Eric's Ultimate Guide to TECs

Ultrasonic2 said:

This isn't the full picture ..i wrote this in the past



Q. I’ve heard that TEC’s are very inefficient is this true. And what about phase change ?

A.
To begin with the Term “Efficient” is NOT the correct term to use when comparing the amount of electricity used to what It’s able to move. As this can exceed 100% by a long way. When talking about moving heat we use the term C.O.P (Coefficient of Performance)

The C.O.P number is ascertained by taking the Qc (Which may NOT be QMax) and dividing it by the input power. Higher the number equals higher the “Efficiency”

Now there is NO blanket rule to whether or not a TEC is “Efficient” (high COP). This is the case because the COP of a TEC changes depending on the input power relative to it’s max. As you increase the input power the COP is reduced (Less efficient).

If we compare the COP of common phase change systems to a TEC system including pumps and fans. The COP can be very similar PROVIDING you have a relatively low input power relative to the max of the TEC. However to achieve a TEC system with such high COP you’ll need to use many under voted TEC’s. Making the unit far to large to direct die cool your PC. You’ll have to make a water chillier for that kind of COP. A Single TEC direct die cooling a PC will require a high input power relative to it’s max giving it a comparably MUCH lower COP than phase change.

For a 360Qmax TEC running at 100% of Umax it would have a COP of about 0.6 (360Q/600)
For a 360Qmax TEC running at 50% of Umax it would have a COP of about 2.3 (269Q/114)

This graph shows you the relation ship between input power and watts moved in Q (orange) and Electricity used (purple) and total watts your cooling system will be cooling.
As long as the orange line is above the purple one the TEC has a COP greater than 1


NOTE: As you decrease the input power relative to the Max of the TEC, the maximum achievable delta is also decreased. So if you have a phase change and TEC system running at the SAME COP the phase change system will always result in lower temps because the phase change unit will have a far greater delta.

This graph shows you the relationship between input voltage V delta

Please note these aren't exact numbers, just used to demonstrate my point

Click to expand...

Good points, I hope this thread gets more attention as I see to many threads with people not understanding TECs in the forum. It is a shame as I would like to see more TEC setups.

The reason TEC setups fail is because people dont understand and dont setup their systems properly.

TEC may not be able to handle modern CPU anymore. Some of the modern CPU puts out a few hundred watts of heat (especially overclocked) and you'd need a rather large TEC which may not work well on a small CPU, or multiple stacked TEC which would suck PSU dry plus you'd need custom mounting to account for tall stacked TEC.

A low power CPU or mobile CPU could still benefit from TEC while overclocked to the extreme but not in a laptop

You can buy ~400w 50mm TECs, just have to come up with a 15-24v power supply for them.

Not everyone is using CPUs drawing hundreds of watts. People simply need to run the calculators to find if they benefit from TECs.

Question... http://www.ebay.com.au/itm/MONSTER-..._Heatsinks&hash=item4153e2fd1f#ht_1850wt_1396

How would this go? (i want to take my i5 2500k to 5.5ghz 24/7) (if i throw the hot side under an amazing watercooler)

HardCaze said:

Question... http://www.ebay.com.au/itm/MONSTER-..._Heatsinks&hash=item4153e2fd1f#ht_1850wt_1396

How would this go? (i want to take my i5 2500k to 5.5ghz 24/7) (if i throw the hot side under an amazing watercooler)

Click to expand...

That unit is 2.5" square: 62mm/25.4mm = 2.44". TECs that large won't work in a direct die setup due to a host of mechanical reasons. And even if you could attach it, too much of the TEC would overhang the IHS on the CPU die, losing 40-50% of the TEC's cooling power.

Thus, this will require a dual loop water setup, one on each side of the TEC, and quite a bit of custom work, obviously. There has been plenty written about such dual loop setups so I won't bother to duplicate the info here.

Do you really want an extra 550w of power draw for temps not that far under zero?

