Project : Tec controller, static and variable response

[ThyKingdomCome]

In an effort to establish need, interest and in response to this thread, I am posting on a project that Joemac and I have completed a year ago but not released in polished format.

We (AquaJoe) have created a board that is capable of automatically adjusting the heat of a cpu with an externally defined variable temperature using a TEC. Our desire was to allow a CPU to remain at a static temperature for testing the max overclock, or maintaining an overclock at a temperature, but not wasting the power of a tec at full power. As you know this power load will change as the environment changes. In winter a house may be cooler to due the AC, or hotter as the heater is turned on. If a user wants to keep a tec at 15C for the simple 'fun' of cooling, this board will allow the user to set a tec of appropriate size and functioning DC power supply to achieve that measurement, and hold it in place.

The benefits are variable, but overall, less power can be used in cooling and benchmarking systems can be calibrated or tested. As jaydee stated in an earlier thread, he could opt to use a controller of this nature (with creativity) to cool water to a set temperature or any other similar application where a set (or variable) temperature is desired.

The first picture is of the temperature was set to 25C which was also the ambient at the time.

The following pic is of the system set at ambient temperature follow, and you can see in the other pic, as the air conditioner was turned on in the house, the case temperature went down, and the cpu temperature followed accordingly. Later, it was possible to have this running with ambient at 25C and temperature at 40C with zero condensation from an un-insulated block. When the temperature was lowered, condensation proofing became necessary.

We used a external Meanwell PSU pushing 24 volts to power our board and TEC. Variable TEC's could be used along with variable power supplies, as the system will automatically raise or lower temperature to accomodate.

I posted to see what you guys thought. This may or not get marketed by us, we just are putting a feeler out. If anything, we created a toy that is fun to test with.

[Ls7corvete]

I could be interested depending on price and such.

How does it work? lowering voltage? switching the unit on/off like the variable PC units? efficiency?

Glad to see something like this put into effect.

[Belenar]

I guess it will also depend on how this whole thing works. If it is using the onboard sensors, I think nobody here would be overly interested, since we know through experience that these can be pretty far off...
But if it would be a complete 'hardware' solution with a seperate temp. probe, I think people building water chillers, TEC systems, ... would be interested if the price is right.
I'm thinking of people building test benches as well. They could use this controller to build a TEC based water chiller to keep their water temps stable.

[ThyKingdomCome]

This uses PWM to control the voltages.
The thermistors used are independent of the computer, and therefore could be used on different applications outside one.

[Ls7corvete]

http://forums.procooling.com/vbb/sho...hlight=tec+pwm

Filtered or un-filtered? what frequency?

[WeaselBlade]

I would be interested, but it would depend on price. Peltier setups get pretty expensive as it is...

[joemac]

Originally Posted by Ls7corvete

http://forums.procooling.com/vbb/sho...hlight=tec+pwm

Filtered or un-filtered? what frequency?

The unit we used for testing was working on an unfiltered circuit. The frequency actually varied depending on the demand required to cool the CPU to the target temp.

The controller did not rely on the motherboard to set or take target temperature the test unit is equipped with it own sensor. The motherboard sensor was only used to plot temperature vs CPU load.

[bigben2k]

There was another product that did that... I'll post a link if I can find it.

[Long Haired Git]

If it could track humidity, and thus "calc" the dew point, and thus keep coolant or a CPU just above the dew point, then that would be useful. Makes running pelts a lot less risky and hassle if you didn't need condensation-proofing....

[bigben2k]

No luck yet, but here's a source for a 437W TEC:
http://www.wintschlabs.com/browse_de...herm._Electric

[bigben2k]

Some other information:

"The Effect of Pulse Width Modulation (PWM) Frequency on the Reliability of Thermoelectric Modules"
http://www.tetech.com/publications/pubs/ICT99MJN.pdf

Here's a commercial/industrial product:
http://www.zionscientific.com/series850.html

Got it! :
http://www.techcool.com/vpc/photo.htm

[joemac]

If you could find that link that would be great. I am curious how their product would compare to our test unit. Weaselblade is right on the money about the expense on getting a pelt system going. In the past I asked peeps about a control unit for pelts that would maintain temperature of a CPU over a wide range of CPU loads. Most responded with - I paid all this money for a pelt setup and now its not going frezze my CPU.

My original vision for this product was not to freeze a CPU with a high power Pelt, but to use a low power pelt to cool the CPU to ambient temperature. I did look in the past for existing products that would meet our requirement (functions, performance and price) but could not find anything that would blend itself well with our crazy (but wonderful) hobby we call WC.

