PDA

View Full Version : Want to build a refrigerator-like cooling for my cpu



PeeWee
01-11-2001, 10:25 PM
I want to build a small cooling system for my cpu inside my computer. A refrigerator cooling system is ideal: not too much noise, GREAT cooling, does not need fans, not very big, cools better than watercooling.

Can I just use an old refrigerator and rip out the compressor, and then attach it to the waterblok on my cpu with some pipes?

If yes, I will ofcourse leak refrigerant. Which refrigerant is a good replacement + easy to find? (I can get some liquid nitrogen, can I use that or not?)

Prof Sporlan
02-11-2001, 06:41 PM
Just curious, but is your PC a Cray supercomputer? :D Your average PC hardly needs a refrigeration unit to cool the CPU. You might be amused to know that some time ago the Prof was involved with component section for a refrigeration unit used to cool a Cray. :)

PeeWee
02-11-2001, 11:48 PM
It's not a cray ;-)
It is an AMD Athlon TB 1.4 Ghz
But I want to overclock it (2.14 volts, at least 150 mhz front side bus, 12.5 multiplier) This produces a LOT of heat! A traditional air cooler is not sufficient! (Everyone who tried this without watercooling burnt his cpu). But I don't want watercooling, I want to be original.

A guy who used a large bowl of liquid nitrogen got his cpu (1.4 ghz) up to 3 ghz! But he had to turn off his pc because the nitrogen boils up very fast

I want to get a record, and actually be able to use my pc normally
This project was acheived by a company called Kryotech some years ago, but it does not exist anymore. So it really is possible, I was just wondering if there is anything special I should pay attention to

Brian_UK
03-11-2001, 01:04 AM
Originally posted by PeeWee
If yes, I will ofcourse leak refrigerant. Which refrigerant is a good replacement + easy to find? (I can get some liquid nitrogen, can I use that or not?)

Leak refrigerant, did I hear you say leak refrigerant ?

Do you know, or care, about the legal requirements for handling refrigerants within the EU ?

Please do not injure yourself or the rest of the world by playing with mechanical things you do not understand. By all means try to fry your cpu but be honest, does it make you type any quicker?:confused:

PeeWee
04-11-2001, 02:55 PM
Obviously you do not study informatics like me :-)
But I didn't study refrigerator school =)
Tell me what kind of technician to hire, tell me what books to read...

Time is not a problem.

Derek
04-11-2001, 05:52 PM
No low temperature domestic refrigerator is going to suit you. The refrigerant temperature will be so low that you will get condensation eg water not recommended.

The reason for the Cray like fluid coolers is that a heat transfer fluid like glycol (used in antifreeze) can be cooled and pumped easily. I'd start with a small water chiller like the ones found in offices and add a circulating pump. That way you don't have to mess with the refrigerants. If you pump a heat transfer fluid like 3M PF 5050 it won't damage your circuit components as they also use it to clean PCB's. Boils at 28-30 deg C.

For a real challenge you could do what I do pump R123 and soon R245fa. These are low pressure refrigertants but finding compatible pumps is a real bitch. Thompson (USA) and Caster (Italy) have worked so far. Remember you still have to ditch the heat you have transferred.

I also cool high temp components with direct liquid nitrogen injection and also direct CO2 lots of cooling but the presures are dangerously high (230bar) but if you can vent through a heat exchanger stuck to the processor you may be able to make this work. These are both oxygen depleaters so plenty of ventilation and some safety devices please.

I will try and find some pics of the stuff that has been tried.

Derek

:p

PeeWee
05-11-2001, 03:38 PM
I will use a home-made waterblock with the lid welded on (should resist a high pressure) What can I use as an expansion valve?
Will a simple small hole somewhere in the tubes (close to the waterblock (or in the block) do fine to evaporate the refrigerant?

