SFF Thermal Solution: Mineral Oil?

[zovc]

Pretty much, as per the title... The idea came to me in some idle time, and I'm not really in a position to do anything with it. It'd be cool to see someone carry the torch, as I'm not sure I'll be building a PC any time soon.

So, mineral oil is non-conductive (or my understanding is that it's otherwise safe), so you can literally submerge your computer components in it. I typically see folks do this without even removing conventional air-based heatsinks, causing the fans to create turbulence and bubbles which definitely looks pretty cool. Pudget Systems has a pretty good page on mineral oil PCs, and I believe they sell (or have sold) a kit. LTT also did a video on mineral oil PCs, as one of Luke's projects.

What I'm really trying to get at here, though, is: Why not try to utilize this in the SFF space? Your case has to be filled with something, unless you're creating a vacuum... why are you creating a vacuum? Nevermind. My understanding tells me mineral oil conducts/transfers heat better than air, and can be pumped directly into a radiator without needing a bunch* of fancy tubing and new water blocks.

Any thoughts?

 

[thewizzard1]

It's messy, and hard to keep sealed. Many oil-cooled rigs are open, and leaky. It also has much more weight that air. Has next to no place in the SFF / Portable realm due to these downsides.

 

[zovc]

It's messy, and hard to keep sealed.

I agree with the principle of these issues, but they don't seem that hard to overcome. This is a community of enthusiasts! I think mineral oil having more weight than air means it's better suited to smaller enclosures than larger ones.

 

[Nightblade]

@zovc the problem with smaller enclosures is that it is even harder to spot any leaks. With less area to cover also means that smaller cases aren't or may not be as durable as a larger case. There lies the conundrum, how do you make a case small enough but big enough to contain all that mineral oil? Its also like previous posters have said, its messy.

 

[zovc]

how do you make a case small enough but big enough to contain all that mineral oil?

The premise of this question kind of confuses me. Like, how do you case small enough but big enough to contain all that [air] in it?

Maybe I haven't put enough thought into the concept, but these criticisms read to me like knee-jerk reactions. Mineral oil cases have been engineered and they do have shortcomings. But the entire point of designing an enclosure is to account for and compensate the traits of things you're trying to bring together.

I'm envisioning basically a solid container that you lower your components into, similar to the Pudget Systems case on the page I linked. There's no reason that particular design can't scale down with size, you may need to make certain alterations in terms of radiator size or how you accommodate a "fill port" if you want to significantly downsize. (Besides just changing from an EATX to an ITX footprint.)

The "container" approach is--effectively--leak-proof as long as the materials are intact and the case stays upright. (Linus, of course, has managed to damage an acrylic one.)

 

[Nightblade]

@zovc All I'm saying is that mineral oil is heavier than air, therefore you must put more thought into structural integrity than you would a case that relies on airflow to cool the system. You can't just make a bucket, stick some components in it, then dump mineral oil in it. More than likely you would end up with micro-fractures that would develop into either structural integrity loss or a leak.

 

[Biowarejak]

Mineral oil wicks up cables, expands as it warms, and must be pumped through a radiator to dump the heat. There's also a company patent-trolling mineral oil enclosures, or at least the one in the LTT series.

I've thought about it, but it seems rather difficult to really do in an effective manner.

 

[jØrd]

 

[kotproger]

And how about using this oil in liquid cooling circuits? Without immersion. It is clear that the characteristics lose to water, but there is no question of compatibility of metals in the circuit and in theory there is no need for maintenance - oil unlike water is inert.

 

[Biowarejak]

And how about using this oil in liquid cooling circuits? Without immersion. It is clear that the characteristics lose to water, but there is no question of compatibility of metals in the circuit and in theory there is no need for maintenance - oil unlike water is inert.

Once again, there is the expansion issue. Plus it can eat certain kinds of gaskets and tubing, which would be unfortunate at best. Maintenance is still required, because it will pick up residue and such from the blocks and radiators.

 

[EdZ]

If you're going for immersion-cooling, and already optimising for minimum volume, then phase-change fluids like Novec start to become viable (cost per volume usually rules them out) which would allow you to pack components in even more tightly, without the need for custom waterblocks. Previous discussion on that here.

 

[thewizzard1]

If you're going for immersion-cooling, and already optimising for minimum volume, then phase-change fluids like Novec start to become viable (cost per volume usually rules them out) which would allow you to pack components in even more tightly, without the need for custom waterblocks. Previous discussion on that here.

Fluorinert is another brand name.

BTW - I'm making an all-soldered copper acetone loop.

 

[jØrd]

BTW - I'm making an all-soldered copper acetone loop.

Is there a link to a worklog or the like for this?

 

[zovc]

If you're going for immersion-cooling, and already optimising for minimum volume, then phase-change fluids like Novec start to become viable (cost per volume usually rules them out) which would allow you to pack components in even more tightly, without the need for custom waterblocks. Previous discussion on that here.

How does this Novec stuff expand, compared to mineral oil? I've never heard of it.

 

[thewizzard1]

Is there a link to a worklog or the like for this?

Not yet, I just build and take pictures.

 

[jØrd]

Not yet, I just build and take pictures.

Let me rephrase. This seems like a really interesting project, do you have anything at all to share about it, be it pictures, links, whatever. Alternatively, do you have plans to share some or more information in the future?

 

[thewizzard1]

Let me rephrase. This seems like a really interesting project, do you have anything at all to share about it, be it pictures, links, whatever. Alternatively, do you have plans to share some or more information in the future?

Oh, sure. I have a backlog of works-in-progress. I have a watercooled B250M-STX, a COM Exp project, and an ITX i7-3840QM which I need to finish.... So much to do, so little time.

 

[Biowarejak]

An additional factor I hadn't considered is that in a closed loop, mineral oil might not flow well in the microchannels. I'm sure it would keep the blocks pretty but it would be less performant, and you might actually need an immersion pump to reduce the stress on that part.

 

[thewizzard1]

An additional factor I hadn't considered is that in a closed loop, mineral oil might not flow well in the microchannels. I'm sure it would keep the blocks pretty but it would be less performant, and you might actually need an immersion pump to reduce the stress on that part.

It's also rather thick for convection, what I've seen is it's actually poured over passive coolers, with 120MM fans for 'convection' cooling inside the oil. The pump / radiator merely takes and outputs from the tank, no direct attachment to GPU or CPU.

 

[EdZ]

How does this Novec stuff expand, compared to mineral oil? I've never heard of it.

Novec (and Fluorinert, and other nonconductive phase-change fluids) boil at low temperatures, there are varieties with boiling points from ~30°C to ~60°C that would be suitable for PC immersion cooling. Because they boil at that temperature, the CPU/GPU/etc will never heat up beyond that temperature while they re still immersed in fluid. The vapourised fluid is captured in the loop and passed through a radiator (air-air heat exchanger) where it condenses and flows back down into the tank. Or it can be actively recompressed to a fluid and the hot fluid sent through the radiator to cool it back down below its boiling point: this is more thermally efficient (the hotter something is, the easier it can pass energy to ambient air) which lets you use a smaller radiator, but you now have to add a compressor to the loop and those don;t scale down to well.