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

Super Cooling: Phase Change + TEC

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

cabe

Registered
Joined
May 10, 2003
here i will outline (and update, as my project progresses) my computer cooling project, involving both vapor phase refrigeration and thermoelectric cooling.

i've been thinking for some time, wondering how i could set up a really badass cooling system. ok here goes.

how about if a pelt were to be used with a vapor phase refrigeration unit? no, not with the pelt directly on the evaporation plate, but how about if the evaporation coils were in a resivoir of liquid that can go to extreemly low temps? in this way, the vapor phase refrigeration unit's evaporation coils would have MUCH more surface area contact with what its cooling, and hence absorb more heat from it ^_^. this super-cooled fluid could then be pumped through an insulated tubing system to a waterblock. that's what has the pelt attached, the water block. and, with applying such a freezing cold liquid to the hot side of a pelt, the cold side would get EVEN colder :D.

with a vapor phase refrigeration unit from lets say.. a refrigerator, (like i just happen to have :), hehe) im sure temps WELL below 0C would be feasable.

the beauty of this system is that you can keep the same vapor phase refrigeration system and peltiers, instead of having to upgrade every time you switch to a new socket (not that often but hey.. how often do you really wanna shell out $600+ for a prometia?). so the only thing that would need to be upgraded is the waterblock and cold plate. other than that, you can keep the whole rest of the cooling system, and who says it only hasta be attached to one thing? you could have it hooked up to a gfx card as well, and supercool that too. (can't do THAT with just 1 prometia ;) )

here is what i've done so far.

1) i have a 1/3 HP compressor removed from a refrigerator
2) i'm 99% sure on using isopropyl alcohol for what i'll pump through the WCing circuit
3) i'm going to get special tubing and possibly coat my pump with some type of laquer to prevent the alcohol from eating it up
4) i want to convert the compressor to use propane (since i believe it uses freon, and freon is no longer readily avaliable) and possibly use a mixture of R134a and argon later on down the road
5) i'm going to use a hard start kit to power the compressor

here are some things i'm either having trouble with / haven't decided on / having trouble deciding on:

1) how long to make the evaporation coils
2) how long to make the condensation coils
3) what diamater capillary tube to get
4) what diamater evaporation coils to get
5) what diamater condensation coils to get (but i'll probably end up buying a large sheet of condenser and cutting it to fit)
6) how to clean the compressor

well that's mainly what i'm thinking about right now.

any comments/suggestions are greatly appritiated! :D
 
just ditch the peltier and it would rock. Pelts dont work worth a damn at extream temperaturs and wont give you much gain. Just stick the evap in a res of liquid of your choise and run it like any other h20 setup except insulate the lines. As for the your diameters, Im not a good person to answer sense Ive never chilled water before. I will say that you will need about .81 meters of .028" capillary or 2.81 meters of .031" capillary if you are wanting it to handle 160watts of heat.

as for the compressor there is no need to clean it because propane works fine with any of the oils. Try and find a condensor with pipe no bigger than 3/8". It doesnt need to be big at all depending on how much surface area it has and how much air you are going to move through it.
 
pelts arent good at those low temps? :( where can i find out more info about how well peltiers work at a given temperature? would a higher wattage peltier work better? if i could chill the alcohol to say -35C then how would a 226w peltier affect temps?

otherwise, i'm going to try to look into a liquid that has a lower freezing point with high thermal conductivty. a 25% R134a 75% argon mixutre would be awesome-- evaporating at -120C! :eek:
 
The thing is the 226 W peltier wouldnt be as good as putting that -35 C sucker in your block then directly on your cpu itself. and at -120...well.. There IS a law of diminishing returns... and a few articles i have read when people used ln2 and such to gooo waaaaaaay below zero it caused the compuer to f^ck up.

Just stick with doign the phase change onto the cpu instead of onto the peltier. You will get better results.
 
then i guess for the waterblock i'll use the whitewater, but i still want to research more into why peltiers dont work well at low temperatures.

if i cant find a liquid that will go to really low (-120C) temps than maybe i just might as well set something up like a prometia / vapochill with the capillary tube dumping straight argon onto the coldplate. i dont know if a 1/3HP compressor is beefy enough for that, but there arent many options at those low temps...

if i can't get the TEC to work with the vapor phase refrigeration, then i might just look into putting a little argon with some R134a to lower the boiling point to about -80c, which is where i believe isopropyl freezes.

does anyone know of any other (affordable, not that $1,000 a liter 3M stuff) liquids that won't freeze at ultra low temps, and still have decent thermal conductivity..?
 
why i'm wondering if pelts would work with a vapor phase refrigeration system, is because shortly after i thought this up, i did some research and found this link:

http://www.octools.com/index.cgi?caller=articles/phasechange/phasechange.html

that guy got it to work! is the reason that i should not use peltiers with the phase change system partially because it would be hard for the phase change system to handle such a high heat load of 226w plus the processor?
 
