You should be able to get minimum period measurements in the vicinity of 20uS or better. I had added the circular queue to allow for a rapid succession of interrupts when I was trying to measure switch contact bounce. Since the hardware does it all, your results should be as consistent and jitter free as your signal is. Since it hijacks Timer1 and changes the prescaler, you won't be able to do analogWrite() to produce PWM on Pin 9 or Pin 10. The other PWM pins will be unaffected. If there is an overflow in the circular queue, a queue full of data (BUFFER_SIZE samples) will be dropped at a time.