using thermocouples to cool, efective?

[athlonnerd]

i know that space probes use decaying plutonium to generate heat, which by means of thermocouples is used to power the space craft. how huch power do thermocouples put out? if its a substancial amount, coulnt we use them to absorb cpu heat, thus cooling the cpu, adn disposinf of the heat, rather than moving it? casue with a total power output of 475 watts, using as pelt, that will notiable rais ones rom temperature, which raises core temperature. but if we absorbe teh energy, then it wouldnt heat the room. what one does with the absorbed eneregy i dont know.

or is thjere another way of doing away with heat, not just moving it? i know that you cant destroy energy.

it would be easier to move the absorbed energy outside with wires, and put the wires on a heating element.

 

[Gravity Man]

As far as I know, thermocouples are used for measuring heat, not absorbing/removng it.

 

[SkiFletch]

yeah, me too, are you sure you have your terms right? as far as i know, thermocouples are 2 wires of known conductance connected ttogether at the tip with a known conductance. a device can then be used to measure the changes in that known conductance with temperature. i've never heard of them being used in active systems....

 

[Bender]

Your all correct in one way or another. Thermocouples are used to measure temperature difference but if a current is run through a thermocouple it creates a temperature change. You are not absorbing heat by doing this, its called the peltier effect. Heat is transfered from one side of the thermocouple to the other. The cooler you keep the hot side the cooler the cold side will get. Peltiers work by using hundreds of thermocouples to produce a temp difference. The more thermocouples that can be packaged in a peltier the more heat it can transfer. Peltiers arent magical heat absorbers, they can only transfer heat.

BTW I used to make thermocouples and design them in turbocad

 

[athlonnerd]

i know how thermocouples work and how peltiers work. maybe i didnt word my question right. i read an article one, which explained thermocouples. http://www.howstuffworks.com/question136.htm go here also http://spacepwr.jpl.nasa.gov/rtgs.htm ,as a quote:

RTGs are a compact space power system which convert thermal energy from the decay of radioisotope materials to electrical current through the use of solid-state thermoelectric converters. These power systems are reliable even when used in the harsh environment of outer space. RTGs are not nuclear reactors and have no moving parts. They use neither fission nor fusion processes to produce energy. Instead, they provide power through the radioactive decay of plutonium. The heat generated by this process is changed into electricity by solid-state thermoelectric converters.

i have yet to read the article, but i do believe that thermoelectric converters are thermocouples.

 

[sillyunclemark]

If you have an older gas furnace, the same idea is used to shut down the gas if the pilot light goes out. Heat from pilot light generates enough power to hold the gas valve open for the pilot and lets the main valve open when needed. I don't have numbers in front of me but, this requires a substantial temperature difference and is rather inefficent as I recall.

 

[Bender]

If I'm not mistaken even a thermocouple will produce minute amounts of current if the 2 sides are held at different temperatures. It would be very inefficient and not wroth using in a PC.

 

[parkan]

If I'm not mistaken even a thermocouple will produce minute amounts of current if the 2 sides are held at different temperatures. It would be very inefficient and not wroth using in a PC.

Isn't that what they use on submarines sometimes? Keep one end near the surface and submerge the other one to a great depth?

 

[MILLTHERM]

I have an old electric handbook from 1914 (AUDEL's I think) that shows a picture of a thermocouple battery which was capable of producing 100W it was pretty cumbersome and only about ten% eff.

I know there was allot of research done on it in the 50's I read some papers on it at some point they could not improve much since the early efforts.

I often thought that with modern photolithography you put 10 million thermocouples of copper/constatine on some kind of substate. you could set them up in series or paraell any way you want.

I know that you can run a wire 100 ft before you get voltage changes but I do not know the lower length is but suspect it is limited by the gradient so you would need a front to back arragement with an insulator in between-maybe epitaxial layers?

 

[Bender]

Isn't that what they use on submarines sometimes? Keep one end near the surface and submerge the other one to a great depth?

If they do I have absolutely no idea why. This one is beyond me.

 

[Diggrr]

Most TEC's (but not all) will produce electricity if they are heated on one side, and heatsinked on another, the same as would a single thermocouple.
I think I get your point...say if one were to take a TEC or other thermocouple array, and put in on the processor, add an HSF on it. Now instead of powering it to use it as a cooler like most do, use it's electrical production to dissipate the heat energy with.
Well, the processor would have to hit self-destructing temps to start the reaction you're looking for. To produce electric, they would have to hit the Delta T difference from hot to cold side. If it's Delta T is 50C as many TECs are, that means 30C coldside, and 80C on the hotside....ouch.

Still an intruguing idea that could be done with something other than a TEC though. The dissimilar metals in direct contact instead of a ceramic interface would help efficiency alot.

 

[athlonnerd]

thats what i was talking about. you coaul always use pelts to increase teh delta t, and if they got 100W in 1914, then im shure the technology has improved some by then, even if not by alot. the energy produced, could be disipated else where, or used to power a fan or something. maye could drive a chemical reaction to store the energy, adn reverse the reaction later to dissipate it, kinda like using a battery for wwhat it isnt meant for. maybe the heat could drive a reaction. isnt there some other way to convert heat into electrisity or some other form of energy?

 

[parkan]

Such system would be inefficient. The energy required to power the second pelt would probably exeed the output of the thermocouple.

 

[athlonnerd]

yeah, but tis just a concept, coolchips are more efficient than pelts. one could always use sterling engines. http://www.howstuffworks.com/stirling-engine.htm i think i might just build one of these, looks cool. i dont know what you woul;d do with the mechanical energy though. just ideas.

 

[Bender]

Very interesting ideas. If efficiency were much higher it would make a good CPU cooler. You could even pump the current generated back into the system for lower power consumption.

 

[MILLTHERM]

Getting a free lunch out of Physics is no easy task. What's the real eff. on a powered pelt, 50%? in passive mode 10%?

That's why I suspect things like the cool chip concept. they may get eff. of 50% in the lab, but they require something like 15 nanometers seperation between the two sides. flat and parael over a one inch span to those specs completlely isolated from one another is a non trival effort, even for the semiconductor industry when any kind of volume comes into play.

 

[athlonnerd]

thanks for all the input, just food for thought. i thought that pelt got like 60% efficiency

 

[Jack Waugh]

www.heatsink-guide.com/peltier.htm
and yes, I see that this discussion is eleven years old.

 

[Crapxxas119]

I am sad to inform you that trying to cool a cpu with any device that generates electricity is still not very feasible. With a peltier you still have to cool the energy emitted by the cpu and then an additional amount of energy that the peltier uses to create the temperature change.

 

[Automata]

and yes, I see that this discussion is eleven years old.

Thank you for taking the time to use the search feature, but there is no reason to bump an old thread like this.

I will officially let this thread rest in peace.