Yaw causes pitch/roll wobble and flyaway in position hold

ArduCopter
1

HI All

I am trying to understand a problem I am having which has caused a few crashes.
I have a Skyjib 8 v1 with 15x5.5 props and xm5015mr-7 motors using a Pixhawk 1 running 3.6.7.

Generally flying in position hold it seems quite stable although I haven’t had a lot of flying time with it.

The issue starts when I push the yaw hard over and leave it there for a couple of rotations.

The first sign of an issue is a wobble in the roll/pitch that increases as it bowls away.
I am flying in a fairly small space so I don’t have much opportunity to recover.

I have uploaded 2 logs available here -
https://drive.google.com/drive/folders/1_DAy1X7FRBgfZoO5MP16H5n6-rRyg0yK?usp=sharing

In log 2.bin the yaw starts at about 13:06:35, the problem starts at about 13:06:39.
By 13:06:44 it has hit the ground.

In log 3.bin the yaw starts at about 10:58:43, the problem starts at about 10:58:49.
At about 10:58:55 it has run into a rope hanging from a tree branch before running off to find a brick wall.
( 8 props, 2 booms and possibly 2 motors to replace :sob: )

Any hints as to the cause of the roll/pitch instability would be appreciated.

I have also noticed that when the yaw control, RCIN c4, is returned to neutral, for a couple of seconds at least there is no reaction seen on RCOU c1 - c8 to indicate it is trying to stop the yaw. In fact the yaw rate increases according to GyrZ.

The activity in log 3 prior to the problem is an auto-tune.
The value of ATC_RAT_YAW_P is high at a value of 2 but not sure how this could cause the problem.
I was wondering if the I value was getting wound up but noticed I_MAX is not very large.
ATC_RAT_YAW_I,0.2
ATC_RAT_YAW_IMAX,0.5

Any help very much appreciated.

Thanks
Duncan

2

For a start, you still have ATC_RAT_PIT_FILT and ATC_RAT_RLL_FILT set to 20, the default.
You should start by reducing these to 10.
I haven’t played with the ATC_RAT_YAW_FILT as I haven’t needed to but autotune has come up with figures around 1. Maybe drop it to 3 to begin with.
The setup is over reactive and you seem to be inducing a yaw oscillation it cannot recover from.

48%20am
Screen Shot 2019-04-30 at 6.54.48 am.png2274×821 287 KB

You can see here that pitch and roll don’t go off the reservation until yaw is totally out of hand.
06%20am
Screen Shot 2019-04-30 at 6.56.06 am.jpg2276×822 616 KB

And here you can see the controller trying to stop the yaw but there is too much power and not enough mass to stop the yaw oscillations.
Notice how the motors are alternately shutting down trying to stop the yaw.

Bottom line:
Read the autotune docs here
Add some weight to get the motors operating in midrange.
Try again.

3

Thanks Mike

It will probably be up to a month before it’s flying again. I’ll try again then.

Cheers
Duncan