designing buck inductor with auxiliary output

Here is a schematic. The parts are difference as the 5116 is a synchronous buck but this gives the general idea.

I've done variations this supply in several portable testers for GenRad...

...Jim Thompson

I think that works. But you may get a huge reverse voltage across the aux diode, not to mention that the secondary of the transformer is going to ring something awful. The diode choice will be interesting.

What's the input voltage like?

I did post a cute little single-inductor inverting switcher circuit here once, or you could do a Cuk inverter, either one off to the side. Using two circuits, each with stock magnetics, might be easier than one circuit with a custom transformer.

If the input were close to +15, you could capacitively "voltage double" off the switch node, with a few gotchas.

John

The input is loosly regulated 15V +/- 15% so the reverse voltage should only be 30V + spikes. I already have a snubber in the design should handle a lot of the ringing problems you're speaking of.

Can those capacitice charge pumps deliver 300mA? I thought they were only good for a few 10s of mA.

I really don't have room to invest the real estate for a formal inverter for the -15V. A slightly larger transformer and a diode is pretty hard to beat size wise.

Thanks

Sure, why not? 2 diodes, 2 caps off the switching node. The major gotcha is that some switching regs with cycle-by-cycle current limiting will see a big peak current spike and shut down; the fix is to add a little impedance between the switch node and the doubler network, a bit of R or L, not a bad idea anyhow. The DC output will be equal to the p-p swing at the switch node, minus the diode drops, minus any charge lost in the caps; so use big caps.

John

Definately something to look into.

thanks

"Jim Thompson" wrote in message >

Jim,

Could you give me a walk through on how this circuit actually works? Is your PWM contrlller looks like a hysteretic controller with very erratic duty cycle changes?

interesting circuit.

I did that about 20 years ago... it runs "free fall" ;-)

Today I would clock it.

Basically the flyback powers the auxiliary supply until regulation is met, then the SCR is turned on to function as a normal flyback diode.

...Jim Thompson

Haven't followed the thread but a charge pump sloshing around 300mA will most likely produce spikes >1A. Which tend to show up on the rails that supply it. If you can't stomach that maybe the classical approach is better.

Personally I use charge pumps only in the mA range.

And hope that the SCR's turn-off delay won't reach into the time slot when the FET comes back on ...

In my day _everything_ was slow... I'm talking 20KHz here ;-)

...Jim Thompson

which classical approach are you speaking of?

Transformer or inductor based conversion. It's really the only way to reliably reign in peak currents.