Liquid Immersion Cooling

[jimmytp]

Check this out. It's Bodosi's project, it might give you some ideas. He dissolved plexiglass, which should work alright. I would consider regular old epoxy, it should work pretty good, is easy to get, and easy to use. It's pretty cheap too. Get the slow set epoxy if you go that route. If you take the time and are real careful on each, I think either will work. A good craftsman never blames his tools, it's how you do it that matters IMO. There are a lot of caps on my board though, and in really bad locations, more so than my other boards, so it's gonna take a while.

 

[Greythe]

I remember hearing that the electrolytic caps are the only ones that pose a problem. Is that true?

 

[jimmytp]

I can't answer that. I have not heard anyone mention anything else, just caps. I would look real close at the board. Caps are covered in plastic, is anything else? What about pci/isa/agp slots? They are plastic, but a more rigid kind. Bodosi mentioned removing the caps to coat them, but you could just move them off the board a little and resolder them. That would give the casing room to expand without prying them loose. I don't know if that would resolve all the problems or not. Some jumpers have plastic bases. What about the ports/usb? Those could be kept outside of the fluid, but it would be hard to create a very tight seal around them, you would have to epoxy the back of them to prevent holes. And even still, a good seal would be hard to accomplish. What about the cards themselves? The interface would need to be sealed too. I intend to submerge the whole thing and get extension cords to keep the connectors that devices attach to out of the fluid. That might not work, but it's how I plan to proceed. Just more stuff to buy though.

 

[Greythe]

What about pci/isa/agp slots? They are plastic, but a more rigid kind.

I think the plastic in question is that rubbery kind... Not sure though.

Bodosi mentioned removing the caps to coat them, but you could just move them off the board a little and resolder them. That would give the casing room to expand without prying them loose. I don't know if that would resolve all the problems or not.

I am certainly not going to be resoldering anything on my board! I completely suck as soldering (lol) so I think that'd be an automatic death sentance.

I intend to submerge the whole thing and get extension cords to keep the connectors that devices attach to out of the fluid. That might not work, but it's how I plan to proceed. Just more stuff to buy though.

Or do what I'm doing, and keep the board vertical and face the ports up. That way, you can set the liquid line just under the point of contact, and you don't get anything messy.

Course I'm building my own case, so...

Anyone have any ideas for what to do with my PSU, hard drives, etc?

 

[jimmytp]

You should read some of our other threads. That's just one, but there are a couple more floating around there somewhere. Some of that is redundant, or maybe some of this thread is redundant, but either way...

HDD were never meant to be frozen, let alone immersed. If you are just going to cool it, not a problem, sub-0 though, they will need to be outside the enclosure. That's way out of HDD specs. PSU should be immersed. It by no means needs to be, but WTF, why not? I would just drop my whole machine in the vat if I thought it would work.

As for the vertical idea, I am still in between on how to do it. I had designed, but not built obviously, a plexi tray that you could slide out, identical to a regular mobo tray. You could probably just use a tray from an old case. It would be vertical and have the connectors not exposed to the coolant. I want it air tight though, so I have come away from that and gone to total immersion. It really depends on the coolant though. Alcohol would need to be sealed, and fluorinert too. Cheaper coolants, or less toxic ones really don't matter. With oil it's not a problem, but with other coolants, water will be absorbed through those holes and then you could be up you know what without a paddle.

Obviously, cdroms and floppies can't be immersed, but they could be frozen maybe. Depends on how nice they are and whether you are willing to try it. I thought about mounting them in my freezer, and again WTFN. Or you could create an external housing for the drives and run cables to it. Then you need to factor in distance between the board and the housing during your design.

 

[jimmytp]

I know it's not lab grade alcohol, and it has contaminents, but I thought I would test the resistance of denatured alcohol. Olive oil and mineral oil too. And why not, water just for a reference. I will test some super pure, lab grade methanol tomorrow at work. I would like to find some DOT5 just to test as well.

Summary table in M ohms

Fluid______________Start___1minute__5minutes

Denatured alcohol____1.6____2.25____3.6
Tap water__________ 0.5____1.5_____2.65
Distilled H20_________1.0____1.9_____2.9
Olive Oil___________ Non conductive
Mineral Oil__________Non conductive

The DA resistance stablized within 5 minutes. Both tap and distilled water were becoming more resistant still at 5 minutes.

Just thought I would throw some numbers your way. I tested my multimeter with 5 different resistors just to verify the solutions were changing resistance, not the meter I am leaving a jar of distilled water with 2 copper electrodes at 5v differential overnight to see if I can pull some ions out of it and increase resistance, just cause.

 

[Greythe]

Sounds like it's worth a look.

I'm not interested really in cooling the external peripherals and storage devices, or the PSU. Since they won't be inside a case, they'll radiate their heat normally. What i meant with "what to do with them" was where am I going to put them? I was considering some type of deal on top of the case, but that's where the lid's going... If I were more industrious I'd build an enclosure for it all, but, well, i'm not more industrious.

I'll probably wind up having them sit on top of the desk. Whatever works.

Also, thanks for those resistance numbers. Useful information I don't suppose you could get the resistance of, say, WD-40?

 

[Diggrr]

Lurking here, thought I'd jump in.

Don't waste time with WD-40, it'll eat rubber and some plastics. I used it once on a $80 motorcycle chain. Two days later I was buying another one. The o-rings flung right out of the joints....just FYI.

Neat thread though.

 

[jimmytp]

I'm at my apartment for school, so no, I don't have much in the way of chemicals. Not unless you bring some down here

As for the housing, if you could find a removable HDD mount, you could attach it to the side. I don't like to leave my drives lying around except during testing. Or hacksaw an old case for a 5.25 in. drive rack to attach somewhere.

 

[jimmytp]

yeah, wd-40 is bad even for regular bicycle chains. sore lesson to learn diggrr. $80 ouch. What kind of bike?

 

[Diggrr]

That was my racing CB-750K (#3). I couldn't even begin to list the mods and overclocks on that bike...it wasn't even really a 750 anymore, more like 1100.

I've had six of them, then moved to Goldwings. Right now I'm bike free and I'm gonna be jonesing when the snow melts.

 

[jimmytp]

Soon Diggrr, soon the snow will be gone. Hibernation will end soon.

 

[jimmytp]

The experiment with the distilled water did not work. It was around 1 M ohm this morning. I will maybe use a multistage setup with salt bridges to draw ions into other chambers, isolating them. Have to go to work now, so that will wait.

Cheers

 

[jimmytp]

Hey diggrr, I got the ins on a what I think is a '99 CBR 900 that is probably gonna go for cheap, you interested? Not quite a goldwing, but a seriously bad *** bike to steady your withdrawl symptoms and just make you twitch. Super low miles too.

 

[Greythe]

Thanks for the tip Diggrr! That coulda been bad... lol

As for your distilled water setup, Jimmy, I'm not sure it's worth the effort. Running that stuff across your mobo will introduce all kinds of impurities into the system, and you'll just wind up having to constantly suck the ions out...

 

[jimmytp]

The reason I'm doing it is to see what happens. I'm the last guy who is going to put his board in water. If for any reason, then to prove to everyone who has thus far suggested distilled water does not conduct, that it will not work. The numbers from before should have already shown that though. To get water of that purity is near impossible, with our resources at least. I am interested in the science of it. How often do you have an excuse to set up some ion transport salt bridges? Maybe this tangent is inappropriate for this thread.

 

[Greythe]

Hey, the best reason for anything is "Just Because," my friend. Whenever I have to explain why I need a certain part (like, for example, when I was asking the people at the refrigerator supply shop about cooling mineral oil), everyone seems to stop and stare for a moment when I say "I'm dipping my computer in oil." Then I get the 'Why's,' to which I answer, "Just Because!"

It's a lot of fun to see peoples' reactions to this sort of thing. Which is why I can't wait to get it finished, and get some pictures out!

 

[jimmytp]

Allright, here it is kids. I just got in touch with Accuity, my local Syltherm FC-72 dealer. The price quote came in at $1928.50 for 5 gallons. That's just under $400 a gallon. So that's out, unless we can all find corporate sponsors....

 

[jimmytp]

Oh yeah, it was like pulling teeth to get a price out of them. After 5 days of talking to operators and techs and being bounced around from person to person, company to company(well only 3), and being put on hold by the last person for several minutes, it finally came out.

 

[Greythe]

Wow, why so secretive?