TEC improving.

[Ven0m]

So peltiers are common and so on, but there's problem with performance.
Lots of heat goes from hot side to cold side inside TEC. Air is a good insulation, but have anyone of you tried to minimize heat transfer in this area?

My new block for GPU is TEC-ready, for VGA isn't finished yet but should also support TECs, but performance of present units greatly prevents me from buying one (additional 200-300 watts is crazy for me to slightly boost cooling).

Anyway what I ask for - is insulation improvement. Maybe some weird gas / vacuum solutions? Vacuum could be nice but I don't know how much it would improve performance. I would need to get TEC, lap it and remove air from inside...

Plz share your opinions and experience.

 

[shiyan]

the heat at the hot side is more than just the heat removed from what you are cooling, and the heat removed from the air on the cold side.

the TEC itself draws a lot of eletricity and acts as a heat source.

so the total amount of heat you have to remove at the hot side in watts is:

heat of what you need to cool + heat from air + Amps x Volts (of the TEC)

I think the main reason most TEC users insulate is not to reduce the heat load that has to be cooled, but to prevent condensation forming. Whenever air containing some water touches the cold side, water could condense there and short out the motherboard or graphics card etc.

 

[Cool Frank]

The hot side of the Peltier is a real problem. I have been thinking about a solution, but it doesn't look that feasible. Basically, it would envolve a stacking of two peltiers. First, put a really thick copper plate on top of the CPU and a peltier right on top. To conduct the heat away to a point right on the PC case, I suppose it might be possible to use another copper bar (or maybe a heat pipe). A last Peltier would transfer the heat out of the case to a really huge heat sink. The drawback of such a hypothetical system would be the very large current drain and requiring the final heat sink to dissipate the heat generate by two peltiers.

Another solution would be to get a good understanding of heat transfer and to use only one peltier with a properly optimized geometry to conduct heat away from the CPU to a single large Peltier on the case with proper cooling on the other side.

 

[shiyan]

I have no experience stacking peltiers, but I don't think it's a good idea to try stacking 2 identical peltiers.

a peltier adds its own heat to pumped heat. A rated 220w peltier and a processor putting out 100w would mean something like 400w of heat on the hot side. There's no way that 220w rated peltier can pump that 400w of heat.

stacking a big peltier with a small peltier is possible, as the combined heat of the small peltier and heat source might be within the heat pumping limit of the large peltier.

using a passive heatsink to cool a peltier is only possible if that peltier is very low powered. if a peltier is not cooled effectively, it begins acting like a hot plate.

 

[gusgizmo]

are you trying to do a passively cooled peltier system? if not, why not add water cooling to your setup. the drop in the temperature on your hot side would reduce your cold side temps as well. this is what most high end extreme cooling guys do with peltiers. also, apparently putting two peltiers side-by-side on a heat spreader is better than stacking them, because of the reduced effiency at lower temps.

edit: also, after reading the reply posted while i was typing, i agree to this point as well. a peltier pulling 120 watts of heat off the processor in addition to the 200 watts it consumes in electricity puts out 320watts, which would require a pretty hefty peltier setup.

and heres a question: is the amount of heat a peltier is rated to be able to pull off of something absolute? or can it be variable depending on the hot/cold side temps? if so, does the difference even matter?

 

[shiyan]

putting them side by side sounds a lot more sane. there are waterblocks that support this, like the Maze 2-2 (to name something really old)

peltier cooling is only worth it if you have watercooling or very high end air cooling.

 

[SpaceRangerJoe]

are you talking about the inefficiency of peltiers? like, a whole lot of heat is transfered from the hot side of the pelt to the cold side, instead of being transfered into whatever medium is cooling the pelt (ie. water cooling?)

if so, this is a design flaw of peltiers. there is a physical metalic connection between the two sides of the peltier. this connection must be present in order for the current to move from one side of the pelt to the other, there for moving the heat.

i read something a while back about a company called coolchips (www.coolchips.com). they have developed a much more efficient thermoelectric cooler. it works on the same general principle of a current peltier. you apply a voltage, and the device pulls heat from one side to the other. however, this device has a small gap between the two sides of the cooler. they developed some way for the electrons to jump the gap as it were. this lack of a substantial metalic connection between the two sides of the chip results in a far for efficient device.

was i just rambling there, or did that help to answer your question?


edit: peltiers will have a max temperature difference between its hot and cold sides (the delta T). as i recall, as you approach the extremes, the real world delta T starts to drop very quickly. for example. if you have very cold water (around 0C) cooling a pelt, compared to room temperature water cooling a pelt (say 22C), the temp on the cold side of the pelt will not be 22 degrees cooler on the chilled water setup. this is because of the inherent inefficiency of the device itself.

summary - a 5 degree drop in temp of the hot side of the pelt, does not necessarily mean a 5 degree drop in the cold side temp.

 

[gusgizmo]

yeah, i think i know what hes saying now. he talking about thermal transfer outside of the peltiers normal operation. example: heat is pulled from the hot side to the cold side, but some of the heat goes back through the peltier in the air between the cubes of thermoelectric material, so it must be pulled back through again.

this in mind, you might be able to get a significant drop in temps, though maybe not worth the effort of vacuum insulating the inside of your tec.

 

[shiyan]

First time I've seen that website. Very cool. Something to keep an eye on I suppose.

 

[SpaceRangerJoe]

yeah. they been talking about patents and stuff for over a year now, and i havent seen anything about an actual product, so... but if they get anything like the results they expect, it could be a big step forward in cooling technology.

 

[Ven0m]

Nice stuff, if tests show results similiar to calculations, it will be new era in cooling (anything)
Anyway it looks like they won't come soon and that they will be expensive due to technology used

Anyway I was intrested with improving efficiency of existing pelts, for example by trying to have vacuum between sides. The problem is that I don't know how much it would give.

If it's not worth tryin,g I'll just complete my conventional water cooling rig and eventually try some phase change in future as unmodded current TECs have not enough performance to make me buy one.