MOSFET vs NPN for 1.7MHz 24V 1A Oscillating Membrane Driver from Microcontroller

Hello,

As an hobby I am developing an embedded application which should drive @ 1.7MHz an oscillating membrane which creates vapor from a liquid (the same tecnology used on the so-called ultrasonic humidifier, even if we are talking about much higher frequencies than ultrasonic band).

Normally such a membranes come with integrated power driver, however for this application I need to drive it directly from the microcontroller.

I am not at all an expert of analog electronic design, therefore I would like to ask you if a common MOSFET is capable of driving such a load, or if a NPN is a better choice.

Are there special things to consider? I would just connect the transistor as an ON/OFF switch...

 

Thank you for your help!

Andrea

 

 

Assuming you are driving a piezo transducer at its resonate frequency, you should be fine with a mosfet at 1.7Mhz. You will have to use a gate driver as the microcontroller will not be able to deliver enough pin current to charge/discharge the gate at that frequency (at least not for a reasonably sized mosfet). You will probably need between 24v and 36v to drive the piezo so using an n-fet as a low-side driver will be the simplest way to go.

The critical parameters for the mosfet will be switch-on and switch-off times and gate charge. You need a mosfet driver that is rated to switch at 1.7Mhz with the gate charge of the selected mosfet. If you are switching a couple of amps or less (it should be much less than this), pick a mosfet rated for 2-3x your peak current with turn-on and turn-off times of a couple of hundred nano-seconds or less and minimum gate charge (or minimize gate capacitance which tracks gate charge closely enough for this application). Then, get a 2A or bigger gate driver and keep the inductance between the gate driver output and the gate of the mosfet to a minimum. A small series resistor between the gate driver and the gate will reduce EMI (by slowing the switching of the mosfet slightly). use low ESR, low ESL filter caps near the mosfet and ideally use a series of caps (0.1uF, 1uF, 100uF would be a good starting choice) and minimize the size of the loop between the filter caps, mosfet source and drain and the piezo.

Must of this advice is EMI and power optimization related, you will probably also get away with ignoring everything except mosfet choice and use of a gate driver.

Be aware that if you are using a piezo, you want to drive it right at the resonate frequency and that frequency will change with temperature and other variables. There are quite a few schematics available online that are self-resonate with the piezo so the frequency will track any changes in the resonate frequency of the transducer. These might be a better idea than driving a mosfet with a microcontroller.

Thank you for the hints!
Of course your initial assumptions is right, I will try to go with the micro (this solution would be much cleaner since I already need the micro for other purposes), then if I find issues with resonance frequency changing during operation I will look for the dedicated circuit.

If you put a current sense resistor between the mosfet and ground, then use an RC filter to eliminate the 1.6Mhz component you can sense the current being sinked by the piezo transducer (google low side current sensing). You can use this information to move the PWM frequency around and find the resonate frequency in software.