As i have a passion for both Tesla coils and audio circuitry,
for a project i decided to combine the both to make a plasma speaker.
On the internet there were several circuits to choose from which i all tried out.
They didn’t really satisfy my expectations.
There were several implementations but they were all missing something.
They were either missing the bass sound and/or hiss due to the arc’s side band production.
So basically i had 2 challenges to overcome to make a plasma speaker that satisfied my needs.
1. Get rid of that annoying hiss sound.
2. Find a way to display reasonable bass.
In my attempt to get such a high frequency that side bands begin being outside of the audio spectrum,
I stumbled upon a class e type of amplifier circuit that was running a small Tesla coil.
So that got me started building which you can read about on the class e Tesla coil page on this webpage.
Since the Tesla coil design used to create the arc runs on a DC voltage from 24v up to 60v,
i decided to go with a buck converter topology which led me to a Class D type of amplifier.
In Class D type of amplification a signal is modulated into the pulse width of a fixed frequency square wave.
This modulation was achieved by comparing a triangle wave of freq. x with the signal of audio.
This means the amplitude of the signal will actively alter the on and off time of the square wave which as passed through a filter creating a controllable output voltage from a DC input voltage.
So what do i need to use this type of modulation in my circuit?
1. A fixed frequency triangle wave
2. A audio signal
3. A buck converter switching topology
Once i had the block schematic down on paper,
things were going to be allot easier as i now got a clear image of whats needed to output sweet audio sounds
So to get a triangle waveform i used a simple schmitt trigger inverter oscillator.
So to pick a frequency we would like for sample rate of the class D PWM wave we should look at a few aspects.
1. The highest frequency you want to display
2. The amount of samples taken from the highest frequenc
For starters i picked a frequency of 80khz to get 4 alias points at the highest frequency which is 20khz.
So a rough estimate on what kind of components we need for the RC network was calculated by using this formula.
f = 1.2 / ( r * c )
f being the outputted frequency.
r being the total RC resistance.
c being the total RC capacitance.
So to make things easy i selected a 3.3nF cap to get me well in range of the selected PWM frequency.
Then i used the reordered formula to calculate the needed resistance for RC network.
r = ( 1.2 / f ) / c
r = ( 1.2 / 80000 ) / 3.3e-9
r = 4545 ohms
So after we set the triangle oscillator to do its job we start to amplify it to a higher amplitude to get some bigger range of tuning into.
I made a simple pre-amplifier made of 2 OPAMP ic’s fast enough to go with the 80khz frequency.
A lowpass filter was applied into the feedback network of the amplfier stage to filter out unwanted EMI
For the audio stage i decided to get my audio source from a cheap china Bluetooth receiver module.
This to get easy access from allot of different devices without the troubles that may occur when using cables.
Those cables can surely act as antenna’s and pickup unwanted EMI screwing with the audio modulator.
Or even worse maybe even fry that laptop you hooked up to the device.
So this took care of getting the signal onto the modulator board,
We will amplify the audio signal the exact same way as we did before with the triangle wave.
As said before we don’t want unwanted high frequency noise to end up into the modulator.
We also use a low-pass filter in the audio feedback network.
This time set at a much lower frequency as we are only interested in the 20-20k band for audio
Since we are now able to control both the amplitude of the audio wave and the amplitude of the triangle wave.
We can start matching them up
So to make complicated stuff fairly simple we fix 1 input at the center of the 5v rail.
At 2.5volt to get a maximum voltage swing between the 0v and 5v input the tl3016 comparator can take.
I selected the higher frequency triangle wave for the fixed position as that is our carrier and must be as stable as practically possible
The audio signal however we want kind of more control over as to set a virtual center line inside the PWM wave.
Therefor we add a potentiometer to dc bias the audio going into the comparator.
This would provide great control for tuning the PWM to the final stage of the modulator.
I put this all on a pcb i designed and this low voltage part of the modulator began to start looking really nice.
Now that we got the actual PWM needed to control the buck converter we can start looking at the power stage of the design.
A Buck converter is a topology used to step down a voltage by charging a cap through a coil. (the low pass filter)
Usually used for lower voltages as this topology needs a switch on the hot side of the circuit which usually ends up with lots of parts and a somewhat complicated gate drive system.
Assuming components behave linear and you selected the right filter with the right frequency you can determine the output voltage of the buck voltage by taking the output voltage and multiplying it by the percentage.
Say you have 100 volts and need 10 volts to be put out.
With ideal components this would mean your in need of a 10% on duty cycle to reach that 10% of a 100 volt.
In the real world it’s far from this.
For calculating the actual low pass filter(1st order) i used the folowing formula.
fc = 1 / ( 2 * π * √ (L * C) )
As i pre picked a capacitor of 1.3uF we still need to figure out the inductance needed to get a low pass filter to pass everything under 20khz.
L = 1 / ( 4 * π² * f² * C )
L = 1 / ( 4 * π² *2e+4² * 1.3e-6 )
L = 48.7 uH
This filter will yield a characteristic impedance of
Z = √ (L / C)
Z = √ (48.7 / 1.3)
Z = 6.12 ohms
6.12 ohms impedance should gives us plenty of available power on the output of the amplifier.
And is well in range of the impedance of the small coil i am using as a load.
As mentioned before the buck converter topology consists of a high side switching device which is different to drive compared to low side switches
There are plenty of ic’s available that work with a bootstrap method of switching.
However i decided to go with a 2nd transformer just to supply the gate drive circuit.
For the ic i used a FOD3184 because of there easy connections and 3A gate drive output which should be plenty peak currents to drive a IRFP260N.
Once i completed the pcb for the buck converter topology i quickly ordered some boards to solder up.
It looked great!
Now to start testing the ouput with a big 200w 6.8 ohms resistor.
But found out pretty soon there were some complications with the snubber capacitor which had to be fixed.
There was a big ringing on the output of the buck converter which i had to dig into.
Allthough it was still well within the limits of the mosfet i decided to find a solution which led me to a RCD snubber network.
With components i had laying around i was able to reduce the ringing quite a bit.
For now it is acceptable.
Which made it time to hook up a load to the output of the circuit.
The first output of the driver was terrible, but after some tuning of the potentiometers on the modulator it started to sound really nice.
Now all that’s left is to design a case for the completed unit.
To be continued.