0
\$\begingroup\$

I need to make a continuous high voltage arc for some project (Birkeland Eyde process).

Most generators use ZVS with mazzilli driver, some use voltage multiplier stages and transformers etc.

Since I have the parts, I'm gonna go the ZVS way, but why do I need a mazzilli driver to control the flyback transformer primary?

Though it's kinda straightforward and needs few number of components and the working principle is known and explained, it seems hard to gain full control over frequency/currents and voltages inside the circuit. Many burnt mosfets and igbt questions implies so.

Why not use a microcontroller (or any other precise pulse-generator) combined with MOSFET driver and a single primary coil on the flyback transformer?

Is it just cost or the mazzilli circuit offers benefits I'm not aware of?

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Have a look at this. It was the design used in Robert Grove's laboratory and shown to a visiting Maxwell (in his mid 30's at the time) in March of 1868. The whole point of the device was to generate a continuous arc in air from a DC power supply. If they could do it in 1868 without any of our modern electronic devices, you can as well. The picture is from "Steinmetz and the Concept of Phasor: A Forgotten Story" by Araújo & Tonidandel, 2012. No MOSFETs or IGBTs to blow up. Just works. \$\endgroup\$
    – jonk
    Aug 31, 2022 at 22:17
  • \$\begingroup\$ Or use one of these with a transformer to generate the HV AC. You'll get a nice continuous arc from that, as well. Just basic mechanical devices that long precede all these fancy and fragile do-dads like MOSFETs and IGBTs. Who needs semiconductors? \$\endgroup\$
    – jonk
    Aug 31, 2022 at 22:23
  • \$\begingroup\$ I think it is the self resonant feature of the circuit, that makes it attractive. Changes in the construction of the transformer or load don't require new frequency calibration. \$\endgroup\$
    – Jens
    Sep 1, 2022 at 0:07

2 Answers 2

1
\$\begingroup\$

For the functionality versus simplicity, it's hard to beat. When operating properly, switching loss is almost nil and conduction losses dominate. Operating range is reasonable, basically it just draws more and more current as load goes up, until the transistors are unable to saturate hard enough (which is the major limitation in high voltage application), then the opposite side transistor fails to turn off, oscillation stalls out and it instantly cooks off.

Protective features could be added, though at fairly substantial complexity for a newbie:

  • It's not easy to control the gates directly, because this is very much an analog circuit: at startup, the gate voltages must change by slight amounts, amplifying initial imbalance or indeed random noise into a full wave oscillation. Attempting to digitize this (say by using schmitt trigger gate drivers) results in a screaming mess, driven by propagation delay and stray inductance/capacitance; it's a dice roll whether it settles into the main resonant mode at all. Once it's running, square waves are fine, but getting there is non-trivial.
  • Digitally mimicking full behavior involves a VCO + PLL, and some means to sense output phase (voltage or current, or a combination thereof). This isn't too hard (e.g. CD4066 + gate driver), but again, it's more complexity.
  • It can be controlled externally, leaving the analog oscillator core alone, and varying its supply with a buck regulator or something like that. This basically needs any CC/CV "bench supply" sort of characteristic, which is pretty simple (well, maybe not to design from scratch, but modules are available), and you can control the setpoint(s) of that with an MCU, and also add output monitoring (overtemp, osc stall, shutdown?). Again, still more stuff -- but this time at least within scope of building blocks and not many lines of Arduino code, say. Do beware of EMI problems with all these switching circuits around: there can easily be enough noise emitted by un/poorly filtered modules, to blow out the logic input levels, or even damage the MCU if hard wired. Good grounding/shielding and filtering are a must. Not to mention the sparking output which likely gives off a lot of nasty EMI or EMP.
\$\endgroup\$
1
\$\begingroup\$

Why not use a microcontroller (or any other precise pulse-generator) combined with MOSFET driver and a single primary coil on the flyback transformer?

The mazzilli circuit isn't good on higher powers. If you want high voltages at higher powers I'd recommend a H-bridge driving a step-up transformer with as many doublers on the output as is necessary to take the several thousand volts from the secondary up to what you need.

I'm currently designing a "several tens of kilowatt device" and wouldn't entertain the mazzilli circuit on anything over a hundred watts. In fact I'd just scale down the H-bridge even on lower powers.

For hobbyists it's fine (if not a little dangerous to be messing with these things) but Darwin's legacy usually evolves us into being more careful!

Is it just cost or the mazzilli circuit offers benefits I'm not aware of?

Well, if it keeps burning MOSFETs and wasting your time then, the cost of ownership might be severely called into question. The benefit to a hobbyist is simplicity and ease of understanding the circuit because, the sophistications of a proper circuit with all the bells and whistles might be lost.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.