Ubuntu Underwater Edition

[gtrawoger]

Hi All,

I just wanted to share a little experiment myself and a friend had done. We all know about computers running submerged in mineral oil. But I thought that it would way cooler (literally) to run a computer submerged in water.

The only thing we needed to do was protect the metal contacts from shorting in the water.

Here is the youtube video we uploaded. http://www.youtube.com/watch?v=gZWdaOV731I As you can see it's running the Ubuntu Live CD and is running GLXGEARS while submerged in water.

Even thought the video mentioned that it ran 41 minutes at the time, the system ran stable for 4 hours underwater. Then the video card gave out somewhere and we lost video. We could still do things (since we heard the CD spin up) but couldn't see.

We replaced the video card (it was still working, but I actually think it was bad from the beginning) and ran the test for a few more hours.

Can you imagine the applications that this could be used for it we can get this to run long time???

[Svartalf]

This one ranks right up there with the other top "Kids, don't try this at home" stupid PC cooling tricks that people have done over the years.

I suppose someone had to try it at some point, just like the Liquid Nitrogen overclockings, and other similar gems. Now, if only I could afford the $2k or more for the Galinstan I want to do MY contribution to this category...

[gtrawoger]

This one ranks right up there with the other top "Kids, don't try this at home" stupid PC cooling tricks that people have done over the years.

I suppose someone had to try it at some point, just like the Liquid Nitrogen overclockings, and other similar gems. Now, if only I could afford the $2k or more for the Galinstan I want to do MY contribution to this category...

Aww... "stupid"? Come on. I admit we used old hardware for now, since if it did fry, we didn't want to shell out too much cash. But we are still experimenting. Just wait for the updates.

The current goal is to have the machine running stable for a week underwater, and have goldfish warm themselves on the chipset!

[Zhick]

A current machine should easily generate enough warmth to boil the water in no time (at wich point the hardware will probably have failed already), so you will have to add some water circulation and radiators to get rid of the heat. And then I really don't see the advantage this has over a regular watercooled rig.

So while this certainly is a cool (hoho) experiment I don't think it's of any practical use.

[gtrawoger]

A current machine should easily generate enough warmth to boil the water in no time (at wich point the hardware will probably have failed already), so you will have to add some water circulation and radiators to get rid of the heat. And then I really don't see the advantage this has over a regular watercooled rig.

So while this certainly is a cool (hoho) experiment I don't think it's of any practical use.

It ran overnight, and to my surprise the water was just lukewarm. I believe it actually is evaporating before it gets warm at all. What it needs currently is something to keep the water level up. In fact, the water level didn't move down much at all.

Pratical Use? Ummm... not sure about that. But something may transpire. Until then we're having fun!

[Svartalf]

Aww... "stupid"? Come on. I admit we used old hardware for now, since if it did fry, we didn't want to shell out too much cash. But we are still experimenting. Just wait for the updates.

The current goal is to have the machine running stable for a week underwater, and have goldfish warm themselves on the chipset!

It's "stupid" as in "Nobody in their right mind would EVER do this...but it's damned cool all the same".

Immersing something in water right after a protracted immersion in water dispersant compound isn't sane. You can't describe it any other way. And I'm waiting to see the end results- believe it. The truth be known, however, I'm not much better, really- liquid metal and vapor-phase cooling with WATER instead of freon isn't exactly "sane" either.

[Svartalf]

A current machine should easily generate enough warmth to boil the water in no time (at wich point the hardware will probably have failed already), so you will have to add some water circulation and radiators to get rid of the heat. And then I really don't see the advantage this has over a regular watercooled rig.

So while this certainly is a cool (hoho) experiment I don't think it's of any practical use.

Uh... Boil the water? Not likely...

Water requires approximately 965 BTUs to boil one pound of water to steam at 212 deg F. If you apply the typical conversions, this translates into something on the order of 285 watts in an hour's time- with your water being held at 212 deg F and NO thermal losses out of the water.

One pound of water translates into a little shy of 1/8th of a US gallon of water. In the video, it looks like there's 2, maybe 2.5 gallons of water. This means that you will need to inject 16 or more times the wattage over time to accomplish the same goal as you would with one pound of the liquid. This means you will need to inject something on the order of 4523 (More if there's more water...) watts of thermal dissipation per hour to accomplish a boil-off of the water in one hour's time, again provided you keep the water's temperature at 212 deg F and do NOT allow it to dissipate ANY of that heat so inputted into the system.

Now, that's not going to happen. There's radiative losses out the sides and bottom of the bowl. There's radiative and evaporative thermal losses from the air-water interface on the top. There's proably at least 300 or so BTUs/hr of loss from those sources- perhaps more. You have to get the water to the magic temp and ditch those losses above and beyond the stated 4523 BTUs/hr to boil that tub of water off like you intimate will happen. If you don't, the system reaches an equilibrium temperature short of the boiling point.

Moreover, this presumes you hit the magic temperature. This would require you inputing 40 BTUs/per pound/deg F temperature raised. If you don't prevent approximately 89600 BTUs from leaving the system within the previous hour, you won't even GET to boiling temperatures. Now, this translates into 1494 BTUs/minute that you HAVE to keep from leaving the system to get it up to even boiling temps.

