Peltier chillers cheap now?

[KazeoHin]

So, Amazon has 150w rated peltier thermoelectric chillers/generators for CHEAP Has anyone tried to use these? I know that a single chiller wouldn't really enable GREAT overclocks rated at 150w; but what would happen if I were to combine this with a traditional heatsync or radiator... would it work? also, would this allow SUPER low-profile builds?

I know the first rule of thermodynamics, the energy has to go somewhere, and these peltier chillers have a cold side and an equally offset hot-side. Would placing one of these between a waterblock and a CPU (using good-quality thermal-paste) actually help to increase the MAX TDP of your system, or would it be unnecessary power wasted?

 

[Arcygenical]

Tecs have always been cheap, it's powering them - and their power supplies - that cost a bunch.

That 150w transfer TEC will consume 300w when moving 150w. That's almost 20a. A 15v 20a supply is expensive. It will also be outpaced by most CPUs that produce well over 150w when overclocked (if placed between the CPU and a Waterblock).

There is no situation where you want to increase the total TDP of your system.

And finally, if you cool water below ambient - with TECs, in a system with radiators, the radiators will absorb heat from the ambient air, and exhaust cooler air into the surroundings, finding an eventual thermal median.

 

[skypine27]

I had a Peltier back in about 99-2000. I was cooling my OC'ed Pentium II or III. Forgot what they were called back then.

Mine looked very much like the one in your link, but was a bit thicker and came pre-wired with a 4 pin molex connector. All I had to do was stick it on there with some thermal paste and plug it into an empty PSU lead. I can't remember exactly how it was mounted to the CPU, like was it just "held" on by a glue type thermal paste, or some kind of clamp setup.

(edit, mine looked like this but thicker:
http://www.dansdata.com/pelt.htm)

One side of it was ice, literally, ICE (as in sub 0 C) cold. The other side of it was scalding hot to the point it would instantly melt any wire (and skin) that was unfortunate enough to touch it. You really have to guard the hot side of one of these things...

It sucked a lot more power of course than a normal CPU cooler with fan setup, but was not a big deal.

It cooled the CPU so much that the BIOS and temp monitoring software rarely showed it to heat up ABOVE 0 degrees C (as in, the CPU temp was always below 0 C when idle or light use). I remember this was a much talked about problem back then, so much so, that there were Prime 95 type programs designed to run when needed to constantly attempt to keep your CPU at greater than 0 C. The theory was that if you allowed your CPU to sit below 0C for too long, condensation would freeze, etc and the socket area would start to look like your freezer, and POP-FAIL.

I never bothered with keeping the CPU above 0. I lived in San Diego at the time and was a fairly well off 24 year old and didnt care if the CPU/mobo failed one day.

And one day, they did. After months and months of near 24/7 use at a far greater OC than I was able to achieve on conventional air cooling, one day, I came home and the system was black, never to boot again.

Anyway, I wouldn't mess with a Peltier these days except as an experiment on a CPU/mobo you can totally afford to lose.

 

[Cozmo85]

I had a Peltier back in about 99-2000. I was cooling my OC'ed Pentium II or III. Forgot what they were called back then.

Mine looked very much like the one in your link, but was a bit thicker and came pre-wired with a 4 pin molex connector. All I had to do was stick it on there with some thermal paste and plug it into an empty PSU lead. I can't remember exactly how it was mounted to the CPU, like was it just "held" on by a glue type thermal paste, or some kind of clamp setup.

(edit, mine looked like this but thicker:
http://www.dansdata.com/pelt.htm)

One side of it was ice, literally, ICE (as in sub 0 C) cold. The other side of it was scalding hot to the point it would instantly melt any wire (and skin) that was unfortunate enough to touch it. You really have to guard the hot side of one of these things...

It sucked a lot more power of course than a normal CPU cooler with fan setup, but was not a big deal.

It cooled the CPU so much that the BIOS and temp monitoring software rarely showed it to heat up ABOVE 0 degrees C (as in, the CPU temp was always below 0 C when idle or light use). I remember this was a much talked about problem back then, so much so, that there were Prime 95 type programs designed to run when needed to constantly attempt to keep your CPU at greater than 0 C. The theory was that if you allowed your CPU to sit below 0C for too long, condensation would freeze, etc and the socket area would start to look like your freezer, and POP-FAIL.

I never bothered with keeping the CPU above 0. I lived in San Diego at the time and was a fairly well off 24 year old and didnt care if the CPU/mobo failed one day.

And one day, they did. After months and months of near 24/7 use at a far greater OC than I was able to achieve on conventional air cooling, one day, I came home and the system was black, never to boot again.

Anyway, I wouldn't mess with a Peltier these days except as an experiment on a CPU/mobo you can totally afford to lose.

The hot side should have been under a heatsink/watercooler of some sort. A peltier is sandwiched between the cooling solution and the cpu. Otherwise the entire thing would be hot and not just one side.

 

[KazeoHin]

I wonder if there were a way to remove the radiator of a WC setup entirely and just use a small heat-exchanger/fluidbox linked up with a couple of adjustable-voltage peltier plates. These plates would expell the heat into the air via HSF, but that air would not be used to cool the system: thus no negative effects of equalization. I could imagine that the effect would be better cooling capability at the cost of higher energy usage+higher heat exhaust.

