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

Theory on adding petier to WC system . . .

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

Farwalker

Registered
Joined
Aug 13, 2005
Location
Gainesville, GA
If I understand correctly, the addition of a peltier such as Swiftech MCW5002-775T Liquid Cooled Thermoelectric assembly http://www.sidewindercomputers.com/swmc7.html (and psu for the peltier) to a water cooling system that has a radiator and fan assembly sufficient to remove the heat; the peltier would lower the temperature on the CPU versus a mere water cooling system with all of the same components.

If this is generally correct, about how much cooler might a e6600 (cooled by the Swiftech assembly) which is at stock be assuming a ambient air temperature of 70 degrees Farenheit. Is it reasonable to expect 10, 20, 30 degrees or more difference?
 
yea, im also wondering about this.

but im guessing an overclocked conroe would probably need a 226 watt pelt.

the only problem is how much colder it would be.



but the reason why i want a pelt is because its a lot more compact. all it is is a small white square and an extra meanwell psu. and thats not very big.
 
You would probably need to look at a 300+ watt pelt and subsequent cooling. There is also some debate over pelt fire hazard issues and horrible power consumption and subsequent power bill costs which put the whole thing in question. Especially when you can get colder with phase change based refrigeration cooling, be it direct dye or chiller, for cheaper in the long run and what some consider safer.

Do some reading first if this is the first time you heard of a pelt or thought about it.

http://www.ocforums.com/showthread.php?t=485785
http://www.ocforums.com/showthread.php?t=481772
http://www.ocforums.com/showthread.php?t=486810
 
If the addition of a peltier has no more effect than lowering an overclocked cpu at full load less than 20 degrees fahrenheit, then I would not want to go to the added expense and trouble.

I would hope that instead of, for example, 110 degrees at full load; I could get down to 90 degrees or lower.

My Freezone cooling system has six solid-state heat pump wafers (TEC) 40 x 40 x 3.5 mm (each). At no load it keeps my OC'd opty 165 (2792MHz) at 70 degrees fahrenheit (19C) or at room temp. Under full load it goes up to about 110F (43C). It only has a single 120mm radiator.

The Swiftech has a single 50mm x 50mm x 3.5 TEC and claims "Up to 50°C drop in processor operating temperature". I plan on having a triple 120mm radiator on my water cooling system.

The cost of a phase change system seems like it would be twice as much as a water cooliing system; $800 versus $400. Additionally, the Swiftech MCW5002 costs $125.
 
Pf.Farnsworth said:
be it direct dye or chiller, for cheaper in the long run and what some consider safer.



hmmm... so could i use my current w/c set-up. but remove the radiator, and replace it with a chiller set-up instead. just put some insulation around the water blocks/pump and it should be ok right?

i thought phase change could only be directly on top of the CPU. but a tiny compressor (not super tiny, but enough to cool a conroe) might be of use. because im thinking that because it will only "chill" the water only, and not the full brunt of the CPU, i might not need a huge compressor - which is what I'm trying to avoid here, below ambient temps, but not freezing, but not bulky. so im sacrificing performance for basically space and portability. hmm.. im still deciding whether or not just to go straight up watercooling.
 
Farwalker said:
The cost of a phase change system seems like it would be twice as much as a water cooliing system; $800 versus $400. Additionally, the Swiftech MCW5002 costs $125.

More like $400-700 for phase. It just depends on where you get it and whats in it.

But thats not the point. Point is in the long run the pelt will cost more though your power bill payments and will ultimately give you much higher temperatures.

darkcow said:
hmmm... so could i use my current w/c set-up. but remove the radiator, and replace it with a chiller set-up instead. just put some insulation around the water blocks/pump and it should be ok right?

i thought phase change could only be directly on top of the CPU. but a tiny compressor (not super tiny, but enough to cool a conroe) might be of use. because im thinking that because it will only "chill" the water only, and not the full brunt of the CPU, i might not need a huge compressor - which is what I'm trying to avoid here, below ambient temps, but not freezing, but not bulky. so im sacrificing performance for basically space and portability. hmm.. im still deciding whether or not just to go straight up watercooling.

Yes you can do that and many people so. I personally will take a chiller over a DD system any day.

