The one advantage I see to Tec's

[pur]

I have always had a soft spot for TEC's, there just so neat. So much cooling power in a small package. But as we can all see by the many posts in this forum, the biggest most complicated tec setups can only handle moderate overclocking, in comparison to phase change.


Right now I have a custom phase change unit I built myself, and I love it to death but it just so heavy! I'd love to go tec and have considered it many times, but I'm not going to be asking any of "those" questions here. This is just more thought I've been having.


Anyways, the only advantage tec's have over phase change that I can see is there portability. To add one to an existing (properly set up) WC loop is easier and possible in most full tower cases.

Would you guys agree?

[flak-spammer]

There are several advantages that TECs have...it's just that nobody plays to those advantages.

[pur]

Such as?

[Uncle Jimbo]

I have always had a soft spot for TEC's, there just so neat. So much cooling power in a small package. But as we can all see by the many posts in this forum, the biggest most complicated tec setups can only handle moderate overclocking, in comparison to phase change.


Right now I have a custom phase change unit I built myself, and I love it to death but it just so heavy! I'd love to go tec and have considered it many times, but I'm not going to be asking any of "those" questions here. This is just more thought I've been having.


Anyways, the only advantage tec's have over phase change that I can see is there portability. To add one to an existing (properly set up) WC loop is easier and possible in most full tower cases.

Would you guys agree?

I think you have it about right. With CPU power dropping (i7, Phenom II), a self-contained portable TEC solution becomes more attractive. The best solutions are when the goal is to maintain near ambient temps on the CPU while overclocking - that makes the installation easier (no need for moisture protection) and reduces power draw.

You can think of TECs used this way as kind of super heat pipes - they move heat from one place to another, but with the big advantage of cooling the load. A smallish WC loop (120 rad & res in the box) can adequately cool a TEC moving 150W across a 15C - 20C differential. A CPU running at 45C on air might get to 25C or lower, which improves OC and extends CPU life. Using a big undervolted TEC, you con get that cooling without big amp draw, probably off of available 12V power. A temp controlled PWM means that the TEC only draws the power it needs, so you don't burn power all the time.

That is not the way most extreme folks want to go - but it is an inexpensive and reliable way to get 25% or better OC for 24x7 use.

[flak-spammer]

In my opinion TEC's would actually be quite a bit more reliable than phase change because of the lack of moving parts. With one 400 Qmax TEC you could pick between several choices of temperatures just by adjusting the voltage given to it, and it's probably 1/100 of the size of the hardware required for phase change. As devices they are incredibly durable and extremely flexible in the ways of use. They pretty much are the dream cooling devices (with exception to power draw if looking for the maximum delta). Personally I would find both to be too much risk for my computer so I doubt I'll ever invest in using them as CPU cooling devices, but I could see using them for several other purposes.

[bluestang]

Uncle Jimbo, I would love to add a TEC to my air cooled cpu heat sink (Zerotherm FZ120) as you described above -- low volts and temp controlled PWM. Do you have some info/setup/plans on how to do it? I don't have to worry about condensation and moisture control? Thanks!

[Uncle Jimbo]

Uncle Jimbo, I would love to add a TEC to my air cooled cpu heat sink (Zerotherm FZ120) as you described above -- low volts and temp controlled PWM. Do you have some info/setup/plans on how to do it? I don't have to worry about condensation and moisture control? Thanks!

The right setup depends a little on your budget. The Zerotherm FZ120 has around .125 C/W performance.

The design I worked up with help from others is at
PWM Controller Design
The thread has a lot of background that might be interesting. Parts for this are maybe $30 if you have some stuff around and are handy with electronics.

The right TEC depends on your budget, CPU and desired overclock. You'll need some thought o work the heatsink mounting. You will need to have the TEC between a cold plate and a hot plate to apply proper pressure, and that goes between the heatsink and the chip.

You might be tempted by a low cost solution such as the 12726 eBay "400W' like Ebay 400W Cooler
Running from a 12V supply, even PWM controlled, that TEC provides 150W across 10C dT, but at 180W in. Too much heat for your cooler.

Somethng a little bigger like FrozenCPU 437 TEC is closer to the mark.

