TEC1-12706 Thermoelectric module

[ThermallyFrigid]

Having issues with TEC1-12706 modules

Use: Igloo Iceless Cooler
Issue: Replaced TEC1-12706 and now cooler does not reach temps as low as before.

Summary:
The original TEC1-12706 finally shorted itself after great service for 10 years.
It had TEC1-12706 printed on the original peltier module.
The original TEC1-12706 was able to consistently get the cooler to or below 42F under the conditions listed below and even in a hot car sometimes.
Not sure how to proceed at this point to get temps back down to where they were originally.
I've tried my best to get a good thermal contact.
Installed new seal around lid.
I've tried raising and lowering the voltage/current with no significant improvements.
Any suggestions would be appreciated.

Trial 1
Replaced with Amazon purchased TEC-12706
Used the same thermal paste that was already on the module
Ambient temp 80F
Cooler empty
Results -  lowest temp achieved 54.1F (overnight run)

Trial 2
Replaced with Amazon purchased TEC-12706
Cleaned all surfaces and used Chinese White Thermal paste
Ambient temp 80F
Cooler empty
Results - lowest temp achieved 56.4F (overnight run)

Trial 3
Replaced Module with Cold and Colder (brand) TEC-12706, cleaned all surfaces with alcohol, applied cheap Chinese thermal plaster (white) (as thin as possible for coverage)
Ambient temp 80F
Cooler empty
Results - lowest temp achieved 46.1F (overnight run)

Trial 4
Replaced Module with Cold and Colder (brand) TEC-12706, cleaned all surfaces with alcohol, smoothed surfaces with 2500grit sand paper , applied Arctic MX-4 thermal paste (as thin as possible for coverage)
Ambient temp 80F
Cooler empty
Results - lowest temp achieved 51.8F (overnight run)

[NiHaoMike]

Try stacking two modules on top of each other, run the first stage at a higher voltage than the second stage. (Experiment to see what works best.) Should get substantially colder than a single stage while using less power.

[coppercone2]

did you check the supply. if it was running on a damaged module maybe its damaged

and are you actually measuring ambient temperature?

[coppercone2]

and try putting alot of thermal paste. if its pressed good it will squeeze out.

Its big, its not a processor. And I still think that 'ultra thin' thing might be a myth, so you are tempted to pay a 'pro' to do simple tasks for computer processors on gaming computers.

Too little is a problem, I never saw a problem from too much, if you tighten it properly and wipe away the excess (messy)


For a big object I can see problems with there not being enough. Big problems. A processor die is like 1/5 the size of the module you have.

And are you sure the mechanical parts are not damaged. Some how its being clamped in there and shit. There is nothing cracked or deformed or warped on the housing? Say something that might make a heat sink tilt over when you tighten it down to the chassis.


I don't see that much of a difference between the expenisve digikey modules and the cheap ones. Like 5-10% maybe. It will come down to reliability over time, not raw performance.

[ThermallyFrigid]

A little follow up.....

So from the time I first disassembled my little cooler, I think I put in about 10 different Peltier modules, each time trying a bit different thermal paste and new foam insulation etc.  basically trying to optimize efficiency as much as possible.  it was a lot of work for a small, $130 thermo-electric cooler.

But in the e3nd here's what I found and what worked.

1). I tried 6 different TEC1-12706 modules, from 3 different suppliers.  The one that finally worked was from Cold and Colder. 
2). I tried 2 different TEC1-12710 Modules, from 2 different suppliers. One came from DigiKey and was the most expensive.  No luck.
3). I finally bought 3 TEC1-12715s from Cold and Colder and was about to try the first one when I finally got Delta Temperature changes closer to the original performance with one of the Colde and Colder TEC1-12706 modules.

I ended up buying a new 1/4 x 1/4 self adhesive seal to put around the top.  It was a roll of 25 feet from Amazon.  SOFT Neoprene (closed Cell.
I also bought a new Insulation pad to go under the Hot side on the top of the cooling assembly.  The one I fabricated was nearly the same thickness neoprene but I cut it larger to better cover the entire underside of the hot side heat sink and it also better closed the gaps going through the cooler opening for the peltier assembly.  An overall improvement.

So, my tests....
12 years ago or so when I first got the Igloo Iceless cooler I ran a temperature test.  Empty cooler, overnight run powered by the AC adapter, room temp at 75F.   At that time it got down to 36F.
The last iteration I did last week with the Cold and Colder TEC1-12706 module resulted in 34.3F with room temp at 75F.

