Sciencemadness Discussion Board

Quad TEC1-12706 water cooler for condensers (FAIL)

Sulaiman - 24-11-2018 at 07:06

This is my 'upgraded' TEC1-12706 based water cooler, after the tests described below.
It is intended to cool water for recirculation through condensers;
QuadTec.jpg - 1.3MB

4x TEC1-12706 with HY810 heatsink paste on both faces,
1x 40mm x 80mm aluminium water block,
2x AK-786 cpu heatsink with fan,

This is an 'upgrade' from my first attempt which used the same heatsinks with
2x TEC and a 40mm x 40mm water block.

I used a DC30A-1230 12V 4.8W brushless pump to circulate the water
which gives enough water flow to make the inlet and outlet temperatures of the water block negligibly different.

The two fans and the pump have one connection to the 12V supply, (600mA total)
and there is a separate feed for the TEC power.
To increase efficiency from the original version I designed this module to run at 12 Vdc with 2x 2-in-series TC for 6V per TEC.

Performance will depend upon water and ambient air temperatures,
and water flow rate if insufficient,
I did a few trial calculations and should give around 100W water-cooling power

All parts via eBay UK from China.
.........................................................
I set up a test using 5 litres of tap water in a plastic bucket
with the submerged pump forcing water through the cooling block.
Ambient temperature 10.5o.
Power for both fans and TECs set to 12.0 Vdc.

With the TECs wired for 6V each, current = 1.8A per TEC, 3.6A total.
It took 55 minutes to reduce the 5 litres of water by 3oC
EFFECTIVE COOLING POWER = 6.24 WATTS :(

This was so pathetic that I re-wired the module as photographed above,
all four TEC modules wired for 12 VDC each.
Current consumption for the TECs was 11.6 Adc total.
EFFECTIVE COOLING POWER = RUBBISH :mad:
I abandoned the test after the water had dropped by about 0.7oC after 30 minutes ... worse :o

At no time did the heatsink faces feel warmer than about 40o.

--------------------------------------------------
I suspect that the TECs are below specifications, to put it politely :P
I'm now considering whather I should buy new genuine TECs
or give up for a while.
---------------------------------------------------
Corrections to any mistake(s) in my calculations or suggestions for a cheap source of 4x 40mm x 40 mm GENUINE TECs ?

Ozone - 24-11-2018 at 08:34

It's been a very long time, but used to get Russian TECs on Ebay that were legit.

O3

[Edited on 24-11-2018 by Ozone]

Ubya - 24-11-2018 at 10:33

every tec should draw 6 amps at 12V so yeah if you are drawing only 11 amps in total 2 are bad and 2 are good, or most probably they are all bad

Sulaiman - 24-11-2018 at 11:02

Before installing the TECs I used a 18650 cell to briefly verify heat pumping operation and orientation,
when the TECs were wired as two in series across 12V for c6V each,
Power supply adjusted for 12.00 Vdc at terminal strip.
one pair was 5v92 vs 6v08 ... 'matched'
one pair was 5v76 vs 6v24 ... not matched but I thought OK
By touch, the heatsink mounting surfaces (leveled and smoothened) never exceeded 40 Celcius.
I can only assume that the TECs that I got so cheaply (£6.32 for four incl.p&p) were not such a bargain :)

WGTR - 24-11-2018 at 11:30

Judging by the data sheet curves, it looks like your current draw is correct for the TEC modules. You may be expecting too much out of them. TECs are not very efficient (down in the single digits). The vast percentage of power used to operate them is wasted heat. That’s why air conditioners still use mechanical compressors. I bought a small refrigerator several years back that used a battery of TEC modules, and it took all day to cool down the half a dozen sodas that it would hold. You could use water cooling on the hot side to improve your performance, but that defeats the whole point.

There is a “sweet spot” on the (non-existent in the datasheet) power vs efficiency curve. Applying too much power can hurt efficiency. The trick is to operate the modules at the most efficient power input, and just add extra modules until you get the cooling power that you need.

[Edited on 11-24-2018 by WGTR]

Sulaiman - 24-11-2018 at 12:55

True, in the datasheets for TEC brand the COP is good at around 6V which is why I intended to operate at around 6V,
I have been using the datasheet from Hebei IT as it is widely available,
the performance curves are not as good as the TEC brand values,
but using the datasheet I expected 100/120/133 W pumped at 30/20/10 oC differential.
I expect the cheaper components to be a little below par.
but the units that I received are effectively faulty - I believe.

monolithic - 24-11-2018 at 13:50

You could probably get really good results with a hacked up air conditioning unit. Extend the tubing on the evaporator coil, to where you can submerge it in a bucket of water that contains your circulator pump.

