Functioning limits of a SSTC interrupter duty cycle and frequency

[Seb555]

After recently getting my first SSTC to produce arcs, I've been experimenting with the interrupter.  My goal is to pulse width modulate the interrupter signal with audio in order to produce music, rather than just making simple 100Hz or 500Hz tones. Such an interrupter signal would need to have a carrier frequency above human hearing (like 30kHz) so that only the PWM aspect is heard.

Because I wanted to first test the capabilities of my SSTC, I constructed a temporary circuit with a microcontroller (arduino nano) to produce the interrupter signal. All it does it produce a square wave with a set frequency and duty cycle controlled by a potentiometer. The problem that I've been facing is that whenever I set the frequency above 1kHz, the tesla coil only comes on for a split second and then shuts off until I disconnect power and turn it on again. This same phenomenon has been occurring for lower interrupter frequencies too, except that it takes a certain duty cycle before the tesla coil shuts off. Here is a list of the interrupter frequencies I have tested and their corresponding max duty cycles (in % on time) before my SSTC spontaneously shuts off:
60Hz - 2%, 100Hz - 4%, 200Hz - 7%, 400Hz - 12.37%, 600Hz - 14%, 800Hz - 1%.
To try to increase feedback and mitigate noise (what I assume to be the issues here) I increased my feedback CT windings to 4 and encased the interior of the wooden box that holds my electronics with earth grounded aluminum foil. These changes barely made an effect on the performance. I also tried changing the topload from a large toroid to a smaller disk (and afterwards tested with no topload) but only got worse results.
I'm just wondering if anyone has ever encountered a similar issue of their SSTC shutting off (without getting damaged) at high interrupter frequencies or duty cycles and what you guys have done to deal with this sort of problem.