Mini High Voltage Supply




About: I'm an electronic engineering student. I don't usually have much spare time but I like to work on random projects to keep myself entertained. I hope you like them!

WARNING: Before you start making anything please take a moment and read this:

  1. This circuit is intended to be used for educational and experimental purposes (electrostatic experiences, franklin bell experiment, plasma generation, gas ionization, electronic igniter, testing of insulating materials...) this circuit should not leave the lab or your house, and it shouldn't be used to harm to anybody, human or animal.
  2. Do not attempt to replicate this circuit if you aren't familiar with high voltages or intermediate electronics, high voltages are very dangerous. This voltage supply uses small capacitors to achieve a desired voltage, remember to safely discharge those capacitors by shorting the output when you finish using the supply to avoid accidental shocks (it hurts, but the energy stored in the 10nF caps is very small to be dangerous). Take into account the sparks and corona discharges generate ozone, so use it in well ventilated areas.
  3. High voltages can disrupt electronic equipment, so don't keep phones, pacemakers or other sensitive electronic devices near the supply.
  4. I'm not responsible for the use given to this device and I've made all what it's on my hands to include safety related information, and safety implementations to the circuit.

Follow the general security measures when dealing with high voltages, here you have a nice safety guide, please read it carefully before you continue.

Although this device outputs a huge voltage, the associated current is extremely small, but it can still be dangerous and produce a quite nasty and painful shock.

In terms of safety, this is one of the safest high voltage sources, since the current output is comparable to the tasers used by the police. It's a safer alternative to more dangerous power supplies made out of TV flyback transformers. Nevertheless the dangers can't be underestimated.

Step 1: Introduction

This high voltage supply is designed to output alternating pulses of DC voltage around 10 to 20 kilovolts, I haven't really measured the voltage, but the spark gap can get as long as 1.5 Cm, this can vary due the different elements used to make the circuit.

The voltage itself can be regulated varying the amount of stages at the Cockroft-Walton multiplier, for example, if you want it to lit a neon bulb you can use 1 or no stages at all, if you want to power a sparkplug you can use two or three, and if you want a higher voltage you can use 4, 5 or more.

Bear in mind less stages mean less voltage, but more current, what could increase the dangerousness of this device. Ironically, the more voltage you get, the less difficult is to be harmed by the supply, since the current drops to a negligible point.

Also, take into account the Cockroft-Walton multiplier isn't ideal, so the more stages you add, the more losses you'll get, until you reach a point where more stages mean a decrease in the voltage, I recommend 5 stages, although I haven't tried to use more.

At the top you have a short video in which I test it:

Step 2: How It Works:

Understanding how this circuit works is not crucial to be able to assemble it and make it work, you can skip this part if you want, but it's nice to know how things work:

After pressing the button, the IR diode is activated and a beam of light hits the sensor of the optocupler, this sensor drops it's resistance to about 50 Ohms, unleashing the energy needed to activate the 2n2222 transistor.

This transistor allows some energy to power the 555 timer. For those of you who don't know what a 555 timer is, it is a chip, that, amongst other functions, can take an input DC voltage and transform it into a square wave. The frequency and duty cycle of the wave can be adjusted modify the value of the components that surround it.

In this case it's adjusted to have a constant 50% duty cycle, while the frequency can be adjusted with the potentiometer.

This square wave, gets sent to a high current transistor, which allows pulses of high current to flow inside the primary coil, and here is where the magic happens. I also added a small 220uF capacitor, it's function is to save some energy while the transistor is off, so it can release current quite fast when it turns on again, this will release some stress from the power supply since the variations of current won't be that high between pulses.

You can charge up a coil with a magnetic field the same way you charge a capacitor with static charges, the difference is when the magnetic field is no longer sustained, it collapses, converting that energy into a voltage spike.

The amplitude of that spike depends on the turn ratio, and since it's quite large we can get big voltage spikes at the output of the transformer. But that voltage is still far away from the voltage we're looking for, so that's why we add a Cockroft-Walton voltage multiplier, which steps up the voltage until its high enough to break air resistance.

You can learn more about voltage multipliers here.