BTW, there is no such beast as a 550W TEC on the market, not a 16V model anyway, and certainly not one for less than $50. If anyone made a 550W TEC it would cost many hundreds of dollars due to super low sales volume. That Ebay sale is a scam. It's very likely the standard low performance 227W pelt that's been sold for years, at reputable places like Danger Den. He's also hawking these things on Facebook and everywhere else. Total scam.

I just checked all the TEC manufacturers' docs and nobody makes a TEC with a Qmax anywhere close to 550W. The largest I found is 288W and it's a 30V model not suitable for CPU cooling, unless you use a specialty 24V PSU:

http://lairdtech.thomasnet.com/item...eries-thermoelectric-coolers/ut15-24-f2-5252?

Even if this "550W" TEC is a large 227W model, it would not be useful to you. The problem is that you NEED to know the exact specs in order to properly design your TEC cooling solution. If it turns out to be a 120W 24V model and you run it at 12 volts, you not only won't chill your Intel i5, but you may likely burn it up as the TEC can't pass the heat.

Only buy TECs for which you know all the specs up front. Since you must go with a dual loop water setup, you should get something like these instead of that scam "550W" unit:

http://www.ebay.com.au/itm/5-x-136-...Electrical_Test_Equipment&hash=item256c5dd916

The advertising on these is a lie as well, 136W when the Qmax is actually 80W. That's ok though, because you get 5 of them for only AU $36. At 12V these 5 give you 320W of thermal capacity which is a serious amount of chilling for a 95W max TDP CPU. The specs for the TEC1-12709 are found in this PDF:

http://www.bkbelectronics.com/pdf files/TEC1-127120-50.pdf

It draws ~8A at 12V * 5 = 40A, so you'll need a PSU with at least 60A on a single rail to power the the 5 pelts and the entire system. A suitable PSU would be the SuperMicro PWS-865-PQ:

http://www.newegg.com/Product/Product.aspx?Item=N82E16817377004

You'll need two of these for the hot and cold side water loops, $500 for a pair:

http://www.customthermoelectric.com/Water_blocks/images/WBA-5.75-1.15-AL-01_tec_700.jpg

You'll arrange the 5 TECs in a cross pattern. I'll let you worry about the rest of the design and assembly of the water system. As always, extreme cooling puts about 3-5 times the dent in your wallet compared to the cost of your high end PC you're cooling. Which is why many folks who have been around this block once simply buy a better/faster CPU/RAM/GPU and use quiet air cooling with heat pipes. It's a lot less money in the long run, albeit with less 'cool factor'.

I'm curious. Several of the posts mention combining TEC cooling with water. Several of the posts also mention the problems with condensation. Has anyone looked at the possibility of combining the TEC system with one of the mineral oil submerged systems? It would be unlikely for the oil to freeze, it would prevent the condensation from forming and the system has been proven (Puget Sound Computers) has been running a submerged system since 2007 on an almost continuous basis.

A thermal electric cooler moves thermal energy by passing current through a p-n junction. This creates a temperature differential between the two sides of the junction, typically between 69F and 71F degrees for most models of TECs. If one were to install a TEC in a direct die configuration and then fill the system cavity with mineral oil, both sides of the p-n junction would be submerged in the same fluid. The fluid would then equalize the temperatures on both sides of the TEC, completely eliminating any cooling effect. This is why nobody does what you suggest.

Condensation is easily avoided by use of dielectric grease in and around the ZIF socket combined with neoprene. With a direct die system there is no issue with fluid freezing as the fluid is cooling the hot side of the TEC, which will be in the 100+F degree range. If using a dual loop cooling system with multiple TECs, such as that mentioned in my posts above, to achieve maximum die cooling performance, and one were uninformed enough to use water in the cold side loop, sure, it would likely freeze. But nobody uses water in either loop because of bacterial growth. Instead they use glycol based fluids such as antifreeze, or water/alcohol mixtures.

If you'd read just a little about how TECs work, and there is plenty of primer information on this site, you'd have answered your own question before asking.

Before you read this, I want you to try to remember what your first reaction was when you heard about somebody water cooling a computer. You probably thought it was an absolutely crazy idea. Am I right? OK.

