View Full Version : NonConductive Fluid
NeoMoses
08-07-01, 10:44 PM
I've got an Idea, but don't know if it'll work. I want to completely immerse my mobo/processor/all expansion cards in a non-conductive fluid, then stick it in a fridge/freezer. Many oils would work (viscosity right, etc...) but I'm not sure if they will degrade the boards or cause any other unseen probems.
If anyone's tried this, or has some junk mobos/pci/agp/isa cards laying around and would like to do some testing, let me know.
It doesn't make a whole lot of difference, is expensive, and difficult. See:
http://www.octools.com/articles/submersion/submersion.html
CrystalMethod
08-07-01, 11:31 PM
3M makes a product called "Florinert". You might wanna looking to it. Only problem is that it's really expensive. I thought about doing the same thing, but I can't afford to invest the $$$ right now. The stuff sells for about $68 Canadian per pound, and weighs almost twice as much as water. You'll need about 2 gallons of the stuff, and there's no real reason to cool it exsessively, seeing as it flows everywhere, including the backside of all the components (providing more heat transfer). I'd give it a shot myself, but I can't swing the cash right now. if your brave enough, a friend of mine brought up the idea of using "acetone" which is also electricaly inert, but I have a problem with this seeing as how it has a tendency to eat through most plastics (like your AGP/PCI/ISA slots). If you can get your hands on the florinert, lemme know how things work out.
CrystalMethod
08-07-01, 11:54 PM
Originally posted by sfa ok
It doesn't make a whole lot of difference, is expensive, and difficult. See:
http://www.octools.com/articles/submersion/submersion.html
I don't know if I'd cool it with LN2, cause that's way extreme. Dry ice and alchohol would do for any Oc'ing purposes. Even just circulating the stuff through a rad/ intercooler would do for those of us that have a grip on reality...
Ridenow
08-08-01, 12:07 AM
I have seen it done with mineral oil. I don't know a lot about it, except it worked.
Crazy Jayhawk
08-08-01, 12:15 AM
Too extreme for me. :p
Is it just the insides of the computer that get submerged? If it's the whole thing it'll be kinda difficult to put in CDs or stuff like that. :p
NeoMoses
08-08-01, 07:48 PM
Well, my Idea was to just submerge the mobo and expansion cards. All the drives and some USB ports would be on the outside. The theory (as CrystalMethod stated already) is that you would get more surface area for heat transfer. I'd probably machine my own EXTREMELY thin copper waterblock and circulate the same fluid through it. (wouldn't have to worry about leakage :p ) Anyway, this is just me dreaming. Thanks for the info, guys!
CrystalMethod
08-09-01, 12:20 AM
You'd probably get better heat transfer without the water block, and just circulating the fluid, through a rad of some sort. Also Any fans would be a no-no, imersed in the stuff. They'ed try to draw too much current tring to spin in a liquid, and probably fry your MB, and power supply. Don't forget that the liquid COMPLETLEY immerses the CPU. drawing away much more heat than just a HSF, which only touches the die of the chip.
Flourinert is 500 dolllars per gallon... 2 gallons is 1000 bucks just to cool it. And you would have to build things to accomodate submerging your parts... too expensive. My dad's company deals with it, but he said he couldnt get me any because of the outrageous price. Oh well
Matt
I've seen this done and it had tons of problems. They guy had his mainboard, CPU, and expansion cards submersed in foam cooler filled with mineral oil. I pump then pumped the mineral oil over the coils to a window type air conditioner. It worked for about 2 hours, then the A/C unit completely froze up. He said it was a huge mess. It's totally impractical.
Blue Jester_2112
08-09-01, 01:32 PM
Well, the AC might have been overkill there, couldn't you just use a normal radiator? Obviously the oil isn't going to dissapate heat as fast as water, but couldn't you just use a couple of good size heater cores to compensate?
What kind of effect will mineral oil have on the components? Is it going to start eating away at the boards and stuff? It might not be prectical but neither are alot of things people around here do. =]
It could be a fun project, just need to work the bugs out first.
Originally posted by Blue Jester_2112
Well, the AC might have been overkill there, couldn't you just use a normal radiator? Obviously the oil isn't going to dissapate heat as fast as water, but couldn't you just use a couple of good size heater cores to compensate?
