I have got a question about my Peltier driving circuit. First of all this is a drawing of my circuit:


simulate this circuit – Schematic created using CircuitLab

I am confused because only 1 of my IRFZ44N gets very hot, even though I ran the same gate signal to both of them. The Peltier that I have uses have maximum current of 10A (TEC1-12710) so I thought that IRFZ44N should be able to handle it (the datasheet says 49A continuous drain current.)

Any idea why only 1 of my IRFZ44N gets very hot?

  • 3
    \$\begingroup\$ Are You really driving them with 1Mhz signal? \$\endgroup\$
    – fifi_22
    Nov 8, 2020 at 19:08
  • 1
    \$\begingroup\$ Are both FET's IRFZ44N? They're not on your schematic. \$\endgroup\$ Nov 8, 2020 at 19:10
  • 2
    \$\begingroup\$ Why the two MOSFETs and inductors with one PWM signal? \$\endgroup\$
    – winny
    Nov 8, 2020 at 19:23
  • 1
    \$\begingroup\$ I agree with the comments. I would not drive the mosfet that fast. Your switching losses are going to be very high with that amount of current. 100khz would be fine with how big of an inductor you're using. You could probably even go down to 50khz and still be fine. I would also only use 1 mosfet instead of two. One reason I can think of as to why one is getting hotter is that one of the mosfets is turning on faster than the other and one of them is seeing more switching losses or conduction losses. \$\endgroup\$ Nov 8, 2020 at 19:55
  • \$\begingroup\$ Both of my mosfet are irfz44n, sorry i made a mistake on my schematics \$\endgroup\$
    – shafiyyah
    Nov 9, 2020 at 4:15

1 Answer 1


Since no MOSFET drivers are shown on the schematic, I assume the FETs are driven directly from arduino pins. IRFZ44 requires more than 5V Vgs to turn on fully. With 5V gate drive it will be somewhere in the linear region. So you must either use a FET with logic level gate drive, and a proper MOSFFET driver with a 5V supply, or keep your FETs and use a proper MOSFET driver with a +12V supply.

Likely explanation for the uneven heating:

FET RdsON has a positive tempco so FETs in parallel share current well. Resistance in the hottest one increases which directs current to the other FETs.

However in linear region it's the opposite. Threshold voltage has negative tempco, so it goes down as it gets hot:

enter image description here

This means your FETs run in conditions that are ideal for thermal runaway.

Additionally, the absence of gate resistors mean they are likely to oscillate due to layout parasitics ; and 1MHz is way too high a frequency. You must calculate inductor ripple current according to frequency and do the math to design a buck converter.

With a proper FET driver chip using 12V power supply you should be able to drive these FETs at 100kHz without trouble.

  • \$\begingroup\$ How about changing to IRLZ44N is it possible to drive it fully on using arduino pins? Since the datasheet says it can handle 80A ID at about 5V Vgs \$\endgroup\$
    – shafiyyah
    Nov 9, 2020 at 20:29
  • \$\begingroup\$ Yeah, it'll turn on at 5V Vgs. It's an obsolete part from 1997 though, with high gate charge, so it'll need a strong gate driver for >100kHz... \$\endgroup\$
    – bobflux
    Nov 9, 2020 at 20:34
  • \$\begingroup\$ Ohh I see, sorry if this is an obvious question, but i want to ask one thing about IRFZ44N. If I follow the chart that you gave me, it seems that at 5V gate voltage we can use up to 20 A of ID. So why do we want another mosfet driver for this case? \$\endgroup\$
    – shafiyyah
    Nov 9, 2020 at 20:39
  • 1
    \$\begingroup\$ I mean you need a driver chip with strong output current to drive your MOSFET because your microcontroller won't have enough output current capability. \$\endgroup\$
    – bobflux
    Nov 9, 2020 at 20:52

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.