PDA

View Full Version : Extreme TEC Cooling


VulgarHandle
09-22-2005, 04:35 PM
Ok, this is an idea i've been cooking up for a while now.

this will involve:
2 single-bay resevoirs
4 copper heatsinks
2 TEC pads(62x62x4.6mm, 14Amp, 120w, max15.4v)
2 pumps
1 radiator (120x240mm)
3 120mm fans
1/2inch ID tubing
1/4inch ID tubing
Power Supply(will have to do research)
and a whole slew of connectors, crimps, thermal wrap, insulation pads, etc....

Cost projected to be about $400USD

the idea:
the 4 copper heatsinks will be used in 2's, two heatsinks back-to-back w/ a tec pad inbetween, totaling 2 units

the bottom of one resevoir, and the top of the other will have 2 holes cut into it, each the size of the heatsinks

the 'cold side' of each unit will go into the top of the bottom resevoir, and the 'hot side' of each unit will go into the bottom of the top resevoir

gettin the idea yet?

now, the 120x240mm radiator will be for the 'hot' loop, each loop will have it's own pump

the cold loop, of course, will be for the vid card, the chipset, and the mosfets

Hot Loop:
Radiator--->Pump--->Resevoir--->Back to Radiator, all 1/2inch ID tubing

Cold Loop:
Pump--->Resevoir--->Split from 1/2inch ID tubing to (2) 1/4inch ID tubing, 1 tube to chipset and mosfets, other tube to vid card--->Rejoin to 1/2inch ID tubing--->Back to Pump

I plan to thermal wrap the cold loop and the resevoirs, as well as insulate between the 2 resevoirs.

My plans fits it all in a Thermaltake Alluminum Xaser III Super Tower, with some mods done for duct work for radiators and added power supply.

i'm plannin on vapochill for the cpu btw

Input, ideas, and suggestions welcome.

EDITED SETUP

Bloody_Sorcerer
09-22-2005, 04:52 PM
a 500watt powersupply can't physically put out 60amps at 15volts. 60a*15v=900watts.
Why would you have a radiator on the cold loop?
a 2x120 rad won't be enough for 4 TECs + their load (whatever is getting chilled).
It'd be much cheaper (both on your energy bill and in your general project cost) and probably more effective to just buy/build a compressor-driven chiller.
Also, this thread belongs in either the TEC forums or chilled liquid forums

VulgarHandle
09-22-2005, 05:18 PM
and this is why i love xs :D

i don't think i could fit a compressor system in the case

do you think 2 tec's would chill the water enough? thus dropping the amps in half

figured the radiator on cold loop would help to cool the water down a bit before hitting the resevoir to help take load off the tec's(maybe more necessary if using just 2 tec modules)

can i move this thread, or do admins need to move it?

Holst
09-23-2005, 09:02 AM
2x120 rad won't be enough for 4 TECs + their load

I think your wrong here.

With a shroud and some decent fans that rad will EASILY keep the warm side loop just above ambient.

My major concern in this project is the heat exchangers.

How will you stop it from leaking?

How will you direct the water flow so that each heatsink is propperly cooled?

I think this will idea is fairly good.

The PSu you have is no good. - 1 500w power supply w/ 60amps on a 15v line

What you will probably see here is that the voltage will drop to about 9 volts (500watts/56amps = 8.9volts)
Or you might just blow up the PSU!!!!!!!!!!!!!!!!!

n00b 0f l337
09-23-2005, 09:29 AM
PSU will blow for sure. Putting heatsinks that are being cooled in the water loop doesnt work. If you read the other tec chiller thread idea's you will get a better understanding I think.

VulgarHandle
09-23-2005, 03:27 PM
attached is a drawing(using my 1337 paint skillz) of the resevoirs using only 2 tec modules w/ 4 heatsinks

leaking, heh, seal it w/ caulking of some-sort of course

what i didn't draw is the insulation that would be between the resevoirs, and the thermal wrap around the resevoirs

EDITED THE SETUP TO 4 HEATSINKS, 2 TEC MODULES, JUST 1 RADIATOR(THE 120x240mm FOR "HOT LOOP"), AND THE POWER SUPPLY

VulgarHandle
09-23-2005, 07:42 PM
ok, been doin some reading, seems my current drawing would do nothing of signifcance as pertaining to chilling the water

have to come up with a design that slows the water enough to cool it down, w/out restricting the flow so far down that even if the water is cooled, it dosn't flow enough to cool the compoments...

while the hot loop may be close to suffecient enough to cool the hot sinks, the heatsinks too would need 'beefing up'

maybe a copper block that fits in each side of the resevoir minus the turn around point, with an intricate path or just drill many lines through the block....

or maybe i'm going about it all wrong... hmmm, more thought needed, bong time

VulgarHandle
09-23-2005, 08:18 PM
would something of this nature be effecient? if maybe 2 are used? 1/4 inch ID tube

http://x4.putfile.com/9/26522190199-thumb.jpg (http://putfile.com/pic.php?pic=9/26522190199.jpg&s=x4)

Holst
09-24-2005, 01:43 AM
No I dont think that will work at at well.

Best bet is to clamp waterblocks against each other. But thats expensive unless you can make your own.

VulgarHandle
09-24-2005, 01:47 AM
ok, how's about this, 2 plates, 1 flat, 1 w/ beaten path, 2 226w TEC modules w/ waterblocks

http://x4.putfile.com/9/26603471947-thumb.jpg (http://putfile.com/pic.php?pic=9/26603471947.jpg&s=x4)

Holst
09-24-2005, 02:01 AM
4 waterblocks will be much better.

Very large area to insulate there and the vast majority of it will do nothing for your temps.

