Mavic Air 2 got crazy and crushed

[sar104]

If you are ever unfortunate enough to suffer a bad yaw error, you probably wouldn't have any time to respond anyway and any attempted control imputs won't have much effect.
The speed is always a big shock.
It's best (and easy enough) to avoid getting into that situation.

That’s when you need ATTI on the mode switch.

 

[GFields]

Great help! This step is included in my pre-flight checklist!

 

[cgmaxed]

Hello everybody yesterday I tried to fly my MA2 around 21.30, it told me compass error re-calibrate so I did and I got the success message, later when I lifted it I saw it didn't follow my commands and when I tried to call it down, it got a crazy turned right, started to speed up and lower and hit a tree. Here is the log file, can you see it and determine what might got wrong?

I doubt it's a compass problem. In my experience, I fly all the time with compass errors on other drones(not DJI). DJI makes you fix the compass before you can fly it. My non dji drones fly normally even with compass errors. However, I fly mostly manually. Maybe if your flying with automated piloting , it may affect it, but I doubt it. I can't tell what happened by the logs. Where you in manual control or was the computer flying it?

 

[Meta4]

I doubt it's a compass problem.

No it wasn't a problem with the compass.
It rarely ever is.

In my experience, I fly all the time with compass errors on other drones(not DJI). DJI makes you fix the compass before you can fly it.

A compass error doesn't mean there's a problem with the compass.
It means the compass is working properly and it's warning you of a problem.

Maybe if your flying with automated piloting , it may affect it, but I doubt it.

Whether you fly manually or not won't make any difference if you have a yaw error.

I can't tell what happened by the logs.

But others can and have ... see post #8.

 

[sar104]

I doubt it's a compass problem. In my experience, I fly all the time with compass errors on other drones(not DJI). DJI makes you fix the compass before you can fly it. My non dji drones fly normally even with compass errors. However, I fly mostly manually. Maybe if your flying with automated piloting , it may affect it, but I doubt it. I can't tell what happened by the logs. Where you in manual control or was the computer flying it?

It was unambiguously a compass error. This behavior has been fully analyzed and documented in hundreds of cases where the DAT files were available, and the signature is obvious, especially since the actual aircraft heading at takeoff was reported by the OP. Your anecdotal accounts of flying with compass errors on non-DJI drones are not relevant, and I suggest that you read up about how DJI FC firmware handles this situation - it's described in the discussion linked in post #8.

 

[cgmaxed]

No calibration of the IMU is needed ... I said IMU initialization.

It's the IMU that tells the flight controller in which direction the AC points during flight ... for this to work the IMU needs to be initialized to the correct & true direction ... that info is delivered by the compass in the power on moment. So if the compass is deflected due to magnetic disturbance during power on the IMU will be wrongly initialized ... in other words, the AC will not have the correct info about which direction it points.

To be sure that the IMU initialization have been done correctly ...

After powering on your drone, connected to your RC/app & placed it in the take-off spot ... but before lift-off, ALWAYS check that the drone icon on the map in your app is pointing equal in relation to other objects in the map as the drone does in reality ... if not, abort launch attempt, POWER DOWN & move away, power up again and repeat.

Below was your situation at take-off ... if the AC was pointed to North in reality, below is a clear indication that the IMU was initialized wrongly ...

View attachment 128910



All a compass calibration does is to let the compass know the amount of magnetic disturbance that the AC internals generate ... so it can be deducted & let the compass measure the earth mag field correctly. So ... if you haven't added or taken off any equipment from the AC a compass calibration is rarely needed (some firmware's request a calibration each 30 days or depending on distance from the last power on location though ... but actually not needed due to those reasons).

So ... always distrust a compass calibration prompt in the app, always power down & move away do another take-off location & try again.

That's true. I notice compass errors whenver I am near metal or attach certain metallic payloads to my DJI drone and other drones. I never correct a compass error unless DJI won't let me take off without fixing it first. But I do move to a location with the least compass interference first. My EVO2 doesn't require compass fix when it says the compass is in error. It flies normally with no issues. However, I don't fly in any automated modes. I always fly manually.

 

[cgmaxed]

It was unambiguously a compass error. This behavior has been fully analyzed and documented in hundreds of cases where the DAT files were available, and the signature is obvious, especially since the actual aircraft heading at takeoff was reported by the OP. Your anecdotal accounts of flying with compass errors on non-DJI drones are not relevant, and I suggest that you read up about how DJI FC firmware handles this situation - it's described in the discussion linked in post #8.

I suppose your correct. Sounds like you have educated yourself regarding this issue much more than I have. May I ask? Why does my EVO2 not need the compass corrected? It flies normally, even when I lift off with a compass error. The error has been sporadic and I have never calibrated it. I guess your saying the DJI compass system in a DJI is much more intertwined with it's function than an EVO2.

 

[cgmaxed]

Actually it has everything to do with the problem, because it confirms the incorrectly initialized IMU heading, which was the direct cause of the crash.

Provided that your luck holds, you may be happier being lucky than knowledgeable.

This is what I read from DJI: "Without a reliable compass signal to the flight controller, the drone loses definintive cardinal sense of direction (North/South/East West). The solid state gyros should sense yaw (course) rotation but not be able to bring it back on a definintive compass/magnetic heading." Sounds like the drone won't know which way is home if you initiate RTH. I read you should fly it manually in cases like that. No pressing the RTH. YOu can lose it I guess.

 

[Meta4]

That's true. I notice compass errors whenver I am near metal or attach certain metallic payloads to my DJI drone and other drones.

That's normal and indicates your compass is working and warning you.

I never correct a compass error unless DJI won't let me take off without fixing it first.

Recalibrating the compass cannot "fix" the problem that the compass is warning you about.
Switching off and moving away from the problem is the proper action.

My EVO2 doesn't require compass fix when it says the compass is in error.

A compass error doesn't mean there's something wrong with the compass.

 

[Meta4]

This is what I read from DJI: "Without a reliable compass signal to the flight controller, the drone loses definintive cardinal sense of direction (North/South/East West). The solid state gyros should sense yaw (course) rotation but not be able to bring it back on a definintive compass/magnetic heading." Sounds like the drone won't know which way is home if you initiate RTH. I read you should fly it manually in cases like that. No pressing the RTH. YOu can lose it I guess.

You are rather confused about the compass and what it does.
Try reading the first post in this thread to learn a lot about your drone's compass.

A short explanation of compass function, calibration and errors.

BACKGROUND The aircraft compass is often misunderstood, both in terms of function and purpose, but it is important to understand what the compass measures, why, and what happens to that information. At the most basic level the purpose of the compass is to measure the earth’s magnetic field in...

mavicpilots.com

 

[sar104]

I suppose your correct. Sounds like you have educated yourself regarding this issue much more than I have. May I ask? Why does my EVO2 not need the compass corrected? It flies normally, even when I lift off with a compass error. The error has been sporadic and I have never calibrated it. I guess your saying the DJI compass system in a DJI is much more intertwined with it's function than an EVO2.

I've never flown the EVO series, but it would be very surprising if the flight control system were not basically similar. These are all derived from the original open-source systems such as ArduPilot, and in order to execute what is generically called "Loiter mode" they have to know both position and orientation. When in flight, the primary orientation data come from the rate gyros, but since those measure turn rather than orientation, they have to be set up with an initial heading, which comes from the compass. It's directly analogous to setting the gyro compass on an aircraft to the actual magnetic heading before takeoff.

If that is done incorrectly (due to magnetic interference on the ground) then the FC is not facing the direction that it thinks it is facing. As a result, when it tries to correct for drift it applies thrust in the wrong direction, and enters a positive, rather than negative, feedback loop.

There are different strategies to correct for this, and maybe that's where the Autel aircraft differ from DJI. DJI aircraft continuously use the magnetic heading to make gradual corrections to the IMU heading, mostly to account for gyro drift and bias. That method fails if the disagreement becomes too large to quickly, such as when abruptly leaving an area of magnetic interference. The Autel firmware may more proactively detect the discrepancy and correct the IMU heading rapidly back to match the magnetic heading. The Mavic 2 aircraft will do that, but only immediately after takeoff. The two strategies have pros and cons.

This is what I read from DJI: "Without a reliable compass signal to the flight controller, the drone loses definintive cardinal sense of direction (North/South/East West). The solid state gyros should sense yaw (course) rotation but not be able to bring it back on a definintive compass/magnetic heading." Sounds like the drone won't know which way is home if you initiate RTH. I read you should fly it manually in cases like that. No pressing the RTH. YOu can lose it I guess.

It's much worse than just not knowing how to get home - it leads to unstable flight whenever the FC attempts to hold position or head in any direction. The only solution is to go to ATTI mode ASAP.

 

[wardtom084]

I am wondering if increased altitude would have avoided the problem long enough to do a quick land and find out what is going on.

 

[Meta4]

I am wondering if increased altitude would have avoided the problem long enough to do a quick land and find out what is going on.

No, it wouldn't

 

[sar104]

I am wondering if increased altitude would have avoided the problem long enough to do a quick land and find out what is going on.

Without switching to ATTI mode there would be no way to do anything resembling a controlled landing.

 

[Kotsos5]

Without switching to ATTI mode there would be no way to do anything resembling a controlled landing.

First of all thank you very much for your help, your experience and knowledge is a great help for me, after learning from my mistake and found out the reason of my problem and what to do to avoid it, I want to ask you at around 10" the controller turned grey and lost connection with aircarft even if it was very very close, could I do something else at that moment? And lastly can MA2 get into ATTI mode?

 

[sar104]

First of all thank you very much for your help, your experience and knowledge is a great help for me, after learning from my mistake and found out the reason of my problem and what to do to avoid it, I want to ask you at around 10" the controller turned grey and lost connection with aircarft even if it was very very close, could I do something else at that moment? And lastly can MA2 get into ATTI mode?

I don't see any disconnect in the log - it appears to be complete all the way to the crash, which was at 14 s.

 

[EricJT]

North

When I bought my M2P 2 years ago, I read all of the Mavic forums and this issue of magnetic interference got my attention.

Sar104 was a terrific source of information....a cardinal rule I follow is I point the AC north while on the ground and then compare that to my RC's display to ensure it indicates north as well. That way I know the AC's FC has the proper information.

BTW, I have a very good innate sense of north and the other points of the compass, so my procedure for takeoff is pretty simple for me. I credit spending my non-adult years living in Northern Ontario for instilling in me a strong sense of compass directions.

 

[Kotsos5]

I don't see any disconnect in the log - it appears to be complete all the way to the crash, which was at 14 s.

Oooh trust me on that my memory is still fresh, after it started to take the "banana" route with high speed the controller turned grey like it is when drone and controller are out of range and it's auto RTH.

 

[sar104]

Oooh trust me on that my memory is still fresh, after it started to take the "banana" route with high speed the controller turned grey like it is when drone and controller are out of range and it's auto RTH.

I can't help you. The controller and app were connected to the aircraft all the way to impact - had they not been then the log would have ended before impact. And I never trust pilot recollection over the logs - it has been demonstrably wrong so many times.

 

[Meta4]

Oooh trust me on that my memory is still fresh, after it started to take the "banana" route with high speed the controller turned grey like it is when drone and controller are out of range and it's auto RTH.

The data is very clear.
It shows that the app was in contact with the drone for the full 14 seconds of teh flight.
It gives a whole row of numbers for 45 data fields for every 1/10th of a second.
The drone accelerating without you touching the right stick would have surprised you.

The drone lifted off at 2.4 sec.
The uncontrolled flight only started at 8.7 sec, slowly at first with the drone 20 metres up.
At 10.2 sec the drone had accelerated to 4.6 m/sec and you started trying to bring it back down and turn it with the rudder.
Your eyes would have been locked on to the drone.
At 14.0 sec the drone crashed and the data stops at 14.2 seconds.

From start to finish the whole incident was over in less than 4 seconds.
It's unlikely that you took your eyes off the drone as it accelerated to 9.8 metres/sec without you touching the right stick, while you were pulling the left stick hard down to bring the drone back.
Your strong memory of a grey screen was most likely the first thing you saw after the crash,
You wouldn'y have had a chance to look at the screenduring the shock of the runaway.