OK, couldn't sleep. Haven't built it but this works in simulation: Both diodes can be 1N4148, it only needs to catch brief spikes. Another diode from Pin 3 to 11V is a good idea; I see spikes in simulation that I don't expect on the bench. Pin 3 is connected to L, there's no transistor. Rb becomes 24k. L needs to be between 250 and 500 microhenries. (This is not a few turns on a piece of ferrite.)
555 needs to be the bipolar (NE555, LM555, etc.), not CMOS.
Regulation is nonexistent, it varies with input and load.