ZVS (Mazilli) Driver

A “ZVS driver” is a mazilli oscillator; a very simple circuit that can oscillate a large amount of power with about 90% efficiency. To the right exists a simplified version of the mazilli oscillator, so take a good look!

When power is applied at +V current starts to flow through both sides of the primary and on to the mosfets’ drains. Simultaneously that voltage appears on both of the mosfets’ gates and starts to turn them on. Because no two components are exactly alike one mosfet turns on a little faster than the other one and more current can then flow through that fet. The extra current flowing in that side of the primary robs the gate current from the other fet and starts to turn it off. A condenser forms an LC tank with the primary and the voltage proceeds to rise and fall sinusoidally. If it were not for that capacitor, the current would continue to increase until the transformer’s core saturated and the mosfets exploded.

Imagine that Q1 was the first to turn on. The voltage at point Y will be at near ground while the voltage at Z rises to a peak and falls back down as the LC tank goes through one half cycle. As the voltage at Z passes through zero the gate current to Q1 is removed and the mosfet turns off. The voltage at point Y is now allowed to start rising and Q2 turns on. That mosfet clamps the voltage at Z to ground; something that makes sure Q1 stays off. This same process repeats for Q2 completing the other half cycle, and the oscillator continues cycling. In order to prevent the oscillator from drawing huge peak currents and exploding, L1 is added in series with +V as a choke. The LC impedance is what limits the actual current (the choke just mitigates current spikes).

A keen eye will notice that this oscillator is zero-voltage switching (ZVS), meaning that the mosfets switch when they have zero volts across them. This is good because it allows the mosfets to switch when they are carrying the least power; something that for the most part eliminates the switching losses which generate huge amounts of heat. This means only small heat sinks are needed, even when oscillating 1000 watts!

Being a resonant oscillator the frequency that the mazilli will run at is determined by the inductance of the transformer’s primary coil and the capacitor. You can use the following formula to figure this out:

f = 1 / ( 2π * √[L * C] )

f  is the frequency in Hertz
L is the inductance of the primary in Henries
C is the capacitance of the capacitor in Farads

Now in reality mosfets are rather fragile components and if the gates are +/- more than 30V from the source the mosfets will be destroyed, or at least degraded significantly. In order to prevent this scenario from occurring we’ll need gate protection; something easily added with a few extra components. See the schematic to the right.

• The 470 ohm resistors limit the current that charges the gates as too much gate current can cause damage.
• The 10K resistors pull the gates down to ground to prevent latchup; a process in which the mosfet gets stuck on.
• The Zener diodes prevent the gate voltage from exceeding either 12, 15 or 18V depending on the zeners you use.
• The UF4007 diodes pull the gates down to ground when the voltage on the opposite leg of the tank is at ground.

One may notice that instead of charging the gates with the LC tank we are instead using +V to charge them up and we are using the LC tank to discharge them via the ultrafast diodes. This improves the overall performance of the circuit.

The following schematic was made very easy to understand, I hope you like it.

Because of a bit of black magic known as resonant rise the voltage in the LC tank will be about pi*vcc, so you’ll need to make sure your mosfets can withstand this tension. A good rule of thumb is to use mosfets that are rated at 4x the voltage you plan on feeding the oscillator and the IRFP250 or the better IRFP260 is a good mosfet for the task. You’ll need some heatsinks for the mosfets, but they do not need to be large. They must not be put on the same heatsink unless insulating pads are used because the metal back of the mosfet is not electrically insulated (it is connected to the drain). Also be sure to use thermal goop when you attach a heatsink else the thermal transfer will be crap. 

The capacitor must be a good one, an MKP, mica or Mylar cap is a good option. Do not use an electrolytic cap, it will without a doubt explode. The two primary windings must also be wound in the same direction or else the oscillator will not function. The oscillator will also fail to function if there is no air gap in the transformer’s core, so always make sure that one exists.

Below is a youtube video of a mazilli oscillator powering a flyback transformer at 12, then 24, then 36V. Skip to 0:47 for the 36V if you are impatient.

 

Problems with the Circuit

The mazilli oscillator has one fatal flaw: it likes to explode above 70V. 60V, does well, 70 is meh… 80 KABLOOEY. The problem is above 70V the powers tend to be so high that the diodes responsible for turning off the gates fail to fully do so, and the oscillation stops with one mosfet left on. That’s essentially a short circuit so the mosfet responds with suicide. To anyone who is reading this article, I propose to you a challenge: fix this problem. First one to do so will receive a present. I don’t know what but it’ll be something. Neon John attempted a fix, but it’s still pretty unreliable…

 

UPDATE: I partially solved the problem by placing a 0.5 ohm wirewound resistor in series with the filter inductor. Now things don’t explode if the load inductance plummets. Still asplodes when VCC>70V though.

31 thoughts on “ZVS (Mazilli) Driver

  1. Sir I’ve built this circuit and it gives really large sparks. I would like to lessen it’s voltage out around 1kv. Is this possible cause I need this driver for another project?.
    Thanks for the info.
    Andrew.

