Alexandre Guimar=E3es wrote:
>=20

> > * R2 is not needed
>=20
>     R2 helps on the turning off time of Q1 !! And it helps a lot actual=
ly. I
> will have to check the value on the working circuit to make sure it is
> right. It makes Q1 BE capacitance have a better path to discharge when =
Q2
> cuts.

The Q1 capacitance is probably picofarads.
At 20kHz or so this thing will run at that doesn't
matter. If you reduce R3 to 2.7k as I said you
will no longer need R2. But I could be wrong. :o)



> > * put a 47uF 25v cap across D3, will help
> > save D3 if you get spikes on 100v supply
>=20
>     Done already :-) Makes sense. Thanks for the tip. What about a  100=
 nf
> there also ?

Sure if you like.

> > * reduce R3 to about 2.7k for better turn off
>=20
>     I will try that one once I build it on the protoboard. I guess I wi=
ll
> have to fiddle with all the values to see how far I can get on the inpu=
t
> range and how efficient it can get on the usual range.

I found the power wasted to run the zener and
transistor biasing was very small, maybe 2% of
total. There was little to be gained from reducing
power in these areas, it was definitely best
making sure the transistors turned on/off
harder, which improved efficiency to the max.


> > * D2 and C4 are not needed.
>=20
>     It will make the ripple lower and let the digital circuits do a mor=
e
> "normal" turning off when the supply input is disconnected. When you
> disconnect the supply no current will flow from the digital power to th=
e
> power supply. The voltage drop gets to be more stable and predictable. =
At
> least that is what I have seen on some measurements I have done on some
> power supplies. They can do pretty nasty things when getting disconecte=
d.

Ok, I understand why you wanted D2 now. :o)
But I would suggest putting D2 in series with
R6, and reducing R6 to maybe 5.6 ohms. Then you
will get less drop there and probably better
regulation.

With the drop from R6 and D2 you will need to swap
the 5.6v zener for a 6.2v or 6.8v zener.

>     I hope ;-) This project might be a nice one and I do not want to ma=
ke
> any silly mistakes. It has to be very reliable.

For super reliability I would be tempted to add
a 6.8ohm resistor between Q1 collector and the
inductor (and D4 cathode) at the top.
-Roman

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu