Thermal Electric Elements?

[EETanner]

With all the heat coming from this hardware, has anyone considered basic thermal electric elements?  These generate energy via the differentiation in temperature on one side over the other.  I would assume placing between the heat sink and the chip would be ideal.

I"ve been you tube videos where the cpu fan was actual run from the electricity created by the temperature differences.

Thoughts?  Cost analysis? 

-ee

[Foxpup]

You do realise that thermoelectric generators insulate the heat source from the heat sink, right? Otherwise the heat just gets conducted straight to the heat sink without producing usable energy. Needless to say, this is precisely what you don't want on your mining hardware, as it will result in greatly increased temperatures, which (since the chip has a constant volume) increases the pressure of the magic smoke. This rising pressure is very likely to rupture the chip's casing, releasing the magic smoke and rendering the chip useless. While you could use the output of the thermoelectric generator to power an additional cooling system, a conventional heat sink provides far more effective cooling at a fraction of the cost.

[Foxpup]

Last time I checked those things had such low outputs that the time for return on investment made them impractical for most applications.

Actually, they are practical for some applications. Just not applications where you absolutely, positively, need to get rid of waste heat.

Would be very interesting if they've advanced much, the ability to generate electricity just from temperature difference has huge potential.

No it doesn't. As I previously mentioned, the heat source and the heat sink must be thermally insulated from each other in order for a thermoelectric generator to work. And if they are thermally insulated and you have no other cooling system, the chip burns to a crisp. There is no potential here.

[Luno]

Not right Foxpup, the power transmitted through the thermoelectric element can be enough to cool the gpu. 100W of heat conducted through a TEC will give you 0.7 watts of electric power: GPU 65 degrees centigrade. ambient 25 degrees. Enough to run a fan on the heat sink. Calculator here: http://www.powerchips.gi/technology/pcalc.shtml

To use 50$ on a TEC to save 0.7 watts of power is not worth your while still.

[Foxpup]

Not right Foxpup, the power transmitted through the thermoelectric element can be enough to cool the gpu. 100W of heat conducted through a TEC will give you 0.7 watts of electric power: GPU 65 degrees centigrade. ambient 25 degrees. Enough to run a fan on the heat sink. Calculator here: http://www.powerchips.gi/technology/pcalc.shtml

To use 50$ on a TEC to save 0.7 watts of power is not worth your while still.

I never said it wasn't possible. I said it was less effective and more expensive than conventional cooling, and that the chip absolutely has to be cooled by something other than the heat sink. It's not impossible, just very impractical.

[MoonShadow]

Those things are terriblely inefficient, couldn't make enough to justify their cost in this scenario.  Better to just use the waste heat to offset building heat needs in winter.

[hardcore-fs]

Wow a workable perpetual motion machine......

[Foxpup]

Wow a workable perpetual motion machine......

No, it's just a heat engine. Heat engines produce energy by transferring heat from a heat source to a heat sink. In the absence of an external energy source, this will eventually result in the heat source and heat sink reaching thermal equillibrium (ie, the heat source gets cooler and the heat sink gets hotter until they're both the same temperature), at which point no further heat transfer can take place and no more energy can be produced. To continue producing energy, you need to either actively heat the heat source or refrigerate the heat sink, both of which require you to put in more energy than you can get out.

Heat engines can be useful for recovering energy which would otherwise be lost as waste heat, but can never produce more energy than you were wasting as heat in the first place (and in practice, they don't even come close to that).

[firefop]

Wow a workable perpetual motion machine......

No, it's just a heat engine. Heat engines produce energy by transferring heat from a heat source to a heat sink. In the absence of an external energy source, this will eventually result in the heat source and heat sink reaching thermal equillibrium (ie, the heat source gets cooler and the heat sink gets hotter until they're both the same temperature), at which point no further heat transfer can take place and no more energy can be produced. To continue producing energy, you need to either actively heat the heat source or refrigerate the heat sink, both of which require you to put in more energy than you can get out.

Heat engines can be useful for recovering energy which would otherwise be lost as waste heat, but can never produce more energy than you were wasting as heat in the first place (and in practice, they don't even come close to that).

google: stirling engine.

