Stacking T.E.C's

[4ghzduo]

Hey guys i doing some tests with T.E.Cs and am new to the game.

I have a 80W air cooled peltier that i am trying out on my old amd system. Now can i add another 80W on top of the other 80W under my cold plate? :confused:

Thanks.

[Gamer67]

Why would you want to do that ?

It is not going to make it any colder and you will use more power.

[[XC] MarioMaster]

problem is peltiers dump the heat they remove from the cold side, plus the power they consume so stacking two of the same wattage will lead to thermal runaway between them and essentially just breaks your tecs and whatever you mount it to, you have to make sure the bottom peltier can pump the heat transferred as well as the energy used to maintain proper cooling

[4ghzduo]

thanks for the quick response guys....well like i was saying new to TEC's :shrug:

The reason i wanted to do that is because the 80W isn't enough for the mobile chip. stock it sits at -8 but when i boot up and OC settings the temp starts climbing like a timer :down:

The cooler is an 80W on a air cooler with a delta fan which is plenty..i also have a spare 80W but with no heatsink just the TEC. I though if i stack them it'll act like a 160W...yes?????? just asking.

(The whole reason for this is to test the effects of condensation.)

[Gamer67]

IMO air cooling a tec is not too effective. Watercooling is best for it.

[4ghzduo]

very true. i was suprised how well this cooler worked on P3's overclocked and they still ran at -3 :) Had massive condensation issues under the cpu though. i applied die electric grease and no more condensation. :up:

If this works gonna try the new water cooled TEC from swiftech on my spare 6600.

[[XC] MarioMaster]

stacking does not double the power, as i said before a peltier's hotside not only has the heatload from the cold side but also heatdump from the energy it consumes

[4ghzduo]

MarioMaster;2326357']stacking does not double the power, as i said before a peltier's hotside not only has the heatload from the cold side but also heatdump from the energy it consumes

thanks guys for your help. only one thing to do......BIGGER TEC.:)

[littleowl]

If you want to just test condensation then you should just get a meanwell psu and a 226w tec off of ebay, a small water loop on it and have fun.

[cyberspyder]

IIRC, what you're talking about is a Peltier cascade---I think... Give this a read: http://www.konradhorn.de/seite4.htm
Brendan

[Stigma]

it IS possible to stack peltiers, but in order to get a useful positive result you have to stack them in a "pyramid" comfiguration. For example have 1 pelt right over the coldplate, then a larger coldplate over that that is again cooled by 2 peltiers of the same strength.

Stacking peltiers scales up very badly however, and in general you are better off just getting the single strongest peltier you can find. If you "need" to get even lower temps, you need to be looking at some phasecooling instead, because when you start to use 3+ strong peltiers in a cascade you are allready using more electricity to run it than a strong custombuilt singlestage phasecooler.

-Stigma

[initialised]

A stack like this could be made to get temperatures potentially as low as 200C below ambient.

62mm Pelt = 450W
50mm Pelt = 226W -70C
40mm Pelt = 120W -140C
40mm CPU = 100W -210C

Total power for the TEC stack = 900W
Total heat load = 1kW

Assuming that Tamb = 30C a set up like this could get down to -180C accounting for parasitic losses a more realistic figure would be 50C per TEC which could still push -120C without the need for an extremely complex cooling system and exotic materials.

Is the economics of TEC really that bad? How do these figures stack up against a condensor system when taken over 2 years of use including the cost of the initial setup?

[cyberspyder]

A stack like this could be made to get temperatures potentially as low as 200C below ambient.

62mm Pelt = 450W
50mm Pelt = 226W -70C
40mm Pelt = 120W -140C
40mm CPU = 100W -210C

Total power for the TEC stack = 900W
Total heat load = 1kW

Assuming that Tamb = 30C a set up like this could get down to -180C accounting for parasitic losses a more realistic figure would be 50C per TEC which could still push -120C without the need for an extremely complex cooling system and exotic materials.

Is the economics of TEC really that bad? How do these figures stack up against a condensor system when taken over 2 years of use including the cost of the initial setup?

In short, yes. Waste of money, waste of energy, and a waste of time. :shrug:

[[XC] MarioMaster]

old thread

[littleowl]

wow talk about raising the dead! :D

[initialised]

In short, yes. Waste of money, waste of energy, and a waste of time. :shrug:

OK so how much would a -150-180C capable condenser system cost, how big would it be and how much power would it use?

[D_o_S]

Hi guys,

I have a similar question... it is easy for me to get 40 W and 80W TECs cheap,and I would like to combine them... so, would it be possible to use the 80W tec I have, and put the cold side on top of a 40 W Tec, which would then go on the CPU? Or would the performance be worse?

