Why not design an ASIC that can be used for AC/Cooling?

[tuaris]

It's not impossible, is it? Don't AC's heat up a gas that then goes through condensation?  My understanding is that the phase change of the gas is used for heat transfer.  The purpose of the compressor in current AC's is just to heat up the gas without fire/heat.  It then condenses into a liquid and evaporate it back into a gas. Or am I wrong in my theory?  I must be missing something because it would have already been done.

[NotFuzzyWarm]

Yes it could in theory be done. There are A/C systems that run off of propane/natural gas instead of using an electric compressor https://www.gassouth.com/blog/natural-gas-air-conditioners and just as they burn gas to power their cooling cycle, ASIC's could in theory do the same thing.

Main issue I see is that high power electronics do NOT like constant cycling on & off as it stresses the solder connections.

FYI: the compressor does not heat up the gas and then condense it into a liquid. It compresses the gas which under pressure of 150-250psi (depends on the refrigerant) becomes a liquid. The liquid phase is hot because the heat energy stored in the (cool) gas becomes concentrated as the large volume of gas is turned into a much smaller volume of liquid.

[MaxMueller]

Yes it could be done. There are A/C systems that run off of propane/natural gas instead of using an electric compressor https://www.gassouth.com/blog/natural-gas-air-conditioners and just as they burn gas to power their cooling cycle, ASIC's could in theory do the same thing.

Thats apple and pears. You cant compare waste heat from an asic to waste heat at a few hundres degrees.

My answer is no. It seems to be impossible to me as the temperature gradient is to small. Stirling engines failed for decades to gain mechanical power from low temperature waste heat. Imagine a stirling engine powered by waste heat which drives the compressor of a heat pump. Thermal compression is nearly the same, it works if you provide high temperature waste heat, but not with asics.

What you could do is to use an electrical powered heat pump to pump the asic waste heat to highter temperature levels e.g. for hot water, with the benefit of having a two phase immersion cooling for your asic chips.

[dimitrylsm]

Good thought. I’m experimenting in a similar line — routing ASIC coolant through membrane distillation so that the system not only cools but also produces distilled water. With right engineering, an ASIC unit could serve cooling + water generation.

[MoonReaver]

It's not impossible, is it? Don't AC's heat up a gas that then goes through condensation?  My understanding is that the phase change of the gas is used for heat transfer.  The purpose of the compressor in current AC's is just to heat up the gas without fire/heat.  It then condenses into a liquid and evaporate it back into a gas. Or am I wrong in my theory?  I must be missing something because it would have already been done.

Imagine having 10 Asics that are mining at 10 TH/sec.
You’d also need to balance out the room temp. to fit the season.

4 out of 10 heating Asics if its summer, 6 if it’s winter (maybe a troll statement but I am trying not to — getting better at it though) 😁

Yes it could in theory be done. There are A/C systems that run off of propane/natural gas instead of using an electric compressor https://www.gassouth.com/blog/natural-gas-air-conditioners and just as they burn gas to power their cooling cycle, ASIC's could in theory do the same thing.

Main issue I see is that high power electronics do NOT like constant cycling on & off as it stresses the solder connections.

FYI: the compressor does not heat up the gas and then condense it into a liquid. It compresses the gas which under pressure of 150-250psi (depends on the refrigerant) becomes a liquid. The liquid phase is hot because the heat energy stored in the (cool) gas becomes concentrated as the large volume of gas is turned into a much smaller volume of liquid.

I tried reading that but my chemistry and physics was very bad in highscool, I literally had a headache trying to understand its complexity.

Did you get an ace in those subjects?

[MaxMueller]

Please don't ride the dead horse again.
Simple answer: its not possible.

Even a perfect heat engine would not surpass carnot efficiency of n= 1 - (T_l/T_h).
Gas Air Conditioners work with high temp. Lets assume burning gas at 600°C you get about n= 1-(300K/900K) =66% as max efficiency. This is the part of energy you can use for clausius rankine cycle to create cold... lets assume 30% efficie ce there, you will end up with a total efficiency of 10-20%.

