DJI Drone Help Forum

Mavic Air went rogue + crashed


#21

I looked at FLY074.DAT some. It doesn’t contain the eventLog stream IMU related errors that FLY083.DAT did. I’m always somewhat reluctant to use the eventLog stream info since it’s different depending on the platform and changes with every FW update. Looks a lot like debug statements that developers use to understand the behavior of the FC. Those developers understand what it means but we’re left trying to guess.

But it does seem that the FLY083 eventLog stream had a lot of ominous sounding statements about the IMU and the problems were IMU related.

I noticed that in FLY074 there were 3 failed attempts at a compass calibration before a 4th successful attempt. Is there something you can add that would explain that?

By way of an example about the pitfalls of interpreting the eventLog stream I had noticed in FLY083 that the calibration was triggered because the MA was 44,586 miles from the last time a calibration had been done. Actually, what it says is that the distance was 71750 km. I suspect that it’s just a units label problem, i.e. it’s 71750 meters (not kilometers). That would be consistent with what’s seen in the M2P.


#22

You could power it up and not fly it and then provide the .DAT. If you’ve traveled back home it will probably ask for a compass calibration.


#23

So, out of curiosity, I’m not sure what this particular portion of the Event Log is stating. The “calibration” point in Long and Lat is where in the middle of the ocean? If this is indeed a “true” Lat and Long value…

157.041 : 7872 [L-COMPASS]mag cali pos and time saved success!
157.041 : 7872 [L-COMPASS][mag_cali_pt]lat:0.452779, lon:-1.398403
157.041 : 7872 [L-COMPASS][mag_cali_pt]height:-4.0, date:20190207


#24

It’s lat/long, but in radians, not degrees. Miami Beach.


#25

I had a feeling that might be either bogus, or a different calculated value but didn’t double check. Thanks. I will put that in the memory bank ( If I remember it’s there…)…


#26

Yes, indeed for some reason it took multiple attempts to calibrate that compass on the '074 flight and not sure why. However it did behave flawlessly during that flight.

Fast forward, got back from my trip today and was able to do a test flight. I again had to try a couple of times before the compass calibrated. I’ll upload the DAT file momentarily. Still not sure it’s behaving 100% normally, but the compass heading seemed correct and it did respond properly to the RC.


#27

Here is the latest file:

I noticed it was saying at one point the RC signal was weak even though I kept the MA directly above me. It’s quite possible that there was some damage from the crash.

Prior to this flight, I did an IMU calibration on a level surface.

For this flight, I was in the middle of an open field, still took a couple tries to calibrate the compass. I did restart the unit in place as suggested. Overall it responded properly although at first did not want to take off (not sure why, plenty of GPS) also did not want to land on the first try.

I did reach out to DJI and send them the DAT and flight record. They said they would be in touch again later this week after the data is reviewed. I’ll keep you all posted when I hear back from them.


#28

Can you provide the .DAT where you did the compass calibration? Probably FLY004 or FLY003


#29

I’ve noticed that the take off altitude is below sea level in all the flights submitted. FLY074 and FLY083 are near the sea so I assumed there must be a high pressure weather system. E.G. FLY074 has GPS:heightMSL = 3 meters and barometric takeoff height = -92 meters. That’s 95 meters difference which is unusually large. But, FLY005 has GPS:heightMSL = 107 meters yet the barometric takeoff height is -70 meters which yields 177 meters difference - I think that’s an error.


#33

Those are the previous files. The system was not letting me link to drop box so I uploaded here. Hopefully that’s ok?

Actual height above sea level here is about 300’


#34

The system flagged your attachments above. I unflagged them, so they should all be displaying now.


#35

Too many .dats on a Sunday eve…Wow…Wave%20the%20Flag


#36

Just an update, I did reach out to DJI. They have been very helpful. I received the following response:

Dear Customer,

Thanks for your patience.

The unfortunate incident that occurred to your aircraft has been confirmed as a warranty case according to our data analysis.