Step 3: The Big List

You will need:

  • 555 timer
  • 8pin chip socket
  • A transformer body and enameled copper wires (winding our own transformer will give us a bit more of power than the 8R:1kOhm audio transformer, also it's kinda difficult to find that specific audio transformer)
  • 220uF capacitor
  • 2x 1k resistor
  • 10x UF4007 or BA159 diodes ( they are faster than the 1N4007, meaning better performance)
  • 10x 10nF* 1, 2 or 3KV ceramic capacitors. I recommend 2kV ones.
  • 2n2222 transistor
  • BD679 or similar high current - fast switching Darlington transistor
  • Small heatsink
  • 10kOhm or 4.7kOhm potentiometer (for more accuracy)
  • IR LED
  • IR light sensor
  • 30 Ohm resistor
  • 2xAA battery pack (it will last longer than just 1AA)
  • Shrink tube (5mm diameter)
  • Two push buttons

Bear in mind you will also need a power source capable of delivering up to 15 volts and at least 1 amp (my circuit needs about 0.60A at full power with a small spark gap).

*EDIT: After many tests and 3 versions of the same circuit I've concluded 1nF caps would be better for this task since the discharge current peak would not be as great as with 10nF, extending the life of the diodes.

Step 4: The Circuit

This circuit will have three differentiable parts, the security part I designed, the frequency tuning or diver part and the high voltage part, I modified the circuit provided by this website to add a bit more of power and to increase the safety of the device (see picture).

You can change the 10k potentiometer for a 4.7k one, that will give you more accuracy when tuning the circuit, but the problem might be the resonant frequency is out of it's limits (See step 9).

You can also modify the value of the 10nF capacitor, by changing it by a 1nF one you can use a 100kOhm pot, or if you want more resolution, as I said before, a 50kOhm pot. The circuit is equivalent.

Step 5: Chosing the Right Transformer.

I couldn't get my hands on a 1k:8R audio transformer, so I decided to replicate it, I though it was going to be a pain, but, in fact,it was quite easy, I re-wound two transformers, and both of them came out successfully with similar results, in fact its hard to tell which of the two performs best, so I ended using the nice looking one.

The first thing you want to look at when choosing a transformer to wind up is the size, the amount of power they can handle is proportional to their size (in 95% of the cases), that said i searched for ones with the size of a large coin (see pictures), that will give us some more power than a small audio transformer.

The first one came out of a big CFL, the second one, the one I'll be using, came out of an old PCB, not sure what it was for, maybe telephonic stuff.

The first thing you want to do is to remove the ferrite core to access the coil, in most of the transformers the two parts are glued togueter, just hold the transformer with a pair of pliers and heat the core with a lighter, careful not to melt the plastic. After a minute or two the glue should melt and you should be able to pull apart the two parts of the core, don't try to force it since ferrite is very brittle and will crack pretty easily, take it with calm.

To wind the primary coil I used enameled copper wire I took from several huge relays, I calculated the width theoretically by measuring the resistance of a 10 meter piece and using an online calculator i was able to obtain a value around 0.15mm.

I calculated the length required to get 8 Ohms, I cut it and winded it.

I was not so careful with the secondary, basically, I wound all the wire I could. I knew from previous results the resistance would be around 300-400 Ohms, and that's what happened, I got a 350 Ohms secondary with the second transformer, 50 Ohms more than the first one.

I don't consider it a failure not to get 1kOhm because the wire width I used for the secondary must be about 0.08mm, somewhat thicker than the ones used with audio transformers, so I had to wind more wire to achieve more resistance, that means the resistance ratio might not be correct, but the turn ratio is pretty similar, and that's what really matters here.

After the two coils are winded you can put the core back again, hold it with tape or glue, it doesn't really matters as ling as it is tight.

NOTE: Transformer winding can be a consuming task, you must have a lot of patience, specially with the primary coil, which must be set up perfectly so the secondary does not appear bulky in some parts, after the primary is winded, you can wind the secondary without caring to much about the disposition, just keep the coil tight, go from one extreme to the other and vice versa and try to maintain a consistent width.

NOTE 2: I've noticed my transformer tends to overheat, and I was wondering about increasing the primary resistance to 20 Ohms or so, this will reduce the heat produced but would also increase the frequency of operation, if you want to do so, reduce the value of the 10kOhm pot to 4.7kOhm for a better frequency adjusting.

TIP: Get a threaded rod and a pair of nuts and washers. Pass the rod though the center of the transformer and with the nuts and washers clamp it in place. Then, attach the rod to a hand drill. This will allow you to wind the coils with ease, but remember to have special care with the primary and to keep a constant tension.

IMPORTANT: Don't touch the transformer while the circuit is running (for example to check the temperature), a plasma arc may be produced between the transformer and your finger, it doesn't hurts, but you can damage the enamel of the coils.

Step 6: About the Optocoupler.

I thought about installing an optocoupler when I accidentally touched the high voltage output while holding the button that powered the circuit. The electrons traveled through my arm and then jumped from my finger to the small pushbutton to get into the positive output of the power supply. Just because you're outputting 15 volts doesn't means the high voltage won't get in there, in fact, it tries to get into anything with lower potential, and that is everything that's not insulating.

If there had been an octocoupler, the electrons would get in my body, but since I can't discharge myself through the button, I would have the same potential as the output, only loosing a small amount of that energy through the air, and that means a much smaller shock, but be careful when touching big metallic structures like windows, since you could drain that charge, and a shock would occur.

This optocoupler will provide you a total independence from the circuit, there will be no electric contact between the pushbutton circuit and the driver and high voltage parts, just a beam of light, that means, if you touch the high voltage output the current can't get back in, so as I said before, a much more smaller shock occurs.

Step 7: ​Making the Optocouppler.

Making the optocoupler is very easy, just grab your IR led and sensor and insert them into the shrink tube, you can use pliers to expand it before inserting them so they fit tighter when you heat up the tube as seen at the second picture.

A black tube will be better to avoid interferences with other IR sources, theoretically, this could be activated with an steady signal emitted from an IR remote control, but the range would be very short.

You might be asking why I included two switches in the schematic, that's because each hand should be touching one at a time in order to activate it safely, it also reduces the risk of accidental turn on. This is a very good practice, since you shouldn't be touching anything else but the buttons. If you want to set up anything do it while the power is off.

I ended installing just one because I didn't had two of them, but it's a good idea to install two.

Step 8: Making the Circuit

The rest of the circuit is quite straight forward I recommend you to test the circuit in a breadboard, but beware of the gaps since there are lots of metallic parts inside a breadboard and if two points with different voltages are too close a spark will jump between them.

When making the Cockroft-Walton voltage multiplier remember to leave enough space between the leads, I don't recommend using a PCB to mount it unless you can leave enough clearance, also, trim the legs of the capacitors in a way no metal parts are overexposed. Trim all the spiky joints since that could lead to corona discharges, which reduce the efficiency a lot.

I highly recommend to isolate all the exposed contacts of the multiplier with thermofusible glue or a similar isolating material and, and after that, wrap it in shrink tube or electrical tape. This will not only reduce the risk of accidental shocks, but it also can increase the efficiency of the circuit by reducing the losses through the air, specially with 5 or more stages.

I also added a piece of foam between the multiplier and the driver.

NOTE: UF4007 diodes are highly recommended, this ones are faster than the 1N4007, meaning your multiplier will be more efficient, I change the 1N4007 diodes of my multiplier by the UF4007 and I instantly noticed about a 40-30% increase in performance.

NOTE2: The resistance of the pot shouldn't be lower than 680 Ohms due to the characteristics of the 555, I recommend you to put a 680 Ohms resistor between the pot and pin 3 to avoid going below that limit.

Step 9: Tuning the Circuit.

Once it's all finished it's time to do the first test, don't worry if it doesn't works, there are many reasons for it to don't work, and the frequency is the first one.

At the right frequency, the reactances of the inductive elements (transformer) and the capacitors will cancel out, producing the largest voltage amplitude to drive the multiplier, producing more sparks between terminals. Sweeping the frequency will let you mentally map out the response curve, and there should be a resonance peak like in the picture below.

The best way to calibrate it is by creating a small spark gap, around 5mm, between the high voltage outputs. Then, by turning the potentiometer you should be able to find the sweet spot, at first some corona hissing may be produced, then sparks may appear, try to find the point where the sparks jump, then carefully move back and forth until the frequency at which sparks are produced is higher. And there you have it, you just adjusted the supply.

In my case, I achieved to tune both transformers with excellent results, the first one, since it has a smaller ferrite core and therefore less inductance, worked both at a frequency of 2.2 kHz and at another one out of the audible range which I have't been able to measure, but I think it must be around 20-30kHz.

The second transformer, since it's bigger and it had a higher inductance, was slower, but that's not an inconvenience, in fact, working at a lower frequency allowed me to tune it more precisely. I made it work with 0.86 kHz and again, with an ultrasonic frequency, which must be around 20-30kHz.

Both transformers showed better performance when working at an ultrasonic frequency, I also could get rid of the annoying hysteresis noise, two birds with one stone.

Step 10: Final Details:

I made a small box to encase the circuit, you can see the process and how I fit the circuit in the pictures, although buying a simple plastic box is better and doesn't requires that amount of work.

Take into account wood becomes mildly conductive with high voltages, so place the high voltage outputs in a safe place. In my case I used a piece of plastic bent in a 90º angle attached to the wood with some screws.

You might think it's quite stupid to put a metal plate on a high voltage supply, but it is completely isolated from the high voltage source and driver thanks to the optocoupler, also, there's an adhesive plastic film behind it.

Step 11: Troubleshooting:

  • It just doesn't works:

Forgetting things is quite a common problem, have you (re)tried that?

Is the frequency adjusted? Don't bring the potentiometer near to zero because that would mean an infinite frequency, you can add a small resistor between the pot and pins 6 and 2, this will avoid you reach a zero resistance. Discard transistor frequency related problems since the BD679 can work up to 1MHz.

How many turns has the primary? and don't tell me ten or twenty please. The primary should have some dozens of turns, around one hundred.

Check the 555 driver, is it working? connect a small speaker to pin 3 and ground with a 1kOhm resistor in between, when you decrease the frequency you should hear sound coming out of it.

555 chips are very sensitive to high voltages, a small spark and they're gone, check if there could be any short or something like that.

  • It worked fine but doesn't works anymore:

Sounds like damaged caps, I know because I had the same problem, I first bought 400v plastic ones, but when I saw lights coming from the inside of them I knew they were dead, buy 1-2kV ceramic ones.

Diodes could suffer damage, you probably wounded a lot of turns at the secondary making it to output more than 1kV, if you have a multimeter capable of measuring up to 1kV DC you can check this. You can reduce the number of turns at the secondary or buy 2kV fast switching diodes.

Also check if you've connected the output of the transformer to the multiplier the other way around.

Check for the 555, as I said before, they're really sensitive to voltage peaks, also reversing the polarity damages them, ST semiconductor and TI work great for me, the ones from Motorola seem to be more unreliable.

  • Everything heats up a lot and it eventually stops working:

Has the primary coil less that 8 Ohms of resistance? If yes, rewind it, use a thin wire of about 0.15mm to get the correct resistance with a relative few amount of turns. You can also increase the resistance of the primary, but take into account that will also rise the frequency of operation.

That might also be because you're running it for long periods of time, with a minimal spark gap. The bigger the spark gap is the better the circuit will work. If it stops working let it cool for a minute or two until the transistor and transformer recover their normal operating conditions, a small heatsink for the transistor is recommended. Also don't short the high voltage output, it makes everything to heat up a lot. A proper frequency tunning is vital to avoid this problem.

Step 12: Extra Content: Franklin's Bell Experiment.

One of the best experiments about electrostatics is the Franklin's bell experiment. As you can see, one can is charged and the other one remains neutral. If the ball isn't charged it will be attracted to the point with more potential, but, once it's charged, it will try to go to the lowest potential, losing it's charge and repeating the cycle again.

As always, thanks for watching, and be safe around high voltages.



    • Backyard Contest

      Backyard Contest
    • Sensors Contest

      Sensors Contest
    • Beauty Tips Contest

      Beauty Tips Contest

    69 Discussions


    4 years ago on Introduction

    You've already very detailed warnings at the very beginning (which is exelent, most people only include them in the bottomline, it not at all) but since safety is a huge deal when working with high voltages I'd like to add a few hints to the construction itself:

    1. PCB Layout:

    When working with low voltages any gap isolates two metal parts, at high voltages you need significant larger gaps to provide good isolation. When working with 230VAC from the grid you should keep a gap at least 5mm (the more the better), at something around 2kV it is recommended (german source) to keep a gap of around 15mm. This mean you have to remove some pads on your board (6 rows = 15mm). If you can, you could also add a cutout.

    2. PCB manufacture:

    After assemby make sure to remove all dirt/ used flux. It has a very high resitance, but that counts at such high voltages.

    3. Wires:

    Some wires are isolated pretty poor, that mean it can spark right through the isolation. If you are not sure about those you have you can isolate them with heatshinking tube, which performs much better.

