Project Overview
Main references for the project taken from the following article with details about a high frequency musical tesla coil '
(https://www.instructables.com/Making-the-Ultimate-Class-E-Musical-Tesla-Coil-Sin/)
The project concerns ourselves with a high frequency musical tesla coil. To understand what this entails, we can decipher this in parts. Specifically, the "tesla coil", "high frequency" and "musical" aspects.
A tesla coil is a resonant transformer used to generate high voltages that are high enough to breakdown air. In addition to being a transformer, the tesla coil takes advantage of resosnance to reach even higher voltages then what is capable with multiplicative coiling ratios of the primary and secondary coils.
Then, a "high frequency" tesla coil refers to the fact that we hope to achieve this resonance at high frequencies in the megahertz range. This is a particular challenge since megahertz frequencies will require special construction techniques to reduce stray capacitances that radio frequency systems sensitive to.
Lastly, the "musical" aspect of the tesla coil comes about by turning the tesla coil on and off at audible frequencies. Though the arcs themselves are operating at well above human hearing range, by playing with the lengths of time that arcs are on and off, we can then generate disruptions in the air that are in the human hearing range. We can modulate how quickly the coil turns on and off to change the pitch we hear and we can change how long each on and off period lasts to change the loudness.
Step-by-Step Instructions
The first step, in no particular order is winding the secondary coil once you have a good estimate of the frequency you want to target.
A popular online tesla coil calculator tool can be used to get an estimation of the resonant frequencies.
https://www.tesla.nu/programs/javatc/javatc.html
This is one way we can target a particular operating frequency of a tesla coil. One can also calculate the inductances oneself but at the end of the day you will also need to empirically measure the resonant frequencies. The online tool is just an estimate.
For this particular instance of the project, 172 turns of 19 awg enamel copper wire was wrapped onto a 2.11cm radius PVC pipe. Which gave an estimate of around 5.6MHz. Once the coil was wound, a frequency sweep was conducted on the coil to empirically find the resonant frequency and it changes depending on the kinds of objects near the secondary and the geometry of the ends of the coil. A mildly sharpened steel bolt has been placed on the end of this current instance to direct a point where the high voltage can escape into the air where it will not damage the coil itself.
Other geometries of coils are possible but the particular 3d files are made for a 2.11cm radius PVC pipe for the secondary and a 4.38cm radius secondary.
Caps have been designed for the 2.11cm PVC pipe to help confine the coil. It is suggested that a notch be filed onto the end of the pipe where the wire can be glued into but the cap is designed with enough tolerance to deform if forced slightly with the wire sandwiched between the can and the pvc pipe. This also helps to hold onto wire as well.
If the primary shell is printed out, 12awg enamel wire can be wrapped around the shell to hold the primary. It if recommanded to first wrap the wire around a cylindrical object of a slightly smaller radius so that when the primary is transferred onto the 3d printer shell it can hold onto the shell tightly to help with tuning. The number of coils can be changed to tune the primary resonance frequency.
Next step is the construction of the circuit. This is done by following the schematic provided
Understanding the Schematic
If one looks at the schematic,it can roughly be divided up to three portions and examined individually.
On the right we have the high voltage portion of the driver circuit that will drive the primary at the resonant frequency of the secondary. There is a capacitor explicitly built into the primary circuit and coupled with the inductance of the primary coil this forms a resonant circuit. The inductance and capacitance will need to be played with slightly in order for the resonance frequency of the primary to match the resonance of the secondary. This is required in order for the secondary to achieve maximum voltage output as a resonant transformer. Assuming we a primary coil at the right resonant frequency, we then also have to use a mosfet to drive this primary resonance.
Why not just use the mosfet to drive the primary coil directly rather than having to set up a resonant primary? Well, at the frequency we are dealing with, the mosfet is less able to efficiently turn completely off and on when the drain and source voltages are at a maximum. In this set up, this is what is known as a class E amplifier and the mosfet would be commanded to switch off and on at the resonant frequency of the primary. This means then we need to then generate a signal to turn the mosfet on and off at the right frequency. This is where the next portion of the circuit comes in.