What parts pf your WC system are ok to use depends on how cold you want to get. You need an all metal block, and if its a 2 piece design then take out the rubber o-ring and use a low temperature epoxy like devcon 250 to bond it shut as the o-ring usually freezes, shrinks, cracks and leaks. Or you can braze it shut, I prefer epoxy as its cleaner.

A pump that can handle the temperatures is also a good idea as they have been known to fail due to shattered impellers and housings as a lot of pumps feature plastics that become extremely brittle at low temperatures. Something like the Panworld 40px is a good value.

Other then that yes, insulate and you are good to go. Just don't use water in a chiller. Ethanol/Methanol and distilled water mix is preferred.

Why you thought phase change refrigeration has to cool the core directly I don't know, but you can go any way about it. Its just phase changed based refrigeration, you can cool anything any way you want, its just a principle of thermodynamics. You might want to look in how it works lol

There is nothing about a chiller that makes a smaller compressor better vs direct die. You can have a small compressor and have a direct die setup too. Direct die will get your better temperatures as its more efficient (unless the evap is a bottle neck then a chiller would do better) with the same compressor. However a chiller is better in many other ways. You can cool multiple heat sources easily. Chillers have better stability when it comes to dealing with heat spikes due to the large thermal inertia of the coolant. If you have to fuss with anything its on the water cooling side and is easy to mess with vs cutting up your phase change loop. They also integrate with an existing wc system given the right circumstances. And you can keep it in a separate case and connect/disconnect using quick connects as you please. Chillers are really the practical way to go.
 
Last edited:
hmm... so would you think an Apogee and a d5 could withstand it? i mean. the temps aren't going to be super chilly as i'm not going to get a super compressor (just a small little AC unit, but an AC unit that can run 24/7). of course ill replace the barbs in the apogee with some metal ones because those are plastic. and i am very handy with epoxy.

what do you think the recommended compressor for a non super chilled set-up is? (just below ambient... not -60c)

i have read up a lot of the stickies and i know how direct die works. its just that most people have direct die set-ups right now. and i've never seen one that chills the water yet. like.. how would i cool to water? could you point me in the way of a sticky or explanation of one that does? would i just create a box. stick the evaporator onto it. and then bam? or is there a more efficient way to do it?

i could be able to do that via my triple radiator. just stick it in the middle somewhere and sooner or later the whole rad will be iced over.
 
You cool the liquid with a heat exchanger much like a interstage heat exchanger on a cascade if you have read anything about cascades.
A plate or tube in tube heat exchanger would replace the evaporator in the system.

Sizing the compressor for warmer temps, I doubt thats going to work, it would be more likely that you would be adjusting the metering divice and charge for that purpose.
 
In a good pelt Setup I think you could get >0C<10C idle and probably <20C at load. Depending on your cpu. I think a conroe would do a lot better than say an X2.
 
darkcow said:
hmm... so would you think an Apogee and a d5 could withstand it? i mean. the temps aren't going to be super chilly as i'm not going to get a super compressor (just a small little AC unit, but an AC unit that can run 24/7). of course ill replace the barbs in the apogee with some metal ones because those are plastic. and i am very handy with epoxy.

what do you think the recommended compressor for a non super chilled set-up is? (just below ambient... not -60c)

i have read up a lot of the stickies and i know how direct die works. its just that most people have direct die set-ups right now. and i've never seen one that chills the water yet. like.. how would i cool to water? could you point me in the way of a sticky or explanation of one that does? would i just create a box. stick the evaporator onto it. and then bam? or is there a more efficient way to do it?

i could be able to do that via my triple radiator. just stick it in the middle somewhere and sooner or later the whole rad will be iced over.

The Apogee has a plastic part besides the barbs that I don't think you can replace that easily:

swiftech-apogee-wb-4.jpg

Window ac units usually use pretty large compressors. Don't worry you wont see -60C. You wont see that on a direct die SS setup let alone a chiller. DD SS reasonable load temperatures are -40C to -50C. Chillers depending on quality and parts -20C to -40C.

Anything below 0C is too cold really so its not really a matter of "super chilled setups". Get a swiftech 6002 block if you can. If not Silverprop makes good copper bottom/brass top blocks.

Cpu:
http://www.silverprop.com/cycloneevose.aspx

gpu:
http://www.silverprop.com/cyclonefusionsl.aspx

all:
http://www.silverprop.com/watercooling.aspx

petra has them for sale at his store:
http://www.petrastechshop.com/siwa.html


What people do if they are broke is dump the evap from the ac in to a picnic cooler chest. But a better way to go about is a proper heat exchanger.

