High Volume Water Cooling, is it an option?

[Odetas]

A while ago I'd posted this idea on Reddit and was directed by an astute member here that it is best discussed on this forum instead.

I've recently picked up a small collection of Antminer S1, S3 and S5 miners.  After going through and cleaning them out, I wondered why a good, efficient high volume liquid cooling solution couldn't be made for these miners and others to increase efficiency and cut down on the noise.  After servicing a few of these, I think that a very simple heat exchanger design could be made that could be an easy retrofit for any miner. When I say heat exchanger, I'm talking about a large heat exchanger, 24" x 48" coupled with an electric pump.  Something like this could easily dissipate 300 KW with a lower noise level and eliminate dust issues.  Concerns regarding water spillage onto boards can always be addressed utilizing different media, such as glycerin which is not electrically conductive. 

Why? There are quite a few of these "antiquated" miners that are going to be on the market, many of them overclockable.  But overclocking means more heat, and air/conduction/convection can only do so much.  If you can take an S9 in the future and overclock it significantly, you might be able to pull some extra life out of it.

Thoughts?

[1714749744]

1714749744

[1714749744]

1714749744

[1714749744]

1714749744

[1714749744]

1714749744

[1714749744]

1714749744

[1714749744]

1714749744

[Turan100]

A while ago I'd posted this idea on Reddit and was directed by an astute member here that it is best discussed on this forum instead.

I've recently picked up a small collection of Antminer S1, S3 and S5 miners.  After going through and cleaning them out, I wondered why a good, efficient high volume liquid cooling solution couldn't be made for these miners and others to increase efficiency and cut down on the noise.  After servicing a few of these, I think that a very simple heat exchanger design could be made that could be an easy retrofit for any miner. When I say heat exchanger, I'm talking about a large heat exchanger, 24" x 48" coupled with an electric pump.  Something like this could easily dissipate 300 KW with a lower noise level and eliminate dust issues.  Concerns regarding water spillage onto boards can always be addressed utilizing different media, such as glycerin which is not electrically conductive. 

Why? There are quite a few of these "antiquated" miners that are going to be on the market, many of them overclockable.  But overclocking means more heat, and air/conduction/convection can only do so much.  If you can take an S9 in the future and overclock it significantly, you might be able to pull some extra life out of it.

Thoughts?

i think this is a great idea using cooling water because i always though of using electricity but i did not know how this idea didn't crossed my mind using other energies so if u can make it work just do it way better than electricity and such Good luck bro.

[Odetas]

Water and other fluids have a high latent heat of evaporation, that means that they can absorb alot of energy, think KW.  Even if the pump fails, the parts will still cool, better than just air.  Once the fluid is made hot, you just have to pass through a heat exchanger.  What you want to be careful of, is if it is too cool and you get condensation, that'll kill electronics.

[leowonderful]

Yes, but taking antiquated miners and OCing them = even worse efficiency = no profit or ROI at the end of the day. S9s also have horrible reliability and do not overclock too well at any temperature AFAIK.

[Odetas]

Yeah, I'm mostly using the S9 as an example.  But I think you get the general gist of it.  I wonder if the S9's are having issues because they're really reaching the thermal limits of air cooling.  They probably had issues with the other high wattage stuff also.  Going down in lithography helps a ton with efficiency, but at some point you kind of hit the limits of convection/conduction through air.

[leowonderful]

Yeah, I'm mostly using the S9 as an example.  But I think you get the general gist of it.  I wonder if the S9's are having issues because they're really reaching the thermal limits of air cooling.  They probably had issues with the other high wattage stuff also.  Going down in lithography helps a ton with efficiency, but at some point you kind of hit the limits of convection/conduction through air.

The build quality of Bitmain products has also gone down over time, with the S9 being the most power dense miner they've made, so it could be an issue but most S9s I've seen do run a little bit on the hot side. Could be why, but could also be bad manufacturing from Bitmain.

[breadginger56]

A while ago I'd posted this idea on Reddit and was directed by an astute member here that it is best discussed on this forum instead.

I've recently picked up a small collection of Antminer S1, S3 and S5 miners.  After going through and cleaning them out, I wondered why a good, efficient high volume liquid cooling solution couldn't be made for these miners and others to increase efficiency and cut down on the noise.  After servicing a few of these, I think that a very simple heat exchanger design could be made that could be an easy retrofit for any miner. When I say heat exchanger, I'm talking about a large heat exchanger, 24" x 48" coupled with an electric pump.  Something like this could easily dissipate 300 KW with a lower noise level and eliminate dust issues.  Concerns regarding water spillage onto boards can always be addressed utilizing different media, such as glycerin which is not electrically conductive. 

Why? There are quite a few of these "antiquated" miners that are going to be on the market, many of them overclockable.  But overclocking means more heat, and air/conduction/convection can only do so much.  If you can take an S9 in the future and overclock it significantly, you might be able to pull some extra life out of it.

Thoughts?

I think Air Cooling is way better than Water Cooling because everytime we think about a slight hardware error and an accident happens, and damage the tubes of a water cooler then it may cause leak and damage everything in your pc which is air cooling is better

[NotFuzzyWarm]

Ya know, before newb's start rehashing old ideas again in another thread for the umpteenth time you might want to first read Why so few immersion cooled custom builds/Mods? to save a lot of brain power...

Pretty much covers most of the in's, out's, and gotcha's. Also a better place to continue the discussion vs making another thread on the same subject.

