enough for todays procs.

Dont expect to be able to run a pelt of your computer power supply. You will need one dedicated to the purpose.
 
The idea was to run off of the many gallons of stored cold water. This would allow me to run for hours while increasing the temp of the holding tank only a few degrees C. I did not run my system 24/7, this allowed time for the holding tank to cool back down. For those that don't run 24/7, this is an easy and cheap thing to do. The power requirements are low ( I used an old 200 watt psu), but the holding tank is somewhat large and must be well insulated. It works ok for me, so whatever floats your boat!
 
It's always been a dead idea.

TEC's go on what you're cooling directly.

Anything more is a fairly flagrant abuse of them; how muhc more inefficient can you get?

If you want to chill something with them, stick them under the waterblocks and get far better preformance out of them.
 
Well I have to disagree.

I can take an 80 watt peltier and use the cold side to chill water and dissipate heat from the hot side with a hsf. Which should (in theory) reduce water temps and increase my overclock. Or, I can take an 80 watt peltier and slap it on my already o/c'd p4 which produces over 100 watts of heat and fry it.

At least that's my idea? Or is all this wrong?
 
I can take an 80 watt peltier and use the cold side to chill water and dissipate heat from the hot side with a hsf. Which should (in theory) reduce water temps and increase my overclock. Or, I can take an 80 watt peltier and slap it on my already o/c'd p4 which produces over 100 watts of heat and fry it.

The reduction in water temps would not be worth it. a 80w would not put the water below ambient. Further, you really need a dedicated PSU for even an 80w. For the price of the psu, you could get a phase change based water chiller and be getting freezing temps.
 
Yes raskren you can pre-chill water with an 80W TEC in a big res. You can use the cold water to chill your 100W cpu for some time, until the water has gotten too warm, then you must give the cooler time to catch up.

I don't remember the formula for heating up water... doc would have to correct me on this one.
 
No, this is not a dead idea, but it is an expensive idea.
The reason commercial TEC water chillers are so expensive
is not greed. They are hard to design properly and no one
has really seen enough volume to get economy of scale.

In most computer environments you can go about 10°C
below ambient before consensation can form. Designing
a system with a radiator to dump the heat and a TEC to
cool the water this much is a very reasonable goal.
 
Tecumseh said:
In most computer environments you can go about 10°C
below ambient before consensation can form. Designing
a system with a radiator to dump the heat and a TEC to
cool the water this much is a very reasonable goal.

Exactly!!! That is what I want to do. I'm not looking to cool water to sub zero temperatures and I'm not looking to cool it far below ambient. I would like the peltier to simply cool the water more than it would be with just a radiator.
 
Still a decision must be made. If you aim to use the 80W your system won't be able to run 24/7. If you set your temp 10C below ambient you can use the PC until you hit maybe 5C above ambient, after that You'd be better off with aircooling alone.
So you've got a temp delta of 15C to play with.


If your CPU radiates 100W it will take 360L of water 1 hour to heat up 1C.
That is if you would shut the TEC off when you start your PC. Since the TEC cools the water at them same time as the CPU warms it the time will be much longer before temperature rises 1C.

How effective the TEC cooling will be is tricky to caculate, since we don't know what kind of cooling it has. Only thing you can bet on is that you won't get 80W of cooling from it. :)

If you get 50W cooling from it, then you've doubled the time to heat up the water (if your CPU generates 100W heat).

So, back to our example. 360L is alot of water and it has to be well insulated!

Let's say you plan to use your PC max 8 hours a day.

Since you've got 50W cooling(from an aircooled 80W TEC) at hand we can half the water.
360 / 2 = 180L
We can take an increase of temp of 15C....
180L / 15 C = 12 liter (now we are getting somewhere!!)
Oh no... we need the PC 8 hours a day though....
8 x 12L = 96 L !!!

Okay.. so 96 L (whats that? 25 US gallons) if your PC runs at 100% at 100W 8 hours a day with a 15 C temp increase... (we haven't even considered pumpheat and insulation leakage)

(Hmm... sounds like alot of water.. I hope someone can correct if my math or formula is failing !!!!! )
(I used the formula 1 Joule can heat up 1 gram of water 1 celsius. And 1kW/hour = 3 600 000 joules)

So... it's doable but bulky (unless my calc is dead wrong somewhere.... arrgh.. "learn first, then speak of it"... someone will probably make me eat my own post...)
 
Well, if you have a radiator before the pelt, that will get rid of a lot of heat. Problem: the water in one of these systems moves fast enough that it is almost the same temperature throughout the system. The closer to ambient you get, the less a radiator can do. In fact, if you are below ambient, a radiator works backwords.

In short, you will reach equalibrium, and it will be cooler than just using a radiator. But you'll never go below ambient. (Unless your pelt moves more heat than your CPU and pump make.) And the same results might be reached with an extra radiator and/or better fans.
 
Back