I have an outboard water cooling system. The hoses to the CPU go in and out of my case. I have an EK Supreme water block. The pump, reservoir, and the radiator are in a box on which the case sits. I don't do LAN parties, so this works fine for me. I get pretty good temps for my Intel Core I5 2500K when overclocked to 4.9. But they rise if I try to go over 5 Ghz. (I don't really like anything over 70 C.)

But what water cooler fiend is ever satisfied with 'pretty good'? I'd like to get over 5 Ghz. Soooo....

I bought some TEC water chillers a few years ago. They are a plastic reservoir with the TEC, heatsink and fan attached to the reservoir. The reservoir probably holds almost two quarts of liquid. I think it was designed for some industrial project - though I have no idea what the actual use was. I experimented with it with an external 12 volt power supply. It would routinely chill water to about 38 degrees in roughly 3 hours. Don't ask me how much current it draws. I haven't measured it yet. I ran it with a commercial 13 amp power pack. It puts out up to 13 amps. It got warm - but not really hot. That's the best I can tell you.

So here is my idea/brainstorm/hairbrained scheme - (And don't try to tell me you've never had any of these!)

1. Run the TEC chiller with its own power supply.
2. Give the chiller a small - low-flow 12 volt pump.
3. Use 1/8 inch ID copper tubing to make a coil within the reservoir of my regular water cooling reservoir. This would be a spiral coil of maybe 8-12 inches. Think of it as a heat exchanger.
4. The chilled water from the TEC unit would be pumped through this spiral coil tubing.

I think this is known as a twin loop system. They use if for some solar water heaters.

Clarifications

1. I would continue to use the radiator and its six fans (push-pull) for most of the needed cooling for my CPU.
2. The TEC chilled water could theoretically remove at least SOME extra heat from the water in my regular reservoir.
3. The extra cooling - I'd predict - would never be near enough to get the regular water cooling system running anywhere close to below ambient. So condensation shouldn't be an issue.
4. The extra heat removed MIGHT trim a degree or three off my CPU when it is under high loads. Thus, I could finally get into the 5 Ghz range.
5. My regular water system contains 1 pint of water - counting what is in the reservoir, hoses and the block.

Yes. This would probably be a plumbing and electrical travesty. But it could be done - especially in an outboard unit like mine. Is it practical - probably not. But I have the chiller and a spare pump. I'd just have to make the copper heat exchanger coil and figure out how to install it in my regular water reservoir. The rest is just tinkering and guesswork.

OK. I'm ready for any observations, advice, or gufaws and tee hees from the sage experts on this forum?

BachOn

That's an office water cooler chiller. Due to the small size of that TEC your dual loop idea might get you a 0.5 - 1F core reduction. It's simply not worth the effort required. You already have a real water loop and PSU. For $20 USD you can get a 168W TEC, 12V, 10.5A, run it direct die, and drop your core temp waaay below ambient. The TEC:

http://www.ebay.com/itm/USA-SELLER-...?pt=US_CPU_Fans_Heatsinks&hash=item4837960b38

Assuming you already have some CPU thermal paste, the only other things you'll need are a tube of dialectric grease and some neoprene, maybe two-four srews 3mm longer for the water block attachment. Full instructions for direct die TEC installation are on this site. It'll run you $30 total and be MUCH easier to do that what you have described. And it'll probably take your core temps down into the 50F range, allowing you to o'clock your I5-2500K to its maximum potential, assuming your mobo is up to it.

Thanks for the response, HardwareFreak. Your idea sounds sooo simple. But the condensation issue in the area of the CPU really scares me. I know many people are able to do it and get phenominal results.

Too, TECs do stop working - often suddenly and without warning.

You said that my idea would not work because the TECs in these chillers are too small. You estimated that I'd get no more than .5-1 degree F improvement. So let me ask this question. If I could put a larger and more powerful TEC on that water chiller - could it come closer to achieving my goal using my Rube Goldberg double-loop design?