Loosing the heat will (hopefully) not be a problem, I got a radiator cooling element + enough fans

Condensation is not a problem. I found a non-conductive transparant spray which is tick enough to stop any condensation getting through. A cheaper solution is also applying some vasoline (This gives you less danger of sticking some elements together, like with a spray)

Abe
05-11-2001, 08:20 PM
Hello peewee,

I mentioned this some time back, some time back I was in Geneva at the huge Telkom Geneva exhibition. I came across telecommunication racks which house small cooling systems to keep the electronics cool. I think youre after the same thing.
I cant remember the name of the company who manufactured the equipement. But try telecom companies, they were 19" standard racks. There was an article in the refrigeration mags a few months ago, Ill try and dig it out.
The system employed was a basic cooling cycle, using hermetic compressors with a forced air evaporator.
You mention what expansion valve to use, do a search on the internet, automotive air conditioning. New cars have a small pencil like contraption which is a brilliant expansion valve. Looks neat and tidy and is simple. Alternatively try using capillery tubing. Ill do some searching myself and if I come across something will post it.
pss. If you manage to develop something suitable, think about the commercial implications!!! With hotter cpu's theres going to be big demand for pc based cooling systems. You might be the Bill Gates of computer cooling!!!

PeeWee
06-11-2001, 08:10 PM
That's great to hear. If I go through with the project I will certainly post some pictures of the result (and ofcourse the overclocking results =)
I've found some links to the kind of expansion valve you mentioned, thanks for the tip!

Next step: finding sponsors :cool:

PeeWee
06-11-2001, 08:21 PM
Aparently a company tried to make this, but they tried to make it fit in a small block. Result: BAD cooling, temps can reach over 40°

Commercial but way too ineffective cooling (http://www.bit-tech.net/?page=article&id=98)

I sure hope I will be more succesfull :)

Derek
06-11-2001, 09:11 PM
Credit goes to whoever posted this on the company site.

Derek
10-11-2001, 11:24 AM
Sorry there should be an photo of one of our direct cooled processors here b***dy fire walls.

PeeWee
10-11-2001, 04:17 PM
What refrigerant did you say you used for that again?

Derek
11-11-2001, 11:09 AM
PF 5050 from 3M its a PCB cleaning solvent from the PFC family. They have some other products in a similar vein. Our main technical contact is 3M Belgium so that may help. Its about 50 Euros a kg (exotic).

You could look at their HFE's less global warming potential.

I understand it is used to cool train gearbox drive motors and a few other heat transfer applications.

Its non toxic in fact you could almost breath it as it absorbs so much air.

For the film fans amongst you see The Abyss. US Navy divers were reportedly using PFC's as deep diving fluids.

:rolleyes:

PeeWee
17-11-2001, 10:12 PM
Somebody who has done this already (http://home.talkcity.com/ConnectionPt/ptcg/athlon2-1.htm)

Weeha, found a nice example =) This is exactly how I am going to do it. The only problem is finding a refill kit.
Thanks for all replies in this post.
I will be posting pics as soon as my project is up & running

Hubris
21-05-2002, 10:31 PM
Hello Peewee. There are a few of us out there that are trying to build something similar to what you've stated. If you want to send me a PM, I can share a couple other good websites detailing computer geeks' initial forays into the world of pc refrigeration.

One question I pose for the group. Copper is fairly stiff...once you bend copper tubing to put inside your computer, you make it fairly difficult to access the motherboard, cards, etc inside.

Steel braided cable is capable of holding pressurized gas and liquid...but remains relatively pliable. Is it possible to build a system utilizing braided SS hoses? Presumably the suction line could be braided since the diameter is less important....but I was intending to use a capillary line to meter refrigerant, rather than a valve. SS hoses could also typically screw together, which is a bit easier than trying to solder copper tubing...especially inside a computer.

Any thoughts?

Thanks,

Hubris

Dan
22-05-2002, 01:15 AM
I hope you fellows are keeping up with Dabit's parallel discussion. Maybe Dez shoul group the two similar discussions into a single entity.

I still think there is merit to exposing the entire motherboard to subfreezing temperatures. Hubris, how often and why would you have to expose the motherboard to room temperature and humidity?
The electrical connections should not be the reason. There are intermediate connectors that could be used to pierce the housing of the freezer. And Hubris, there are all sorts of flexible connections suitable for the refrigerant connections. They may permeate somewhat, but I don't see that as a problem.... but I still think that putting the entire motherboard in a freezer has considerable merit. Wild idea time:

Take an old home fridge with a cold wall freezer at the top that utilizes that cold wall refrigeration for cooling the refrigerator. Put the motherboard in the freezer section. Cut a hole in the door so you can view and/or access the motherboard... maybe a double thick plexi thermopane and two smaller gasketed holes where you can get your hands on the motherboard and put it in the refrigerator portion without opening the door.

