Submerged mineral oil PC - but EE what would say on it? Finally...

[Zucca]

You know what I'm talking about:



Maybe I will overclock my PC in a nasty oil bath... just because I want to see myself how it is.

After some google, I found a lot of feedback from IT/Gamer/PC Overclocking Guys/Kids.

It would be interesting to see what an EE (like me but I have no experience on this) i.e. the person I trust the most when it gets down to HW stuff, have to say about it...

Do you guys have any experience or suggestions?

Few inputs from my side I found, (everything to be confirmed)....

1) TIM paste on Heatsink will dissolve in Mineral Oil, replace it with thermal pads

https://hardforum.com/threads/thermal-compound-oil.1321446/

2) Some Elco Caps rubbed sealed have hard time since the rubber seal is eaten by the oil



to prevent that:



no way!!!

3) Submerged cables will lose flexibility and become hard, here the isolation plastic is not happy in oil. On top of that oil will creep to the top of the cable because of capillarity.

Interesting here, seems to work somehow...
https://www.pugetsystems.com/mineral-oil-pc.php

This is the oil I will use: Addinol WX15:

http://www.insist2.com/admin/upl/779016_wx_15_gb-en.pdf

The heat capacity should be about (Paraffin)  2.00–2.9 J g-1 K-1... which is OK, of course not like water...

[Howardlong]

While I understand it "works", the practical aspects would frustrate the hell out of me!

One further comment is that I would imagine that surface traces on boards designed for specific impedances will be directly affected by the permittivity of the oil, so you may find that under the oil some things won't work, or will negotiate back speed such as on PCIe traces, and in marginal cases something that works in air simply won't work under oil.

[dmills]

I don't see that case doing much for either interference or immunity either.....

Many plastics do not do well in mineral oils, so making your cables up out of teflon insulated might be a good idea (Also, watch the fishtank plastic, it may eventually deplasticise and become brittle, real mess potential there).

CPU heatink fins run the wrong way for convective cooling (And pull the cpu fan, it will not make an effective oil pump and just blocks the oil flow).
 
If you must use silicone gunk, make sure you get the electronics grade stuff that does not emit acids as it cures.

I would be really nervous about the PC DIY crowd pulling the guts out of a PC power supply like that, those things do not take prisoners, and the case is actually quite important to both safety and RFI screening.

Also, oil immersed electronics is a messy pain in the arse (The mucking stuff gets everywhere), been there commercially, damned if I am going there on a hobby job, I would almost rather deal with DI water. 

Also, remember that you still have to get the heat out of the oil eventually....

73 Dan.

[RGB255_0_0]

LinusTechTips has a few videos on this. The case eventually leaked. Also, due to not wanting to be sued, companies don't offer these any more so there isn't as much long-term testing or reliability done as there has been for watercooling.

They do look absolutely fantastic though as a showpiece.

[Codebird]

Coolant options, in sort:

Mineral oil: Cheap, works, damages a lot of plastics. Will degrade insulation over time, so don't immerse anything expensive in it.
Silicone oil: Bit more viscous, so takes more to keep circulating, but also less destructive. Costs a bit more.
Perflurocarbons: The coolent supreme. Safe on all components, low-viscosity, cools great. The cray ran on this stuff. But the price is... well, you can't afford it.

[Berni]

Thing is oil cooled PCs are ONLY for the cool factor and thats it.

Most kinds of oil don't have very good cooling properties. While yes they do have more heat capacity than air, but its still far from the heat capacity of water and being so thick means they don't flow easily enough. So in the end you don't really gain much cooling performance compared to air, but you gain a lot of pain in the process. As you said things degrade in oil, tweaking components is a pain as everything is messy, if a component dies you can't warenty it back etc....

If you want overclocking performance then good water cooling blocks are the way to go. But its still so expensive that it only makes sense to do it on top of the range hardware. Otherwise the price difference is better spent simply getting faster components rather than overclocking cheap ones using expensive watercooling.

I don't knock anyone for building one, i think they look really cool too. Just don't do it exclusively for better performance as you will be disappointed.

[tszaboo]

So a BIG cpu cooler has some 30, very thin sheet metal part, 120x120mm, giving a surface area about 800.000 mm² area.
The aquarium has 13.8 x 12 x 7.5 inches (whatever that unit is) dimension, giving 460.000 mm². The thing is ultimately cooled by air. And there is a video card, etc in your system.

So brilliant idea. Also, having 6 gallon (again, whatever that unit is) of 88 Celsius hot mineral oil on your desk, above your reproduction parts, sounds like a good way to get a Darwin award.

[RGB255_0_0]

Also, having 6 gallon (again, whatever that unit is) of 88 Celsius hot mineral oil on your desk, above your reproduction parts, sounds like a good way to get a Darwin award.

Sure, if you don't pump it to radiators.