I'm guessing I'd need a more powerful fan. And the heatsink might have to be machined to accomodate a larger TEC. I'm more comfortable with the idea of using the chiller to boost my cooling slightly. Using the TEC right on the CPU, frankly, is just beyond my comfort level. I guess I'm just too much of a "wuss".

Finally, you said the TEC on eBay was 168 watts at 12 volts and needed 10.5 amps. 12 watts at 12 volts equals 1 amp. So I think the current draw on that TEC would be closer to 14 amps. Or am I missing something?

BachOn

BachOn said:

Thanks for the response, HardwareFreak. Your idea sounds sooo simple. But the condensation issue in the area of the CPU really scares me. I know many people are able to do it and get phenominal results.

Click to expand...

This is how you eliminate condensation. It's not difficult at all.
http://www.overclock.net/t/410339/info-how-to-insulate-your-mobo-for-tec-peltier-cooling

Too, TECs do stop working - often suddenly and without warning.

Click to expand...

This is a red herring. You're trying to come up with excuses for why you should avoid doing a direct die TEC. CPU, case, radiator fans, and water pumps fail far more quicky and frequently than TECs do. TECs are made of transistors, same thing CPUs are made of. The failure rates are about the same. How frequently do you see CPUs dying? How frequently do you see fans dying? I rest my case.

You said that my idea would not work because the TECs in these chillers are too small. You estimated that I'd get no more than .5-1 degree F improvement. So let me ask this question. If I could put a larger and more powerful TEC on that water chiller - could it come closer to achieving my goal using my Rube Goldberg double-loop design?

Click to expand...

Do you want to fart aroud trying to make something work, that will never work, or do you actually want to chill your CPU? For a dual loop system you need a real heat exchanger, like the one I posten in this thread. What you have will not work and cannot be made to work. Scrub it from your gray matter.

I'm guessing I'd need a more powerful fan. And the heatsink might have to be machined to accomodate a larger TEC.

Click to expand...

Plz post brand/model numbers: the fan, rad, reservior, and pump. If you have a single fan radiator then you very likely don't have enough capacity for cooling a TEC, especially if it came from one of those cheap all-in-one kits. You'll either need to add another rad and fan, or swap to a dual core or dual core/dual fan unit. Please post a pic of the custom "under teh PC" radiator box and the radiator itself so I know what you have to work with.

I'm more comfortable with the idea of using the chiller to boost my cooling slightly.

Click to expand...

This is due to vanity, and lack of education, nothing else. This idea originated with you. So you naturally want to use it in preference to anything else, regardless of the fact it simply won't work. Again, do you want to screw around for months on an idea that will never work, or get sometihing done in a day that will literally freeze your CPU?

Using the TEC right on the CPU, frankly, is just beyond my comfort level. I guess I'm just too much of a "wuss".

Click to expand...

It's above you comfort level simply due to your lack of education WRT Thermal Electric Cooling of CPUs. Educate yourself. There's plenty of information easily available beyond my replies here. Note that the condensation problem exists whether you go direct die or if you use a dual loop. What matters is the temperature at the junction of the CPU and the heatsink device directly above it, whether a TEC plus water block, or just a water block. A properly designed dual loop setup using two of these 168w TECs will take your i5 CPU at full load down to the -32F to -12F range, ~20 degrees colder than a direct die, as you'll see below. Again, it doesn't matter where the TEC is installed. What matters in the temp drop achieved.

Finally, you said the TEC on eBay was 168 watts at 12 volts and needed 10.5 amps. 12 watts at 12 volts equals 1 amp. So I think the current draw on that TEC would be closer to 14 amps. Or am I missing something?

Click to expand...

