I mentioned before that I need to supply 5v@<10mA to a current shunt and ADC which may be located on the positive side of a high voltage battery (maybe 250v). The ADC output will then pass through an optoisolator.
I have +12v versus ground readily available.
The 5v for the ADC cannot be generated by a linear reg. Both the voltage gap, inefficiency, and heat generation are too high.
I came up with the idea of putting a lithium battery on it, but it will always be drawing a few mA and will not last a very long time.
I evaluated switching reg configurations. They're looking a bit too complicated, especially since I was thinking of making it support both high and low side shunts so I would need 2 types of switching regs here.
I had some ideas of using a capacitive charge pump to charge a cap off of 12v, disconnect it, and reconnect it to the ADC/shunt board. The isolation is a bit tough here and I don't really like the idea.
I think the best idea is the 12v board- which reads the isolated signals off the ADC- could power a small isolation transformer. I think that's the best idea. Now I have only a small amount of board space and watch my component costs, so I'm not interested in using a 12v-to-120vac inverter and running a 120vac-to-5vdc transformer. And we need only a few mA. I think the best thing is to give it around a a 1:1 ratio, rectify it on the other side (1.4v FWB loss), and run it though a reg (0.5v drop if it's an LDO reg).
Perhaps I can find a small toroid here. Should I find a driver chip or just one of those two bipolar transistor oscillators? If it's a toroid, won't its low inductance require it to oscillate at fairly high freq if we can't replace some of the pulse width with dead time? Otherwise I think the current may get pretty high. I guess I don't have anything against higher freq but I'll need to use fast diodes for the FWB.