Triode-like behaviour of bottom SiC MOSFET endstage. Topside MOSFET - gyrator and output. CREE SiC 1700V (C2M1000170D).
First tests are done, connected to signal generator, sinewave looks nice on scope. Unfortunately these SiC MOSFETs can be easily damaged by ESD, then will pull big gate current! I took new transistors from ESD bag, shorted G+S with aligator clips; and soldered 9V Zeners directly to transistor pins (near the body). Now they can be handled safely. Do this soldering at ESD area, if possible. Unfortunately I killed 3 transistors with static discharge, about 20€ loss....
Here is the "AmberBridge", in this raw shape it does 350Hz..250kHz (-3dB). Yes, no bass here (now), tested it with signal generator through µ47 cap. It´s the one of reasons for having no bass.
With next update, I will go with line transformer, because winding is floating source of signal, and feedback network can be wired better. Lundahl LL1545A (mu-metal) costs about 80€/pc. Transformer benefits: 1) exact phase (and same amplitude) splitting for each stator; 2) high voltage isolation (1500V) to input.
As you can see, this has minimum components and is simple to build. Compared to original Stax amplifiers, it will be very cheap. Next version with driver transformers will increase price a bit, quality will be improved too (with different feedback network).
One critical spot here - feedback network and correct frequency compensation. This is the main reason, why I selected lowest capacitance MOSFET.