Powering multiple isolated MOSFET drivers

[TrickyNekro]

Hello people,

I need some opinions on this matter. I need to power 8 isolated MOSFET drivers. They are part of 4 half bridges, so we are talking about 4 low side MOSFETs and 4 at high side.
Bootstrapping is better to be avoided, achieving 100% duty cycle is a design goal.
The load ranges from 10mA in quiescence to dynamic 150mA (let´s say 200mA to have a nice safety net).
Voltage input for the IC that will do this "job" will have to range from 3V (lower is better) to 30V (higher is better). If unavoidable a regulated source can be provided.
The solution has to be as small and efficient as possible. Feedback is a question which will probably remain rhetoric cause of the extra space it can take.

So far I have thought of two possible solutions both include flybacks or a converter of similar type:
One is that I power all low side MOSFETs together providing feedback and wiring the transformers for the high side drivers in series with the main transformer.
The other is that I wire 8 transformers in series but then feedback is kind of impossible.
In any case, the voltage to each individual mosfet driver will be regulated for achieving dynamic current consistency.
How would you approach this problem?

Thank you in advance for your opinions and time.
Best Regards, Lefteris
Greece

[Kleinstein]

I would consider using a forward converter, starting from a regulated input voltage. This way no extra regulation is needed and coils for separate outputs can be added at will. So only one transformer with several secondary windings - likely 1 for the low sides and 4 for the high sides. Output voltage will not be perfectly stable, but maybe +-10%, which should be good enough.

If a flyback is used one could use parallel secondary coils. If needed feedback could come from one output (e.g. the low sides) - the other should be still reasonable stable unless the load is very low - so zener clamps for those outputs could be an idea. 

[peter.mitchell]

i think the simplest would possibly be a self oscillating halfbridge driver, from a somewhat regulated input, zener clamps on the secondaries, and size the series capacitor such that the power transferred is limited.

[TrickyNekro]

The forward converter is indeed very appealing but the higher part count does mean more space of the PCB,
a combination though of the forward and flyback converter though a center-tapped transformer, well that´s interesting.
And rather clever, having the flyback resetting the core but also without the fuss of the extra LC filter.
I just found an article about doing this and indeed is a very good idea to be tested.
http://electronicdesign.com/power/low-cost-high-performance-dc-dc-converter-topology

I am uploading a hand drawn visualization of what I have in mind of achieving, just to make my case clear and the main
reason I need it to be like that is extra lazy but also quite neat layout. I will not be using a main transformer core,
rather than that I will be wiring transformers in series.

I will do some simulation on this for a while, if it works it can be very interesting.
Thanks for your reply and the mental stimulation 

Cheers, Lefteris

[TrickyNekro]

i think the simplest would possibly be a self oscillating halfbridge driver, from a somewhat regulated input, zener clamps on the secondaries, and size the series capacitor such that the power transferred is limited.

Series capacitors are also a good idea, never quite thought of it. We are actually on a good way to get somewhere.


I am also thinking of a bit of "positive feedback" in a way, something like ramping up the PWM based on the activity of the drivers since I have access to their pre-isolated input.

[Marco]

Why would you put the primaries in series? You get no cross regulation that way, or rather you are making cross regulated current sources rather than voltage sources.

I'd buck-boost to some intermediate voltage and then use half bridge. Every transformer gets it's own DC blocking capacitors, but the other end of the primary you connect to a single half bridge, continually oscillating at 50/50 duty cycle (with a little dead time). Good regulation and no feedback necessary.