Medium voltage (~800v) SMPS

Well, you should now be aware of a few probelms with the transformer:

1) lots of leakage inductance, as you cannot interwind so many output turns to so few input turns. 2) Lots of capacitance to drive, as the inter-winding capacitance of the secondary is going to severely reflect back to the primary. After all, a 5KV secondary is going to have a *lot* of turns. Be careful with insulation, you will need some between the secondary and the coer itself. Have you picked a core and determined volts per turn for a "final" prototype?

I still haven't come up with a way to find the ultimate minimum turns per volt for the primary.

I've been rethinking the secondary output voltage. 10kV capacitors are expensive but 6.3kV capacitors are more affordable. I may want to build a 3kV secondary with more cascade sections than use larger more expensive capacitors and fewer stages. More stages with a 3kV input will result in less output ripple too.

I still have not made the SPICE simulation work. I need to choose a capacitor value and fill in all of the characteristic information in.

If I use a form and a split core then will the form be the insulation I would need?

Grant

"logjam" skrev i en meddelelse news: snipped-for-privacy@d55g2000hsg.googlegroups.com...

Usually, what one does to measure HV is to feed a *current* into a buffer amplifier.

You cannot get a transformer that small to work: *if* it does not arc over, you will loose all the power in inter-winding capacitances (before it arcs over).

However, there is a much easier way: Use an automotive ignition transformer in either a flyback topology (easiest) or a resonant converter (more efficient). Ingnition transformers are cheap and available (unlike attractive wimmen)

An typical ignition transformer should be good for about 100 W at 25 kHz and probably give you 25 kV!! Enough to require EHT diodes (BYW 95 ??) and oil & silicone potting for the rectifier stack. And deadly, really it is!!

Sorry; all else being equal, the number of stages is not related to amount of ripple. Try load current, stage capacitance, load capacitance for starters.