I've been thinking about developing a water cooling system that uses a stirling engine to cool the chip for a net gain. the only issue I see with it... is stirling engines make noise.

[Foxpup]

I've been thinking about developing a water cooling system that uses a stirling engine to cool the chip for a net gain. the only issue I see with it... is stirling engines make noise.

There's an even more obvious issue: cooling a chip and using it as a heat source for an engine (of any type) are mutually exclusive goals. If you have an effective cooling system, the chip won't get hot enough to provide enough energy to run it, as preventing the chip from getting hot is the whole point of having a cooling system in the first place. If it does get hot enough to produce a practical amount of energy, your cooling system is obviously not doing its job. It's most unlikely that any sort of heat engine will allow for more effective cooling than a simple heat sink; that's simply not what heat engines are designed to do. Like I said: possible, but almost certainly not practical.

[hardcore-fs]

Wow a workable perpetual motion machine......

No, it's just a heat engine. Heat engines produce energy by transferring heat from a heat source to a heat sink. In the absence of an external energy source, this will eventually result in the heat source and heat sink reaching thermal equillibrium (ie, the heat source gets cooler and the heat sink gets hotter until they're both the same temperature), at which point no further heat transfer can take place and no more energy can be produced. To continue producing energy, you need to either actively heat the heat source or refrigerate the heat sink, both of which require you to put in more energy than you can get out.

Heat engines can be useful for recovering energy which would otherwise be lost as waste heat, but can never produce more energy than you were wasting as heat in the first place (and in practice, they don't even come close to that).

google: stirling engine.

I've been thinking about developing a water cooling system that uses a stirling engine to cool the chip for a net gain. the only issue I see with it... is stirling engines make noise.


Some people need too develop a better sense of humour. (The only reason a photographic memory is useful)

http://dilbert.com/strips/comic/1993-10-10/

[Luno]

I've been thinking about developing a water cooling system that uses a stirling engine to cool the chip for a net gain. the only issue I see with it... is stirling engines make noise.

There's an even more obvious issue: cooling a chip and using it as a heat source for an engine (of any type) are mutually exclusive goals. If you have an effective cooling system, the chip won't get hot enough to provide enough energy to run it, as preventing the chip from getting hot is the whole point of having a cooling system in the first place. If it does get hot enough to produce a practical amount of energy, your cooling system is obviously not doing its job. It's most unlikely that any sort of heat engine will allow for more effective cooling than a simple heat sink; that's simply not what heat engines are designed to do. Like I said: possible, but almost certainly not practical.

I'm sorry Foxpup this sounds like flaming but you still don't get it: You can get useful power from waste heat from a processor without raising it's temperature:

1.; CPU/GPU 70 degrees / standard aluminium heat sink 40 degrees. air 20 degrees
2.; CPU/GPU 70 degrees / TEC hotside 70 cold side 40. makes power but have to have a larger aluminium heat sink than a processor without a TEC, as it provides thermal insulation.

Summary: If you remove 100w of heat power from a CPU to keep it at 70 degrees and there is more thermal resistance because of the power producing TEC, you just have to make the cold side either better, bigger or colder to remove the same amount of energy to maintain 70 degrees at the CPU. around 15 times as large a heat sink in this case.

Phone charging camp stove: http://www.biolitestove.com/campstove/camp-overview/features/

Tea light (ca 40W) drives a LED. As this is a standard Coke cooler TEC the temperature on the hot side is less than 70 degrees, or it would be destroyed:

[Foxpup]

I'm sorry Foxpup this sounds like flaming but you still don't get it: You can get useful power from waste heat from a processor without raising it's temperature:

1.; CPU/GPU 70 degrees / standard aluminium heat sink 40 degrees. air 20 degrees
2.; CPU/GPU 70 degrees / TEC hotside 70 cold side 40. makes power but have to have a larger aluminium heat sink than a processor without a TEC, as it provides thermal insulation.

Summary: If you remove 100w of heat power from a CPU to keep it at 70 degrees and there is more thermal resistance because of the power producing TEC, you just have to make the cold side either better, bigger or colder to remove the same amount of energy to maintain 70 degrees at the CPU. around 15 times as large a heat sink in this case.

