What happens to Tec 172W 24V if 12V?

[Skee^]

yeah what happens.

i have a PSU that can give me 50A in 12V but now i want to use two 172W tec but i only see them with 24V.

what happesn if i put two of thoose to my 12V PSU, will it be half the performance or the same?

will this be better than a 226W tec?

[Since87]

You can use the Kryotherm software downloadable here to calculate the information you want depending on your system setup.

Because TEC's operate much more efficiently at voltages well below Vmax, You may get much better cooling with two 172 Watt TEC's at 12V than a single 226 Watt TEC at 12V. It is very dependent on your setup though.

If the setup is going to be CPU-Coldplate-TEC(s)-Waterblock, then the added distance the heat needs to travel in the coldplate, to reach the TEC's, may offset any efficiency gain.

[Skee^]

Thanks


one more questoin. if i dont want to use th TEC's can i just turn them of and use ordinary Watercooling while they are inbetween the cold plate and WB?

hope you understand

[Since87]

No, you can't just turn the TEC off.

TEC's have relatively low thermal conductivity. (Otherwise the heat pumped to the hotside would just conduct back to the coldside.)

Because of this, an unpowered TEC is effectively an insulator between the coldplate and the waterblock.

[Skee^]

i read that when going dow in voltage the Watts also going down.

"78 watts at 14 volts - at 12 volts, it is 67 watts "

I did not understand that Kryotherm ware hehe..

[Since87]

The wattage dissipated in the TEC will generally go down as the voltage goes down.

The current draw of the TEC is a function of the hot and cold side temperatures as well as the applied voltage though. It's not possible to accurately say what the current draw will be without knowing a lot of details such as:

Heat generated by the CPU.
C/W of the coldplate. (Including TIM joints.)
C/W of the watercooling system. (Including TIM joint.)
C/W of the heatpath down the CPU pins to the motherboard.
Ambient temperature.

With the Kryotherm software set up to model "a hypothetical water cooled TEC system" with two 172 Watt TEC's:

Vtec=12V, Itec=10.84A, Tcpu=12.9C
Vtec=16V, Itec=14.4A, Tcpu=10.4C
Vtec=20V, Itec=18.06A, Tcpu=11.3C
Vtec=24V, Itec=21.76A, Tcpu=15.1C

[Skee^]

BUt now i want to know

will two 172w@12V give me more cooling then one 226w@12v

will it give me double the power off 172w or 2/3 increase. any 1 know?

[Since87]

First of all, I assume you are talking about parallel pelts with Maze2-2 type watercooling. Stacked pelts used this way definitely won't work.

If you want to have an idea what goes into calculating this stuff, have a look at this thread over at overclockers.com. Be warned, it's a lot to wade through.

To sum up, there are a lot of unknowns that I can only make guesses at without doing a lot of testing and calculations with specific hardware. A lot of the unknowns are related to the C/W of the hot side cooling system and the C/W of the cold plate.

The cooling through the coldplate is going to be less efficient with dual 172's because of the increased distance the heat must travel from the CPU to the pelt.

The cooling provided by the Maze2-2 waterblock is going to be more efficient because of the larger surface area to be cooled in a dual 172 setup. (Of course if you are going single 226, a Maze3-1 will likely give better hot side cooling than a Maze2-2.)

The most important issue that affects the comparison of dual 172's at 12V vs a single 226 at 12V is that, the dual 172 watt pelt will consume about half the power of the single 226 and the dual 172's will have somewhat more cooling capacity (Q) but lower dT. The lower heatload on the watercooling system probably cancels out the loss of dT of the dual 172's at 12V for the majority of watercooling systems.

My best educated guess, (With a lot of help from the Kryotherm calculator and a spreadsheet I wrote.) is that the dual 172's at 12V will get the CPU about 2C colder than a 226 at 12V. If you run your system 24/7, the money saved on power will pay for the second 172 in much less than a year.

Also, if you decide you want more cooling, you can connect a 5V supply in series with your 12V supply, to power the dual 172's at 17V. Increasing the voltage of a 226 above 12V will increase the CPU temperature with most water cooling systems.

[Skulemate]

Here's a quick question for the pelt guys here. Skee^ asked about running the block with the pelts off, and you told him that won't work since the pelt is an insulator when it's off. Here's the question I have: could you set a voltage for the pelt that would allow it to adequately move the heat through it while maintaining a low delta T (i.e. simulate an "off" condition with a much lower voltage than normal operation)?

[Since87]

Originally posted by Skulemate
Here's the question I have: could you set a voltage for the pelt that would allow it to adequately move the heat through it while maintaining a low delta T (i.e. simulate an "off" condition with a much lower voltage than normal operation)?

Yes you can. (Depending on what you mean by "a much lower voltage".)

I've attached a screenshot of the Kryotherm calculator showing a 172 Watt pelt pumping 50 Watts of heat while powered with 2.76 Volts. The calculated deltaT is 0C and the pelt power consumption is 4.55 Watts.

I'm curious. Why might you want to?

[morephyous]

At 12 V for Dual 172W you will get in the 20C under load.

At 20-22V for Dual 172W you will get 0C under load.

[Since87]

Originally posted by morephyous
At 12 V for Dual 172W you will get in the 20C under load.

At 20-22V for Dual 172W you will get 0C under load.

Wow Morephyous, how do you know this?

I take it you know:

How Skee^ is going to be cooling the water in his water cooling system.

How many Watts his CPU is going to dissipate when overclocked to the maximum stable speed with pelt cooling.

The C/W of the Maze2-2 waterblock on dual 40mmX40mm pelts.

The C/W of the Maze2-2 coldplate when mounted between Skee^'s CPU and dual 40mmX40mm pelts.