[Zucca]

Silicone oil: Bit more viscous, so takes more to keep circulating, but also less destructive. Costs a bit more.

Very interesting....

Just a little  research:

Mineral Oil, viscosity @20C  35,3 cSt   about 3,45€/Liter  Specific Heat @25C  2.0-2.9  J g-1 K-1
Silicon Oil, viscosity  @20C  350 cSt    about 4,50€/Liter  Specific Heat @25C   1.46?? J g-1 K-1
Silicon Oil, viscosity  @20C  100 cSt    about 8€/Liter        Specific Heat @25C  1.5 J g-1 K-1
Silicon Oil, viscosity  @20C   20 cSt     about 11€/Liter      Specific Heat @25C  1.6 J g-1 K-1

ah very important, the silicon oil  smell a little bit but I read it is not dramatic...

Very interesting:

http://www.silitech.ch/upload/complement_info_fournisseur_d/32.pdf

c Effects on rubbers
v Oils with viscosities of less than or equal
to 10 mm2
/s have a detrimental effect on
rubber immersed for prolonged periods.
v Materials containing very little or no plasticizers
that are compatible with Rhodorsil® Oils 47
are not affected by immersion.
Example: neoprene, butyl, natural rubber,
fluorinated rubber.
v Materials containing plasticizers can be
affected by immersion in a silicone oil.
The effect will vary according to the oil viscosity,
the material composition, the immersion
temperature.
The main effect is a loss of weight and volume,
as well as an increase in hardness due to
extraction of the plasticizer by the silicone oil.
These effects can only be seen after prolonged
immersion at temperature and do not prohibit
the use of the silicone oils for certain
CO applications, such as demolding, release, etc.

Maybe the low viscosity is the real enemy for rubber/plastic?

it looks like yes:

https://www.shinetsusilicone-global.com/products/type/oil/detail/about/index2.shtml#fushokul

Silicone fluid has no adverse effect on metals, nor on most other substances. However, it may reduce the volume and weight of some rubber and plastic compounds due to extraction of the plasticizers when subjected to high temperatures. This tendency is especially strong for low-viscosity fluids. Particular care must be exercised when silicone fluid comes into contact with rubber sealing materials.

https://www.shinetsusilicone-global.com/products/type/oil/detail/about/p2401.html

https://www.shinetsusilicone-global.com/products/type/oil/detail/about/p2402.html

Anyway it looks like silicon oil has less heat capacity then the mineral one... so mmmm...

[Zucca]

Most kinds of oil don't have very good cooling properties. While yes they do have more heat capacity than air, but its still far from the heat capacity of water and being so thick means they don't flow easily enough.

Look at the data, (J g-1 K-1):
water         4
mineral oil  2-2,9,
silicon oil  1,5
air              1

but do not forget that there are a lot more molecules of oil touching the heat sink than air... and this is not taken in account by looking at the heat capacity number.... so this will also improve drammatically the heat taken away...

correct me please, this is just what I got from he web... not at school.

[dmills]

note that those specific heat numbers are per gram, you need a lot fewer litres/second of oil for a given delta T then you do air, because the oils are far more dense.
The difference is far more then the specific heat numbers on their own would imply.

I would not bother for a computer, but am thinking about a water loop for an RF power amplifier, the power density of modern LDMOS sand makes cooling painful (1,500W in a few square cm).

73 Dan.

[CatalinaWOW]

This double ee would say forget it.  Not because of component damage or any of those things.  It will probably work fine for quite a while.  But the cool factor is small, the trouble factor is big.  It is now heavy, has to be right side up, is a royal PITA to service, and doesn't buy much if any on performance.

If you want cool, set up a CO2 tank to periodically spray dry ice on the machine, or set up a LN2 spritzer.  (be very careful if you do this.  It isn't that hard to kill yourself by suffocation.  Frostbite also sucks, and too much spritzing will kill your computer.)

[FrankD]

Don't use skydroll or break fluid.

[Berni]

Yes those heat capacity figures make air seam better than it actually is since a given volume of air at sea level has very little mass, but air does have one large advantage, its viscosity is very low. This is best demonstrated by compressed air plumbing, the tiniest of gaps result in quite a sizable air leak while if you put water at presshure in the same piping it would result in a very slow drip or no leak at all. A lot of gases even when liquefied at low temperature or high pressure maintain there very low viscosity. Liquid helium is even famous for having an incredibly low viscosity. But even water flows a lot easier than most oils.