[ijl53]

Its an option just not a great option. Get a better a/c and fans. Pay for the electricity or pay for the headaches and failed hardware

[fanatic26]

I dont understand why so many people keep bringing up liquid/immersion cooling. Think on it for a good 30 seconds and you should be able to come up with half a dozen reasons why it is such a monumentally stupid idea.

[rockminer1]

Hello.

In Russia there is company. The company produces the watercooling boxes 5-50KWt for cooling any sort of chips (NVIDIA, Asic and other)
Chips need to be placed inside the box into dielectric water (Novec).
I want introduce it on bitcointalk.







P.S. I'm not native speaker. Sorry for mistakes.

[NotFuzzyWarm]

Hello.

In Russia there is company. The company produces the watercooling boxes 5-50KWt for cooling any sort of chips (NVIDIA, Asic and other)
Chips need to be placed inside the box into dielectric water (Novec).
I want introduce it on bitcointalk.







P.S. I'm not native speaker. Sorry for mistakes.

Interesting.
Please re-post this information here Why so few immersion cooled custom builds/Mods? so all information regarding liquid cooling stays in just 1 major thread.

[Odetas]

Ya know, before newb's start rehashing old ideas again in another thread for the umpteenth time you might want to first read Why so few immersion cooled custom builds/Mods? to save a lot of brain power...

Pretty much covers most of the in's, out's, and gotcha's. Also a better place to continue the discussion vs making another thread on the same subject.

Thanks for the link to the other thread, I read through that and it was informative.  Mostly about immersion cooling though.  I was actually thinking of making a heat exchanger that you could direct mount the hashing boards onto, and replace air cooling.  If it already exists, cool.  The thing with this approach is that it can be pretty cheap and potentially open for DIY using off-the-shelf parts.  Immersion cooling and the like all look to be quite expensive.

[NotFuzzyWarm]

Ya know, before newb's start rehashing old ideas again in another thread for the umpteenth time you might want to first read Why so few immersion cooled custom builds/Mods? to save a lot of brain power...

Pretty much covers most of the in's, out's, and gotcha's. Also a better place to continue the discussion vs making another thread on the same subject.

Thanks for the link to the other thread, I read through that and it was informative.  Mostly about immersion cooling though.  I was actually thinking of making a heat exchanger that you could direct mount the hashing boards onto, and replace air cooling.  If it already exists, cool.  The thing with this approach is that it can be pretty cheap and potentially open for DIY using off-the-shelf parts.  Immersion cooling and the like all look to be quite expensive.

In most ways, immersion is cheaper and a lot easier as the same pumps/fluid/air exchanger setup is used but:
 No time spend tearing apart miners and installing cold plates.
 No plumbing manifolds with hoses using dripless tubing connectors feeding each miner/hash board
 Just drop in the miner and trick the fan speed monitor into being happy

Using cooling blocks are a different story. Yes they can be used on S1 through the s5's. There are already folks here that made them long ago. Do a search for 'Antminer cooling block' and it should turn up folks that might still sell them. On the commercial product side, search Wakefield Cold Plate for their offerings and a ton of information regarding sizing everything.

As for S5+, S7 and S9/T9, forget it. The chips are in strings and there is a serious likelihood of causing a short-circuit. Then there is the issue of removing the original heatsinks used in them - very difficult to do without causing damage.

As for the heat exchanger you mentioned... Physical size tells next to nothing. You need to find out what its thermal specifications are! Being very familiar with various cooling systems including industrial chiller systems and outdoor cooling towers I can say that 24"x48" is NOT going to be able to dissipate 300kw... depending on fluid flow rate, airflow and delta-T between the air and fluid - 10kw maybe.

Lastly, the S9 and T9 CANNOT be overclocked. They are already pushed to their limits (which is 1 reason the s9 has such bad reliability) and their freq/core voltage cannot be changed.

[rockminer1]

Interesting.
Please re-post this information here Why so few immersion cooled custom builds/Mods? so all information regarding liquid cooling stays in just 1 major thread.

Ok I will. It need to be translated but I will try and do it soon. Thank you for helping

[rockminer1]

Interesting.
Please re-post this information here Why so few immersion cooled custom builds/Mods? so all information regarding liquid cooling stays in just 1 major thread.
[/quote]

I did big post. https://bitcointalk.org/index.php?topic=1867203.msg20263037#msg20263037

[flameruk]

To give you an idea of how big a unit you would need for heat disipitation.

See the pictures here :-

http://www.pwmaintenance.co.uk/air-blast-cooler-service/


Scroll down the page and take a look at the photos.
This is a pair of 400kw air blast coolers we installed a while back.
Each one will dump 400kw of heat from a water cooler compressor.
They are sized to give a 15 Deg temp differential.

Big cooling needs big coolers.

[Odetas]

As for S5+, S7 and S9/T9, forget it. The chips are in strings and there is a serious likelihood of causing a short-circuit. Then there is the issue of removing the original heatsinks used in them - very difficult to do without causing damage.

As for the heat exchanger you mentioned... Physical size tells next to nothing. You need to find out what its thermal specifications are! Being very familiar with various cooling systems including industrial chiller systems and outdoor cooling towers I can say that 24"x48" is NOT going to be able to dissipate 300kw... depending on fluid flow rate, airflow and delta-T between the air and fluid - 10kw maybe.

Damn those individual heatsinks.

I think I may have underestimated those heat exchangers.  I was going off of experience where I used to run a 100 gallon tank to cool an absorber (automotive) that took in 600kw, without a cooler and it took the water a while to heat up.  However, that wasn't 600kwh.