I am trying to build a powerful and not lethal high voltage arc lighter. I revisited the project and have not had much luck. A simple relay circuit worked, but it was not practical or efficient at all. I made a MOSFET circuit with a shunt and LM358 op amp. It sort of worked but often the fets (IRFZ44N) would heat up way too much and fail. Maybe there was also inductive spiking. Well, regardless, I had to replace them like fuses. It also didn't seem to work too well with the flyback diode. I got two seconds of arcs powerful and then it failed. If I was lucky. It was about 15V and 4A.

Then I got a mosfet, 500ohm pulldown resistor, and arduino. I did a frequency sweep. It made a sound, but no arcs or significant output power. Still pretty cool (not what I was going for though). I also tried fixed frequencies up to 120kHz.

https://photos.app.goo.gl/2hyzPxoCnPNdnxVA7

The primary voltage stayed around .6-.7V (DC measured on my meter). But it did seem to show some switching too. So I am wondering if the reverse recovery of my diode is an issue. And I tested the transformer after, and it made some arcs when briefly applying a DC voltage to the primary (caused by inductive spiking). Yes, I am sure the diode was reverse biased. It is a 10A silicon diode.

 

a powerful and not lethal high voltage arc lighter

This is a contradiction in terms.

 

circuit with a shunt

You keep saying that in many of your posts, why do you think you need a 'shunt'?? Like when you joined, it seems like you hear a word (that you didn't know before) and then put it into everything. Do you know what a shunt is or does?

 

You keep saying that in many of your posts, why do you think you need a 'shunt'?? Like when you joined, it seems like you hear a word (that you didn't know before) and then put it into everything. Do you know what a shunt is or does?

Yes, of course I know what it does (it has a voltage proportional to the flowing current). The circuit I was talking about amplified a shunt voltage, then fed that to the inverting input of a comparator. I used an LM358 dual op amp. The comparator output goes to a mosfet with pulldown resistor. So when enough current flows, it turns the mosfet off. But then there is not enough current, so the comparator turns the mosfet back on. It was pretty basic but required minimal parts and worked.

 

This is a contradiction in terms.

If it's 15-20kV and 4mA then that's about 70-80 watts of heat, but not lethal. You just need a current limit

 

The circuit I was talking about amplified a shunt voltage, then fed that to the inverting input of a comparator. I used an LM358 dual op amp.

That's your problem, right there.

Do you know the difference between an op amp and a comparator? They're two VERY different beasts, and generally not interchangeable.

Operational amplifiers have an output which is slew-rate-limited, a consequence of their design for stability (i.e., won't oscillate) in a feedback circuit. The maximum slew rate of the op amp's output (about 0.5 V/μs for an LM358) determines how long it will take for the output to get from one voltage level to another, irrespective of input drive. And that, in turn, will determine how long the output takes to transition your MOSFET from full OFF to full ON-- probably at least several microseconds. During that period the MOSFET is partially ON, and dissipating a lot of power, which is why it's getting hot and failing.

A comparator, on the other hand, is not designed for use in feedback circuits and therefore has no internal frequency compensation circuitry to limit slew rate; it is designed to have its output switch from one state to the other as fast as possible. Some of the slower comparators, like an LM393, take several hundred nanoseconds for the output to switch ON or OFF; faster ones can switch in just a few nanoseconds.

Use a proper comparator to drive your MOSFET, instead of an op amp, and your problem will go away.

 

I know an op amp has many limitations here, but I thought the delay might mean the mosfet would switch at a more manageable frequency. I kind of assumed it could handle the time in the ohmic region (I know it leads to big losses). But it made the sound and failed when the mosfet was driven by an arduino (with pulldown resistor, and I know a few 10s of mA isn't good for charging the capacitance). I tried 100Hz-120KHz, and it failed. No voltage on the output.

But I will get some comparators and will see if it works.