Overclock.net banner

Best coolant for ultra low temperatures?

17K views 28 replies 7 participants last post by  Puck 
#1 ·
Im new here but not exactly new to computer modding,
So hi im Kamacho, nice to meet you lovely people. Ive been nosing around all day, and I have a grand plan that I can't find anyone that has done it. It will get my CPU temperatures down to as low as -90 Celsius with zero condensation, no insulation required
thumb.gif
. Only problem is car anti freeze will freeze at -30 roughly, water much much less than that. I have found some space age technology from 3M that would cost me upward of $2000 to fill my reservoir and I would have to replace it every three years but there has to be a better solution. Realistically if I go through with this I can theoretically cool to -90 without having to worry about condensation but if I could get anything below -20 I would be happy with this project. It seems all the companies that make coolant for PC's are selling out of dyed RO water for premium prices and the only information I can dig up is how well the water looks after 3 years of use. Im beginning to think I should invest in some RO systems.. Does anybody have any input on the best ultra low temperature coolant?

Kamacho
 
See less See more
1
#4 ·
#5 ·
Im just realizing that is a problem, I read some bad TEC chip information earlier that claimed alot lower temps than are possible. The peltiers I am looking at is the 545W one that is on ebay right now. Aparently it is good for -30 if the hot side is properly cooled (Sounds more realistic than what I read earlier). I plan on running at least three of these to keep my temperatures very low. Here is the idea here. Im thinking about building a custom case with a closed loop nitrogen gas blanketed pod in the bottom. The MoBo, liquid cooled video cards, SSDs, and the liquid cooled CPU would all be in the inert/dry gas environment. The PSU and the DVD drive would be in a separate compartment. The blanket gas environment would get hot so to counteract this I plan to run the peltier cooled liquid through a single or two large radiators that are also going to be in the dry gas environment with triple fans on each one pulling the heated air through the peltier cooled radiators. I want to put a peltier between the coolant reservoir and the radiators, from there the coolant will be pumped through another peltier cooler setup, than to the CPU, than through another Peltier and on to the Video card than back to the reservoir.

So thats it, probably not going to reach -50, but it would be possible without condensation this way.
 
#8 ·
Quote:
Originally Posted by kamacho970 View Post

Just realized your the vacuum chamber guy, I read your project earlier and I gotta say, very awesome idea.
well with TEC's, antifreeze will most likely be fine for the temps there capable of.
I have not finished my final build yet, but I have done a test build, unfortunately the ice chest on that build collapsed, after I had drilled several holes (12) into one side of it.

if you get a chance to read OCaddicts build log, he has lots of good info on how TEC's function. his rig has been running without condensation for 3 years now, and his setup is the reason I started mine.
http://www.overclock.net/t/1390376/peltier-tec-hybrid-water-cooling

it has taken me four years to build up to this kind of build it is expensive, and time consuming. I have been working on it a few hours almost every day to get it complete and it is still going to take me 6 months to actually get it all done.

this project seems to be about the same size as mine, so don't get discouraged, when I started I gave myself 6 month to a year to finish the project.

life will get in the way several times before something like this will get completed, but if you peck away at it every chance you get it will get done.

so ask questions along the way, on any issue that may arise, and there is most likely someone that knows a way to deal with it.

good luck, and I will definitely be waiting for the build log on this project.
 
#9 ·
Keep in mind antifreeze gets pretty thick for computer pumps below 0.

I ran an Ethanol + distilled mix, but you have to be careful with the parts used since it can be corrosive. It stays nice and thin though, and wont lower your flow rate as temps drop.
 
#10 ·
Yea I know it will be a big project. I whipped up a dirty cost analysis already and it's defiantly going to cost a pretty penny without unforeseeable expenses. Thanks for the encouragment. I actually read his build as well, between his and yours, I got my inspiration. Than my mind keeps running wild with new ideas.s I still can't help but dream of incorporating a small home built AC unit into the case as well to blow ultra cool air through the heat sinks or radiators cooling off the hot side of the peltier's. I like OC's setup with the heat sinks and the fans but it seems like radiators would cool better. Have you thought about 3/4" to 1" thick expanded acrylic (plexiglass) to solve your case collapse woes? its pretty cheap (sorta)..
biggrin.gif
 
#11 ·
yes, I even paid mountain mods for a design sketch, for making a see through vacuum chamber, they wanted $3000 to $5000 in development costs, because they had never attempted something like this, but there price was too high, but even for me to get enough 3" thick plexiglass to build one, would be a $1000 for material, and another $1000 for a shop to cut, drill, and tap all the sides for me
biggrin.gif

this will never happen, but I can always dream.
 
