I have read a paper where they used an FPGA to measure the currents @20MHz, but I also think
that deadtime is too short and full of noise because of the switching FETs.
Than you inject some kind of short pwm pattern like INFORM method and calculate the current derivative, right?
Do you have two...