Plasma Speaker 2

My last plasma speaker, which used only one mosfet, got a bit too hot to run without a bucket of ice water nearby. To cut down on that heat I decided to use a half bridge switch to drive the flyback this time. A half bridge is a topology that uses two mosfets that flip flop, each taking a turn conducting current. This reduces the load on the FETs during their switching times, and thus reduces the heat.

Below is a pictorial schematic of the device, and here is a regular schematic for the electrically inclined. For those that like instructables, here’s one.

This circuit is considerably more complex than the other one, but it still uses the same principal:

By sending more current through an arc, you produce a hotter arc. Hotter arcs will have a higher pressure, and by varying the current through an arc you create varying pressures, something commonly known as sound.

Now this circuit uses PWM to accomplish that. By shortening the width of the pulses to the mosfet’s gates you can reduce the current that flows through the flyback. By changing those pulses to the tune of music, the arc’s current and pressure will vary in tune with the music, thus producing music!


Some important notes about this circuit:

    • The 10kΩ pot still controls the frequency, and a lower frequency means higher voltage. The sweet spot was around 50kHz for my flyback, but it may be different for yours. It’s a trivial matter to turn a knob though.
    • The 1uF cap must be either an MKP, mylar or foil cap. No electrolytics! Its function is to block excess DC current from flowing through the transformer, since capacitors block DC yet pass AC.
    • There is a gate drive transformer. What this does is provide an isolated gate drive for the upper mosfet preventing it from dying, and it is required for the circuit to work properly. A GDT is simple to make: all you do is take a bundle of 3 wires and wrap them around a ferrite toroid. I used 14 turns on a 2.5 centimeter toroid, but other designs may work better for you. Ferrite toroids can be bought for about a dollar online if you don’t have one laying around.
    • The GDT must be phased correctly for this to work, otherwise your mosfets blow up. What is phasing you ask? Phasing is the direction of the coils in a transformer. For this to work, one of the coils on the mosfet side must be ‘reversed’. On schematics phasing is marked by dots, and you can get a better picture of what’s supposed to be done by looking here. A properly phased transformer looks like this on a scope; one wave is inverted.
    • Too loud a signal can and will overload the chip. When you put audio in, set the volume to 0 then slowly turn it up. When the music becomes distorted that’s as loud as it’s getting. It’s not terribly loud, but it is still easily heard in a quiet room. You must have a clean audio source too, no electrical noise. I used my computer, and it would appear that iPods don’t work too well. It’s trial and error with the sources.

My speaker:


Youtube video of it playing some supertramp. It sounds much better in real life (more bass). Sorry Germany, UMG blocked the video in your country.


26 thoughts on “Plasma Speaker 2

  1. Almost the same question as the guy above me^

    i keep burning out my fly back everything works, but i keep burning out my flyback when i boost the input voltage to 48 volts.
    since it’s just an autotransformer are there any other alternatives?
    Could I just use another High voltage transformer that has a higher VA rating? 8-10 kv range?

    • It’s not an auto transformer, but there is no reason why another high frequency transformer wouldn’t work in lieu of a flyback.

    • I don’t believe such a transformer would work then, unless you use a large current-limiting impedance on its output.

      Otherwise you burn out the diodes inside.

  2. First mine works great except I seem to be not getting such a big arc because of only using two capacitors. If I add more capacitors such as you have in your video is it just two banks of 5 capacitors in sereis then in parallel leading to the flyback 6 turns?

  3. I build this , but when i’m connect circuit to 12V Battery there’re nothing out from FBT and IC (SG3525) get hot! My GDT use 28awg wire and wind for 14Turn, Phase is correct

    What am i wrong or missed? Please answer

  4. Thank for your reply. Now I know solution! I read comment on instructable and i found guy having problem same as me, Chip getting hot because he used yellow toroid core, Yes i’m same as him , I use core from old computer power supply then i will change it and try. Thanks :)

  5. I have fixed my previous problem by using a bigger transformer.

    I am now trying to Run this with a 150Vdc input.
    i’ve bought 2x200v caps and have made a suitable linear powersupply and used a power resistor and shunt regulation to mitigate the cap Inrush current at the begining.

    my problem now is that my mosfets have blown up, i used two irfp350N s. i now realize that the specs where too low in terms of current.
    What type of mosfets would you recommend to run this at 150Vdc?
    something like:

    Vdss: 500V
    Id: 55A
    Rds(on): 80mΩ
    Pd: 600W,K)55N50,%2050N50.pdf
    what do you think the maximum voltage rating should be ?
    Whatwould the maximum Rds that wyou could get away with be?
    Or can i get away with lower values mosfet, any suggestions?

    • Twice the expected voltage is usually a safe bet, but what you have chosen there is fine. An 80mohm RDS on is okay, but I wouldn’t go above 0.1 ohms if you can avoid it.

  6. I am Running this circuit with 70v DC
    But i’m finding the switching controller chip is actually heating up.
    I’d like to buffer the signal before it gets to the inductor.
    I’ve fashioned a simple push pull amplifier, the output waveforms look pretty identical although now the arc doesnt ignite.
    Any thoughts on buffering the push pull signal?

    • td8f4 A simple AB amplifier should be just fine. Remember though, mosfets are switched by voltage, not current like BJTs.

  7. Hello, i might need some help.
    i have buld part of the circuit, but am unsure of if my GDT is what is required here.
    here is my circuit, it is still missing the mosfets, zeners, caps, and 22 Ohm resistors.
    That is how it looks like at the moment:
    That is what i get out of the ic:
    and that is on a other turn of my GDT /my minus is still the one of the psu/
    Since the images are not so clear the both measurements ar at 10micro seconds. the firs one is 2v per devision and the second .2V per devision.

    Does it seem ok to you?

  8. Hello

    Im curious what the maximum input current is for this circuit??
    i have a 24 volt 20Amp source will that work

  9. Meant to build this like 2 years ago when this used to be teravolt and finally getting around to actually potentially building this, was wondering though while I gather the missing parts I need, would it be even more effective to use a lm7824 or would the additional 6 volts would just go to waste since this extra voltage is only fed to the lg3525 ic and the transistors go with the 35 volt straight from your source anyway?

  10. Hi, another forum member, found your old simulation done in LT Spice IV. It shows three separate waveforms Vgm1, Vgm2, HVout.

    I’m very new to LC Spice and I would like to know where in LT Spice to setup the command etc.

    Thank you.


    • What command?

      LTspice is interesting in that it supports “forced” spice directives. Instead of fiddling through 1,000 menus to find what you need, in LTspice all you do is paste in for example, “.function normal(nom,tol) nom*(1+gauss(tol))” to define a random, Gaussian value function.

  11. Adam, I found out how to create new Plots and how to name and setup my traces, with a right click.

    Thanks for the response. I’ll be building this now and I hope I’m able to get a good transformer. Any suggestions?

    I’ll by a new one if they have it, somewhere.

  12. Thanks for posting your build! I never could have done it on my own : )

    I am getting a lot of static when inputting the audio. The initial arc without audio is basically silent but when adding audio a static sound drowns out most of the audio. Have to put it full volume just to sortof hear it. The source audio is clear in headphones or speakers. Any ideas what this might be?

  13. I finished this finally had some trouble with the Core I was trying to use from a Computer Power Supply. Once I had the one suggested all went as expected. I stress tested it for 6 hrs. It did just fine. Good circuit.


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>