Yes, you're missing something. TECs advertised as 12V are "nominal" 12V. They will operate from 0-15.1 or 0-16 volts, depending on the manufacturer. "What? Zero volts?" you ask incredulously. Yes, TECs operate in both directions. You can attach one to a heat source and generate voltage and current instead of using it to move heat. Some people have built solar panels out of arrays of them. So with low heat input you may generate 0.001 volts. The VMAX is usually 15-16 volts. IMAX varies based on the capacity of the unit. This unit is QMAX 168 watts, IMAX 10.5 amps at VMAX 16 volts. At the 13 volts your external PSU puts out, it will be able to move approximately

((13/16=.08)*168)= 136 watts

The max TDP of the i5-250K is 95 watts. Using the formula in Eric's guide, at the top of this thread, and assuming your loop can dissipate 231 watts, your temp difference between the hot and cold side of the TEC will be approx:

i5 @95W: 70F
i5 @65W: 97F

At ambient temp of 71F, the cold side will be approx:

i5 @95W: 1F
i5 @65W: -26F

For a calculated actual CPU core temp estimate you'd add 10-20F to each of these figures to account for thermal losses in the system, including the IHS nickel alloy and water block copper, the two layers of TIM, the tubing, etc. In other words, you'll never get a 100% efficient cooling loop. So when we add 10-20F to each of these we get a CPU core temp estimate of:

i5 @95W: 11 to 21F
i5 @65W: -16 to - 6F

This below zero Celsius level of cooling performance can be yours for ~$30-40 USD for the TEC and install materials, plus ~$50 for a second single rad/fan and fittings/tubing to join it to the existing one, and a few hours of install time. Far less time than the idea bouncing in your head. A litle more money maybe, but look at the performance/dollar. You may be able to get by with your existing rad, but I need to see it first. If your current rad can keep the water temp below 110F with this pelt installled you should get max load temp of around 40-50F with ambient 71F. If that's good enough for you then you save $$ on the 2nd rad. Lots of if's.

Note that the equations don't take into account the performance of your water loop--the equations assume 100% efficiency on the hot side of the TEC. If your loop can't move the heat fast enough then the CPU core temp will not drop nearly as much. Also note that at idle the i5 will be shedding much less heat, and especially if the BIOS drops the CPU clock in power saving mode. Your water loop needs to be able to move 231 watts for max efficiency, which is the combined output of the CPU at max TDP, and the TEC.

Thanks for the "tough love" response. I really will think about it.

FYI:
Everything in my cooling loop is in an external wooden box except the waterblock and the tubing that goes to and from it. I have a smaller mid-tower case.

My radiator is a fairly recent 3x120 Swiftech unit (seems like is was the MCR 320).
I have six Zalman fans mounted on it in a push-pull configuration. I control the fan speeds with a six channel fan bus (Logitech). I normally run them at about 50-60% power.
The 12 volts for the fans comes from a very quiet computer power supply.
I have an Eheim 1046 aquarium pump to move the fluid. (I'm considering moving to a 12 volt pump.)
I use medical grade silicone tubing with an ID of 1/2 inch. It's almost 7/8 OD. It doesn't kink.
The waterblock is an EK Supreme I bought earlier this year.
The water reservoir is an electrical junction box. I have 3/8 inch ID fittings on it. (Some flow loss here.)
The system (full) holds just a little over a pint of liquid. I use distilled water with some algaecide in it (I forget the brand, but it's one I've seen mentioned on this board - maybe PT something?).

BachOn

BachOn said:

Thanks for the "tough love" response. I really will think about it.

Click to expand...

I'm no politician. I'm straightforward and honest, sometimes brutally so, and offer no BS assessments and advice. I call it like I see it.

FYI:
Everything in my cooling loop is in an external wooden box except the waterblock and the tubing that goes to and from it. I have a smaller mid-tower case.

Click to expand...

Already got this part from your previous post.

My radiator is a fairly recent 3x120 Swiftech unit (seems like is was the MCR 320).
I have six Zalman fans mounted on it in a push-pull configuration. I control the fan speeds with a six channel fan bus (Logitech). I normally run them at about 50-60% power.

Click to expand...

The MCR 320 and 6 Zalmans is more than sufficient for 231 watts. However if your GTX 560 Ti is in the loop that adds 170w, for a total of 401w. This is right on the borderline of the Swifttech test setup. See below.

I have an Eheim 1046 aquarium pump to move the fluid. (I'm considering moving to a 12 volt pump.)
I use medical grade silicone tubing with an ID of 1/2 inch. It's almost 7/8 OD. It doesn't kink.
The waterblock is an EK Supreme I bought earlier this year.
The water reservoir is an electrical junction box. I have 3/8 inch ID fittings on it. (Some flow loss here.)