It also gets cold FAST now which  I love.
I can't say for sure if the Peltier module that I ended up using made the most difference or if my upgrades to the insulation made a bigger difference.  Maybe both contributed?

I did a resistance test on ALL of them and got measurements all over the place.  But the one I used had the lowest resistance of the bunch at if IIRC was around 3 ohms or so.
So if I ever need to replace it again I'll definitely measure the resistance first and choose one with a resistance at 3 ohms or so.  Just keep in mind the resistance changes significantly with temperature so if you measure the resistance in a HOT room the resistance will be much higher.  IIRC, I measured the resistance right around 80F room temp.

Anyway, I'm enjoying my little thermo-electric cooler again and hoping it lasts another 15 years.   The only other item that could fail is the fans and the original fans are unobtanium, so I will have to engineer a new fan setup (hopefully more efficient than the original setup) to replace the unavailable fans.  Shouldn't be too hard. 

OH...btw....
I used several different thermal pastes and one thermal plaster (good God that was hard to get off!)   I also tried Arctic MX-4, expecting a great result.
I can't say why but none of the tries with thermal paste ever resulted in temps below about 44F.
On the LAST attempt, I used NO THERMAL PASTE at all.  I just cleaned the surfaces spotlessly clean after polishing them as best I could with a 1000grit buffing paste.
Now it gets down to 34F at 73F room temp.  Any colder and things would start freezing solid.

Hope this helps someone.

[coppercone2]

I am suspicious you are chasing a ghost related to voltage/current. The main difference between the peltiers of different designation (for the same footprint) is their thickness as far as I can tell, and this effects all of their parameters.

I got marginal improvements from using digikey modules, and marginal improvements from using premium thermal paste, leading to a very small improvement after alot of effort, for the same class of device, but I was always measuring current and voltage (power)

If you still have the broke one, measure it with a caliper to see thickness compared to the one you installed and the ones that have performance problems.

IMO the problem you have with grease must be related to pressure, basically you gotta be willing to crack a module or two to figure out what too much pressure is when it has a thick thermal grease. This is the same, there is some region where you get nominal performance, and beyond that you might squeeze a tiny bit more of it by putting it under high pressure at the risk of cracking it, but if you don't get to that nominal point (with thick grease that is difficult to 'feel') then it will have decreased performance


Now, if there is substantial impedance variation between the modules, and someone set the device power supply to some resistance they got at 100,000 quantity of modules, I can see how tolerance on the module impedance can cause you a problem from another batch. The performance of the device really depends on current flow. When they mean a 12V module, its nominal, and they probobly expect a trim to get your desired nominal operating current.


I suspect that any module of the same designation will work, if you can adjust the power supply a little, to get the same nominal current, and when you use as 12.0v rail, you are totally at the mercy of the manufacturing tolerance and you are left with trying to match the device to the rail, rather then matching the device to the expected trimmable power supply.


And the same thing when you adjust current, you get to a nominal point where the system is reasonable, and then if you increase current more, you get a small increase in the cold properties of the cell at the expense of alot of extra heat and inefficiency. 


Think about LED's (diode), they get very non-linear, and running them based on a voltage is not going to get the most performance out of them, since you end up with a diode curve, and really you have to adjust the voltage (and so current) to get the most out of each LED. And LED are highly efficient devices, unlike peltiers.


This might be useful to know if it breaks again, so instead of trying to match more modules, you can give a voltage adjustment (with current measurement) a try. And measuring the current use of this module (if its accessible) might help you get a 'specification' for future repairs, so you know what current you should expect.

 I realized when I search this thread for the word 'amps' what the problem probably was




The thing I am curious about now, does that power supply have an adjustment potentiometer on it for voltage? I wonder if that is why these devices have shit reviews, since some of them might be getting under-powered if their impedance is higher and not cooling. Or maybe they just get the modules sorted to work at exactly 12v to get rid of the adjustment circuit

[ThermallyFrigid]

Nice reply.  Thanks.

I did try MANY different voltages and currents because a lot of the information I found said that could make big differences.
In my case, the best results seemed to occur right around car battery voltage of 12.6-12.8v.

I did not check thicknesses.  They all seemed to be the same thickness.  I would need to check them with my caliper.
What I looked for was a difference in wiring.  I was expecting the TEC1-12715 modules to have a slightly increased AWG wire size.  They did not.

[coppercone2]

Mine have a substantial thickness difference, but I forgot what designations I have. Also some might have different amount of junctions.


