Slashdot Mirror
Liquid Nitrogen Cooling at Home?
newell98 asks: "Given the rise in popularity of water cooling systems for home computers, I was wondering how many slashdotters have played with the idea of cooling their system with liquid nitrogen? Lots of super-comps use them (or used to at least), and I'm curious about who's played with the idea of taking home computing to the same level?" The thing to remember about Liquid Nitrogen is that this stuff is generally not safe for home use. It must be stored and used with care or serious injury can result. I think this is why not-too-many people use such in overclocking. Water is by far more easier to obtain and is harmless to boot. Now, after saying all that, have any of you tried using liquid nitrogen in cooling a home or garage-built computer rig? What kind of safety precautions did you take, and how well did your cooling system work?
Seriously, this is overkill. Excellent cooling can be achieved with a traditional water pump and thermocouples (TEC's, Peltiers, whatever). At least that way you don't risk freezing your fingers off.
-Billco, Fnarg.com
While liquid nitrogen is sexy, almost any coolant in a refrigeration system will keep almost any system cool very effectivly, and save quite a bit on not needing special insulation, and atmospheric needs associated with cooling.
Of course if you live in the north right now you could just stick the thing outside, its supposed to be quite cold for the rest of the week.
Degaussing scares the bad magnetism out of the monitor and fills it with good karma.
I've seen this discussed on places like [H]ardForum before, and the general consensus seems to be that its TOO cold. Ceramic on the chips can become extreamly brittle and you have to worry about condensation, which will easily kill your system unless you use some type of non-conductive grease.
The major problem I can see with liquid N2 cooling is the formation of condesation on the CPU cooler. As soon as one drip hits the Motherboard the whole computer is out. :-(
Water doesn't have this problem because the temperature is always above room temperature. So condensation cannot form.
Get an electrically inert liquid (they're expensive,) a cooling coil, and a pump system. Suspend your mobo over a large resivoir, fill place the cooling coil in the resivior, fill with the liquid, and use the pump to draw it up, and let it flow over your mobo. It's like a waterfall!
Vintage computer games and RPG books available. Email me if you're interested.
Just use R-134a, an evaporator, a condenser, a throttling valve, and a compressor. Create a sealed system with the components in the right order, and you're set.
It's called a refridgerator, and it's much easier to use to keep your components cool enough. Keeping liquid nitrogen liquid, plus the hardware to pump it is way too expensive. Seriously, what a silly idea. Liquid Nitrogen, pfft!
One future, two choices. Oppose them or let them destroy us.
This guy uses Fluorinert and Liquid N2.
s ubmersion/submersion.html
http://www.octools.com/index.cgi?caller=articles/
S
Most/all mini-fridges lack sufficient heat removal capacity to keep up with the heat production of a fairly hot computer. Think of their intended design: how often does your food heat itself to 30 C while sitting in the fridge?
You could tinker with the pressure ratio, and perhaps even install a stronger/more efficient compressor, but you might as well go with a custom built unit, and only cool the hot spots with a small heat exchanger/[evaporator].
My 2..
One future, two choices. Oppose them or let them destroy us.
Put down the crack pipe. What on earth do you need to do that requires liquid nitrogen to cool it. This has got to be the worst "ask slashdot" ever, except for the one where the kid wanted us to do his homework for him.
A supercomputer your PC ain't. Cool it the sane way.
- A.P.
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
As anyone that's worked with extreme cooling knows, its the condensation that kills. Ever look inside a chest freezer? There's ICE on the walls.
I've seen (and tried) lots of peltier combinations, cases in mini-fridges, etc. But as soon as you get far enough below ambient, you risk condensation on the components. I've cooled systems to about 30F(ambient about 70F) and fried an AMD 'cuz its pins were soaked in water.
With N2, its LOTS colder than ambient, so condensation turns to ice VERY fast. So you have to insulate the shiat out of your proc.mobo. But from what i've seen, the o/cing advantage isnt a whole lot more because you're limited by sheer quality of the silicon, and the design of the transistors. They're just physcally too big to switch that fast, etc.
my $.02(.01CAN)
For the cost of the liquid nitrogen and the precautions needed to handle it safely you could just go out and by a faster CPU and motherboard. You're not running a supercomputer.
