[EDIT: Found a couple of 12715 units in the pile that did suck the full amperage, but despite excellent cooling of the hot side, could not deliver much cooling.]

I'm waist deep in project to keep shot glasses full of beverage near or below zero °C.

After a reading of the charts of the easily available TEC1 series, I chose the TEC1-12706 and TEC1-12715 for my experiments.

As I understand it from the tech sheets, a module should can come close theoretical dT as the maximum heat absorption should also be available as long as full voltage and amperage are applied and the hot side is maintained cold enough.

No worries there. A simple cooling loop or even decent air cooling computer heat sink keeps the hot side cold and we get 95% of dT. Achieving cold side temps of -15 °C is no problem. In fact, we have achieved lower temps on the cold side simply by chilling the cooling loop of the hot side.

Clearly temperature is not an issue for us to achieve, but where we run in to issues is current and Wattage. For instance, a TEC1-12715 connected to a 12 VDC PSU capable of supplying 150 Amps is drawing a mere 60 W. Why is this?

In terms of heat dissipation, we have even gone so far as to use Stannol Kristall 511 solder that binds ceramic very well to aluminum and copper for very high W/mk. We have also (for reference) tried high W/mk standard CPU/GPU creams such as Grizzly Kryonaut, Grizzly Hydronaught, and Noctua's best. We also, since this is not a high compression project, tried silicone rubber heat transfer glue of 2.9 W/mk and a special heat transfer epoxy of 8.7 W/mk.

We have tried bonding in every way conceivable - again, there is no problem with temperature, but dissipation of Wattage and module power consumption are low. I think we can all agree that a single TEC1-12715 affixed to cooling source maintaining 30 °C and the cold side being at -15 °C is normal. Literally soldering or using a high W/mk glue or grease to a milled and polished aluminum shot glass should freeze 25 °C liquid in minutes or less. At best, we can remove 1 °C every 3 minutes and at best achieve 8 °C at the coldest.

In tests, the 12715's did cool the intended liquid much faster than the 12706's. Ideally, the TEC units will cool a basin of salt water to sub-zero C and our shot glasses will sit in the water, but we are open to other ideas.

We've seen TEC machines that can cool a bottle of wine or soda to 7 °C in 12 minutes. Clearly we re missing something.

So... what are we not understanding or do simply have a bad batch of TEC units?

  • 1
    \$\begingroup\$ Are you sure the glass isn't absorbing heat from the room? Have you done your tests with the outside of the glass insulated, or with it bare to the air? \$\endgroup\$
    – TimWescott
    Commented Jun 6, 2019 at 4:43
  • \$\begingroup\$ Smart question. In answer, yes, we have tried 3" of PU on all 4 sides and top, Of course, we do need to factor in the joules required for the cooling of the material of the container and have done so in the advanced calculations. The issue here is current. Why would a TEC1-12715 only draw about 5A@12VDC? Do we need to get up the Vmax to get the full 15A? If so, are we talking about a Boost controller (can't find one for high amperage). I suppose we could rectify and ripple stabilize the 220VAC local power and use TECs in series to get high amps... \$\endgroup\$
    – J. Little
    Commented Jun 6, 2019 at 5:11
  • \$\begingroup\$ is this for a contest of some type? \$\endgroup\$
    – jsotola
    Commented Jun 6, 2019 at 5:14
  • 1
    \$\begingroup\$ Sure. Put any liquid you want in the glass. It just changes the physics. Here... let's take an example of water since it works well for Joules and Watts. (But, remember, the real help we need is why are the TEC units sucking only 5A of power?) Water: 4J per gram per degree - 1ml=1g. To cool 10 C then for 40ml water: 4*40*10 = 1600 J To determine watts, realizing that 1W changes 1J by 1 C in one second, divide by the amount of time you will allow for the cooling to happen \$\endgroup\$
    – J. Little
    Commented Jun 6, 2019 at 5:20
  • 1
    \$\begingroup\$ Two things. (1) TECs really suck at efficiency. If you are going to the bother of circulating water, then just get a standard compression pump cooler to cool the water, higher capacity, way lower power consumption. (2) TECs are fragile inside, heat/cool cycles eventually detach the elements. Strive for minimum dT possible that still meets objectives to minimise failure on cycling. Are you getting them from a reliable source? If you pay little enough money (zBay?), then they can come pre-detatched, or ready to fail at a moment's notice. \$\endgroup\$
    – Neil_UK
    Commented Jun 6, 2019 at 5:57

1 Answer 1


For instance, a TEC1-12715 connected to a 12VDC PSU capable of supplying 150 Amps is drawing a mere 60W. Why is this?

Because (if the data sheet can be trusted) at the maximum delta-T, the thing draws 15A at 16-17V. When you have a passive device, the current drawn is a function of the device's characteristics and the applied voltage. Usually the current drawn increases with applied voltage.

  • \$\begingroup\$ TW - Yes, we are assuming the lower current draw is because we use 12VDC and don't have a ready way to draw hundreds of amps at greater than 12V. We thought of buck converters, but the price for high power ratings is way too high. Specialty 17VDC supplies are also out of reach. At this point, we are seriously considering rectifying the 220VAC with a ripple cap using high amperage diodes and using the TEC units in series. \$\endgroup\$
    – J. Little
    Commented Jun 8, 2019 at 12:21
  • \$\begingroup\$ TW - We are not in a passive operation. We have water, devices, etc. that need cooling on the cold side. That does not increase the current but it does raise the cold side temp (of course.) So, are you saying to do some experiments at 17 VDC and see the difference as the graphs show? If so, couldn't we simply add some diodes to the negative lead headed to neutral (say if 1.5v forward voltage diodes then we would need 3 diodes to raise the voltage to 16.5VDC? \$\endgroup\$
    – J. Little
    Commented Jun 8, 2019 at 12:26
  • \$\begingroup\$ @J.Little: How do you expect adding diodes to increase the voltage to the peltier device? Are you talking about modifying the power supply in some way, or just putting them in series with the peltiers? \$\endgroup\$
    – JRE
    Commented Jun 8, 2019 at 13:21
  • \$\begingroup\$ I'm not sure where you're getting your power supplies, but if there are any surplus electronics houses you can buy from, you might be able to find something. \$\endgroup\$
    – TimWescott
    Commented Jun 8, 2019 at 16:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.