OK, 44KV is cool and dandy, but for a Tesla coil the 50 Hz is way too low.
To increase this frequency we use a spark gap oscillator.
The spark-gap oscillator consists of 3 parts; the primary coil, a capacitor and of course a spark-gap.
Someone once asked me this question: If I have 2 identical capacitors, one charged and one empty, and I connect them in parallel so that the charge gets evenly distributed over both, the voltage will be half of the original voltage, because the voltage is linear dependent on the charge. But that means that the energy in each capacitor is one quarter of the original energy, because this energy is linear dependant on the square of the voltage. So the total energy in the end is half the original energy.
Where did this energy go?
Well, the problem here is that what you think would happen does not actually happen. To connect these capacitors you will need a piece of wire (= inductance and resistance) and when you connect them the charge of the full capacitor will flow to the empty one until they are both half-full. But that is not the end of the story, because of the inductance of the connecting wire the current will continue to flow until ALL of the charge is in the empty capacitor (which is then full, of course).
And from there it will start to flow back, again until all charge has been returned. And this process would continue indefinitely if it wasn’t for the resistance of the wire. This resistance makes that each time some energy gets radiated as electromagnetic waves and some gets converted into heat.
And so that is where the energy goes.
A spark-gap oscillator works on the same principle, but instead of 2 capacitors we have only one of which each time the polarity changes. By shorting a charged capacitor, the charge gets evenly distributed over its plates, but then the inductance of the connecting wire kicks in and transfers the remaining charge as well leaving the capacitor oppositely charged instead of discharged, and so the process repeats.
This way we get a chunk of charge that is moving very rapidly back and forth through an inductor between the two plates of a capacitor.
Normally a spark-gap oscillator consists of these three parts all in series. Then the capacitor is charged by a suitable voltage source until the spark-gap is triggered and it can discharge through the inductor (primary coil).
This is the simplest form of a spark-gap oscillator, but we use a slightly advanced version. We charge another two capacitors and when their voltage is high enough the spark-gap is triggered which causes a voltage reversal across our primary circuit (consisting of our primary capacitor and primary coil). Because this happens extremely sudden, the primary coil blocks this current and we charge our primary capacitor. And consequently an oscillation is started in our primary circuit.
The advantage of this method is that the spark-gap is no longer part of the oscillating (primary) circuit. Thus we have removed a large portion of the resistance from this circuit allowing it to oscillate much longer.
In this post I also briefly talk about this advanced oscillator.
So with this thorough understanding that we now have, let’s turn to the parts.
The Primary Coil
Our primary coil consists of 10 strands, each 25 mm² copper, making just 1 turn.
So we have a total of 250 mm² copper in our primary winding. That seems a bit excessive but I want to stay close to anything Tesla wrote, whether or not I fully understand or even agree with it. If we want to prove his theories we must follow his lead.
As you can imagine the resistance of this primary coil is very low and its inductance is little over 2 μH as is measured here.
Tesla wrote that in order to have a most efficient transformer, the amount of copper used in the primary should equal the amount used in the secondary.
I can think of one logical reason why that has to be true: In a transformer the product of voltage and current has to be the same in both coils. Now, the voltage is determined by the number of turns and the current by the thickness (actually it is the area) of the wires used. In an average transformer the turns are of equal length, so if the product of turns x wire area is also the same, then there is a same amount of copper in the primary and secondary coils.
Although this makes perfect sense, I have a feeling that this is not what Tesla meant when he said this. Anyway, to avoid endless discussions, as I said we should just do as he said.
The Capacitor Bank
Our capacitor bank is made up of strings of 17 polypropylene capacitors, each good for 3,000V.
In the primary capacitor we use 2 of such strings in series and 10 in parallel. Giving a capacitor of 44.1 nF and 102,000 V.
As extra capacitors we use 20 of these strings in parallel, resulting in 2 capacitors of 176.4 nF and 51,000 V. These two are used in series, so they are actually 88.2 nF and 102 KV.
In total our bank consists of 1,020 polypropylene capacitors.
I have chosen the Cornell Dubilier 940 series because in that series there is a 3 KV version. With 1,440 V/μs this capacitor is a bit slow compared to the Tesla Coiler’s favorite 942 series which does 2,879 V/μs. Though this looks a bit poor at first, since we have 34 in series and each of them does 1,440 V/μs, their combined speed is 48,960 V/μs which is good enough.
The Spark-Gap
For a 44KV spark-gap we go for a multiple-gap-in-series approach, and with a stepper motor and arduino controller I made it in such way that the gap can be adjusted to any required size.
In the back of this picture you see a quite important detail of this spark-gap; a fan.
The powerful discharges in this gap produce a hot plasma between the electrodes. Without the fan this plasma would cause a permanent short of our high voltage source and consequently a poor performance of the coil.
With the fan the gap is extinguished as soon as there is not enough current flowing through it.