How to Make a Musical Solid State Tesla Coil (SSTC) That Plays Guitar!

This is my first Solid State Tesla Coil (SSTC), which I think has turned out pretty well! My intention was more towards it playing music rather than huge sparks, but I got about 6in sparks from it too which is a bonus.

The aim of this instructable is to explain what Tesla Coils (TC) are, how they work, a couple of variations and and also how to make one! To make a TC you really need to know exactly how they work (you can't build one successfully by blindly following instructions). I did a lot (and i mean A LOT) of research into how to make one, and I feel like I should give back some information that I discovered whilst making mine.

I first want to say that you'll be playing with very high voltages (and high current!), and this of course is/can be very dangerous! Always carry out the correct safety procedures when dealing with high voltage and high current circuitry.

Secondly, Tesla Coils take a lot of work, it's unusual for a new Tesla Coil to work 100% on it's first light (first light is a term given to a Tesla Coil when it first turns on). So stick with it!

Step 1: What Is a Tesla Coil?

Tesla Coils were invented (as the name would suggest) by Nikola Tesla in the 1890's. Tesla did not want to create music but rather transmit electricity wirelessly - and to some extent he did! There are rumours that he turned on 100 light bulbs 26 miles away using Tesla Coils! However, there isn't enough evidence to support this claim, but it would be an awesome idea!

Essentially, a Tesla Coil is a high voltage, step up transformer. They take in relatively low voltages and step them up to hundreds of thousands of volts. But I've got a transformer at home and it looks nothing like a Tesla Coil I hear you say? Well you're right, mobile phones chargers, laptop chargers, Xbox's they all have transformers! But there are different types. Your laptop charger will more than likely be a switching mode power supply - which is basically a very efficient transformer (it probably falls under the category of rectifiers but still) and old mobile phone chargers are normally just a small step down transformer. The reason they weigh so much is that there's a fat lump of iron in them!

The issue with these transformers is inherent with the design. A conventional transformer has a ferrite core with copper wire wrapped around either ends (see picture). When you pass a current through a coil of wire, you get what's called electromagnetic induction. Basically, you generate a magnetic field. Likewise, if you pass a magnetic field through a coil of wire then you induce a current! So, stick a ferrite core in the middle to conduct the flow of this magnetic field we're generating and there you have it, a transformer! The issue with this is that you have got a big piece of metal that also conducts electricity really well next to some wire with a lot of voltage, if the voltage is high enough it'll short through the ferrite core; not good. Alas, this is where Tesla Coils come in.

Tesla Coils are loosely coupled transformers, there's no ferrite core to conduct the magnetism (and so conventional transformers are tightly coupled). This means that there are two coils of wire in fairly close proximity to one and other but with an air gap between them. Air is a very poor conductor of electricity, but it's also not brilliant for conducting magnetism and so it takes a few oscillations to do it. I'll explain this more in the next step.

Step 2: Spark Gap and Solid State Tesla Coils (SGTC & SSTC)

There are a lot of different Tesla Coils out there, but for the scope of this I'm going to talk about 2 - Spark Gap and Solid State.

As far as the design goes, Tesla Coils are fairly simple. If we look at the SGTC (the first image), it consists of a high voltage step up transformer, a spark gap, a capacitor and two coils. As I mentioned earlier, we want to generate an electromagnetic field to induce a voltage in the secondary winding's. This voltage is going to be stored on the top load (usually some metallic object, to act as a capacitor) and eventually we'll pass so much electrical energy to the top load it'll ionize the air around it (basically tearing the electrons from the atoms of oxygen/nitrogen etc. causing a flow of charge) and produce a spark!

The more current we pass through the primary coil (and the quicker it flows), the greater the magnetic field we produce and hence the larger voltage we induce in the secondary. This is the reason for the gap in the electrical circuit. We charge up the high voltage capacitor (using the high voltage transformer, this steps up mains voltage to around 30kV). As the capacitor is charging up the potential difference (or voltage) at the top of the spark gap gets higher and higher. Much like the voltage on the top load, when it's sufficiently high enough it will ionize the air between it, then short the circuit! The gap basically acts as a switch, and when it's 'closed' the capacitor rapidly discharges causing a HUGE current to flow through the primary.

