Submerged Phasechange cooled

[intomwetrust]

Alright guys, me and my best friend have taken up a project, to create a phase change unit.

Our first step was finding a fridge on the side of the road, taking it apart and testing it out to make sure it all worked, and it did.

Now the task was to cool the cpu. And so after a bit we thought of way to cool the cpu and re gassing the system to make a direct die block was looking a tad difficult, then it came to us. Submerge the computer in canola oil, then use the phasechange unit from the fridge to cool the oil.

It really looks like a super easy task. Build a case out of angle aluminum, and sheets of aluminum, insulate the "aquarium" part, and put the radiator in the oil and put a lid on it. No worry about condensation if we fill it basically to the top with oil. We also tested how cold canola oil gets before it freezes and it doesnt freeze at -15*C, apparently it freezes at -18*C to -20*C according to www.canolainfo.org

The thing about cooling the oil would be you would cool EVERYTHING, ram, chipsets.

I just have not found any flaws yet. What do you guys think?

[baldrick]

I would think that it would be easier just to go the tried and tested route and build a water chiller.

http://forums.extremeoverclocking.co...d.php?t=101902

[intomwetrust]

I'd rather venture into new territory though
i guess no one has done this before.

[Faaip De Oiad]

It will look horrible, it will be a ***** to clean up, forget about ever selling any of the components on to anyone else after you have put them into that system, and I would seriously question the capabilities of a fridge compressor in removing that much heat from such a large volume of oil.

[intomwetrust]

well from what i heard turning compressors on and off is bad for it.
so it is just going to run 24/7
so the only worry is what is the temperature when it hits equilibrium

[Shyfty]

Sounds like an absolute nightmare... Why dont you build a chill box...? That way you can chill the comp at a much lower temp. I don't think that oil has very good heat transfer properties and you probably will just end up over heating. Think about it...no fans are going to work (for long) in oil. you are relying on completely passive cooling. The only way it could possibly work (in my mind) would be if you were planning on going the DD route.

[Lurid]

/second the chillbox idea

If i'm not mistaken a full chillbox, with the PC enclosed in it would also require much less (if any) condensation proofing.

[sluflyer06]

Originally Posted by intomwetrust

well from what i heard turning compressors on and off is bad for it.
so it is just going to run 24/7
so the only worry is what is the temperature when it hits equilibrium

actually refrigerator phase setups are made to turn on and off alot and NOT to run 24.7. Also i believe it has been found that mini fridge size compressors can only remove about 15-30 watts of heat....but i say go ahead and try and see how it works...though keep in mind you'll never be able to clean those parts once they have been submerged in oil so you'll never be able to sell them.

[Drewmeister]

Given this oil is compatible with all the electronics and metals...the actual core temps will suffer greatly due to the lack of circulation and poor thermal properties of the oil. That oil freezes at too warm of a temp so would be almost impossible to pump at sub 0c. You would need an oil pump($$$) and better oil for beginners. Even pumping -20c oil around all the components is going to be very inefficient. I don't see this outperforming a chillbox or even mild chilled liquid. You would need to go very cold with the oil to be of any real benefit.

[johnny45]

Alright, I don't know about chilling the oil part, but I've seen the entire computer submerged in oil before.

Its a successful (albiet messy) way to silently cool a computer. Like I said though, I don't know what the effects would be like trying to chill it.

[nd4speed]

Sounds like an interesting idea, but with such high viscosity finding a pump that can circulate the oil is going to be a real challenge. With poor thermal transfer properties of oil you'll probably end up latent pockets of heat (unless you have really good circulation, back to problem 1). Also I would think you'll have to keep the oil totally pristine, any contamination could introduce the risk of unwanted conductivity or capacitance. Like others I think you'll probably have much better success if you go the chiller route or chillbox.

[Panthols]

Hrmm... for silently cooling a PC, please correct me if I'm wrong but dont most phase setups run 30-40db+?

[baldrick]

Mostly depends on the compressor but yes.

[intomwetrust]

oh that reminds me, how hot can the compresser get before it starts to malfunction?

Hmm so i would need a rather heavy duty pump then.

EDIT: Does anyone know of a less viscious fluid that doesnt conduct electricity that i could use?

[Lurid]

3M makes something along the lines of what your looking for. Though its ridiculously expensive....

[intomwetrust]

yeah flourenert...
and idea was using something like alcohol seeing how it has a really low melting point, its really really thin so moving it wround would be great. And whe you pull it out a videocard or something the alcohol would evaporate.

[Shyfty]

Yeah using alcohol based liquid can be a real safety hazard...not too sure your parents would appreciate that...

How hot can a compressor get? Depends on the type... recips, most suggest staying under 80C, but rotaries can go to 100C or so. BTW compressors are made to cycle...just make sure not to short cycle.

I really don't know why you are going down this path...sounds pretty risky trying to avoid contaminating the oil (or whatever "inert" liquid you choose) with very little in terms of "reward". Also, all that cooling capacity would be wasted trying to cool an oil-like fluid.

Well live and learn...

[johnny45]

Ok, Have you thought of maybe trying to chill the container from the outside, and using the oil only as a thermal conducter inside the box(like as5 between a cpu and HSF, except larger scale). Though the thermal conductivity of oil isn't that great, a phase change setup, i think, would be able to remove the heat out of it.

Or maybe try sticking heat removing things down inside the oil (like blocks of metal that are being cooled) and insulating the outside of the computer.

I think that trying to circulate the oil is going to be a dead end.

Just throwing out a couple ideas to play with, maybe they are dumb.

[Tanner8]

I never knew you can run a whole computer under oil. Thats so cool, how wouldnt it short it? And after your done, there is no way you can just rinse off the parts with water or somthing. Its gonna be all slimy and sticky for ever. Ewww, anyone got pics of comps running under oil? sounds intresting.

[sdat1333]

I third the chillbox idea.

All the advantages of oil, yes, you do have to insulate and seal the actual box, but think of a chillbox as basically your idea, only instead of oil you use air. You can switch out parts in chillboxes relatively easily (you just have to wait for it to warm back to room temp before opening it, and this can take hours. Of course you can always keep a nice 100w lightbulb or a heating element in there to warm it back up when you so desire), it doesn't get any oil on your parts, and as long as you make it reasonably airtight you won't have problems with short circuits. Also, any leaks are easy to spot becuase ice starts forming all over the place.

But I do understand the feeling of wanting to go where no one has gone before