Mavic drops out of sky - crashes - why?

[pjurka]

Hi,
I'm new to this forum and new when it comes to use drones.

I've recently bought an Mavic Pro Platinum (only six days ago) and yesterday it crashed and are now totally demolished.

Yesterday I flew at a few different places without any problems. On my last flight I sent it up in the air to let it hover and just a minute and a half later it came falling down. I didn't see any warnings before or during the flight. When I check the flight at Airdata it all seems good. No warnings as far as I can tell.

The conditions were good. Not much wind. I see on the log in Airdata that the drone have been moving around, even though it should be hovering in the same place.

I don't think that I've made any mistakes, but since I'm new to this there might be something that I've missed. I've downloaded the .dat-file and the .txt-file in the hope that any of you wizards might help me give an explanation at what went wrong. The files can be found here (I hope I've downloaded the right files): MAVIC_PRO_FILES

Any help is much appreciated.

All the best,
/p

 

[sar104]

The motors were shut down by the FC at 99.283 s when the roll exceeded 70°:

98.181 : 6883 [L-FMU/FSM]near ground
99.103 : 6929 [L-FMU/MOTOR]safe_near_grd:true
99.206 : 6934 [L-FMU/FSM]not near ground
99.265 : 6937 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
99.283 : 6938 [L-FMU/MOTOR] Stop. reason:roll over or atti fail
99.283 : 6938 [L-FMU/MOTOR]Total start times: 1, time: 99.52
99.283 : 6938 [L-PWM]set actuator mode:IDLE
99.283 : 6938 [L-FLYMODE]CTRL reset all by motor stopped
99.285 : 6938 [L-RC]craft ctrl failed!!!
99.286 : 6938 [L-FMU/FSM]state changed. cur: motor off
99.287 : 6938 [L-FDI][CTRL]: fault off, tilt_ctrl_fail​

It's not completely clear what caused the attitude excursion that trigger shutdown, but the motors were behaving strangely before it happened. Still looking...

 

[MavicCF]

The motors were shut down by the FC at 99.283 s when the roll exceeded 70°:

98.181 : 6883 [L-FMU/FSM]near ground
99.103 : 6929 [L-FMU/MOTOR]safe_near_grd:true
99.206 : 6934 [L-FMU/FSM]not near ground
99.265 : 6937 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
99.283 : 6938 [L-FMU/MOTOR] Stop. reason:roll over or atti fail
99.283 : 6938 [L-FMU/MOTOR]Total start times: 1, time: 99.52
99.283 : 6938 [L-PWM]set actuator mode:IDLE
99.283 : 6938 [L-FLYMODE]CTRL reset all by motor stopped
99.285 : 6938 [L-RC]craft ctrl failed!!!
99.286 : 6938 [L-FMU/FSM]state changed. cur: motor off
99.287 : 6938 [L-FDI][CTRL]: fault off, tilt_ctrl_fail​

View attachment 36045

It's not completely clear what caused the attitude excursion that trigger shutdown, but the motors were behaving strangely before it happened. Still looking...

Well that saves me calling you in! Was that info in the AC DAT file? I can't find those messages in the tablet DAT I was using.

 

[sar104]

Well that saves me calling you in! Was that info in the AC DAT file? I can't find those messages in the tablet DAT I was using.

I was working with the aircraft DAT, but they are in both DATs. Slightly different timing data in the tablet DAT:

98.439 : 6883 [L-FMU/FSM]near ground
99.361 : 6929 [L-FMU/MOTOR]safe_near_grd:true
99.464 : 6934 [L-FMU/FSM]not near ground
99.523 : 6937 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
99.540 : 6938 [L-FMU/MOTOR] Stop. reason:roll over or atti fail
99.540 : 6938 [L-FMU/MOTOR]Total start times: 1, time: 99.52
99.541 : 6938 [L-PWM]set actuator mode:IDLE
99.541 : 6938 [L-FLYMODE]CTRL reset all by motor stopped
99.543 : 6938 [L-RC]craft ctrl failed!!!
99.544 : 6938 [L-FMU/FSM]state changed. cur: motor off
99.545 : 6938 [L-FDI][CTRL]: fault off, tilt_ctrl_fail
99.558 : 6939 [L-IMU]set imu switch level = 1

 

[sar104]

The only obvious issue is the response of the right front motor at 99.14 s. The FC requested higher speed, but the data indicate that the motor slowed down. That appears to have been the start of the attitude excursion.



I've not seen anything quite like that before.

 

[pjurka]

The only obvious issue is the response of the right front motor at 99.14 s. The FC requested higher speed, but the data indicate that the motor slowed down. That appears to have been the start of the attitude excursion.

View attachment 36058

I've not seen anything quite like that before.

