• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Fluorinert or equivilent for SUBMERGING my RIG

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

BlueBall

Member
Joined
Feb 1, 2003
Location
The Hammer, Ontario
I am looking for a liquid in which to submerge my entire system. I need to decide which medium to use. I have found several options ... oil ... Flourinert and FluidXP. If anyone has any suggestions, I'm all eyes.

I have read a few articles from a couple years ago, but nothing current.

I don't want to go broke buying the medium, but if one product is superior to the others, I will bite the bullet and use that.

I won't be using LN2 to cool the medium, I was hoping to use air or water, so gelling up at -150C is not a concern.
 
theres one problem with this though, the system that gets submerged will never be 100% free of the medium ever again - in the past AFAIK, everyone used older systems that they wouldn't mind if they screwed them up

there was one guy who submerged everything in vegetable oil or mineral oil or something like that... i really don't remember too well

there was also a guy in the classifieds selling a container of fluorinert a little while ago
 
Is this just for ****s and giggles? And are you just looking to burn money somewhere?

Never saw a submersion product that didn't fit both of those categories...

I dont think fluidxp stays non-conductive. Might be another liquid but I remember something about it stripping ions away from whatever it touches and eventually becoming conductive.

Since subersion projects are only for a limited time and never for 24/7 setups (unless you are making one of those mainframe 1980 boxes that were submerged, but in that case you'd be using the 3m stuff) fluidxp would probably work for a day or two.

Now are you sure you know what you are doing? The line about you wanting to air or water cool the submersion fluid makes me raise an eyebrow...

Air cooling is pretty much out of the question. Im not going to dig it up but there was some dude with a VERY clean submersion setup. I remember it looked like a normal case with a side window with a completely submerged inside mobo. only problem was in order to air cool it he had to use some 1kw+ TEC and a heatsink as tall as the case to keep an old crappy P3 at decent temps.

Once again... All the submersion projects Ive seen were temporary. They'd dump ice or dry ice and styrofoam peanuts into a cooler of the submersion fluid while it was being circulated.

I guess you could dump a radiator inside a cooler and have it in a loop with another radiator thats being cooled by a fan else where. But that sure seems like a lot of thermal interfaces... Would be a giant waste of energy.

But hey as I said you must be doing this for ****s and giggles.
 
Flourinert, no question. You could toy with solvent but I can imagine you will have problems with components under a solvent.

Your gonna have to have some kind of pumping to keep things cool. I dont think its gonna be worth it for you though, your gonna need plumbing in there to keep things cool, non-moving liquid sucks for cooling as much as non-moving air and a pump and tubbing is more complicated than a fan.

I guess we need to know more about this system you have planned.
 
ls7corvete said:
I guess we need to know more about this system you have planned.


I am going to submerge a S939 Gigabyte SLI mobo and cheap video card to start with... In a clear acrylic case horizontally.

I am still going to use my typical home-made waterblocks for everything. CPU / GPU / Mosfets / Chipset. My idea is to pump the medium from a resevoir below the mobo (mobo is suspended in the medium) to a radiator, then to all of the blocks ... from there the medium will need to circulate AROUND to BELOW the mobo to reach the resevoir again (cooling the entire mobo surface and keeping the medium moving).

I had thought about using an A/C unit to chill the radiator in summer and use a valving system to use an exterior radiator in winter. I am not decided on this issue.

The NSA use Chilled water to cool their super computers submerged in $400/gal Fluorinert.

I also read about short term experiments using distilled water and additives. I'm not interested in short term.
 
Last edited:
BlueBall said:
...
I am still going to use my typical home-made waterblocks for everything. CPU / GPU / Mosfets / Chipset. ...


wth?

You are going to have waterblocks in a loop with your submerged setup?

You realize that defeats the purpose, right? If you were having the fluid at subzero then yeah maybe but otherwise thats stupid.
 
If you're thinking of using fluorinert you might read this thread.

I've listed all the good and bad parts of using the stuff I could think of and others chimmed in with what I didn't think of.

I'll say this, if you are thinking of using it do somemore research. Fluorinert isn't like anything you've come into contact with before unless you already work with it.
Best of the Good:
It's inert and nonconductive.
Worst of the bad: -Price
This stuff is expensive. Pound per cubic foot it weighs close to the same as concrete, and it's normally sold by the pound.
You'll need fully metal pipes with metal crush washers in order to use it with out refilling it monthly. It will permeate pretty much all organic and hydrocarbon based tubing, this includes even the Tygons inwhich are rated for it.
You'll want a fully metal centrifugal pump as you don't know what type of plastics or adhesives this stuff might deteriorate. Yes, I know both I and 3M say it's inert, but for the most part this means it's nonflammible and only explodes when in contact a couple of other compounds inwhich are also completely man made. However this doesn't mean that it wont eat some plastics and polymers.