Yoda of Borg am I! Assimilated shall you be! Futile resistance is, hmm?
Sure, it's harmless to *boot* Aqua. But God help you if you attempt to install iTunes.....
- undoware.ca
Speaking as a chemistry major who's spent entirely too much time in lab screwing around with the liquified gases, don't worry. Seriously. I mean, I guess unless you didn't have _any_ ventilation in the room where you were fooling with it. Still... You can hold liquid N2 in your hand, no problem. (there is an interface layer where your body heat has converted the N2 liq->gas, so it's "floating" on a cushion of gaseous nitrogen; basically just don't try to drink it or something retarded like that) I actually did the superconducting ytterium compound/levitating magnet thing once using superconductor pellets held in my hand with liq N2 poured on top. :-) So don't freak out, the liquified gases are pretty benign compared to some of the shite us chemistry f00ls mess around with... (as a still-not-so-evil example, 18+ molar mineral acids, so concentrated that they behave more like jelly than liquid) probably the only "liquid air" compound I'd be cautious around would be liq O2 and that's only because of the flammability aspect... I guess the main thing with these compounds is be careful about really really super cold metals because you could frostbite off of them (boiling, and thus gas interface, point is high)... wear work gloves or something if you're handling things in direct thermal contact with the gas (i suppose I should say liquid).
sorry for any incoherency or bad grammar/spelling; very, very, very drunk at the moment ;-) wot can I say, chimay belgian ale is a great way to celebrate a radical success day at work...
if you want to mess around with ReallyCold(tm) but don't have access to liquified gases, try dry ice + (alcohol, usually ethanol or something more esoteric than that but methanol would do in a pinch (non-chemist-speak: grab some dry ice from Kroger's or a party store and mix it with rubbing alcohol, waaaay colder than you'd get with normal ice and water. in a pinch, just find the cheapest nastiest grain alcohol you can... (everkleer, what?) ;-))). in lab I've hit the -20/40 F region using just dry ice and some not-to-uncommon alcohols. as I recall the F scale itself was calibrated using esoteric mixtures of dry ice and ethanol. which, if you think about it, highlights the total absurdity of the non-metric temperature system. always kinda wondered if the water-bases liquid cooling systems could be adapted to use this class of stuff as a coolant...
Have fun, and play safe! (which, with these compounds, really means don't snort the vapors or drink the liquid)
News for Geeks in Austin, TX
The big safety hazard is if you have poor ventilation and end up with low levels of oxygen in the room. Another hassle is that the liquid conducts elecricity. The biggest problem, however, would be your PC icing up. Thermal stresses could also be a big problem.
Ultimatly, the question is why do it? If you have electronics that operate best at low temperatures then it makes sense, but PCs have components made to run at room temperature. Semiconductor behavior is temperature depenant, so the machine may not run as intended at low temperatures (the CPU may end up being made of a lot of resistors instead of transistors). Tin-Lead solder has known cracking problems at sub-zero temperatures, and not a great deal of strength anyway (Scott of the Antarctic got to stay there forever after the solder failed in fuel tins at low temperatures). Delamination of tracks from the fibreglass base could also be a problem if the board gets very cold - copper and fibreglass shrink at different rates.
All of these things can be designed around. the easiest solution that I can think of off the top of my head is a great big lump of metal in contact with a resivior of liquid nitrogen under a feeder tank. When the CPU gets too hot, drip in a bit more liquid - just keep the liquid a long way from the CPU to keep all the ice and water away. a watercooled block would, of course do the same job if somewhat less efficently - or my favorite: airconditioning, big heatsinks and big fans. That way the user doesn't overheat either.
Ask Robert Wagner about water being harmless.
Visit dhmo.org for more information on the dangers of this all-too-common substance.
