> I do have a working circuit in my hand. The worst case efficiency at
> full load (360mW = 9V x 40mA) is about 45% across the temperature range 
> and
> voltage range. Not really fantastic but it is okay. I do have some 
> unsolved
> issues. One of the strange problem is the start-up problem depending on
> how I place the inductor. It can start and stay in linear mode
> if I place the inductor in the wrong direction. I suspect that the 
> inductor
> current generates magnetic field which disturbs the feedback loop (zener
> diode).

As I noted, I never had startup problems at all. Sounds like you may have 
the wrong subspecies of Lorentz butterfly :-). I'd investigate closely (if 
possible) what the marginal condition is caused by. Reversing the polarity 
of a two terminal inductor is very cute! (and extremely marginal ;-) ).

What have you got on the output side of the inductor?
How stiffly is it loaded?
The circuit depends on the point where the zener connects to the output 
being able to have some ripple (small but finite). A smaller rather than 
larger capacitor here will assist this.  Also, if all else failed you could 
add a small series resistance in the output after the zener diode tapping 
off point. I didn't need any of that, although the current sense resistor 
obviously helps to isolate the zener feedback point from downstream 
filtering. (AFAIR it always worked OK prior to adding the current limit.)

If you want smaller than this you are almost certainly going to have to 
integrate things in an IC as there's not a lot there. The two low side NPN 
transistors could be in a dual pack provided that it has enough voltage 
rating. An opto driven FET isolator may have enough power capacity to act as 
the output switch and this would reduce drive dissipation and probably 
slightly reduce component count.

The reference zener MAY be able to be removed with the two divider resistors 
providing one Vbe at the transistor when the output is correct. This may 
have a worse variability with components than with a zener, but maybe not. 
depends on how tightly specced you need the output voltage.

So far:
Dual transistor (high voltage)
Opto coupled FET (can be N Channel !!!!)
Inductor.
Flyback diode
A few resistors.
Filter caps in and out.
Maybe an RC for regenerative switching assistance.

Sounds good :-)


        Russell McMahon


_______________________________________________
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist