ZVS driver charging capacitor bank

Question

I built a ZVS driver to make some big, high-voltage sparks a while ago. It worked off three voltages, 24 V, 50 V, and 80 V, and had no issues. The only part of it that got really hot were the 1 kΩ, 3 W resistors.

Recently I decided to play around with coil guns so I used it to charge up a capacitor bank as shown in the picture. The rectifier diodes are just regular 1N4007s.

The problem I'm having is this: when I charge up the capacitors it doesn't make a high pitched whine like it normally does, it (the transformer, which is a ferrite core from an old TV) makes this really awful chattering sound that starts off really rough-sounding and eventually turns into the whine as the capacitor fully charges. It seems to be a current issue, where it's drawing more current at the initial stages of charging, and less as the voltage in the capacitor rises.

However, it doesn't do this when I simply short the secondary, which should draw maximum current.

After a few charges, for the first time ever, the MOSFETs were actually really hot and the chattering sound became random and the MOSFETs failed (due to overheating).

So my questions are:

Does anyone know what went wrong?
Why is the transformer chattering and how do I make it stop?

I think I should have used UF4007s for the rectifier and I'm wondering if 1 kΩ at 24 V input is too high and I should go with the typical 470 Ω. Maybe the high resistance is allowing the MOSFETs to stay on too long which might cause poor oscillation and cause the chattering.

Answer 1

The oscillator is essentially shorted out at low output voltage. It is therefore unable to start up, the transistors sit with essentially full supply voltage on them, and cook off in a few milliseconds. I'm impressed the transistors didn't simply self-destruct the first time (but then, they did in later configurations).

Some current limiting, particularly at the output, is called for. Specifically, series inductance, so that the oscillator always has an inductive load to work with: either the transformer magnetizing inductance (open-circuit load), the series inductance plus transformer leakage (short-circuit), or something inbetween, with a generous resistive component (which needs to be light enough not to quench the oscillation; this is done by keeping the ratio between inductances fairly modest, i.e., using series inductance not many times smaller than magnetizing).

Other answers suggesting thinner wire = higher resistance, likely have a similar effect, the downside being extra heating in the transformer instead. The inductor however has much lower losses, so can sustain a short circuit indefinitely, i.e., there's no limitation on what output capacitance can be charged.

The downside of this approach, for general power conversion applications, is the inductor increases output impedance (Thevenin equivalent resistance of the DC output). So it takes a long time to charge to peak voltage; or, say you wanted a stable output voltage independent of load current, you'll need some kind of controller instead (which is roughly where LLC type resonant converters fit in).

Answer 2

I suggest that you improve the output diodes, because they are far too slow. Now remember that the capacitor bank will look like a short circuit when starting from a fully discharged state. You could use reactance to limit the current. This would change the oscillation frequency, but sensible calculations will keep things in the ball park. If you use an inductor that has the same inductance as your transformers' secondary magnetising inductance, then you will get about 1.4:1 frequency ratio from fully discharged to fully charged, which your circuit will cope with. Your MOSFETs will now be more reliable, allowing you to fine tune the system. It is feasible to make your transformer very leaky, allowing lower cost. I use a current-limited buck converter to precede the Royer and thereby protect the FETs, but this costs more, and is only suited to high reliability applications.

Answer 3

I tested some of my theories, ( and blew some mosfets up...) but in all of my circuits the problem was that the secondary windings were made out of too thick wire, so the resistance of the secondary was around 1.1ohms, I tried to connect (highest wattage I found was 470ohms and 3w (was too small, some bluish smoke came out)) and it started the oscilation. then I re-wound the transformer, with much thinner wire, the resistance now was like 34 ohms. now even with the capasitors attached, it starts the oscilation. the transformers had (the first with too thick wire, 5+5 on primary and about 260 on secondary, the second transformer has 6+6 on primary and around 250 on secondary. Hope this helps

4LEKS1