With a Miners heat source at ~50°C you get about n=1-(300K/330K) =10% for the first step, which need to be multiplied with the efficiency of cooling cycle.
With all additional losses (=dissipation heat) you will not be able to create any cold.  

[philipma1957]

well if you boil water with asics and

1)put it under pressure (large pressure cooker) you have steam under pressure
2)which could run a turbine which could make some power
3)which could run an ac


but getting the water hot enough means you need chips that can take 125c not 80c

so good luck with that.

[Nexus9090]

well if you boil water with asics and

1)put it under pressure (large pressure cooker) you have steam under pressure
2)which could run a turbine which could make some power
3)which could run an ac


but getting the water hot enough means you need chips that can take 125c not 80c

so good luck with that.

Water's pretty much out of the question for a thermal transfer media due to its boiling point.

Something with a much lower boiling point could be used to carry the heat and this is typically the type of fluid used in submersion cooling PFPE or Perfluoropolyether, these fluids are available with a wide range of low boiling points, also they're not conductive and have a very high dielectric constant.

However it would only be useful in as much as a thermal transport media, in other words for moving the heat from the ASIC to somewhere else. By itself, it would not be practical for generating power or providing cooling other than at the point of use.

I do not know if PFPE boiling point alters under pressure like water does, but potentially the vapor generated during the boiling process could perhaps drive a small turbine with the exhaust gasses then being condensed and returned to the system.

[philipma1957]

well if you boil water with asics and

1)put it under pressure (large pressure cooker) you have steam under pressure
2)which could run a turbine which could make some power
3)which could run an ac


but getting the water hot enough means you need chips that can take 125c not 80c

so good luck with that.

Water's pretty much out of the question for a thermal transfer media due to its boiling point.

Something with a much lower boiling point could be used to carry the heat and this is typically the type of fluid used in submersion cooling PFPE or Perfluoropolyether, these fluids are available with a wide range of low boiling points, also they're not conductive and have a very high dielectric constant.

However it would only be useful in as much as a thermal transport media, in other words for moving the heat from the ASIC to somewhere else. By itself, it would not be practical for generating power or providing cooling other than at the point of use.

I do not know if PFPE boiling point alters under pressure like water does, but potentially the vapor generated during the boiling process could perhaps drive a small turbine with the exhaust gasses then being condensed and returned to the system.

if you designed one and got it to work I know you would not get 100% efficiency as that would be like a perpetual motion concept.

But I doubt you could salvage 10 to 15 percent via some steam style turbine.

Basically 10 submerged 3k miners heating something into a gas driven turbine to give a 3k output.

which would run an efficient ac .

And as I said I do not know what you could use to get hot enough to make steam/gas with enough pressure to run the turbine.

It would cost big bucks you need a vat for the immersion you need pumps to flow the coolant and I still think the coolant won't run a turbine well enough to recoup the cost to build it.

[colinistheman]

I find these illogical ideas quite amusing. Does an ASIC have a compressor inside? Even when they get very hot, they dont work like a compressor or a heat pump. To use it like an air conditioner you would need to add refrigerant gas and a powerful compressor inside, otherwise the ASIC chip will never be able to push the heat from the inside to the outside. Doing all these bizarre things would only increase the cost

The best thing an ASIC can do is keep your room warm in the winter, lol. Anyway, the point is, to keep the miner cool you can use liquid or hydro cooling which is much safer and more practical

[FP91G]

Any equipment must meet customer demand; otherwise, manufacturing it is unprofitable. The bulk of Bitcoin mining is industrial, and all company resources are devoted to meeting this demand. Household mining devices have a number of noise and power consumption restrictions, meaning they are less efficient.

[MaxMueller]

if you designed one and got it to work I know you would not get 100% efficiency as that would be like a perpetual motion concept.

But I doubt you could salvage 10 to 15 percent via some steam style turbine.

Basically 10 submerged 3k miners heating something into a gas driven turbine to give a 3k output.

which would run an efficient ac .

And as I said I do not know what you could use to get hot enough to make steam/gas with enough pressure to run the turbine.