Right now, I don't know of any machine that cranks out 4.6-5kW of thermals right at the moment. You'd need that to boil the water in that tub. And, this doesn't even come close to talking to what the thermal losses take away from you on this one with this situation.

Now, having said all of this, can it keep up and keep something reasonably cooled? Perhaps. It's all in how many watts of heat that the system makes and how many watts are lost to thermal losses getting to equilibrium. I suspect all but the high-end gamer systems would just warm up the water a bit and run fine, sans shorting out. You get high point heat temps and thermals that can boil because you're using a constrained heat exchanger in most water cooling rigs (to keep the water away from the system, something he's not bothering with...) which means you've got a smaller volume of water that you're directly heating- which CAN cause boiling and you need the radiator, etc. to ditch the heat.

I agree with the conclusion you arrived at- it's not practical in the slightest. But the grounds and the premise you use I don't at all agree with.

[gtrawoger]

Just a quick update for those who are interested:

So the company KROWN got back to us and promised to sponsor some more experiments, since until now we used old junk.

I went out and purchased a P5N-MX with an E1200. Got wireless keyboard and mouse, and a 8GB flash stick (which got Ubuntu 8.04 on it). Also, got bluetooth and wireless networking.

The current setup has an integrated video, and only the power and the video cable going to it.

The first test run was pretty impressive indeed. The processor never went over 30*C (which isn't great since I didn't overclock, but the fan couldn't spin in the water... so it kinda made up for it) at 100% load. I also loaded the video card to the max, which would have heated the chipset up pretty good. But the chipset never went past 14*C, even after 12 Hrs of running.

I think if I could get the fan to spin on the CPU I could bring the temp down there as well.

Overall the system ran stable for 12 Hrs @ 100% load. I had a temperature sensors in the water, and the water never went past the temperature of the chipset (always a little lower than the chipset).

There were two mayor issues though, which happened in the previous tests as well:
One issue that transpired was electrolysis. Even though the KROWN product protected the contacts from shorting, for some reason there were 4 SMT Capacitors that had one pole of it eaten away (if you looked at the process of electrolysis you'd see that it eats the metal in the process) almost completely (system still ran, but we shut it down).

The second issue was the CMOS battery. Because the housing of it is made in a stainless metal the protective product would not stick to it (asking KROWN about that they confirmed that KROWN T-40 does not stick to plastics or stainless metals ... what would be the point anyways!)
So we got some bad corrossion on the battery. We tested one battery with KROWN and one without, and the one without would COMPLETELY rust within an 1 1/2 Hrs... crazy electrolysis!!! With the product it at least lasted longer, but still not satisfactory. That is why we currently run the tests without a battery until we figure out how to fix that (but this means no overclocking).

It has been very interesting so far to see what happens. Our goal is still to have it run underwater, but we may need to help things on a few parts of the board (some other coating??)

Oh, BTW, we submerged one board in water without anything on it. It wasn't pretty. First, it shut off almost instantly and the proceeded to *decompose* (strange green slime) within a couple hrs. Funny enough the PSU still worked after. It must have had some protection inside it.

I need to take more pictures and videos and then edit together another YOUTUBE of this new experiment. We checked the PH level too, BTW, so that if we have some Goldfish in there they won't die on us.

[Svartalf]

Just a quick update for those who are interested:

So the company KROWN got back to us and promised to sponsor some more experiments, since until now we used old junk.

It'd be a BIG ad and PR source for them now that you've done the initial insanity. I'm not at all surprised...

I went out and purchased a P5N-MX with an E1200. Got wireless keyboard and mouse, and a 8GB flash stick (which got Ubuntu 8.04 on it). Also, got bluetooth and wireless networking.

The current setup has an integrated video, and only the power and the video cable going to it.

Nice choices. The on-board video, while not impressive on performance, lowers the contact point count down so the stunt you pulled here will work better.

The first test run was pretty impressive indeed. The processor never went over 30*C (which isn't great since I didn't overclock, but the fan couldn't spin in the water... so it kinda made up for it) at 100% load. I also loaded the video card to the max, which would have heated the chipset up pretty good. But the chipset never went past 14*C, even after 12 Hrs of running.

Don't overclock. You're hitting thermal equilibrium with the water because of the radiative and evaporative losses you have with the lashup. Overclocking may put enough thermal dissipation in the system to offset your losses and it'll heat up more on you- perhaps too much.

I think if I could get the fan to spin on the CPU I could bring the temp down there as well.

While it'd look cool, it's not going to help as much as using something like an aquarium filter pump or small fountain pump to force convection in the bucket. That is, if it'd not cause other problems. The rust-inhibitor/dispersant is adhering to the board and parts- unfortunately, too much convection may peel enough off that it fails early. You want some heat exchange off the sinks that're there on the motherboard (incl. CPU)- but if things run nicely without the convection it's doing well enough on it's own.

Overall the system ran stable for 12 Hrs @ 100% load. I had a temperature sensors in the water, and the water never went past the temperature of the chipset (always a little lower than the chipset).

