submersive cooling Q?

[adamtekh]

ok so as a project for my web design class @ school my and my budy sinister will doing a page on overclocking ( where did we come up with that idea? )

and well we have a mmx 200 to trash while we cant OC the beast of computing power we can still cool it

so distiled water submersion is what we will do or just maybe throw it in a freezer

we dont care if it dies , its more of a test to see how long it lasts

but i have a question about the HD would it work if it was under water?

and how cold is to cold for the HD

 

[overdoze]

From what I know HD does not work well at cold temp. Check the operating temp on the HD website you will know the exact temp for your HD brand. AND i better not work under water unless you seal it from entering the motorize inside your HD phew what a crazy idea hehe?

 

[adamtekh]

well its a really old western digital 2.5 gig caviar 32500

but i also have a pair of 1.12 gig 21200

i dont care about the 1.12s

 

[craigiz1]

There was a thread started in the forums somewhere about this topic. The guy was going to submerge the M/B in transformer oil.
Find it and see what he did.
Good luck.

 

[deez]

yeah I saw something like that with some type of oil submersion. I think the system lasted a few minutes maybe. If you try that it has to be nonconductive. I would think a mobo in a freezer may last a few hours till the condensation kills it

 

[Diggrr]

Good luck finding water that don't conduct! Even the stuff I distill at home conducts...less than the store bought stuff, but it does.
Plug the psu into a powerstrip with a breaker in it, that'll save the trip to the basement to reset the main breaker.

 

[mustangman]

You're braver than I am even r/o water has some conductivity and it's hardness is normally pretty low out of a good r/o system. A good way to test the purity of your water may be to put low voltage to it and let the neighbors cat drink from it if he starts freaking it may not be good enough for your project! Good luck!

 

[jimmytp]

adam,

I too am trying that same thing. I bought a used chest freezer last week(that's a fun story about a long road trip), and have been looking at fluids for a while. I have spent the last 2 months looking into this pretty seriously. Don't use distilled water. Theoretically, it would be fine. In practice, you'll toast you components in no time flat.

Leave your HDD out of the freezer and fluid too. No matter how new or old it is, it was not meant to operate in those conditions. I wish I could find Balazs Bodosi's article on his oil cooled Power supply right now, but you can put those in the fluid/freezer. I will look for that page tomorrow, but look for it yourself in case I can't find it.

Transformer oil will degrade your components in a year or so, so for your project it might be fine. But if you intend to freeze it, it is meant to be used at higher temps, and will be very viscous at room temp or below. Mineral oil will be similar, but less viscous.

Syltherm is a good coolant and is a non-conductive silicon based oil. It looks good, but is expensive and hard to get a hold of in small quantities. If you are interested, let me know though. I posted a thread a while back to see if I could find anyone to split a jug of it with me, no one but you responded. There are a few places that sell it in 5 gal. jugs.

Balazs Bodosi suggested brake fluid to me. DOT5 is silicone based and I think it would work, but it is pretty pricey too, $35-80 per gallon. It is what I have on the top of the list right now though. I will use more or less coolant depending on the price of what I use. He also suggested using a coating to put on the board to isolate the components from the coolant. His experience with dissolved plexiglass does not appear to be the answer, but it is a good idea if we could find an alternative.

I was also thinking alcohol since it is really cheap, dielectric and extremely fluid at low temps, but it is volatile and should be used with caution. Don't breathe the vapor from any alcohol for very long, all alcohols are toxic. I can't recommend alcohol, though it is an option.

I need this semester to end before I really finish this thing, but I am collecting pieces and working on what I can when I can. I have an old 486DX2 or a P200 that will test the system for a while. Let me know what you come up with and/or how it goes. Let me know if you have any other questions too, I might have come across something in my research.

 

[jimmytp]

This is a link to an article by Bodosi. There is a link to his earlier article that goes into the project a little closer in the first few lines. Thanks for all your help B.

