Mavic Air looks unstable in flight. Why? (demo video inside)

JoelP · Mar 18, 2020

I had a similar problem with my MA where it would yaw spontaneously without any stick input making it unstable in flight. I also tried recalibrate my compass and IMU with no resulting improvement. Then I recalibrated my sticks and the problem finally was resolved.

sar104 · Mar 18, 2020

JoelP said:
I had a similar problem with my MA where it would yaw spontaneously without any stick input making it unstable in flight. I also tried recalibrate my compass and IMU with no resulting improvement. Then I recalibrated my sticks and the problem finally was resolved.

Different problem.

Dimedrol · Mar 20, 2020

sar104 said:
I'd have to take a look at some of my compass calibration logs...

Where can we find them?

And BTW, I have calibrated IMU and tried to fly once again.
Being on start location, I have calibrated my compass - 1st time - calibration failure, 2nd time - [looks like] OK.
Flight conditions was even worse, to be honest... It was yesterday's evening and it was quite dark already (and still windy as my previous attempt)
Here's the log: https://yadi.sk/d/vmIqCFidPPLpYw

Deleted member 94047 · Mar 20, 2020

Dimedrol said:
Where can we find them?

And BTW, I have calibrated IMU and tried to fly once again.
Being on start location, I have calibrated my compass - 1st time - calibration failure, 2nd time - [looks like] OK.
Flight conditions was even worse, to be honest... It was yesterday's evening and it was quite dark already (and still windy as my previous attempt)
Here's the log: https://yadi.sk/d/vmIqCFidPPLpYw

I can't do what @sar104 did in post #15 but going by what is reported in the DAT event stream log, your second compass calibration returned an estimation error value of 0.0.

-27.927 : 5516 [L-COMPASS][scale cali(0)] fill num:[149]
-27.927 : 5516 [L-COMPASS][scale cali(0)] estimation error:[0.0]
-27.927 : 5516 [L-COMPASS][scale cali(0)] succeed! bias:2393.2 -1864.3 -1143.3 scal:3.000 2.633 2.849
-27.927 : 5516 [L-COMPASS][save data] app cali all success
-27.927 : 5516 [L-COMPASS][save data] in user index mode

The first calibration attempt which failed returned an estimation error value of 55.2 and was rejected by the FC.

-51.347 : 4342 [L-COMPASS][scale cali(0)] fill num:[141]
-51.346 : 4342 [L-COMPASS][scale cali(0)] estimation error:[55.2]
-51.346 : 4342 [L-COMPASS][scale cali(0)] status: estimation err large|
-51.346 : 4342 [L-COMPASS][scale cali(0)] status: estimation err large|
-50.607 : 4379 [L-FMU/LED]action changed. mag calibration: error0)

So we can safely assume from the above that the lower the estimation error, the better the accuracy of the calibration. And going by that logic, your second calibration was a very good one.

Now, as to why the flight was unstable, could be a combination of many possible factors. The wind you mention is certainly one factor, GPS signal quality could have been another (you only had 8 GPS sats locked although NavHealth was at 5 which is as good as it gets; so I am not sure if this was an issue here). Any way, I would say leave the compass be and give the bird one or two test flights in a more ideal environment (calm conditions, good GPS reception) before you do anything.

Out of curiosity, I would like to know what you did differently between the first and second calibration attempts. Do you recall anything?

sar104 · Mar 20, 2020

It's clear why the first calibration was rejected and the second one worked:

As for the result, with the calibration points excluded:

It's better. Not perfect, but note that the obvious 2π (360°) periodicity is much smaller, indicating that the aircraft's own magnetic field has largely been compensated.

Here's the previous version again for comparison:

I would check the flight characteristics in light winds before deciding if there is still a problem.

Dimedrol · Mar 20, 2020

Doppler said:
Out of curiosity, I would like to know what you did differently between the first and second calibration attempts. Do you recall anything?

Of course I do. (I wasn't drunk ;-) )
So, at first compass calibration attempt, I tried to hold my drone (for the second "circle") like this: (found video)

"Well, isn't that cute... BUT IT'S WROOOOOONG!!!" (c)

The correct position is -

...like it was written in App: "eyes down".
So, the second time, I've positioned the drone so it "looks down" and the attempt was successful.

It's a windy coronavirus weather last several days in Latvia, so I need to wait a bit for a good conditions.
Guys, thanks so much for such feedback!
I'll keep you informed.
ASAP