WGTR - 24-11-2018 at 15:40

Looking at the datasheet that you're referring to, it looks like you're right; something is off. The current is somewhat low, as if there's a large temperature differential. Assuming it's not caused by a poor thermal connection to the heat sink, perhaps several junctions are making poor thermal contact with the alumina substrate, or there's otherwise high electrical resistance in a junction.

Before writing them off totally, I'd probably check the heat sink temperature with a thermocouple just to be sure that there's enough heat sink for the job. One interesting thing to try would be to repeat your test, cooling another 5 liters of water, but this time run the TECs at half power or less. If the water still cools down by 3° in 55 minutes, then that signals an obvious thermal conductivity problem somewhere.

RogueRose - 24-11-2018 at 16:58

How long do you need to have cooled water for the condensor? I've found the easiest way is to use frozen bottles of water placed into a larger bucket of water, wrap the bucket with insulation and put some foam or insulation on top of the water and it stays cooler much longer.

It's easiest to keep the water frozen in the winter obviously, just keep it outside or keep the water bucket outside if possible.

Sulaiman - 24-11-2018 at 18:02

I have a few options available for water cooling
- but TEC cooling would be cool :cool:


I wrongly thought that I could do this cheaply.
Even if the quad TEC1-12706 module worked as intended it would not have had the cooling power for my 24/29 kit at full throttle.
I can't afford/justify genuine TECs for the forseeable future,
so I've decided to abandon TEC-based cooling for now :(

Update

Sulaiman - 3-12-2018 at 02:04

This morning I referred the TEC purchase to eBay resolution centre
and half an hour later the full refund ammount has been received.
Hooray for eBay Resolution Centre.
(they have helped me a few times before, which is why I still shop via eBay uk)

now I'm looking for cheap Original TECs :P

elementcollector1 - 3-12-2018 at 09:50

Quote: Originally posted by WGTR  
Judging by the data sheet curves, it looks like your current draw is correct for the TEC modules. You may be expecting too much out of them. TECs are not very efficient (down in the single digits). The vast percentage of power used to operate them is wasted heat.


This strikes me as a bit strange - a TEC I've had my eye on claims a rough efficiency (Qout over Pin) of somewhere around 57%. Granted, I don't know enough about Peltier modules to know if that's a good metric to go by, but it seems to be reasonable to me at least.

WGTR - 3-12-2018 at 13:36

One way to measure efficiency on these would be to mount two of them in parallel, in between two heat sinks that are reasonably well insulated from the outside world. Apply power to one of them, apply a load to the other, and measure power in vs power out. Since the efficiencies would be squared, taking the square root of the total efficiency should give you the efficiency of each unit. I'm pretty sure that efficiency number will be in the single digits, but I've been wrong before.

If the temperature differential is very minimal across the junctions, it may be possible to pump more energy from the cold to hot side than what you're putting into it, achieving a COP greater than 1. However, even comparing it to a mechanical heat pump it is rather inefficient, as these can have COPs of 3-4, at a temperature differential of 10°C or more. A simple air-air heat exchanger is more efficient than a mechanical heat pump, even. The more expensive Santa Fe dehumidifiers use a heat exchanger in front of the evaporator core, to obtain more latent heat removal for the same amount of energy.

WGTR - 3-12-2018 at 17:18

I had to test this, because I just gotta...do I need a better reason?

I was expecting efficiency to be 1-2%, but it was about 10%. This means that I am a very wrong little squirrel. Sorry.

Anyway, I had some Tellurex CZ1-1.4-127-1.65's laying around the lab, so I decided to check them out. These would be considered some of the "good ones", if one was comparing brands. Basically, I drove one TEC with an adjustable power supply, and placed a 3Ω load (measured initially to calibrate) on the output of the other TEC, and measured the voltage across it. I was able to either directly measure or calculate voltage, current, and power input; as well as voltage, current, and power output. From this, I could calculate total efficiency from input to output, as well as estimate efficiency for each TEC by taking the square root of the overall efficiency. This assumes that both TECs are closely matched, or course.

The hot side aluminum block weighs 344g, and the cold side 353g.

Results are plotted in the chart located in the attached image.

For fun, someone can try calculating the COP from this data, if it is possible.
IMG_4629.JPG - 1.4MB IMG_4630.JPG - 1.6MB

Sulaiman - 9-12-2018 at 02:02

The TECs you are using have 127 couples, just as a TEC1-127xx
You have cleverly operated at much lower than 12V,
which is the common operating voltage - but gives very poor COP.

My first prototype used two TEC1-12706 at 12V each ... poor performance
My second prototype used four TEC1-12706 with each operated at about 6V for better COP / heatsink utilisation.
The four new TEC1-12706 were faulty and I got a full refund.