So you are saying, even when you varied the voltage somewhat, the replacement 12706 peltier would still not perform to the requirement of your box? I.e. between 10-14 volts

I figure that the one that had 12V initially, the replacement 12706 might need.. say 13 volts to match the performance, or less then 12 if you had a bad one in the original

and you are powering the module directly right? Like test leads -> module leads. Your banana cables look big but you would have to check the voltage drop, you might be suprised how much you can lose on something like 18awg test lead (common enough). I was using 18 AWG alot for low power things, and i forgot, then when I had to deal with a few amps at low voltage I thought i had bad devices until someone reminded me about voltage drop, of course when I went to the 10AWG test leads the problem went away. I don't like using the thick ones if I don't have to and then forget

I would just be very surprised if the requirement of a little refrigerator was so strict that basically any off the shelf replacement within some adjustment range is not capable of repairing it. Usually they are like cheap/bad parts.

[ThermallyFrigid]

So you are saying, even when you varied the voltage somewhat, the replacement 12706 peltier would still not perform to the requirement of your box? I.e. between 10-14 volts

Yes.  I had to test quite a few to find one that did the trick.  VERY laborious.
One batch I bought on Amazon (that was actually rated highly) was so bad I had to send them back.  They were all over the place.  Two of the five were open.  Infinite resistance.

I think these things are pushed to their limits basically.  If everything is 100% right on, they can get cold to specifications.  But if one component is not up to par it can easily fail to perform as expected.  I would think there is VERY little room for error.  It has to be pretty much perfect.

I also found that if you have one of these and you get condensation inside after it's been running a while, you have an air leak in the seals somewhere.  That's critical because these units lack enough overall capacity to cool the inside AND fight against outside ambient air entering.  Said condensation produces latent heat in the process of condensing that robs the unit of it's ability to cool so extremely important to check for air leaks.  I used my thermal camera to make sure there were no leaks.  I had to re-do the seals a few times to get it just right. it might actually be too "just right" now since it gets very close to 32F.  I don't want stuff freezing in the little Coleman cooler.
My new rule of thumb is if there's more than a few drops of condensation inside, somethings wrong.

I do also have several Euhomy 37 quart compressor refrigerators that I use for longer trips.  Those can get down to sub zero temps fast.  But they are bigger and bulkier (24 x 33 x 23ish).  For in town stuff and trips less than 300 miles the little Coleman is perfect.  I actually just bought another to keep in my other vehicle.

[Simmed]

while scouring around for TEC a long time ago
i remember coming to this chinese site
maybe you can find some helpful data
thickness = 3rd column from right (mm)

they have 4 versions of 12706 for some reason 

http://www.tecooler.com/products/tec.ht.html

[coppercone2]

Ok it makes sense the one that worked is the same model# from cold and colder a reputable manufacturer

But, I still think that most  12706 from CC would work (you only tried one right)?.

 I can understand that the other part # don't work, becuase IIRC their thickness or element count or both is different, so it has different performance. I forgot what the thickness has to do with the behavior, some things are obvious, like if its thin, and you turn off power, it is less insulated (thinner 'insulator' that is un-powered diode material), so it heats the cold side up more quickly (the heatsink basically dumps heat into the cold side through the thinner diodes).


Dare I say the thicker ones, have a lower temp, but less heat flow, and maintain 'insulation' better, despite being less efficient? But they also generate their own heat from the inefficency, so IDK how much of a role it plays.

And the diode count is similar, in the sense that.. more diodes = more thermal linking. Less diodes = less thermal linking. Longer diodes = less thermal linking. Few long diodes = 'good insulation' and alot of stubby diodes = 'thermal jumper'.  Just from the perspective of the unpowered diode material.

 They are hard to think about because their performance is poor and it feels like its a 'hack' with electronic components

I am almost sure that the high heat flow (but low max temp delta) one had ALOT of short diodes, and the "very cold" one with low heat flow had fewer diodes that are longer. The high heat flow one is for like bulk low grade cooling, while the low heat flow is for cooling some low power device like a CCD, to get the absolute lowest temp.


You have to get a big object down to a lower-temperature, but you don't have active heat generation, so that is a 'refrigeration' application, while cooling a overheating junction box is a high heat flow application, while cooling a tiny CCD is a maximum low temp application.

There is still a design choice though, of how fast you want your drinks to cool vs how much power it has to use on the long term, I think. Because there is the control system too, you can run that shit really hard to cool it down then.. barely run it so it does not freeze, to shorten the time it takes.


