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

Peltier questions...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.


The devil’s advocate
Dec 23, 2001
Where the hottest chicks reside!
'lo guys,
Ok, new to pelts, Just built my wc setup, thinking to inject a peltier with it, my question is, what type of temps will I expect with 226w pelt? .. or is that too much? .. 172w is "ok" ? .. totally blinked out when it comes to pelts, your input is highly appreciated .. thanks :)
Pelt performance is highly dependent on hotside cooling and applied voltage.

Do you know how much power your CPU dissipates loaded, and what temperature your CPU runs at when loaded for a long period?

Those two pieces of information provide a rough C/W of your watercooling system.

With that data I can give you some ballpark estimates of pelt performance.
Still need to know the CPU power consumption. The Computer Nerd calculator can give you a ballpark number.

Also need to know ambient.
I calculate this stuff using an Excel spreadsheet I created.

You can read details and download the spreadsheet here.

The one I use is actually updated a bit from that one, but I'm not going to post the newer version until I get a chance to do some testing to tune its accuracy. I'll probably ask that it be made a sticky then.

The biggest issues affecting its accuracy are: how well the C/W of the hotside cooling system is known, and how well the loaded CPU power consumption is known. Another big issue is the C/W of cooling/heating through the CPU pins and motherboard. I just kind of tack that on as a fudge factor to the number the spreadsheet calculates.

There are so many variables in watercooling systems and CPU temperature sensors, that a C/W guess I would make from a description of someone's WC components probably has at least a +100%/-50% margin of error. And the C/W value for the WC system has an enormous effect on the final result.

The best I can do is say that; IF the CPU temp sensor is accurate, and IF the Computer Nerd Thermal Calculator is accurate about loaded power consumption, then the actual CPU temperature will probably be between 3 and 10 degrees higher than the number my spreadsheet calculates. (Those are both very big "ifs".)

So, to answer your question, yes there is an equation, but it is one big ugly MOFO.
Perhaps the most important factor in pelt performance is the
quality of the mounting and proper pressure. For proper operation
the pelt must be sandwiched at a MUCH higher pressure than
you would ever mount a CPU, even a P4.

This will make all the difference between success and failure.

I'd downloaded the software a while back but failed to find the calculator. I just thought it generated graphs.

That calculator is pretty handy. I compared my spreadsheet to it and was pleased to see how close I came. For a 172 Watt pelt, I came up with a number 0.3C higher than Kryotherm at 75% Vmax, and 5.1C higher at 33% Vmax. When I came up with my equations I deliberately ensured I would err on the conservative side.

My spreadsheet doesn't have a cell for Rins. I maxed that value out on the Kryotherm calculator to do my comparisons. I'll have to add that to my spreadsheet. (As I mentioned earlier in the thread, I've just been swagging that number in.) I should be able to get a ballpark number for Rins by analyzing Joe's heatdie versus CPU data points. The low accuracy with which actual CPU power dissipation is known severly limits how accurate I can get though.

One advantage my spreadsheet has, is that it is somewhat more general purpose. The Kryotherm calculator requires that you choose from among specific pelts that Kryotherm sells. I didn't find a 220W or 226W pelt that could be selected. With my spreadsheet you just put in Qmax, dTmax, Vmax, and Imax for any pelt, and it calculates the results based on that.