In a power circuit once the power requirements of the transistor are meet , which is a dump of thumb rule to select or discard a MOSFET transistor based on the switching frequency?

For example for the IRF540, what are the frequency bounds to work with this transistor?

I guess the switching losses emerges as the more relevant factor if a specific efficiency requirements have to be achieved, but I'm looking for a simpler general rule, something like:

The cutoff freq of the LC filter should be 10 times smaller that the switching frequency in a DC - DC converter


My rule of thumb is to add the rise time and fall time (as specified in the data sheet), multiply by ten and take the inverse to find the maximum frequency. So for the IRF540, there is a 44 ns rise and 43 ns fall time giving you 87 ns. Multiply this by 10 to give 870 ns then invert to get 1.15 MHz as the maximum operating frequency.

But, this is a very general rule and there is nothing stopping you modifying this to be a factor of 20 or even 100 and I will break my own rule on some jobs where I need a little bit more efficiency.

  • \$\begingroup\$ Do realize that those 44 ns and 43 ns switching times are for the test conditions as specified in the datasheet. Note how the series resistance through which the gate is charged/discharged is 9.1 ohms. This resistor needs to be low because you want to charge/discharge the gate quickly and that gate behaves as a capacitor of up to a few nF. So up to what frequency you can switch also heavily depends on choosing the proper way to drive the MOSFET. A simple and weak inverter from some logic chip will not cut it ! \$\endgroup\$ – Bimpelrekkie Jun 2 '16 at 7:29
  • \$\begingroup\$ Oh, you use the same rule! Interesting. \$\endgroup\$ – winny Jun 2 '16 at 7:31
  • 1
    \$\begingroup\$ @FakeMoustache I totally agree but rules of thumb are meant to be simple I suppose AND may or may not contain "smallprint". \$\endgroup\$ – Andy aka Jun 2 '16 at 8:35

That's difficult to answer in general since it comes down to how it's used, mainly soft versus hard switching but for a simplified hard switching topology, you want to limit the switching losses to some resonable level, say 10% switching losses and 90% conduction losses, then you can take a look at the rise and fall times of 43/44 ns. Ignoring many many important things, ideal versus real drive would be the top of which, you end up with a maximum of 1 MHz, a number you won't be able to reach in real life. 500 kHz perhaps if you do everything correct and think about everything.


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.