I think there are some modules with protection, maybe even control(?!) onboard? So it depends, if that's the case. But if the module is as the OP's circuit, ya nah, turning it off suddenly will most likely brick it, among a variety of other things.
at power off sequence circuit self power by resonant coil
This is true, after the inductor discharges, assuming it happens faster than the output takes to ring down. But therein lies the problem: as soon as power is removed, the inductor flies back, supply reverses (there's not even a clamp diode to catch this; the 470R resistors might just get popped in the process..?!), gate voltage goes to zero (well, slightly below -- clamp zeners doing their job), and avalanche ensues.
Which, maybe the transistors can take a few hits of that, maybe millions of hits, maybe zero. Who knows.
Using MOSFETs rated for avalanche current of at least as much as the maximum design input current, and using inductor(s) smaller than that used in the avalanche test, is necessary to survive at least one event. How many more you get after that... who knows.
The trouble is that it is too simple, and does not have proper gate drivers, and when the fets are partially open, they have both voltage over and current through them, and they overheat extremely quickly.
Therefore this circuit NEEDS to be switched on "instantly". So put an extra switch between the power supply output (and extra buffer elco's?) and the ZVS circuit itself and then turn on this switch last.
Alternatively, you can use the MOS fet's themselves as a switch. First keep the gates shorted to the sources, and then remove these shorts when you want the circuit to oscillate.
This is tricky -- using them to turn on, is indeed okay. You can start in an open-circuit (non-oscillating) condition, enable gate bias, and it rings up.
You cannot stop it by removing gate bias, lest the above happen!
Just making this perfectly clear, in case any readers might make that inference -- being able to do the first step, absolutely does not mean the second is also allowable!
How then? Disconnect supply to the inductor(s) first, leaving gate bias on. Only once energy has dissipated, remove gate bias.
(This will then also work with less robust devices e.g. IGBTs.)
If the input has a big enough capacitor on it, that the supply is not allowed to dip out due to inductor flyback -- then it can indeed be somewhat self-powered by the resonant tank, and energy will dissipate gradually. Self-power...doesn't really mean much, since such a capacitor will dominate total energy storage, anyway; but, in the sense of contributing to total energy, we can go with that.
The capacitor needs to be larger than the oscillator's equivalent capacitance (the transformer/coil center tap (if present; note we can still imagine a virtual one even if using a 2-terminal coil and two supply inductors) presents an equivalent R || C impedance at frequencies much lower than Fosc), and needs to store several times more energy than the inductor(s) at maximum nominal supply current.
Which should be the case when a PSU is attached: its output bulk capacitance is presented to the load.
Actually, even that isn't sufficient help, because some PSUs use very little output capacitance indeed, even in quite high ratings; e.g. LLC and compact (high frequency) types can use surprisingly little. More capacitance might be needed to adequately dampen the oscillator in this way. (I really doubt you'll find a random -- cheap and available -- PSU that actually has this little output capacitance; this is more to say: it's still a technical possibility to need more external capacitance.)
In any case, tossing in a few 1000 uF at the input to one of these circuits, is probably not a bad idea. Low ESR types should be good enough even for fairly powerful units.
Note that, if you switch into said capacitor -- now you have its inrush to deal with as well. Now switching on is harder than switching off. So, again, probably better to let the PSU handle it, and switch its input instead.
But for a proper design you need lots of extra's. and most of that stuff is protection to keep the MOSfets from blowing up. Some idea's:
* Real gate drivers.
* Undervoltage lockout.
* Temperature monitoring and fan control.
Also great ideas; but maybe not so feasible? -- It's one of those problems where, obviously the full design is best, but if one doesn't know how to do it, maybe it really is better to just throw the dice with the simpler design and have a bucket of spare transistors handy.
I fairly recently explained this elsewhere, actually, for those curious about more details why:
https://electronics.stackexchange.com/questions/633217/why-zvs-driver-mazzilli-is-necessary-preferred/633239#633239Tim
Home
Help
Search
Login
Register