Just some thoughts.

Edited for typo
 
Last edited:
Try mineral oil - it's cheap. Unless you are dumping some form of cooler in there, all you need is a pump and a radiator. Heatsinks are advised as well as fans, they spin really slowly.
 
karl pell said:
Ouch! Not only harsh, but incorrect ... the plan, as I understand it is to cool the fluid ... assuming he gets the pressure/flow right to the waterblocks, their heat dissipates in the larger volume of fluid that is cooling the board and that volume is subsequently cooled by radiator.


The heat from the water blocks dissipating into the rest of the fluid is the whole "dumb" part. Just like any loop the water is going to reach an equilibrium. The liquids temp is going to be a certain number of degrees above ambient. No matter what radiator he uses this is true. Think about it. In a normal water cooling setup the water in the loop is hotter than the mother board except where he has waterblocks.

If you have waterblocks on everything that creates heat, why spread it to the rest of the components that dont?

As I said it would be worth while if the fluid temp was below zero or even just below ambient. If you measure the temp of the motherboard thats not right by the VRMs or CPU etc it will always be a tad bit higher than ambient. So I could see it a tad bit benificial.
 
Some other thoughts:

If you were to do a setup based on submersion I'm not sure you'd even want to use normal waterblocks. You'd probably just have the circulation pump output over whatever normal air HSF's You'll probably want to use HSF's that take advantage of the center the hot spot with fins directly on the base and fan them out more then they originally were. You'll want a fairly high GPM pump (600+ ish). Also if you use a thicker fluid (mineral oil) you'll probably want a pump with a higher head pressure.
A larger pump will make much less noise if submerged in the liquid.

Now, heres the fun part. For a permanent solution you're going to need to take into account the FULL wattage output of the CPU, RAM, motherboard, and graphics card when looking at pumps and radiators. Now this at first thought may not seem all that much different then what most people are already doing but it really is. Consider all the regulating devices on the board like the MOSFETs. Also you must consider that much of the heated liquid will not make it straight into the radiator after being heated and instead will circulate through out the case capturing more heat. This is the same as what normally happens in a case with air. This is why NB temps will very (sometimes greatly) with proc load. If you're looking for exceptional temps you might want one of the new quad 120mm radiators or bigger.

Yes, the fluid will get warmer then ambient but you have more thermal capasitance to work with. Look at the way many motherboards are designed today. Yes, the air hitting the MOSFET's and RAM has been heated by the proc, but the air is still cooler then the surface of these items and still cools. This is also the same concept that allows the heated liquid from the proc to cool the NB in many standard watercooling systems.

You will have a greater thermal capacitance to work with but, if you're looking to get lower temps you'll want to have a good (planned) circulation path for the liquid. If you use a thinner liquid like fluorinert you can use standard air fans to help out. The added thermal capacitance should make up for the extra heat the fan will generate and keep it alive in an environment it was never designed for. I say this because I have tested fans under normal water for extended periods of time to test this. I cannot stay the same for a thicker liquid and although you might try it there’s no guarantees.

You'll want to think of what you're going to do with your case and get cable extensions for pretty much every thing. Consider this, you start overclocking and it finally tops out. You get so far and then it locks. A restart shows the dreaded black screen telling you something in there doesn't want to play anymore. The one thing that was forgotten was the jumper for clearing CMOS and it's now under half a cubic foot of either very expensive or very messy liquid.
You'll want to have extensions to and insure you have long enough cables to keep the floppy, HDD, CD/DVD ROMS, and PSU out of the liquid and with the example above you'll want to extend all jumpers for Pwr, reset, and clear CMOS at the very least. I'd recommend setting it up to where all the lines and extensions exit the case and end at a socket, so that when/if something dies you can replace it without opening the submersion section. This could make for easy upgrades later.
Although you could submerge the PSU, I wouldn't recommend it. You'll be adding heat to the liquid pretty much needlessly and also have to do extra work insuring it too has direct circulation.
No, your HDD is not sealed. It has an air filter in it and there is atleast one hole that allows air to go from outside directly in. If you think I'm wrong look it up or email your HDD's manufacture before you say it.
Not submerging the other drives is common sense I guess. No one wants to have a "salivating" drive.

Lastly, remember everything that moves can and will die given the time. This could be 30 minutes or 3 years+, but alittle thought before implementation will save you time in the long run.

I was once considering doing the submersion thing and got allot of it lined up, but in the end it was go nuts with it with time and money or make the new server I needed. The server idea won and I dropped it, but the above were the things I figured I needed to take into account to make everything operate smoothly over time.

Sorry for the length, hopefully you'll find some of the info posted useful

Good luck du,
YEL
 
I think I have the solution...

OK.

