Extreme pitch/roll/yaw when changing modes in forward flight

I posted a similar question on DIY Drones but since then have had a crash (a combo of this issue and me being dumb). I was in fast forward flight and it switched to RTL at about 30’ above the ground. I didn’t have the GoPro on the front (luckily) and when it switch to RTL it pitched nose up and all the way over and in trying to recover without the necessary altitude it powered into the pavement. I replaced all of the props, tested the motors, fixed the GPS post and everything else was fine as it was inside the WQ-550 body. Given the opportunity I also moved the Pixhawk forward slightly to allow more room for the battery which puts the back edge of the pixhawk about center front to rear on the frame.

I’m thinking that this is at least partially due to the aircraft not being balanced. The 3300mah 3S LiPo is in the rear which ends up helping to balance things when the GoPro is on the nose although I have the problem GoPro or not, but it usually recovers. It “feels” like it is more severe when transitioning out of nose down pitch than it does out of nose up pitch…which seems to fit with the tail heavy config.

It flies very well in Stabilize but the Log Analysis shows pitch/roll problems if I go full pitch forward to gain speed then drop the stick to full pitch backwards it will get almost vertical for a moment and flatten out. When in auto or switching to auto, RTL, or loiter that attempts change the forward motion it will do the same but not recover as well.

Attached is the log from a few flights after the crash. Following autotune I like the way it flies in stabilize but it has the issue in loiter. At line 73812 is where I was in fast flight and it hit battery FS and showed the problem but did recover…it is heart stopping if the GoPro is onboard or you are above a lake though.

I’m not sure what to do to try and tune this out. Does it make a difference how well the Pixhawk is center front to rear? It is centered left to right. I can fly it just fine in stabilize mode although it does pitch back slightly if I go from full forward and let go of the stick.

I was just about to order a Tarto 2-D Gimbal today until I had the crash, now I want to get this figured out better before I add the gimbal. I was also looking at upgrading to Tiger MT2216-11 900Kv motors due to the extra weight.

WQ-550 Frame
Turnigy 2216-800Kv
1147 Props
Pixhawk w/ 3.1.5
uBlox GPS + Compass
Afro 30A ESC
3DR Power Module
Total liftoff weight is 1668g w/ GoPro.

diydrones.com/forum/topics/flies … e=activity

Any help would be appreciated.


It seem to be an known issue, with fix in coming firmware release 3.2. I can only recommend to keep CG in between motors so they are evenly loaded (thus more or less in same rpm range in steady hover) and to slow down before entering for example Loiter mode from Stabilise. In your case with aft heavy craft and then going into RTL, I recon the copter has massive increase of RPM to play with, on forward located props ,and I take it the flight controller in pre 3.2 update has a too coarse algorithm to neatly adjust pitch rate needed to stop.

Sent using Tapatalk

We’re definitely getting rid of a lot of transition jerks in 3.2. That said, jerk normal, crash not normal.

Is this copter autotuned? Did you do the autotune with a payload (gopro)? If so, redo autotune with no payload.

Also, it looks like RATE_PIT_D hit the upper limit of 0.015 during the autotune. It is possible that it can go higher. You definitely have massive overshoot in the pitch axis when pitching backwards. I’d also suggest increasing RATE_PIT_IMAX to 4500 - the constant offset towards being pitched back that you see in the attached graph contributing is significantly to the overshoot.

Thanks for the help, I look forward to the full release of 3.2 and hope that it clears some of the issue up.

jschall: I looked at the RATE_PIT_IMAX param and it is set at 500. According to MissionPlanner the valid settings are 0-500 and it warns me if I try to set it any higher so I didn’t try it. Is this an outdated reference in MP or was there an extra 0 when you said 4500?

I did another few flights today after another autotune this time with the GoPro case on (no camera, didn’t want to take the chance again yet) and it certainly made an improvement.

Yes, the copter was fresh off autotune in the log that I posted. No, there was no payload in the autotune from yesterday. The log file from the new autotune with payload is at droneshare.com/mission/12285

I’ve uploaded the last log from today to droneshare at droneshare.com/mission/12284 (is that easier than attaching here?) and I will also upload a video to show how it was flying today with the additional weight in the front which was better, but I think still a bit harsh and will probably be cleaned up in 3.2 it sounds like.

My other observation from the flights today is that the transition from forward flight to Loiter is by far the worse, followed by forward flight to RTL. It seems like it does a much better job between waypoints even if it has to turn around because it yaws towards the next point rather than trying to “brake” with pitch.

I think my crash yesterday was because it has been tuned with the full GoPro payload and that’s how I had been flying for weeks. I then tried full fast forward flight and flipped to RTL to low to the ground for it to recover even though it was trying to.

I can probably mark this as solved and wait for 3.2, but I am curious to find out if I should be tweaking the RATE_PIT_IMAX value to 4500.