Mineral oil

[tkal]

I am tring to find a good cheap supplier of mineral oil. I plan on using it in a immersion cooling style computer project. I have look at 3M's web site and called some of their distributors. While I have not been able to track down any mineal oil with them they do have fluorinert, unfortunaly the cost is rather prohibitive at $3,000 usd. If you are wondering I was looking at FC-77 if you care to glance at it. In my searches I have not been able to track down a mineral oil supplier. If any of you know where I can get this I would be greatly appreciate it.

As I understand it, grocery store bought mineral oil does not have a high enough dielectric strengh to use this in an immersion cooling type project.

Another thing about the FC-77 is that it will "eat" certain plastics. PCB? The tank portion of the case will be made out of lexan or plexiglass.

 

[d23]

tkal:

I would guess that any oil distributor company in your area can provide "transformer oil." If you can't find one local try transformer manufactures like GE, Southwest, Howard etc.

Good Luck

 

[tkal]

Ok, got it. I will try that. THANKS!

 

[busbar]

Available in barrel-to-railcar quantities—

http://www.shell-lubricants.com/Electrical/tests.html

http://www.chevrontexaco.com

http://www.mobil.com/USA-English/Lubes/Your_Industry/Power_Generation_Utilities_Collection.asp

http://www.intech-edm.com/pdf/SpksFl00.pdf

 

[fuggeos]

Hello,

I am also involved in a project whereby some piece of electronics will be (hopefully) cooled by oil filling a pressure vessel that will be submerged in water. We're looking at DC200 silicon oil by Dow Corning, the 20 cs viscosity type. Shell Diala has also been used in these applications.

I haven't looked at costs yet, but you can get in touch with Dow Corning for price info.

Hope this helps

 

[Comcokid]

The Dow Chemical oil that fuggeos suggested is excellent. It is used as a base in many cosmetics and in many electromechanical applications as a dampening fluid.

Another possibility is Baby Oil just like you find at the local drug store. The added fragence may get annoying. These are fine as long as you don't intend to go to high temperatures. We use it for in-house engineering test of sensors used in diesel - much safer from a flammibility standpoint.

If you are trying to over-clock a high speed processor, you also need to be aware that the presence of oil instead of air above any high-speed busses that are on the surface of your circuit board may have their impedance affected resulting in signal integrety problems.

 

[ScottyUK]

What the....?

 

[busbar]

Hopefully the CPU fan won't make too much foam. ;-)

 

[busbar]

Now add mushrooms, shallots, garlic cloves and toss with 4 tablespoons of balsamic…

 

[JavaMoose]

Lol, no CPU fan...actually the only fan is on the PSU. Could have submerged it, but decided not to.

Lol also at the receipe...the Mineral Oil says it's laxative, think I'll save the dressing for the In-Laws!

So...you guys like? Like I said, not totally done yet...more coolness to come!

 

[Nurotic]

Wow, this is a really cool project. I currently work tech support for a University's libraries, and can obtain a lot of crappy computer equipment (500mhz at best). I'm wondering how much I could overclock something if I attempted what you did. Do you have a way of cooling the oil?
Also, (more importantly), how the hell did you construct your case?

 

[melone]

While I admit that this is a COOL project, I don't know how practical it is in overclocking schemes. My experience with micros and DSPs, is that very little heat is transferred out of the package via the mold compound. The vast majority of the heat comes out of the package through the solder balls, usually through ground. The reasoning behind it is that ground should be the biggest piece of copper on the PCB, and could therefore pull the heat out of the device and distribute it across a much larger surface.

My point is that if the copper around the processor isn't getting too hot, I don't know if you will be able to "pull out" any signifigant amount of heat through the mineral oil. (Replacing a very poor thermal insulator, air, with a better one, mineral oil)

Are my assumptions / statements valid?

 

[Nurotic]

You're right, oil is a better insulator than air. I saw someone else's project, however, that involved much the same thing, but he filtered his mineral oil through an air conditioner somehow to make it really cold. He claimed he'd gotten it down to -30 (Farenheit, I assume). He overclocked a 300mhz machine to a 1Ghz.
I'm just wondering how easy that is to do. I remember him having a problem of water condensation on the air conditioning unit, which freaked him out 'cause he didn't want to fry his board. So basically I'm wondering if there's a way to pump the oil out of the case, cool it significantly, and pump it back in, setting up a loop to get the oil really cold. This might make it loud, which would defeat the purpose of having a fanless, quiet machine, but at least you could overclock the hell out of a system.

-Nuri

 

[busbar]

Small aside — There may be an eventual degree of softening/swelling in oil-immersed PVC wire insulation.

 

[Cannibal]

WOW! Very nice. I wonder if a Peltier attached to a copper tongue, with one end immersed in the fluid would achieve the oil cooling you are looking for. I don't think the Peltier will work correctly if it were immersed itself, but I may be wrong there. You may have to add some type of mechanical stir to the oil also, as I am not convinced that convection would effectively distribute the cooled oil throughout the box.

This is an awsome project and something I have wanted to try myself for sometime. I just didn't think it could be done effectively without the FC-77, which I believe has a much thinner consistancy than mineral oil. Hey, maybe I can even put my TV in an aquarium now, plastic plants and fish of course.

Please be sure to keep us posted.....

Cheers,
Cannibal

 

[RAMConsult]

There are quite a few cooling schemes for overclocking, but most liquid cooling systems use direct contact heat transfer modules piped to external cooling devices in place of the standard fan/heatsink combos. Submerging the board in the cooling medium subjects the board, components and connectors to contamination from a possibly incompatible fluid.

See

http://www17.tomshardware.com/search/search.html?category=all&words=water+cooler

for a comprehensive review of competing methodologies.

 

[ScottyUK]

Has anyone considered a "heat pipe" for this sort of application? I think they are an interesting and little-used way of getting large amounts of heat away from a source and into a large heatsink or cooler.

 

[RAMConsult]

ScottyUK, there are a few suppliers of heat-pipe coolers specifically for CPU cooling but they lack flexibility in mounting and require high fan speeds (=noise) for adequate heat removal.

 

[melone]

Wouldn't it be WAY easier to buy a faster processor?

 

[Nurotic]

Probably, but that would defeat the purpose of having a project. Price is an issue for me, but mostly I just want to tinker.
I hope JavaMoose writes back soon. I really would like to know how he built his case.