1. Mini Fridge http://www.minicoolers.co.uk/products/comparison.htm

Find a way to butcher this thing so that it cools the fluid. They are cheap and would cool the fluid.

2. Use all "heat pipe" style coolers. Ones with fins i.e.: http://overclockers.com/articles1263/

[edit]3. Pump within the fluid like the ones in fish tanks... Do a waterfall thing or something. This would be so much fun to build. [/edit]

The tank or enclosure is something I'm still thinking about.

The fridge may also be an idea in itself... Don't know...
 
YouEatLard said:
If you're thinking of using fluorinert you might read this thread.

You'll need fully metal pipes with metal crush washers in order to use it with out refilling it monthly. It will permeate pretty much all organic and hydrocarbon based tubing, this includes even the Tygons inwhich are rated for it.
You'll want a fully metal centrifugal pump as you don't know what type of plastics or adhesives this stuff might deteriorate. Yes, I know both I and 3M say it's inert, but for the most part this means it's nonflammible and only explodes when in contact a couple of other compounds inwhich are also completely man made. However this doesn't mean that it wont eat some plastics and polymers.

I've read that it is totally non-corrosive, non-conductive, non-toxic, etc. Isn't an oxygenated version used for liquid breathing experiments. It is also refered to as Dry Water.

Here's two other mediums I found ...
http://hardforum.com/showthread.php...light=Submerged
http://www.midel.com/English/first.htm
 
Just get the liquid cool enough so that it's better then room temp. There is so much surface area in contact with a cooler that this isn't really a concern. Although, I don't think I'd attempt this by abolishing all forms of heat sinks. Processors are a lot hotter these days. I think I'd still use a heat-sink of some form.

YouEatLard said:
You will have a greater thermal capacitance to work with but, if you're looking to get lower temps you'll want to have a good (planned) circulation path for the liquid. If you use a thinner liquid like fluorinert you can use standard air fans to help out. The added thermal capacitance should make up for the extra heat the fan will generate and keep it alive in an environment it was never designed for. I say this because I have tested fans under normal water for extended periods of time to test this. I cannot stay the same for a thicker liquid and although you might try it there’s no guarantees.

I was thinking he can use that mini fridge idea or something like it. Get the liquid really cool. Possibly a submersible pump would do the trick for this project.

YouEatLard said:
You'll want to think of what you're going to do with your case and get cable extensions for pretty much every thing. Consider this, you start overclocking and it finally tops out. You get so far and then it locks. A restart shows the dreaded black screen telling you something in there doesn't want to play anymore. The one thing that was forgotten was the jumper for clearing CMOS and it's now under half a cubic foot of either very expensive or very messy liquid.

Route a switch to the back of your case for this.. I'd make the switch one that is hard to "flick". nub :p

YouEatLard said:
You'll want to have extensions to and insure you have long enough cables to keep the floppy, HDD, CD/DVD ROMS, and PSU out of the liquid and with the example above you'll want to extend all jumpers for Pwr, reset, and clear CMOS at the very least. I'd recommend setting it up to where all the lines and extensions exit the case and end at a socket, so that when/if something dies you can replace it without opening the submersion section. This could make for easy upgrades later.
There are now modular PSUs making this task much easier in so many ways. With both ends being something which you can manipulate, I think this task just got a lot easier.

Long 36" round cables... And special order SATA cables.

YouEatLard said:
Although you could submerge the PSU, I wouldn't recommend it. You'll be adding heat to the liquid pretty much needlessly and also have to do extra work insuring it too has direct circulation.
I agree... I think it's an unnecessary risk. Also, this just adds so much more heat into the picture.
YouEatLard said:
No, your HDD is not sealed. It has an air filter in it and there is atleast one hole that allows air to go from outside directly in. If you think I'm wrong look it up or email your HDD's manufacture before you say it.
Not submerging the other drives is common sense I guess. No one wants to have a "salivating" drive.
I personally wouldn't submerge any of the above mentioned either. I'd just stick to circuitry and the parts which you are most concerned with cooling for stability purposes.

Good ideas...
 
Nice additions ProperMethodz.

I also looked into those mini fridges. You'd need an army of them to cool a dynamic load like a computer. I never found the wattage of the peltiers that they run but, from what I've read it'd probably under 50 watts even for the big ones.
Then I looked into the smaller compressor based keg sized fridges. I never worked the numbers for my self but from what I've read they *might* be able to take a normal processor, but no where near the active wattage of a full computer. The normal refrig is designed to keep fairly static materials cool in an insulated enclosure. A full sized fridge or freezer compressor might be needed for a dynamic applications with a high demand load such as this.

BlueBall:
Yeah, I've read the same and then some with the added pain/bonus of working with it on a daily basis.

