Ok, so need some assistance with tuning yaw - see the attached vid (you won't need to watch more than a minute or two to see the problem. You can ignore the pitch issues which I'm addressing via the gimbal PID's.
Basically, I can't get the damned thing to fly straight without the nose bouncing around. Speed doesn't seem to matter - loiter or high speed, the bouncing is the same.
But it appears yaw can't be tuned the same way as pitch and roll - I've tried:
- detuning P (to 0.1)
- uptuning P (to 0.35)
and various settings inbetween.
I've also tried combining it with Stab Yaw as well - from as low as 4.5 to 9.5. And tried doing some I tuning, but it didn't really seem to have any effect.
If I detune it, it just bounces left/right slower, and has a lot of overshoot. If I tune it up, it bounces less, but more more quickly.
Where should I be tuning? Should I even look at D to control the overshoot?
I redid the compass calibration at the field before I shot this video. Compassmot is 9%, and I have tried running with and without it enabled, and it makes no difference. All motors are perfectly aligned, and seem to be running normally. Running APM2.6 and 3.1.5
Replies
So are you relying on the supplied tools only, or do you know anything about PID?
I think you should understand PID first, try Wikipedia.
In general, you would want to have D low, almost 0, I also low, but sometimes it helps, and P is the one that makes the fastest moves, because it is simply P, which is proportional to the difference between set point and actual point.
Your overshoot is being caused by I or D, but most likely I because it takes some time for the Integration to accumulate, and then it will oscillate. Cmon R&D.
If you want it to be as smooth as possible, try a low or medium P setting, no I , no D. And get back to me, if you care.
Lol - did *you* read it?
Answer me this mr PID-expert-who's-annoyed-I-critised-his-airframe-design...why is there no D on yaw in arducopter, and why in most craft is it not needed.
Because you don't need to use any differential feedback for small craft with good response.
If it was a larger craft the inertia of moving would cause it to overshoot which is why a negative D is added, to reduce the speed as it approaches the target point. I went through 2 years of just PID.
So if there is no option for D,
this assumes you are moving a small mass, and that you need to keep P low.
I (integral) can help if there is any sensor offset or drifting.