What kind of effect will mineral oil have on the components? Is it going to start eating away at the boards and stuff? It might not be prectical but neither are alot of things people around here do. =]
It could be a fun project, just need to work the bugs out first.
The oil itself won't hurt the components. Cooling the oil is the challenge. Pumping it through a radiator will take a massive pump and I doubt air will be able to cool the oil fast enough. That's why this guy used an AC coil.
His biggest complaint was the mess. His animals in the house kept drinking the spilled oil and getting the runs all over his house YUCK!
My biggest problem with this idea is that once you submerge your components in this oil, what are you gonna do if it doesn't work. Mineral oil is not easy to clean off of all these components if you change your mind.
Blue Jester_2112
08-09-01, 02:03 PM
Good point, I think the first thing you'd need to do is get a solid pump that can do the job, probably a bit pricey, but it would just take some looking.
About the heat, what about attaching a couple heavy duty pelts to the heater cores?
As for the cleaning the components, that would pretty much be out. I'd definately test it out with some cheap older components like a S7 board and a cheap k6-2. That would probably be a good tst too considering how hot those old chip use to get. Then get some new parts to put in there after it done right.
I would never do this to my main rig. I would probably make it look real nice and set it up in my living room as a conversation piece. A nice piece of compu-art. =]
I think I heard somewhere of someone submerging their PS in some kind of liqud. They seemed to have problems with the caps swelling then exploding from the liquid seeping in. If this would REALLY happen, I would coat all the caps with a thin layer of some spray paint, or something to seal the liquid out. How much heat can caps generate?
JigPu
Crash893
08-09-01, 04:47 PM
why submirge it in the first place
why not get a water bater and put it on the mother board that and have a in out.
you could add something there for condinsation but why go with the high tech wen you can use something like a ziplock bag or something like that
or you can do the opposit you could put the mother board in a bag and sink it in just plain water or something like that
seall it where teh cords come out.
ill draw a picture later.
Highly purified water is very non-conductive, believe it or not. I designed a water coolant purification system for a huge 250,000 volt vaccum tube (not a typo) used in a radar. The tube was the size of a water heater and produced a 6 megawatt pulse, but that is another story..... To quote Clint Eastwood 'It will blow your head clean off' ;-)
Anywho, it was important to keep the conductivity of the water down below a certail level to keep the tube from shorting out. The water flowed through the tube assembly to cool it.
Soooo, a water immersion system incorporating the right kind of purification cartridge could work. Corrosion could be a problem, but the purified water helps with that too.
Hmmm..........
Blue Jester_2112
08-09-01, 06:06 PM
Tell me more about this water purification. What exactly are we talking about? What kind of setup would be necessary? This is starting to get me thinking and interested in trying this in the near future. =]
Boomerang
08-09-01, 07:46 PM
The aforementioned guy with the foam cooler and window A/C used mineral oil. Apparently it worked, but as stated before, was a HUGE mess.
As for the option of using super pure water, it's pretty infeasible for the average overclocker. You're looking at something like $200 just to filter the stuff. Plus, you gotta have special pumps that won't pollute it in any way, and all kinds of stuff. very complex, and VERY expensive. I would use the mineral oil first.
A third (and yet unmentioned) option is pure antifreeze. I did an experiment the other day by immersing a multimeter in antifreeze with no casing whatsoever for a couple of days to see if it would still work. It was fine, and there were no corrosion problems. Apparently, the conductivity of antifreeze is very close to zero. One thing to keep in mind though, is that this was Zerex brand antifreeze. I don't know if any other brand would work the same way. YMMV. Hope this is helpful.
Blue Jester_2112
08-09-01, 10:05 PM
I'm liking this...I have alot of free time at school, this might be fun to try. Keep this thread going with ideas and info. Who has more info on antifreeze?
Revelation
08-09-01, 11:59 PM
This sounds really cool(no pun intended)...
So lets see, you fill something with non-conductive fluid, then you place the computer parts in the liquid, then you cool the liquid, seems simple enough... :)
Is there any way to use liquid nitrogen to cool the fluid WITHOUT the LN2 evaporating? My guess is no.
What other ways are there to cool the liquid?
Thelemac
08-10-01, 12:05 AM
A phase change system, some TECs, something along those lines...
BrianCapps
08-10-01, 01:22 AM
Why not just throw your comp in a lake? that would keep it cool. But seriously... If you could figure out some way to do it, I have the perfect case in mind: Get an aquarium, some fake plastic fish and plants, and an air pump. Wa la! Aquaman's computer!