Edit: the setup would be describe as such: instead of a radiator, the system pumps water to another CPU (or universal) waterblock strapped to a peltier plate (that itself has a sizable heatsync). The whole idea is that the peltier plates can withstand much higher temperatures than CPUs or GPUs, so even though they in-the-end increase the system's gross heat, they actually keep the actively cooled components (CPU, GPU, VRMs etc...) cooler and even below ambient temperature. Yes, the heat output would be ridiculous, but you would never have to worry about your CPU or GPU temps again.

 

[PliotronX]

Holy shitsnacks, 10A on the 12V rail. I wouldn't mess with it for that power draw and also when I played with them in the P2/300A days, condensation was a problem and the resultant water dripped on and almost fried my Voodoo2 card. Just not a good idea for 24/7 use, I relegate it to LN2 areas..

 

[Deleted member 88227]

Hmm that does look like some interesting tech to play with.

 

[KazeoHin]

I see some on eBay rated up to 240w, with over 400w max consumption! That's pretty damn substantial. I could imagine that would catch fire quite quickly if left alone during a malfunction...

I'm thinking maybe using TECs connected to a temperatre-controlled fan controller. Set up the profile so they only kick in when the system gets really hot. Use my before-mentioned dedicated TEC waterblock idea and I would have a system that would be relatively low-power consumption (versus the 'always on' approach) and still be able to chill the hell out of the system during heavy gaming/ computing. This system would also reduce the effect the TECs would have on ambient temperatures, as they would only kick-in temporarily, and the ambient air would not equalize as quickly, unless you were doing some marathon gaming session or mining, what have you.

 

[PliotronX]

I see some on eBay rated up to 240w, with over 400w max consumption! That's pretty damn substantial. I could imagine that would catch fire quite quickly if left alone during a malfunction...

I'm thinking maybe using TECs connected to a temperatre-controlled fan controller. Set up the profile so they only kick in when the system gets really hot. Use my before-mentioned dedicated TEC waterblock idea and I would have a system that would be relatively low-power consumption (versus the 'always on' approach) and still be able to chill the hell out of the system during heavy gaming/ computing. This system would also reduce the effect the TECs would have on ambient temperatures, as they would only kick-in temporarily, and the ambient air would not equalize as quickly, unless you were doing some marathon gaming session or mining, what have you.

The TEC has to be active in order to pump the heat, if you remove the current it's going to insulate the CPU and possible kill it. Not to mention it is going to draw far more wattage than any fan controller ever conceived can supply. I recall some folks used the cold plate methodology but that was to reduce chances of condensation or act as a buffer on the hot side. It's just not worth messing with.

 

[KazeoHin]

The TEC has to be active in order to pump the heat, if you remove the current it's going to insulate the CPU and possible kill it. Not to mention it is going to draw far more wattage than any fan controller ever conceived can supply. I recall some folks used the cold plate methodology but that was to reduce chances of condensation or act as a buffer on the hot side. It's just not worth messing with.

I see exactly what you are saying, I do: but I just can't believe that an active heat pump that can chill below ambient is worthless to an overclocking enthusiast. I'm not saying you are wrong, but rather that someone hasn't been clever enough to make it work.

Edit: also: look a this badboy right here. This thing can do up to 10A at 25+v, more than enough to power two 12v peltiers in series. Hook the cold-sides to a couple universal waterblocks, hook the hot sides to a couple Hyper212 HSFs... It would not be cheap, over $100 in the end, and that is if you already have the WC pump, res and tubing. But it would sure as hell have cooling capacity. Maybe even enough to remove the radiator from the loop...

 

[PliotronX]

It would definitely not be [H] of me to advise against it but I'm just saying there are many hazards to this ageold technology. If it can work for you, all the better though you best post pics and results especially longterm stability. An independent powersupply makes the most sense but if that gets unplugged or fries itself it may take the CPU with it...

The radiator would still be absolutely necessary. You don't get cooling capacity for nothing with these, whatever the cold side puts out must be dissipated on the other side just as air conditioners require exhaust and collectively contribute to global warming

 

[wrangler]

Been a while but I do remember these sandwiched in between a waterblock (with a pump and radiator and water flowing through the tubing of course) and the cpu as a way to get lower than ambient but as always condensation and powering the peltier were problematic. I'm sure nothing has changed.

But, a lot of people did play with them back in the day.

 

[Arcygenical]

There's nothing wrong with using a Tec, but in over clocking and computing - cooling is just as important as efficiency sometimes.

An unpowered TEC is a substantial insulator (if it conducted heat, it would let its own heat back to the cold side!) and it consumes a huge volume of power.

Something like a large phase change chiller can do what you ask, with a much higher efficiency rating (in the order of more than 80% efficient where a tec is 45-50% efficient)

In your proposed setup, yes, those two tecs (not recommended to run in series though, run them in parallel with a 16v supply for maximum efficiency) would have a heat dump of around 250w or so... IF hooked up to a 16v 20a supply. the rig would,comsume around 320w to do this....Which is 1 r290 under load with no CPU...