Need help tuning very large quadcopter

Hi,

We are having trouble tuning a very very large octocopter. To give an idea of the size, we are using Turnigy RotoMax 150cc motors. We have been able to make this drone fly more stable in previous iterations, so we are afraid the new oscillations could be caused by a damaged motor, or frame miss-alignment. We are experiencing over corrections problems from the roll and pitch axis and some sort of harmonic oscillation in the yaw axis after an input. We also think vibrations from the frequency of the 8 motors could be causing the yaw oscillation.

We are hoping to gain some insight into which layer of the controller loop could be responsible for the instability, sensors, filtering, rate PIDs, angle PIDs, or all of the above. Any help is appreciated, thanks!

Latest-Log

What battery power and prop size? Why don’t you configure the Dynamic Notch Filter before getting too much further? Set INS_LOG_BAT_MASK to 1 and make an AltHold Hover flight for ~1 minute. Then review the FFT and/or post that log.

Not too much to learn from those short hops. Big overshoot on pitch and roll. The motor outputs show a pitch oscillation.

We are using 14s (2 7S batteries in series) and 32 inch propellers. I’ll give the Dynamic notch filtering a try and post the logs here. Thanks.

We attempted to configure the Dynamic Notch Filter following the instructions on the Wiki page. We did not really observe much or any improvement on the stability of the drone. The logs of the filtering tuning process can be found in the same location as before. logs

We also experienced a sudden drop of thrust when switching to altitude hold on our last flight, which resulted in a small accident. We are not sure what caused this sudden decrease of thrust, since the hover thrust parameters remained unchanged from previous flights, in which we were able to successfully use altitude hold. The log can be found in the same folder: “2021-04-16 19-15-24-Small-Accident-anon.log”. We believe this is unrelated to the filtering changes we made.

Any recommendations regarding filtering and PID tuning would be greatly appreciated, Thanks.

The notch filter looks fine. These look low:
ATC_ACCEL_P_MAX,10500
ATC_ACCEL_R_MAX,10500
ATC_ACCEL_Y_MAX,4666.66

Did you run Autotune with these values and the aggression at .05 or did Auto Tune set them this low?

Saw you posted in another thread. I’ll stop responding here.

Hi Dave, this thread is unrelated to the other one. We arrived at those numbers first by looking at the graphs in the tuning process instructions wiki for our prop size. Then the values were decreased even further to compensate for the empty weight to max weight ratio, as suggested in the same page.

image

I’m not sure that’s a wise thing to do before tuning. I have seen on at least one other large craft where these values are too low and a crash occurred due to lack of authority. One was a large X8 and the other was similar to yours on 12S.

I would set them back to the Initial tuning values, set aggression closer to .1 and try Auto Tune in the unloaded condition.

Sounds good, we will try increasing the values.

Is auto tune relatively safe on a drone this large? We were hoping to obtain decent control through manual tuning first, before attempting auto tune, we were not completely confident we were at that point yet.

In my experience autotune doesn´t work that well on large frames, but is a good starting point if it can already fly without a lot of oscillations.

I think that’s a fair statement. The last 32" quad I helped tune, which I would say is still a work in progress, required some tweaking after Auto Tune. And I think the goal is to arrive at an aggression setting of 0.1 or just start there.

1 Like