Hi bdn,
Please ignore what I said about the Spin-Up / Down delays, it will make no difference – My head was still full of another post I had made regarding over-burn with the use of the M3 / M5 command set.

I think you have identified the cause. My tube requires a minimum of 4mA to fire instantly at it’s first trigger (this would translate to 20% PWM). Thereafter it will fire at 2mA if the next pulse is within approx 100mS of the previous pulse.
The RF tubes have a slightly larger threshold so they overcome this problem by providing the 5kHz ‘tickle pulse’ before and in-between triggers.
I don’t fully understand the physics behind this but it appears that the various gasses within the tube need to be mixed / agitated in order to keep the tube in a state or readiness so you could perhaps try modifying your GCode to include an initial pulse or two (outside the work area) in order to prepare the tube before starting the cut.
Tweakie.