Oil Cooled Motherboard

[IRJack]

Question: Anyone here ever tried to submerse a motherboard in oil for cooling?

I work with electronics that are submersed in a synthetic oil, but I never had the guts to do it to my mobo.

If you put the mobo in a tank and circulated the oil through some type of radiator to remove heat, would it be effective enough to cool the cpu (with a passive heatsink), vid card, etc. Obviously, you wouldn't want to submerse your cd/floppy/hard drives due to the optics and the moving parts.

Some of the drawbacks I can see already are:
1) It's a darn mess
2) The system really needs to be sealed to avoid contaminants


What do you think?

 

[dakels]

I guess the oil doesn't conduct electricity at all so it wouldn't interfere with the miniscule amounts of volts flying around the MB eh? I'm sure this would be doable but it seems to be more a question of practicality to me. That sort of a system would cost alot more then a few heat sinks and fans. That brings me to a topic I wanted to post recently. Are water cooled systems really necessary, and in this case, oil as well? It seems to me more show then necessity. What on our current consumer systems is getting so hot that it can't be controlled with our standard arctic sliver, heatsink, and fans? Nothing that I know of yet. There maybe some new experimental stuff that may require more agrressive and radical cooling measures like water or oil but for now, who needs it? Even the water cooled systems have alot of case fans, which to me seems to defeat the purpose of being "fanless". The water tubes also take up alot of internal space from what I have seen. The risk of a leak, albeit extremely slim, is also a really scary thought and potentially harmful to the users.

Most of our systems target the chips that get excessively hot in our computers. In most cases the CPU, GPU, and sometimes RAM. Is there a real need to cool the entire motherboard and all its circutry?

IIRC the first Pentiums ran so hot and without the invention of heatsinks or CPU fans yet, the R&D crew resorted to using some liquid nitrogen or freon system? I guess if a new series of semiconductors running at 1000ghz ran so hot that no combination of adhesive compounds, heat sinks, or fans could disperse the heat, then maybe we would have to consider a more radical approach, but for now it seems we really aren't near that need yet.

Just my thought. So who hear does use a noncoventional cooling system like water and how necessary is it from a cooling standpoint (not sound)?

 

[FuriousBroccoli]

I'm building a watercooling system right now, and you are completely glossing over the two key factors. Sound, and overclocking.

Most high-end PCs are damned noisy. I've already taken many steps to quiet my computer, but when I switch it off at night, the silence is golden. Well I want that all the time, in fact I'd often like to sleep with my PC on, and since it's in my bedroom, that's even more reason. Listen just before and after you've shut down your PC today. Was it nice to hear thoses fans fade away to silence? If no, then fine, stick to air cooling.

Now, overclocking. That's pretty self-explanatory. You get quite a bit further with water, and it improves the life of your components. If none of that matters to you, then again, stick to air.

BTW, you're wrong about lots of fans. You need one single fan for a watercooling system, and you can purchase that fan in a speed that is completely silent. That's what I'm doing.

A leak is also a non-issue; you can get deionized water; most components dry out fine; watercooling systems never leak unless you set them up wrongly (that's why you test); and most importantly, you can't harm yourself from a leak, the highest voltage line is 12V in your computer, which you couldn't even feel if you stuck one end up your arse and licked the other.

 

[KnightMarauder]

The mention of oil cooling brings up an interesting hybrid: oil in a watercooling setup.

Since oil does not conduct electricity (they use it to seal those green electrical break boxes on the corner of my street) it would be nice to use as a replacement for water. You'd just have to get something that isn't so viscious that it won't flow easily. Admittantly, I don't have a watercooling setup, but I'd think that the pumps would be able to handle an increased viscosity of an oily liquid.

I don't know if this helps or hinders my point, but what's the difference between the heat capacities of oil and water? Is water the greater of the two or is oil? A higher heat capacity would mean you'd be able to pump less liquid through the system to dissipate the same amount of heat.

 

[FuriousBroccoli]

Most oils are semi-decent thermal conductors. There are a few problems that spring to mind:

-The heat could build up around the CPU unless you've got good flow over it.
-You'd wouldn't really want to cool the oil by pumping it, your pump wouldn't last too long for one thing.
-If what you use has any solvent or hydroscopic properties, your motherboard's connections will be erroded over months.
-Over time perhaps, the oil could act like a capacitor, build a potential difference, then short out your motherboard unless discharged.

The best liquids that come to mind would be fluroinert (or whatever 3M sells now, HFE?) and mineral oil. Personally though, I think you're nuts.

 

[FuriousBroccoli]

Water is a much better thermal conductor than almost all oils, and flows a lot more easily, so oil is a really bad choice for closed cooling system.