This will severely limit the condensation problem if you want to add or remove components from the board or change switch settings etc. Leave enough lengths of wires with intermediate connections for whatever you think you need. Basically, all you have done thus far is create a minus 10 to 20 deg F ambient around the operating board, and a 35 to 45 deg F ambient to service the board.... and all you did was buy a used refrigerator and cut some holes in it! Call it an environmental chamber; it addresses a lot of concerns without a whole lot of expense or design.

If that isn't cold enough, now we can think about how to make it colder, knowing that we can do anything we want in a controlled environment that will always keep the entire motherboard cold and will not cause condensation when we need to work on it. It is more fun to hit the gate running at subzero temperatures than it is to hit the gate running at room temperature and humidity:}

Dan

Hubris
22-05-2002, 01:36 AM
I did see another thread....I'm following that as well.

One downside of using an actual freezer....it's somewhat less portable than a refrigeration unit mated to a standard computer case. It also loses somewhat on the 'cool factor' - we aren't out to build pretty computers.....but all the pc's built into fridges I've seen look like a tornado happened to hurl the pc through the door of the fridge.

If you were going to talk about extreme cooling....there are parts of a computer that don't like getting too cold....capacitors....cmos battery, hard drive, power supply, anything with moving parts etc.

Just how much permeation are we talking about? Most of us who don't actually work in the trade are going to have to get a tech to apply vacuum and charge the system....we don't want to have frequent maintenance.

Gary
22-05-2002, 03:11 AM
I still think there is merit to exposing the entire motherboard to subfreezing temperatures.

I'm with you, Dan. I would separate the parts that need cooling and build a cascade system around them, -80C would be no problem at all.

Dan
22-05-2002, 03:13 AM
Okay. I suggested and you answered. Certainly the batteries wouldn't work all that well. I am intrigued about the capacitors, however. I thought we were just worrying over a mother board, not the drives.

I also had no clue that we are doing something that is supposed to be portable. Even now, that makes no sense after seeing the designs proffered.

I, again feel that I am proposing all the correct answers to all the wrong questions.:)

Is there a contest with a set of rules that we are attempting to accomodate?

Give me a scope of work and a budget.


I have this feeling like when I talk to my managers.........:

They say "Guess what's on my mind."

Then they say "No, wrong, guess again."

You quietly have dismissed a simple,inexpensive step toward making something cold with what's available, because it is not portable. Thus you abandoned the next step to make it colder.

Regarding permeation, just get a good hydraulic hose and the fittings you need.

Have a schrader valve avaiable for checking pressures or adding gas. Permeation is insignificant, unless the rules of what you are doing have to make this whatever you are accomplishing last for a year or more.

Sorry if I sound sour. But I do not comprehend how a home refrigerator is less portable than a gobbledegook of independant/dependant devices.

If we ask each other the proper questions, perhaps we can get to the proper answers. I sense I am failing with my answers, and that's not fair.

What do you really want?

Hubris
22-05-2002, 05:10 AM
Indeed, perhaps your questions are entirely TOO good.

What do we want. We want something that will cool our processors to very low temperatures, and allow extreme overclocking. We want something that is lightweight and portable. We want something that is appealing to the eye. We want something that is easy to build. We want something that is very inexpensive.

I suspect you are finding a fault in our planning stage. The Vapochill (or Kryotech) is a very pretty, very functional...very expensive solution that matches all conditions above...except one. We're trying to home-build our own version of these commercial products....perhaps even expanding upon them to cool additional devices, while still being at a lower cost than the commercial versions.

There is nothing wrong with building a pc into a fridge....I first saw it done a few years ago. My vanity and ego would like me to build a more precisely-targeted system with better looks than a fridge.

Currently, I have built an enclosure for my compressor, separated it from the condensor, and have everything mounted in a box on casters. I plan on running lines up into the pc case specifically to the video card and CPU, utilizing evaporators I have built. The pc case will be mounted on the cooling unit, and thus be semi-portable for the 4-5 times per year my machine has to move.

I believe I need to finish sealing up the cooling enclosure, run a capillary line that splits, giving 1/3 refrigerant to the video and 2/3 to the CPU (100 watts cpu, 50 watts video). These lines need to be connected to the evaporators...and suction lines need to rejoin before they return to the compressor.

