Copter going crazy after failsafe rtl landed

Hi there, I just finished tuning my 7’ long range fpv drone, it has Mamba H743 V4 with dual MPU6000 running copter 4.3.4 and M8Q-5883 GPS.
I did a radio failsafe test which I unpluged the radio module out of my transmitter. It was all good as I observe it doing a radio failsafe in my goggle, after it touched down, I was walking towards it planning to pick it up but the motor suddenly starts going crazy. And I do not have the time to plug back my radio module which I could manually disarm it. It disarmed itself after 10-15 seconds (log saying 15s). Burnt my ESC and 1 motor.
I was thinking that I forgot to change the DISARM_DELAY paramater but when I came back and looked into the blackbox log, everything seems fine except the status show it thinks it has not landed after it landed. So the problem might not because of the DISARM_DELAY but something with landing detection.
Vibe data and Position estimation data all looks fine and I have no idea now of why it happened.
Here is my log.

Did it actually bounce on touchdown as the log seems to show? Or propwash affecting the baro?

It really touchdown as shown in the log, but before the actual ‘landing’ phase, by looking at the acc data you could see a jump in value followed by near 0 value. I didn’t here any propwash as it descend, but since I increased MOT_MIX_MAN to 0.9, the vertical descend does seem to be slower as the copter trying to increase throttle to maintain stability.
Here is a video I took as it doing failsafe. Unfortunately I ended the video after it landed and soon after it starts going crazy.

Right. I have a 7" LR also and I initially had this at 4 copying the parameter from Andy’s 7" LR build Blog. I set this back to .5 after the 1st flight. Good for Acro flying at a higher value but perhaps not so much for other styles of flying. I’m not saying that’s the cause of no Landing Detection but perhaps it’s related. In any case it was still running and the EKF became unhappy with resulting lane change and the motors took off. Not sure why that happened.

I was just about to test if I change my switch arm to arm/disarm instead of arm/disarm with airmode and enable airmode for acro mode, since it do appear affecting stablized mode descend. But it crashed after failsafe test and I’m still waiting for parts to get fixed lol

I don’t see anything unusual in the video for landing. I thought maybe with a big bounce the Landing detector wouldn’t trigger but it didn’t do that.

Landing in airmode will do this, although if it failsafed it should switch out since airmode only applies to manual modes.

Landing detection is very, very hard - most likely the landing was not detected

Ok then…So this means there isn’t actually a fix for this but be careful about it? As long as I have airmode enabled.

I mean, if you try landing with airmode enabled there is a good chance that the copter will bounce and the landing won’t be detected. I always have estop on a switch for copters that I use airmode and high MIX_MAN on.

To be clear. Arimode will not be engaged when failsafe RTL because you are in a non-manual mode (unless its a bug). So if you are flipping in this instance its not airmode causing it

sorry, I was thinking you are pointing a possible cause to me.

As far as I can tell your throttle is not at zero which will mean the landing detector won’t kick in. Good question though - it’s not at zero because you had a failsafe - will have to figure out whether it handles that properly.

Hi, I just found that my PILOT_THR_BHV is set to 0, will it possibly fix the problem if I set to 4?

That won’t change the Land detection behavior. It will only cause a Disarm after Land Detection.