Would it be possible to use a standard 12V DC LED dimmer together with a thermoelectric element to control the temperature? I'm talking about a very simple application with no need of fine tuning. Assuming that the output current provided by the dimmer is enough to drive the TE, I'm interested in knowing if the use of a PWM LED dimmer could possibly affect the life of the TE. A number of studies have shown that the PWM frequency has little or no significance on the life of the TE.

  • \$\begingroup\$ "A number of studies have shown that the PWM frequency has little or no significance on the life of the TE" there is your answer. Can you link to these studies? \$\endgroup\$ Aug 20 '13 at 17:43
  • \$\begingroup\$ tetech.com/publications/pubs/ICT99MJN.pdf \$\endgroup\$
    – Francesco
    Aug 20 '13 at 18:12
  • \$\begingroup\$ Are you thinking of incorporating this technique into a product that may sell several-to-many units a year. If so then, in the absense of hard facts about its viability, prototype testing is the key to understanding any problems and modifications. This ultimately means, for a one-off, in the absense of good data, try it and be sure to let us know what happens. \$\endgroup\$
    – Andy aka
    Aug 20 '13 at 21:03

Yes - you could use a dimme to control a TEC. But do note the following:

With a TEC / Peltier cooler the aim if not using DC drive, is to not introduce significant thermal variations with the driving waveform. If the periods of the lowest frequency components of the driving signal are short compared to the thermal time constants then the TEC will not "notice" changes in thermal input.

If you lowered the PWM frequency such that a control cycle was say 10 seconds long, or if you used an on/off thermostat control (electronic or mechanical that cycled power based on measurable variations of temperature, then you will shorten lifetime. This is due (mostly, probably) to mechanical contraction/expansion cycles "working" the many bonded junctions and their bonding materials).

Olin has on several occasions made the interesting and seldom made point that if you want operation at less than 100% thermal output then you are better off using filtered DC than PWM. See this answer here for his reasoning.

RC PWM filter have low efficincy (Efficincy ~+ PWM duty cycle):
Worth noting is that PWM with smoothing using an RC filter is equivalent to using a voltage divider as load with TEC as bottom element and a virtual resistor dropping the supply voltage to the operating point as the upper element. The power dissipated in the virtual resistor is lost - unless you use eg a buck converter or an LC (nominally lossless) filter to reduce the voltage. Note also that an RC filter with PWM will have an efficiency of about the same value as VPeltier / Vsupply = the PWM duty-cycle (eg 30% for 30% on PWM ) and this will apply regardless of filter size. If you use NO formal series R and charge the capacitor with short pulses Murphy will not be fooled - the very large currents that will flow for very short periods when very short pulse and no formal R are used will dissipate just as much energy overall as if a formal larger value resistor had been used.

  • \$\begingroup\$ I'm sorry, but you have gotten what I said backwards. You DO want to drive Peltier devices with average DC, not by banging back and forth between full on and full off, like you could do with a motor, for example. See electronics.stackexchange.com/a/28637/4512 for a good example of where I mention this. \$\endgroup\$ Aug 29 '13 at 20:04
  • \$\begingroup\$ @OlinLathrop - Don't be sorry, it's my mistake :-). Having inverted your advice, It didn't make total sense so I explained why I thought you'd have said it :-). However, the point that I made re efficiency is one which may bear thinking about - and one I don't recall seeing mentioned before. Imagine a say 20% on PWM cycle and a say 10V supply. If a Peltier device was a true constant current load then PWM with no filtering gives 100% energy transfer to the Peltier. A 1A CC Peltier would dissispate 10W/2W at 100%/20% PWM. BUT as soon as you add filtering you significantly affect efficiency ... \$\endgroup\$
    – Russell McMahon
    Aug 30 '13 at 8:12
  • \$\begingroup\$ @Olin - with say 20% duty cycle 10% supply RC filter abd cap with delta ripple small. Vcap = 2V. Iout mean = 1A for CC. Iswitchj mean = 1A but when on Iavgon = 5A. Filter R dissipation = Vres x Ires x duty_cycle = 8 x 5A x 0.2 = 8Watt. Pload = Vl x Il = 2 x 1 = 2W. Z = 20%. ie with an RC filter = duty cycle. You need an LC filter or a buck converter or equivalent converter // This sounds wrong and AFAIK nobody mentions it but I can't instantly see that I've done something silly. // \$\endgroup\$
    – Russell McMahon
    Aug 30 '13 at 8:21
  • \$\begingroup\$ @OlinLathrop - Back of brain suggests that your arguments re PWM versus filtered driving are in part based on how you derive the filtered voltage. If I get to it I'll go over it with some real world parameters and see how it looks. \$\endgroup\$
    – Russell McMahon
    Aug 30 '13 at 8:28
  • \$\begingroup\$ @OlinLathrop I've modified my answer to reflect your actual comments - and noted that filtered DC needs and LC filter or buck converter if efficiency is to be maintained. \$\endgroup\$
    – Russell McMahon
    Aug 30 '13 at 8:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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