I'm trying to switch a high-current, 12V load (large parallel LED strip, with the anode connected to LOAD-) using a PWM chip. My approach right now is to use a Power MOSFET to switch the load; and due to the high capacitance of the gate and higher gate threshold, I'm using a second MOSFET to drive the first MOSFET.

This is the schematic:

MOSFET Circuit Schematic

5V is the output of a linear voltage regulator rated for 800 mA, which I don't think is an issue in this circuit.

The PWM Driver (a WS2801) is pull-down, so both FETs have a pull-up resistor to return the gate voltage to 5V.

I'm seeing that Q1 behaves as expected: when there's no signal applied, Q1 is pulled up high by R8, and conducts. Q1 conducting pulls Q2 low, so Q2 stays "off" in this state.

When a PWM signal is applied, Q1 turns on and off per the signal. However, the Q2 never fully conducts. Upon further inspection, the gate of Q2 never goes above 700 mV or so. It definitely pulses between 0 V(ground) and ~700 mV in a way that matches the PWM signal, so Q1 is definitely conducting, but for some reason the gate of Q2 is never pulled up high to 5V.

I feel like I'm missing something fairly obvious here, but can't figure it out -- why isn't Q2 switching as I expect it to?

  • 1
    \$\begingroup\$ Are you sure that Q2 isn't reversed (source and drain swapped)? The body diode of a reversed Q2 would conduct continously, dropping the 0.7V you are seeing. Also, 10k is way too much resistance for pulling the gate high - Q2 will take forever to switch on fully, heating up in the process, which is really bad considering you want it to switch at a non-zero frequency (PWM). A gate drive of merely 5V doesn't help either, as it slows the switching even further, why not use the 12V you already vave available. Q1 also seems redundant to me. \$\endgroup\$ – jms Oct 31 '19 at 1:05
  • \$\begingroup\$ I don't think it is -- I definitely see the voltage on Q2's gate pulse at the same timing as the incoming PWM signal, but the voltage never goes higher than 700 mV or so. I haven't seen any issues with timing from 5V or 10k resistor; the edges of the pulses are quite clean and square. I tried changing the rail voltage to 12V and saw the same behavior as well. \$\endgroup\$ – nathan lachenmyer Oct 31 '19 at 1:07
  • \$\begingroup\$ What resistance would you recommend for a pullup resistor in this case? At some point in the past it was recommended to me to use 10k as a standard value for pull-ups, and I never bothered to recalculate a more optimal value. \$\endgroup\$ – nathan lachenmyer Oct 31 '19 at 1:09
  • 1
    \$\begingroup\$ 10k is indeed good as a pull-up for small signal lines e.g. reset pins - but definitely not gates of power MOSFETs. You want as low as possible, e.g. a 560 ohm resistor from a 12V supply. Even lower values (like 100 ohms) are better, but then you'd need to use a power resistor. Ideally a gate should be driven with a push pull driver - a transistor pulling the gate low, another transistor pulling the gate high, likely both integrated into a purpose made gate driver IC. That said, if the frequency or current is low enough, you can probably get away with using a pull-up resistor. \$\endgroup\$ – jms Oct 31 '19 at 1:23
  • \$\begingroup\$ Ah, I see -- that makes a lot of sense. The PWM signal is pretty slow -- I'm driving LEDs, so the frequency is around 2.5 kHz. I've seen other designs that were similar to this, so assumed it wasn't problematic, but your points are well taken. \$\endgroup\$ – nathan lachenmyer Oct 31 '19 at 1:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.