Maybe the underlying question is what the Voltage-Current curve looks like. Can I drive it from a voltage source (like you drive a heater) or from a current source (like you drive an LED)? Or even different than those two options?

ADDITIONAL1: Say (hypothetically) I have two commercially available identical Peltier's, they are spec'd 6V/3A. Can I connect these in series to a 12Vdc power supply without any worries?

CONCLUSION1: Current/Voltage load curve is reasonably linear, both driving from current or voltage source will do fine as long as the device is operated within its spec. (Olin Lathrop, Russell McMahon)

CONCLUSION2: Don't drive a Peltier from PWM, power loss due to current increase, grows more rapidly than cooling power. (Olin Lathrop)

CONCLUSION3: Beware the mechanical wear of the device with continues cycling. Eg. don't use a thermostat on/off controller. (Russell McMahon)


3 Answers 3


Peltier devices work on current, but usually have significant enough resistance so that voltage control is possible.

Peltier devices are one of the few things you do not want to run with pulses, particularly in cooling applications. The cooling effect is proportional to current, but the internal heating due to \$I^2R\$ losses is proportional to the square of the current. Starting at 0, increasing current causes increasing cooling. However, at some point the resistive heating due to more current outweighs the additional cooling power of the higher current. More current beyond this actually therefore causes less overall cooling. The maximum cooling current is one of the parameters that should be supplied by the manufacturer.

While maximum cooling occurs at some specified current, efficiency steadily decreases with increasing current. Therefore you don't want to PWM a peltier cooler between 0 and the maximum cooling current. Driving it at the steady current to produce the same overall cooling is more efficient.

Of course the microcontroller regulating the temperature will still produce PWM pulses. These pulses need to be filtered so that the Peltier device sees relatively smooth current. The general rule of thumb is to try to keep the ripple below 10% of nominal, but of course that is just a tradeoff someone picked. Fortunately, this is usually not a difficult requirement to design to.

  • 5
    \$\begingroup\$ I like the addition of your PWM-warning. In this microcontroller-age we live in, driving load with PWM is often a very obvious choice. \$\endgroup\$
    – jippie
    Commented Mar 24, 2012 at 15:09
  • 1
    \$\begingroup\$ Interesting to see our slightly different perspectives but similar end advice. I see your point in calling them current driven but think it's probably somewhat moot - not quite like deciding whether a resistor is current or voltage driven. Maybe a thermistor for comparison. ie controlling either is doable enough but one may have an advantage. . \$\endgroup\$
    – Russell McMahon
    Commented Mar 24, 2012 at 15:15
  • \$\begingroup\$ Reviving an old question: I am confused by the comment about not using PWM. If I have a Peltier device which I want to control from an MCU (I'm interested in heating), how would I control temperature except by PWM? \$\endgroup\$
    – spring
    Commented Mar 23, 2013 at 14:06
  • 1
    \$\begingroup\$ @skinny: I thought I answered this in the last paragraph. What specifically is unclear? \$\endgroup\$ Commented Mar 23, 2013 at 14:19
  • 1
    \$\begingroup\$ @skinny: Read what I wrote. As I said, the PWM output needs to be filtered before being presented to the Peltier device. \$\endgroup\$ Commented Mar 23, 2013 at 16:18

Added 12 years on: I've rereread my and Olin's answers.
I agree with Olin (no surprise).
Do what I say and then also do what Olin says. ie

  • If using PWM, filter it so that the Peltier element sees lower voltage DC.
  • Never use on/off switching at rates that allow the element to thermally cycle as this causes mechanical stresses.
  • Even very low switching rates (the order of minutes with an occasional cooling topo-up) are less efficient than using continuous low voltage DC instead, if this can be provided efficiently.

Peltier devices have a reasonably linear load curve.
They are voltage driven devices but you could safely enough drive them from a voltage or a current source as long as you do not exceed their maximum voltage or current ratings. eg a 12V, 5A 60 Watt unit may be operated at any steady voltage <= 12V or any steady current <= 5Aas long as neither maximum value is exceeded in any case.