That one gets you 120W cooling across 25C with input of 9A at 10V (90W). You will see a 25C rise above ambient with your heatsink and that load, so you would keep a 120W load at ambient.

What happens at other CPU power? Well, if you run up to say 160W, the rise is around 30C, and dT drops to 20C. Now your CPU is 10C above ambient - but that's not bad for a big OC. If you are running from a bigger supply than 12V (say 24V), PWM would ramp the voltage up to maintain ambient, but the increase in power makes that a losing game - the heatsink rise goes up faster than the increase in TEC cooling. So 12V max is the right max voltage.

At idle power, maybe 30W for i7 or Phenom II, you get rise of 15C, and a dT of 35C. So without PWM, your CPU is 20C below ambient. To maintain ambient, the PWM would ramp effective voltage down to 4V at 3A, a total of 12W in.

So with those parameters, you could drive this from your PSU if you have 10-12 amps to spare, using a PCIe power connector for exampe. Installation would be clean, and you would maintain the CPU at ambient up to some modest overclocks, and keep it fairly cool even with big OC power.

[pur]

Ok Jim, is there a diagram that illustrates all of what you just said (or some). Perhaps its my state of mind, but whenever you post on TEC's im a little baffled as to what your saying.

[Uncle Jimbo]

Ok Jim, is there a diagram that illustrates all of what you just said (or some). Perhaps its my state of mind, but whenever you post on TEC's im a little baffled as to what your saying.

There is a lot of background on this forum on how to mount TECs between copper plates, how to torque, etc. So make a sandwich of cold plate, TEC, and hot plate, and put it between CPU and cooler.

If the PWM controller is intimidating, just hook up a simple thermostatic control and relay to reduce the TEC power when the temp gets down to ambient. I would not recommend turning it off, since that will induce a lot of cycling. Switching between 12V and 5V should do the trick. If it still gets too cold, go to 3.3V.

[kintaro]

this is my new T.E.C project with a 535 watt peltier.
so far i've only test it with the 5v line and the temps i'm geting are extraordinary for me considering i spend only 30 euro.
my main interest now is to control the volts somehow in order to have the desired temps whithout changing psu lines for varius temps.
right now at idle i have around 0 C and is no good for 24/7 due to moisture consideration.

for more pics:http://www.hwbox.gr/showthread.php?t=4858

[Ultrasonic2]

how did you attach the heat sinks to the hot plate ?

[flak-spammer]

By appearance there is no sign of a clamping device. I should remind you that TECs require clamping force for optimal lifetime as well as the fact that thermocouples when excited vibrate a little and it also reduces the gaps between the thermally conductive surfaces.

[kintaro]

By appearance there is no sign of a clamping device. I should remind you that TECs require clamping force for optimal lifetime as well as the fact that thermocouples when excited vibrate a little and it also reduces the gaps between the thermally conductive surfaces.

there is no clamping devise because simply by turning upside down the the construction of 1st foto and put it on cpu i'm using the normal cpu 4 screws with backplate. is this the required force or i need more?
i apply AS5 on all surfaces.

[Uncle Jimbo]

there is no clamping devise because simply by turning upside down the the construction of 1st foto and put it on cpu i'm using the normal cpu 4 screws with backplate. is this the required force or i need more?
i apply AS5 on all surfaces.

AS5 is no good for cold side of a TEC as it will never get hot enough to cure. Use Arctic Ceramique.

The maximum allowable force on a modern CPU is around 50 psi and mounts are designed to apply no more than that to avoid damage. The MINIMUM torque for a TEC is three times that - 150 psi - and 250 psi is better.

Also the plates need to be flat to .001 in / in and the torque needs to be applied evenly, usually across the centerline.

If you don't torque adequately or evenly with flat plates which are at least 1/4 inch and preferably 3/8 inch to reduce bowing (.001 in / in is not much!), the internal stress of the elements heating and cooling will cause them to fail pretty quickly. Maybe in a week, maybe in a month - but not long.

So basically, you have to make the plates larger than the TEC, enough to allow them to be bolted together with the correct 150-300 psi torque. That high torque 'sandwich' then goes between the heatsink and CPU, which can have the desired 30-50 psi.

[kintaro]

AS5 is no good for cold side of a TEC as it will never get hot enough to cure. Use Arctic Ceramique.