 

[rayster]

Isn't this kinda what you're talking about?
Submersion Cooling
They use HFE, but as long as the fluid is dielectric, I think anything would work.

 

[glugglug]

Pure water is an electrical insulator, not a conductor. It becomes conductive only after it has dissolved other materials (dirt, etc), whose molecules separate into ions when dissolved (the dissolved ions are what actually conducts).

I suspect the exact same thing will happen with most oils, so the oil will have to flow within its own airtight system (or at least not directly on the mobo), just as you do with a water-cooling system - if the oil dissolves dust/dirt from your case, it will no longer be such a good electrical insulator.

 

[PowerEngineer]

Rayster already posted the link to the same submersion cooling article I read a while back.

While pure water is a good insulator, you're never likely to run into water that pure. Solid impurities do not break down oil's insulative properites nearly as quick; in fact, it's water vapor getting into the oil that is the biggest problem. The windings in electrical transformers are isolated in a tank of oil that provides both insulation and a method for cooling (due to resistive winding losses and hysteresis losses in the iron core). The temperature difference is enough to cause convection currents in the oil that transports the heat away from the windings for conduction through the tank to the air outside. Of course, natural convection can be augmented by oil pumps to circulate more oil, and conduction through the tank walls can be augmented by radiators attached to the sides of the tanks to expose more surface area to outside air.

My worry about submersion cooling would be how readily convection currents would form in the liquid used. Certainly the liquid is likely to absorb more heat than air before its temperature rises a degree, but if that heat isn't readily convected (or conducted) away then your component temperatures could still rise to damaging levels. (Note that the article employs a pump for forced convection.) I'm thinking it makes more sense to apply herioc cooling measures to the critical components (e.g. CPU, clock chips, memory, ?) and leave the rest to air cool as they were designed to do.

Just my two cents...

 

[imgod2u]

The problem is, oil is not very easy to move around. Without circulation, pockets of the oil can get very hot. Although I have heard of people using mineral oil to cool their computers.

 

[dejitaru]

Most high-end PCs are damned noisy. I've already taken many steps to quiet my computer, but when I switch it off at night, the silence is golden. Well I want that all the time, in fact I'd often like to sleep with my PC on, and since it's in my bedroom, that's even more reason. Listen just before and after you've shut down your PC today. Was it nice to hear thoses fans fade away to silence? If no, then fine, stick to air cooling.

Beh, I use a Mac. My Moto chip needs no fan.

-Over time perhaps, the oil could act like a capacitor, build a potential difference, then short out your motherboard unless discharged.

I hadn't considered that. Sounds like fun.

Try mercury cooling.

 

[IRJack]

Originally posted by: imgod2u
The problem is, oil is not very easy to move around. Without circulation, pockets of the oil can get very hot. Although I have heard of people using mineral oil to cool their computers.

Moving the oil around is not really a problem - you can get an car oil pump from any auto parts store fairly cheap. Actually cheaper than some water pumps. The oil I'm using at the moment is a fairly expensive synthetic though - about $10/quart. It's used in a non-commericial system for cooling about a dozen or so circuit cards - it's not actively circulated. You could think of the cards as just sitting in a sealed bucket of the oil. About 2 gal. of the oil will hit around 50C after an hour or so of running the system. If I were to pump the oil and circulate it through a radiator, there would definitely be a temp drop. I'm estimating the power output of the cards at about 90W.

What I'm actually interested in is something like KnightMarauder suggested: Oil in place of water for my pc. I'm not satisfied that cooling with water is safe.

 

[FuriousBroccoli]

Safe in what regard? To you it's safe (you CAN'T hurt yourself on the low power going around a PC). To your components, well, you'd have to be extremely unlucky for a sealed TESTED watercooling system to first spring a leak, then leak all over your critical components, then to build enough conductivity to short out something before you noticed, THEN to actually have done permanent damage. Personally I think I'd rather have water leaking in my PC than oil!

 

[IRJack]

I've been mixing oil and electronics for a few years now and I would definitely feel safer with oil than water. I don't think twice about submersing a circuit card in oil and turning the power on. If someone came to me with a container of deionized water and said,"Here try this.", my response would be "No Thanks."

I assume you're talking about deionized water; tap water is a definite no-no. But your deionized water will disolve contaminants (ie dust) on a board much more readily than oil would, thereby becoming a conductor much easier. Oil's big problem is picking up debris that is already conductive, and depositing it somewhere it can do harm (ie between pins on an ic).

No it's not safe. But to me, oil is more safe than water.

But to the big question: Would oil be able to provide adequate cooling for today's (and tomorrow's) CPU's and GPU's (or VPU's for you ATI folks)? The big deal is that we need more cooling today than we did three years ago, and cpu's and graphics chips use more and more power each year. Process changes and die shrinks are not solving all of the power probs, so it becomes a cooling prob.

