The ultimate quiet computer?

[JeremiahTheGreat]

Was just wondering, what would the technical barriers be to have a computer (mobo, HD, CPU, video etc) completely submerged into liquid, a sort of complete water cooling?

I'm sure its not as simple as getting a large fish tank, filling it up with tap water and chucking in the computer case.. but if one of you want to try it, i would like to know the result

 

[Chooco]

interesting concept, but the problem remains of how to keep the heat sync exposed to water but not the circuitry.

perhaps if the case was welded to the heat sync with the heat sync protruding from the outside. how would the case sustain the pressure and remain easy to open to upgrade stuff?

 

[zhena]

water does not transfer electricity. the salt in the water transfers electricty. regular tap water has a lot of salt in it thats dissolved.
if you fill a case with distilled water and dump your working mobo in it.. nothing will happen.

and it's already been done a bunch of time.. i think except with an oil solution instead of water

 

[ttn1]

Actually even distilled water will conduct a little bit. You would need very pure water, to get small enough conduction to allow a working motherboard to be submerged in it.

For the ultimate quiet computer, why not just run without any fans and with a solid state harddrive. If you design your passive cooling solution well enough you would be set. I actually have a small, slow computer that is like this, and it is really nice.

The biggest problem I have seen with submerged solutions is that the board material for the cards and MB begin to absorb the liquid. When they delaminate everything starts to have problems. This seems to happen even if the people seal everything very well. That being said, I believe, CRAYs are submerged in liquid nitrogen. Of course, Crays, are not anywhere near silent.

Another problem is removing the heat from the water. If you submerge a computer in liquid that is in a sealed container, you will just heat the liquid up. You have to have a way to remove the heat, which is where the noise usually comes in.

 

[J.Zorg]

You could use some kind of dielectric oil. It's used for cpacitors and tranformators i think. But you will run in quite a few problems i think... You will still have to cool the liquid. And HDs don't like liquids. .

 

[Belegost]

While I don't have any links right now, it has been done.

Just thinking about it, here's how I'd do it:

Take all the boards, mainboard, video card, etc. Mask off connectors and spray them with a conformal coating. Since we'll be using an insulating liquid we don't want to accidently create capacitance between components.

On the top of the radial caps put a dab of silicone RTV or similiar sealant, this way the dielectric won't seep inside the cap.

The hardest part will be making the connectors and backplane watertight so that your cabling doesn't have to go through the liquid.

Put the cards in, and seal the outside of the PCI/AGP slots where the card slides in, probably with some sort of insulative grease, that way you can remove the cards later, but the fluid won't affect the connection. I'd suggest sealing the floppy and IDE cables in with silicone, and sealing the power connector well. And cutting little pieces of hard foam to cover the jumper blocks.


Fill the CPU socket with dielectric grease, if it uses ceramic packaging, probably coat the outside with conformal coating to keep it from absorbing the fluid.

To cool the liquid I'd send it through a radiator, I'd pump it into the system right over the video card and CPU, and pump it out from the side of the RAM furthest from the CPU.

I'd probably leave the power supply outside, but I suppose you could put it in the fluid as well, just make sure the fluid has a dielectric breakdown voltage much higher than any inside the PSU.

All the connections to the outside would have to be well-sealed, don't want to lose any oil to evaporation.

Hard drives and other drives should not be placed inside.

I think the expense would outweigh the benefits, though.

 

[hoihtah]

interesting. makes me wanna try.

but, if silence is what's desired...
ear plugs might be a better solution.

 

[saab9kt]

OK... a couple of things here.

- There is not such thing as "pure" water if you are talking about putting it in a fish tank and then submerging computer stuff... so don't even think of using water this way.

- There are a number of dielectic fluids that could be used. Some use mineral oil. Engineers use Fluorinert or other similar high performance cooling fluids available from 3M, Dupont, etc.

- Once you stick your computer stuff in a tank of liquid, you still have to get rid of the heat!!! The tank of liquid does not just absorb it! If you continue to add heat to the liquid, it will boil away. Of course, a tank of liquid has a large-ish "thermal capacitance"... so you could run your computer for a short while before the liquid started to boil.

- The ultimate "destination" for the heat is the ambient air. The tank of liquid in this example doesn't change that... it just adds another part to the path.

- If you want a quiet computer and cooling system, you have to eliminate the fans. That means the you're dependent on natural convection to transport the heat to ambient. Natural convection heatsinks are pretty big.

- Another method of getting a quiet computer would be to route the liquid to a heat exchanger that's in a different room than your ears!

 

[littlezipp]

I remember reading an article a few years ago about someone who submerged their computer into some sort of oil.
The main problem with that is, you can never use those components in another system again!

 

[jot027]

i remember seeing a guy who had submerged the various components of his PC in an old enamel sink, stuck in the coil off of the back of his fridge and covered the whole lot in a fine grade mineral oil, it was pretty silent if i remember. (if a little bulky)

 

[ttn1]

Well, here is the final solution that I am going to implement. There is a wonderful software called linux terminal server project. Basically, you have fast server machines that do all the work, and small, slow, diskless clients display the desktop. There are quite a few problems that crop up, but if you don't do much gaming, it is a very good solution. My first major use will be a silent computer for use in my entertainment center. The servers store all the movies and mp3s and the client just displays an interface and ports the video and sound. It is still in the development stage though.

 

[Shimmishim]

if you want pure pure water, i work in a lab that produces deionized, distilled water...

 

[Moohooya]

Several questions, comments and opinions.

Why worry about the HD? Since a spec of dust will ruin a HD, are they not pretty well airtight? Seems that spinning a disk at 15K per minute, 250 revs per second, will create a fairly large pressure difference between the inside of the platters and the outside. As the air at the outside will be near the seal, that had better be good not to let any air out, or else when the disk spins down the vacuum will suck in air and dust.

