DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Mavic Air drifted aggressively during RTH

Tiago Miranda

Member
Joined
Feb 13, 2019
Messages
13
Reactions
5
Hi,

I have a Mavic Air for almost 16 months. Last week I have experienced an abnormal behavior for the first time.

I was flying at about 130 feet high with little to no wind condition, good satellite and remote control signals. I was in the field side of the city with no wifi noise. I was taking some pictures and then decided to return home pressing the RTH button in the remote. The RTH altitude was set to 164 feet so I was expecting that the aircraft would climb to the desired altitude and return straight home. That was not what happened.

Right after pressing RTH the aircraft started to drift sideways and the camera view indicated that it was tilting aggressively. Than I've lost direct line of sight when a Mountain became between me and the A/C. For around 20 seconds I've lost control and video feedback.

After the blackout the video was live again while it was returning home. Another strange thing is that despite the log replay indicates that it returned home facing straight forward, it actually returned sideways.

After landing I've inspected the entire device and there was no signals of bird strike nor mechanical failure. Unfortunately It was not recording during the event.

I kindly ask for help to try to understand what may have happened.
 

Attachments

  • DJIFlightRecord_2020-03-23_[07-36-29].txt
    1.4 MB · Views: 16
  • DJIFlightRecord_TXT_DAT.zip
    4.8 MB · Views: 9
Last edited:
This is an interesting one. That kind of behavior generally results from a yaw error, and that's exactly what happened here but with an unusual cause. Looking at the IMU, magnetic and computed gyro yaw values shows the following:

IMU_calcs1.png

Disregard the vertical offsets, but note that at 434 seconds the active IMU (1) shows a significant CW rotation, clearly based on the IMU1 magnetic yaw, while IMU0 magnetic and gyros, and the IMU1 gyros, all indicate a CCW rotation.

That disagreement implies that IMU1 misinterpreted the magnetic field due to incorrect pitch and roll readings, and that turns out to be the case. Comparing the IMU1 and IMU0 attitude record, there is reasonable agreement until around 400 seconds, after which the IMUs start to disagree:

attitude_overview.png

Looking closer at that time period shows the pitch and roll discrepancies that led to the magnetic yaw discrepancy:

attitude_mag.png

So there appears to have been a problem with IMU1. It may be hardware, or it may have been a computational problem of the kind that has been seen before on the M2 and, if I recall correctly, the MA as well. I would suggest an IMU calibration but it is also possible that it was an isolated FC processor issue.
 
This is an interesting one. That kind of behavior generally results from a yaw error, and that's exactly what happened here but with an unusual cause. Looking at the IMU, magnetic and computed gyro yaw values shows the following:

View attachment 97814

Disregard the vertical offsets, but note that at 434 seconds the active IMU (1) shows a significant CW rotation, clearly based on the IMU1 magnetic yaw, while IMU0 magnetic and gyros, and the IMU1 gyros, all indicate a CCW rotation.

That disagreement implies that IMU1 misinterpreted the magnetic field due to incorrect pitch and roll readings, and that turns out to be the case. Comparing the IMU1 and IMU0 attitude record, there is reasonable agreement until around 400 seconds, after which the IMUs start to disagree:

View attachment 97815

Looking closer at that time period shows the pitch and roll discrepancies that led to the magnetic yaw discrepancy:

View attachment 97816

So there appears to have been a problem with IMU1. It may be hardware, or it may have been a computational problem of the kind that has been seen before on the M2 and, if I recall correctly, the MA as well. I would suggest an IMU calibration but it is also possible that it was an isolated FC processor issue.
Saw this already late yesterday evening ... & became really confused & chickened out, instead placed this thread under monitoring waiting for you to chime :D

My bet yesterday was also a IMU(1) failure ... but more from a large extent gut feeling then totally out from pure log data.
 
  • Like
Reactions: sar104
Thank you so much Sar104 for spendding your time in such a great analysis.

After reading your investigation I think I have some additional information to include.

That drone has spent about 40 hours in air. Half of it's life flying with Master Airscrew propellers. The gains were changed as manufacturers suggestion. Pitch from 100 to 85 and Yaw from 100 to 95.

No crashes registered in it's record, except for a little bird strike that only produced a little glitch in the recorded video. Actually two attempts in the same flight.

In the first flights I've experienced an uncommanded yaw of around 10 degrees to the right when commanding full pitch forward right after take off. That was solved after an IMU calibration.

I will try another IMU calibration as suggested and keep it close until become confident again to push the envelope.

Thanks again for your contribution.
 
...an uncommanded yaw of around 10 degrees to the right when commanding full pitch forward right after take off.

Just that small yawing is somewhat a common (typical..?) behavior from the MA ... have my self experienced it with my Air both during manual ascent from start but also during auto take off totally without any stick inputs at all.

If you do a search for "uncommanded yaw mavic air" you will find numerous of cases ... haven't found a single example that this have causes any incident, it's more of a annoyance.

Here is one thread about it --> Mavic Air turns (yaws) by itself - no indicated stick input in logs
 
  • Like
Reactions: Tiago Miranda
Hi,

Today the same issue occurred at the same location. This time I have the recorded video also.
I would appreciate very much if you could help me figure this out.
 

Attachments

  • 20-08-04_Logs.zip
    4.1 MB · Views: 6
Hi,

Today the same issue occurred at the same location. This time I have the recorded video also.
I would appreciate very much if you could help me figure this out.
This is mainly a copy of your earlier flight you did in March ... the flight path went over the same area as earlier but this time you flew a shorter distance out until the problems started.

You might as well look back at @sar104 conclusion in post #2 as exactly the same happened in this flight as in the one in March. And even though you haven't provided the .TXT log to check which IMU that was active I believe it again was IMU1 that failed in this flight also as that IMU doesn't match the gyros, the compass & IMU0 ... so by this, setting off a yawerror of a half turn at worst.

Have a hard time to think that the area itself have anything to do with this, it was just a coincident ... it's more probable that the IMU1 is faulty.
 
  • Like
Reactions: Tiago Miranda
Thanks a lot for the analysis slup. I will try to replace the IMU. I'll post my results when I get it back in air.
 
OK. It was not an easy task to get another IMU. I've first bought an used one from China but it wouldn't start. Then yesterday I've received another one from eBay and I'm finally back in the sky.
Spent 3 full batteries this morning in the same area and couldn't notice any issue. I will keep monitoring and reporting my results here.

Thanks again for your collaboration.
 
  • Like
Reactions: dawgpilot and slup
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,585
Messages
1,554,102
Members
159,586
Latest member
DoubleBarS