Influence of parameters - Plasmatree/PID-Analyzer GitHub Wiki

The tuning process itself is still a manual process. The availability of the characteristic curve makes it easier to put in relation to the generic PID controller as described on literature. For a useful overview just have a look at Wikipedia. The following examples show deviations from a neutral tune I set on a 3” quad and their influence on the resulting response. For simplicity only roll is shown. There is nearly no influence on the other axis.

These are just examples I produced with a single 3" quad. Other quads might behave slightly different or show other artefacts. Feel free to submit your own findings and help improve the wiki!

Influence of P

P is the proportional gain of the controller. Increasing it makes the controller react faster, but can lead to unwanted oscillations and over shoot. A defined edge by a minimal overshoot is often favourable compared to a smooth converging one. It relates to a faster and more defined reaction of the quad (‘locked in’-feeling) and leaves less work for the I-term to do.

Influence of D

D is the antagonist to P. It’s often falsely called ‘damping’ (which describes what it does in a certain range), while the correct name is ‘derivative’. It acts on the rate of change (which is the derivative) and thus counteracts the fast change produced by a high P. A too high D can lead to instabilities and oscillations as the system is perturbated. These will be visible by a ‘rougher’ plateau.