I'm sorry I never took a picture of the Cray 2 that was upstairs. There was a very pretty fountain designed to give a visual indication that the coolent was in fact circulating.
If you were lucky, you could see a little bubble rise off the circuit boards.
Of course, the Connection Machine over next to the Cyber was more flashy, with 64K little red LED's flashing all the time...
Why, yes. I am a rocket scientist.
Bob-
The Ludwig von Mises Institute. The reasoning individuals economics
chongo (was here)
Yes, they do (Apple users have fans, that is). At least some of them have. Even some PC users have fans (like this guy ;-).
Lars T.
To the guy who modded me down from perfect to terrible Karma - Apple haters still suck
First install your mobo and proc inside this custom case. Then use the patented location cooling system to lower the temperature to 4Deg about absolute zero. Now you only need to find 7.5 billion miles of CAT 5 ( and some batteries that last longer than 2 hours). Sure you can play Qake at 900fpss, but the latency is a bitch.
SD
âoeWho knew something as harmless as willful ignorance could end up having real consequences?â
Search google for liquid nitrogen ice cream for recipes.
Try the string liquid nitrogen amateur scientist for experiments you can try at home.
Unless you've got a farm, farm uses aren't interesting. However, if there are farms near by that use it, there may be a liquid nitrogen vendor who delivers to your neighborhood.
You could use water to cool the motherboard by immersion or a spray bath too... if you were brave (and lived in a clean-room).
Pure water is non-conductive. Yes, it is. Really, it is!
However, if you add impurities to the water (mostly salts), it rapidly becomes conductive due to the free ions in salt.
So, as long as your MB didn't have any contaminants on it, and you sealed it into a container, you COULD cool your MB/CPU directly by spraying cold water on it.
Now, as for actually trying this with MY computer... ah, no.
MadCow.
I used to have a sig, but I set it free and it never came back.
I'm going to try it.
Conveniently, I drive past two large scrap yards every day on my way to work. It just so happens that I want to pop into one of them to grab a central locking module, and the larger of the two has a fine collection of Citroens.
So I'm planning on setting up a large tank of a light oil of some kind (probably diesel, it's pretty runny but not volatile - a bit hygroscopic though). I would use LHM or something but it's pretty inflammable, and *very* expensive (about £5/litre).
So, pump it with a car fuel injection pump (about 20 gpm or so!) through a couple of oil coolers, blow through the coolers with a car radiator fan, and you've got nice cold oil...
I need a PC to test it with, but I'll try overclocking the old P60 that's lying, dusty and forgotten, in a cupboard to, oh maybe, 66MHz? It should kick out plenty of heat...
Oxygen has a higher boiling point (-183 C) than nitrogen (-196 C), and forms highly explosive mixtures with many organic materials.
I do not deploy Linux. Ever.
My primary work machine is a water cooled athlon. It runs pretty damn near ambient, and it runs cooler than it did with a heat sink even if I turn off the fan (e.g. silent running). This has been running 24/7 for about 4 months now with no problems whatsoever, so just go with that.. I have the chip overclocked as high as the motherboard multiplier will let me.
http://www.nyx.net/~smanley/watercool
That said, this has to be the most pointless ask slashdot I've ever seen. Why would you ask that? Niquid N2 is expensive and dangerous. Most motherboards and memory can't clock high enough to overheat a watercooled system. If you're going to spend 5 large on a N2 cooling system, why not just get a half dozen watercooled machines and cluster them?
My next box is a Powermac on OSX, and the linux box can move to the closet. Quiet, and clustering is as easy as 1-2-3.
..don't panic
I setup a water cooled system on my highly overclocked CPU some time ago with some interesting results.
;-)
;-)
The water block was made from a cutoff PCV piping cap mounted over an old heatsink. I used a surplus Peltier from All Electronics and I used my computer's power supply to run the Peltier. The water pump was a fountain unit sourced from an online supply company and was intended for 24X7 use squirting water in the air of someone's small pond.