Like the thin one, despite being efficent and having lower waste heat, has to fight against the hot heat sink that is backflowing. The longer diodes prevent the backflow somewhat. I think its a factor, but its confusing. Like if you made a refrigerator with 1 inch thick copper pipes, the radiator would keep heating everything backwards and it would not be a very good separator. Because the diode material is a poor conductor compared to the heat sink

[ThermallyFrigid]

That's a lot to digest there :-)

IIRC, I tried 4 different Cold and Colder modules until the one worked great BUT.....

Keep in mind that I was making improvements to the insulation etc each time I removed and installed a different one.
Basically, I trust Cold and Colder and will likely get any Peltier devices I need from them.
Also, the ONLY one I did not use thermal paste on was the last one and that one is the one getting super cold now.  Go figure.
Coincidence?  Seems like it would be but on the other hand....imagine if I had used thermal paste....odds are it could not have gotten any colder than it is now without it.

Leaves me wondering if I truly got authentic Arctic MX-4 thermal paste through Amazon?

These thermo-electric devices are definitely NOT the pinnacle of efficiency in refrigeration by any means.  But the Igloo cooler I'm using is small and lightweight making it perfect for trips to town to get groceries or short trips to keep drinks and a sandwich cold.  Mine pulls a pretty consistent 4Amps while running (a bit more when the vehicle is hot inside).
By comparison, my larger (but more efficient Euhomy compressor refrigerator/freezers) have a brief surge when starting up then settle down to IIRC only 3Amps or less draw.

So for overnights when I sleep in the vehicle I definitely want the Euhomy rather than the Igloo if I'm running on batteries.

[coppercone2]

I don't know, I think it has to be a pressure thing, usually the difference between thermal paste and no paste is substantial and the differences between brands of thermal paste are rather small, with the processor its a little more pronounced because the power density of the die is huge, and sometimes the heat spreader they have on the lid is not sufficient

I doubt you would see too mcuh of a difference if you got low-grade thermal paste in a mx4 tube on a peltier (but you probobly would see a difference on a high end CPU working hard)

What would be nice for these things is a big ass caliper or micrometer thing that can measure the 'stack' height between both heat sinks, but due to the price of such a tool everyone is working in the dark. That is a variable between all the compounds, that you can't easily measure the thickness, and you need to rely on torque, and that is kind of shit, especially considering usually people are threading to aluminum plates and things bend and all the compounds have different viscosity  etc

I have thought to get some cheap woodworkers caliper, and then do some tricks with gauge blocks to try to measure a weird heat sink/exchanger arrangement, but its just not useful to me to try to do something high performance with peltiers.

I made a little cold plate with a dual stack (water cooled) that can get to ~-40C without insulation, just a bakelite bracker so you can put a beaker or something on top (i.e. weak cryotempering of a tiny gear can be done, but dry ice would be better and ln2 would be best), or some experiments with crystals and solvents. It loses ALOT of performance to condensation heating, and I agree with you that trying to work out the moisture seals is a ridiculous pain in the ass. It got to -55C with out a copper heat spreader IIRC, if you just set things on ceramic, probobly because there was less surface area for condensation to form. One day I might try to seal it up but I hate working with styrofoam/sealants, and it would need some 'bell jar' shit on top. God cutting styrofoam is awful  . I hate every single manufacutring process related to peltiers, its goo, foam and glue goo. If I could get a precision machined aerogel form that I can gently put into place and seal around with a tiny syringe of sealant, that would be alot more pleasant

[ThermallyFrigid]

Now that you mention it......

You're right and I forgot about that.

PRESSURE made a BIG difference.
I remember when I first unscrewed the original module those two screws were so tight I wondered if I would be able to get them out.  I had to get a better, fresher phillips to get as good grip in those screws to remove them initially.

On some of the trials when i got temps that weren't even close, I'd try torquing down on those two screws a bit more and EVERY time it made a difference in the temperature change.
For example it would bottom out at 55F and then I'd torque them down harder and it would start dropping again like 5F more.

So yes.  The amount of pressure I applied to the screws squishing the module, the colder it would get....like you said earlier, probably right up to the point where something failed.

[Geoff-AU]

I remember when I first unscrewed the original module those two screws were so tight I wondered if I would be able to get them out.  I had to get a better, fresher phillips to get as good grip in those screws to remove them initially.

There's a saying in the car world that "no matter which tool you use to tighten a bolt, you'll need a bigger tool to loosen it again". 

[ThermallyFrigid]

For anyone curious, here is a temperature drop vs time plot that shows what a healthy Igloo Iceless Cooler (28QT) looks like when functioning properly.  All parameters included.