The first circuit shown is very similar to the idea which I have used a number of times over the years. Maybe not of the most efficient designs but I have usually used old computer power units for supply (generally 12V at several amps capability) so don't have to think about battery life. I have generally used a multivibrator, regulating the supply to 5V for the multivibrator. I use logic level mosfets for transformer switching, and supply, either the 12V to the transformer or, regulate the 12V downward for a more precise output voltage. It appears feasible for feedback to the regulator (12V side) for better regulaion but I have not had a need for it with my applications. I have mostly used more complex output filtering to reduce spikes etc. I use normally available mains transformers. I have done a little on measurement of efficiency but would have to lookup to check my figures. Just to say that the efficiency seems to be very much dependent on load and can vary quite a bit.
I have spent a little time looking into things and have planned to use an Arduino with carefully controlled switching times for comparison (to minimise cross over reduced efficiency and spike effects) but have not got round to it yet. Obviously this would directly replace the multivibrator, the output switching logic level fets being driven directly. Another thought thas had ocurred to me that needs a look is that a failure in the drive could cause problems. I have had an occasional blown transistor on tests but some of the supplies I have experimented with have been reasonable high power - In one case a twin parallel transformer design which drew an initial surge switching off the old Xbox psu I was using (rated at 14A I think). In the end, before I pretty much finished with >100V supplies, I changed to using fairly efficient voltage multipliers instead as the switch on surge from low voltage is a tricky problem whereas multipliers are more straighforward, in my experience. Obviously for most puposes, and especially for battery circuits. a simple invertor is a good move.
Tracy
I have spent a little time looking into things and have planned to use an Arduino with carefully controlled switching times for comparison (to minimise cross over reduced efficiency and spike effects) but have not got round to it yet. Obviously this would directly replace the multivibrator, the output switching logic level fets being driven directly. Another thought thas had ocurred to me that needs a look is that a failure in the drive could cause problems. I have had an occasional blown transistor on tests but some of the supplies I have experimented with have been reasonable high power - In one case a twin parallel transformer design which drew an initial surge switching off the old Xbox psu I was using (rated at 14A I think). In the end, before I pretty much finished with >100V supplies, I changed to using fairly efficient voltage multipliers instead as the switch on surge from low voltage is a tricky problem whereas multipliers are more straighforward, in my experience. Obviously for most puposes, and especially for battery circuits. a simple invertor is a good move.
Tracy
