Ingenious modder converts countertop ice machine into an RTX 3060 GPU cooler with a beer fridge thermostat — mod reduces temps by up to 62% in games, Cyberpunk 2077 runs at 22°C

I feel like this should result in the same issues that come with trying to use a refrigerator or freezer for this sort of thing - all of these devices are designed primarily to keep things cool, not to cool continuously, which is also why the cooling component works in conjunction with a heavily insulated box.

All compressors have a rated duty cycle, after which they need to shut down so they don't end up overheating and breaking down, and devices like this don't really require a particularly high one for their intended role. As such, I get the impression that this system will either fail at effectively maintaining stable temperatures, or simply die within a relatively short period of use.

This is why you frequently see Peltier coolers/TECs used for cooling scenarios like this, despite being significantly less efficient overall. Unlike how vapor compression cooling systems operate on cycles, a TEC will continuously transfer heat from one side to the other for as long as it is powered (in addition to the heat generated from the electrical process). The hot side naturally still requires a way to effectively disperse the generated heat in order to prevent it from self-destructing, but, since it is continuously generating heat across a predictable range, you can calculate the maximum potential heat output, and assemble a solution capable of dealing with it.

Of course, TECs have their own myriad issues on top of their inefficiency, especially if being used in direct contact with components inside of a computer. While I'm sure everyone here likely already knows this, the cold side of a TEC simply gets too cold, particularly when the component(s) to be cooled are in an idle state and barely putting out any heat. The resulting temperature differential with the ambient air causes condensation to form on the sub-ambient surfaces, and then, you have uncontained water inside your computer, one of the Top 10 places you do not want uncontained water to be.

It would be interesting to try this same basic setup, except replacing the compression cooler with a TEC (either one big one, or several smaller ones) to keep the water/coolant chilled. I suppose at that point you really just have an external, well-insulated TEC-cooled reservoir (which may or may not already be something readily available for purchase today - it's been around a decade since I built my last custom loop, and I haven't really kept up with developments since). And with the TEC(s) then needing their own beefy traditional water cooling loop (or one seriously powerful air cooler), it certainly wouldn't be as neat and tidy as just having a modified ice maker, in addition to using more electricity. But, assuming the math all works out, it should actually be able to work on a continuous basis without breaking down or failing to cool sufficiently.

Then again, you could also simply replace the compressor with a larger, more powerful one, and use it in conjunction with a larger tank, so you would have a coolant volume sufficient to take the heat input for longer before seeing any real temperature change. That should also be enough to address the duty cycle issue. Maybe something like a used commercial-grade Slurpee/Daiquiri machine? Those should have the required volume, compressor, fairly exact temperature controls, plus a solution for continuously agitating the coolant to avoid hot spots.

Anyway, I better stop this flow-of-consciousness brain-dump here; I'm sure plenty of people far smarter than I have already come up with far better solutions long ago. This just reminded me of the olden days, fresh out of college and working at Microcenter, when we would all hang out after work discussing our progress/latest painful failure of our latest unnecessarily complex builds, go running around hunting for parts and pieces at second-hand stores, scrap yards, public auctions, etc, and generally live our best geek life. Really makes me want to start on a new stupidly complicated build, but it's a lot less fun without a crew who are all into the same sort of thing, and I sadly have exactly zero people like that in my life these days...

purposelycryptic said:

I feel like this should result in the same issues that come with trying to use a refrigerator or freezer for this sort of thing - all of these devices are designed primarily to keep things cool, not to cool continuously,

Yep, 100% agreed.

purposelycryptic said:

I feel like this should result in the same issues that come with trying to use a refrigerator or freezer for this sort of thing - all of these devices are designed primarily to keep things cool, not to cool continuously, which is also why the cooling component works in conjunction with a heavily insulated box.

All compressors have a rated duty cycle, after which they need to shut down so they don't end up overheating and breaking down, and devices like this don't really require a particularly high one for their intended role. As such, I get the impression that this system will either fail at effectively maintaining stable temperatures, or simply die within a relatively short period of use.

This is why you frequently see Peltier coolers/TECs used for cooling scenarios like this, despite being significantly less efficient overall. Unlike how vapor compression cooling systems operate on cycles, a TEC will continuously transfer heat from one side to the other for as long as it is powered (in addition to the heat generated from the electrical process). The hot side naturally still requires a way to effectively disperse the generated heat in order to prevent it from self-destructing, but, since it is continuously generating heat across a predictable range, you can calculate the maximum potential heat output, and assemble a solution capable of dealing with it.

Of course, TECs have their own myriad issues on top of their inefficiency, especially if being used in direct contact with components inside of a computer. While I'm sure everyone here likely already knows this, the cold side of a TEC simply gets too cold, particularly when the component(s) to be cooled are in an idle state and barely putting out any heat. The resulting temperature differential with the ambient air causes condensation to form on the sub-ambient surfaces, and then, you have uncontained water inside your computer, one of the Top 10 places you do not want uncontained water to be.

It would be interesting to try this same basic setup, except replacing the compression cooler with a TEC (either one big one, or several smaller ones) to keep the water/coolant chilled. I suppose at that point you really just have an external, well-insulated TEC-cooled reservoir (which may or may not already be something readily available for purchase today - it's been around a decade since I built my last custom loop, and I haven't really kept up with developments since). And with the TEC(s) then needing their own beefy traditional water cooling loop (or one seriously powerful air cooler), it certainly wouldn't be as neat and tidy as just having a modified ice maker, in addition to using more electricity. But, assuming the math all works out, it should actually be able to work on a continuous basis without breaking down or failing to cool sufficiently.

Then again, you could also simply replace the compressor with a larger, more powerful one, and use it in conjunction with a larger tank, so you would have a coolant volume sufficient to take the heat input for longer before seeing any real temperature change. That should also be enough to address the duty cycle issue. Maybe something like a used commercial-grade Slurpee/Daiquiri machine? Those should have the required volume, compressor, fairly exact temperature controls, plus a solution for continuously agitating the coolant to avoid hot spots.

Anyway, I better stop this flow-of-consciousness brain-dump here; I'm sure plenty of people far smarter than I have already come up with far better solutions long ago. This just reminded me of the olden days, fresh out of college and working at Microcenter, when we would all hang out after work discussing our progress/latest painful failure of our latest unnecessarily complex builds, go running around hunting for parts and pieces at second-hand stores, scrap yards, public auctions, etc, and generally live our best geek life. Really makes me want to start on a new stupidly complicated build, but it's a lot less fun without a crew who are all into the same sort of thing, and I sadly have exactly zero people like that in my life these days...

The components in conventional coolers like fridges and ice makers are only designed to handle as much as they need to handle. So it may be possible to replace whole components or motors with more powerful ones or ones designed for continuous operation. Maybe a surplus industrial cooler would work better.

Instead of all the futzing with metal to metal heat transferring, the CPU makers should simply include the coolant pipe inside the CPU itself as part of fabrication.

Since they want to move to wafer stacking to increase density, this is the clear opportunity to design an internal cooling channel at inception.

Just a warning for those who place their CPU or GPU through constant temp cycling at such low temps will stress the supporting substrate over time and it will finally crack and the CPU will brick.