In search of non-conductive liquid for liquid cooling setup

[rawlic]

I'm in a search for a non conductive liquid that I can use for a full liquid cooling setup (ie plunk the whole motherboard in a liquid to run through a radiator/compressor setup). I've heard that mineral oil fits the bill, just don't know how to really test the theory.

 

[mechBgon]

To test the theory, you could get a multimeter, set it for resistance (ohms) and put the probes in the fluid. If the resistance is very high, that's what you want.

A material that came to mind, besides mineral oil, is silicone oil, like here. However, I don't know about the environmental aspects of that stuff, whereas mineral oil is pretty harmless.

 

[Freshbrain]

Destilated water. Pure water is a non-conductive liquid. What makes water conductive is all the "extra" stuff in it, like dust, other liquids, Cl, etc.

Whatever you are to use, you should make sure it doesnt go getting dust cause its gonna make it conductive, or renew the liquid once every two weeks or something.

I recall reading a tomshardware.com article about building a home made water cooler where it explained the whole "conductivity in liquids" thing quite well.

 

[FenrisUlf]

THEORETICALLY distilled water would work, but in reality it won't. Pure water is non-conductive, but impurities such as dust or anything in the case or on the motherboard would significantly increase conductivity. I have a water distiller and have tested the output - with a 30 v source I can get about 2-3 ma across a 2" distance. Even a single grain of salt in a pint of distilled water will greatly increase the conductivity. Now think that many of the traces and leads at 12 volts are only 0.1" from ground. As soon as this happens, atoms of metal from one terminal are released and migrate to the other (it's called electromigration, the process used for electroplating, also this is a way to turn water into hydrogen and oxygen gas called electrolysis). When this happens you get MORE conductivity due to the impurities in the water, leading to greater current flow, more electromigration, more impurities, etc. If you want to run a test of this: get a 12v source or even a 9v battery and connect them to two pieces of bare copper wire about 6-8" long using alligator clips. Put the wire in a glass (use real glass, it's more transparent and is non-conductive) and make sure the wires don't touch, about 1-2" apart. Fill glass with distilled water and put something over it to keep and dust out. Let it sit overnight. Observe what would have happened to your computer had you used distilled water. There will be a big mess of oxidized copper from the oxygen generated and a bunch of copper that has come off the other wire. Your battery will probably be dead because the low resistance of the contaminated water has caused such a large short-circuit. I've done this with two 6" silver wires to increase surface area and about 30v (30v produces the least electrolysis) and after about 30-45 minutes you can see one of the leads is heavily tarnished and the other has a visible trail of microscopic sized metal particles falling to the bottom of the glass. There was about 25-40 ma flowing through the circuit - enough to light up an LED and more than enough to cause all kinds of problems with sensitive electronics.
Nobody uses distilled water for an external liquid cooling setup (at least not for more than a few minutes or with the same equipment twice).

There are only two reasonable choices that I know of, mineral oil and Flouronet (sp?). Flouronet has better thermal tranfer properties and can get very cold before thickening, and cleans up easily but it costs several hundred dollars per gallon. Mineral oil doesn't transfer heat as well and will not easily come off of components, but it is cheap. Do a google search and you will find examples of both. Personally I watercool and have no need for anything more extreme.

 

[WarCon]

If you do decide to go with either the flourinert or the mineral oil, I would make sure you seal the project from exposure to air. Condensation will allow water to build in the liquid and will eventually find its way to your components.

Dr. Freeze (no idea his real name) used mineral oil and hung the freezer coils in the mineral oil. It eventually shorted out from water build up in the mineral oil. Wish his website was still up, but I haven't been able to find it.

 

[rawlic]

Wow. To be honest, I wasn't really sure how to test the conductivity of a liquid, but you guys helped a lot. I don't really need the setup as I don't even OC my components. I just thought that it would look cool and the geek factor of it all would be nice.

 

[socketman]

several people have looked into alternative fluids from overclockers.com and HardOCP. In all cases, the liquids they found were VERY expensive. Also, the manufacturer would not sell to individuals. They only sold it by the 55 gallon drum, which cost several thousand dollars.
These liquids had zero conductivity but had a high heat capacity (good for carrying heat away from objects). SOrry, i forget the names of the liquids. But since you couldnt get them anyway, doesnt matter.

PS: forgot to mention they were considered hazadous materials. So unless you had a hazardous materials storage permit, you couldnt get the stuff delivered to you anyway.

 

[strikermike]

There was a segment on the screen savers a while back where they submerged a motherboard in a non-conductive liquid. It worked pretty well. I don't remember the name. It was available in one gallon jugs. I believe the cost was about 200.00 USD per gallon and it evaporated quite rapidly. You might do a search on their website.

 

[TJ69]

3M makes a non-conductive liquid for your purpose. I believe this was the same stuff they used on the Screen Savers segment.

 

[nemo160]

sounds like the last two posts are talking about Fluorinert
i just really can't see using a liquid that costs a few hundred dollars a gallon and evaporates just for cool factor, which seems to be your goal since you don't want to oc
i've read about mineral oil being done before but it doesn't strike me as a good idea, kills resaleability and any prayer of warranty
also you would have to have enough flow throught the case to cool the cpu, doable but difficult
work on making a lava lamp case window or something instead

 

[Den]

Forget where I saw it, but I read an article about a couple of guys who used flourient (sp?) and Liquid Nitrogen to do an extreme overclock. It was a pretty cool little experiment. Cost em like $500.

 

[Richardito]

The best thing is to get silicone fluid. Check the Corning website for more detailed information. Just order a couple of quart samples and you should be ready to rumble.

 

[Jaxidian]

Originally posted by: Richardito
The best thing is to get silicone fluid. Check the Corning website for more detailed information. Just order a couple of quart samples and you should be ready to rumble.

Could you give a little more info on this? I did a search for silicon and found nothing like this.

Thx...

 

[MassiveUnit]

The guys who used Fluorinert and liquid nitrogen were stupid because they didn't read the .pdf sheet that describes the gelling tempurature of the liquid as much higher than liquid nitrogen's temp...thereby smoking their pump and wasting a decent bit of liquid nitrogen. If you do use Fluorinert, you need to have a completely enclosed setup because the Fluorinert evaporates. If you go to 3M's website, there are also other non-conductive fluids used for cooling that are not as toxic as Fluorinert. Another suggestion is if you really wanted to submerge your entire setup in fluid and didn't want to be able to resell it later, simply set up your computer with all the cards in it and no processor (be sure to cover the socket with tape) and spray the hell out of it with conformal coating. Then install your processor and a heatsink on it and seal around the heatsink and the socket (you need to fill the center of the socket with silicon) with caulk and you should be good to go. You also need to cover your hard drive connectors and such before you spray and once you connect the cables to them, seal them with silicon as well. If you do a good job of it, you could put the thing in basically any non-caustic, non-acidic, non-reactive, non-etc. liquid you wanted and you would be fine. That having been said, unless you really want to spend both a lot of time and a lot of money for something that is a one-time deal (or maybe more than one time, if you have the money for Fluorinert) I would suggest staying away from complete submersion simply because the results don't outweigh the complexity involved in setting up this particular project.