Rubber degradation in different mineral oils

[Latango]

I'm looking at making a submerged PC, which is part for fun and part so that it is completely silent. What that basically means is the motherboard, plugged in expansion cards, and the power supply are all submerged in a dielectric liquid - generally oil.

However, a lot of the capacitors in the system have rubber seals at their bases, where they are soldered to the pcb's. Is there a mineral oil that doesn't attack rubber? Possibilities available to me are liquid paraffin (known as Mineral Oil in USA I think), FR3 from Cooper, and R-temp from Cooper. I know for a fact Diala B inhibited transformer oil attacks rubber, so that's not an option.

If anyone has any experience or knowledge, I'd appreciate a heads up. I don't mind experimentation, but it's for personal reasons so I can't write off any failures, which could be costly. I have emailed Cooper about R-temp and FR3, but it takes them a few weeks to get back to me usually. Wanted to check feasibility before that.

Thanks guys/gals

 

[Latexman]

What kind of rubber are the seals?

Good luck,
Latexman

 

[Latango]

From what I can find, it's high purity "gum rubber" ... does that sound like something that makes sense?

 

[quark]

Teflon and Viton are good, though viton is expensive. We have bad experience with silicon and EPDM when used for liquid paraffin. Ofcourse, our application is different.

 

[MiketheEngineer]

Why in the world are you doing this?? Even the PC board itself may come under attack??

 

[ports394]

Partly for fun people do it... but it's great for overclocking and higher performance because of the heat transfer properties. Fans and air are nice... but a cold liquid!

 

[Latango]

Mike, Ports394 summed it up quite well. Fun, and heat dissipation in a liquid is far more effective than in air.

Quark, Teflon and Viton would be great to make capacitors out of, however I'm not making the capacitors. They are already made.

I was just wondering if the chemical world knew of a liquid with good dielectric properties that didn't degrade rubber and didn't evaporate at 40-50 degrees C. Doesn't look good for my little experiment. There are companies that will sell you an oil immersion kit for your pc, but I am not sure on their longevity.

Thanks though. If I take the plunge and just try it with mineral oil I'll post a picture some day.

 

[Latexman]

I did a search on "Parker's_O-Ring_Handbook.pdf". Natural Gum Rubber is NOT satisfactory for many, many "oils", but there are a few it performs satisfactorily in:

Castor Oil
Silicone Oil
Ucon Lubricant 50-HB-100
Ucon Lubricant 50-HB-260
Ucon Lubricant 50-HB-5100
Ucon Lubricant 50-HB55
Ucon Lubricant 50-HB-660
Ucon Lubricant LB-1145
Ucon Lubricant LB-135
Ucon Lubricant LB-285
Ucon Lubricant LB-300X
Ucon Lubricant LB-625
Ucon Lubricant LB-65
Ucon Oil Heat Transfer Fluid 500 (Polyalkalene Glycol)
Ucon Oil LB-385
Ucon Oil LB-400X

You can download that reference and look for other fluids that are safe and convenient to you.

Good luck,
Latexman

 

[moltenmetal]

Start with the cap manufacturers: find out what rubber they're using. It might be nitrile, in which case you're safe with just about anything. It might also be EPDM or even thermoplastic elastomer, either of which will swell in hydrocarbon oils.

Most of them will be OK in paraffin oils until the oil is very hot- above 60 C. The degradation mode isn't corrosion or embrittlement- it's swelling, which will probably just make the static seals at the bottoms of the capacitors seal tighter than they did before. They're soldered to the board, so there really isn't anywhere for them to go, so there's no risk of them popping out etc.

 

[MiketheEngineer]

Why not use ice-water or Freon??

Now I do understand why you are asking - mostly I was just curious!!

 

[Latexman]

Gum rubber is good in water. It is good for some Freons and not others. Check the compatability first.

Won't you still have to remove the heat or will it just build temperature until heat losses of the enclosure equal the heat gain?

Good luck,
Latexman

 

[EdStainless]

1. Puget Systems will build you one of these.
2. Yes it works
3. You can use pure mineral oil, though we gave up because at low temps the oil clouds.
Thought about silicone, flourinert, and krytox, but they are too expensive.
We ended up with a very low viscosity PAO with no additives. We stripped all fans and heatsinks. Built a chiller from an old freezer. We could run the thing at about -25F. Overclocking old AMD Thunderbird CPUs.