A low viscosity not only means that its easier to push the same amount of fluid trough the heatsink using the same fan but it also makes the fluid not stick to the fin surface as much. Only the fluid that is making contact is useful so if the same fluid sticks around on the surface while the fluid above it moves then your thermal transfer is not very good. The way PC watercooling combats this is by making the fin gap inside the waterblock very narrow and use a decent amount of pressure to force liquid trough them. This makes for a high flow velocity trough the restricted path so it can drag the sticking fluid along as well as not give it much room to flow anywhere but right next to the fin surface. This also allows for a tiny heatsink size so less copper is needed to spread the heat out to a larger area (copper also has thermal resistance, even if its small)

Maybe you could get great cooling performance if you use good low thermal resistance heatsinks with high power fans in oil. Those high power server/industrial fans make hell of a racket in air but would likely run silently in oil. That way you overcome the viscosity issue with raw power, get the benefit of a smaller heatsink to reduce the path for the heat to travel and still have it silent.

But in the end of the day i think its very hard to beat a properly designed watercooling setup, if you want to go more extreme than that you can always use phase change refrigeration to cool your PC components below ambient, or even deep freeze them for huge overclocking, tho i don't think even an air conditioner compressor would be powerful enough to cool a gaming PC down deep negative and keep it there all day. Also you would need to carefully control it and regulate it as freezing CPUs too cold makes them crash and unable to boot. That is ignoring the practical challenges of needing a outdoor heat exchanger to get rid of the kilowatts of heat, keeping moisture from condensing on your components or the long term reliability concerns of running computer hardware at there very limits continuously.

[james_s]

I've used mineral oil to submerge HV transformers and the stuff is a real pain. It gets *everywhere*, it's almost impossible to prevent all leaks. Anything it gets on is then virtually impossible to get adhesive or sealant to bond to. If you use wires, especially stranded wire in it the oil wicks up under the insulation. I can't imagine what possible advantage there would be to immerse the whole PC. Air cooling works just fine and can even be made very quiet. For those who just have to be different, water cooling with water blocks on specific components is viable.

[R005T3r]

Why on earth would you want to submerge the pc on oil?

[DimitriP]

Why on earth would you want to submerge the pc on oil?

So you can make a youtube video about it, and ask viewers to comment , like and subscribe.
After it leaks,  you can make a video about that and ask viewers to comment , like and subscribe.

[R005T3r]

right....

[avrishuvorlaz]

Some folk ought to find something more useful to do with their obvious surplus of spare time... wow... 

[avrishuvorlaz]

LinusTechTips has a few videos on this. The case eventually leaked. Also, due to not wanting to be sued, companies don't offer these any more so there isn't as much long-term testing or reliability done as there has been for watercooling.

They do look absolutely fantastic though as a showpiece.

I would have thought a more elegant and efficient way to solve this "problem" (that no one except those who can never stop fiddling around inside their PC - CAUSING more problems that never existed apart from in their head - like an obsessive mechanic "boy racer", seems to have) would be to get one's nose down to the drafting table and come up with a custom made airflow and heatsink solution, find a used CNC machine and built it from scratch - a bit like that amazing guy "Tesla500" does with his "Chronos" camera - THAT level of from-scratch engineering is something to revere, not some dude who shoves a load of oil into a tank, dunks a PC in and expects praise.

[Bud]

I have a PC with a CPU liquid cooler , i can say it works great, very quiet and efficient, the CPU never gets over 50C. I've been having it since 2012, for 5 years or something, no signs of leaks and the tubes remain nice and soft.

[Zucca]

Thanks for all the Infos... I probably go for water cooling solution then,

I have a ASUS Sabertooth 990FX R2.0, and I will buy replace the stock heat sinks with those:

North bridge:
https://shop.anfi-tec.de/ucdnano.html

South Bridge:
https://shop.anfi-tec.de/ucdsb.html

VRM:
https://shop.anfi-tec.de/pwm068.html

Feel free to tell me I'm crazy... 

[Berni]

By the time you put everything together it still won't be cheap but it should have amazing cooling performance if done right.

Oh and if you go this far the graphics card should really be watercooled too. The GPU burns significantly more power than the CPU when working hard (Unless you are overclocking your CPU to hell and back or running one of those ridiculously hungry AMDs).

[Codebird]

I can't imagine what possible advantage there would be to immerse the whole PC.

Records, mostly. You can overclock on water cooling, but that only does spot cooling on your processor and GPU. As you push the hardware harder in your quest for the ultimate in performance you start to see overheating of the buck converters, memory, then all sorts of glue chips you never thought about before. If you are into extreme overclocking as a hobby and want the bragging rights that come with an eight-gigahertz eight-core monstrosity, you have to look into extreme measures, and immersion cooling is one way to do it. Along with phase change cooling.

I don't know about today, but as of a few years back the highest speed record was held by a Pentium 4... immersed in some liquid I can't recall, with dry ice floating in that to bring the temperature down, and with a special CPU cooler that allowed liquid nitrogen to be just poured over it. It'd only run so long as you kept the coolant pouring in, but it was the fastest processor in the world. That's an engineering achievement to feel proud of.

[RGB255_0_0]