I'm just interested in how some people have tackled their cooling problems and throw out an idea for an alternative cooling method and beat the snot out of it.

 

[CTho9305]

scroll to the very bottom of the page. It is slow to load.

edit: big version

 

[everman]

I"ve seen an oil submersed system a while ago (probably over a year or more ago). The guy put the system in a cooler of mineral oil I believe and used a pump to bring the oil through a radiator where it was then deposited in a flow right over the cpu. He didn't run it for very long though as it was just for a fun experiment, but it did work. You would definitly get very low temps if you did use such a radiator cooling system.

 

[Tweakmeister]

my only concerns with oil would be if it seeped into the capacitors.

 

[kag]

I can't give you more info really, but I know it can be done. I remember seeing a picture on the front page of HardOCP of a oil submerged computer.

I don't remember what type of oil was used, but it looked more opact than those in the pictures above.

 

[ElFenix]

that was a good article.

 

[grant2]

But to the big question: Would oil be able to provide adequate cooling for today's (and tomorrow's) CPU's and GPU's (or VPU's for you ATI folks)? The big deal is that we need more cooling today than we did three years ago, and cpu's and graphics chips use more and more power each year.


4 years ago my computer had a regular heatsink & fan

1 year ago my new computer ... still had a regular heat sink & fan

the "we need more cooling" probably relates more to overclocking than new technology.

 

[SharkyTM]

Originally posted by: FuriousBroccoli
Water is a much better thermal conductor than almost all oils, and flows a lot more easily, so oil is a really bad choice for closed cooling system.

oil is more viscous... meaning it is MORE resistant to flow... it flows less easily, you need a massive pump to move oils, even thin oil, like Marvel Mystery Oil. I worked using oil as a non-compressible liquid to fill motor housings with when i was building ROVs (remotely operated vehicles). You would fill the motor housing with an oil, then seal it from the rest of the craft, and the water with a grease tube. Then, we would submerge the motors (on the ROV) in 100m of water...

If the oil was conductive, the motors would blow up -literally- if they were too thick, the motors would catch fire- literally- so, we had to find a thin, nonconductive liquid... we tried brake fluid... nasty stuff, would burn your hands... so we used MMO.

 

[0roo0roo]

if you used a decent sized tub to hold the oil, wouldn't natural circulation from heat currents and stuff be adequate for cooling?

 

[LordRaiden]

Huh, I've actually tried something like this and ended up nuking a number of cards and several motherboards plus like 2 PSU's and 2 HD's which I'm still wondering how I pulled that off.

You're right about the oil cooling. I never successfully pulled it off myself, but you are right about one thing. You need to caulk the living heck out of that ah heck, and caulk, and caulk, and caulk. Yet my problem was, no matter how much I caulked, I still had the problem of seepage, and consequently, oil quickly finds its way between all of the contacts and within 15 minutes your system is dead.

But if someone does come up with a working method for full system emertion oil cooling, I'm game to listen.

My biggest problem with full emertion oil cooling is that once the processor heats up the oil, then what? Your board will cook, and so will all your cards. Plus, you have to consider the volumn of oil you have to forcefully exchange and effectively cool. Unless you could create a huge temperature differential between the incoming oil and what was in the case and what was being taken out.

I'm not saying it's impossible. I'm just being realistic.

 

[PowerEngineer]

Originally posted by: 0roo0roo
if you used a decent sized tub to hold the oil, wouldn't natural circulation from heat currents and stuff be adequate for cooling?

That's the big question. Yes, there will be some natural (convection) circulation caused by the temperature difference between the oil that has had its temperature raised by the heat from your device and the cooler oil beyond, but how much? Let's assume your device is producing a steady amount of heat. It and an evelope of oil around it will continue to rise in temperature until the heat being produced is balanced by the heat convected away by the oil's circulation. Assuming the "oil beyond" is at room temperature, how much higher above room temperature does your device have to get before this balance with convective cooling is struck? I don't know the answer, but the higher viscosity of any liquid when compared to air would make we worry that benefit of higher specific heat of the oil would be more than offset by much slower natural convestion. The few examples I have seen all use pumps for forced rather than natural convection. Even then, you'd have to be sure that the pumped oil swept around all the "hot" components.

Just my two cents...(again)

 

[0roo0roo]

ok finally loaded the pic of the oil cooling system, seems he only uses mere inches of oil, no wnonder he needs radiators. i was thinking more along maybe a foot or two deep or even more! with a simple cpu fan reversal it would shoot hot oil straight up. not to mention an aluminum bin? might help transfer heat better, or is that dangerous somehow'?