• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

TEC/Peltier MATH XD

C

ColinB

n00b
Joined
Apr 16, 2013
Messages
27
If I wanted to use multiple TECs to chill a water cooling loop, while only air cooling the hot side with some large heat sinks and powerful fans, what type and spec and number TECs would you recommend I implement? And what PSU?

I will be cooling 4x AMD Radeon R9 290X and an Intel i7 4960X Extreme Edition overclocked to the max, (whatever that may be, I'm guessing around ~4.6GHz).

I do not want to go far below ambient, most likely stay away from the local dew point, (http://weatherspark.com/averages/30928/Modesto-California-United-States), and maybe dynamically change the temperature with a standalone Rasberry Pi with a link to live weather data. Also, I might be using modded silica packets around the case just in case.

Thanks for the thoughts! :)
 
if you just use more liquid say like a water jug size pump that to the rad gravity fed even you wouldn't need to worry about dew point and such and it would stay quite cool for quite a long time provided you have a good rad, it also wouldn't suck back tons of power.

TEC are great in that they can get very cold but are also wicked inefficient, if you are trying to get extra cooling you could just run an ac type such as a water cooler or similar to force cool the mass amount of liquid even more, just a thought.
 
Could you be a little bit more clear? I'm having some trouble understanding what you are trying to say. Are you just recommending standard water cooling with a large reservoir?
 
Xtremesystems has what you want... they've done this. I believe power / temperature can be regulated via PWM. Instead of using the weather-service, it may be better to use the temp + humidity inside your case - but I can see this getting a bit involved with a fair amount of testing required.

Generally the strategy is to use multiple large TECs, but at low voltage. Both hot and cold sides are generally water-cooled.

At low voltage, TEC-chilling becomes very power-efficient, possibly even surpassing phase-change efficiency. The delta-T (between hot / cold side of the TEC) will thus be low (you're not getting more than about 15-20C delta before it becomes a better idea to go phase-change).

Because you do not want to go below the dew-point, this may be a good fit for you.

But...if dew-point is all you want, then evaporative cooling may also be a good fit. You'd encounter some water-loss, and you wouldn't be able to put any toxic additives in (maybe silver only), and the system would be open-loop, and maintenance might be more annoying, but it'd be inexpensive.

I believe "bong" towers popularized the evaporative-coolers for PCs, but there have been some other highly successful designs that may have been better - some were housed in those large insulated ice-chests.

Anyway...you have some reading to do.

Edit: Evaporative cooling will not work if you live in a humid area, or if you allow the cooling system to humidify your room... there may be some room-ventilation required, especially judging by your heat-load.

CRAP... yea... I'm looking at that heat-load... that may be a lot of water-loss. You'd have to make sure to use a large volume of water, and probably maintain a high flow-rate. I'm not sure...I'd have to think about this... perhaps evaporative cooling is not a good fit. I may have recommended a sauna to you :)

Edit 2: I wouldn't dismiss the simplicity of a phase-change chiller setup (basically the guts of an A/C or fridge). It can be temp-controlled in the way you want. The efficiency will be decent. And really the only reason I dislike these setups = size and noise (not horrible noise... you've been around air-conditioners). Because you wanted to air-cool the hot-side of your TEC setup, I'm assuming you won't be bothered by the noise.

Edit 3: Also when you begin pricing out the cost of a multi-TEC setup for your heat-load... you may just be forced to use phase-change (or evaporative cooling). It's going to be expensive. You can save some money by using fewer TECs at a higher voltage, but then your power-efficiency will not be as good as phase-change. Because it is preferable to use TECs at low voltage, and because your heat-load is so high, you're going to need more than a few. You may need... 8 of them... I don't know. With Phase-change... there are some inexpensive yet powerful A/C's out there. You may want to start at the 10K BTU range... look for a high EER, and MAYBE you'll get a good compressor in there...who knows. Research is required.

Koolance makes some nice plate heat-exhangers for water chillers. That'd be the expensive option. The cheap option is to just place the A/C's evaporator into a reservoir, or replace that evaporator with a coil of copper (so there there is no aluminum in the water-path - not a requirement). Tube-within-tube (copper evaporator tubing within copper water tube) can be done as well... might want to find an HVAC-guy for some help with the brazing and refrigerant-charging, and that may be necessary to source / optimize (ratios, types) refrigerant.
 
Last edited:
Yeah AMD is not the most chill silicon at the party. I have seen some evaporative cooling before, I believe using 3M Fluorinert. Interesting to say the least.
 
Here is a crude drawing of an elegant PC.
Scan+1.jpeg
 
I feel that phase change will be really more risky than what I'm looking for (in terms of condensation, condensation, condensation, etc). Multiple low voltage TECs seem quite appealing, as they don't appear to be VERY expensive (both in the short and the long run).


[However, if I was to use a phase change system, I'd want the unit to be compact and I would need to explore insulation options, (maybe two layers of tubing with a vacuum in between), then I would still need to cool the GPUs and the motherboard, something a bit foreign to me.]
 
yeh what I was saying
use a larger volume of water/liquid as said like a water jug (5 gallon) then you can force cool the water if you please via say a water cooler or whatever.

