Computer that boils in water

[batteksystem]

Anyone see this kind of things? Very strange that computer should short circuit in water, right?

https://youtu.be/JxzWo1I9v58

[Whales]

Short circuits are unlikely.  The VRMs on the motherboard would be boiling instead.

Processors can easily hit 100degc.  Modern ones try to throttle before then, bu many older ones (eg early Athlon 64s) didn't care:



Similar for the graphics cards.


EDIT: In other words, the computer in that video could be working perfectly fine.

[John Heath]

Just had a thought. If this was a bitcoin mining reg then the steam could be vented to a vacuum type steam engine. A vacuum type steam engine condenses the stream , 15 pounds per inch , to a vacuum. 15 pounds per inch is a reasonable amount of energy for a 1 foot diameter piston with a 10 foot stroke. A mining reg that uses it's own heat energy to offset the electric bill , hmmm.

[nctnico]

Anyone see this kind of things? Very strange that computer should short circuit in water, right?

Physics 101: Water doesn't conduct electricity.

[Echo88]

That isnt water, its Novec by 3M. A nonconductive fluid with a low boiling point used as a cooling solution.
For example: https://multimedia.3m.com/mws/media/121372O/3m-novec-7000-engineered-fluid-tds.pdf

[SiliconWizard]

Just a couple facts.

1. Pure water has low conductivity  (but it may still mess with some non-hermetic parts in other ways).
2. This is not water.
3. This is not boiling.
4. This is most likely mineral oil, and the bubbles come from oil circulation through a pump, much like a fish tank.

This kind of PC cooling is a thing for a very few enthusiasts:
https://www.pugetsystems.com/submerged.php

[ebastler]

That isnt water, its Novec by 3M. A nonconductive fluid with a low boiling point used as a cooling solution.
For example: https://multimedia.3m.com/mws/media/121372O/3m-novec-7000-engineered-fluid-tds.pdf

Right. Besides water's potential conductivity (@nctnico: how would you keep it from dissolving unwanted contaminants?), its boiling point would also be unsuitable for this cooling approach. You would not want 100°C environmental temperature for your chips, I assume. Novec comes in various formulations, with boiling points starting at 34°C.

[Gyro]

That looks pretty inefficient, the die dissipating area is way too small. Vapor cooling only works efficiently if there is a constant supply of liquid to the surface to vaporize, most of the surface area in the photos looks obscured by bubbles.

I think heat spreaders are needed.

[wraper]

Anyone see this kind of things? Very strange that computer should short circuit in water, right?

Physics 101: Water doesn't conduct electricity.

Except it does as long as it contains dissolved ions. And even if it does not contain ions (DI water) it will suck them in once you place it on metal surface.

2:20 for start of experiment

https://youtu.be/1DEwjJe3d1o

[RobK_NL]

Nothing new. Cray Research already did this in the mid 80's:
https://en.wikipedia.org/wiki/Cray-2

https://youtu.be/PTVfnd4KsLU?t=60

[james_s]

Water cooling is/was not uncommon with arc lamp pumped industrial YAG lasers. The arc lamps including their electrical connections are submerged in the cooling chamber through which deionized water is pumped. They require special filtering to keep the water in the cooling loop deionized which keeps it from conducting.

[bloguetronica]

And I though that the computer was vaporized by the somewhat hot temperature of the water. I say that judging by the title.

Anyway, to answer your question, distilled water conducts electricity pretty much badly. But you still have ions to conduct, because the water is never 100% pure (especially after being contacted by the motherboard itself, that carries tiny amounts of various salts and minerals), and the water itself carries very small ammounts of H3O+ and OH- ions (both the pH and the pOH of pure water are 7, which means that pure water has ions).

That isnt water, its Novec by 3M. A nonconductive fluid with a low boiling point used as a cooling solution.
For example: https://multimedia.3m.com/mws/media/121372O/3m-novec-7000-engineered-fluid-tds.pdf

Still not a good idea. I imagine the quantity of noxious vapors filling the air.

Kind regards, Samuel Lourenço

[SiliconWizard]

Still not a good idea. I imagine the quantity of noxious vapors filling the air.

Yes. As I said above, the only current way that is "routinely" done and that works in the long run is using mineral oil, pumps, and an efficient cooling system. Mineral oil is safe, relatively inexpensive, has ultra low conductivity and absorbs little moisture.

"Liquid"/oil cooling is good but won't do much unless you can dissipate heat inside it. So you still have to add an heat exchanger (radiator), a pump to circulate the oil and fans.

Getting back to water, not only wouldn't it work but the fact it would "boil" is absurd. No CPU will get hot enough before throttling and eventually shutting down, unless it's actually heating because of short circuits.

Back to mineral oil again, CPUs need a passive heatsink on top of them. A CPU without any heatsink will not transfer enough heat to the oil to keep its temperature to a reasonable level.

[amyk]

Fluorocarbon coolant is inert and rather nontoxic. Observe in the datasheet linked above: "LD50 > 2000mg/kg".

[JohnPen]

An even earlier demo of water cooling transistors was back in the 60s.  This was a Mullard demonstration of a number, cannot remember how many,  of OC16 germanium power transistors  furiously boiling the water. 

[Mr. Scram]

I've seen a demo of servers being run inside this kind of fluid. The fluid they used is supposedly non-toxic or harmful, as people would need to work with and around it with some ease. It is frighteningly expensive though, and it reportedly has a limited lifespan of about 5 years.

[ogden]

That isnt water, its Novec by 3M. A nonconductive fluid with a low boiling point used as a cooling solution.
For example: https://multimedia.3m.com/mws/media/121372O/3m-novec-7000-engineered-fluid-tds.pdf

Still not a good idea. I imagine the quantity of noxious vapors filling the air.

Unless you want to refill tank often, you want heat exchanger in the system. More detailed video and comments as well:

[ebastler]

Unless you want to refill tank often, you want heat exchanger in the system. More detailed video and comments as well:

In principle, that's what happens in a heat-pipe cooling system (which must be what the low-boiling-point version of Novec is designed for): Have the fluid evaporate at the CPU or other heat source; keep it contained in a closed system; have it condense at the other end of the system, where you can attach a radiator and fan.