by Octal450 » Mon Jun 17, 2019 4:27 pm
Here is the most common issues I've seen:
* PID where no PID should be used
* Using integrators to try and catch up on lagging systems
* Duplicated controllers for the same thing (can cause handoff instability between modes
* Totally missing filters for when A/P is disabled, makes the PID initialize in a bad place
* Poorly tuned
Yes, correct. The generic A/P really needs to be redone. I would correct it and offer one if I had the time, but I just don't have time right now.
Essentially, think of it something like this: the ideal system (or theoretical system) has no turbulence, no weight inbalence, no disturbances, and possibly even reduced aerodynamics or mechanics of the system. But the actual system is much more less "ideal". So to have to build the system with all that in mind. What may work theoretically may not always work that well in actual aircraft FDM even if you tune it.
Plus, tuning your control loops is a complex process. If you want good results, don't just throw number at it and see what happens. Especially with PID elements. The entire system has to be carefully designed to achieve it's purpose. It also helps to learn the math and theory behind the PID and various components.
Kind Regards,
Josh