Have fun
Man, Id like to lug that thing to a lan party..
Why not use real fish? Your water should be warm enough and some air isn't bad :) I just wonder if a goldfish would survive in distilled water... You probably would have to refresh the water alot.
Someone care to try? I'd be happy to read about that here :D
Blue Jester_2112
08-10-01, 07:05 AM
Originally posted by BrianCapps
Why not just throw your comp in a lake? that would keep it cool. But seriously... If you could figure out some way to do it, I have the perfect case in mind: Get an aquarium, some fake plastic fish and plants, and an air pump. Wa la! Aquaman's computer!
Have fun
Actually, that was exactly what I was thinking. I was thinking I could take a 10 gallon aquarium and modify it to suit my needs. =]
TEC cooling and a couple of heater cores seems like the way to go for now. What about the liquid? Mineral oil seems to be the most likely candidate right now. Someone mentioned anti-freeze, does anyone know anything about the conductive properties of it?
Any other candidates for a coolant?
Originally posted by Boomerang
The aforementioned guy with the foam cooler and window A/C used mineral oil. Apparently it worked, but as stated before, was a HUGE mess.
As for the option of using super pure water, it's pretty infeasible for the average overclocker. You're looking at something like $200 just to filter the stuff. Plus, you gotta have special pumps that won't pollute it in any way, and all kinds of stuff. very complex, and VERY expensive. I would use the mineral oil first.
A third (and yet unmentioned) option is pure antifreeze. I did an experiment the other day by immersing a multimeter in antifreeze with no casing whatsoever for a couple of days to see if it would still work. It was fine, and there were no corrosion problems. Apparently, the conductivity of antifreeze is very close to zero. One thing to keep in mind though, is that this was Zerex brand antifreeze. I don't know if any other brand would work the same way. YMMV. Hope this is helpful.
Keep in mind antifreeze becomes acidic as it ages. I check the antifreeze in my car with a multimeter. Once it registers a certain value (can't think of it now) it's time to replace it.
Crash893
08-10-01, 09:08 AM
my idea.
im sure you could also use a water block for the cpu and have the water hoses go up threw the same water tight seal as the other cables.
just a rough draft let me know what you think.
also i would probably recomend putting some of thouse dehumity packs in the bag with the mother board.
As for the option of using super pure water, it's pretty infeasible for the average overclocker. You're looking at something like $200 just to filter the stuff. Plus, you gotta have special pumps that won't pollute it in any way, and all kinds of stuff. very complex, and VERY expensive. I would use the mineral oil first.
Actually it is not that bad. For example, special pumps are not required. The system I worked on had a standard cast brass impellor pump. Most of the stuff we used to build it was inexpensive. The purification cartridge can be an standard off-the-shelf item, that is what we used, though it was a high-temp version. Low temp versions can be bought at the hardware store. It would be best to implement a conductivity meter though to know when the cartridge has been used up.
But, antifreeze is also an interesting idea. It has corrosion inhibitors in it, and is not too expensive. As far as the acidity problem it could just me monitored with a cheapo pH meter and changed when it was time. The computer environment would be pretty mild compared to a car and thus it may not break down into acids as quickly.
Revelation
08-10-01, 06:32 PM
:bump:
Revelation
08-11-01, 01:57 PM
Are there any non-conductive liquids that wont freeze/turn to gellitine when it's as cold as N2? :D
*spazzed*
08-11-01, 03:01 PM
I HAVE to try the antifreeze cooling :D My copper/aluminum/silver HSF wasn't very good, so i have to make something that works. Anti-freeze sounds good:cool:
Crash893
08-12-01, 12:19 PM
isnt it a bit thick
Check out www.evanscooling.com.
Propylene glycol is a "food grade" antifreeze (read your twinkie wrapper) that auto manufactures are increasingly using. Car companies mix it with water but this company has mixed it with other "proprietary" (non aqueous) components. It is completely liquid to -50F after which becomes less fluid but will never solidify (so say they). It is reasonably cheap ($25/ gallon) and non-conductive however i have no idea what the "propietary components" might eat or be absorbed into on the MB. Small company though so they might be willing to discuss it with you.
The thermal conductivity is decent but not even close to water.