You seem to think that you have to keep the water in contact with the TEC blocks for a long time... you dont.
Providing that your getting a good transfer of heat from the water into the TEC blocks (cold side) then your temps will go down untill you hit equilibrium.

Using more lesser wattage TEC will help your efficency with this system.
4x120watt will be more effective than 2x226 IMO.
Problem with that is that 8 waterblocks are expensive!

Starting to understand why an effective TEC chiller is hard to make now?

Most cost effective method will be to refine your original idea to use heatsinks to cool the water. I think this can work pretty good if you come up with a better design.

VulgarHandle
09-24-2005, 02:20 AM
well, guess i should give some more insight on my plan

the cpu will be vapochilled, not exactly cheap

this chill system is intended for a single video card, mosfets, and chipset, again, 3 water blocks, not exactly cheap

aside from gettin stupid on the cost, the point is to make an effective chilled water system that can be contained neatly inside the case, which will be sittin on the vapochill LS of course, what case that is is undecided, obviously one that fits dimensions of the vc ls

now, the other idea some ppl have been pushing is evap on these as well, this sounds much more expensive and complicated, which is why i've decided on TEC's

TEC on-di on the mosfets sounds complicated too, considering the insulation aspect

I guess i could do on-di for the vid card and maybe the chipset, and just make sure the mosfets are first in line in the loop.....

On the other hand, i just thought it'd be cool to setup a compact, and effective chilled water system, and 2 evaps don't meet the compact part.

now, maybe just evap on a large resevoir and waterblock the cpu too..... u think that would be effective enough for even -1C on cpu, and still effectively cool the vid card/chipset/mosfets? seems it would be hard to hit -1C on cpu like that

and dangit, i really liked the plate idea, lol :wiggle:

VulgarHandle
09-25-2005, 03:12 PM
Ok guys, how about this

(2) 1/2 inch thick aluminum blocks with 1/4 inch deep paths(w/ milling machine)
(1) 1/8 inch thick aluminum plate to be between the 2 blocks
(4) water blocks sitting on 80w TEC modules

http://x2.putfile.com/9/26719534684-thumb.jpg (http://putfile.com/pic.php?pic=9/26719534684.jpg&s=x2)

n00b 0f l337
09-25-2005, 03:22 PM
Yup thats basically how I'm going to do mine. But that much coppers going to get expensive. And your going to need a hell of a 2nd waterloop.

VulgarHandle
09-25-2005, 04:29 PM
yay, my 1st positive responce, now we're getting somewhere

yeah, the hot water loop will be fun, and at the same time, why i chose 80w

as far as the cost of the copper, this could be true, and i'll probably opt for aluminum, i could imagine when cooling the water is concerned, aluminum probably won't be beat by copper all that much

it's cooling the alluminum that might be more significantly hindered

so for the sake of it all, i might just do the water blocks as alluminum too

i think a heater core would be sufficient enough for the hot loop radiator

as far as the cold loop, won't know till i build it how effective it is, and wether or not a radiator on the cold loop before it goes into the cooler would help, i already have a black ice extreme

i think this will be the design i'll go for first

draw up some plans in autocad, take em to a local machine shop(since i don't have a milling machine lying around, lol), and if too much, i think i can go back to my old college and see if my old teacher will let me get on one of their machines(or ask him to do it :D )

n00b 0f l337
09-25-2005, 04:57 PM
Cant mix metals without erosion. So either get a copper cooled side with acopper block or make it alluminum and get an aluminum block somewere.

VulgarHandle
09-25-2005, 05:20 PM
i'll be doing my own water-blocks, like i said, i'll do all aluminum, guess i'll redo the drawing to show this

Holst
09-26-2005, 03:44 AM
Those blocks look very good :D

You can insulate it failrly easily, just wrap the entire lot up with neoprene foam :D

You might want to consider running the sytem the opposite way around..

So you make the hot sides of the TEC run into the alu block and the cold sides into the copper.
Might be slightly less efficent this way but you can then use copper CPU and GPU blocks.
Plus you can limit the corrosion issues with the aluminium if you use an aluminium rad as well for the hot side.

Better to go all copper but that seems an ok compromise to me.

n00b 0f l337
09-26-2005, 11:50 AM
You may also be able to consider independent air cooling if you use a lot of tecs at are 40 watts each. The only thing is water or air your ambients are going to quickly start rising. VERY quickly. Thats what tec's these days are better wtih chilled, and normally the chilled is better than the tec's efficiency.

VulgarHandle
09-26-2005, 04:16 PM
yeah, room temperature will be gettin warm

I plan to duct all exhaust outside, yet another plan to come up with.

I do plan to wrap the unit in neoprene, and all cold loop tubing.

Do you think I'll need to wrap the blocks that will be on the components?

If so, would it be dangerous to wrap around the mosfets? I mean, if cold enough, won't i need to grease around the mosfets, and might this cause problems?

VulgarHandle
09-26-2005, 04:30 PM
You might want to consider running the sytem the opposite way around..

So you make the hot sides of the TEC run into the alu block and the cold sides into the copper.
Might be slightly less efficent this way but you can then use copper CPU and GPU blocks.
Plus you can limit the corrosion issues with the aluminium if you use an aluminium rad as well for the hot side.

you think the main block itself would be a good enough cooler for the 4 blocks if i do reverse it?

I could see an advantage to this, as i could have, say, 2 blocks for the biggest heat producer(i'm thinking the video card), and the other 2 blocks on the 2 lesser heat producers(i'm thinking the mosfets and chipset)

guess it'll be something i test

i'd love to do all copper, might just bite the bullet and do it :D