QX6700 TEC installation

[alexankius]

hey,

I have 2 226 watt tec's lying around from the athlon days..

Do you think that both of them combined would do the trick with an ocked QX??

And if so one on top of the other or both combined on the same waterblock?

By the way i can't seem to be able to find tec's larger than 226 watts? any links???

[p8ntslinger676]

no, bad idea. but you could use them on your graphics cards :D

[littleowl]

hey,

I have 2 226 watt tec's lying around from the athlon days..

Do you think that both of them combined would do the trick with an ocked QX??

bad idea IMO. The heat off of a QX is to much for a tec. You would be better to go with a chiller or ss.



And if so one on top of the other or both combined on the same waterblock?

By the way i can't seem to be able to find tec's larger than 226 watts? any links???

Good luck on finding anything larger then a 226 but you can look on ebay or search for the email of wince labs (wrong name) ask NOL where to get them he can tell you.

As for stacking the 226w tec you would have to have 3 psu's running that system. 1- pc 2- half volted tec 3- full volted tec

[cruise]

I have a 320 watt 50mm by 50mm, I know the ideal place is on the GPU but I was wondering what would be the best way to mount it? What WB?

I stupidly brought two danger den 8800GTX blocks, they are probably not suitable for the tec.

[Cronos]

littleowl;2209769']
As for stacking the 226w tec you would have to have 3 psu's running that system. 1- pc 2- half volted tec 3- full volted tec

This can't be done.
Half-volted TEC will have 1/2Qmax cooling power. So, two 226w TECs stacked will have 113W cooling power (and 375W consumption). Lower TEC will work as thermal isolator for any CPU with power dissipation >=113W.

[Holst]

hey,

I have 2 226 watt tec's lying around from the athlon days..

Do you think that both of them combined would do the trick with an ocked QX??

And if so one on top of the other or both combined on the same waterblock?

By the way i can't seem to be able to find tec's larger than 226 watts? any links???

One ontop of each other wont work at all.
Stackng TEC reduces the wattage they can cool, but increases temperature differential.

But, you could put them in paralell and it would probably work pretty good, even on quad core.
Paralell will double the possible wattage to cool, which is what you need with quad core.

Problem with this is that you need a special mount and coldplate to mount two 220wat TECs onto one CPU.
Some guys did do this in socket A days and I have been considering making something similar for fun when I get back to the UK.
But unless you can make things yourself this will be difficult.

[Cronos]

The only practical way to use several TECs is parallel is water chiller.

[alexankius]

thanks for all replies. :)

I think I'll go the chiller root for my cpu; I bought a normal water chiller that I integrated as the reservoir in my watercooling system. It currently has a thermostat at 5 celcius but I will change that to about -10; I need to find the right fluids.. any suggestions?

I'll use one of the tec's to cooool my GPU; this should be sufficient (??) :confused: I will try and mount it on the swiftech 8800GTX Waterblock. I will post pics if i manage to do it. :D

The other tec's i'll try and sell...

The parallel idea seems interesting but how would it work in theory as the tec's are 50x50x3.10 and therefore 2 of them would be at least 110x50x3.10, which I suppose means you have to sandwitch the tec's in between two copper plates, one of them being 110x50x3.10 and the other being the dimmensions of the cpu.

The problem would be that in most 775 mobo's you wouldn't find 10cm's in any direction so as to fit the damn thing.. Unless i've got something wrong in my calculations

By the way the tec's are potted. I need to clear the surface of the tec from the old thermal paste. I'm thinking of using ArcticClean 1 & 2. Any reasons why i shouldn't do it?

[Cronos]

Actually, the best advice is not to use TECs at all.
You will be disappointed as almost everybody else.

[alexankius]

Actually, the best advice is not to use TECs at all.
You will be disappointed as almost everybody else.

You say this because of the hastle / power requirements of the tec installation or because of the results??

Because I would have thought that a 226watt tec for a 8800gtx card should do more than fine (the hot side of the tec will be water chilled) ????

