Momentary attitude instability during auto mission

ArduCopter Copter 4.6
1

Hi all,

I was out flying my Holybro X650 quad this morning - mostly running some mission tests to dial in things like CIRCLE_RATE and doing a little tuning.

During one of my missions, the quad experienced a moment of oscillation between approaching the ‘circle’ waypoint, and beginning the execution of the circle.
Visually, the quad slowed down as it approached the perimeter of the circle, it began oscillating as it slowed (most noticeable in roll), and continued to oscillate whilst it was ‘paused’ hovering at this point. Then, once it began tracing the perimeter of the circle, the oscillation subsided and the rest of the flight felt fine again.
You can see the instability in roll in the below plot.

newplot (43)
newplot (43)1170×576 45.8 KB

I’ve had this setup flying for a little while now, though in the last month I’ve only logged about 30mins total flight time since I’ve been away for a bit. But until now, the flight performance has been pretty good.

One thing I realised is that I’ve forgotten to setup my ESC telemetry based notch filter. I have the parameters ready to upload but in this flight I’m currently still using throttle-based. I don’t see why this would cause a momentary loss of stability but felt worth mentioning.

I’ve tried to use the PID review webtool and got the following outputs at the moment of the loss of stability in roll. The first plot shows the actual roll rate overshooting the target at this point during the oscillations.

newplot (44)
newplot (44)1200×450 72.5 KB

newplot (45)
newplot (45)1200×450 79 KB

I haven’t yet run an autotune on this kit, but after this incident I’ve moved it up in my priorities, and I’m hoping to get that done in the next week or so.

If anyone has any insight on why this happened when it did, and why it seemed so intermittent, it’d be great to learn more about it!

Kit:

Frame - Holybro X650
Flight controller - Pixhawk 6C
Battery - Zeee 6S 8000mAh
GPS - Holybro M10

Log:

.bin file

2

Running auto missions on a quad that isn’t tuned yet is risky. The good news is the quad is flying well for not being tuned so you’re in a good place to finish things up.

The filter settings you have aren’t bad, but you could simplify them a lot and still have good results. I’ve attached my suggestions based on that flight.

newplot(9)
newplot(9)1200×450 13.1 KB

filter(1).param (335 Bytes)

There is a slightly higher than ideal Z vibration. Not the worst by a long shot, but might be worth making sure you don’t have anything loose bouncing on the controller or maybe a prop with a damaged tip.

Once you’ve got that done you could run the quick tune script and then likely go to auto tune.

3

This is helpful, thanks. Quite similar to the params I have lined up for my next batch of tests:)

I’ve not noticed anything - props are in good shape and cables are relatively well managed. I’ll give everything a once over and tighten anything that could be a problem. Thanks for the heads up

:ok_hand: I’ll drop the results here in a few days once I’ve got round to it! Thanks for the insight:)

1 Like