Tube or plate. Look up what Doucette has.

Plate hx:
http://www.doucetteindustries.com/brazed.html

Coaxial 1:
http://www.doucetteindustries.com/coaxial.html

Coaxial2:
http://www.doucetteindustries.com/suction.html

You can find good deals on ebay for plate hx.


You can read up about chillers and see alot done at XS the phase change pc cooling central lol
http://www.xtremesystems.org/forums/forumdisplay.php?f=155
 
thanks for that defiantly makes it a lot clearer. well, i could sell off the apogee. and get a new one. then get a better block...

but as of right now. i'm thinking about just sticking with w/c. and then chilling it later once i get bored.


but it would be fun to try and stick a whole chilling system into a wooden case. but i think ill work on that at a later time. i think first things first is to go water cooling. and then chill it later.
 
You probably could replace the top of the Apogee with a 1/4" thick copper plate tapped for the NPT thread size of your choice. I haven't really seen the underside of the plastic top to see if it really does anything besides keeping the water in and holding the barbs.
 
Those parts on the apogee are made out of Delrin which is good for atleast -40°c. Some of my friends that used it for Antenna covers for an airport survailence project say that it was totally fine down to -60°c.
 
speed bump said:
Those parts on the apogee are made out of Delrin which is good for atleast -40°c. Some of my friends that used it for Antenna covers for an airport survailence project say that it was totally fine down to -60°c.

wow... this makes this idea of phasechilling my computer a lot more interesting... if i can keep the d5 as well. (im hoping it can handle -20 on idle).


i think i might do it... i just might....

have you seen anyone fitting a whole waterchilling system into a full tower case? because i think i can do it the dimensions of the case are 27.1x24.2x7.5 inches. and thats pretty damn big compared to the stacker and other cases.

only problem is by doing this. i could only fit 1 or 2 hard-drives (i was hoping to fit 3, but i can sell off my old ones) and i would have to get rid of the floppy drive. and one of the cd-drives. i could just mount the hard-drives along the top of the case with the PSU.

that would leave... if the mobo is 9.6 inches.... 17.5 inches for the phase change. if i put it in a system which is taller than long. so minus some space for the cd-drives..... 20 inches or so upwards. it would be a tight fit. i could fit the radiator on the side fan. only problem is that fan would have to run 24/7 then (which is kinda loud, but i can handle it). but if i do that then the routing of the loop might get complicated as once insulated the pipes will be a bit larger than 3/4 of an inch OD. might have to get a bit fancy when turning curves. but honestly at -20 i could care less about restriction. so 90 degree angles should be of no problem. the d5 should handle it.


(sorry for the thread steal)

but just to try to stay on topic. after seeing the facts. i think peltiers are definatly out. phase change is in.
 
speed bump said:
Those parts on the apogee are made out of Delrin which is good for atleast -40°c. Some of my friends that used it for Antenna covers for an airport survailence project say that it was totally fine down to -60°c.

Thats conditional. If it shrinks at a different enough rate then the rest of the block the block will leak. Just because something doesn't crack easily from impact/vibration/stress at a given temperature doesn't mean it will work in the context of the rest of the system.

However I think darkcow should definitely try. Just try testing before you put it on your pc for a few days and see what happens.

darkcow said:
have you seen anyone fitting a whole waterchilling system into a full tower case?

Yes, I have seen a few chillers in the lian li V2000 type cases.

darkcow said:
i think peltiers are definatly out. phase change is in.

Welcome to the dark side ;)
 
well, i think it can't hurt all that much to try it. i mean... there is the posibility of cracking the plastic. the plasitc top is actualy bigger than the copper bottem. the question is, how much faster would the copper expand than the delrin. i don't think it could be much.

but to be on the safe side. i think i would put some epoxy around the edge of the thing because i can imagine if there is some rubber on the inside of the pump it could shrivl up.

but i kinda don't want to destroy the block. i guess i could sell it off again. and try and find a used swiftech 6002. im sure someone would probably want to trade. but i think im going to try it anyways. i've seen people with dangerden waterblocks (which are acrylic) in a super chilled state.


but honestly, to be on the safe side. i think im going to put the chiller on so the water only gets to -10c and no more. i'd rather save money on not having to by a new pump and block. then getting -20c
 
Back