[littleowl]

This can't be done.
Half-volted TEC will have 1/2Qmax cooling power. So, two 226w TECs stacked will have 113W cooling power (and 375W consumption). Lower TEC will work as thermal isolator for any CPU with power dissipation >=113W.
very true you will not get the cooling you are desiring at all but yes I have seen this done.


The only practical way to use several TECs is parallel is water chiller.

That is not good advice either because tec chillers have been proven not to work very well.

[littleowl]

You say this because of the hastle / power requirements of the tec installation or because of the results??

Because I would have thought that a 226watt tec for a 8800gtx card should do more than fine (the hot side of the tec will be water chilled) ????

a 226W on a 8800gtx would be sweet. the problem is you would have to make your own waterblock and cold plate. you will not find one premade to fit the 50mmX50mm tec.

[Cronos]

littleowl;2211086']
That is not good advice either because tec chillers have been proven not to work very well.

Properly designed, TEC chiller is the only TEC-based device wich can be made at least reasonably effective. It is possible to reach COP ~1, 20-30C water cooling with ~300-400 total cooling power. This will require 9x 300W TECs, undervolted to 1/4-1/3 maximum voltage. I had the idea, but eventually decided it is not worth investment.

[alexankius]

I will try and make a block; if i get around to doing it i will post pics; it should work very nice.

[Fr3ak]

Properly designed, TEC chiller is the only TEC-based device wich can be made at least reasonably effective. It is possible to reach COP ~1, 20-30C water cooling with ~300-400 total cooling power. This will require 9x 300W TECs, undervolted to 1/4-1/3 maximum voltage. I had the idea, but eventually decided it is not worth investment.

That sounds interesting, but wouldnt be a compressor-based chiller be the better option power consumption and temp wise? Cost of the unit would be very similar I suppose.

[ViperJohn]

This can't be done.
Half-volted TEC will have 1/2Qmax cooling power. So, two 226w TECs stacked will have 113W cooling power (and 375W consumption). Lower TEC will work as thermal isolator for any CPU with power dissipation >=113W.

You forgot your Watt's Law. When you halve the voltage you also halve the current so 1/2 voltage = 1/4 Qmax.

Viper

[Cronos]

You forgot your Watt's Law. When you halve the voltage you also halve the current so 1/2 voltage = 1/4 Qmax.

Viper

No, i am right.
The sticky here is screwed up in the TEC Q (cooling power) calculation.
TECs cooling power is linearly proportional to current,
Q ~ Qmax*I/Imax ~ Qmax*V/Vmax

1/2 voltage ~1/2 current ~ 1/2Qmax cooling power.

But power consumption W is proportional to V*I, so it drops quadratically,

W~V*I ~V^2

1/2 voltage ~1/4 power consumption.

This is the reason for increased TEC efficiency at lower voltage.

[ViperJohn]

No, i am right.
The sticky here is screwed up in the TEC Q (cooling power) calculation.
TECs cooling power is linearly proportional to current,
Q ~ Qmax*I/Imax ~ Qmax*V/Vmax

1/2 voltage ~1/2 current ~ 1/2Qmax cooling power.

But power consumption W is proportional to V*I, so it drops quadratically,

W~V*I ~V^2

1/2 voltage ~1/4 power consumption.

This is the reason for increased TEC efficiency at lower voltage.

Sorry you are incorrect. A 226.1w Qmax TEC at a Vmax of 15.2 and an Imax of 24.0 would draw 364.8 input
watts for an input power to Qpumped efficiency of 61.98%.

If you drop the TEC input volts to 7.6 the input current will fall to 12A and the input powerwill fall to 91.2 watts.
Now going by your theory Qmax would now be 113.05 making for an input power to Qpumped efficiency
of 123.96% which my friend is impossible lol.

Input power and Q lower at the square of the ratio on the voltage change. The reason for the increase in heat
pumping effciency, which only increases slightly as you reduce voltage, is due to lower couple self heating
from smaller I/R losses at reduced voltage.

Watts are watts and it doesn't matter if it is power or heat. They all are locked into abiding by watts law and
physics 101.