I will see if I could find more charts to post later, Its pretty strange to see an almost straight line for CPU temperature regardless of where CPU usage is.

long_hair_git we never had a trouble with humidity in our test environment, we set the target temp (the temp we wanted our CPU)a few degrees above desire temp (Temp we where trying to match). I want to clarify that our unit has two sensors and not one. What this means is if the desire temp (say room temp) changed from 80F to 73F our target temp would change from 80F to 73F with no user input required.

[bigben2k]

Originally Posted by bigben2k

...

Got it! :
http://www.techcool.com/vpc/photo.htm

10 char

[Holst]

Im thinking about going back to TEC in the near(ish) future.

As I only have one PC and I want to cool the GPU this type of automated system would be ideal for me. As I dont fancy having my GPU sit at -10 when in windows all the time.
If I could set the max allowed temp to a couple of degrees less than the GPU's max load temp with the TEC on full power (or better a few degrees less than the maximum stable artifact free temp) then I could save allot of money on power and have the cooling power of TEC when Im using the GPU.

I would prefer this to be a seperate box rather than running under windows on the host machine as I use multiple OS and installing this on them all might be a pain (especially in DOS memtest)

[link1896]

I've drawn up so far a simple block diagram of such a system with a humidity sensor to keep the TEC just above dew point using a microcontroller. Probably best for me to first get my water cooling monster of a system into action. http://forums.procooling.com/vbb/sho...?t=7442&page=9


Nothing is built yet, though I have got the humidity and temp sensors. A real voltage out humidity sensor is not cheap, eg the Honeywell ones are 50 bucks each. Capacitance style ones need calibration and are a pain. I think the humidity sensor of choice is the Honeywell HIH-3610

No plans for commercialization, just for my own fun. Last time I looked there were a few commercial offerings, but none were cheap. If someone had a 350-400w controller with dew point control for a sane price, they could move watercooling to a whole new level.

If all goes wrong with my own project, I'll have a hell of a coke can cooler.


link1896

[Shoebox9]

PLEASE keep working on this project. Such a monitoring/control system would have mass appeal to people after cool/stable high powered systems.

I'll buy one from you today!

I need/use a very fast/quiet rig for my business, and see TECs as the future, once control issues can be sorted out. I'm not about to home-waterproof my system & play till it pops.

As a condensation control measure, I'm considering building my my next box inside an old bar fridge (because it's air tight) and attaching a 350w TEC to a large sheet of copper to attract/freeze the condensation out of the air. A water block, opposite the TEC on the back of the copper sheet, would cool a water loop for CPU etc.

I'd rather buy what you are describing however.

Cheers,
Shoebox9

PS Comments on my condensation attractor idea are very welcome. :-) I figure inside an air sealed system (with an external cooling radiator) a limited amount of water vapor exists which will be "attracted" to condense on the coldest thing, thus reducing available water held by this packet of air to condense on other bits. Am I mad, will this work?

[LPorc]

Originally Posted by Shoebox9

PS Comments on my condensation attractor idea are very welcome. :-) I figure inside an air sealed system (with an external cooling radiator) a limited amount of water vapor exists which will be "attracted" to condense on the coldest thing, thus reducing available water held by this packet of air to condense on other bits. Am I mad, will this work?

I think the issue is your "attract" is in quotes for good reason, there is no attraction of water vapor, just an increased local probability of condensation at your "attractor". In the strictest sense it will reduce the amount of water vapor available to condense elsewhere because the system is closed (materially, but impossible to do thermodynamically, best you can do is buffer), but does not reduce the probability of condensation elsewhere.

You could get your own little weather system going in that case. Unfortunately it would be a mild one, so no use having a case window to display "a cloud in a box". Even if you accumulate a lot of ice at the "attractor" with the energy inputs into the case there will be sublimation and redistribution as everything tends towards entropy.

[Shoebox9]

If my "attractor" (a sheet of metal that's the coldest thing in the sealed air space) reduces the relative humidity at increasingly cold temps, by continuiously removing (condensing) water vapor from the air, wouldn't this keep my CPU water block significantly safer?

I've read of people loving really cold days as they provide the best O/Cing potential, but why not replicate this within a sealed packet of air? (If this can genuinely be done?)

Perhaps the trick is manually balancing power input to the TEC so it is always on, (rather than say using a fridge thermostate with a relay) while the PC is running, but a total freeze-over doesn't occur? (If the TEC were to create a set air temp and cut out, my sheet of copper would warm, water would evaporate, other case items may now be colder, etc.) I imagine keeping the TEC always on would mean constant manual voltage adjustment, as the TEC's cooling ability would be related to how much it is being cooled, which would vary with ambient temps. (Dam TECs, why can't someone invent a monitoring/control unit?!)