Although it is non-corrosive, non-conductive, and non-toxic it is still mildly destructive to skin (it makes skin very dry and flaky also bad for eyes which I've taken someone to the hospital for) and many organic and hydrocarbon materials will deteriorate in it over time.
Now with this in mind, is the formation of a chemical bond occuring (like styrofoam and gasoline)? To my knowledge no, not at all. The best I can tell the fluorinert may be acting as a catalyst, much the same as the sun or regular air flow will effect the same materials over time. The difference between air, the sun, and fluorinert is that this will occur at a much faster rate with fluorinert. This is what I ment by "eating".

This with the added bonus of loosing a measurable amount per week due to permeation through organic and hydrocarbon tubing (plexiglass too) means that you'd want a fully sealed all metal system (copper tubing is easy to work with and everything else aluminum).

I haven't personally tested it with copper. I have only read the MSDS's and other information put out by 3M and other manufacturers.
As for aluminum it's pretty awesome. I've inspected much of the tubing and units our fluorinert runs through and I have yet to spot any corrosion. This may not sound like all that big of a deal, but this tubing and other units have had this stuff pumping through it for 30+ years while enduring temperatures from -45°C to 55°C. I'm not talking about gradual temperature change either. I'm talking about doing that full range (from 55° to -45° and back to 55°) in as little as 8 hours.
 
BlueBall said:
Isn't an oxygenated version used for liquid breathing experiments. It is also refered to as Dry Water.

Sorry, I just read that last part - it's alittle late.

Although, I haven't read anything to say either way my guess would be no. If more then a pint of this stuff spilled in an enclosed area our procedures require us to leave the area for an amount of time to insure the vapors have cleared. Although this might be just incase as long term human exposure may not have been fully tested, from what I've seen it do to skin would tell me this stuff would be very hard on the human respiratory system during evaporation and would be likely to dry much of it out.

No, it doesn't burn the skin. It has more of the effect of dumping alcohol over the skin a few times which will make it dry and flaky. Basically when it evaporates much of the water in the surface goes with it. Main difference is that this stuff evaporates at several times the rate of alcohol so the effect is worse. Last I checked the human race really enjoys having their internal organs (lungs and the rest of the respiratory system included) hydrated.

So, all that being said, I highly doubt we intentionally put any variant of fluorinert in our bodies.
 
This is my previous OverMOD project in Flash (1.7mb)
http://www.ocforums.com/showpost.php?p=3036775&postcount=427

I made every component I could think of. I have no choice but to out-do myself on this submerged rig.

Besides the evapouration part, I think Fluorinert is the best. I might try to find a Chemist or even a 3M rep. to find out more info.

Next week I'm going to buy the Acrylic for the setup. I can't wait to start modding again.
 
karl pell said:
We'll agree that the rig and heat exchanger are operated at a constant condition that will, over a period of time, yield an equilibrium temperature for the medium, when measured at some point. I would argue, however, that this is a dynamic equilibrium for the system ... different equilibrium temperatures at different points in the system, i.e. the equilibrium temperature at the waterblocks will be higher than the equilibrium temperature at the cooling radiator outlet. The dynamic equilibrium for the system is attained when all sources of heat input are balanced by sources of heat removal.
... The water temperature difference from any component in a normal watercooling setup is never more than 0.1c... So yes you are correct its different, but it doesnt matter. His system would only be different if the flow rate is extremely low. Sort of like the thermosyphons.


But you do not have waterblocks on everything that creates heat (see this , for example) and that was the point of the original exercise that inspired blueball. Inexperienced users ...

Ok lets nitpick some more. Did he have blocks on absolutely everything that creates heat(anything that consumes electricity)? No. Does he have blocks on everything that reacts to lower temps? Yes. NB, CPU, GPU, VRMs... My NIC isnt going to act any different from 50c to 30c. Neither is my SATA controller or USB... They will work or they wont.
The only thing that you could argue is ram. If he's running some bh5 at 4v or something else equally crazy then sure, that would benefit some.

That bigbruin site brings up a good point. Look at all the pics. The majority of the motherboard is in the blue color temp spectrum with little spots of red. Think its better to have that or the entire board to be closer to green as well as the same red spots?

Only way he'd have better peformance than just putting blocks on the important components and using a normal H2O setup is if he has an extemely large direct flow and a giant radiator. Like a suburban's radiator style.
Even then I wouldn't be surprised if he was only on par or below par a normal h2o setup.
 
karl pell said:
... The goal was not to improve operations at higher volts/temps, but to enable operations at insane volts/temps.

Back to my initial thought, Submerging is only going to accomplish this if you do sub ambient liquid. Even if you focus the flow extremely well (just having one general flow over the entire setup won't cut it) and have a gigantic radiator you aren't going to lower the temps of anything very much. But you will definitely raise the temps of other components...
 
Back