general question about a tec, heat generation

[gbomb944]

If I have a tec cooling a stream of water that is more or less ambient and the tech is powered at 72watts, should I expect the hot side to generate around 200 watts of heat? I'm guesstimating using the 30% efficiency rule that I've heard used a few times.

I'm thinking of having like four tecs chilling my cpu loop so if my cpu produces say 100w that would give me about 50w of excess per tec. Would it do anything?

[gbomb944]

I'm thinking 12v x 6 amps for the tecs, any psu suggestions would be appreciated also. Cheap being the primary consideration.

[ls7corvete]

12*6=72*4=288w fromt he tecs, plus the 100w from the cpu=388w total

288W of TEC power cooling 100W is pretty reasonable, you have a am64 pic on your sig so I doubt you are actually putting out 100w unless you are dual core.

4 TECs give you lots of options, you can run in series and use a 48v 6a PSU. You can also run them parrallel off of 12-15 then switch to series on 12-15 to drop the voltage to each TEC for idle or when your not trying for extreme clocks.

I would browse the surplus electronics sites(mpja, goldmine, all-electronics etc.) and see what you can find in 12v/15v/24v/48v PSUs, go with what you find.

[Holst]

corvette, please remember that 288watts is the output heat NOT the pumping capacity. I also think that your estimate of 6amps at 12V is VERY innacurate!!! , TEC are normally WAY WAY more inefficent than that.

http://www.crazypc.com/Merchant2/merchant.mv?Screen=PROD&Store_Code=120199&Product_Code=5910
This is the only 72watt TEC I could find easily, and thats rated at Vmax 14.4v Imax 8.5amps.
This gives maximum heat output of 14.4x8.5 122.4watts, four of these give 489.6watts..!!!
Quite a bit more than the 288 that you esimated, and at a much more awquard 14.4volts.

The max theoretical cooling capacity will be 75x4 = 300watts, but thats at 0*c deltaT
I normally recomend that you halve this to get decent temps, but thats still ample at 150watts

I think this system will work pretty good if you can keep the TEC hot sides cool.
Even at 12v you should still get ok performance.
Personally I would buy an ajustable 15v PSU to give you more control over the system.

Please note that your TEC may be 12v Vmax, and as such will have higher Imax value .. you will need to research this before you buy a PSU, or just overspecify things by allot :)

Good luck.

You have the makings of a good TEC chiller here, but its a challenge to get it working good.

[Holst]

I just re-read your post.

Cheap being the primary consideration.

this wont be "cheap" unless you make the blocks yourself.

A 12v PSU that can run 40amps can be had fairly cheaply on ebay. Then you will have to insulate your CPU block and tubing as well.
2 pumps are not too expensive, neither is a big rad to cool them

But 8 waterblocks will be expensive. Homemade is the way to go with this part.

This will be noisy as well. potentially 500watts to cool, although it shouldnt be that high, but at least 350 IMO.

[ls7corvete]

corvette, please remember that 288watts is the output heat NOT the pumping capacity. I also think that your estimate of 6amps at 12V is VERY innacurate!!! , TEC are normally WAY WAY more inefficent than that.

http://www.crazypc.com/Merchant2/merchant.mv?Screen=PROD&Store_Code=120199&Product_Code=5910
This is the only 72watt TEC I could find easily, and thats rated at Vmax 14.4v Imax 8.5amps.
This gives maximum heat output of 14.4x8.5 122.4watts, four of these give 489.6watts..!!!
Quite a bit more than the 288 that you esimated, and at a much more awquard 14.4volts.

The max theoretical cooling capacity will be 75x4 = 300watts, but thats at 0*c deltaT
I normally recomend that you halve this to get decent temps, but thats still ample at 150watts

I think this system will work pretty good if you can keep the TEC hot sides cool.
Even at 12v you should still get ok performance.
Personally I would buy an ajustable 15v PSU to give you more control over the system.

Please note that your TEC may be 12v Vmax, and as such will have higher Imax value .. you will need to research this before you buy a PSU, or just overspecify things by allot :)

Good luck.

You have the makings of a good TEC chiller here, but its a challenge to get it working good.

I agree but....

If he says they draw 6a at 12v then I will figure accordingly, not gonna be searching the net for what chips he really has to come up with calculations. Not gonna be working backwards to estimate pumping power from input power for a simple question like this(he didnt ask about cooling capacity).

could either be a 72w unit(most likley) like you said (14.4v 8.5a) or a smaller unit that draws 72w at 12v and 6a like OP said(if OP says something I usually use that data). Either way makes little difference to the question.

[gbomb944]

I can make homemade blocks for about a buck each. If I ever get my digicam ready I may post a work log of it as it may help those with ghetto tendencies save some $.

I'm planning to use 4x of these http://www.frostytech.com/articleview.cfm?articleID=1568 on the hot side of each tec. In the review it kept 1oo w pretty chilly. If you go around the site and look at a newer review he tried it on a 125w source and it kept it 18c which I guess was his ambient.

[ls7corvete]

I can make homemade blocks for about a buck each. If I ever get my digicam ready I may post a work log of it as it may help those with ghetto tendencies save some $.

I'm planning to use 4x of these http://www.frostytech.com/articleview.cfm?articleID=1568 on the hot side of each tec. In the review it kept 1oo w pretty chilly. If you go around the site and look at a newer review he tried it on a 125w source and it kept it 18c which I guess was his ambient.

I havent read it yet but Im pretty sure the dT was 18c, ie 18 above ambient.

Your gonna use a 1$ water blocks for the cold side but use a HSF for the hot side?

Why not water cool the hot side for 1$ per waterblock?

heck, Ill give you 2$ for a water block :toast:, 100% profit for ya, lol.

Anyways, neat project, post some pics so we can see how it goes.

[Holst]

The DT at 100watts is 24.4 *c above ambient using that heatsink

So at 140 watts I estimate that temps would be around 30*c , possibly slightly better (heatsink will become more efficent the higher the wattage)
140watts is the TEC + approx 1/4 cpu wattage.

Thats pretty poor hot side temperature.

I doubt that you will gain very much from running the system in this way.

Hard to say exactly what temps you will get, but suffice to say that they will not be below ambient.

my watercooling the TEC you should be able to get the hot side temps down to around 10*c above ambient.
Thats an instant 20*c drop in temps.

Basically what im saying is that aircooling TEC of this wattage is fairly pointless.

If you can get the wattage per heatsink down to about 50watts (including CPU) then you can aircool and get good temps. But thats going to take ALLOT of heatsinks and low voltage TEC's

Im not saying dont build this ... im saying if your going to do it your way dont expect it to work too good!

[Holst]

And if you can post a design for a working $1 TEC waterblock ill send you $10 to cover your costs.
And ill build myself a TEC chiller as well just for the fun of it... i think i have everything but the waterblocks allready.

Working meaning I can make one at home and it will work, for only $1

[gbomb944]

They are a variation on the mt. dew can block but they use little lexan jars from wal-mart instead of a pop can bottom. The fittings are just pipe instead of commercial barbs held in with epoxy. Nylon tie-locks hold the tube to the fitting and usally 2 are all you need per barb. The bottom is aluminum that I got from an aluminum yard stick from home depot. I just cut a square and lapped it then lapped the little jar and stuck it on with silicone. I used it for six months or so before my mobo went. Gave pretty good temps with a danner mag 7 so it's definitely high flow rated.

[Bloody_Sorcerer]

powerdraw has nothing to do with heat generation... I'm fairly certain that the only heat a peltier generates is its QMax that it is always pumping, in addition to the wattage load on the side getting pumped from; they're just ludicrously innefficient and take loads of power to do this pumping

[ls7corvete]

powerdraw has nothing to do with heat generation...

Um, how not?

Power draw+ heat pumped from cold side=heat at hot side.

