I was trying to generate plasma inside a cylinder for a hobby project, so I decided to use a flyback transformer from an old TV and drive it with a 555 timer as shown in the schematics. enter image description here

It solves my problem and the plasma looks something like this. Almost evenly distributed across the cylinder.

enter image description here

But, sometimes due to foreign objects like metal pieces or mosquitoes inside the cylinder the entire discharge happens through a single spark (which is obvious). But that's not what I want; I want an even distribution of the plasma. I want to detect these sparks and shutdown the flyback for say 1 sec and then start it again(like in a hiccup mode) or report a fault with an LED.

I tried measuring the current via the R10 shunt resistor to find a change in current waveform during normal operation and during a spark, but I couldn't see much difference between the both. Is there a way to find the high voltage sparks and shutdown the circuit?

PS: I have limited knowledge of electrical engineering.

  • 2
    \$\begingroup\$ Looks cool. Can't help with your question. Can you tell us what you are up to? Curiousity is killing me. :) \$\endgroup\$
    – JRE
    Commented Aug 11, 2017 at 7:32
  • 1
    \$\begingroup\$ 0. This looks awesome, what are you planning to do with that? 1. When you say measure, do you mean using a multimeter or a scope? 2. you might want to add a grid in front of this tube, it looks like a huge safety hazard. \$\endgroup\$
    – Sclrx
    Commented Aug 11, 2017 at 7:58
  • 2
    \$\begingroup\$ Just an idea if nothing measurable comes up in the circuit: does the sound change? In plasma mode it's silent and in zap mode it makes a zap-sound? I guess that would get quite involved but you could detect the zap sound and shut down the system. \$\endgroup\$
    – Arsenal
    Commented Aug 11, 2017 at 8:58
  • 2
    \$\begingroup\$ @Arsenal has a good idea - if you can't sense the rogue discharge at the source-side, sense at the plasma-side. Optical detectors, magnetic field (loop antenna), electric field (probe antenna) also come to mind. What irks you about rogue discharges? \$\endgroup\$
    – glen_geek
    Commented Aug 11, 2017 at 13:28
  • 1
    \$\begingroup\$ I'm not entirely sure, but I think the infrared photodiode should detect the sparks. But I do not know if it will react to the plasma. You would have to experiment. \$\endgroup\$
    – sigaris
    Commented Nov 17, 2017 at 7:42

3 Answers 3


Do you have 300MHz BW on your scope? That's what you will need I suspect. See my other answer on what to expect. Parasitic Capacitor current of an inductor Put one or more turns around the secondary into high quality coax ( not RG58) or SAT. coax and terminate with 50 or 75 Ohms to match the coax. This is a 1:1 CT.

The energy stored into the falling dielectric starts as a rising voltage relative to the kV/mm in the plasma , then it breaks down and explodes inspite of the low capacitance but large V^2. The plasma waves are semi-conductive ( negative resistance ) rather than insulative air dielectric.

Then use a peak detector that has 300MHz BW (CML comparator) to amplify and peak detect the pulse with a low pF high speed diode to small cap to stretch it but not current limited then out to trigger a lower (20MHz) BW 1-shot to 1 second.

There may be a PD pulse before a full breakdown arc and if you are fast enough you can divert the energy, somehow.

For commercial and researcher solutions, use google scholar on "PD detector" which tend to cover a wide range in costs. You just want to detect the secondary stored energy pulse which does not show up on primary.

My signal came from an invisible contaminant in a jar of oil at about 20kV with a 1mm gap.


I do not have a great solution to this problem, but this is a common issue in fast high voltage switching (pulsed power). This issue is usually referred to as 'multi-channeling'. Multi-channeling lowers the inductance of an arc channel by distributing the current between multiple arc paths. The way to determine weather or not multi-channeling is achieved is by observing the rise time of the current in the discharge path. This is typically in the 100's of nanoseconds depending on the auxiliary circuitry (probably slower for a flyback). If you compare a single channel breakdown with a multi-channel breakdown, the rise time of the current for the multi-channel will be faster because of the lower inductance of the current path.

The best way to measure the current of the signal is to use either a current transformer such as a Pearson or a rogowski coil. I'm guessing most commercial options will be out of your price range. You can try to make one of these diagnostics, or attempt to make a D-Dot probe to measure the rise time of the electric field in the region of the breakdown capacitively.


In order to detect the spark in a current circuit or in a high voltage transformer, do like this: If you have an old radio, place it where the spark is. At that point, the radio signals will not be working and the radio will not play properly

  • \$\begingroup\$ Hello. I think the original poster (OP) wanted something more automatic to shut down his/her circuit. \$\endgroup\$ Commented Nov 3, 2018 at 15:07

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.