 

[mustangman]

Fluronert is something you could look into although it is very expensive, I think around $500/gal. It's used for testing electronic components in submersed conditions. Sorry right off I don't remeber who makes it. I would be cautious of brake fluid even the high quality dot5 is somewhat corrosive although it's boiling point is quite high around 600-700f if I remember correctly and does a good job of dissipating heat.

 

[adamtekh]

what about coolanol

 

[jimmytp]

Fluorinert would be ideal, but it is expensive from what I hear. Ramil Tranquilino of OCTools is in NZ, so I don't know what it would cost in the states. Also, Fluorinert is a family of compounds from 3M, so the cost might vary. FC-72 looks like the stuff we would want, at least if you plan to chill it. Otherwise, there is quite a range. FC-87 would work better, but it boils just above room temp. There are also other compounds from 3M that would work, but they have material comptibility issues associated with them. Some might work though. I will call tomorrow about costs and quantity.

From what I found, DOT5 used to be pretty incompatible with polymers, but recent formula's aren't a problem now. I don't know if I want to put my board in that stuff until I see someone else do it, but it looks like it would work well. It's purple too. That's a cool color and all, but I want to see the board.

Coolanol is 50% water, so that's out. Evans has NPG+, a nonaqueous coolant. They were real helpful when I talked to them, but we never resolved anything. I will call them again if nothing else pans out, but those kinds of coolants are designed for high temp apps, not low temp. I also contacted Glenn Cannon about Biotrans, an oil for use with transformers. Same thing, good for high temp apps, but pretty thick near 0C.

I've been back and forth on all of these. I will get a hold of 3M tomorrow if I remember. I want to find some place that sells DOT5 for a reasonable price. I think Syltherm is a pretty good option too, but I would need to find it cheap. I am still thinking alcohol as well. I have a feeling I am going to end up with isopropyl because of the cost factor alone, $5 a gallon for the good stuff through my work. That's pretty hard to beat when the others range from $30-$500/ gallon. The head of our analysis department told me to call a few places and see if I could get some samples. He said Dow or 3M might send a gallon if I told them I was evaluating coolants, and left out all the details. If they give small samples, then what we need to do is have all our friends call too. We might get enough altogether to put one of these things together

I just realized that Ramil did a follow up to his first project. It's been months since I looked at that site. I'll post again tomorrow if I find anything out.

 

[RoadWarrior]

I'd have to do a heck of a lot of testing before I trusted anything to brake fluid, even if some other crazy assed foo' did it first. Like boil scraps of PVC, PCBs and tubing etc in it on the stove for a while.

 

[Diggrr]

For testing, you could use an old AT psu and a tupperware container (buy one...Mom or the Mrs. aren't very understanding here). Put an old hsf and a small pelt on it sitting on the psu just out of the oil to keep it cool and test it's cooling capacity at the same time. The psu would probably cost less than the brake fluid.

 

[adamtekh]

yeah gots a atx 300 watt extra

just jump the green wire

 

[jimmytp]

I share your sentiments warrior. I have a 486 that is going to get "donated" to project immersion as a guinea pig. But I doubt that it will be in long enough to show degradation before I get bored and put one of my good machines in it though. I'm real impatient.

That's a good idea diggr, I have scrap PS's that I was thinking about demoing first anyway. I need a spare pelt though. My mom said she just bought one of those pelt cooled coolers, I wonder if she'd know it was missing...

Do you guys have any ideas/suggestions on other coolants?

 

[Diggrr]

You could also do it without the pelt, I just thought it might accelerate the test by puting a good load on the psu. A plain hsf would work too...remember the fan in the psu isnt' gonna be pushing air, just fluid. You might also want to make sure the psu fan is totally submerged, you don't want it half in the fluid and slinging it about the room. Opening it up before testing and brushing out the killer dust bunnies wouldn't be a bad idea too.

 

[jimmytp]

I was going to take it out of the metal box and just submerge the board. I don't need to burn out a fan trying to "sling fluid about the room." lol That should work, right? Do I need anything to circulate the fluid, or can I let convection work it's magic? In a setup with a machine, I will have a pump, but I .... have a little spare pump that would love to bathe in purple fluid. I still like the pelt idea.