#13 ·
Forget using TEC, its just added complexity and power draw that you don't need if your really going for sub-zero. TECs are 100% efficient at half Qmax. So if your using a 400w TEC it will move 200w of heat at 200w of power. Increase past 200w and your efficiency falls below 1:1. Keep in mind that you have to cool BOTH the heat load being transferred as well as the power being drawn by the TEC. So for the above example you will need 400w of cooling ability.

Unless you have access to a commercial gas refinery, closed loop liquid nitrogen is not feasible and definitely not safe for an indoor environment.

Your best bet for running 24/7 is to use cascade phase change. Basically the idea is you chain two or more AC units together using heat plate exchangers. The bottom level one (running the coolest) has a phase block which attaches to your component. This can run 24/7 sub-zero but it won't be cheap on power and will be the size of a small desk. Plus you will need to mount the radiator for the top level AC unit outside or else you'll essentially be running a giant space heater in your computer room. With this you need to cover your mobo in some sort of insulator to prevent condensation.

If your interested in the best temps/power/size ratio, then your best bet would be TECs with a water loop. You'll still have a space heater but your temps will be marginally better than WC and you'll probably only need 2-3 rads depending on how many components you cool.
 
#14 ·
I beg to differ about the closed loop Nitrogen. I work in the oil industry, not here but in new mexico they use closed loop nitrogen systems to hold down the VOC's (volatile organic compounds) that permeate off the flowback and oil tanks. They use this because it is a cheap solution. I think it would be feasible to run a Gaseous (key word) closed loop nitrogen system not for the cooling aspect but for the dry atmosphere aspect. If the atmosphere is 100% dried nitrogen than there will be no condensation because there is no water molecules in the air to condensate. Thus allowing suepr cooling of any and all components without fear of them becoming rusted or shorted. No ice, no water, nothing like that. If you know about nitrogen you know it is not a toxic gas, it is very inert, non flammable, and from what I understand it is an insulator rather than a conductor (no static, or at least less static). It is heavier than air so it sits on the bottom of whatever container you put it in. If I end up building this I will be using silicon to seal around the mouse ports/usb ports etc than I will fill the case with nitrogen from the bottom purging the natural atmosphere out of a purge valve on top. I will close the purge valve and have a small psi meter installed into the top of the case to close the pressure in for a night or two. 2-3lbs will be enough 5lbs will be overkill so I will go for 5 to make sure everything is sealed. If the case holds the pressure, it will surely hold the nitrogen at normal atmospheric pressure. The thing that makes nitrogen dangerous is the fact that it displaces oxygen, so for the oxygen to be displaced in my room that I will have this system in, it would have to be enough nitrogen gas to settle on the floor of the room and fill the room to head level. This would be a lot more than would be required to fill a computer case. Of course once the case is purged, the nitrogen bottle will have to be stored outside making sure the valves are closed before unhooking it. Than it's just a matter of recirculating the nitrogen in the case through cooled radiators (or even an ac condenser) inside of the case with the heated radiator fluid leaving the case and dumping its heat outside the case somewhere before being recirculated.

Anyways, so 1" Plexiglas wouldn't work? I would think 1" thick Plexiglas would hold a pretty good vacuum, how much vacuum are you putting on this thing? It was a thought because it's easy to work with. I've cut it myself with a circular saw before, I know it can be drilled with a hand drill, and I would imagine tapping it would be a simple matter with a tap and dye kit purchased from a hardware/auto store and maybe some wax or lube of some sort. Sheets can be purchased online, I did a rough estimate for making my case out of it and for a 17" long by 12" wide by 17" high case it would cost less than $300 for the outside glass, inside glass (to hold the SSD's etc) and the .75" square extruded acrylic rods to attach the sides with. Drill some holes, seal everything with a bead or two of silicon before squeezing it together with some good screws and presto!. Im a DIY kind of person myself haha.
 
#15 ·
If you intend to keep the atmosphere dry you don't need to use nitrogen then. If its airtight you can just run a dryer on the internal air.
As or pressurization you'd have to build very custom case to hold any pressure. If you assume your case is 2 feet tall by 2 feet long then

24 in x 24 in = 576 sq in. x 3 psi = 1728lbs of force on the side panel at 3psi of pressure
If you were to scale it up to holding back a full vacuum of 14.7psi then your looking at almost 8,500lbs.

Sure you can do it in a compressor tank, but those are round and not square. What I'm trying to say with all of this is that there are a lot better and easier ways to get sub-zero temperatures while preventing condensation. If your really looking for an innovative way to prevent condensation I would suggest making an oil-submerged, phase change hybrid. Submerge the whole computer in mineral oil, then crank the phase change up as high as you like because the oil will prevent air from reaching cold components. A lot easier than trying to make a computer in a pressure vessel and still epic enough to definitely make the front page banner if you build it.
 