The maximum allowable force on a modern CPU is around 50 psi and mounts are designed to apply no more than that to avoid damage. The MINIMUM torque for a TEC is three times that - 150 psi - and 250 psi is better.

Also the plates need to be flat to .001 in / in and the torque needs to be applied evenly, usually across the centerline.

If you don't torque adequately or evenly with flat plates which are at least 1/4 inch and preferably 3/8 inch to reduce bowing (.001 in / in is not much!), the internal stress of the elements heating and cooling will cause them to fail pretty quickly. Maybe in a week, maybe in a month - but not long.

So basically, you have to make the plates larger than the TEC, enough to allow them to be bolted together with the correct 150-300 psi torque. That high torque 'sandwich' then goes between the heatsink and CPU, which can have the desired 30-50 psi.

thank you for yours usefull info.
seems i have to make another pair of plates large enough to add bolts.
there is a restriction at this area as t.e.c is 62x62mm.
i'm thinking also to make a gup 1mm both plates at the dimentions of t.e.c in order to keep it at the desired position.
what is yours position regarding the thiknes of cooper. is it helping at the stability of loads. what's yours recomendation?

last night i attempt with 12v. -15 idle.
after 10 minutes i'v noticed the 12v line is geting warm.
an old psu 500w was the next step. i cut all (7) 12v wires and attached all in one in order to combine 12v1 + 12v2. i did the same with 7 grounds.
after 1 h of testing there are no signes of warming on neither cables nor psu itself.

[Uncle Jimbo]

thank you for yours usefull info.
seems i have to make another pair of plates large enough to add bolts.
there is a restriction at this area as t.e.c is 62x62mm.
i'm thinking also to make a gup 1mm both plates at the dimentions of t.e.c in order to keep it at the desired position.
what is yours position regarding the thiknes of cooper. is it helping at the stability of loads. what's yours recomendation?

last night i attempt with 12v. -15 idle.
after 10 minutes i'v noticed the 12v line is geting warm.
an old psu 500w was the next step. i cut all (7) 12v wires and attached all in one in order to combine 12v1 + 12v2. i did the same with 7 grounds.
after 1 h of testing there are no signes of warming on neither cables nor psu itself.

That's the right way to get the power to the TEC. Don't do a cup - you will have a hard time machining to .001 flatness, and the TEC will not move at all with 250 psi on it. Torque carefully - get it tight then gradually increase, going back and forth, until you get to the desired torque.

For what you are doing, 3/8 copper is about right - anything more will just get in the way. The hot and cold plates don't need to be square - you can do a rectangle to fit in your heat sink. Put the bolts on the centerline of the TEC (in the center of the long part of the rectangle). You can machine in a little outside of the actual TEC area to fit past your heat sink mounts.
Read these assembly tips:
http://melcor.com/procedl.html

[kintaro]

this is an update with the hot!! QX9770 half an hour OCCT.

[Ultrasonic2]

what voltage are you running this thing at ?

[kintaro]

what voltage are you running this thing at ?

it's about 11.3 v from an old 450w psu.

[Ultrasonic2]

cool i was actually meaning what voltage was the CPU running at .. but it's good to know what tec voltage your running it at to .. im gonna make one of these too.. and im not just talk i have my own CNC MILL

while im asking questions .. what cooler did you use ?

oh and what fan/s did you use

[kintaro]

cool i was actually meaning what voltage was the CPU running at .. but it's good to know what tec voltage your running it at to .. im gonna make one of these too.. and im not just talk i have my own CNC MILL

while im asking questions .. what cooler did you use ?

oh and what fan/s did you use

cpu is runing defult volt arount 1.25v
cooler is http://www.plaisio.gr/product.aspx?p...758BB7255737E7

2 fans push/pull format.

[Ultrasonic2]

[Ultrasonic2]

cpu is runing defult volt arount 1.25v
cooler is http://www.plaisio.gr/product.aspx?p...758BB7255737E7

2 fans push/pull format.

What fans did you uses ?

[kintaro]

What fans did you uses ?

2 x 12 cm Zalman

[Ultrasonic2]

does the copper blocks try and slip out from under the cooler ? especially if the motherboard is mounted on it's side ?