3M Novec submerged computer

That's actually very cool. Somewhat impractical in a full submersion application, but I could see it as the basis for a phase change system in a conventional water cooling loop. Set up correctly, you wouldn't need a pump.

Actually that idea reminds me of a little company called Alam Thermal (whose website is now dead, sadly. Here's the archive.org cache) that produced and sold a small prototype passive phase change cooler (they called it "Thermoloop"). It had an evaporator block, much like a water block, connected by flexible tubing to a standard refrigerator-style condenser. The loop was sealed and the working fluid was presumably water operating at a lowered pressure to facilitate the phase change process at a moderate operating temperature range. There was a rubber bulb in the loop that acted as a passive pump by expanding and contracting using only the changing pressure in the loop. Sort of like a heat pipe using flexible tubing and a passive pump instead of a wick. Pretty cool little device. Hmm... found a pic, and it looks like they may still sell them. IIRC, the price on the old website for a kit was about $100.

My thought was, a cooling set up like the Thermoloop, using Novec instead of water allowing a non-sealed loop at normal pressures. That would make possible building out a loop using standard WC parts, and the absence of a pump means it could be virtually silent.
 
that's really cool, I had previoulsy seen systems that used non-conductive liquid to submerge the board in for insulation purpose in extreme ocing but this puts a whole new spin on it!
 
It's not toxic unless it thermally decomposes (from putting out a fire.)
 
It would make cooling all kinds of things easier. Rather than constantly buying new HSF/water blocks/ram sinks/heat sinks for motherboard components you just fill a sealed case with a uniform liquid. No worries about breaking board/clips/bolts/screws. No worry about thermal paste application, air flow. You know that the liquid will touch almost ever part of the board due to gravity and the way liquids work.
 
Hmm... that is awesome...

They are using a TEC to re-condense the vapor, so that is the method of heat removal I suppose.

I wonder what is required at higher heat-loads? And what temperature differential is needed for that condenser?

I bet I could find some of those specs/answers on the 3M website... it'd be different for each of their fluids

I guess the entire top of the case could be a giant TEC-heatsink: multiple low-voltage TEC's in parallel, as seen in chiller arrays, set for a 15-20C temperature delta perhaps, and water-cooled, perhaps.

The problem with that: You could use that hardware to chill the water loop directly, and possibly achieve lower temps.

The liquid-Core delta seems impressive. I guess I'd have to analyze more detailed performance numbers to make an assessment.
 
Hmmm... I was going to make a thread about which oil was the most thermally conductive but I think I may go with 3M Novec instead (well with the heatsinks installed of course).

Hey, I do have the sudden desire to build an oil PC from scratch.
 
My only concern would be what it would do to higher TDP processors. This was only an i3 hitting those kinds of temps.
 
My only concern would be what it would do to higher TDP processors. This was only an i3 hitting those kinds of temps.

According to Corvalent, 3M have done tests to demonstrate it will be a while before TDP exceeds the capability of the fluid
Corvalent said:
3M did a successful test cooling 4000w, w/only 200ml of fluid. (same as ~30 i7-940's @ 100% load)

Some people have fish tank's in their offices, if I win the lottery, I want one of these /w quad sli/xfire in mine :D
 
This is indeed really cool.

i don't think "normal" people can just go to 3M and buy this stuff

I don't see why not, other than 3M or a reseller giving you a hard time being a consumer or requiring a certain amount to purchase there should be no reason not to sell it. It is not toxic or anything.
 
3M is an awesome company! I'm still trying to find something they don't make. Thanks for the information. Is this liquid similar to 3m's fluorinert?
 
Saw that video a few months ago... this is actually exactly my Phd research is related to (immersion cooling)... I'm pushing for my professor to let me build a system eventually, and actually put our research to test. It's going to be so fun if/when I do.

We actually ordered this fluid after seeing the video. Phase change cooling at above ambient is pretty awesome. :D
 
I wonder if you could use Novec in a closed loop pump-less setup...cpu block with a dual 120 rad/res combo up top...
 
I'd be somewhat concerned about the long term effects of running a setup like this. Its basically a fluorinated solvent so it might solvate some of the things in the plastic over time, like the plasticizer in cables and whatnot.
 
I am the engineer who built the systems in the 3M videos (GTX-480 machine was built by a customer) and you are correct, organic contamination can be a issue but not for its effect on the material containing the organic. Polyvinylchloride (PVC) wire insulation, for example, contains a lot of dioctylpthalate (DOP) plasticizer, a clear white oil. However, the PVC will insulate a function just fine with the DOP extracted. However, the DOP, once in the fluid, will be deposited on the high performance boiling surface and foul it. For this reason, we use carbon adsorbents to remove it in situ. You can learn more at: http://www.youtube.com/watch?v=Wn6baA0rKbI
 
Why does everyone do this stuff with no heatsink? They should put a big fanless heatsink on the CPU and it would work way better..

EDIT: It appears they are using some small heatsinks on there, good! Many setups like this in the past I have seen just running the cpu bare.
 
Why does everyone do this stuff with no heatsink? They should put a big fanless heatsink on the CPU and it would work way better..

EDIT: It appears they are using some small heatsinks on there, good! Many setups like this in the past I have seen just running the cpu bare.

I saw a video of a guy explaining this. He said that the small heatsink helps the boiling process and reduces temps. The CPU is sans heatsink and he said they did that just to show you could
 
There have been some questions about the performance and how we achieve it. The lids of the GPU and CPU are modified by soldering onto them a coupon of copper that has been modified through the application of a porous copper boiling enhancement coating. In the case of the GPU, that coupon has small pin fins on its surface but they would not have been necessary and they do not, in this case, enhance performance.

