1

**Transformer (ferrite core) / Re: Royer oscillator (ZVS) driven Jacob's ladder, E80 core transformer**

« **on:**December 12, 2019, 04:51:08 AM »

Yes. I've built 5 or 6 ZVS circuits and ran lots of simulations. When power is applied, current in the power-feed inductor(s) from the supply ramps up fairly quickly. Initially there's no oscillation, so both FET drains (or IGBT collectors) are at 1-gate-voltage above ground. Eventually the positive feedback gets oscillation going. By that time the power-feed inductor current is much higher than in normal operation, perhaps even saturating and/or shutting down the DC input supply. If the supply doesn't shut down, the high current causes a separate issue. All that stored energy causes the oscillation voltage to spike way above normal. (The oscillation amplitude rings for several cycles.) To handle this voltage, either the FETs need to be rated for much higher voltage than needed during normal use, or TVS diodes are needed to clamp the voltage. I've done both.

Before this design I played with lots of ideas to kick-start the oscillation, from just unbalanced part values to extra circuitry to hold off one gate signal until power is up, or hold of both gates with staggered release. At best, these got the oscillation voltage overshoot (and ring) down to 2x rather than 3-5x. At this point, the only solution I've found is what's in this design: Let the oscillation start slowly at low current, feed by bleed resistor, R2 in this circuit. Once oscillation is going, applying full voltage isn't so problematic. It takes much less time for the oscillation to ramp from the low amplitude than it does to start with no oscillation at all. I still have TVS diodes, as I'm pushing the voltage, 600V rated IGBTs for a normal-operating peak of 533V (PI * sqrt(2) * 120V, for 120VAC line power input).

I'm about to build a small low-voltage ZVS with current-limited buck-regulator input, using the ZVS input inductor for double-duty as the buck-regulator inductor. Of course, the buck-regulator itself isn't zero-voltage or zero-current switching, so somewhat defeats the point of ZVS.

Before this design I played with lots of ideas to kick-start the oscillation, from just unbalanced part values to extra circuitry to hold off one gate signal until power is up, or hold of both gates with staggered release. At best, these got the oscillation voltage overshoot (and ring) down to 2x rather than 3-5x. At this point, the only solution I've found is what's in this design: Let the oscillation start slowly at low current, feed by bleed resistor, R2 in this circuit. Once oscillation is going, applying full voltage isn't so problematic. It takes much less time for the oscillation to ramp from the low amplitude than it does to start with no oscillation at all. I still have TVS diodes, as I'm pushing the voltage, 600V rated IGBTs for a normal-operating peak of 533V (PI * sqrt(2) * 120V, for 120VAC line power input).

I'm about to build a small low-voltage ZVS with current-limited buck-regulator input, using the ZVS input inductor for double-duty as the buck-regulator inductor. Of course, the buck-regulator itself isn't zero-voltage or zero-current switching, so somewhat defeats the point of ZVS.