VCC for Isolated SMPS Topologies

[__pmos]

I am currently working through the details of a DC/DC push-pull converter, and am curious about different methods for obtaining VCC for the associated PWM controller and FET drivers. I assume that in a lot of circumstances where the controller and drivers can't be directly powered by your input, they would be powered by a "start-up" resistor at first, and then your aux winding takes over once switching begins. I've seen that method in a few of the reference designs for isolated power supplies that come up readily through a google search. Using a FET for the high power startup that can be cut off by the aux winding also looks like an established thing.

But what if I want to have a FET driver on my secondary side for synchronous rectification? What if I want to use a really nifty part like the ADuM3190 and need a Vcc/dd on that side too? Am I essentially just tying into the secondary winding and relying on the first pump of the transformer being all that's needed to get me going?

I'm still pretty new to the serious power supply design game, and I'm looking to put a few more wrinkles in my brain. Any reference designs, application notes, datasheets with examples, etc that show this would be greatly appreciated, especially if they are push-pull!

[mag_therm]

Years ago I did an isolated DC_DC that had control board entirely powered from its inverter secondary.
To cold start, there was a short time rated dropping resistor
( Millisec) and electromechanical break before make 2 pole relay.
There was a specified input to output withstand of 5kV or so, needing a special relay and power transformer interwinding insulation  layers.

That configuration required gate transformers but allowed simpler dc current and voltage feedback, also the control board load helped with regulation to no load step, avoiding bleed resistors.

[David Hess]

But what if I want to have a FET driver on my secondary side for synchronous rectification? What if I want to use a really nifty part like the ADuM3190 and need a Vcc/dd on that side too? Am I essentially just tying into the secondary winding and relying on the first pump of the transformer being all that's needed to get me going?

Wouldn't the diodes in parallel with the synchronous switches provide lower efficiency rectification until the synchronous control circuit is powered?

[__pmos]

Years ago I did an isolated DC_DC...

I haven't really worked with electromechanical devices, so hearing about these types of different techniques is really cool! I suppose I'm lucky in that I don't need to plan for a very high withstand. That's also an interesting detail regarding your no-load condition. Thanks for sharing!

[__pmos]

Wouldn't the diodes in parallel with the synchronous switches provide lower efficiency rectification until the synchronous control circuit is powered?

I think you're right! I suppose synchronous rectification really is just an efficiency boost, and I got caught up in thinking the whole topology needed to be switching. That would also answer my question about that error amp as well, but I'll have to dig more into the datasheet to see what the consequences of compensation components being on the secondary side are.

[Faringdon]

You can have a little TNYswitch Flyback to give you some secondary bias power if you want. They are nice and cheap.
Basso has good stuff on having error amp on sec side.

https://www.researchgate.net/publication/280308828_Designing_with_the_TL431_-_the_first_complete_analysis/link/55b11ab408ae9289a084b79b/download

[David Hess]

Wouldn't the diodes in parallel with the synchronous switches provide lower efficiency rectification until the synchronous control circuit is powered?

I think you're right! I suppose synchronous rectification really is just an efficiency boost, and I got caught up in thinking the whole topology needed to be switching. That would also answer my question about that error amp as well, but I'll have to dig more into the datasheet to see what the consequences of compensation components being on the secondary side are.

Synchronous rectifiers almost always have a parallel diode, whether the body diode of a MOSFET or an external diode, to catch the time while the synchronous rectifier is switching.

[T3sl4co1l]

Switching on both sides of isolation can be synchronized by transformer coupling, for instance (2-switch or bridge forward, most often?). It can also be done passively (no signal isolation required); see e.g. NCP4306.  This one can still use signal coupling, see the application section; it wouldn't be used in DCM or QR flyback, but might be handy in CCM forward/flyback.

Most sync rect controllers have wide supply ranges, so they can be self powered -- not that it's something you necessarily want to make use of because of the extra power dissipation at high voltages, but when you aren't driving big transistors, it can definitely get you there.

There's also the even more passive option, a gate drive winding on the secondary, driving a MOSFET directly.  This is more of a forward converter thing(?), and easiest for modest output voltages (3-12V?).  Note that the wide voltage swing between Vgs(on)/(off) acts like reverse recovery of a PN diode, but much worse (it's not time dependent, but voltage dependent!), so it only works in certain cases (here, the output inductor allows voltage to fall to zero).

Tim