I'm sorry, but I really don't get it. I'm agreeing with you. I never said it couldn't be done. I said it couldn't be done without a vastly improved cooling system, which isn't practical in most cases. Unless you think using a 15 times larger heat sink to get a measly 0.7 watts is practical (I don't).

[DannyHamilton]

. . . you can pick one up on ebay for something like 50c that will run for about half an hour on the heat from a cup of coffee and put out enough power to put about a days worth of charge into a mobile phone . . .

*Dubious

[DannyHamilton]

. . . the torque output and duration is sufficient . . .

Source?

[MrTeal]

Not right Foxpup, the power transmitted through the thermoelectric element can be enough to cool the gpu. 100W of heat conducted through a TEC will give you 0.7 watts of electric power: GPU 65 degrees centigrade. ambient 25 degrees. Enough to run a fan on the heat sink. Calculator here: http://www.powerchips.gi/technology/pcalc.shtml

To use 50$ on a TEC to save 0.7 watts of power is not worth your while still.

Good luck finding a fan/heatsink combo that will keep the cold side at 25C on 0.7W.

[DannyHamilton]

You could have picked a better time, bit hazy in here and google seems reluctant to give simple answers lately . . .

I didn't pick the time of your reply, you did. 

My charger says 0.55 amps max output at 5v so that's 2.75 watts so would need to run for just over 40 minuets to consume 6615 joules.

That's assuming a perfectly insulated coffee cup so the time would be shorter, maybe half an hour for a travel mug and 5 min for a glass mug, chargers are generally into the 90% efficiencies so not much lost there.

. . .you can pick one up on ebay for something like 50c that will run for about half an hour on the heat from a cup of coffee and put out enough power to put about a days worth of charge into a mobile phone . . .

I haven't double checked your math or assumptions yet, but...

Would the 5 minutes of charge that you are going to get at 2.75 watts will be enough "to put about a days worth of charge into a mobile phone"?  Or the half hour?  Or the 40 minutes?  My phone usually takes longer than that to charge (and I rarely get a full day's use from a single charge).

Furthermore, your calculations assume that you use the full amount of energy from the mug (minus losses), 6615 joules, within the given timeframe.  It seems to me that the heat won't move through the sterling engine that fast.  As the heat moves through the engine, the thermal differential will drop further reducing the power output.

[DannyHamilton]

. . . you'd just have to make your coffee hotter

Agreed.

[hardcore-fs]

Not right Foxpup, the power transmitted through the thermoelectric element can be enough to cool the gpu. 100W of heat conducted through a TEC will give you 0.7 watts of electric power: GPU 65 degrees centigrade. ambient 25 degrees. Enough to run a fan on the heat sink. Calculator here: http://www.powerchips.gi/technology/pcalc.shtml

To use 50$ on a TEC to save 0.7 watts of power is not worth your while still.

Good luck finding a fan/heatsink combo that will keep the cold side at 25C on 0.7W.

It's a myth anyway... you cannot fan cool below ambient, and IF your ambient IS below 25 then you don't really need a fan, just a correctly designed heat-sink.

[firefop]

I'm sorry Foxpup this sounds like flaming but you still don't get it: You can get useful power from waste heat from a processor without raising it's temperature:

1.; CPU/GPU 70 degrees / standard aluminium heat sink 40 degrees. air 20 degrees
2.; CPU/GPU 70 degrees / TEC hotside 70 cold side 40. makes power but have to have a larger aluminium heat sink than a processor without a TEC, as it provides thermal insulation.

Summary: If you remove 100w of heat power from a CPU to keep it at 70 degrees and there is more thermal resistance because of the power producing TEC, you just have to make the cold side either better, bigger or colder to remove the same amount of energy to maintain 70 degrees at the CPU. around 15 times as large a heat sink in this case.

I'm sorry, but I really don't get it. I'm agreeing with you. I never said it couldn't be done. I said it couldn't be done without a vastly improved cooling system, which isn't practical in most cases. Unless you think using a 15 times larger heat sink to get a measly 0.7 watts is practical (I don't).

I think the best approach to this idea is using multiple heat exchanges. Standard, water block to a large water system, then couple that to another liquid loop that's much smaller - using that to power the stirling. It's the same idea used for some solar hot water heaters, massive loop on the roof, tiny one that's used in a hot water tank as a heating element.