The C/W of the heatpath down the CPU pins, and the temperature of the motherboard around the CPU socket.

etc. etc....

Please share this information with the rest of us, because this is extremely difficult information to come by.

I imagine Skee^ would be particularly glad to know how high an overclock he is going to get.

[Les]

Originally posted by Since87

..........
The C/W of the Maze2-2 coldplate when mounted between Skee^'s CPU and dual 40mmX40mm pelts.

The C/W of the heatpath down the CPU pins, and the temperature of the motherboard around the CPU socket.
.........

I am not actively engaged in playing with Peltiers at the moment.
However my currently favored Kryotherm parameters may interest :-
Rc=0.19 - 0.26 (for 40x40x6 mm Cold plate with a 10x10mm heatsource)
Based on:
1) C/W(TIM,Die/coldplate) ~0.1 - from Bill Adams work(e.g http://www.overclockers.com/articles654/index02.asp ) ,making an allowance of 0.05c/w for die thermocouple offset.
2) C/W(Coldplate) ~ 0.09 to ~ 0.16 -- from Waterloo calculations using,possibly mistakenly, an h=10,000 - 100,000W/m*m*c as corresponding to the conditions for a 172W Pelt cooled by a wb with a C/W=0.00625 - 0.0625(cooling a 40x40mm heat source).

Ri=1
Based on my current attemps to cross-correlate Bill Adams data with CPU(Diode)/"Radiate calculated Watts"

All are very iffy and only the thoughts of a dilettante.

The apparent,empirical relationdship between "Radiate Calculated Watts" and "In Socket Die Temps" for a variety of Peltier cooling platforms may interest.


The data is mainly from old (6months +) threads in a UK forum http://www.bigforums.co.uk:81/viewforum.php?f=18 .

[Skulemate]

Since87

Well, it was really just curiosity, more than anything else. Skee^ did ask if he could run with the pelt off... running at the voltage you gave would simulate that condition closely enough without drawing too much current, wouldn't it?

[morephyous]

Originally posted by Since87
Wow Morephyous, how do you know this?

Through observing it in practice, not theory.

[Since87]

Originally posted by morephyous
Through observing it in practice, not theory.

Wow, truly amazing. You've observed Skee^'s hypothetical cooling system in practice.

Too bad us mere mortals are limited to theorizing about hypothetical systems.

[morephyous]

I had a dual 172W system on an XP2000 and those were the results I got.

I'm now on a P4 system and will cool it down with only one 172W unit in the New Year when I'm on holidays.

For a dual 172W system you really need two radiators.

[bigben2k]

Originally posted by morephyous
I had a dual 172W system on an XP2000 and those were the results I got.

...

For a dual 172W system you really need two radiators.

Good show.

Which rad did you use?

[morephyous]

2 car heatercores

[Since87]

Originally posted by Les
The data is mainly from old (6months +) threads in a UK forum http://www.bigforums.co.uk:81/viewforum.php?f=18 .

Les,

Thanks for the data. Can you point me to the thread where this is discussed. I looked around the site. A lot of interesting stuff being discussed there.

I'm not following everything you wrote. I haven't been able to find a reference to "Waterloo calculations". Is the graphed data arrived at by running the processors at different frequencies to get different "radiate" Watts?

Having seen you mention it somewhere, I have been using 0.1C/W as a single pelt/die Rc. I had written a crude pelt calculator spreadsheet before I learned of the Kryotherm calculator. With it, I was able to match JoeC's test results for the Swifttech air cooled pelt fairly accurately using 0.1C/W for Rc, and 0.22C/W for Rh. (Compare to 0.26C/W for JoeC's die simulator measurement of the MCX462+ alone.)

[Since87]

Originally posted by Skulemate
Well, it was really just curiosity, more than anything else. Skee^ did ask if he could run with the pelt off... running at the voltage you gave would simulate that condition closely enough without drawing too much current, wouldn't it?

For the system I calculated with, at 2.76 volts (4.55 Watts) the system would behave almost like the pelt wasn't there. The voltage that "nulls out the pelt" is very dependent on all the system characteristics though.

Because of the low power involved, it wouldn't be hard to make a little circuit to maintain the deltaT across the pelt at 0 for a relatively wide range of conditions.

[Since87]

Originally posted by morephyous
I had a dual 172W system on an XP2000 and those were the results I got.

You really got a 20C spread when the only difference was whether the pelts were powered with 12V or 20V?

When you did this testing, did you allow enough time for all the temperatures to stabilize?

[morephyous]

The temps were stabilized.

The 172W modules don't work too well at 12V.

[Since87]

Sorry morephyous, I simply don't believe you.

There is no way this product could even come close to working unless peltier's can become much more efficient (in terms of heat pumped vs power conumed) at voltages well below Vmax. Your results are inconsistent with every reputable source of information I've seen.

Fortunately, I'll soon have things setup to where I can test this myself.

[bigben2k]

I believe Morephyous: all similar rigs I've seen have similar results, and the setup of 2 * 176 W pelts should be able to provide 0 or 20 deg C, depending on the power applied to the Pelts.

Since87: I think everyone can appreciate the work that you're doing here, but Les' graphs seriously contradict your statement, where you say that 2 * 176 might be 2 deg cooler than a single 220, where Les' graphs show a difference of 0 to 20 deg C, depending on who's rig is observed. (voltages being different though).

I think that there may be a lot more to this than you are looking at. Maybe you could explain how you figure that there would only be a 2 deg diff:

My best educated guess, (With a lot of help from the Kryotherm calculator and a spreadsheet I wrote.) is that the dual 172's at 12V will get the CPU about 2C colder than a 226 at 12V.

I can use Kryotherm too, so maybe we can try to relate?