Yesterday’s idea for a Cuk converter synchronous rectifier drive (right) seems to work, according to my use of LTspice.
But it needs a rather low value (100Ω) R1 resistor to pull against R2, the 330Ω resistor that turns on (via Q3) the p-channel mosfet represented by the capacitor.
As the gate of a mosfet is essentially a capacitor, high current is only needed briefly to change the gate voltage, after which a higher impedance can maintain the gate voltage (Aside: unless there is are transients on the drain, which can get onto the gate through internal device capacitance).
The improved circuit on the left, replaces 330Ω R2 with a 10k holding resistor, and adds another npn (Q5) to briefly connect an even lower value turn-on resistor (220Ω R5) just for the brief moment when high current gate drive is needed.
Now R1 can be increased to save power as it is no longer fighting 330Ω full time, and the circuit still switches quickly (LTspice plot below).
The values of C2 and R4 are guessed and not at all optimised. Also, a series diode-resistor pair across R4 would allow this extra circuit to be controlled more finely.
In a real circuit, the microcontroller (pulse source V1) would need enough drive to handle both R3 and C2-R4 – an emitter-follower drive to C2-R4 might be needed as a buffer.
Also, 10k-ish hold-off resistors across the base-emitters of Q1 and Q5 would be needed to keep the circuit ‘off’ until the pulse source was up and running.
Ok, so now there are five transistors, but a week ago there didn’t appear to be a solution to this problem anywhere in all of Internet-land.
Hats off to ADI for keeping access to LTspice easy and free.