From what I experienced, playing with various transformers, its generally around the winding resistance and amplifier interactions. You can shift around and manipulate the frequency response by using a series resistor on the primary side and playing with some parallel resistors on the output taps if you need to go even further. It’s basically creating RC / RL networks. You might be able to level out the LF rolloff and HF rise. I prefer not to use resistors at all personally but depends on the transformer.
You will be shocked what even a 1ohm resistor series primary winding resistor will do and completely flip things around.
Load a 20hz sinewave and run up the volume while watching the scope and see what sort of voltage swing you can get also before the sine wave falls apart (minding what you’re doing haha) that seems to be where transformers show their capabilities and worst case scenario for how much voltage drive you have
Transformer input protection is interesting and something STAX seemed to struggle with also (those garbage thermistors). I have been playing with some ideas of zener clamps / voltage monitoring / MOSFET switches and its possible but my first iterations of all this were to maintain the purest signal path possible with minimal component count and just being mindful of the power available and keeping gains limiters in place in my signal chain.
I have implemented zener crowbars on the bias side though to end the show if any overload occurs on that side. Which should be impossible as is with the voltage regulator safety features.