As the author says, there are some flaws, I am working on improving it by using slightly different routines and Pic usage, also higher resolution pulse which allows for much shorter duration between reads.
Well, you said it yourself, not as good as PID. So why not implement a proper PID controller? Modelling of a DC motor is very simple, and you should be able to tune all the PID parameters from a simple simulation in Simulink or other programs. Quick tip: velocity control of DC motor -> PI controller. Position control of DC motor -> P controller.
True, but irrelevant. The control signal is still proportional to the error. As an example, position control for a DC motor: if the error is 10 nanometers, and the control signal is 10 nanovolt, we would consider that the reference is met. In MaxHeadRooms case he has conditional statements: if error>0, do something, if error=0, do nothing. In this case the "do nothing" might never be executed!No, but an analog PID approach is never EXACTLY zero either due to offsets and other errors.
It engineering you get the error low enough to meet the specs and that's where you stop (or shoot the engineer).
Interesting. I used a like control system for a positioning system in the track_motor function a while back. The error signal from the set-point was a turns count for a encoder or a voltage differential for positioning pots instead of speed but the 'fuzzy' logic was much the same because it was emulating a primitive motor controller. It required quite a bit of code fine-tuning to make the control system smooth and accurate.
Very, very nice work ...
Facebook
Google
GitHub
Linkedin