If we look centrally, we have the six in one schmitt inverter and gate driver.
We are using two of the six inverters on the IC, one of them put into an oscillator configuration and the other takes the output of the oscillator and inverts it and squares it up through the schmitt triggering mechanism. The potentiometer attached to the first stage oscillator inverter is used to tune the frequency of the oscillation and this is how we will tune the signal's frequency. Remember that we need a signal that will command the mosfet to turn on and off at the right frequency.
However, this signal is not strong enough to directly drive the mosfet gate. This is then where the gate driver IC comes in. This will then turn out 5V output square(ish) signal from the schmitt oscillator into 10-15V signal that will effectively drive the mosfet gate. There are two gate driver circuits within the same IC and we are using both in parallel. Both receives the input from the schmitt trigger and both will output the a signal that will, in combination, drive the mosfet.
Lastly, towards the left of the schematic is the audio signal processor. This uses a 555 timer to take an audio signal and processes it into a signal that will modulate the on and off periods of the tesla coil. This is where the "musical" aspect comes from. It takes whatever audio waveform that comes in and then modulates the tesla coil arcing in a way that corresponds to the audio signal such that we hear it through the arcs. The signal coming out of the 555 timer is fed into the enable pins of both drivers in the gate driver IC. This is how the 555 timer will turn the tesla coil on and off. Without an audio signal, the 555 interrupter can also be tuned using the potentiometers to adjust its duty cycle and base interrupter frequency which it will be outputting despite the lack of an audio input. It is possible to set this frequency to ultrasonic frequencies so the tesla coil once again becomes inaudible, at least without an audio input.
Circuit construction guidelines
It is advised that we employ the "dead bug" circuit building technique where IC and components are soldered directly onto a piece of copper clade board instead of using a non-soldering breadboard or solderable prototyping board with routing wires. Due to the high frequencies we are operating, these prototyping methods of circuit construction introduces a lot of parasitic capacitances that would interfere with the signals we are hoping to generate. Short of ordering custom PCBs, which is entirely possible with the PCB prototyping services, we will essentially make our own PCB, except its not printed.
This method done in this particular instance of this project is by carving channels into copper clade boards. This gives us sectioned areas on the board where we can wire components together. This does mean however that we will do a substantial amount of soldering but the benefit is that we can worry less about stray capacitances if we are using the copper clade as a big ground plane
After the coils have been wound
Then a frequency sweep should be conducted on the secondary in order to find the resonant frequency. Once this is found, we then tune the potentiometer on the schmitt oscillator so that the schmitt oscillator is outputting at the same frequency that the secondary is resonating at.
Finally, before we should get everything set up in the circuit and powered up, we should see that the driving signal integrity is maintained. We should probe the the waveform of the various versions of signals we are sending. At the very least by using oscilloscope, we should ensure that the schmitt trigger oscillator is provide the square(ish) waveform to the gate driver, then the gate driver should also be checked that it is outputing the same square(ish) signal.
After these have been verified, one could then get everything set up and carefully powered at the low voltage in order to proceed with tuning the primary.
As of right now, the first attempt at this project does not work.
Though the resonant frequency found for our particular secondary coil was 5.555MHz, though this changes easily with different constructed coil geometries and condition, setting the schmitt oscillator and tuning the primary to a frequency close to 5.555Mhz, no sparks were produced and high current consumption by the primary was observed.
Further investigation showed degradation of gate drive signal from the gate driver IC once it was connected with the mosfet. It is unclear as to why this is happening and could be due to multiple reasons. It could be that the current configuration of components still introduces enough stray capacitance to disrupt the gate drive signal. Though this 2nd iteration of the circuit already produces a clean schmitt oscillator output, it still sends the output from the gate driver across a pair of unshielded wire onto a separate copper clade board where the mosfet is situated. Another possibility could be a bad gate driver IC that is not able to provide sufficient signal strength to maintain the signal integrity once it is loaded by the gate capacitance of the mosfet. It has also been suggested that mosfet itself is unable to switch at the speed desired but the mosfet has been showed to be able to operate at at least 1MHz and the insufficient switching speed is already observed at lower frequencies with a constant voltage applied across the drain and source.
