Can Water be too cold in a Peltier System?

[vfrjim]

I was thinking of running a custom setup of cooling through a baseboard radiator sections (12 feet of length) 3/4" copper in a crawlspace and pumped back into my room, total lift on the water will be about 2 1/2 feet, so flow should be fine. In the wintertime, temps of the coolant will be ranging from 0 F to 40 F, will this be too cold to be used in a dual 156watt peltier system? Because I read about chillers in watercooled peltier systems and they recommend that you don't run it flowing fully through the chiller and to use some sort of restrictor on it. Anybod know this answer? Thanks!

Jim

[hielko]

Water for a peltier system can't be too cold. (unless it's colder than -120 C, at that point the peltier effect no longer exists...).

[futRtrubL]

Or lower than 0C at which point water flow no longer exists. You might then consider adding antifreeze. There really is no reason to bypass (or slow water down in) a chiller. Get low enough flow and the chiller will ice up and not be usefull. The pelt will draw more heat from a faster flow as average temp. in the chiller will be higher causing greater efficiency. People say that lower flow will help as the temp of the water will be lower when it reaches the cpu (and it will be) but this water spends more time in the cpu block too rising to a higher temp.
Edward

[g0fvt]

Yes basically the colder the better!

I have seen a so-called commercial water-cooling site that uses a bypass circuit around a peltier assembly... this is wrong and only works because there is a very fundamental error elsewhere in the circuit!

They talk about the peltier being "overwhelmed", too much pseudo science!

[Kevin]

If an inline system has no reservoir, then 0 degrees water is probably okay... even a bit under. It is hard for water to freeze if it is constantly flowing (note the rivers in Colorado and stuff). A few drops of anti-freeze should make it all good though.
-Kev

[Butcher]

At 0C or just below with no antifreeze you might get ice particles in the water which could collect somewhere (typically a joint or corner) and form a blockage which would reduce flow, it's unlikely it'd ice solid though unless you're talking major cold, I think 0F is probably major cold as it's -18C which is below what most antifreeze recommends. That is assuming the weater gets that cold given it'll be running over a cpu / pelt to keep it warm

[inda]

They make a pink water based anti freeze for winterizing water systems it can go down to -40c and is cheap and non toxic you can pick it up at most hardware stores for a bucks on the gallon works really good

[Archon77]

Well, from my knowledge of TEC's and the peltier effect, there is an optimum temperature for the hot side of the TEC. It depends on what kind of TEC your using.

For the kinds of TEC's used in cooling computer systems, you achieve the greatest delta-T when the water is ABOVE freezing. There are a few reasons for that, one of them was mentioned. Ice crystals can form, even with antifreez, and that hampers water flow. Second, the effecience of the TEC decreases as the hot side temperature drops.

I've found, and even read elsewhere, that you achieve optimal delta-T and efficiency when the hot side of the TEC is 10-15 degrees below ambient, but above freezing. With ICE-71 peltiers (71 degree DeltaT, 13.4 volt, 8amp TEC), I try to sustain a 10-12 degree hot side temperature, and my cold-side easily sustains -20 under a full, sustained cpu load. It drops close to -52 when the system is completely idle.

I'm not achieving a 71 degree deltaT, its more like 60 degrees, which is expected. Its impossible to achieve perfect operation in a real-world situation.

Also, some other tips. Optimal TEC operation depends more on voltage than hot side temperature. For most high-efficiency TEC's, you need at least 13 volts for them to operate efficiently. Supplying less voltage quickly reduces efficiency. My ICE-71's, for example, operate quite poorly at 12 volts (I can hardly sustain sub-zero temperatures under load, and my lowest was -12 idle). Providing too much voltage can have a detrimental effect as well, lowering efficiency while producing too much heat. TEC's like to have stable current and voltage, and they have a very short "sweet spot" where their effeciency is optimal.

Hope this helps. Most of this comes from experience, but I learned a lot just by searching the web for information, much of which I found either from links here on ProCooling or on Overclockers.com.

[vfrjim]

I do appreciate all the input, but as of now, I decided to go water only on this system, with water holding 14C, under max load on a Tbird(1.55 @1.95v) core is staying at 17-18C, which to me is cool enough, maybe in the summer time, will go back to pelts. But I wonder how cold this water will get this winter, it seems like the water stabilized at appox 3C above air temp, do you think this will stay true for the rest of the winter? If so, damm its gonna get cold!!!

Jim

one suggestion... use greeen (glycol) antifreeze, and not too much of it. Red (dexcool) antifreeze has much worse thermal transfer efficiency

[WaterPog]

Archon77,

You may be correct that the pelts may have a "sweet spot" where the Delta T from hot side to cool side is the largest, but what you are forgetting is that it doesn't matter what the delta T of the pelt is.

Your goal isn't to optimize the pelt, it is to optimize the cooling of the processor so the delta that matters is between the cool side of the pelt and the Die in order to maximize cooling of the die.

[GigaFrog]

You can't get the 71°C in a normal setup, not because it is real-world, but because the rated delta T is what is achieved with no load. Once you put a load, the delat T decreases.
That's how I understand the way a Pelt is specified.

[PsyKoRealm]

Water can never be 2 cold in a watercooled/pelt system. As long as it flows its fine1!! My water temp are around -21C.

[Brad]

I'd have thought that as low as possible before flow rates died would be the best bet myself..