The words dont try it come to mind. If you did pull it off you woudl be in trouble. The CPU actually starts to rip itself apart bettween -40C and about -50C to a point where it wont work
 
Lynx said:
The words dont try it come to mind. If you did pull it off you woudl be in trouble. The CPU actually starts to rip itself apart bettween -40C and about -50C to a point where it wont work

i'm not going to try to disagree with you, but where did you get your information about this?
 
Last edited:
it will only rip itself apart if you go from from hot to cold too fast. It wont do anything to the chip if its at even -270c temp as long as it takes awhile to get there.
 
yea thats what i was thinking...

i've heard rumors that 'some of those computers with ln2 have stuff go wrong with them if they get too cold' but have never seen anything to substantiate these rumors. anyway i wont be doing anything like liquid nitrogen any time soon.
 
A CPU will have thermal stresses long before - 270C and does not depend on the rate at which it is cooled. Sometimes someone will get away with cooling down to LN2 temps, but how many times do they do it 4,10 maybe 20, then the next thing ya know it wont boot up. Thermal stress is a cumulative process and should not be taken lightly. I dont know about you guys, but I like the idea that I will get more than 20 boot ups out of my CPU.
 
you know i've really only heard rumors and speculation about low temps damaging CPUs, so where can i find some more info about thermal stress on CPUs?
 
Soz I cannot remember where i got that titbit of info from but what I think basically happens is that the small solder balls that connect the core to the packaging get stressed (as teh packaging contracts a different amount to the core) untill one or more loose contact or dont have a good enough contact for the cpu to run propperly. But if you Phasechange you are ok as long as you dont plunge the depths of -50, -60. The LN2 guys PCs only usually last for about 2-3 hrs before the CPU craps out. Phasechange will probably last a few years.
 
yea that makes sense. sucks though :(. i'm hoping i'll be safe at around -50 to -60 with the project i'm working on.

i'm going to keep looking for info on the thermal stress though.
 
i'm now trying to find what would be the best balance of the pressure difference between the high / low pressure side. i guess that would have to be determined more by trial and error.

now in my original plan i planned to use peltiers, but here are a few reasons why i think i may actually get better performance if i optimize the system for non-pelt cooling:

1) i'll be able to use a smaller diamater capillary tube (lower pressure on the low pressure side, making the propane boil at a lower temp)
2) because of the smaller diamater capillary tube, the system will not be able to handle as much heat, but that is not as much of a problem since such a system would not use peltiers
3) peltiers lose a lot of their efficiency at ultra low temps, so that means a greater chance of the compressor alone (with lower pressure on the low pressure side, although lower heat capacity due to restriction) surpassing the power of the combined peltier / vapor phase refrigeration.

those peltiers just add so many watts of heat, a 226w on the CPU and a 172w peltier on the graphics card, plus the heat output of a overclocked CPU and overclocked graphics card... i'm beginning to wonder if a 1/3HP Compressor (approx 3000BTU / 770w) could handle all that heat.

so mainly what i'm wondering is which setup would get me lower temps:

[1]
original idea with a high capacity for heat, but with higher coolant temps. the coolant would be pumped around to the processor via normal WCing system and dumped onto the hot side of a pelt, where the pelt, with about 50% of its efficiecny would cool the processor down some more.

-or-

[2]
much smaller capillary tube than first idea, much lower pressure on the low presure side, much lower boiling point for the propane. lower flow of propane through the system, and thus lower heat capaicty, but this wouldn't be a problem due to the greatly reduced heat output of the computer due to the removal of the peltiers.

please tell me if/where i went wrong in anythign i described, and which setup looks best. as i mentioned before i have a 1/3HP compressor removed from a '70s refridgerator. i've talked to warlord and he believes my compressor is around 3000BTUs (it doesnt have a LRA number, but aparrently 'fridges have compressors around 3000BTUs). also we discussed that this BTU rating translates into about a 770w rating.

i know the theory of vapor phase refrigeration, but beyond that... i need some help! :confused:

as before, any comments GREATLY appritiated!:)
 
awesome. someday i hope to LN2 cool my CPU, so i'm wondering, how did you do it? what temps did you get? was it stable? what vcore?

hopefully, i'll eventually (years from now, lol) be able to get a ln2 plant and run it on ln2 all the time! that would basically be my final ultimate cooling setup, with the possible exception of a helium cascade system.
 
What if u got like a 800watt peltier(you would need a **** load of power) but i'm pretty sure that would bring your cpu down to like -100C, somewhere around there.
 
Back