9 volt overly complex(I think) power supply

[Mooly]

This is the sort of thing that can be tweaked indefinitely... and it certainly looks like it would work OK.

A couple of things jump out,

1. The only source of base current for Q1 is R1. Perhaps a darlington for Q1 and run the error amp at lower current. Do you really need two LEDs in series... it will work OK but looks "messy"... use a zener oround 4.7 volts.

Same goes for Q3 as long as the input voltage is sufficient to overcome the added volt drop of a darlington.

2. Fuses are good... you could always incorporate simple current limiting (another transistor and low value resistor). However good fuses may be they are horribly non linear and in small values have significant resistance, so if you move R3 to the other end of the fuse it's (the fuse) is now within the NFB loop and it's non linearities disappear.

One other thing, C3 and C6. They couple the output directly via the B-E junction of Q2 to ground... and that's not good. You could easily see fast transients that exceed the rating of Q2. Test the regulator under dynamic conditions, eg a pulsed load drawing max current for say 1ms and see what happens. Repeat for 10ms, 100ms. C3 looks massive, you are after filtering HF noise, not adding a time constant

[grufti]

Everybody is going to have a different opinion:

1. darlington yes, but keep the LED's.

... Q3 might be disposable

2. add simple current limiting ... heed the R3 advice