CrystalMethod
02-11-02, 05:47 PM
It's not the anti-freeze on it's own that's conductive, it the water you mix with it that is conductive. Run the mixture through a system with copper or aluminum, and you get and electrical charge. Like AMDGuy said, that how he checks his coolant, by the electrical charge the mixture produces. Oh yeah, If I remeber correctly, it's 0.18 V/DC when you should change it, but you can let it go until you get to 0.23 V/DC.
Originally posted by CrystalMethod
It's not the anti-freeze on it's own that's conductive, it the water you mix with it that is conductive. Run the mixture through a system with copper or aluminum, and you get and electrical charge. Like AMDGuy said, that how he checks his coolant, by the electrical charge the mixture produces. Oh yeah, If I remeber correctly, it's 0.18 V/DC when you should change it, but you can let it go until you get to 0.23 V/DC.
I don't know if the above was directed at me or not but the point to my post was that the evens cooling system does not dilute the propylene glycol with water at all ut rather other _secret_ ingredients. It is supposed to be completely non-corrosive. check out the web site.
jimmytp
02-12-02, 07:49 PM
That evans coolant is a good find, but take a look at the viscosity at 10° C on this page (http://www.evanscooling.com./main3.htm). NPG+@10° C = 58, H2O@10° C = 1.2. It's over 48 times more viscous than water at 10° C. If you were just going to submerge your board and use a radiator to keep it near room temp, it might work. If you were going to try and get sub-z, I think it would start to get a little thick. It would be nice if they had some data for lower temps or a chart to see the curve, but I couldn't find any. I emailed Evans about more data, so we'll see what they have to say if I hear back from them. It's a good conductor of heat, but how well could it be pumped? This stuff is meant to perform best in really hot racing engines, not cold applications. I think I'm gonna order some for my '90 Camry. For some info on viscosity, and a nice chart of relative viscosities, look here (http://xtronics.com/reference/viscosity.htm). Thanks to rockstarbob who posted some helpful data here (http://www.speakeasy.org/~language/pictures/superchart.gif). I think the NPG+ would work, but I just don't know how well pumps handle olive oil. Hmmmm.... I'll get back to you on that.
As an alternative, most alcohol is non-conductive, right? I'll stop by the chem library on my way home to get a reference book of liquid conductivities. It's not corrosive either, is it? It can't be worse than acetone. Well, maybe to your liver, but... no I don't drink acetone. A bath of pure methanol would be very viscous at low temps. It might not have the same thermal capacities of water, over 50% the heat capacity and just under 50% the thermal conductance of water, but the board would be submerged with good convection and coolant flow, so how important is that? It would also clean up a lot easier than mineral oil. As long as it was pure, it would just evaporate with no residue. Would that work?
jimmytp
02-13-02, 01:12 AM
Alright, I did the test on the oil. I used Best Yet Vegetable Oil, with a viscosity I estimate around 80 cP. It looked to be the thinnest at the store. Real objective, I know. I wasn't going to use my good pump to do this, so I scrounged around for an old 40 gph I had in the basement. It might not have the torque that it needs to pump such a heavy fluid, so a better pump might yield better results. I submerged the pump, let it run for a while (it took some time for the oil flow to become steady), then I timed how long it took to fill a 16 fl. oz. measuring cup. The water went so fast, I did it a second time using a larger container.
Water = 23 seconds to fill 1 liter, or around 34 fl. oz.
Veggie= 68 seconds to fill 16 fl. oz.
I ran each of those a second time and got almost identical times.
Water 41.5 gph @ 3 in.
Veggie 6.6 gph @ 3 in.
That's a pretty serious decrease in flow rate. 84% to be exact, but like I said, it was a pretty weak pump.
Also, I have already received an email from Doug Evans, from Evans Cooling. He guessed I was interested in electronic applications and said that the NPG+ coolant was completely non-conductive and non-corrosive. He also said he would have the tech dept. send some more in depth tech specs in a day or two, so I will post those when I get them. And I need to eat my foot, 'cause there are all kinds of charts on the NGP+ specs page (http://www.evanscooling.com/html/npgPls2.htm) that I linked to previously. The viscosity increases logarithmically with linear decreases in temp. A viscosity of 500 cP at -17 C would never work for sub-z.
