@soren Good that it works. But omitting the sleep() commands are hardly the reason. I checked with an oscilloscope that PWM output stays when sleep() is performed. It's more something else you did. I worked with these servos decades ago, and remember that the pulse width range they expect is pretty small, from 1 ms to 2ms, at a 20ms period, resulting at a PWM duty cycle range between 5% and 10%