Before I go further....is my theory at least sound? As I've said before....I'm a computer guy...I've done watercooling, I've used thermoelectrolytic coolers...now I'd like to try my hand at phase-change cooling. You guys are the experts here....and I'm asking the favor....that you help me with my project. Hopefully I'm not coming off as demanding...that's not the intent.

Thanks,

Hubris

Gary
22-05-2002, 05:41 AM
I believe I need to finish sealing up the cooling enclosure, run a capillary line that splits, giving 1/3 refrigerant to the video and 2/3 to the CPU (100 watts cpu, 50 watts video). These lines need to be connected to the evaporators...and suction lines need to rejoin before they return to the compressor.


If you are going to use two cap tubes, they should be identical, regardless of the load. And think of them as resistors. If they are placed in parallel, the resistance is halved, and the flow is doubled. This is not a good thing. If this is done the cap tube lengths will need to be doubled in order to balance the system.

Better to split the refrigerant flow after the (single) cap tube.

Don't be concerned about the type of refrigerant used. Unless/until you are aiming for a much lower temp, the refrigerant makes little difference.

The cap tube sizing, however, is crucial. A TXV is much more forgiving and versatile.

Hubris
22-05-2002, 10:01 PM
Is there an easy and cheap source for an appropriate valve? Cap tube is easy to come by....all the equipment I've used for parts has included capillary tubing....so I have plenty to spare. I have no worries about using a valve....what issues does it bring to the equation? Cap tube is static...you aren't getting enough flow, you chop a length of tube......a valve has to be configured for proper restriction, no?

Thanks again for the help.

Gary
22-05-2002, 11:59 PM
Since your heat load is fixed, a cap tube would work very well. The trick is figuring out the ideal cap tube size. This could take a great deal of hit and miss experimentation. The downside is that you don't have the equipment to evacuate the system each time you change the cap tube sizing.

The TXV will adjust itself to the load, so you don't need to experiment with it. But a cap tube is cheaper.

Each has it's pros and cons.

Hubris
23-05-2002, 12:41 AM
CPU head load is generally constant....my systems are almost constantly under 100% load. The average computer out there will vary probably 20% in heat output depending on the amount of work it's doing.

Gary
23-05-2002, 02:55 AM
I'm wondering if a needle valve might be the ideal metering device for this type of system? Or possibly a needle valve upstream of a short cap tube? Anyone have any opinions on this?

Hubris
14-06-2002, 10:06 PM
Just hoping to keep this discussion alive.

Since I've already built a cooling box utilizing a matched compressor and condensor from a winecellar cooler....I believe all I need to do is run copper or some other kind of line to connect everything.

http://athlon.rungerealm.com:8080/pics/Cooling%20box/

One fan will be cooling the compressor, since it will have to run 100% duty cycle, the other fan will be pushing through the condensor.

So far when I've brought up this subject at local refrigeration shops they have been somewhat less than helpful. Perhaps if you kind folks could suggest what I need to ask....they will provide better assistance.

Since I want to have flexible tubing after it enters the computer...I either need to have a junction after the capillary tube from which flexible tubes sprout and enter the evaporators....or else I need to use flexible tubes all the way from the condensor back until it hits the compressor. Since I'm making the evaporators....I can put whatever fittings are required. Are we talking AN fittings.... any suggestions?

Thanks again.

AndrOvr
14-06-2002, 10:30 PM
Yeah.. I need flexible tube too, here can I find that!?

iceman8691
16-06-2002, 09:34 PM
i think you'll findsomeone has already done it, a colleuge of mine repaired one for somone recently, i'll see if i can find out who made it.:D

Hubris
17-06-2002, 12:28 AM
If you're referring to a Vapochill case....yes it's commercially made. It's rather pricey....and limits you to only cooling the cpu, not the video.

If you're referring to someone else making a home-made cooler using flexible lines.....please definately let us know the details.

Thanks!

Jesse

DaBit
28-06-2002, 11:34 PM
Can't you use the refrigerant hoses used in cars? Be sure to get barrier type hoses for R134a, since it's molecules are quite small.

