smps problem

Oooops...... programmable power supply with auxilliary taken from the main transformer. Bonk! :-)

Depending on how you define fast and cheap then you might have to split the auxilliary winding on your tranformer in two and connect the taps as required.....

DNA

How much does the input voltage vary? During the "on time" the transformer has the input voltage for both modes. If you invert the connections of the AUX winding, the AUX supply will be unregulated but not change when you change output voltages.

or:

You could make a low powered zener regulator and use a highish voltage MOSFET as a source follower.

or:

I throw this circuit out there for folks to have fun with:

ASCII art

C1 -------!!------- ! D2 ! D3 ! --->!--+-->!--+--Vb ! D1 ! ! Xformer ---+-->!----+ --- ! --- C2 R1 !!- ! Vb -/\\/\\-+---+-----!! Q1 GND ! ! !!- D4/-/ c\\! ! ^ !-/\\/\\-+---+----------- Vout To load ! e/! R2 ! Vout ! Q2 --- /-/D5 --- C3 ^ ! ! GND GND

D1 and Q1 have to be modestly large.

C1, D2, D3 makes C2 have a highish voltage on it that is always there.

Q1's drain gets its power from the unfiltered diode D1 and thus can only charge C3 during the flyback time. If Vout tries to go more than a diode drop above D5's rating, Q2 yanks the gate of Q1 down to prevent this.

During the primary side's on time, the load discharges C3 slighty turning Q2 off and allowing the gate on Q1 to rise until, perhaps, D4 limits it.

At the instant of primary turn off, Q1 is biased on. The transformer flies back until D1 starts feeding current to C3. The transformer's swing pauses at that point until Q2 shuts Q1 off.