[littleowl]

IMHO it wouldn't be worth the trouble! you would be better off looking on ebay and getting a 226w tec for 10$

[initialised]

Hi guys,

I have a similar question... it is easy for me to get 40 W and 80W TECs cheap,and I would like to combine them... so, would it be possible to use the 80W tec I have, and put the cold side on top of a 40 W Tec, which would then go on the CPU? Or would the performance be worse?

That would work to get down to -100C but only if the chip you're cooling emits less than 40W so it'd be ok for a 35W TDP CPU. The problem comes when you want to overclock as P=f(V^2)/2 so anything more than a 10% OC would push it beyond to TEC stacks capability. It could be an effective way of cooling a north or southbridge or VRM.

[Duh]

IMHO it wouldn't be worth the trouble! you would be better off looking on ebay and getting a 226w tec for 10$

Id rather get a 437 for $50 and forget about it :cool:

[n00b 0f l337]

That would work to get down to -100C but only if the chip you're cooling emits less than 40W so it'd be ok for a 35W TDP CPU. The problem comes when you want to overclock as P=f(V^2)/2 so anything more than a 10% OC would push it beyond to TEC stacks capability. It could be an effective way of cooling a north or southbridge or VRM.
From how your describing temperatures basically working 100% on a delta I dont think you understand the flaws of TEC's, nor have used them in this application. It takes 2-3 stages of TEC's to hold a solid load in the labs for -30C, much less -100C. There far from a perfect 69C delta. Far. From.

[littleowl]

Id rather get a 437 for $50 and forget about it :cool:

well that would be nice but seems how we don't have blocks that are 62mm X 62mm we will have to stick with a 50mm tec and that would be the 226.:up:

[Duh]

well that would be nice but seems how we don't have blocks that are 62mm X 62mm we will have to stick with a 50mm tec and that would be the 226.:up:
maybe you and me dont have 62x62 but you know they do exist although expensive ;)

[littleowl]

maybe you and me dont have 62x62 but you know they do exist although expensive ;)

I didn't say they didn't exist but it the real world I live in they are very hard to come by and not every one has a CNC sitting in there living room! :rofl:

[Duh]

CNC :confused:

no cnc, just $200
http://www.arcticspider.com/


cheers

J

[littleowl]

CNC :confused:

no cnc, just $200
http://www.arcticspider.com/


cheers

J

yes I know about the spider and again we are talking a cheap setup not one that cost a fortune! your talking 200 for a tec water block and 300 for psu and then electric cost not including neoprene and dielectric and conformal. verses a setup that will work just fine for 98% of chips out today that is 65 for clock and 10 for tec then another say 30 for cold plate. again not including conformal and neoprene and dielectric.

I am not rich I will go the cheap route!! :D

[pH(x)]

it IS possible to stack peltiers, but in order to get a useful positive result you have to stack them in a "pyramid" comfiguration. For example have 1 pelt right over the coldplate, then a larger coldplate over that that is again cooled by 2 peltiers of the same strength.

Stacking peltiers scales up very badly however, and in general you are better off just getting the single strongest peltier you can find. If you "need" to get even lower temps, you need to be looking at some phasecooling instead, because when you start to use 3+ strong peltiers in a cascade you are allready using more electricity to run it than a strong custombuilt singlestage phasecooler.

-Stigma
Very good info, Stigma!

That has made me reconsider carefully a cascade TEC setup :shakes:

[naPS]

yes I know about the spider and again we are talking a cheap setup not one that cost a fortune! your talking 200 for a tec water block and 300 for psu and then electric cost not including neoprene and dielectric and conformal. verses a setup that will work just fine for 98% of chips out today that is 65 for clock and 10 for tec then another say 30 for cold plate. again not including conformal and neoprene and dielectric.

I am not rich I will go the cheap route!! :D

Ummm... actually, your numbers are a little off. The $200 includes the neoprene and dielectric. Properly set up, you don't need the conformal on the bottom of the board if you use the neoprene instead.

PSU is $145, not $300.

[littleowl]

Ummm... actually, your numbers are a little off. The $200 includes the neoprene and dielectric. Properly set up, you don't need the conformal on the bottom of the board if you use the neoprene instead.

PSU is $145, not $300.

Right! well I want to back of my board air tight and you will not get neoprene to stay on the back of a motherboard without conformal or some other epoxy. The little tube of dielectric grease is not enough IMHO.
Also I was talking about the most common price for a meanwell not a disscounted one.