>>2952487
Thanks anon!
>any reason you aren’t using a gate drive transformer or photovoltaic gate driver?
I thought the idea could work in theory if the opto is fast enough. The rv1s9160 is very fast indeed. In my test circuit I put together on breadboard the opto (I presume?) caused lots of oscillations especially on the high-side N-MOS gate.

Yes, eventually some kind of switching power supply will be unavoidable. Probably a transformer supply with 3-4 outputs. But for now a lab supply and some linear regulators will do for testing purposes.

>stop 36V from backfeeding into the 24V source
Q102 is backwards exactly for this reason. In theory 0V at output should never happen. Replacing Q102 with a diode is a good idea but it might slow down the 36->24 transition too much when high-side N-MOS switches off. The low-side NMOS can conduct current in both directions when switched on, unlike a diode.