We wanted to use water but we could not get a good ion exchange unit to keep the resistivity above 10Mega ohms. I have done this in a electric motor at high voltage.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[ports394]

Mike, ice water is conductive...you would short circuit electronics on the computer. If he's doing this for a home computer, he won't have a good way to ion exchange it, RO, etc.

Freon, I think goes a bit too far, because it creates water condensation, and then you risk short circuiting again. Same if you run freon through a heatsink as indirect cooling.

 

[Latango]

Ports: Apparently some people use oil baths for temporary condensation prevention during dry ice / lN2 cooling of components, but that's obviously not long term.

Latexman, thanks for that list. I have a few k-litres of silicone fluid here I didn't even think of - providing it's the same stuff that was popular in some european transformers pre-FR3 days. Of all the listed oils that would probably be my choice (doesn't smell, for one thing. Castor oil I'm not sure of). Thanks a bunch for the list.

Moltenmetal: It's the swelling that's caused problems in the past. Someone once tried this idea with Diala B oil, and it attacked the capacitors so viciously that they literally popped off the pcb, which destroyed the system. The 60 degree problem is my issue - still building the tank in my off time (3 kids + 50+ hr a week work + wife + training = slow process)

Mike: Port pretty much summed it up. Even with demineralised / distilled water, impurities, dust, heck even electrolysis I think, cause the water to be conductive enough in a pretty short time.

Latexman: I don't think it will build up enough heat. It's my home pc, and it doesn't stay on for more than a couple of hours at a time. After the kids and wife go to bed I only have so much time, still need to sleep a bit.

Ed: I did see Puget's systems, but they like to put them in fish tanks and have bubbles and fake fish and stuff in it. I also see a lot of home built units out there using plain roof and gutter silicone / bathroom silicone holding the tanks together. They don't last long before the silicone breaks down and the tanks leak. (Probably can't get clear polyurethane?) I'm not aiming for freezing temps after all, or a gimicky fishtank. Just to eliminate those damn annoying fan sounds.

Thanks guys, I'll start fabbing the tank up soon and try it out. How hard can it be?

 

[MikeHalloran]

Fluorinert is the only thing I'd trust for this.
Everything else mentioned will dissolve _something_.


Mike Halloran
Pembroke Pines, FL, USA

 

[Latango]

To be honest I DID look at Flourinert, but it would mean a hermetically sealed case, which is difficult with all the multicore cables that need to come out, and at $840 per US gallon here in Australia, it would be cheaper to replace the pc every year and use plain old mineral oil.

 

[MikeHalloran]

It was once a trade show staple to have a television (with tubes, of course) running normally in a transparent tank of Fluorinert. ... with an open top.

You only need a bucket big enough to hold the motherboard. The cables can come out through the surface. Just don't tip it over.

If it's a custom bucket, it can be shaped to fit the computer with just a little clearance, and equipped with cooling fins on the outer surface.

Then you just need a source for less than a gallon of Fluorinert.



Mike Halloran
Pembroke Pines, FL, USA

 

[Latango]

The locally available Flourinert has a boil point of 56 degrees. Without actually building the tank and running an experiment or 2, I'm not sure how the heat dissipation will go, thus Tmax (I'm sure it's calculable but there would be a lot of variables, and I'd only have to be wrong a bit to start boiling off what is more or less an ozone eating gas - at big cost, without sealing the tank perfectly). Also, cables have this annoying propensity to wick fluid up between the cores, resulting in little puddles of liquid in unexpected places.

Think I'm going to risk the Si oil. The worst that will happen is I have to upgrade my computer after it blows up - thank goodness for Acronis and external hard drives I guess.

 

[EdStainless]

Clean water is not conductive. You water cool the targets on x-ray tubes and the water insulates the 60kV.

Latango, In the end I bought a high performance PSU that will run over 90% eff at about 30% of load, the fan barely turns. Added a huge CPU heat sink and a GPU with passive cooling. I move air through the case with a couple of 140mm fans that will move a reasonable amount of air at low speed. Now the loudest part is the HDD (still waiting for SSD prices to drop).

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[ScottyUK]

Synthetic transformer oils, either silicone or something like Midel 7131 would be worth a look. Plenty medium-sized electrical overhaulers out there - some will be willing to supply it in fairly small quantities. Midel is a proprietary product and has full technical and MSDS information available, so you can have a look and decide whether it is a good match for your requirement.


----------------------------------

If we learn from our mistakes I'm getting a great education!