What I am stating is simple, TEC and other methods of cooling can get complicated, expensive and power hungry, pumping liquid into a rad would be less complicated and still give ample enough cooling.
 
ColinB said:
Yeah AMD is not the most chill silicon at the party. I have seen some evaporative cooling before, I believe using 3M Fluorinert. Interesting to say the least.
Click to expand...

Did you see the temps the current Intel chips can hit and their power draw when clocked up. Problem of modern tech when shrunken, sure at idle doing nothing they have great temps and power use, when operating and bump even a bit above their "spec" these numbers go out to pasture
 
ColinB said:
[However, if I was to use a phase change system, I'd want the unit to be compact and I would need to explore insulation options
Click to expand...

As I was saying, you could use the phase-change system as a water-chiller for the entire system loop (CPU, GPUs, anything with a water-block). You can temperature-control this system so that, for instance, it comes on at 3-5C above dew point, and shuts off at dew point. Using a large amount of water (5 gallons or more), this will ensure relatively stable temperatures, and no insulation will be required inside the PC.

Unfortunately...this is a lot of trouble considering that regular water-cooling will get you a coolant temperature close to ambient already.

One solution that has been built is to reduce the temperature inside the case...also using phase-change (as an A/C). This allowed for cooler water temps w/o the need for insulation. This required some significant sealing of the case of course...
 
Archmage said:
As I was saying, you could use the phase-change system as a water-chiller for the entire system loop (CPU, GPUs, anything with a water-block). You can temperature-control this system so that, for instance, it comes on at 3-5C above dew point, and shuts off at dew point. Using a large amount of water (5 gallons or more), this will ensure relatively stable temperatures, and no insulation will be required inside the PC.

Unfortunately...this is a lot of trouble considering that regular water-cooling will get you a coolant temperature close to ambient already.

One solution that has been built is to reduce the temperature inside the case...also using phase-change (as an A/C). This allowed for cooler water temps w/o the need for insulation. This required some significant sealing of the case of course...
Click to expand...

That reminds me. There are compact ACs made for cooling industrial cabinets with huge servo/VFD drives in them.... Check those out. Might be sticker shock since they are a niche item. :)

http://www.hoffmanonline.com/product_catalog/section_index.aspx?cat_1=34&cat_2=2383&SelectCatId=2383&CatId=2383
 
Last edited:
About the compact ACs, that price seems quite unrealistic. However, they seem like they would be a decent fit.
Won't using large, underpowered TECs be just as if not more efficient than phase change solutions?
 
Last edited:
Some links to get you started: I'd pay attention to the work of Ultrasonic2, mindchill, and krownik. When I was considering such a project, Ultrasonic2's work was the most complete.

http://www.xtremesystems.org/forums/showthread.php?67927-TEC-undervolting-charts - looks like you'll need to run those big 437w TECs at 10v or so to get useful cooling out of them for your large heatload. I remember thinking I wanted to run them at below 9v for efficiency reasons - I remember thinking that if I went much above this, I may as well go phase-change. There are efficiency spreadsheets out there correlating deltaT, qMax, power consumption, etc... you'd have to find them or find the information to do the math yourself.

http://www.xtremesystems.org/forums/showthread.php?38367-The-Ultimate-Guide-to-TECs - general guide

http://www.xtremesystems.org/forums...rasonic2-s-TEC-liquid-chilled-pc&daysprune=-1 - Ultrasonic2 project 1

http://www.xtremesystems.org/forums...ource-DIY-TEC-and-Fan-controller&daysprune=-1 - Ultrasonic2 project 2

http://www.xtremesystems.org/forums...ec-Chiller-w-pwm-controller-24-7&daysprune=-1 - mindchill small chiller

http://www.xtremesystems.org/forums...In-One-Liquid-Chiller-Build-Off!&daysprune=-1 - krownik makes appearance...

http://www.xtremesystems.org/forums...=&pp=25&daysprune=-1&sort=lastpost&order=desc - TEC forum, past forum threads showing.

It is possible to match or beat the power - efficiency (or lets say Coefficient of Performance (COP) - because you'll run into this term in your research) of even the well-designed phase-change setups. It's probably possible for your setup because you don't seem to want to go far below ambient. I'm not positive you'll reach this target. I actually think that, once the inefficiency of the power supply is factored in, it'll be difficult. It looks like you'll need too many TECs, too much copper for the waterblocks, too much of everything, to match phase-change. But, I may be wrong.

You'd need to find out how many of which TEC to run at what voltage (and the series/parallel configuration you may want to use), the power supplies (24v? 30v? 36v? meanwell? cosel? = search terms for google/ebay) and possible power supply voltage controllers (like the 6020s and 6020d dc-dc controllers (don't buy this item from companies selling at a huge discount = not legit), which would allow for great flexibility in configurations (but perhaps lower reliability and significantly increased complexity) - can power really anything staying within that 20amp limit) you will use, the PWM and temp control solution, and...you may need some custom waterblocks milled... need to size the blocks of copper, make sure you have a solution to optimize clamping pressure between them (which is important for TECs).

I've done this all before...long ago... just not with these larger heatloads.
I can't be of much more help w/o re-researching everything.

Hope that helped you assess the scope of the project.
 
Last edited:
You must log in or register to reply here.
Back
Top