I am trying to make a n-channel mosfet switch a load from a high voltage (250V). I am using the n channel C2M0280120D mosfet. the circuit looks something like below:


simulate this circuit – Schematic created using CircuitLab

so in this configuration, when ever the uC pulse rises (uC generating pwm between 5V and 0V), the Vgs on M1 would rise too. I measured the Vgs on M1 and its a pulse at 5V peak for 250ns. the datasheet states that maximum threshold voltage is 4V, so one would assume this would be sufficient, but when I measured the source of M1 (node S2) against power ground (pgnd), it is a pulse but with a peak of only 30 odd volts, instead of 250V.

I thougt it might be the case that the mosfet is not properly overdriven and it still is in its linear region, so I tried chopping a 15V supply to make the gate signal with a higher amplitude, this time my Vgs is a pulse with 10V peak. The circuit is shown below:


simulate this circuit

I also put a 1uF cap (as it is in the first schematic) just before the trasnformer, but I still get less voltage than expected on node S2. (waay less for that matter.)

so if this is not for the larg resistance between drain and source (since it is turned on properly) what else could be the issue?

Thank you in advance for your time.

  • \$\begingroup\$ What is your desired pulse waveform, including rise, fall times and pulse width? How do these compare with gate drive pulse waveform? From the values of C1.C4 I suspect you are trying to generate a short pulse with slow turn on/off times. You'll need good simulation technique and a fast sampling scope when you get to real HW to see and understand what you're doing. \$\endgroup\$
    – user16324
    Jul 24, 2020 at 12:02
  • \$\begingroup\$ @BrianDrummond I am trying to make a pulse with about 9ns rise time and 50-100 ns fall time. the simulations are fine (did it in ltspice), and I don't know the specs of my scope but my coworkers use it for rf high frequencies (GHz and above) so i'm assuming thats fine. would the problem be entirely because of the mosfet? I tried applying 15V (Vgs) pulses to drive my mosfet, but I still get a pulse that has apeak of 30-40V instead of 250. I tried increasing the load to reduce the current to see if that would allow the voltage to rise further, but to no avail! \$\endgroup\$
    – NeuroEng
    Jul 24, 2020 at 13:27
  • \$\begingroup\$ Try reducing R3, or otherwise increase drive current to overcome the gate threshold step faster. 9 ns risetime is quite challenging even with only 260pF Cin. \$\endgroup\$
    – user16324
    Jul 24, 2020 at 14:00

1 Answer 1


I measured the Vgs on M1 and its a pulse at 5V peak for 250ns.

Just look at the graph in the MOSFET data sheet. This one: -

enter image description here

With 10 volts on gate w.r.t. source it's barely breaking the sound barrier in terms of performance and it's not even listed for a \$V_{GS}\$ of 5 volts.

the datasheet states that maximum threshold voltage is 4V, so one would assume this would be sufficient

Always best to check the graphs.

  • \$\begingroup\$ I don't know if its my mosfet or what, but I tried applying a pulse of 15V and I still get the same output (pulse with a peak of 20-30V rather than 250V or sth close). when i simulate the circuit and replace the mosfet with just a switch, the simulation results are all fine, does this mean that it's definitely my mosfet thats causing the problem? \$\endgroup\$
    – NeuroEng
    Jul 24, 2020 at 13:29
  • \$\begingroup\$ What transformer are you using - link the data sheet please. It's not definitely the MOSFET but it's likely. It could also be the 1 nF capacitor - can you look what happens to the voltage on S1 and S2 and take a screen shot? \$\endgroup\$
    – Andy aka
    Jul 24, 2020 at 13:41
  • \$\begingroup\$ I am using this transformer: docs.rs-online.com/89cb/0900766b80d6b42d.pdf. its worth mentioning that the signal that is fed into the transformer is a: 15V peak that is on for 100n and a period of 200us. \$\endgroup\$
    – NeuroEng
    Jul 24, 2020 at 13:45
  • \$\begingroup\$ You should be able to model that transformer in LTSpice by the way. It does look suitable however. \$\endgroup\$
    – Andy aka
    Jul 24, 2020 at 13:48
  • \$\begingroup\$ apologies I must have forgot to accept the answer \$\endgroup\$
    – NeuroEng
    Jan 22, 2021 at 12:13

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.