M PSU documentation specifies a 10-50dC ambient operating range. What would happen to it if the air was dry but a lot colder? Would it go pop?

[LPorc]

The large cold plate as a condensation control is like adding a second larger goat to a field and expecting the first smaller goat in the field to stop eating grass because the second larger goat is eating grass.

Unfortunately the probability is high that both goats will eat grass. Furthermore, I'd think the assertion that the grass is somehow safer from being eaten by the addition of the second goat would be considered dubious at best. And even if the original goat eats less grass, there's still going to be a lot of goat waste in the field.

[Shoebox9]

Hmm, not just a scientist, but a philosopher as well…! By comparison I'm a mere grasshopper- this is my first step into water cooling, let alone TEC's.

I currently have a h/drive sitting in the fridge. The drive is quite cold, but so far I haven't been able to find any condensation on it, even with the fridge door being constantly opened, letting in warm moist air. I wipe it with a tissue but no water. Yet there's tons of condensation on the cooling plate at the back of the fridge- some times icy, some times it just beads or runs down to a collection tray. If the h/drive was warm/hot, this should work even better (I imagine).

I guess I'm thinking of turning my case into a fridge. But what happens when I power down? As the drives, PSUs, etc, cool, and the water re-evaporates into the air from the cold plate, will dew form anywhere nasty then? I don't know is if I need a large chunk of metal inside, to act as a cold storage buffer, or a really thin cold plate, so it will warm up at a similar rate to the components.

Any ideas?

[LPorc]

Ok, I think I understand your point a little better. You've only got one goat, and you want him to have free run of the field, but only eat the grass on one side of the field. Because the conversation started about TECs and variable control of TECs I thought you were talking about a condensation control measure for TECs, not a condensation control measure for a PC in a refridgerator case.

How about turning off the fridge and opening the door when you turn off the PC? It really breaks the goat analogy but seems like it would work.

Also I think that the typical PC will rapidly overwhelm the cooling capacity of the typical mini-fridge.

The relationship of how much heat will move depends on the difference in temperature, a 10 degree C difference will move twice as much as a 5 degree C difference. Eventually the PC and the Fridge would find an equilibrium if the fridge was sufficiently capable of pumping that much heat. However, the fridge has limits. It is not designed for continuous operation, and was designed to overcome the leakage of heat into the cabinet and a couple of openings of the door throughout part of the day. I suspect the heat produced by the PC continously is likely more than a mini-fridge can handle.

In effect your TEC "attractor" plate would not be so much for condensation as to make up for the shortcomings of the fridge. So things have just gotten more complicated and using more energy to run, and condensation is still an issue. Really, phase change is an issue, and the water vapor in the air could easily see three phases in your sealed cabinet.

So, how do we use your big TEC to help out? Say we keep the case sealed and we force the air to circulate through a heat exchanger that your TEC is pumping heat out of. Now we've cooled the air, and since we are forcing all the air past the coldest point, we can design ways to prevent the phase changed former vapors from circulating. Think about a shop vacuum, but on a much smaller scale as we only need to deal with the vapors in a fixed volume. The liquids and solids are separated from the gasses, and in our (sealed) case the cooled gasses are forwarded through the case again.

Sure, we've got inefficiencies all over the place, mulitple heat exchanges going on. Your TEC still needs to be cooled on the other side. But, we've now got cool dry air with no condensation worries around the parts we care about. Well, as long as we don't cool below the condesation point of all the vapors and we maintiain a decent vapor pressure, anyway. But that's been the crux of the problem all along, hasn't it?

Hmm... An air cooled system is immersion cooled, just immersed in a gas, eh?

[redleader]

I currently have a h/drive sitting in the fridge. The drive is quite cold, but so far I haven't been able to find any condensation on it, even with the fridge door being constantly opened, letting in warm moist air. I wipe it with a tissue but no water. Yet there's tons of condensation on the cooling plate at the back of the fridge- some times icy, some times it just beads or runs down to a collection tray. If the h/drive was warm/hot, this should work even better (I imagine).

The hard drive doesn't get wet because its warmer then the air. This is equivilent to trying to condense water with a heatsink on a CPU. You need to cool at least below ambient to condense anything. Obviously your pelt will be below ambient (at least it better be!) so its not comparable.

[Shoebox9]

Sorry guys- I haven't been very clear about what I'm planning to do, and what my goals are. Because this is a major digression I'll start a new thread, with a more detailed explanation. :-)

LPorc, I'm quite taken by the idea of forcing chilled air through a vacuum bag (or something similar) to extract water, and would like to explore this more.

Cheers,
Shoebox9