It would cost big bucks you need a vat for the immersion you need pumps to flow the coolant and I still think the coolant won't run a turbine well enough to recoup the cost to build it.

Lol. Just cheat physics / thermodynamics and try to ride all the dead horses. 
Steam engine will not work either, because the temparature gradient is still too low, and clausius rankine cycle is less efficient than carnot.

In summer when its hot outside (the time when you need AC) a lot of miners already struggle to get waste heat away. if another temparature gradient is added for whatever purpose, that increases 1:1 the asic chip temp which is at least for bitmain antminers a very bad idea, as it decreases hashing efficency.

[Nexus9090]

if you designed one and got it to work I know you would not get 100% efficiency as that would be like a perpetual motion concept.

But I doubt you could salvage 10 to 15 percent via some steam style turbine.

Basically 10 submerged 3k miners heating something into a gas driven turbine to give a 3k output.

which would run an efficient ac .

And as I said I do not know what you could use to get hot enough to make steam/gas with enough pressure to run the turbine.

It would cost big bucks you need a vat for the immersion you need pumps to flow the coolant and I still think the coolant won't run a turbine well enough to recoup the cost to build it.

Lol. Just cheat physics / thermodynamics and try to ride all the dead horses. 
Steam engine will not work either, because the temparature gradient is still too low, and clausius rankine cycle is less efficient than carnot.

In summer when its hot outside (the time when you need AC) a lot of miners already struggle to get waste heat away. if another temparature gradient is added for whatever purpose, that increases 1:1 the asic chip temp which is at least for bitmain antminers a very bad idea, as it decreases hashing efficency.

You could perhaps recycle some of the heat using a peltier device to generate some voltage/current that could be recycled into powering the system boards. I doubt it would be particularly efficient maybe 2-5% at best so 60-150Watts for a 3KW system.

But as you correctly point out thermodynamics are in play. you still need a cold junction in order for it to work.

[MaxMueller]

You could perhaps recycle some of the heat using a peltier device to generate some voltage/current that could be recycled into powering the system boards. I doubt it would be particularly efficient maybe 2-5% at best so 60-150Watts for a 3KW system.

Lets assume a setup with thermoelectric generator between asic chip and heatsink. Lets assume working temparature diffence of 20-30 degrees to archive those values (i think its very optimistic but at least not totally unrealistic).

On the other side the chip temp of asics is 20-30 degrees highter, which increases energy consumption by +15%. Finally it makes no sense.

Today and yesterday i did some tests.


Temp Range about 30-60°C, max_temp=62°C, stable hashing @12.45V @354 MHz (Preset profile 12.40V 345MHZ)


Temp Range about 18-37°C, max_temp=38°C, stable hashing @12.15V @354 MHz (Preset profile 12.40V 345MHZ)

[Nexus9090]

You could perhaps recycle some of the heat using a peltier device to generate some voltage/current that could be recycled into powering the system boards. I doubt it would be particularly efficient maybe 2-5% at best so 60-150Watts for a 3KW system.

Lets assume a setup with thermoelectric generator between asic chip and heatsink. Lets assume working temparature diffence of 20-30 degrees to archive those values (i think its very optimistic but at least not totally unrealistic).

On the other side the chip temp of asics is 20-30 degrees highter, which increases energy consumption by +15%. Finally it makes no sense.

Today and yesterday i did some tests.


Temp Range about 30-60°C, max_temp=62°C, stable hashing @12.45V @354 MHz (Preset profile 12.40V 345MHZ)


Temp Range about 18-37°C, max_temp=38°C, stable hashing @12.15V @354 MHz (Preset profile 12.40V 345MHZ)

No one said it would be easy.

[MaxMueller]

It is easy. Trust in physics and thermodynamics and do the opposite of those suggested AC/cooling concepts.

Increase cooling of asic chips and save 15% energy consumption / waste heat production due to highter efficiency...

[OgNasty]

Thinking about it...  You could probably use a geothermal cooling system and have the heat produced by the miner to act as an exhaust vacuum that brings the cold air into a house or whatever.  I would think the issue would be that it would be expensive, not very effective, and wouldn't be the most efficient use of energy to generate the cooling.