#16 ·
no, I would need the 3" because of insulation, at the 1" mark, the outside would sweat from the cold on the inside, and I did not want to need to cover the build with another insulator, that would have defeated the reason to make it see through.
biggrin.gif
 
#19 ·
Well I know that, ive seen them already. I just assume keep all water out of components as well as mineral oil, and mostly not fond of doing stuff that has already been done.. Im not really fond of the idea of dielectric grease hence why I would like to find a way to dry everything and keep it dry without using it. As far as size and or power limitations, I would like that case to be no taller than 24", no wider than 12". Power is still under debate in my mind, its easily possible to go wild and have a 1000W computer with 4 545W chips, and everything that entails with a AC unit built in that sucks as much or more on a different breaker but I would like to have something alittle more practical, exact numbers I havn't planned yet. Ideas and this issue has been rolling around in my mind on and off though.
 
#20 ·
Phase change definitely isn't going to fit in a normal computer case. Especially since the compressor itself is going to be bulky. You know the compressor that powers your fridge? Phase change is basically using one of those, only bigger. And cascade uses multiple loops. Click below for an example as to the size of the thing:

http://www.overclock.net/t/1469603/three-stage-cascade-build

As for trying to keep it in traditional case your only real bet is use TECs. You'll want to have separate power supplies for the TECs. Look up meanwell PSUs for that. Cool the hot side of the TECs with water cooling. Unfortunately sub-zero really isn't feasible for a traditionally sized PC.
 
#21 ·
Im building my own case.. Thats the whole point of this, there are small ac compressors that fit in the palm of your hand that ive already looked at. I think Phase change is out for me also because I can't see the temps getting as low as I want to get. Ive thought about using phase changed air or air conditioning whichever you prefer to blow through the radiators or heat sinks of the tec coolers, but thats more space and electricity. I watched a linus video where he used a commercially made phase change case (it costs like $1200) and he was getting down into the single digits but nothing negative. I would like to be very negative in my temperature and I think a series of TECs if done right can take it there.
 
#22 ·
Quote:
Originally Posted by Puck View Post

Keep in mind antifreeze gets pretty thick for computer pumps below 0.

I ran an Ethanol + distilled mix, but you have to be careful with the parts used since it can be corrosive. It stays nice and thin though, and wont lower your flow rate as temps drop.
well 6 hours in, and it's looking like the anticongelante antifreeze is the answer to this issue, the 60% antifreeze, 40% water, is the only one not thinking up at all, it seems that the other additives are working against the antifreeze, and are starting to gel. who would have guessed
rolleyes.gif
I will leave in it until tomorrow to make sure, but there you have it.
 
#23 ·
Quote:
Originally Posted by toolmaker03 View Post

well 6 hours in, and it's looking like the anticongelante antifreeze is the answer to this issue, the 60% antifreeze, 40% water, is the only one not thinking up at all, it seems that the other additives are working against the antifreeze, and are starting to gel. who would have guessed
rolleyes.gif
I will leave in it until tomorrow to make sure, but there you have it.
smile.gif
 
#24 ·
You can definitely get super cold with cascade phase but it takes BIG compressors to do it. Nothing you'll be able to fit in the palm of your hand. TECs will get you nowhere near close to phase temperatures. See the quote below:

Quote:
Originally Posted by Sdumper View Post

R14 arrived and im in the process of tuning the charge or Argon/R14. Right now im idiling at around -137c but would like to idle closer to -140c and load at -130c...

Good thing its still cool in the mornings because the rest of the day is useless for tuning due to the outdoor temps...
Now read this for more info on TECs:
http://www.heatsink-guide.com/peltier.htm

The maximum difference you can have between the hot and cold sides is called the delta tmax. For pretty much any TEC that's about 70C. Which means for you to achieve -90C on the cold side you would have to cool the hot side to -20C. If you read the link above you'll see that this is not even possible because:
Quote:
The specified maximum value of delta T only occurs when the peltier element does not transport any heat
So basically to achieve -90C you have to cool the hot side of the TEC to -20C and have the cold side not even on the CPU. Now sure you could cascade the TECs on top of your CPU but they don't make TECs past about 500w and you'll quickly outstrip the ability of the top-level TEC to transfer heat (since it will have to transfer the load of the CPU PLUS the load of all the TECs running below it) Sorry but what your suggesting just isn't possible.

I'm not trying to shoot down all your ideas and its commendable that your have the inspiration to do all this but what your suggesting is beyond the capability of TECs.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top