Compass Error & ATTI

[pamisano]

So, had my Mavic out, looking to get some good video of a waterfall at work. Everything looked good, plenty of satellites. Took off, moved to a good hover location, went to line up the video, and all within 6 seconds got Compass warnings, switched to ATTI and took off back towards me. Obstacle avoidance didn't work and it went head first into a pole. At the last second, I was able to turn the only thing that hit, that I'm aware of, was the right front arm. Even the blades missed and are still in good shape.

Is it normal when there is a compass error / ATTI (although I don't understand why it switched to ATTI when the log shows 14 satellites) for the Mavic to just take off like that? Now, this wasn't a return to home, it was a straight line back.

Any help with the attached log and advice/counseling would be appreciated.

Thanks in advance.

 

[pamisano]

Problem? Fly away

Was unit in a crash? Not bad, but yes.

What have you tried so far? Nothing, trying to understand what happened and avoid it from happening in the future.

What device are you using ( iphone , ipad, Samsung , etc)? Note 8

What firmware are you running ( aircraft, remote controller)? Latest

What Go app version are you using? Latest

Any modification? No

Did you change anything or install any apps? No

Do you have a video or pictures of the problem? (if so post link) Log file attached above
Any help would be greatly appreciated.

 

[aokusman]

Please correct me if I’m not understanding you correctly. You said you flew it back to you, so why did it crash ?

 

[pamisano]

No, it took off on it's own. I had it hovering preparing to take some video, then it started right back at me. I was able to get it to turn slightly before the crash so it didn't smash head on into a pole.

The piece I am curious about is why it went to ATTI mode when it had 14 satellites and when it took off on it's own.

 

[sar104]

It sounds like a fairly classic case of magnetic interference at the takeoff location. The IMU initialized the yaw at -8° (so just west of due north), but after takeoff the aircraft's subsequent motion was not fully consistent with that yaw value - was that the direction it was facing at takeoff?

In order to determine exactly what happened I'd need to see the mobile device DAT file that includes the magnetometer data as well as the IMU yaw values.

 

[pamisano]

Says, the .dat extension is not allowed. Is there a trick to uploading the .dat files?

Yeah, pretty close to north, from what I recall.

 

[sar104]

Says, the .dat extension is not allowed. Is there a trick to uploading the .dat files?

Yeah, pretty close to north, from what I recall.

Most people upload them to Dropbox or similar and then post the shared link. Or since they are the smaller mobile device DATs, you could just change the file extension to .txt.

 

[pamisano]

Here it is w/ .txt extension.

 

[sar104]

Here it is w/ .txt extension.

Right - that shows the problem clearly. The IMU initialized the yaw value to -8° because that was the compass reading on the ground at takeoff. However, there was obviously something distorting the magnetic field at that location - as soon as the aircraft took off and climbed to a height of a couple of meters the compass heading changed to 62°, which was likely the correct value. The IMU rate gyros had not detected any rotation, and so the IMU yaw did not change and the two values now disagreed by about 70°, causing the "compass_error_large at 4.4 s and fusion failure at 8.9 s:

8.942 : 3376 [L-FDI]NS(0) FUSION(0): fault on , magn_heading_err_large



If you had checked that the aircraft orientation arrow matched the actual direction that the aircraft was facing at takeoff you would have detected the problem.

 

[pamisano]

Thanks for the info, hadn't really checked the way the AC was facing versus compass heading in the past. I will definitely do that in the future.

When you say fusion failure, what does that mean or refer to?

Also, from what you stated the issue was most likely at the take off location, not at the hover location, correct?]

And one final question, what tool did you import the .dat file to get that information?

Thanks again for the help!

 

[sar104]

Thanks for the info, hadn't really checked the way the AC was facing versus compass heading in the past. I will definitely do that in the future.

When you say fusion failure, what does that mean or refer to?

Also, from what you stated the issue was most likely at the take off location, not at the hover location, correct?]

And one final question, what tool did you import the .dat file to get that information?

Thanks again for the help!

Fusion is the flight control term for combining high-sample-rate data from the IMU rate gyros and accelerometers with slower data from the magnetometers and GPS to counter the drift and bias in the IMU measurements. It stops working if those data sources disagree substantially.

Yes - the interference was at the takeoff location.

I use DatCon to translate the DAT files.