2x80watt peltier stack?

[Silver]

The TEC cooling the TEC that is cooling the heat load absolutly has to be larger than the first TEC. For the cold side of a TEC to be colder than the Hot side, the TEC has to be pumping less heat than it is rated for. If a 100 Watt TEC is given a 100W load, the Cold side won't be colder than the hot. If you try to get a 100 watt TEC to cool a 150 Watt TEC and a heat load, well you won't be cooling a thing. All the the pelts and the heat load will quickly heat up to the point of them all frying.

Gotta tell ya I agree with you in most everything except that the tec cooling the tec. Lets reduce the heat load that the second tec sees and we should have a viable solution. Call it augmented tec cooling if you will. If the second tec only sees a 60 watt load then all is good (w/ an 80w tec). It is happy and the 226w sees a hot side that is better than ambient would be. Possibly an air cooled second tec.

 

[squeakygeek]

Gotta tell ya I agree with you in most everything except that the tec cooling the tec. Lets reduce the heat load that the second tec sees and we should have a viable solution. Call it augmented tec cooling if you will. If the second tec only sees a 60 watt load then all is good (w/ an 80w tec). It is happy and the 226w sees a hot side that is better than ambient would be. Possibly an air cooled second tec.

Do you mean cooling a 226watt pelt with an 80 watt pelt? That's definately a recipe for disaster. Having a 226 cool an 80 might work if the original heatload was low enough, but it won't work for a cpu. Even then, you aren't going to be able to cool the stack with regular air cooling because you're going to have a heatload of over 500 watts. You'd likely get better temps with a single 226.

 

[matttheniceguy]

maby I'm not reading what you are saying right Silver, but I think you are way off on how TEC's work. They are heat pumps, and they do not remove heat, they transfer some heat and generate some more heat.

Imagine you have a TEC on a heat source of say 100 watts. To cool this heat source you need a TEC that can pump at least 100 watts, and for there to be any difference in temperature between the cold and hot side, you need a bigger pelt than this. Let's say we are using a 150 watt TEC. Now, the TEC pumps heat away from the heat source, and also generates some heat. This means that the heat generated at the hot side of the pelt will be equal to the heat source plus the generated heat from the pelt. This means that there is actually something like 200 watts of heat ariving at the hot side of the pelt, which must be delt with. This is where watercooling normally comes in, but if you want even colder temperatures, you *could* use another TEC. Unfortunatly this second TEC has to cool a heat load of 200 watts now, so you will need a massive TEC rated at something like 350 watts. This massive pelt will then be pumping 200 watts and generating another 200 or so, and you now have 400 watts of heat to deal with on the hot side of that pelt.

The idea of putting pelts in series or cascade like this does work, but only for very small heat loads. If you only have a 30 watt heat load then you only need a 50 watt and 100 watt TEC to cool it. If your heat load is 100 watts, then you will need 200 and 400 watt TEC's and it just isn't feasable.

 

[Silver]

Do you mean cooling a 226watt pelt with an 80 watt pelt?

Of course. Common sense though would indicate that in order to make a cascade of sorts by using a lower wattage tec then supplementary cooling would be needed as well. How about we try the 80w tec directly cooling the 226w as a supplement. Of course a cold plate would be required between tecs.

 

[squeakygeek]

maby I'm not reading what you are saying right Silver, but I think you are way off on how TEC's work. They are heat pumps, and they do not remove heat, they transfer some heat and generate some more heat.

Imagine you have a TEC on a heat source of say 100 watts. To cool this heat source you need a TEC that can pump at least 100 watts, and for there to be any difference in temperature between the cold and hot side, you need a bigger pelt than this. Let's say we are using a 150 watt TEC. Now, the TEC pumps heat away from the heat source, and also generates some heat. This means that the heat generated at the hot side of the pelt will be equal to the heat source plus the generated heat from the pelt. This means that there is actually something like 200 watts of heat ariving at the hot side of the pelt, which must be delt with. This is where watercooling normally comes in, but if you want even colder temperatures, you *could* use another TEC. Unfortunatly this second TEC has to cool a heat load of 200 watts now, so you will need a massive TEC rated at something like 350 watts. This massive pelt will then be pumping 200 watts and generating another 200 or so, and you now have 400 watts of heat to deal with on the hot side of that pelt.

The idea of putting pelts in series or cascade like this does work, but only for very small heat loads. If you only have a 30 watt heat load then you only need a 50 watt and 100 watt TEC to cool it. If your heat load is 100 watts, then you will need 200 and 400 watt TEC's and it just isn't feasable.

The only thing I add is that you're being very optimistic with the heat that peltiers generate themselves. At max voltage, a 226 watt peltier will generate 374.4 watts of heat itself.

 

[matttheniceguy]

wow..... I didn't realize they were that inefficent at peak voltage. OK, so what I said above still stands, but now it is a 100 watt heat load, a 226 watt TEC, and a 600 watt TEC. Anyone finds a 600 watt TEC they can give it a go, otherwise just forget it.

 

[Silver]

I've heard this sort of thing before however.....Following the 'norm' is to get normal results. Gotta think outside of that. At the time this idea was rather noval (sp). Have had several others, some worked some did'nt however ones own logic has to superceed that of the normal thinking.

http://www.ocforums.com/showthread.php?t=108848&page=1&pp=30&highlight=dual+tecs

Let's pose this another way, you are given a 226w tec and told that you must cool it by using a 80w tec. Anything else may be used as well however phase change may not be. How would you solve this? Through another 226w on top of the 80w? Cooling would be a real b......... Let's come up with some workable solutions. I have one that I think is valid and workable however maybe some others would have a better solution. Do put them forth.