Thanks for taking your time. I've been in contact with DJI to request repair (even though there's not a whole lot to repair - the drone is just completely wrecked), so I'm hoping they will find that there was a malfunction on the aircraft.
/p

 

[joejaxvic123]

I've not seen anything quite like that before.

No way!

 

[laurens23]

Just a guess but here it goes:

facts:
you were taking off in a small clearing between the trees.
you were flying at roughly 20 m (barometric altitude)
@97.5 sec you give full down throttle.
@98.0 sec the first (small) oscillation start (first pitch then roll)
@98.2 the throttle is back to neutral
@98.5 throttle is going full up
@99.2 the oscillations escalate and @sar104 posted the resulting steps ending in motor shutdown.

my guess:
you flew/hovered just above the tree tops.
@98 the small throttle down made the drone clip the tree
in the following second the drone is fighting the tree
@99.1 sec the tree wins the fight, the right front motor gets stuck and it's rpm drops.
What follows is the sequence at posted by @sar104

This is supported by the ultrasonic sensor intermittently detecting something (leaves) between 0 and 30cm below the drone starting @98.1 sec.

 

[swooshdave]

Just a guess but here it goes:

facts:
you were taking off in a small clearing between the trees.
you were flying at roughly 20 m (barometric altitude)
@97.5 sec you give full down throttle.
@98.0 sec the first (small) oscillation start (first pitch then roll)
@98.2 the throttle is back to neutral
@98.5 throttle is going full up
@99.2 the oscillations escalate and @sar104 posted the resulting steps ending in motor shutdown.

my guess:
you flew/hovered just above the tree tops.
@98 the small throttle down made the drone clip the tree
in the following second the drone is fighting the tree
@99.1 sec the tree wins the fight, the right front motor gets stuck and it's rpm drops.
What follows is the sequence at posted by @sar104

This is supported by the ultrasonic sensor intermittently detecting something (leaves) between 0 and 30cm below the drone starting @98.1 sec.

I guess people don’t realize how much data is available. A simple cross reference of the location shows how much info was left out of the original story.

 

[sar104]

Just a guess but here it goes:

facts:
you were taking off in a small clearing between the trees.
you were flying at roughly 20 m (barometric altitude)
@97.5 sec you give full down throttle.
@98.0 sec the first (small) oscillation start (first pitch then roll)
@98.2 the throttle is back to neutral
@98.5 throttle is going full up
@99.2 the oscillations escalate and @sar104 posted the resulting steps ending in motor shutdown.

my guess:
you flew/hovered just above the tree tops.
@98 the small throttle down made the drone clip the tree
in the following second the drone is fighting the tree
@99.1 sec the tree wins the fight, the right front motor gets stuck and it's rpm drops.
What follows is the sequence at posted by @sar104

This is supported by the ultrasonic sensor intermittently detecting something (leaves) between 0 and 30cm below the drone starting @98.1 sec.

That's a credible hypothesis.

 

[Simmo]

Guys whats the waveheight? @sar104

 

[RayOZ]

Ah.. so that log about "[L-FMU/FSM]near ground" and "[L-FMU/FSM]not near ground" is that it detected something underneath. I initially thought it went up and hit something on one motor. I guess it's the other way round. I wonder of DJI will come to a similar conclusion.

 

[swooshdave]

Ah.. so that log about "[L-FMU/FSM]near ground" and "[L-FMU/FSM]not near ground" is that it detected something underneath. I initially thought it went up and hit something on one motor. I guess it's the other way round. I wonder of DJI will come to a similar conclusion.

They will now.

 

[pjurka]

Just a guess but here it goes:

facts:
you were taking off in a small clearing between the trees.
you were flying at roughly 20 m (barometric altitude)
@97.5 sec you give full down throttle.
@98.0 sec the first (small) oscillation start (first pitch then roll)
@98.2 the throttle is back to neutral
@98.5 throttle is going full up
@99.2 the oscillations escalate and @sar104 posted the resulting steps ending in motor shutdown.

my guess:
you flew/hovered just above the tree tops.
@98 the small throttle down made the drone clip the tree
in the following second the drone is fighting the tree
@99.1 sec the tree wins the fight, the right front motor gets stuck and it's rpm drops.
What follows is the sequence at posted by @sar104

This is supported by the ultrasonic sensor intermittently detecting something (leaves) between 0 and 30cm below the drone starting @98.1 sec.

Thanks for your reply and ideas.

Just a few clarifications (I don't know if they have any affect on your hypothesis, but I give them anyway):

The spring here in Sweden is very late - there are no leaves whatsoever on the trees (not even the tiniest).

I was standing on an pathway in the forest where it was quite clear from trees and branches all around me. No branches straight above. I had visual on the drone the entire time. I basically flew it straight up and let it hover there. I never flew it over the trees, or moved it a lot sideways. In my opinion (but I'm a beginner) I was clear of the trees the entire time and as far as I can tell I never hit any branches. However I noticed that the drone itself was drifting when it should be hovering (in retrospect, I should have taken the drone down when I noticed that, but I didn't).

Since I'm a beginner I'm here to learn from my mistakes, and hopefully avoid them next time, so here are a few questions to help me understand better:

- If I were flying too close to the trees/branches, wouldn't there be some kind of warning/notification?
- When I check the flight at Airdata UAV I see that the altitude of my last known location was 21.1 m - doesn't that indicate that something happened to the drone?
- If you where to fly at a location like this, would you have any particular settings to avoid ending up in my situation?

Once again, thank you for taking your time. I really appreciate it.
/p

 

[RayOZ]

The sensors can't really pick up on small twigs. Perhaps it hit one of those. Have a look at the front right motor, to see if it hit something, prior to flipping over. Another scenario could be spider webs?

 

[laurens23]

- If I were flying too close to the trees/branches, wouldn't there be some kind of warning/notification?

The sensors are not reliable at detecting small objects like tree branches or power lines. (I turend mine off as they cause more trouble than they are worth for me. They are not reliable enough to trust. I see them more as a marketing gimmick)

- When I check the flight at Airdata UAV I see that the altitude of my last known location was 21.1 m - doesn't that indicate that something happened to the drone?

Something definitely happened to the drone. I still think it clipped a tree branch. leaves or no leaves does not change my thought.

It looks like the Airdata log ends when the power to the motors is shut off. What happened after that is not recorded in the txt log. It might be in the aircraft .DAT file, but those files are big and downloading time in China at my work is slow, so I only use the txt logs.

- If you where to fly at a location like this, would you have any particular settings to avoid ending up in my situation?

Personally I feel safer flying higher. Where I live there are not many trees, but plenty of power lines, tall buildings, phone towers, large bill boards, etc. I rarely fly below 50-60m altitude, usually quite a bit higher.

In your case I would have climbed through the trees and gone up well above the treetops. maybe 30-50m. Branches etc are hard to see even on the camera.

Once again, thank you for taking your time. I really appreciate it.

you are welcome. remember this was just speculation, but I'm afraid DJI will come to the same/similar conclusion (pilot error) after looking at the logs. Hopefully the repair cost is not to bad.

edit: do you have a video of the flight? that might give more details.

 

[pjurka]

edit: do you have a video of the flight? that might give more details.

Unfortunately not, I didn't even manage to take a photo before it came hurling down, it all happened so quickly.

The thing that still makes me wonder if there could be some kind of malfunction is that the drone was moving around, when it should be hovering in the same spot. What do you think about that?

 

[laurens23]

Unfortunately not, I didn't even manage to take a photo before it came hurling down, it all happened so quickly.

The thing that still makes me wonder if there could be some kind of malfunction is that the drone was moving around, when it should be hovering in the same spot. What do you think about that?

I don't know. It was drifting a bit more than I would expect, but nothing too crazy. In total it moved 15 meter in ~100 seconds (red line). Around half of that distance was from control inputs (mainly at 20-28 and 55-60).

When it was drifting it was so slow the velocity parameter did not even registered it (light green line was at 0m/s) But the drone should keep position based on GPS. The GPS signal was good and the GPS did notice the small drift, but no compensations was made to counter it.

I don't know why the drone did not compensate, or if it should have done so in this case. But it is nothing like the "blown away, fly away or out of control" I have seen in other logs.

 

[sar104]

The sensors are not reliable at detecting small objects like tree branches or power lines. (I turend mine off as they cause more trouble than they are worth for me. They are not reliable enough to trust. I see them more as a marketing gimmick)



Something definitely happened to the drone. I still think it clipped a tree branch. leaves or no leaves does not change my thought.

It looks like the Airdata log ends when the power to the motors is shut off. What happened after that is not recorded in the txt log. It might be in the aircraft .DAT file, but those files are big and downloading time in China at my work is slow, so I only use the txt logs.



Personally I feel safer flying higher. Where I live there are not many trees, but plenty of power lines, tall buildings, phone towers, large bill boards, etc. I rarely fly below 50-60m altitude, usually quite a bit higher.

In your case I would have climbed through the trees and gone up well above the treetops. maybe 30-50m. Branches etc are hard to see even on the camera.



you are welcome. remember this was just speculation, but I'm afraid DJI will come to the same/similar conclusion (pilot error) after looking at the logs. Hopefully the repair cost is not to bad.

edit: do you have a video of the flight? that might give more details.

The mobile device and aircraft DAT files recorded past the motor shutoff, as shown in post #5. Was there anything in particular that you wanted to see?

 

[sar104]

Guys whats the waveheight? @sar104