What's the goal of diode and resistor in parallel on SMPS?

The idea is to get the MOSFET to turn off more quickly than it turns on. When the MOSFET is driven "on" the gate charge is supplied through (say) R915 + R917 = 51.7 ohms.

When it turns off, the gate charge is sucked out through the diode in series with the 4.7 ohm resistor.

You can think of the gate as looking a bit like a large capacitor (gate-source capacitance plus a typically much larger component from the drain-gate capacitance, the latter has a bigger influence due to the Miller effect- the drain typically changes in potential by a much larger amount, multiplying the effect of the drain-gate capacitance.

In the case of the FMV111N60ES, the gate charge can be as much as 73nC.

This can be used to help prevent two MOSFETs from being "on" at the same time, causing shoot through (which wastes power and can damage the MOSFETs) or just to control the waveforms a bit better.


In addition to Spehro's excellent answer, there are a few other considerations.

RF emissions from circuits increase with fast-switching devices, but there's also the gate driver limits to account for. As the transistors drive inductive loads, faster switching will not actually increase performance for a given circuit. The circuit is tuned to operate at a certain frequency, so faster switching can lead to greater driver cost with no benefit.

The context changes dramatically when you replace the MOSFET with a GAN-HEMT transistor, as they can handle higher loads and switch at far higher speeds, 500kHz switching of KW range supplies are not unheard of. This is when ground bounce and RF emissions can become a serious design headache.