  2. At 60V supply the Vdss of IRFP260N will be near the 200V breakdown limit.
    So exceeding the 60V supply voltage limit the Vdss go over the 200V maximum and the mosfet will be damaged.
    amz.

  3. You might fix this Mazzelli converter by adding two small signal transistors and three resistors to each mosfet’s drive circuit.

    Connect an npn transistor with its collector to the gate of the mosfet and its emitter to the source. Then connect a pnp transistor with it’s emitter to the gate of the mosfet and its collector connected to the base of the npn though a resistor. Connect a resistor from the gate of the mosfet to the base of the pnp and another resistor from the base of the pnp to the anode of the diode.

    What happens… the cathode of the diode goes low and it pulls the base of the pnp down causing it to conduct which turns the npn on clamping the mosfet gate to source.

    The problem is the combined forward drop of the diode plus the voltage drop across the mosfet’s Rdson on isn’t low enough to reliably keep opposite mosfet off. The npn transistor clamps the mosfet off effectively.

    Also connect the two 470 Ohm resistors to a regulated voltage. The zener current climbing as V+ is cranked up isn’t helping.

  4. It might help to make the inductor at least 4 times larger than the primary inductance of the transformer and use a freewheeling diode. Connect the diode’s cathode to the V+ side of the inductor and connect the diode’s anode to ground. The diode should be rated to carry the input current. A large bypass capacitor in parallel with the diode is a good idea too.

  5. One way to fix the circuit is the get better MOSFETS with a higher voltage and current rating upgrade to 1 watt 15 volt Zener diodes and get faster, higher voltage rated diodes. And lastly upgrade to 5 watt 470 ohm resistors

  6. I’ve recently taken to experimenting with flyback transformers because of my interest in lightning and other natural phenomena. The website link that I posted will show a circuit that is designed to provide a sort of pulse width modulation. The tuning effects are interesting, however, the site that this circuit came from is inoperable at present and so the corona demo (along with sound effects) is not available. I have just recently built the ZVS driver that is described here after winding my own flyback for the cause. This was a fun project and I want to thank you for taking the time to make this an informative and interesting project. Having “cut my teeth” thus far I plan to revisit this PWM design. Any comments appreciated…thanks.

  7. try to use other mosfet transistors and parallel them about 2*10pcs irfp460n and better gate drivers with separated 15v supply voltage and try to slowly increase supply voltage to less then 1/3 max rated voltage of mosfet transistors used and not to forget to first powerup gatedrivers and then apply and slowly increase suply voltage to the inverter

  8. SO, explain this to me. I’m in the process of building this but I can’t possibly see how it won’t catch fire. I’m using a small, 12v lead-acid battery and other parts are as spec’d, but the resistance of the mosfet when on is tiny, and the resistance of the choke plus the 5 loops of wire wrapped around a ferrite core is also tiny. At the moment the mosfet is fully on, I have trouble believing the total resistance seen by the battery is going to be much over 0.1 ohms, giving a peak amperage of 120A. Why don’t the wires melt and the mosfets explode? Insight welcome.

    What I need is a bright, hot noisy spark that can bridge about 1/2″, and run more or less continuously without melting down, from a 12v battery and a portable circuit. I’m hoping this circuit it is, but suggestions are welcome.

  9. I, think, you can fix problem running in 70 volts using not 470 omh resistors, but about 3 kiloomhs. And using not 10 kiloom resistors, but 20-30 kiloomh. On 13-19 volts I try succesful with 1 kiloomh resistors 5 W (not 470 omh) and with 5 kiloom resistors betwen gate and sorce (not 10 kiloomh). (Zener I use 16 volts. Mosftes that use, I now not remeber.)

  10. Hello
    Thank you for this beautiful article. I’d like to know which has been your minimum limits in capacitance to try to reach the maximum frequency on this circuit?
    Thank you in advance for any answer

  11. Hi,

    Beautiful circuit. Thanks for this article.
    I wonder if it would be possible to adapt this circuit like a Meisner circuit, where the load, say a battery and it’s capacitance provides the feedback to switch the fets?

  12. Is there anything special about the 47 – 200 uH inductor? Just needs to match up with the current draw? And the current draw depends largely on the inductance of the primary coil and size of the capacitor?

  13. I’m thinking of using this as the power supply for a 40 kV Cockroft-Walton voltage multiplier, that’s going to be driving a pulsed DC load, with time average power of about 14 – 30 W. Is there anything I should worry about when running a flyback at such low loads? Peak output current of about 5-20 mA, but as it charges my pulse capacitor, it naturally drops off.

    • Certainly not; these transformers after all were designed to run at about 30W. This mazilli circuit pushes them much further than that, however.

  14. Can I substitute GP15D diodes for the UF4007′s. The GP15D has a Vrrm=200V, If=1.5A and trr=3.5uS. I dont expect to run the ZVS at more than 48V supply.

    John.

    • (3.5 microseconds)^-1 = 285kHz, so as long as you stay under 30kHz or so, they should work.

      A good rule of thumb is to never have the diode recovery exceed 10% of your switching time.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>