3.6V to +-50V ~50mA isolated DC converter(s)

[Stephen37]

Hello and happy new year,
I am looking forward for my first project in 2019 and I will require a rather special DC converter for an OPA454 opamp:

1. The converter will be powered by 2xPanasonic NCR18650B 3.7V batteries (I could go for a 4.2V battery if that would be better),
2. The converter will need to output 3-4W and have isolated outputs. (the load will be variable from 0.1W)
3. The current should be configurable according to the impedance of the opamp output. (for energy efficiency in a constant current source, so the opamp won't need to dissipate the extra voltage as heat)
- This means that I need to switch the duty cycle of the switch, a simple 50% duty cycle reduction might be enough if the opamp is supplying less than X mA.
4. Over 100kHz switching (the OPA454 will be used in the audio range)
5. Low ripple for audio and no ground loop noise with linear regulation across a 0.2mA-50mA load in the +-20->+-50V range.
6. 80% efficiency would be excellent, but I doubt it's possible.

Here is what I have put together so far in my mind:
For the first stage I am thinking about implementing a simple boost converter for 3.6V to 25V before the battery charging controller and before the protection. Maybe with the cheapest switch, inductor and diode from Shenzhen.
For the second stage after the 25V add another stage with coupled inductors in an isolated flyback configuration with a PWM controller.  This raises the following issues :
1. Choosing a suitable PWM controller (switch) for generating/regulating +-50V in a flyback configuration from ~25V DC. An IC with an integrated mosfet would be better.
2. Choosing a suitable schottky or ultrafast diodes for the best efficiency.
3. Choosing a coupled inductor for the isolated boost topology and for the negative rail. I would even try to get 2x coupled inductors with +-50V output from a single controller for two isolated sources.
4. Choosing a suitable mosfet. (should be a small smd package)
5. Choosing suitable capacitors.
6. I should be able to put it all together without access to an oscilloscope. In the worst case I could use an old tv for debugging. I will build a programmable test load with a microcontroller and a digital pot

Block diagram from my mind:


Does anyone have an idea for a suitable PWM controller for boosting a flyback to +-50V from 20-30V, what type of coupled inductors would you recommend, which diodes ? Thank you.
I am thinking about LM5020,LM5022. LM5022 has a reference PMP7931 converter, but I need to scale it up to my requirements.

[nctnico]

I'd go from the low voltage to +/-50V directly. I'd look at readily made transformers. Wurth and Coilcraft have several of these. A PoE transformer might have the right winding ratio but it will need to have 3 windings: one for the input, one for the output and one for the voltage feedback (it won't be accurate though). If you use a forward converter approach you could do with one transformer and two inductors per output voltage.

I'd also invest in a scope. A $50 DSO nano or old analog scope will be infinitely better than nothing. I doubt you'll get a circuit like this working without an oscilloscope.