As aforementioned the spark gap acts as a 'switch', you can adjust how often it 'turns' on by changing the distance between the two points. If they're closer together then the time to charge the capacitor to the breakdown voltage is less, hence the we energize the primary coil more often (I believe a general rule of thumb is that 30KV is needed to bridge a gap of 1cm). Conversely, if we make the distance greater it takes longer to charge the capacitor, ergo, we energize the primary less often, but when it does break down the spark gap we get a larger flow of current. But we now live in a age of semiconductors, so it seems fitting to replace the spark gap with solid state switching devices - and this leads us onto SSTC (Solid State Tesla Coils). The circuit it very much the same (the circuit design is shown in later steps and thoroughly explained) but we need to add in some switches. These can be MOSFETs or IGBTs - it's usually the latter due to their high current handling properties. A high level abstract of myMusicalTesla is shown in the image above. We have three main components - the rectifier (to change the AC mains to DC) the switching circuitry (to energize the primary coil) and the interrupter (this is used to create music, more on this later). Introducing these switches also means that we no longer need such high voltages - because we don't need to bridge an air gap. This reduces the size and maybe more importantly, the cost.

So far it seems fairly simple, but the complexity comes in when we see that this is an LRC (inductor - resistor - capacitor) circuit, loosely coupled with another LRC circuit. Because it's loosely coupled not all the energy is passed per cycle. What's more is we want to be able to pass as much energy per cycle as possible - and this can be done by energizing the primary coil at the resonant frequency of the secondary. Before I explain how we calculate the resonant frequency (see the step regarding the secondary coil), I'll explain what resonance actually is.

I would highly recommend visiting Richie Burnett site - he explains in a lot more detail how Tesla Coils work and he really gets down and dirty with the math - he also includes a lot of 'scope traces to further back up the theory, a must read for Tesla Coil enthusiasts!

Step 3: What Is Resonance Frequency?

Resonance is key to building a Tesla Coil that makes large sparks - without it you'll be lucky to see anything (although you may still hear it!). So what is it? Resonance is a natural phenomenon that basically means you can provide energy at a certain period (or frequency) to something that's currently oscillating and even though you're providing the same energy each time, the amplitude of the oscillation gets higher and higher. The best example I can give of this is with a swing, When you push someone on a swing you normally wait until they have swung all the way back and briefly stopping until they continue there swing and forwards - this is the point you push them. You're providing energy to the system at regular intervals and the person is getting higher and higher. To further this analogy (and tailor it slightly more to Tesla Coils) imagine that when the person on the swing reaches a certain height they jump off and land on some grass somewhere. We can't give one single push straight away for them to reach this height, we're just not that strong! But after a few pushes (or oscillations) they're high enough to jump! This is how Tesla coils work, we run them at the resonant frequency to top the voltage up on the top load quickly and with little effort until it becomes so high it ionizes the air and causes sparks! That's pretty much it!.

Continuing with this example, if we push them when they're just after the halfway point of they're swing, we going to need a lot of force and all we'll do is slow them down.This is similar to you running your Tesla Coil out of resonance and you're likely to draw too much current and blow something in the circuit.

Step 4: The Circuit

I should mention here that my design is slightly different to most in the sense that I haven't used a normal micro-controller, I've used the National Instruments myRIO - a great piece of kit that allows me to control the Tesla Coil wirelessly (due to a built in wireless adapter) and have deterministic switching using it's real-time operating system and on-board FPGA. This meant I could have a pretty cool user interface and perform FFTs on an audio signal to obtain it's frequency - I understand not everyone has access to this so I'd advise something like the STM32F4 Discovery board as an alternative. It's cheap (around £10), it has an on-board audio jack like the myRIO and can be programmed in embedded C. If you're interested in the LabVIEW code and how that works, check out my link here https://decibel.ni.com/content/docs/DOC-40979.

Now, lets talk circuits! Please note my design was based off Steve Wards (see the final image), I made a few alterations to this to suit my needs. Please check out Steve's site, he's a member of the infamous ArkAttack and he really knows his stuff!

The first image shows you the whole system, as I mentioned before there are 3 main parts:

1) Switching circuitry - This energizes the primary coil by switching mains voltage across it.

2) Rectification - Since we want to be switching with DC voltages and mains voltage is 240V RMS (I'm in the UK) then we rectify it to DC, ~340VDC.

3) Interrupter - Although you can run SSTC in what's called "continuous mode" I would advise you against is. When you do this you simply turn the Tesla Coil on/off at it's resonant frequency (usually in the order of 100Khz) and since IGBTs/MOSFETs loose power as heat when conducting, they get really hot! An interrupter does what it says on the tin and simply interrupts the signal, giving the TC a rest. You can also incorporate PWM (pulse width modulation) to keep them IGBTs running cooler.

Since I made my circuit very small (10cm/10cm) I sent the Gerber files off to a company to have some PCBs made - I got 10 for £26 which I thought was great. This makes it a lot easier to solder together, I did make some prototypes on strip board but after the 3rd one exploding I decided against it :P

In the next step I'll go into more detail about the circuits core components.

Step 5: Switching Circuitry - SSTC Half Bridge

As far as I'm aware all of the SSTC use the same idea of either a half bridge or full bridge of transistors to switch the current through the primary coil. I used a half bridge (because it's cheaper, you only need two IGBTs rather than 4) and so I'll talk about that one instead of the full bridge.

From the image you can see that the rectified DC voltage is on the top rail and we have our two IGBTs and finally a terminal block where we connect our primary coil. As well as this we have two high voltage PP capacitors. It took me quite some time to understand/discover why these are actually needed. It seems that we need to hold one end of the primary coil between the two rails so that we can pass the current through it either way when we're switching. Of course, if we just connected it without these capacitors, we would short the power rails. Hence, the capacitors simply act as a DC decoupler. But it comes at a price, because of this we end up with a potential divider and half the available voltage we can switch across the primary. Most people (I believe in the US) incorporate a voltage doubler circuit, but I didn't feel the need since the US use 120V, half of the UK already, therefore I would be switching the same voltage as them without a doubler circuit.

See the hardware list at the end for the components I used, really the IGBTs you use don't matter, so long as the peak pulsed current is high enough and the voltage is high enough. Usually, the voltage is the easy one - the higher the current you want the more you'll pay!

It's note worthy to mention that most IGBTs have internal flyback diodes, but I (and many others) have added in some dedicated flyback diodes, in parallel to the IGBTs. These are very important for the TC and I'll explain why.

When we switch the two IGBTs (I will explain this in the next step) we do so in compliment - IE when the top IGBT is on, then the bottom one is off. It's no surprise that these don't instantly switch, they take a finite amount of time. There's now 3 states that the switches can be in; both on, both off, or complimentary of each other. We always want to know the state of the switches, and so you never switch them both at the same time and so we introduce what's called dead-time. The reason for this becomes clear when we think about the 3 different states - if the switches are both on, we're shorting VCC through the IGBTs, drawing a LOT of current (I=VCC/R where R is the resistance of the IGBTs and the coil, so not a lot). This results in dead IGBTs (trust me, I've got a bag of them!) and sods law has it that when one fails, it fails short and kills the other one! So option 1 is a no-no. Option 2 is to have them both off, well if we do that we can't short out the supply voltage right? True, but we've also got an inductive component that's storing a lot of energy and now has no path to ground. Eventually it'll find a path to ground which usually results in a spark which can go anywhere and damage anything. Another no-no. Finally, option 3, switch them at exactly the same time and hope for the best - hmmm maybe not. What do we do!? The answer is, option 2 BUT, we add in flyback diodes to give that energy stored in the coil a path to ground. So those diodes are very important!

Step 6: Switching Circuitry - IR2110

Since I used the IR2110 high/lows side MOSFET/IGBT driver at university, it seemed fitting to use it here since I understand it quite well. The reason we need this is that most micro-controllers have 3v logic - that is they can only provide 3v when their outputs are high. This isn't nearly enough to fully switch the gates on the IGBTs, nor do they supply anywhere near enough current to turn them on quickly. With IGBTs they only use power when they switch, hence the longer it takes to switch the more power they use (P=I^2*rDS(on)). With my design I don't need very large heat-sinks because they don't get very hot due to the speed in which they switch.

I found this link here that explains the IR2110 extremely well and why you need the bootstrap capacitors. Essentially you provide your switching signal to the input of the IR2110 - this comes from the uC (micro-controller) or in my case, the myRIO. It then steps up the voltage and essentially amplifies the current. What's great about this chip is that it has a shut-down pin, meaning we can control when the spark occurs. In the second image you can see the response from the IGBTs switching, there is a fair amount of ringing on the signals, this is the reason for the 5ohm resistors on the outputs of the IR2110, it reduces the ringing but it does increase the time taken to switch - so it's a bit of a trade-off.

What happens is we produce a two square waves (at the resonant frequency of the TC) 180 degrees out of phase of each other and input them into the corresponding high side or low side input. This will then amplify the voltage and current and switch the IGBTs nanoseconds later and hence energize the primary coil. if we didn't turn the output on and off the TC would run in continuous mode and we would get a flame like spark (see picture) and this doesn't give the circuit a rest and so the IGBTs get really hot. This is also what allows us to play music (more on that later).

Step 7: Rectification - Full Bridge Rectifier

As I mentioned before, we need a DC voltage to energize the coil. But we also need a source of high voltage and high current, and the mains voltage is a readily available source. But as most of you know, mains is AC. Therefore we have a simple circuit to change the voltage into DC - it's called a full bridge rectifier. You can make one of these with 4 diodes, but it's easier to just order a semi conductor package. Just make sure it's rated for 400V or more and 8+A.

Step 8: MUSIC!

Now we're getting on to the cool stuff! We know we can interrupt the signal to the TC . Using this knowledge it makes sense that if we interrupt the signal 100 times a second we get a 100Hz tone! Pretty simple, but during my research of musical TC I was under the impression you switched the Tesla Coil at the frequency of the tone you want to make - but of course this means it will be well below the resonant frequency at you get a tiny spark!

So what I found out is you incorporate what's called pulse repetition frequency modulation. This means you produce a square wave at regular intervals. I feel this is best explained with images, so if you look at the first image you'll see we have two waves (not to scale). One that's our resonant frequency (for the coil you will make, it'll be around 115kHz if you follow my instructions) and this is almost 10 times above the audible spectrum of humans (who can hear anything from 20-20kHz). We then have out audible tone (for example 100Hz). We essentially logically AND this two waves and get the output shown in the second image. And there you have it, the spark will occur 100 times a second!

But why does this make music? With conventional speakers they produce sound waves by forcing the air up and down with a speaker cone - thus creating longitudinal waves at a desired frequency. With the TC, the spark heats the surrounding air causing it to expand, and when the spark decays and stops, the air cools. As it cools, it contracts. This expanding and contracting of air creates the longitudinal waves much like a loud speaker. The only difference is that the loud speaker can add in different harmonics - IE sine waves on top of sine waves. We're using square waves and so we can't do this. This results in a low fidelity sound.

Step 9: Lets Start Building! the Secondary Coil

Lets start with the one component of a Tesla Coil that takes the longest time to make. The secondary needs to be wound full of enameled wire. Mine has about 1000 turns on a 4.5in PVC tube (the same stuff that's used for drainage pipes). I wound about 25cm up the tube, as this gave me a resonant frequency of 115kHz which is about the range I wanted. Any lower and I need a bigger TC (which wasn't what I intended to make) and any more you have to start switching at very high frequencies which has a whole host of other problems.

You need enameled wire because otherwise the wires would short (since they're touching each other!) so ensure you buy the correct one. I got mine from eBay, but I bought 2 lots of 250g and had to solder the wire in the middle because I ran out - so buy around 400g and you should be good.

I also learnt that it's best to tape down the winding's you've already done after about 4cm, because if you drop the roll you're going to have a bad time :P

Final point to mention, I would leave a lot of wire at either end - one end needs to connect to your top load and the other needs to be connected to a grounding post - so leave yourself plenty of room. It's also a good idea to varnish/lacquer your secondary afterwards to stop the turns of copper sliding over each other. It also look nice and shiney!

Finally, to make things nice and easy, Bart has written an application to calculate the resonant frequency of a TC if you give it all the dimensions and such (called JavaTC) - I would defiantly recommend using this as a bench mark. It won't be 100% accurate but it's definitely within the right range. It also extremely useful for calculating how much wire you're going to need, and if you're building a DRSSTC, what primary capacitor you need to achieve resonance.

I'd like to mention here to (because again, I struggled to find this simple bit of information that's usually skipped over in these how to's) that the secondary's top wire connects to your top load and the bottom wire connects to your ground pole Not ground itself. I have read that this isn't a good idea as the voltage builds up on the earth wire of your mains and can spark and cause a fire. So find something metal, like a pole (or bed post, but note it does pit the metal - not that I did that...) and connect it to there. This will give you nice long sparks!

Step 10: The Primary Coil

As we've mentioned before the primary coil is needed to generate the magnetic field to induce the voltage in the secondary winding's. What I wasn't aware of until after making myMusicalTesla is that the coupling coefficent between the two coils should be between 0.13 - 0.22. Barts JavaTC application calculates this for you (which is why it's so damn useful!) and if you click the subheadings of each section, it normally explains a little more about it. For example, this is what happens if you click the 'Coupling Coefficient' header in JavaTC - http://www.tesla.nu/programs/javatc/script/out24.html. I'd read through it, it explains the coupling coeffienct extremely well.

With my TC, I wrapped some speaker wire (because it's fairly think ~16AWG) around the base of the secondary and taped it into place with black electrical tape. I know understand that I shouldn't have coupled the primary so closely to the secondary, if you look at the output of my JavaTC run, you'll see my k value is far too high.

The great thing with JavaTC is that you can play around with the values until you see what you want, for example if I change the radius of my primary to 15cm (instead of 6cm, the same as my secondary diameter) then achieve a 'k' value of 0.156. I'm going to make another TC (a slightly larger one) and I'll be sure to make this change and hopefully the sparks will be better!

One thing I'd like to mention here is how confused I was with the 'Primary Capacitor' value on JavaTC. This is only for DRSSTC, but you will get an output in JavaTC nonetheless saying it's 100% detuned - don't worry about that. If you see caps in SSTC designs, it's likely that they're only DC blocking caps, unless they're class e SSTC. This link here explains the whole scenario :)

Step 11: The Top Load

The top load is pretty easy, it just needs to be something fairly smooth and round. Torus (or doughnut) shaped top loads are usually chosen but spheres are also used. The torus shape helps to shape the electric field that's generated. Spheres don't work as well because you still have to mount it at the bottom. With the torus, you can mount the secondary coil in the middle, where it's fairly well shielded - thank you smlizz, member of ArkAttack for explaining why the torus shape helps. As long as they're metal (meaning they can store the voltage) and they're smooth (so that they can build up enough voltage before sparking) then you're good to go. What is normally done is to add a break out point on the top load, this creates an area of high field stress, to break down the air (at atmospheric pressures) you need a field strength of ~3x10^6V/M (again, thank you smlizz for correcting me - I thought it was an increase in the concentration of electrons due to adding a metallic object with a large surface area!). It means you can then direct the spark to your grounding pole - this increases the spark length. See here for some more information on electric field strengths - http://en.wikipedia.org/wiki/Electrical_breakdown#Mechanism

I made my top load with two polystyrene half doughnuts glues to a paint lid and covered in silver tape (you can also use aluminium foil). After making it it seemed that it didn't have enough capacitance and so I bought a 50mm aluminium ducting (1M) and curled it back on itself and taped this to the top. This worked well, I added a break point (a screw driver) and all was good!

Step 12: The PCB

Attached are the Gerber files for my PCB - I would advise making your own and tailoring it to your specific needs - but it's here nonetheless. All you need to do is simply solder it up, if you don't want to use a PCB then strip board works fine just the same, although it's a little messier!

Step 13: The Final Design

My LabVIEW code is attached, I haven't written any C for this since LabVIEW was much quicker and easier. My design will work with the STM32 (or simular controller) with a few adjustments, but the reason for this instructiable is to give you the tools and knowledge to make your own - not blindly make one from some instructions because it doesn't work with TCs! You know, teach a man to fish and all that!

As I mentioned before, I now want to build a slightly larger TC, and make some changes to try and increase spark length. I also want to change the code to allow for better FFT resolution at higher sample rates. I'm also aware that these guys managed to make a polytonal Tesla Coil! Which amazed me and I'd love to know how they did it, so if you have any ideas - please comment!

Step 14: Components!

Since you've read this far, you deserve the hardware list! Enjoy!

And thanks for reading! Comment and Favorite!