Hubris
02-07-2002, 07:03 AM
Anybody ever ask about using additives in the refrigerant? I've heard this one is supposed to make 134a almost as effective as R12 was.

http://www.epatest.com/qwikboost-press-release.html

abcdefg1675
23-07-2003, 12:18 AM
Originally posted by Hubris
I did see another thread....I'm following that as well.

One downside of using an actual freezer....it's somewhat less portable than a refrigeration unit mated to a standard computer case. It also loses somewhat on the 'cool factor' - we aren't out to build pretty computers.....but all the pc's built into fridges I've seen look like a tornado happened to hurl the pc through the door of the fridge.

If you were going to talk about extreme cooling....there are parts of a computer that don't like getting too cold....capacitors....cmos battery, hard drive, power supply, anything with moving parts etc.

Just how much permeation are we talking about? Most of us who don't actually work in the trade are going to have to get a tech to apply vacuum and charge the system....we don't want to have frequent maintenance.

What about using an old dehumidifyer? I havent opened one up and measured the evaporator temperatures, but some dehumidifyers are small, they have wheels and most people can pack them around.

I dont know much about that 3M liquid that has been mentioned, but would the PCB board break down eventually? Ive seen some PCB boards submerged in transformer oil and the layers start to seperate at the ends. Would this also happen after a long time with this 3M liquid?

Arcan3
12-09-2004, 05:03 AM
the 3m liquid is for testing pcbs... its non conductive but its only meant for tests... not regular use... else it would be used in supercomputers IMHO

mech890
12-09-2004, 04:48 PM
What about a mini absorbtion system like this http://www.nh3tech.org/absorption.html
I think you could make it entirely self contained w/in the case using electric heat instead of propane.

kevinmcggg
04-09-2006, 10:57 PM
why dont you try the new intel core duo cpu. i hear it runs cool even overclocked so all you need is a standard cooling fan. or you could build a small refrigeration unit that pushes very cold air into the case and keeps the interior very cool.

frank
06-09-2006, 09:40 PM
or you could build a small refrigeration unit that pushes very cold air into the case and keeps the interior very cool.

What about condensation?

US Iceman
06-09-2006, 09:47 PM
What about condensation?


Frank, I guess you are saying there might be a problem with water and electricity???

Is this something new?:p

frank
07-09-2006, 09:36 PM
No - I was thinking more of where he would put the condensate drain pan :p

alpha
12-09-2006, 09:25 AM
pc's and refrigeration, oh what fun :)
airtight enclosure>evap on cpu>air evap coil>fill case with dry nitrogen=No condensation and no need for drain pan.
Only problem to keep eyes on is static with such a dry enviroment but if all is well earthed nothing should die :D

frank
12-09-2006, 08:16 PM
pc's and refrigeration, oh what fun :)
airtight enclosure>evap on cpu>air evap coil>fill case with dry nitrogen=No condensation and no need for drain pan.
Only problem to keep eyes on is static with such a dry enviroment but if all is well earthed nothing should die :D

I would be interested to know all about how you build/get an airtight enclosure and also how you insulate the casing with all the wires coming out.

As they say - you learn something new every day.

alpha
12-09-2006, 08:49 PM
Hi there, I think that in the next three months the final build will be up and running and ready for show, I shall put up a post and reveal all in due course, can't give too much away just yet :D

I have just finished 4months of testing on version1 and the results were well above what I expected, it's all a bit rough and untidy the test bed but the final version should be very neat, may even build a beer can cooler into it for those long winter nights playing games..

Will update with more info and pictures very soon!

puterg33k
21-09-2006, 09:28 PM
I think the idea of building a unit like or better than the one that is 900 dollars is a great idea. I was wondering if there's some parts list that we could create? I am really not sure about this, nor have I ever delt with refrigeration. I think perhaps a parts list and details on where to get the parts would be a good start for us newbies :) .

Gohwoz
28-09-2006, 09:32 AM
Where you get it???

scrapen
17-03-2008, 01:56 PM
I dont know how small it has to be, but if you want to use tubing you dont necessarily need a compressor, just a pipe w/ a fluid in it. (aka Heat Pipe) Add fins at the cool end and maybe a fan to condense the working fluid.

I wish Stirling Coolers could be mass produced already. Using one of those cooled a 300 Degrees C computer for the military. (Before they had a messy spray some mixture cooling method) And a quote from the engineer who designed that military application "You could run a 3GHz processor at 15GHz"

While still expensive $500 bux for 100 W Stirling Cooler (100W lift at O C) and i dont know if 40W is available (what i would use), i would just use that coupled w/ a heat pipe and it would be a done deal.