BTW, unworkable is not a solution to the problem and would definately indicate to me that no thinking on the problem has taken place. Anyone?

Ahhh...to make the tec work you connect one wire to .... no ...its' watts vs....no, I always get confused at this point.

 

[Silver]

wow..... I didn't realize they were that inefficent at peak voltage. OK, so what I said above still stands, but now it is a 100 watt heat load, a 226 watt TEC, and a 600 watt TEC. Anyone finds a 600 watt TEC they can give it a go, otherwise just forget it.

Definately true, on the efficiency. I have found mine to perform best at right around 13v. Kind of a balancing act. Can't find a 600w tec (at least that I could afford) though there is a Russian (?) site I believe offering a 274w. At any rate forgetting it does'nt work for me. They go to bed with you.

Bad me, double post.

 

[zabomb4163]

phase change seems a lot more practical.

 

[Silver]

Does, does'nt it.

 

[zabomb4163]

lets see.

226watt pelt = 40$
360watt pelt = 60 (at least)
power supply for both = 150$ (at LEAST)
cooling system for pelt = 100 (I think that is a low #)

=350$ total for -20C coldplate temperatures at best.

vs

http://www.xtremesystems.org/forums/showthread.php?s=&threadid=36930
400$ for gary's chiller that will take a 250watt load down to -40C. pretty sure you would make up the difference in cost in a month saved on electric bill. Temps are much lower as well.

 

[Silver]

I must admit that I find the chillers one of the more interesting avenues. Not so much in that it can get really extreme cpu temps (though very good) as that it can supply effective cooling to both the gpu and cpu at a reasonable operating expense. This makes it a very acceptable solution to extreme cooling in my opinion. Down side is finding pumpable liquids at very extreme temps, pumps that can handle it and insulation on two lines to both cpu and gpu. What is Gary getting now? Last I heard Loaded temps on GPU and CPU overclocked were in the range of -20 or so?

 

[matttheniceguy]

Let's pose this another way, you are given a 226w tec and told that you must cool it by using a 80w tec. Anything else may be used as well however phase change may not be. How would you solve this? Through another 226w on top of the 80w? Cooling would be a real b......... Let's come up with some workable solutions. I have one that I think is valid and workable however maybe some others would have a better solution. Do put them forth.

I looked through that thread, but I think the pictures have been removed or something, I didn't really get what the idea or solution was.

The bit in the quote about a 3 pelt stack with an 80 inside 2 226W... well it's not just impractiacal, but fairly impossable. An 80 watt pelt can only pump 80 watts of heat with no temperature difference. To have the full 226 watts of heat going through it, the cold side would have to be much much hotter than the hot side. This means the pelt is doing more harm than good.

I'm all for using pelts in new ways to cool things, but physical limits make things difficult. The fact that pelts actually generate heat is a big problem.

To get better Temps with a pelt, you have to get the hot side of the pelt cooler. Putting a MASSIVE pelt on it can do this, but there are easier ways. Improving your watercooling will help, and if you want really good temps, running chilled water is a good solution. Hey, you could cool that 226 watt pelt by running a regular water cooling system, with the 80 watt pelt hooked up after the radiator to chill the water a little. The 80 watter can't handle the 226 all on it's own, but at least that way it can chip in.

edit:

Whoops, I missed the second page stuff. Yep, phase change chillers are deffinatly a good way to go. They are perfect for people who know a fair bit about phase change cooling, but don't know enough or have the tools to make a direct die system. They can also cool more than one thing at a time which is plus.

 

[Silver]

I think I would like to give a go at using water cooling on a 226 and then using a small tec to cool the all copper block. This has to be focused at the core only as augmented cooled. It would be directly cooling the 226 however only a small portion of it relative to size. The water should be able to handle the majority of the cooling with the smaller tec serving to only remove a portion of heat from the larger tec which could far exceed the wattage given out over the area covered. I think this would allow for a small portion of the 226w to see hot side temps equivalent to chilled water. Ideally the small tec would be air cooled though this may make it much more difficult. Then there is the design of the water block. Allow for enough cooling to cover 100w of so then leaving 60w or so for the 80 to cover. Problems, temp bleed off to water, relative vivinity to cpu core (read tec) and of course temp bleed off to the block itself. However if an end result could be a 'focused temp to the 226w over the core in an area of say 5mm x 5mm then the temp difference given the same area on the 226w would be exceeded directly over the cpu core and allow the 226w to see cooler hot side temps. I really do believe this could be done. BTW, dual 80w tecs on a verticle cold plate, air cooled is another one that I think would work well in regards to a "standard" tec cooling application. One fellow was going to try it but never completed. Dual 60's would probably work in that application and easily handled on a system with dual psu's, albeit with only 90w or so effective cooling.

In the foregoing the secondary cooling would be augmented cooling though still direct. Intent has nothing to do with cooling the water (a very poor means of cooling water tecs are) but of cooling a very focused portion of the lower tec while water carried the bulk. The water block serving as a cold plate of sorts to the smaller tec. Advantages to tecs of course is the relative ease, clean setup (regarding lines and line insulation). Downsides are max temps of -40 (though in most cases not met), power usage.

BTW, on the dual 156's as well as single 226w tecs (re 234w and 169w roughly) bong coolers proved very effective. Attained idle temps were -15 with load temps in the range of 9c (dual 150s). Took the 2.8c to 4Ghz (D1 stepping). At the time, very close to mach 1 cooling without refrigerated cooling. DD did not follow up on it though Swiftech did make a dual tec for Intel as a special project of sorts at Intels request.

Something along these lines (I recognize this presently is not workable as it is not focused enough to allow for small tec usage).