[n00b 0f l337]

Bloody corvetes right on this. Also, tecs at there maxes are inneficient. Once i finish getting the deisgns done (thanks overconfidence a lot) for the tec chiller, you'll see what tecs are good for..

[Bloody_Sorcerer]

hmm, yes, i suppose i'm just being dumb and illogical... long day...

[n00b 0f l337]

Dun worry I'm not done wiht you yet.
The days about to get a lot longer.

[Bloody_Sorcerer]

sounds like someone is gonna start pumping some custom stuff.... *wink*

[gbomb944]

An extremely easy way to water cool multiple tech's would be to go here http://www.onlinemetals.com/merchant.cfm?id=70&showunits=inches&step=2&sortby=od

Buy and appropriate piece of this channel. This is al which is what I would use but they may have copper if thats your thing.

All you'd have to do is cut one in half or buy two and you could fit an arseload of tecs, braze in some fittings on the end. Clamp them together with your pelts in the middle and boom you got a multi pelt hot and cold block. I was going to use this for my cold loop but I guess I'll spill the beans in case someone else wants to try it. I'm starting to wonder if I want to do mine or not. This definitely beats the fark out of having multiple blocks.

[n00b 0f l337]

Here's just a base pic of the design so far... The heatsink gray thing is there for size and it may be on the pelts....

[Holst]

Looks good noob but how will you make it?

That copper channel idea looks interesting though.

[Holst]

powerdraw has nothing to do with heat generation... I'm fairly certain that the only heat a peltier generates is its QMax that it is always pumping, in addition to the wattage load on the side getting pumped from; they're just ludicrously innefficient and take loads of power to do this pumping

Im sorry mate.

You need to reread the stickys.

Amperage drawn is determined by the TEC's design (duh!) and the temperature differential across the two sides.

The temp differential depends on the wattage you apply (CPU wattage)

At the TEC's max pumping wattage the temperature differential will be maximum and the ampage drawn by the TEC will be Imax (maximum)

With less than the TEC's max rated wattage the temperature differential will be increases, and drawn wattage decreased.

This can be quite a significant change in power draw when using TEC with low wattage heats scource (or a giant TEC)

[ls7corvete]

Here's just a base pic of the design so far... The heatsink gray thing is there for size and it may be on the pelts....

Looks good, The design I keep thinking about is a cross-hatched design like in this block:

http://forums.procooling.com/vbb/showthread.php?t=11318

for pelts I think it is in our best interest to increase the turbulence because of the large size of the water block, water moving through channels in a tiny WW style block is moving fast and cooling well. When you scale the channels up to a 50x50mm or 62x62mm size the flow is much slower.

Anyways, how are you getting these done?

[Bloody_Sorcerer]

Im sorry mate.

You need to reread the stickys.

Amperage drawn is determined by the TEC's design (duh!) and the temperature differential across the two sides.

The temp differential depends on the wattage you apply (CPU wattage)

At the TEC's max pumping wattage the temperature differential will be maximum and the ampage drawn by the TEC will be Imax (maximum)

With less than the TEC's max rated wattage the temperature differential will be increases, and drawn wattage decreased.

This can be quite a significant change in power draw when using TEC with low wattage heats scource (or a giant TEC)
yes yes yes i know... i had a brain fart dammit! :(

[n00b 0f l337]

The crosshatch design we've thought about. Its a matter of mass vs surface. :( Hard decision, we may try both.

[ls7corvete]

The crosshatch design we've thought about. Its a matter of mass vs surface. :( Hard decision, we may try both.

I would like to see that...

Plenty of cpu block information out there, not much out there to help someone looking into TECs. People need to look into stuff like this more for TECs so they can become more mainstream.

[n00b 0f l337]

Dun worry, I like tec's, there like kinky sex. You can think of your own description words.
Otherwise, I'm also working on a gpu block for 50mm tecs that can cool gpu and its ram or just gpu depending on what base you use.