Nice.

There were two mayor issues though, which happened in the previous tests as well:
One issue that transpired was electrolysis. Even though the KROWN product protected the contacts from shorting, for some reason there were 4 SMT Capacitors that had one pole of it eaten away (if you looked at the process of electrolysis you'd see that it eats the metal in the process) almost completely (system still ran, but we shut it down).

Heh... That means there was a faint conduction path there. The KROWN didn't stick as well as it ought to have on those parts. Not a short that prevented operation, but a conduction path all the same- which means you've got a lot more electrolysis potential going on there than one would like to have. You can at least stretch this out further with distilled water which will be almost non-conductive for the purposes of this little experiment. The KROWN will prevent leeching into the water of most of the copper, etc. which causes distilled to become conductive (Pure water is an insulator...)- but it won't completely keep it from doing it. That should handle running for at least 12-24 hours or more immersed that way.

The second issue was the CMOS battery. Because the housing of it is made in a stainless metal the protective product would not stick to it (asking KROWN about that they confirmed that KROWN T-40 does not stick to plastics or stainless metals ... what would be the point anyways!)
So we got some bad corrossion on the battery. We tested one battery with KROWN and one without, and the one without would COMPLETELY rust within an 1 1/2 Hrs... crazy electrolysis!!! With the product it at least lasted longer, but still not satisfactory. That is why we currently run the tests without a battery until we figure out how to fix that (but this means no overclocking).

Perhaps covering the battery with wax or something like Vaseline will work to prevent water getting at it. It rusted because of galvanic action from off the battery itself.

It has been very interesting so far to see what happens. Our goal is still to have it run underwater, but we may need to help things on a few parts of the board (some other coating??)

Conformal coating for the problem parts that don't come out, probably Vaseline or similar for the ones that do... Past that, I don't know what else would work right at the moment.

Oh, BTW, we submerged one board in water without anything on it. It wasn't pretty. First, it shut off almost instantly and the proceeded to *decompose* (strange green slime) within a couple hrs. Funny enough the PSU still worked after. It must have had some protection inside it.

Heh... A predictable result. By the by... You should show the immersion of an unprotected board and a protected board.

I need to take more pictures and videos and then edit together another YOUTUBE of this new experiment. We checked the PH level too, BTW, so that if we have some Goldfish in there they won't die on us.

The temps might present a problem for some fish, and the KROWN on the surface may also be a problem. Fish might not be a good idea, but everything else is just cool.

[gtrawoger]

Heh... That means there was a faint conduction path there. The KROWN didn't stick as well as it ought to have on those parts. Not a short that prevented operation, but a conduction path all the same- which means you've got a lot more electrolysis potential going on there than one would like to have. You can at least stretch this out further with distilled water which will be almost non-conductive for the purposes of this little experiment. The KROWN will prevent leeching into the water of most of the copper, etc. which causes distilled to become conductive (Pure water is an insulator...)- but it won't completely keep it from doing it. That should handle running for at least 12-24 hours or more immersed that way.

Yeah... something was happening there. I couldn't figure it out, since I had not problem anywhere else. Only the 4 Capacitors on the Power Circuit on the Mobo. Maybe too much power for the KROWN to insulate it from. I guess submerging the PSU is out of the question then ...

I really don't want to use distilled water. I am not sure why but it feels like ... ummm... *cheating*. In any case, by now the company wants us to show the potential it's product has to protect electronics (which are in cars too) from water and salt. And going distilled is not really showing that off.

BUUUT... they have another product, which they say is thinker and so will put a bigger barrier on the metal. The only thing is that it will not "creep" as fast (supposedly the KROWN product actually will move along the metal to the inward parts and protect there too.)

So more test to be done. Actually we got another Mobo (same model) which will be outfitted for the next test soon to come.

Perhaps covering the battery with wax or something like Vaseline will work to prevent water getting at it. It rusted because of galvanic action from off the battery itself.

Yeah, something like that. I was thinking of covering it with silicone and permanently mount it on the mobo. It will still last a few months, even though the computer is not plugged in. Then I would have to cut it out and try again.

Vaseline might work good too though.

Heh... A predictable result. By the by... You should show the immersion of an unprotected board and a protected board.

We are going to do that. We used 6 old motherboard in different scenarios' (all are retired now, can't really reuse them in the test since they have some KROWN left on them). When we submerged the one motherboard without protection on it we were still using our ... ummm... *old* cameraman (who has since been retired too) who didn't know what the record button would do... grrr.. so we will repeat the test with another board. These Asus boards are only $50, so it will be worth the cost to see what happens... now to find a reliable cameraman!

The temps might present a problem for some fish, and the KROWN on the surface may also be a problem. Fish might not be a good idea, but everything else is just cool.

Yah, we are holding back on the fish idea until we are ABSOLUTELY sure (we have both a PH Meter and a Particle Counter) that it would be save for them. We might have fresh water flowing through when we do the fish test. I am wanting to build a nice plexi-glass *case/aquarium* for this. Still on the planning stages there.