The mobo is all epoxy-resin is it not? What are boats these days built of, fibreglass and epoxy? Of course, when building a boat and a mobo you'd look for different properties, but I would assume the stuff to be pretty durable in water.

As said several times above, it is the salt that conducts. So use distilled water. But there will still be some small amount of salts correct? Sweat from hand on the mobo when putting it in the tank etc.

So what if there are salts. Get a 12v battery and stick leads in from both ends into the water until all the salts have reacted with one of the leads. (I forget which lead the will react with, I never was good at chemistry.) Now I'm not sure if something will slowly react and continue adding salts into the water, this could be an issue. If you really care that much, get a step up transformer and use 1KV (I think dc only, but perhaps AC is fine), and you could even leave it in their forever to slowly absorb any salts. Just make sure it is limited to only a few micro amps so you won't kill yourself if you stick your finger in the tank.

But, having said that, given average distilled water, (any one have numbers), we must have a chemist here who could calculate the current between two pins with a voltage differential of say 3.3v and a spacing of 1mm. Is this enough to mess up the flow of electrons for the computer, I'd guess not? So we have a 12V and -12V line going in. I'd guess they are much farther apart but we could check out the ATX power header and calculate that current also.

With a little smart thinking we can make the mobo cause a convection current in the tank. The warm water would rise up to the top of the tank where if would go through an upwards leading hose to a radiator that could be outside through a window. Then the cooler water return from the radiator would feed back into the bottom of the tank.

I forget the physics, but 1 calorie (approx 4.19 joules) will heat 1 gram of water 1C. 300 watts is 300 joules per second. So we'd need approx 4.3l per minute flowing through the radiator losing 1C, or 430ml per minute losing 10C. Either way seems like a fair bit and may well involve either 1 huge radiator, or a small pump, or a fan on the radiator. I think I'd go for a small 12v marine pump.

So who wants to try this

 

[ttn1]

The biggest problem with water is everything would react with it. You would have to completely seal the tank and basically jump through hoops, when there are lots of dielectric oils out there that would work much better.

Harddrives are not air tight. You would definitely destroy it by putting it in a fluid. The manufacturers use filters to assure that any air that gets into the drive has most particulates removed. I don't know a whole lot about the aerodynamic effects inside the drives, but I do know they are not airtight.

Most computer boards are multi layer. The glue holding the layer together is not very liquid resistant. The boards delaminate and become unusable unless sealed. Most people doing this seal them with spray on polyurethane or the like.

As for using electrolysis to remove impurities from water, it is possible, but my gut feeling says you'll fry lots of computer boards.

Passive liquid cooling systems usually have to be prohibitively large. When you add a fan or a pump to decrease the size, you add noise which defeats the purpose here.

I still think the best way to do this is sacrifice speed for lower power dissipation.

 

[Evadman]

3M makes a liquid that you can submurge electrical components in without problems. If I remember correctly it started turning into a slush around -50 degrees C. Either way it could not be used on a Hard drive as the friction and lift on the heads in the drive would render the drive useless.

I will try to find a link. It was called hydro[something]ether. I am looking.

<edit>
Got it. Called hydrofluoroether


Linkys
Link to 3M "HFE" properties and costs. $220 per gallon in 5 gal quantity
Article in Chemical Processing ( cooling properties )

Other cooling liquids for use with electronics: Florinert

OCTools Article on cooling system with florinert & Liq Nitrogen..
</edit>

 

[Moohooya]

Kind of pricy. My fish tank at home is 10 gallons, and the idea of putting a board in something less than half its size sounds tricky.

I guess you could put the board into a small tank of hydrofluoroether which in turn is in a much larger tank of water. This way you still use el-cheapo water mostly, but protect the board from the water. As long as the inner tank was not a thermal insulator, you should be set.

Bummer about the HD. Oh well. Guess you could always put them in a plastic bag!

 

[Evadman]

It is possable to get sealed Hard drives for use in high elevations. Those should be able to be imersed.

 

[CQuinn]

<< Why worry about the HD? Since a spec of dust will ruin a HD, are they not pretty well airtight? >>



http://www.pcguide.com/ref/hdd/op/packAir-c.html



<< Hard disks aren't sealed, because they have to be able to pass air between the inside of the drive and the outside, in order to equalize any air pressure differential that may exist between the two environments. This allows the disk to maintain proper equilibrium when the weather changes, or the drive is moved to a different altitude; if pressure is not balanced the drive might not perform properly and damage could even result. You can actually see the small breather holes in the cases of many drives, placed there for this purpose. Of course just putting a hole in the case would cause contamination of the drive, so the holes are covered with a breather filter which lets air pass through slowly but not dirt or dust. These filters are placed permanently and do not need to be serviced or replaced.

>>

 

[CQuinn]

If you are familiar with TechTV, they had some examples of what you guys are talking about on the show
"The ScreenSavers" back in April.

submerging your pc in a non conducting liquid

buildin a submergence case



Note also: even deionized, distilled water molecules will eventually bond with any excess mineral content,
(leads and traces) on a board, giving it enough conductivity to do bad things with your setup.

 

[ttn1]

CQuinn, thanks for the links. I was pretty sure the harddrive issue was ambient pressure changes, but I didn't want to offer anything I wasn't absolutely sure about.

 

[etech]

If you want to read some of the trials and tribulations of someone who has gone through the process of submerging his motherboard, Dr. Freeze was one of the first people I heard of that had done it.

"Dr. Ffreeze Project"

It's a bit dated but still has some good information for someone determined to "sink" his motherboard.