Menno:
PWM is PWM however it is implemented. You are describing one method of implementation.

Note that the PCA being setup in full 16 bit "High Speed" output mode can generate stable output waveforms that are jitter free and isolated from software interrupt latency variation. The only limitation in this mode is that the narrowest widths that can be attained with the PWM waveform at near 0% and at near 100% duty cycle are regulated by the sum of the worst case interrupt latency time plus the length of time it takes to execute the interrupt service routine. You can still have "full control range" if you do three things.

- For 0% level have software disable the PCA channel and drive the output bit to the OFF state.

- For 100% level have software disable the PCA channel and drive the output bit to the ON level.

- For intermediate levels first find out the number of ticks of the selected PCA clock that are more than the interrupt latency time plus the interrupt execution time. Lets say for discussion you determine this to be 8 clocks of time worst case including a comfort margin. You then divide the 16 bit resolution if the PCA channel up by the 8 counts. This derives your basic resolution for the PWM. (I.E. 65536/8 = 8192 steps in the PWM).

So you can control the PWM in steps from 0 to 8192. Full control at that resolution of 8 PCA timer clocks.

When you use the High Speed output mode though you are free to use almost any total period you want. You dont need to just use a full 16 bit range. You can use a shorter range such that you have control over the PWM period which in some cases is mandatory due to time constants and filtering characteristics of what ever the PWM is controlling. (Whether those characteristics be electrical in nature or mechanical in nature).

Michael Karas