The PWM circuit makes no sense as drawn, please review. I guess the µC PWM output should be controlling the switch, not feeding a switch terminal.
It's necessary to analyze the exact low pass transfer function instead of assuming a theoretical "cross over" frequency. If there are PWM residuals left after the filter, the ADC should be synced with the PWM, otherwise expect interferences and alias signal genertion.
The final control loop bandwidth will be established by the control algorithm, according to application requirements and hardware capabilities.