Hello Hans,
first of all, I am not the inventor of this solution, but I think it is a good one.
I'll try to answer your questions :
1) No, I have not tried it and I am reluctant to put my QMX to the test. I would humbly suggest that you do it, if you agree to my arguments below
2) if you read the text it is proposed to put a resistor in series with the GPIO coming from the processor. It is not in the diagram though.
3) I do not agree with your analysis.
As soon as the zener starts to conduct, the PWM signal is clamped to ground. So no more energy is fed to the 330uH inductor.
The current through the zener will be quite low, so wether the zener nor the transistor will suffer.
There will be no current spike.
As soon as Vcc goes below the zener+base/emitter voltage the clamp on the PWM will be removed and the processor will continue to adapt to the new input voltage, not even knowing the the over-voltage protection was engaged
br SM5EIE/Gunnar
