Sudden instability

ArduCopter Copter 4.4
1

After having correctly performed the autotuning, and correcting the values according to https://ardupilot.org/copter/docs/autotune.html to obtain a less aggressive response, I have the problem that with the payload of this quadcopter after remaining stable in Loiter, destabilizes without even giving it any order in pitch or roll.

I have ruled out that the flight controller is defective by installing a new one, the chassis is rigid enough, in fact it is the second drone with it and the first one flies perfectly.

I have observed testing each of the 4 Hobbywing X9 motors on a test bench that they all have a slight vibration at 1430 PWM, and this should be filtered by the FFT that is also activated. However, it is strange that it destabilizes after a few seconds in hover.

I have been testing with 4.4.0, 4.4.3 and 4.4.4 firware versions.

Please, if someone can review this log and comment on what is incorrect it would be of great help to me.

Thanks!

https://drive.google.com/file/d/1028mYK-JztjIm_QuRj-I-AeaqH92NGSW/view?usp=sharing

2

Flash to V4.4.4 and stay there. Until V4.5 is stable anyway. The tune is very poor with low ATC_ANG pitch/roll indicative of an Autotune aggression <.075. The vibration levels are very high with clipping. You have to hammer a CubeOrange+ pretty hard to clip. Can’t check the Notch filter you have batch logging disabled.

3

Thanks for your answer. So, do you suggest that the loss of stability can be solved with finer tuning?

Video of this flight:

https://youtu.be/dM0-xeoPjfo?feature=shared

4

Yes, unless the base vibration levels are too high. Remove the payload and run Auto Tune again with an aggression of 0.075 (never go below this). This assumes the notch filter is set correctly. You may want to just use a throttle based notch rather than FFT.

5

Thanks again. I will do that.

6

Hi, please check notch filter, this log has imu logging:

Log

It was previous to change to Thottle based mode. Vibes are under 20 mostly.

Looking at the pre-filter data, do you think these parameters are more appropriate?

INS_HNTCH_MODE 1:Throttle
INS_HNTCH_FREQ 33Hz
INS_HNTCH_BW 6Hz
INS_HNTCH_HMNCS 15 (up to 4th harmonic)

7

I would try these and make another flight to check them out. Disable FFT FFT_ENABLE,0

Notch

You will need to run another Auto Tune with a higher aggression. .075 to .100. The tuning is poor at present.

1 Like
8

Hi,

The parameters were modified and this flight was made.

Flight 1

On the next flight AUTOTUNE was performed with AUTOTUNE_AGGR=0.100 and this was the result

Flight 2

A flight with the payload was then carried out. It aborted immediately because the vibrations came back.

Flight 3

The payload was removed and it flew again. In theory the result should be the same as flight 2, since no parameters were changed, however strong vibrations were observed again.

Flight 4

What do you think?

9

You have 48Hz for the Notch filter freq so it’s missing the largest mode.

10

You are right, flight 4 has 48Hz for Notch filter, but you can compare flight 2 vs 3, without and with payload which both have 38Hz filter.

Flight 3 has vibrations above 30 m/s/s constantly, however Flight 2 barely reaches 30 m/s/s.