I was looking for original TECs but
I've decided to try cheap eBay peltiers one last time,
so I now have 4x TEC1-12715 on order to use at around 4 to 6 Volts.
Hopefully I will have a working water cooler around Xmas :D
- or I'll be applying for another refund. :(

I suspect that the efficiency as TEC & TEG are NOT the same
(I think that the TEC should be more 'efficient' than the TEG)
so SQRT(overall efficiency) may not be accurate - but a good indication.

WGTR - 9-12-2018 at 08:13

Quote: Originally posted by Sulaiman  
I suspect that the efficiency as TEC & TEG are NOT the same
(I think that the TEC should be more 'efficient' than the TEG)
so SQRT(overall efficiency) may not be accurate - but a good indication.


I suspect that you are right! In my experience, a TEG will only get about 1-2% efficiency. 10% seems way too high. In fact, doing some more reading shows that others have calculated the efficiency by measuring direct resistive heating input vs TEG output, and it's only about 1-2%, as expected. So my assumptions in calculating the efficiency are almost certainly incorrect. I'm still trying to wrap my head around how this works.

It's interesting though, that whether resistive heating or a heat pump from another TEC is used to power the TEG, the overall efficiency from input to output seems to be very similar.

Well, hopefully your new coolers will be very cool indeed...

Quad TEC1-12715 water cooler for condensers (FAIL)

Sulaiman - 16-12-2018 at 09:36

Good News;
. The 4x TEC1-12715 arrived safely
(from https://www.ebay.co.uk/itm/TEC1-12715-Heatsink-Thermoelectri... )
. In my test rig, my so far limited measurements agree closely wth the Hebei datasheet
... HOORAY!

Bad News;
I calculated the thermal resistance of the heatsinks that I am using (AK-786) to be 0.25 oC/W measured with 12 Vdc supplied to the fan.
(0.225oC with fan at 13.2 Vdc)
this is equivalent to each TEC module having its own 0.5 oC/W heatsink.
The TEC1-12715s have a slope of -1W/oC (dQc/dT)
so for low voltages where the COP is >=1, lower than ambient temperaures can be acheived,
at about 7V per device the cooling power of my heatsink(s) becomes less than the heat generated by the TEC(s),
so the minimum available temperature INCREASES slightly with increasing voltages.
The TECs are performing as per specifications
- the heatsink(s) that should be OK for 1x TEC1-12706 or less each,
are completely inadequate for 2x TEC1-12715 each.

For distillations such as ethanol and water the setup would be
more efficient just as a water-cooling radiator without the TECs :(

Overall quite a disappointing exercise,
but on the bright side,
I have invented a new type of 'do-nothing' machine :D

I now think that
. my original 2x TEC1-12706 were not performing to specifications
. the four new TEC1-12706 were useless
. the four new TEC1-12715 perform close datasheet graph values
. minimum acheivable temperature is not a good measure of usefulness for distillations but is indirectly useful
. TECs need LARGE EFFICIENT heatsinks - especially when using 'not-necessarily-full-specification' modules.


... now working on 'Plan D'

[Edited on 16-12-2018 by Sulaiman]

wg48 - 16-12-2018 at 17:47

Quote: Originally posted by Sulaiman  


. TECs need LARGE EFFICIENT heatsinks - especially when using 'not-necessarily-full-specification' modules.


... now working on 'Plan D'

[Edited on 16-12-2018 by Sulaiman]


More specifically you need a heat-sink with a thermal resistance such that its temperature rise when dissipating the heat from the TECs is small compared to the temperature difference of the TEC or you will use up the available TEC temperature difference getting heat to the heat sink and or more power. Probably with those TECs 0.1C/W or better, Thats a big expensive heat sink if its aiir cooled.

Dual TEC1-12715 (Partial success)

Sulaiman - 31-12-2018 at 06:11

Using my two AK-786 cpu heatsink/fan assemblies (each <0.25 oC/W measured)
with one TEC1-12715 per heatsink,
both TECs wired in series and connected to 15 Vdc,
Ta = 23 oC, TEC current = 6.3 Adc ;

Approximate heat pumping capacity :
00W @ -10oC
40W @ +20oC
at other temperatures t.b.d.

NOT powerful enough for cooling condenser water on its own. :(

So, if anyone is contemplating similar,
do as WG48 suggested and get MUCH better than 0.25 oC/W heatsinks .
.. e.g. 0.1 oC/W ... or better :P

OR ... convert an old refrigerator for the purpose.
TEC cooling does not seem cost-effective for cooling condenser water.
They may be suitable for an m.p. determination machine,
or very low power / slow distillations near 0oC ?