Viper

[Cronos]

Sorry you are incorrect. A 226.1w Qmax TEC at a Vmax of 15.2 and an Imax of 24.0 would draw 364.8 input
watts for an input power to Qpumped efficiency of 61.98%.

If you drop the TEC input volts to 7.6 the input current will fall to 12A and the input powerwill fall to 91.2 watts.
Now going by your theory Qmax would now be 113.05 making for an input power to Qpumped efficiency
of 123.96% which my friend is impossible lol.


Sorry, but you have to educate itself a little bit in the area. Efficiency of heat pumps (coefficient of performance or COP fort short) can be more that 1, and in theory can be arbitrary high (depending on temperature difference). All air conditioners operate with COP higher than 1. It is not against thermodynamic laws :)

By the way, your calculation is correct, with one important correction -the Qmax value correspond to 0C difference between hot and cold side. And if you
want to cool something with TEC, usually you will be taking 1/2 Tmax for cooling power estimation, which will give you 1/2 of cooling power at 0C(if you know how TECs work).
So, for your 226W TEC at 7.6V and 1/2Tmax =35C difference between hot and cold side, you will have 1/2 *1/2 *226W=57W only cooling power and 1/4*365=91W consumption, which amounts to 0.62 COP.

To reach COP 1 at 1/2Tmax, which was my ultimate goal, you'll need further decrease voltage, for 1/3Vmax or lower.
This will give 1/3*1/2*226W=38W cooling power
and 1/9*365W=41W power consumption, COP~0.93.
Actually, the voltage should be decreased even further. As you see, we need
many TECs if we want to achieve COP close to 1 and still have enough cooling power, and this can only be achieved with water chiller, hence my initial statement.

But, then you realize that the same temperature difference can be achieved with small phase change device with COP at least 3 (properly designed), you see how terribly inefficient TECs are in the first place!

[littleowl]

Properly designed, TEC chiller is the only TEC-based device wich can be made at least reasonably effective. It is possible to reach COP ~1, 20-30C water cooling with ~300-400 total cooling power. This will require 9x 300W TECs, undervolted to 1/4-1/3 maximum voltage. I had the idea, but eventually decided it is not worth investment.


That sounds interesting, but wouldnt be a compressor-based chiller be the better option power consumption and temp wise? Cost of the unit would be very similar I suppose.

Yeah I would like to see a few people do this and keep up with the electric bill that comes along with it!!!!!!!!!!!!!!!!
I could have 3 chillers or SS's running and still be better then what your going to get with this idea.

[Holst]

The only read advantage of TEC chiller is that it is significantly easyer to build than phase change.

For phase change you need gas, torches and guages.. plus the knoledge to build it.

For TEC you just need waterblocks and a power supply, practically anybody can make a TEC chiller. You just need the money.

[Cronos]

Yes, the main challenge in building phase change device is pipe connection, thats why we don't see everyone building them.

And TEC chillers are so inefficient that we don't see many of them either (even less, actually)

[L33T]

Good luck on finding anything larger then a 226 but you can look on ebay or search for the email of wince labs (wrong name) ask NOL where to get them he can tell you.

As for stacking the 226w tec you would have to have 3 psu's running that system. 1- pc 2- half volted tec 3- full volted tec

The Wince Labs (I'm not sure how to spell it either hehehe) doesn't sell them too often I believe -- they're always out of stock. I've called them three months ago and still have gotten no answer and no eta. Anyway, FrozenCPU does sell the 437W TEC.... $49.99 + shipping. They're basically the only place that carries them now.

[Burn]

Wintsch Labs has been making TEC waterblocks for years now- I bought an Arctic Web off them a year or so back, and it works wonderfully. I love how everyone jumps down the throat of someone who is trying to make something work, that's always nice :)

I wouldn't use two TEC's to try to cool of a QX6700, I would just buy a 437W peltier and jack it up to 24V. At that point you should be able to get some decent temperatures and still maintain electrical efficiency. One peltier is inefficient enough, just think of what two would do!

And Devin (Owner, Wintsch Labs) is bringing the 'Web back into the market. I would stay tuned for some updates, I hope he will be around more often.