As for alcohol, the book in the library was gone, but I found a reference to the conductivity of ethanol, or the lack thereof. It was listed as non-conductive, but had no other information beside that. Has anyone tried alcohol (not the drinking kind) or know that it won't work as a submersing coolant?
WOW thats a crazy idea I have never heard of that. I dont think I would be brave enough to submerge my computer parts in fluid... If you do it though you WILL HAVE to post pics
Pure glycol or antifreeze actually has no "anti-freezing" properties until it is mixed with water so I'd be careful about trying to go sub-zero with it in it's pure state.
UserName
02-14-02, 12:05 AM
LN2
Weel normal water is not so conductive!! the only real problem is the corosion... and from my experience normal water and some prestone in it can prevent it from coroding. I've never try a full system on in water ... but I've done it with a cpu with really god result (as son as the pump don't die)
can found my pic here :D
http://jfpayeur.com in (overcloak) (Voir les sytemes) second one
it in freech for now but I only intend you to see the pictures
the k6-2cpu/imerge system did work for me for a month ... and sudenly my pump fried .... so i don't know the long therm effect
oups ... you have to chose a mirror befor selecting the second one ...
mirror 2 is not working realy often
direct link will be simpler
http://colba.net/~magimagi/tijeff/over/2/1/index.htm
http://colba.net/~magimagi/tijeff/over/2/2/index.htm
http://colba.net/~magimagi/tijeff/over/2/3/index.htm
jimmytp
02-14-02, 01:21 AM
I just got a chance to check it out. Nice site. Nice work. The set up I would like to use for project immersion involves a 500 K6-2 as the test subject. I notice the 350 didn't mind taking a little bath. What kind of pump are you using? What is the AF/H2O mix ratio? And you mentioned corrosion, can you go a little more in depth? Also, I didn't notice any kind of cooling, are you just letting the water pour into the resevoir, or did I miss a radiator or bong? Again, nice site.
sorry for the delay
Well this is a really old sytem ... one of my first one in fact. I was only working with a big reservoir of water. I was talking about corosion because water tend to corod things ... specially metal. And the cpu is having some metal contact on the top .... so it's not really good.
If you have other question contact me on icq
5520014
Captain Slug
04-17-02, 09:17 PM
I've been trying to research precision instrument oil for this purpose. It's thin, non-conductive, and non-corrosive. The only thing I havn't been able to find out is it's thermal conductivity. It's extremely cheap compared to flourinert or anti-freeze but it's dang hard to get solid information on...
Malpine Walis
04-17-02, 09:50 PM
Originally posted by Quaky
Why not use real fish? Your water should be warm enough and some air isn't bad :) I just wonder if a goldfish would survive in distilled water... You probably would have to refresh the water alot.
Someone care to try? I'd be happy to read about that here :D
Actually, that would be a problem. I would be surprised if the air did not raise the conductivity of the water (O2 being a negative ion). And if that does not do it, the fish is going to be contributing it's own ionic compunds to the water. Think about what fish do in water and you get the idea.:D
Crash893
04-18-02, 12:35 AM
to have the watter pure enogh to keep it not conductive you basicly have to seal it off from any outside contamonation
i heard somewhere that if you stuck your finger in that was contamonation ( electrolites and amino acids) to cuase errors due to conducction.
so a fish living in it would definelty screw it up
but it would be coold to have a fish live in a res tank
well, just covering up the metal contat with some epoxy or anything really tough will do the job. After this plain water will be really nice, my system did run with some for a month with no problem. For the pum i was using a garden spring submesible water pump.
But this is definitly one the the most risky way to cold down a cpu, one of the most efficient to (without going under the room temp).
Go to see tnis old topic alive :P
You can go to see my new system :P
http://colba.net/~magimagi/tijeff/over/
My Dad is a Chemical Engineer I brought this up to him and the first cheap non-conductive liquid he could think of was gasoline. so anyone that serious could do that. And about the Aquaman PC goldfish excreet stuff that could be conductive after some time. though there are those fake electric fish that you have to put themin distilled water so that they work and you could use some of those!!
Aquaman78
01-25-09, 04:20 AM
Recipe:
1: Submerge it all in a vat of vegetable oil
2: Crank up the voltages and frequencies as high as they will go
3: Place french fries in frying basket
4: Run P95 while playing Crysis for 30 mins
5: Drop basket of fries in oil and cook until golden brown
6: Enjoy!
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.