Another area for further investigation could the strangeness within the primary coil when a field sweep is performed. Though the inductance of the primary coil was measure and a corresponding capacitance required was calculated and closely implemented, once the primary coil circuit was soldered, the frequency sweep could not find the resonant frequency that would be predicted by the inductance and capacitance values. The primary coil with an inductance of between 3-4 microhenries and a capacitance of 0.25nF should have a resonance frequency of between 5.0-5.8Mhz but a frequency sweep of the primary coil showed a resonant range at 12-14Mhz which is strange.
If so far one had been successful, tuning the primary would be the next step. Since we could have calculated and soldered up a resonant capacitor of an appropriate size, we should be somewhat close to the secondary resonant frequency. To tune into the secondary even more precisely we can either stretch of compress the primary to change the inductance and this will change the resonant frequency of the primary. The point of tuning the primary is to also establish what is know as class E amplification where driving a resonant circuit with a mosfet at resonance, we limit the power dissipation on the mosfet and reduce mosfet heating and reducing the switching time as well. We know that the primary coil is tuned to the right resonance either by seeing the mosfet cool down or by probing the drain and source voltages and compare it to the gate voltage signal and see that the gate opens when the drain and source voltage is low.
As of right now, the first attempt at this project does not work.
Though the resonant frequency found for our particular secondary coil was 5.555MHz, though this changes easily with different constructed coil geometries and condition, setting the schmitt oscillator and tuning the primary to a frequency close to 5.555Mhz, no sparks were produced and high current consumption by the primary was observed.
Further investigation showed degradation of gate drive signal from the gate driver IC once it was connected with the mosfet. It is unclear as to why this is happening and could be due to multiple reasons. It could be that the current configuration of components still introduces enough stray capacitance to disrupt the gate drive signal. Though this 2nd iteration of the circuit already produces a clean schmitt oscillator output, it still sends the output from the gate driver across a pair of unshielded wire onto a separate copper clade board where the mosfet is situated. Another possibility could be a bad gate driver IC that is not able to provide sufficient signal strength to maintain the signal integrity once it is loaded by the gate capacitance of the mosfet. It has also been suggested that mosfet itself is unable to switch at the speed desired but the mosfet has been showed to be able to operate at at least 1MHz and the insufficient switching speed is already observed at lower frequencies with a constant voltage applied across the drain and source.
Another area for further investigation could the strangeness within the primary coil when a field sweep is performed. Though the inductance of the primary coil was measure and a corresponding capacitance required was calculated and closely implemented, once the primary coil circuit was soldered, the frequency sweep could not find the resonant frequency that would be predicted by the inductance and capacitance values. The primary coil with an inductance of between 3-4 microhenries and a capacitance of 0.25nF should have a resonance frequency of between 5.0-5.8Mhz but a frequency sweep of the primary coil showed a resonant range at 12-14Mhz which is strange.
If so far one had been successful, tuning the primary would be the next step. Since we could have calculated and soldered up a resonant capacitor of an appropriate size, we should be somewhat close to the secondary resonant frequency. To tune into the secondary even more precisely we can either stretch of compress the primary to change the inductance and this will change the resonant frequency of the primary. The point of tuning the primary is to also establish what is know as class E amplification where driving a resonant circuit with a mosfet at resonance, we limit the power dissipation on the mosfet and reduce mosfet heating and reducing the switching time as well. We know that the primary coil is tuned to the right resonance either by seeing the mosfet cool down or by probing the drain and source voltages and compare it to the gate voltage signal and see that the gate opens when the drain and source voltage is low.
