Extreme water cooling

[c64]

Is it OK to submerge a working PCB in distilled water?

[ejeffrey]

Not really.  Distilled water is corrosive, It will attack metals and other materials on your board and quickly become not distilled.  At that point it is conductive plus you have corroded your board.  Immersive cooling is generally done with non-polar liquids.

[Berni]

If the 220V power supply is left outside it would probably work for a while, but then the water would leech some of the metals and other crap into itself and become conductive again.

But even then you would get worse cooling performance compared to normal waterblock cooling solutions. Heatsinks designed for air don't work so well in water.

[ChristofferB]

I dont think it's so much heat sinks being worse while submerged as much as the fluid (air or liquid) being still, as opposed to a forced flow

[Berni]

Well for natural convection cooling id guess an heatsink designed for air also works well for water.

But when it comes to forced cooling the large long straight surfaces of an air cooled heatsink is far from ideal for water. The long fins present a lot of restriction to flow and the water loves to form a laminar flow along the surface so very little water comes into contact. The large heatsink also presents a lot of unnecessary thermal resistance.

For this reason heatsinks designed for water that are used in waterblocks are very different. They have lots of very tiny short fins to keep the useful surface area up despite laminar flow. The short fins also make for less thermal resistance from the base to the tip of the fin, the small size also makes it sensible to make it out of solid copper to reduce thermal resistance even more.

[OM222O]

1) that's not "extreme water cooling", you'd need chilled water for that

2) if you want submersion cooling, M3 makes a specific liquid that boils at around 50 or 60c, used for phase change submersion cooling

3) if you want something off the shelf, mineral oil is the usual go to for submersion cooling.

[RenThraysk]

Yeah, mineral oil was what Puget used to use to cool entire PCs in an aquarium tank.

Maybe bit of publicity stunt, but they looked good at least.

[langwadt]

1) that's not "extreme water cooling", you'd need chilled water for that

2) if you want submersion cooling, M3 makes a specific liquid that boils at around 50 or 60c, used for phase change submersion cooling

3) if you want something off the shelf, mineral oil is the usual go to for submersion cooling.

I think it is Dupont that makes the stuff they use to make up mosfet datasheets, it boils at 25'C

[SiliconWizard]

As said above, mineral oil would be more adapted to the task and less risky.
That said, it still depends on the complete BOM of your board. Some components may not like being immersed into oil and may stop working properly - Some sensors, or oscillators that wouldn't be fully hermetic, come to mind.

This has proven to be workable for whole computer systems though - I don't know for how long, but at least for a while.

Keep in mind it'll be very hard to remove if you ever want to get your board out and clean it.

As to efficiency, the obvious: it may be more efficient at transmitting heat than air, but you'll need to cool the oil. Otherwise it will end up reaching the temperature of an uncooled board. Cooling down mineral oil is a tad more difficult than cooling down water. This requires specific pumps to handle the task.

Usually not worth the trouble IMHO.

[Berni]

1) that's not "extreme water cooling", you'd need chilled water for that

2) if you want submersion cooling, M3 makes a specific liquid that boils at around 50 or 60c, used for phase change submersion cooling

3) if you want something off the shelf, mineral oil is the usual go to for submersion cooling.

I think it is Dupont that makes the stuff they use to make up mosfet datasheets, it boils at 25'C

At that point you probably need a sealed system to make sure it doesn't boil away.

But if its sealed off then you could just add a refrigeration compressor to pull a partial vacuum and lower its boiling point to say -30°C, run the output of the compressor back around trough a condenser to liquefy it again and you got a refrigeration cycle.

But yeah oil submersion cooled PCs are purely for show. There is no performance advantage to doing it while also being very inconvenient and messy to service. The true practical extreme PC cooling is running sub zero refrigerant trough a CPU block. It works, is practical to run as a daily driver and provides a significant performance improvement in overclocking (But is still very expensive and power hungry for how little gain you get so is very rarely done).

[schmitt trigger]

The Cray 2 Supercomputer utilized Fluorinert liquid cooling back in the mid 1980s

However, because of its harmful atmospheric effects, I ignore whether it is still available.

[TimFox]

The CDC 6600 had a pressure gauge on the front panel to monitor the Freon refrigeration system.

[c64]

I dont think it's so much heat sinks being worse while submerged as much as the fluid (air or liquid) being still, as opposed to a forced flow

Actually, I was thinking about no heat sinks at all. Just water

[Berni]

I dont think it's so much heat sinks being worse while submerged as much as the fluid (air or liquid) being still, as opposed to a forced flow

Actually, I was thinking about no heat sinks at all. Just water

It would work if you live on Mount Everest where the boiling point of water is about 70°C. But at sea level in normal air pressure nope.

EDIT: Unless you are cooling something like a reasonably power saving laptop CPU or one of those low clocked low power binned Intel desktop chips. Those might produce so little heat that it works. But then again you can cool these CPUs without even using a fan if the heatsink is fairly big.