EDDC
02-04-2008, 08:12 PM
back to the Nitrogen. a well ventilated area should be used for safety since one pure wiff supposedly will make your body think it doesn't need to breathe & you won't. i'm still alive so i can't speak from experience.

EDDC :D

The MG Pony
02-04-2008, 11:32 PM
back to the Nitrogen. a well ventilated area should be used for safety since one pure wiff supposedly will make your body think it doesn't need to breathe & you won't. i'm still alive so i can't speak from experience.

EDDC :D

Not quite, pur N2 will not trigger the alarms in your brain that let you know you aren't geting oxygen, thus you wont panick, you'll simply pass out, and unless you get air you'll die from no O2. Your body never evolved an alarm for N2 due to the abundance to it in the air, only CO2 will set off our inbuilt alarm.

Ferdx
10-11-2008, 05:05 AM
Flooded system

I was thinking of completely submerging a PCB in a pressure vessel and using circuit board laquer and maybe foam filler on the main board to insulate the board electrically. Exposing only a passive heat sink to the refrigerant. Without testing I can almost see some kind of chemical reaction taking place. I do know of a liquid that is used in the particle accelerator of the ANU that is non-conducting. I am unsure what kind of properties it has as a refrigerant, but it does boil at room temp as they have to pump down the system on a yearly basis for maintenance.

Glycol

My other idea is to use a glycol system from a beer chiller to pump through a standard water cooler. Set glycol temp, start pump isnt that ez'r?? Short of that, I have an orange juice cooler, commonly found in service stations that i may use for the glycol.

Phase change

Again using the circuit board laquer and foam filler, to avoid the question of condensation (also that upgrades are impossible once system is set up), I am thinking of drilling 1/4 inch(or smaller) holes through a passive heatsink and doing the whole tx thing, by extending the copper just far enough away from the alluminum to avoid melting. Or try to get a copper cooling block.... Electronic tx valves running off a bridge rectifier should work for tx controll, if u can get a small enough valve. I am not really concerned about flexible hoses, as this system is not able to be upgraded once installed.

The MG Pony
10-11-2008, 07:15 PM
Flooded system

I was thinking of completely submerging a PCB in a pressure vessel and using circuit board laquer and maybe foam filler on the main board to insulate the board electrically. Exposing only a passive heat sink to the refrigerant. Without testing I can almost see some kind of chemical reaction taking place. I do know of a liquid that is used in the particle accelerator of the ANU that is non-conducting. I am unsure what kind of properties it has as a refrigerant, but it does boil at room temp as they have to pump down the system on a yearly basis for maintenance.

Glycol

My other idea is to use a glycol system from a beer chiller to pump through a standard water cooler. Set glycol temp, start pump isnt that ez'r?? Short of that, I have an orange juice cooler, commonly found in service stations that i may use for the glycol.

Phase change

Again using the circuit board laquer and foam filler, to avoid the question of condensation (also that upgrades are impossible once system is set up), I am thinking of drilling 1/4 inch(or smaller) holes through a passive heatsink and doing the whole tx thing, by extending the copper just far enough away from the alluminum to avoid melting. Or try to get a copper cooling block.... Electronic tx valves running off a bridge rectifier should work for tx controll, if u can get a small enough valve. I am not really concerned about flexible hoses, as this system is not able to be upgraded once installed.

did you post blind folded or did you roll dice and then go "Ah HA that thread has nothing to do with what I want to say so I'll say it any ways as the dice landed on 2!"

How did you manage to join up even?

Please read the posting rules and be for hitting "Submit reply" look at the thread does it make sense what you say there? If not make your own new thread.

The MG Pony
10-11-2008, 07:20 PM
FYI You'll seriously regret the day you attempt submerged nothing but prblems and bills.

http://www.ocforums.com/showthread.php?t=494311

http://www.ocforums.com/showthread.php?t=373263

Now in the future make your own new thread please!

camerond
13-11-2008, 06:37 AM
Why don't you use liq nitrogen for the cooling block on ur chip, but instead of just watching it evaporate just condense more out of the air? It would use a lot of power though. Just a thought