The ultimate water-cooled PC?

[lake_wrangler]

Just watched this, and it tickled my fancy...

[FreeFlier]

A few years ago someone built a directly liquid-cooled PC . . . by putting it in an aquarium that they filled with mineral oil.

They reported using standard components, and the only modifications they needed to make was to alter a few of the incidental adhesives to something that would not dissolve or soften in mineral oil.

They also needed to replace conventional spinning-disc hard drives with SSDs . . . the spinning-disc drives weren't sealed well enough. (I'm not sure if this was actually tested or just suspected.)

The only other issue they had was oil wicking up the cables.

Performance was reported to be outstanding . . . to the point they really couldn't test the cooling because the thing really didn't heat up.

--FreeFlier

[Dave]

That's pretty slick!

On first reading of the title, I was concerned that they were going to try to cool the components directly with water. In theory you could do this without shorting things out, if you used ultra-pure distilled water (very low ion content, and hence not very conductive). It wouldn't work for long. Ultrapure water is actually quite corrosive and would start dissolving the metal components, and it wouldn't be ultra-pure or nonconductive any more. Fzzt.

Using normal cooling (closed-loop liquid, and air) and sinking the heat out into an external body of water seems like a great solution.

I've seen liquid-cooled systems in the past, at trade shows. They all used Fluorinert or a similar freon-type coolant. Some immersed the boards completely. Others used a spray of liquid to cool the boards, and circulated the runoff through a closed-circuit cooling loop. It definitely works, but it's expensive and ecologically not-so-nice.

As to hard drives - most of them aren't hermetically sealed, and have a "breather" arrangement which allows the drive's internal pressure to equalize with that of the outside air. The breather has a very fine, dense filter which prevents any dust particles from getting in. Oil would almost certainly soak through and ruin the drive. Some of the newer high-performance drives are filled with helium, and these do have to be hermetically sealed and it isn't easy - helium loves to diffuse through a lot of materials.

[FreeFlier]

As I understand it, helium diffuses easily because it has the smallest molecule of any substance known to exist - While hydrogen has smaller atoms, hydrogen takes the form H2 -one molecule composed of two hydrogen atoms - and helium does not bond and thus remains monatomic - and smaller than an H2 molecule.

--FreeFlier

[Dave]

Yup, that's correct.

This has led to some interesting technical failures. A few years ago, a bunch of iPhones gave up the ghost at a medical facility... they just stopped working. It turned out that helium escaping from the superconductor cooling systems in the new MRI machine was hanging around the facility long enough for some of it to diffuse into the MEMS oscillators which act as the phones' frequency standards. The presence of the lighter gas changed the oscillation frequency enough that the phones crashed.

Although hydrogen molecules are larger than helium atoms, hydrogen diffusion can still be a big problem. It can diffuse into steel, eventually changing the crystal structure enough to make the steel become brittle. Bridges and buildings have failed, or come close to failure when their steel cables or beams have suffered from hydrogen embrittlement. The SF Bay Bridge has such a problem about ten years ago.

[lake_wrangler]

This! This is one big reason I like this forum!

I love to see the tangents and the bunny trails that result from a post, and all the various fields of knowledge that come together on any given subject by the various forum-goers.

[Alkarii]

There's a YouTube channel called Life of Boris, and in one of his videos, he made a vodka-cooled PC. Or maybe he just put a liquid cooling system in what he already had, but he put vodka in it instead of the water.

And yes, he's a Slavic dude. He's pretty funny.

[FreeFlier]

Yup, that's correct.

This has led to some interesting technical failures. A few years ago, a bunch of iPhones gave up the ghost at a medical facility... they just stopped working. It turned out that helium escaping from the superconductor cooling systems in the new MRI machine was hanging around the facility long enough for some of it to diffuse into the MEMS oscillators which act as the phones' frequency standards. The presence of the lighter gas changed the oscillation frequency enough that the phones crashed.

Although hydrogen molecules are larger than helium atoms, hydrogen diffusion can still be a big problem. It can diffuse into steel, eventually changing the crystal structure enough to make the steel become brittle. Bridges and buildings have failed, or come close to failure when their steel cables or beams have suffered from hydrogen embrittlement. The SF Bay Bridge has such a problem about ten years ago.

Anything electroplated can have significant issues due to hydrogen embrittlement. This is why guns are usually plated using electroless nickel.

--FreeFlier

[Warrl]

Anything electroplated can have significant issues due to hydrogen embrittlement. This is why guns are usually plated using electroless nickel.

--FreeFlier

And gun barrels have rather short active lifespans anyway. I did the calculations once (using various plausible but unreliable estimates) and it came out to about 30 seconds. That is, time during which an actual round being fired is traveling through the barrel.

[FreeFlier]

Anything electroplated can have significant issues due to hydrogen embrittlement. This is why guns are usually plated using electroless nickel.

And gun barrels have rather short active lifespans anyway. I did the calculations once (using various plausible but unreliable estimates) and it came out to about 30 seconds. That is, time during which an actual round being fired is traveling through the barrel.

The electroless nickel is on the outside . . . barrels are either are bare steel (sometimes stainless steel or other things) or hard-chromed using special techniques for erosion resistance . . .

My numbers suggest that for a .45 ACP that would work out to around 6000 rounds . . . which seems awfully low, though jacketed bullets would be harder on the barrel. For a rifle, OTOH, it would take fewer rounds, though it still seems to be a significantly low number . . . Of course,I don't know what your assumptions were.

FreeFlier

[Dave]

Speaking of helium leaks...

https://home.cern/news/news/accelerator ... photostory

[FreeFlier]

This! This is one big reason I like this forum!

I love to see the tangents and the bunny trails that result from a post, and all the various fields of knowledge that come together on any given subject by the various forum-goers.

You should try the Cross-Time Cafe . . . topic drift is a calling there.



--FreeFlier

[Typeminer]

Speaking of helium leaks...

https://home.cern/news/news/accelerator ... photostory

Went through this pretty fast, but I think the takeaway is that CERN doesn't keep duct tape handy.

And I've stayed out of the CrossTime Cafe on purpose, because I'm still technically employed (until January, anyhow), and I already waste enough time at the Doc's Machine Tinker's Guild.

[lake_wrangler]

This! This is one big reason I like this forum!

I love to see the tangents and the bunny trails that result from a post, and all the various fields of knowledge that come together on any given subject by the various forum-goers.

You should try the Cross-Time Cafe . . . topic drift is a calling there.



--FreeFlier

I'm familiar with it (as well as with the drift-meter)... but there's only so much time in the day... I haven't even read the comic-related threads on this forum, since the comic restarted!

You know how it is... I've got my country's 500th anniversary to plan, my wedding to arrange, my wife to murder and Guilder to frame for it; I'm swamped.