This is the schematic in its basic form. It's an nMOSFET that gets driven by a pin of an Atmega328p, this nMOSFET on his turn, drives a big pMOSFET. The source is connected to a voltage regulator (9V6), the drain is connected to the backlight of an RGB screen. The Atmega generates a PWM-signal (100kHz, duty-cycle is variable) so one can change the brightness of the screen.


However, when looking at the drain, I noticed that the OFF-part doens't go to ground and there's lots of ringing.


The ringing I could damp by adding a snubber network to the drain of the pMOSFET. But the OFF-part still doesn't go to ground and there is still a sudden peak in voltage that I could not remove. It seems something is pulling the voltage back high.



  • How comes that there is still a DC component in the OFF-part of the PWM?

  • How can I remove that part so it goes to ground?

The only other question I found, is this one: High side mosfet source voltage does not switch back to ground

So I added a parallel RC network at the output.


Measuring now, it goes down to ground, but there is still a peak present in the signal.


Can I clean this up further?

  • \$\begingroup\$ 1. Why high side driving? 2. Why 100 kHz with passive pull-up? \$\endgroup\$
    – winny
    Commented Oct 8, 2020 at 14:04
  • \$\begingroup\$ @winny 1. No reason in particular. You think it would be better to switch on the low side? 2. 100kHz cause otherwise you could hear the switching, and we thought it would be enough. In the first iteration, the switching was a lot slower and the passive pullup was ok but we had the noise. I'm testing now different values and layouts with air soldered through hole components. \$\endgroup\$
    – Swedgin
    Commented Oct 8, 2020 at 14:14
  • 1
    \$\begingroup\$ Then I recommend bog standard low side switching. Simpler and cheaper. \$\endgroup\$
    – winny
    Commented Oct 8, 2020 at 14:21
  • \$\begingroup\$ @winny bog? Thank you for your recommendation. \$\endgroup\$
    – Swedgin
    Commented Oct 8, 2020 at 14:24
  • 1
    \$\begingroup\$ Yes. If you assume LED- can be disconnected from GND in order to tap in your new driver, but it has some connection elsewhere not seen, your new backlight driver will not work. \$\endgroup\$
    – winny
    Commented Oct 9, 2020 at 7:12

1 Answer 1

  1. When the P-MOSFET is off, it runs no more current into VLED. It is possible that the input of the LED does not provide a pull-down path, and the drain voltage just keeps floating above ground. You "fixed" this with the pull-down array that discharges C62 and C63, but such does not seem necessary to operate VLED.

  2. The negative peak might be due to the inductor L3. When the current through L3 collapses, the inductor reacts with a voltage spike of opposite polarity. Can you try adding a flyback diode across L3 (me not knowing its purpose to begin with)?


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.