But -

  • A Peltier device MUST NOT be constantly thermally cycled as part of the control scheme

What you must avoid is controlling them with step changes in voltage with control signal time constants that are long compared to the device's thermal time constant.
That is - if a the control scheme is aimed at holding a device or a controlled space or an object at a constant temperature it should provide an essentially DC voltage and current or PWM in the 10's of Hz plus range and NOT PWM or thermostat or on off switching such that noticeable heating and cooling occurs during the on and off times of the control cycle. If thermal cycling occurs as part of the control process the device will rapidly be destroyed mechanically.

Power supply ripple at mains frequencies is acceptable.

Thermal Cycling:

Say >= 1 kHz PWM rate needed - Nord ferrotech Peltier manufacturers

Note that Peltiers used fro test cycling SMD components also fail due to cycling

IGBT thermal cycling testing - nbo Peltiers BUT seems to directly apply

Peltiers specially made to allow long thermal cycling lifetimes - UWE

Same - Ctech

Same Ferrotec


  • \$\begingroup\$ Wait... so turning one on/off at 3 minute cycles would destroy it quickly? \$\endgroup\$ Commented Aug 6, 2014 at 21:06
  • \$\begingroup\$ @AnnonomusPenguin - Hmm ......... 3 minutes is quite long. You'll very obviously get substantial thermal cycling but it may be bearable. I do not know what the limits are. 3 min//cycle = 20/hour = 480/day =~ 175,000/year. I'd say that it may be getting fairly exercised by then. \$\endgroup\$
    – Russell McMahon
    Commented Aug 6, 2014 at 21:36
  • \$\begingroup\$ so would it be better to do 1/2 a minute increments? 10 minutes? I don't want to go too short as that will loose efficiency, but too long will be a lot of thermal changes. Likewise, it is probably not the best to turn it on and off every ten seconds, either. \$\endgroup\$ Commented Aug 6, 2014 at 21:39
  • \$\begingroup\$ @AnnonomusPenguin - As noted above and elsewhere - PWM of say 10 Hz or faster frame rate with adjustment of the mark-space ratio to adjust cooling is easy and avoids the thermal cycling issues. There are many minor mentions of thermal cycling issues on web but I've yet to find a good discussion. | See addition at end of my answer. \$\endgroup\$
    – Russell McMahon
    Commented Aug 6, 2014 at 21:59

Most of the microcontrollers I am familiar with do PWM in the hundreds of cycles per second range. No thermoelectric module is going to be able to distinguish that from a steady voltage.

Also, (google it) there is a paper out there where they tested PWM cycling with rates of 1/10s all the way up to 1000/1s rates and the peltiers did not exhibit any decline in performance over thousands of hours. The one that cycled every 10 seconds did exhibit temperature fluctuation due to the slow response time.

In any case, PWM is utterly safe for controlling a peltier.

I fear this thread has a bunch more misinformation in it, but I'm not knowledgeable enough to point it all out.

  • 2
    \$\begingroup\$ Could you add a link in your answer to the paper you're referring to? \$\endgroup\$
    – jippie
    Commented Oct 16, 2013 at 5:22
  • 1
    \$\begingroup\$ rittal-corp.com/literature/pdf/… \$\endgroup\$
    – jippie
    Commented Oct 18, 2013 at 18:28
  • 3
    \$\begingroup\$ "No thermoelectric module is going to be able to distinguish that from a steady voltage" --- Are you sure? Have you worked out how much parasitic inductance and capacitance is present, and calculated that it effectively sees a DC voltage/current? Because if not, then it may likely experience fluctuating voltages/currents, and at that point the I2R power loss points discussed in other responses become valid. Try PWMing a power resistor vs. driving it at an equivalent constant current to see the difference. \$\endgroup\$
    – gbmhunter
    Commented Sep 30, 2015 at 4:18
  • 1
    \$\begingroup\$ This answer sounds very wrong. Would you PWM a diode or an LED without an LC filter? \$\endgroup\$
    – Navin
    Commented Jul 29, 2019 at 7:28
  • 1
    \$\begingroup\$ Actually yes, there would be no problem with PWMing a LED without a LC filter, as long as you PWM it fast enough that the eye cannot discern any flickering... 1KHz is usually flicker free. The problem with Peltiers is that as soon as current turns off the hot side of the element starts cooling rapidly and the thermal shock may damage the ceramics in it. 1KHz is 1ms per cycle, so 0.5ms cooling down every cycle. Above a certain frequency it won't have time to cool down enough to cause damage. \$\endgroup\$ Commented Dec 11, 2019 at 3:03

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