    4. Component placement:

    Don't stuff everything into the box, keep the isolation in mind. You definitly don't want to touch a button at 2KV due to a fault. (The optocouplers only help if used correctly)

    5. Do not breadboard:

    Breadboarding seems like a smart idea, you can check if everything works before assembly. But if you don't know the exact specifications about max. voltage it can handle at least I'd avoid that. You possible ignore me anyway, so stay at least at a reasonable distance.

    6. Acrylic glass...

    isolates very well. Consider to use it for the top plate and/ or the case.

    7. Overvoltage protection

    In the case that the isolation breaks down you want to make sure the high voltage doesn't get in touch with the switch or the input terminals. This can be done by adding a transient voltage suppressor diodes across the input and the switch (such as these, about 0,30€ ea.). If the voltage extends the rated voltage at any time it gets conductive and thus limits the voltage.You should note that this is not the perfect solution, but a good starting point.

    Please, don't get me wrong, you did a good job and for you everything went perfect, but with some bad luck this may be risky. This is not critisism to your project, it's rather a safety advice to anybody who wants to attempt this.

    Thumbs up for you and fingers crossed for everybody who wants to build this. Good luck :)

    1 reply

    Reply 4 years ago on Introduction

    Thanks for the advices, every contribution to the safety of any maker counts. I featured your post so everyone can see it.

    I'll edit my instructable to add some of those points.


    3 years ago

    what is the current on this how would you make a 70V 200A DC power supply for a zvs driver for a induction heater with that input I think I should get around 14KW

    1 reply

    Reply 3 years ago

    This circuit doens't uses more than 10W of power, I don't know why you've asked me about building a 14kW ZVS driven induction heater. Googleing it would be a good point to start since I haave no idea on how to build an induction heater with such a massive power.


    3 years ago

    what is the current on this how would you make a 70V 200A DC power supply for a zvs driver for a induction heater with that input I think I should get around 14KW


    3 years ago

    What if I want to start with a 2kV microwave oven transformer and get out 9kV to power a Ozone Generator? Thanks

    1 reply

    Reply 3 years ago

    A microwave oven transformer works with hundred or thousand times the current this circuit does, I don't really know how I would proceed in order to step up the voltage in that case, you might still be able to use a Cockroft-Walton multiplier with bigger capacitors and diodes rated more than 2kV, but again, I'm not sure. I haven't worked with microwave oven transformers, they are scary stuff.


    3 years ago on Introduction

    Does this work with these TV transformers? I got 3 now (from a printer, one seems to have a voltage doubler)

    1 reply

    Reply 3 years ago on Introduction

    You mean flyback transformers? No, although the driver circuit is extremely similar there are some important variations, I recommend you to check my instructable on a 50kV PSU using one of those transformers.

    The one from the printer might work with lower power, so this circuit might be useful to power it, but still I would recommend changing the BD679 darlington transistor by a more powerful one.

    Arduino Tech

    4 years ago

    Can I use a ldr and an led instead of the optocoupler and can I use a 12v 5amp power supply ??

    1 reply
    Victor805Arduino Tech

    Reply 4 years ago on Introduction

    Yes, the LDR should work just fine, just a little slower, but you won't tell the difference, just make sure light can't get in by accident.


    4 years ago

    Can I use a 12 volt car battery on this circuit or do I need to make modifications

    1 reply

    Reply 4 years ago on Introduction

    Of course, 12 volts is perfect, the circuit should work with any voltage between 5 and 18 volts, but I recommend powering it with a voltage from 9 to 14 volts, anything below 9 volts would make the output much lower and if you use anything above 14 volts you risk frying components of the multiplier.


    Reply 4 years ago on Introduction

    I would say around 20€, the most expensive parts were the capacitors and the diodes because I bought them at an electronics store, and it was more expensive than get those components online.


    4 years ago

    Could u use a phototransistor and laser diode


    4 years ago

    have you thought about putting this all inside of a RadioShack project box, sealing it and then filling the entire thing with mineral oil? it would greatly reduce internal arcing, and does an astonishing job keeping everything cool, even the transformer once the oil has seeped into everything.

    1 reply

    Reply 4 years ago

    I know mineral oil is a great way to isolate the components and to cool them down, but I think it's a bit overkill for a voltage supply like this one, maybe if it was powered by a TV flyback transformer I would consider doing that, but so far there are no signs or arching or any similar losses in the circuit. Also it's quite messy, so if I wanted to make some modifications I would have to clean it to do so.