Pixhawk crash after takeoff in auto mode

I am hoping that somebody can assist me to understand what happened during this flight. I have a quad setup running pixhawk with arducopter 3.1.5. I have performed successful auto tuning and completed hundreds of fully autonomous flights. The day before the crash I completed a 10km autonomous flight with no problems.

On this flight just after raising the throttle in auto mode took off and violently move horizontally to the right and then started to climb to the target alt of 40 meters, but looked lower than that. It then looked like it was starting to move in an incorrect direction (not in accordance with flightplan) It then rapidly started to fly away rolling way past 45 degrees starting to dive towards the ground. I tried to recover by selecting loiter mode and throttling up, but this did not help and the quad crashed. I think my saving grace was the tree it went through as I was able to fix the quad with minimal spares. (just on arm and gps support broke! Unbelievable given the severity of the crash!) When I started up the pixhawk I downloaded the log but before that I retrieved the waypoints that was loaded. Something that was strange was that the first waypoint was missing with only home and waypoint 2 and onwards present on map.

Dont know if this navigational issue could have caused the drone to crash. Surely the build in limitations for pitch and roll angle should have prevented this?

Other information is that I later saw some high power lines and a telecoms tower not too far away. No sure it this could have influenced quad?

Thanks for helping a newbi…

I am really trying to get to the bottom of this crash. I am including the results from the autoanalysis here. Can somebody perhaps at least comment on whether the reported IMU mismatch could be the cause of the crash?
Thanks guys.

Log File C:/Users/Jasper/AppData/Local/Temp/tmp7794.tmp.log
Size (kb) 558.87890625
No of lines 7785
Duration 0:00:44
Vehicletype ArduCopter
Firmware Version V3.1.5
Firmware Hash ee63c88b
Hardware Type
Free Mem 0
Skipped Lines 0

Test: Autotune = UNKNOWN - No ATUN log data
Test: Balance/Twist = GOOD -
Test: Brownout = GOOD -
Test: Compass = FAIL - Large compass off params (X:-156.64, Y:-133.48, Z:-186.40)
Large compass offset in MAG data:
X: -156.00
Z: -186.00
Moderate change in mag_field (34.06%)

Test: Dupe Log Data = GOOD -
Test: Empty = GOOD -
Test: Event/Failsafe = GOOD -
Test: GPS = GOOD -
Test: IMU Mismatch = FAIL - Check vibration or accelerometer calibration. (Mismatch: 2.16, WARN: 0.75, FAIL: 1.50)
Test: Parameters = GOOD -
Test: PM = GOOD -
Test: Pitch/Roll = FAIL - Roll (151.46, line 7744) > maximum lean angle (45.00)
Test: Thrust = GOOD -
Test: VCC = GOOD -

Test: Compass = FAIL - Large compass off params (X:-156.64, Y:-133.48, Z:-186.40)
Large compass offset in MAG data:
X: -156.00
Z: -186.00
Moderate change in mag_field (34.06%)

Test: IMU Mismatch = FAIL - Check vibration or accelerometer calibration. (Mismatch: 2.16, WARN: 0.75, FAIL: 1.50)
Hey sorry you waited so long.
The way it looks to me is that your compass was way off, thats beeing said the Controller thought it was some where else.
Not having the right Parameters frpm the compass will let it go haywire.
Also the IMU mismatch is caused from to much vibrations. that is also an issue for it.
The Mismatch triggered the fail safe as the log says.
If the Controller thiks its somewhere else and goes to failsafe it will fly to the coordinats it has.

Test: Pitch/Roll = FAIL - Roll (151.46, line 7744) > maximum lean angle (45.00)
This line right here only indicates the max lean angle which followed because it banked to the right.

Hope you had any luck with your IRIS

Thanks for taking the time to respond. I still wonder what could have caused the vibration levels to be that high, without having changed the props and the controller being mounted on a damped plate. Possibly some other part of the copter pushed against the controller or something. Eitherway, in the end I was able to repair the quad by replacing one arm only! All four props intact after crashing from about 40 meters full power through a tree onto the ground. And that same pixhawk is still flying today without any further incident. Vibrations levels below 3 and compass properly calibrated!

Greetings

[quote=“TheColonel”]Test: Compass = FAIL - Large compass off params (X:-156.64, Y:-133.48, Z:-186.40)
Large compass offset in MAG data:
X: -156.00
Z: -186.00
Moderate change in mag_field (34.06%)

Test: IMU Mismatch = FAIL - Check vibration or accelerometer calibration. (Mismatch: 2.16, WARN: 0.75, FAIL: 1.50)
Hey sorry you waited so long.
The way it looks to me is that your compass was way off, thats beeing said the Controller thought it was some where else.
Not having the right Parameters frpm the compass will let it go haywire.
Also the IMU mismatch is caused from to much vibrations. that is also an issue for it.
The Mismatch triggered the fail safe as the log says.
If the Controller thiks its somewhere else and goes to failsafe it will fly to the coordinats it has.

Test: Pitch/Roll = FAIL - Roll (151.46, line 7744) > maximum lean angle (45.00)
This line right here only indicates the max lean angle which followed because it banked to the right.

Hope you had any luck with your IRIS[/quote]

You know it might sound weird, but lets say one or two screws on a arm or a mount are a bit loose.
Through the vibration the unscrew them self and create even more vibrations.
And think about props too, if they are a bit off and with the Rpm´s they fly, they create so much force that a small ofset on something important can cause vibrations.
Im Working on Military Airplanes and you wouldnt believe what kind of weird stuff happens if only one small part is not in the right place.