With my small FPV LR copter (7" props) I still have a tuning problem, that I could not sort out so far: on (very) low throttle, there is considerable twitching. The copter is defnitely overpowered, as most racers are and hovers at a very low throttle value (12%).
The twitching is there not only during descent, the same happens when suddenly reducing throttle during a rapid climb, while the vehicle is still moving upwards due to inertia in a zero-g situation. So I guess it is definitely not related to the dreaded vortex ring state often seen by heavier copters while descending in a hover.
The twitching is also very prominent when soaring down from altitude with a high forward airspeed and low throttle.
What I have tried so far:
I tried to lower MOT_THST_EXPO after reading a few related threads here. But with 0.57 I still see the twitching.
The MOT_SPIN_xxx params seem to have no real influence on this issue. Due to the BDShot ESCs used, the motor test even spins the props at 1% throttle. If have set MOT_SPIN_ARM to 7% for flying, as the props are fully stalled and do not create any noticeable lift below.
An Autotune was done, just to be sure - again no change in behavior.
Lowering the PIDs determined by AutoTune by 30% - the vehicle still flies fine, but same problem with low throttle oscillations.
MOT_PWM_MIN…MAX is set to 1000…2000. Lowest value in the graph is exactly 1070. However, this value is exactly MOT_PWM_MIN plus 7 %, which is in turn is MOT_SPIN_MIN.
I might try to do a test flight where I do not lower the throttle completely down to zero in the next days.
I am a bit hesitant to go further down with MOT_SPIN_MIN. At 1% or 2% I can see the prop doing one rotation per second or so. It is defintely completely stalled and producing zero thrust here, at 5% as well. The value of 7% is the value where I begin to feel some kind of prop wash 5cm under the prop.
Tried that and there is indeed some improvement. The amplitude of the oscillations is smaller, and the new lowest value of RCOUT is - big surprise - 1050 now. However, the RCOUT signal does not look like there is some clipping at 1050. The oscillations are also more damped now, start later and stop sooner (just a gut/finger feeling). I am unable to trigger any oscillations when throttling down to 10% instead of 0%.
Unfortunately there is a catch, that I almost expected to happen: when suddenly throttling down at high forward airspeed, there is sometimes an uncommanded roll or pitch deviation now. I have seen this before and that was one of the reasons not letting the motors spin down to a speed that clearly stalls a prop.
Maybe there is a need for scaling the PIDs not only by battery voltage, but also by throttle output?
