I am panning to build a soldering station and would like get myself clear about driving MOSFET to control heating element using microcontroller.

Microcontroller will be ATmega328p (Arduino Pro Mini) and MOSFET planning to use is IRLZ44N. Heating element rated 50W at 24V .

  1. I am planning to use PWM frequncy around 30 kHz. Is there anything bad with this frequencey particulary for this application?

As per MOSFET datasheet.

  • \$t_{d(on)} = 11 ns\$
  • \$t_r = 84 ns\$
  • \$t_{d(off)} = 26 ns\$
  • \$t_f = 15 ns\$.

If I want to limit total ON and OFF time within 1% of a PWM cycle. Then least PWM cycle druration I can have will be (\$t_{d(on)} + t_r + t_{d(off)} + t_f) \times 100 = 13600 ns\$. So I can use PWM frequencey upto 73529 Hz (73 kHz) without problem. Could you please confirm my calculations are correct?

  1. Is it necessary to have MOSFET driver circuit instead of driving MOSFET directly with MCU output pin for this application?

My calculation tells I can directly drive MOSFET with MCU output pin. As per the datasheet \$Q_g = 48 nC\$. With PWM frequencey 30 kHz, maximum current required to charge/discharge MOSFET gate will be:

\$48 \times 10^{-9} / (1 / 30000) = 0.00144 A = 1.44 mA\$.

(Update: My above calculation is incorrect as per @Unimportant's comment below. Without resister it would take \$48/95 = 505 mA\$ to charge the gate, which is undesirable for ATmega328p IO pin. So, there should be a resister or gate driver)

So, current will be well within the allowed limit (40 mA) and I think MOSFET can be directly driven by IO pin in my application. Could you please confirm my finding?

Thanks in advance.

  • \$\begingroup\$ PWM cycle duration you have mentioned is for the condition given in the datasheet. In practical total switching time will be more than you calculated. \$\endgroup\$
    – CNA
    Commented Mar 28, 2018 at 6:52
  • \$\begingroup\$ @Dhans However, it would be safe to drive it at 30KHz, isn't it? \$\endgroup\$
    – Junaid
    Commented Mar 28, 2018 at 7:23
  • 5
    \$\begingroup\$ Is there a point in driving the heating element that fast? With the heat capacity of the tip, I would think 100 Hz or even less would be plenty, and save you some switching losses. \$\endgroup\$
    – Dampmaskin
    Commented Mar 28, 2018 at 8:04
  • 1
    \$\begingroup\$ way too fast for a high-inertia system like a soldering iron. if you want good perf, it's far better to focus on PID tuning than overclocking the heater drive. \$\endgroup\$
    – dandavis
    Commented Mar 28, 2018 at 9:59
  • \$\begingroup\$ But you cannot spend the entire PWM period transitioning, so you do not have 1/30000 transition time. You want the transitions to be fast, say 300ns for 30Khz for example, that makes 48nC/300nS = 160mA. \$\endgroup\$ Commented Mar 28, 2018 at 10:42

1 Answer 1


Your heating element at 50W rating driven at 24V will be drawing just a small bit over 2A. So the MOSFET that you linked to should be more than suitable to switch this load.

There is no need to overdo the PWM switching frequency since the load is a thermal load with a long time constant. You could very comfortably be able to use a switching frequency of 500Hz to 2kHz which can easily be generated by the ATMega328. There is no reason to have to go into computing the maximum switching frequency of the MOSFET for this application.

In analyzing whether the ATMega328 pin can directly drive the MOSFET study Figure 1 graph of the MOSFET data sheet. With a gate drive of 4V (lower than the 5V that the MCU can actually provide) the 2A load line is way below the related VGS curve. This means that the MCU can comfortably drive this MOSFET for a 2A load. A MOSFET like this still has a pretty high gate capacitance and so it would be advisable to put a small values resistor in series between the MCU and the gate to limit the peak transient current some during switching.

enter image description here

  • \$\begingroup\$ In my opinion an additional mosfet driver would still be a cleaner solution though. Especially when switching at higher frequencies, otherwise you will spend a relatively big part of the switching period on transition time. Also possible ringing on MCU pin may be reduced. \$\endgroup\$
    – Jogitech
    Commented Mar 28, 2018 at 10:36
  • \$\begingroup\$ What sort of Graph is that? Log? \$\endgroup\$
    – MaNyYaCk
    Commented Mar 28, 2018 at 11:21
  • \$\begingroup\$ 2D Log Log Chart. \$\endgroup\$ Commented Mar 28, 2018 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.