My first problem was SEALING the damned water block. Pinhole leaks were a PITA but with time I got a good seal using some SERIOUS automotive gasket sealer (Tough Stuff I think it was called). Nowadays you can buy built water blocks - I'd suggest doing this! They're not THAT expensive and you really don't need a super big one for this project. The second problem was that the water supply got VERY warm over the course of about 8 hours. This was 5gallons of water in an old cat litter bucket - sealed. There was actually steam rising from the water! Solving this was as simple as buying a trans cooler at the local auto-suply store. The water still got hot until I put a small FAN on the trans cooler - this is a must! With no airflow the trans cooler acted as a very poor heatsink. Melted quite a bit of snow on it though
The above setup allowed me to nearly double the clock speed of the CPU I was using - an older Celeron. I don't recall my highest peak but I'm pretty sure I was hitting a Gig long before Intel released one. I ran Distributed Net's program so the CPU was tasked 100% 24X7. This program raises CPU temps noticably! Temp on the CPU sensor was nearly freezing measured from just under the CPU - I used a special add-in board to measure this as MBs at that time had no temp sensing. Unfortunatly this nice board is no longer sold or supported - it was called Heat Sentry I think.
The best part about this whole system was the silence. I mean ZERO noise except from the power supply fan. The pump barely hums when it runs submerged and I was able to remove ALL of my case fans - there were many of these. I loved this system! I found that a small amount of Clorox would kill the bacteria that quickly formed (ick!) and had no need to experiment with Water Wetter or dishwasher spot remover in order to up the cooling ability - my CPU temps were stable and reliable.
I did have some ongoing problems and one that finally killed the project until I have more time. First - condensation UNDER the CPU. Temps beneath the CPU were nearly freezing and certainly below dew point. Condensation would occasionally form around the pins of the CPU leading to an occasional system crash. Fixing this would sometimes require a hairdryer to dry the socket (sigh). I never tried dielectric grease or foam insulation under the CPU. Both of these ideas should be used if someone attempts to duplicate what I did. The second and worst problem was what occured when the system must've locked up early one day while I was at work. With no load to heat the Peltier it cooled WAY down. I came home to a dead system that had a solid block of ICE around the CPU socket area and onto the water block. As the ice furthest from the Peltier melted it dropped water onto my video card! The video card ended up fried but no other hardware was damaged. There were some tense moments with the hairdryer though! At that point I decided that the added frames in my game of the week and the extra keys in RC5 simply weren't worth the trouble of insulating everything. I pulled my fan\trans cooler out of the window and put it all away for another day. I DO still have the Peltier, tran cooler, water pump\resevoir, and a brand new aluminum water block on my shelf though. My unlocked AMD CPU now runs a bit over 1.4GIG with a good copper heatsink (aircooled). At some point when I've got time I'll water cool again as it was lot's of fun but I've simply not got the time right now.
A feedback system to monitor Peltier temps would've been a nice addition as would a seperate power supply to run it. I was using a +12 and a -5 lead if memory serves. Without water flowing the CPU block's water hoses got pretty pliable with the heat and I was sure the water in there was nearly boiling. Monitoring the pump and the Peltier would be good ideas especially with today's super fragile CPUs. I noted that my plastic hose deteriorating over time from having been submerged in water for so long. I also had to cut the mounting ends on my trans cooler in order to up the water flow and lower back pressure. No antifreeze was EVER used - it lowers cooling ability as any gearhead can tell you. I used tap water with infrequent water changes once the system was closed up.
Straight water does a VERY good job of removing the heat compared to convection and I predict that it won't be too long before we start using more advanced methods of cooling other than simply bolting on a still bigger block of heavy crap onto our CPUs. Cooling pipes at the very least look interesting as does liquid cooling sometime in the future. Lookup amatuer astronomy and how they cool the Peltiers that cool their CCDs to below freezing for ideas. These guys were way ahead of the Overclockers and I learned a great deal from some of their projects....
P.S. A shame I didn't see this article when first posted. No Karma for me this time
Build it, Drive it, Improve it! Hybridz.org