HV Joule Thief

A joule thief that can produce a high voltage output for powering neon bulbs

Similar projects worth following
I really like the concept of joule thief circuits. They have very few parts but they work well and can run on very low voltages.

This circuit is a variation of typical joule thief designs you can find online. It's essentially a regular joule thief with an extra coil of many turns which produces the high voltage output.

On this page I will post information about how I built this circuit and how I tuned it to get the best results
  • Understanding the circuit

    Adam Oakley03/28/2016 at 19:27 0 comments

    At this point I have built this circuit according to the schematic. It works well enough to drive single neon bulbs off a single AA cell, though I have only tested with 1.23V minimum as that is the lowest voltage that my crude bench supply can provide. For phosphor coated bulbs or with multiple bulbs connected it is usually necessary to increase the input to 1.5V or so.

    When I increase the input voltage I get an increased HV output, which is nice. However, once I increase the input voltage high enough, the circuit will stop operating correctly and won't provide enough output voltage anymore. My goal right now is to understand why this happens and if possible, find a solution for this problem. It is nice that this circuit could run off a single cell battery, but I would like to be able to run it on better power sources if I want more output. I will try to get more information about this issue with my oscilloscope, and I will post an update with some scope captures.

View project log

  • 1
    Step 1

    Here's how I built this circuit. If you wish to make your own, you may need to do things differently depending on what materials you have.

    Parts needed:

    • Ferrite core. I used a toroid core salvaged from a computer power supply.
    • Fine magnet wire. I used 30awg enameled magnet wire for the HV coil
    • Thicker magnet wire. I used 24awg enameled magnet wire for the primary and feedback coils
    • Power transistor. Try to use a transistor that can handle at least 500mA. I try to pick transistors that still have decent gain at 100mA or higher
    • Fast diode. Although a standard 1N400x diode will work, you will get more performance with a fast diode on the secondary. Make sure the diode can withstand the high voltages generated on the secondary.
    • Resistors and capacitors. It's best to have a variety of parts on hand so you can see what works best in your circuit.

    Build Process:

    1. Wind the HV coil. I found it easiest to first wrap enough wire on a small stick. You can pass this stick through the toroid core when you wind the coil instead of passing a huge length of wire through each time. The goal is to get as many turns as possible on this winding, so make sure you wind the wire tightly to fit as many turns in as possible. Don't wind the coil completely around the core. There will be a high voltage present on the ends of this coil, and you should keep some distance between the ends to help isolate the HV.
    2. Wind the primary coil. Using the thicker magnet wire, wrap about 8 turns around the core. Since the insulation on magnet wire is not perfect I tried to keep the primary coil from physically touching the HV coil. If your core is too small you can't avoid them touching. This was the case for me, so I applied a coating called "Corona Dope" on the HV coil to keep it well insulated.
    3. Wind the feedback coil. Start winding the feedback coil where you finished the primary coil, and keep the polarity the same. The end of the primary and start of the feedback coil will be connected together so it is convenient to place them next to each other. Wind about 4 turns for the feedback coil.
    4. Strip the coil wires. There are several methods of stripping magnet wire. I prefer to scrape the enamel insulation off with a sharp knife. Strip off enough insulation from the end of each coil wire so you can plug the coils into a breadboard. I like to solder pin headers to the fine gauge wire ends, otherwise they slip out of a breadboard easily
    5. Prototype the circuit. Following the schematic, start building the circuit on a breadboard. Put the primary and feedback coils of the toroid in first, they will help hold the toroid in place.

    At this point you should be able to power up the circuit and check the voltage of the capacitor on the secondary. You could do this with a multimeter but I prefer to use an oscilloscope to get more information. If your circuit has no output or it is very low, the polarity of the secondary could be incorrect. Swap the ends of the coil around on your breadboard and try again. With my circuit I was getting an output of about 40V for 1.5V input. Interesting, but not enough to drive a neon bulb. There are a couple solutions to this. By tuning the circuit I was able to get the output closer to 70V, still not enough but a big improvement. I solved this issue by winding another coil of fine magnet wire on top of the secondary. This gives you two HV coils, which you can wire up in series to get higher output voltage. I recommend insulation between these two HV coils, such as the Corona Dope coating I mentioned before. With the two HV coils and proper tuning I was about to get up to 200V output using 2.8V input. I will describe the tuning process next.

View all instructions

Enjoy this project?



Yann Guidon / YGDES wrote 03/23/2016 at 18:55 point

Hi ! Just a little note : HV lights like neon bulbs require alternating, balanced voltage to prevent electromigration and breakdown of the electrodes. However the typical Joule thief is unbalanced so it's not suitable for long term use...

But yeah, I love the Joule thief too ;-)

  Are you sure? yes | no

Adam Oakley wrote 03/23/2016 at 19:13 point

You mean balanced AC? I rectify the secondary into a capacitor so it's fairly smooth, but it is DC. I'll post the schematic tonight

  Are you sure? yes | no

Yann Guidon / YGDES wrote 03/23/2016 at 19:23 point

Yes, I mean balanced AC, so the same RMS current flows in both direction, which keeps the electrodes from deteriorating. This is why CCFL's electronic ballasts have a 2-transistor oscillator with a 3-windings transformer.

I must have a CCFL inverter somewhere with a single transistor but it uses a 3-windings transformer too, one for low-voltage power input, the other for high voltage ouput (many thin wires) and a 3rd for voltage sense.

How many BJT did I kill with bad design of my "experimental Joule thieves" ? :-)

  Are you sure? yes | no

Adam Oakley wrote 03/23/2016 at 22:22 point

Interesting. I never knew that. I wonder if that's what limits the lifetime of nixie tubes?

  Are you sure? yes | no

Yann Guidon / YGDES wrote 03/23/2016 at 23:05 point

Maybe. I was thinking about them while writing. However they work at lower power levels and their electrodes are probably different. Neon bulbs however are meant for AC.

  Are you sure? yes | no

Martin wrote 03/24/2016 at 08:31 point

@Adam Oakley:

Nixie tubes are essentially neon lamps. In a neon lamp only the negative electrode lights up. This is what you want for a nixie. If you drive a neon lamp with AC, both electrodes light up (alternating). You do not want this in a nixie tube, so you can not drive it with AC.

Most sensitive to DC drive are fluorescent tubes with oxide coated electrodes, which can sputter away. But there are small hand held 8W fluorescent tube working lights with a DC drive voltage doubler circuit and often no electrode preheating. Yes they have lower lifetime than AC drive, but they work for relatively long time. It is less than ideal, but CHEAP.

@Yann Guidon / YGDES

If you have a CCFL in a display, it is not replaceable. So you really want the longest possible lifetime and drive it with high frequency AC.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 03/24/2016 at 11:21 point

Thank you Martin for the details :-)

  Are you sure? yes | no

K.C. Lee wrote 03/23/2016 at 22:29 point

BTW You might have seen  my project:
I bumped up the efficiency by using a BJT/MOSFET hybrid to increase the efficiency and also implement a voltage feedback.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 03/23/2016 at 23:02 point

LEDs are polarized elements so the issue of AC is not present there.
But yeah, nice circuit :-)

  Are you sure? yes | no

K.C. Lee wrote 03/23/2016 at 23:12 point

I have designs for driving VFD filament, but it is in the pipeline.  The new design uses a single ended driver, but more or less balanced AC due to the resonant design.

  Are you sure? yes | no

Adam Oakley wrote 03/24/2016 at 00:36 point

Cool project! I've seen some hybrid circuits before but I've never built one.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates