Mavic Mini strange behavior after take off, drifted away.

[sar104]

I'm very happy for the analysis that you guys all provided, thank you very much.
But now I'm not sure what to do, should I keep flying? Contact DJI?

I think a few basic tests are needed first. For example - in a location well away from any potential magnetic interference, power up the aircraft facing a known direction, let it sit for a few minutes to get position etc., then pick it up and rotate it slowly through several rotations, then put it back down facing the original direction. After that, power it off and retrieve the DAT file.

 

[BudWalker]

I think a few basic tests are needed first. For example - in a location well away from any potential magnetic interference, power up the aircraft facing a known direction, let it sit for a few minutes to get position etc., then pick it up and rotate it slowly through several rotations, then put it back down facing the original direction. After that, power it off and retrieve the DAT file.

And, @rafareusch don't calibrate the compass before this test.

 

[rafare]

I think a few basic tests are needed first. For example - in a location well away from any potential magnetic interference, power up the aircraft facing a known direction, let it sit for a few minutes to get position etc., then pick it up and rotate it slowly through several rotations, then put it back down facing the original direction. After that, power it off and retrieve the DAT file.

I'll do.
Should I consider doing a compass and IMU calibration again?

EDIT:
Just saw BudWalker's post, so I'll do without any type of calibration. I'll keep you guys posted.

 

[Zbip57]

I'd just like to add, although I don't (yet) understand half of what's being said here or displayed on these graphs, how very much I appreciate reading such intelligent and thoughful analysis. You guys are GREAT!

 

[rafare]

Hey guys, Done 2 small flights, hovering at the minimum altitude before landing. Done a lot of rudder actions to try to reproduce the problem.
both flights were 100% normal.
After first flight it asked for a compass calibration. I restarted the Mini to avoid the calibration as you guys said and continued with the second.

(Google earth shows this is a crowded parking lot, due to quarentine there was no one there but me)

 

[sar104]

Hey guys, Done 2 small flights, hovering at the minimum altitude before landing. Done a lot of rudder actions to try to reproduce the problem.
both flights were 100% normal.
After first flight it asked for a compass calibration. I restarted the Mini to avoid the calibration as you guys said and continued with the second.

Did you do the manual rotation test that I suggested?

 

[rafare]

Did you do the manual rotation test that I suggested?

I did, I don't know why but the fly app decided to not log the "flight" that I manually moved the mini around.
The first thing I did was to try the manual rotation you said. I put the controller about 1m away and when I rotated about 90 degrees it asked for a compass calibration. Then I restarted the drone to avoid the calibration, and then continued with the manual rotation.

I tried with turning autosync off to prevent the app from erasing the DAT file. but theres only one file, which is when I actually took off.

I can try it again tomorrow morning, if it would help.

 

[sar104]

I did, I don't know why but the fly app decided to not log the "flight" that I manually moved the mini around.
The first thing I did was to try the manual rotation you said. I put the controller about 1m away and when I rotated about 90 degrees it asked for a compass calibration. Then I restarted the drone to avoid the calibration, and then continued with the manual rotation.

I tried with turning autosync off to prevent the app from erasing the DAT file. but theres only one file, which is when I actually took off.

I can try it again tomorrow morning, if it would help.

There should be a DAT file for each power on event, but txt logs are only created for actual flights.

 

[rafare]

There should be a DAT file for each power on event, but txt logs are only created for actual flights.

Yes, I think I was slow to turn off autosync and it unfortunately deleted the first file.
I'll try tomorrow morning again.

 

[rafare]

Did you do the manual rotation test that I suggested?

Okay, with autosync turned off I managed to get the logs.
Here is the DAT file of me manually turning the Mini about 3 ft from the ground, without doing any compass calibration.

the magnetometer data are horrible and it's rather remarkable that it flew as well as it did.

I'm a bit scared to fly it again considering this statement, if you guys could make an analysis on this log and see if something is really wrong with it or it was just an sporadic thing.

 

[BudWalker]

Okay, with autosync turned off I managed to get the logs.
Here is the DAT file of me manually turning the Mini about 3 ft from the ground, without doing any compass calibration.


I'm a bit scared to fly it again considering this statement, if you guys could make an analysis on this log and see if something is really wrong with it or it was just an sporadic thing.

The magMod data has a wider range than normal, but apparently not too excessive. Not nearly as bad as the incident flight. Has a compass calibration been performed since that incident flight - FLY036 at 2020-5-27 23:1:55 GMT?

And, I know you've answered this but just to be clear. Has the MM ever had an external device attached such as a tracker or strobe?

 

[rafare]

The magMod data has a wider range than normal, but apparently not too excessive. Not nearly as bad as the incident flight. Has a compass calibration been performed since that incident flight - FLY036 at 2020-5-27 23:1:55 GMT?

And, I know you've answered this but just to be clear. Has the MM ever had an external device attached such as a tracker or strobe.

No, Yesterday it asked for a compass calibration, I avoided it by resetting the drone. Still haven't done it.

About something attached, no. I never attached anything but the blade protector that comes with the fly more combo, and the last time I used was at least, 2 months ago.

 

[Deleted member 113929]

I want to thank @rafareusch for the descriptive post, for everyone's comments, with great knowledge and seriousness. I learned a lot from this topic. This forum is special.

 

[BudWalker]

No, Yesterday it asked for a compass calibration, I avoided it by resetting the drone. Still haven't done it.

About something attached, no. I never attached anything but the blade protector that comes with the fly more combo, and the last time I used was at least, 2 months ago.

The problem may be that the magX gain is set too high. Can you do a test where the MM is held level at arms length and pointing magnetic north. Then raise your arms so that it's pointing up 40° and then lower it so it points down 40°. Do this slowly and repeat a couple of times. We're trying to see what the X magnetometers' response is relative to it's alignment with the geomagnetic field. At your location the geomagnetic inclination is about 40°.

I noticed this test was done in a parking lot. I'm hoping there were no cars within 5 meters or so. Can you move to a better location? There could be all kinds of metal under that parking lot. Maybe the open area SE

 

[rafare]

The problem may be that the magX gain is set too high. Can you do a test where the MM is held level at arms length and pointing magnetic north. Then raise your arms so that it's pointing up 40° and then lower it so it points down 40°. Do this slowly and repeat a couple of times. We're trying to see what the X magnetometers' response is relative to it's alignment with the geomagnetic field. At your location the geomagnetic inclination is about 40°.

I noticed this test was done in a parking lot. I'm hoping there were no cars within 5 meters or so. Can you move to a better location? There could be all kinds of metal under that parking lot. Maybe the open area SE
View attachment 102943

I’ll try the test you suggested.

About the parking lot, due to quarentine it is 99% empty, I only have access because I work with healthcare.
Yesterday I parked my car about 5m away, today I parked it more than 10m away from the drone. There is no other car within 50m away. I went away from any pothole or anything metal that I could see. There was nothing around it.One thing that could be a factor is my apple watch, I’ll try to remove it next time.
The open area in SE is the entrance/exit of the parking lot, and today there is a bunch of cones there.
I really wanted to test it on a flat, grass park, unfortunately a bunch of parks here are closed to public and the few that are still open, are crowded.

 

[rafare]

Can you do a test where the MM is held level at arms length and pointing magnetic north. Then raise your arms so that it's pointing up 40° and then lower it so it points down 40°. Do this slowly and repeat a couple of times. We're trying to see what the X magnetometers' response is relative to it's alignment with the geomagnetic field. At your location the geomagnetic inclination is about 40°.

Here's the test you suggested.

