Flybuck converter funnies

Nearly every wallwart uses an optoisolator + TL431 for isolated feedback.

Might help your situation?

Cheers, James Arthur

The Microchip MCP16312 seems to work okay, hurrah. Amazingly small--an MSOP-8 with no power pad. (It has thermal shutdown, which helps.) It has some loop stability problems, but then I'm testing it in a proto with longish connections to some nodes. Changing the bypassing improves things, so on a proper board layout it ought to work fine.

The idea is for one of the +5's to power a RPi, the analogue stuff, and the logic side of an isolated RS485 transceiver. The other one powers the line side of the RS485 (160 mA if double-terminated).

I'm using some of those Bourns coupled inductors we were talking about a couple of months ago, two 100 uH with their primaries in parallel, connected with 20 uF to ground from the cold end (i.e. the normal buck output).

Each secondary gets its own Schottky rectifier and filter, making two isolated supplies--isolated from each other as well as from the input.

If the coupling coefficient were above 0.99, the cross-regulation would be good enough by itself, but with those parts it's more like 0.95. Thus the isolated voltage varies a fair amount--from about 5.5 to 8V or so at the lightest loads. The measured output impedance is about 5 ohms, which isn't terrible.

However, that's not a bad match for an LDO apiece--a 20-cent AOmega part. (This is not an ultrasensitive application, so SMPS-quiet is okay.)

Cheers

Phil Hobbs

Thanks. If I have to use two separate controllers, I'll probably do that.

Cheers

Phil Hobbs

It's close but LM536x5?

If you can use an async buck, the LMR140x0 works.