The resultant thermal performance is quantified through the case-to-fluid or junction-to-fluid thermal resistance which, in the case of the NVidia GF-100 GPUs, is about 0.076C/W. At 250W, the junction layer of those chips is 19C hotter than the fluid boiling point or 34+19=53C. These have a grease interface. The CPUs do better with solder as the interface. Their Rjf is 0.060C/W and they run about 100W so that their junctions are only 34+6=40C.

If you want to learn more, here is a webcast.
 
The resultant thermal performance is quantified through the case-to-fluid or junction-to-fluid thermal resistance which, in the case of the NVidia GF-100 GPUs, is about 0.076C/W. At 250W, the junction layer of those chips is 19C hotter than the fluid boiling point or 34+19=53C. These have a grease interface. The CPUs do better with solder as the interface. Their Rjf is 0.060C/W and they run about 100W so that their junctions are only 34+6=40C.

WOW! :eek:

I'm really surprised that a company hasn't tried to market one.
 
I am the engineer who built the systems in the 3M videos (GTX-480 machine was built by a customer) and you are correct, organic contamination can be a issue but not for its effect on the material containing the organic. Polyvinylchloride (PVC) wire insulation, for example, contains a lot of dioctylpthalate (DOP) plasticizer, a clear white oil. However, the PVC will insulate a function just fine with the DOP extracted. However, the DOP, once in the fluid, will be deposited on the high performance boiling surface and foul it. For this reason, we use carbon adsorbents to remove it in situ. You can learn more at: http://www.youtube.com/watch?v=Wn6baA0rKbI
thanks for the response. I hadn't thought about the fact that you might accumulate the plasticizers in places they shouldn't go, I was definitely thinking more about the results of plastic itself degrading. Interesting.
 
May seem counterintuitive, but a heatsink (HS) would make matters worse. In addition to the required thermal interface for HS attachment, there is the conduction resistance within the heatsink itself. In this case it would be higher than the boiling resistance. Remember, the lid of the processor is only 2-3C hotter than the fluid if the boiling enhancement is used.
 
Would it lower temps more if you took the lid off the processor and put the boiler plate on the cpu die?
 
WOW! :eek:

I'm really surprised that a company hasn't tried to market one.

Because this kind of research, while it has been more focused on high powered electronics for a decade now, hasn't done much for applicability yet. People are enhancing surfaces, making them perform better and better all the time... but some issues

1) Liquid aging - this is a documented and known effect
2) Surface oxidation (if using copper) - again a documented and well known effect - it helps remove more power but lowers your Twall to CHF (#4)
3) Durability of enhanced surfaces is still pretty unknown
4) Film boiling - CHF - if you are not careful about how much power you dissipate, pushing your chip too far via overclocking will eventually result in something called film boiling, and you will likely burn out your CPU. We use heaters in our experiments that are good up to 200C or so, but when this film boiling occurs, we see temperatures go from +15 Tsat to... +80 or more in just seconds (experiment will auto shut off when this happens). This could be lethal to a CPU or GPU.
5) ... and finally manufacturing. Do you really think companies will sell full sized desktops like this? NVDA's vapor chambers are a good start. :p

Why does everyone do this stuff with no heatsink? They should put a big fanless heatsink on the CPU and it would work way better..

Pin-fin effect. Large heatsinks won't fully boil. 5mm tall pins are about the limit of completely diminished returns. Looking at the video, the pin sizes seem to be in the right range for something like this.

This was the first video I've seen surface enhancement used, which was very impressive. I expect soon to see Ivy Bridges tested out because those can be de-lidded just like those GTX 480s were. But maybe the Ivy Bridge CPUs get a little bit cheaper (used that is). But again, if someone is going to do this and overclock the hell out of Ivy (I can see IB/SB easily getting into 5ghz and beyond with this kind of cooling), you absolutely have to know where CHF is with your setup, or you might lose the chip.
 
Last edited:
shut-up-money.jpg
 
I saw this the other day. Such a cool idea... I saw a different submerge PC that was put into a fish tank of a non-conductive oil. Why anyone would want to do that is beyond me, but still very cool :p
 
Design: the liquid boils at a fixed temp, so, if the liquid is designed to boil at, say 100F, where we see all the bubbles, all parts are kept at 100F at least externally where the devices is in contact with the liquid.

If the device puts out more heat energy than the liquid can remove (at the rate it evaporates, creates air bubbles, floats off and fresh liquid comes in), there are still risks of overheating, though almost impossible.

This is interesting, when I first saw it. But after giving it much thought, there are higher costs, and hazards to it.

If you keep the case sealed, and the active cooling fails or cools insufficiently, the pressure will build up, and your case will explode. Now we need a pressure fail safe mechanism to shut down the PC.

If the case is left open, the vapor will escape, and the liquid needs to be constantly topped up.
 

This is a very interesting idea :D. Cool idea using the change of phase to get rid of the heat and then recondensing it :D

Not sure if it has been asked, but could you use a similiar liquid to go below ambient? like if you condensed the fluid to below ambient (in an insulated case of course) as there would be no need to worry about condensation.
 
yeah, one of the earlier post mentioned that we should use the condensation plate to cool the entire case.
This idea might work, because as the bubble forms on the surface of the devices, it will condensate within the liquid, and the bubble should disappear before surfacing. BUT, condensation takes time, and for some bubbles, it may not have enough before hitting the surface and escapes.
 
Back
Top