In the start I let the drone sit on the ground pointing north (I used my iphone as reference)
The Mini was pointing north during all the test.
After some time I held the drone with my fingers with my hand being on top of the drone (trying to touch it the less I can). I held it arms length as far as straight as I can, as you said.
( I can't imagine what someone walking nearby would imagine trying to figure it out what I was doing pointing to the sky and the ground over and over again, lol)

At some point during the test, I started holding it from the bottom, to see if that has any effect.
At the end I turned it upside down to power it off while holding it above ground.

My car was at least 10 m away and I left the controller and my apple watch inside the car. There was nothing around the drone but me.

The test produced this results:

 

[BudWalker]

Here's the test you suggested.

In the start I let the drone sit on the ground pointing north (I used my iphone as reference)
The Mini was pointing north during all the test.
After some time I held the drone with my fingers with my hand being on top of the drone (trying to touch it the less I can). I held it arms length as far as straight as I can, as you said.
( I can't imagine what someone walking nearby would imagine trying to figure it out what I was doing pointing to the sky and the ground over and over again, lol)

At some point during the test, I started holding it from the bottom, to see if that has any effect.
At the end I turned it upside down to power it off while holding it above ground.

My car was at least 10 m away and I left the controller and my apple watch inside the car. There was nothing around the drone but me.

The test produced this results:
View attachment 102975

OK, the magX values appear to be correct or, at least, consistent with magZ. First, confirming that the geoInclination is 40°. MagZ is close to 0.0 when the pitch is close to 40°


If pitch is 0.0° then the angle between the Z axis and the geomagnetic field is 50°. Then the field strength = magZ/cos(50°) . Here magZ is -851 yielding a value of 1329 for the field strength. Likewise, magX is 1098 which yields a value of 1433 for the field strength (= magX/cos(40° ). The reason these are slightly different and less than the magMod of 1488 is that the MM isn't perfectly aligned with the geoMagnetic field (magY = 534)


What does all this high school trig mean? There doesn't appear to be a problem with the X magnetometer like there clearly was in the incident flight. Since then there hasn't been a compass calibration or HW change affecting the magnetic properties of the MM. Maybe it's time to suggest a compass calibration and call it quits. That is, unless @sar104 can see something else.

 

[rafare]

OK, the magX values appear to be correct or, at least, consistent with magZ. First, confirming that the geoInclination is 40°. MagZ is close to 0.0 when the pitch is close to 40°
View attachment 102980

If pitch is 0.0° then the angle between the Z axis and the geomagnetic field is 50°. Then the field strength = magZ/cos(50°) . Here magZ is -851 yielding a value of 1329 for the field strength. Likewise, magX is 1098 which yields a value of 1433 for the field strength (= magX/cos(40° ). The reason these are slightly different and less than the magMod of 1488 is that the MM isn't perfectly aligned with the geoMagnetic field (magY = 534)
View attachment 102984

What does all this high school trig mean? There doesn't appear to be a problem with the X magnetometer like there clearly was in the incident flight. Since then there hasn't been a compass calibration or HW change affecting the magnetic properties of the MM. Maybe it's time to suggest a compass calibration and call it quits. That is, unless @sar104 can see something else.

Interesting. Anyway, thanks a lot to you everyone that helped with this awesome analysis.

Out of curiosity I did the same test you suggested earlier but on the same location that the incident happened. I’ll compare later with the previous test and see if there’s a difference.
Also, on the first part of this test I did what you suggested and on the final part I did some rotations as suggested by @sar104.
I still haven't calibrated anything, will do it tomorrow.
But anyway, thank you guys very much, I’ll try to calibrate IMU and Compass on a very open space.
You guys are awesome.

 

[Prismatic]

I looked at this yesterday and still haven't figured out what is going on. I've never seen magnetometer data that look like this. ... I'm still trying to figure out what kind of linear combination of magnetic fields could produce that. It looks unphysical to me so far.

You're rarely stumped, @sar104, so that alone makes this event beyond peculiar. But ... it's 2020; the way things are going, I'm not sure I'll be surprised if a fleet of space aliens take over. Actually, I'm not sure they haven't! ?