Click to expand...

The 1046 is rated at 80 gph at 3ft head pressure. So your real world gpm after losses is probably going to be 60 gph, or 1 gpm. Using Swiftech's test setup, this puts you right at 400w capacity.

http://www.swiftech.com/mcr320-qp-stack.aspx

You have twice as many fans as the test setup. Running them at full RPM may get you another 50w or so. Swapping the pump for a 12v with double the flow rate will get you another 50w, for a total of 500w of capacity.

So as you can see, water and air flow rates don't do much for increasing cooling capacity. Rad fin surface area is the key. Adding another MCR 320 will double your capacity to 800w using the 6 fans you already have. As it turns out, simply adding another MCR 320 is cheaper than swapping pumps, which run ~$80+. The MCR 320 plus fittings should run ~$60.

The system (full) holds just a little over a pint of liquid. I use distilled water with some algaecide in it (I forget the brand, but it's one I've seen mentioned on this board - maybe PT something?).

Click to expand...

Since this is a hot side only loop you can stick with this coolant solution.

In summary, your current water loop should be sufficient to easily carry the extra 136 watts of the TEC. So ~$40 and a few hours should get your i5 core temp near or below 32F without the GPU in the loop, and probably somewhere in the 40-50F range with the GPU in the loop. I'm referring to max CPU TDP of 95w here. Temps will be much lower at idle. If temps are higher than you hoped for, add another MCR 320 and you'll get there instantly.

So what are you waiting for? As Nike says, just do it.

HF,

Thanks for the excellent analysis of my equipment and great information and suggestions. Your enthusiasm is very clear and quite infectious. I do appreciate the input on my double-loop idea. As I said, I will consider it all, research it further, then make a decision when I'm ready.

BachOn

Posting this with the hopes that someone will find the information contained in the threads to be of value to them.

Two different PC systems, one using 2x 50mm TECs for chilling the water, the other using 2x 62mm TEC to cool the CPU via a cold plate.

Both of these projects cool the hot side of the TEC via air cooling...specifically a pair of Thermalright Ultra Extreme (TRUE) heatsinks.

This linked thread is to a currently functioning "hybrid" TEC chiller system, using two TEC units to chill water which is then stored in an insulated reservoir for use in cooling the CPU thru chilled water running inside insulated tubing. The hot side of each TEC is cooled by a TRUE heatsink with a single fan in a pull configuration.

http://www.tomshardware.com/forum/282844-29-peltier-water-cooling#t2033960

This second link is to one of my all-time favorites...an older project that I saw over on XS, the similarities being that it is also using a pair of TRUE for cooling the hot side of the TEC units.

http://www.xtremesystems.org/forums...RUE-TEC-Module-(with-pics)&highlight=TRUE TEC

The other similarities in these two projects is that both of these guys were told by the "experts" that it couldn't / wouldn't work...in some cases even after it was fully functional. Go figure!

Brilliant thread !
Ok my crazy idea is to cool the hot side of the peltier unit through ignition. I figure I'll attach the peltier to a custom reservoir or radiator and when the loop inside by desktop hits ambient +10C the system kicks in and starts cooling. I'll use butane, an isopropyl alcohol mixture, Diethyl ether or any other substance with a super low ignition temperature to continuously coat the hot side of the peltier. I can make it look fun like an exhaust flame thrower kit for a car with small flames shooting out the top of the case - while controlling for particle size so fire alarms don't go off
Let me know if I'm nuts or if something like this can work please ! Cheers guys !

Hey guys I've been reading this post some and an idea popped into my head. Please bear with me. Wouldn't it be possible to use a TEC on a water block after the radiator in the loop in order to slightly chill the water before it returns to the cpu?. Assuming that the TEC isnt high wattage amd has its own heatsink.

It seems to me that if the heatsink hept the TEC cool enough it could drop the temp a little bit but majorly. But it could still be useful.