Mavic Air 2 unstable flight on return path

[Meta4]

I saw the drone shaking with my own eyes, as I was looking at the 90% mark,

I've gone over the flight data for the whole flight and can't see any instability in the pitch, roll and yaw data.
All the high values correspond to your joystick inputs and there's nothing erratic there.

and also saw the gimbal on screen do weird stuff too. Can't explain why, unfortunately.

It's common for the gimbal to act up in strong side winds, so seeing that isn't anything to be concerned about.

I will conduct a test flight tomorrow and see if I can replicate its behaviour under similar conditions, as it'll be windy during the morning. I will do the same under no wind conditions later in the day, as it will be windless at about 1800, this time recording screen too.

Wind doesn't normally make the drone's flight unstable.
I doubt you'll see anything out of the ordinary.

 

[slup]

...my drone suddenly became unstable...
...A wind warning was issued, despite winds being not that strong, and the drone became jumpy when coming back
...On landing, it would not stop the motors, so I had to stop it with CSC.

Any ideas as to what could have caused it?

The wind was affecting your craft more than you think ...

This was the wind speeds your MA2 faced up there ...



Looking closer into the return path where you said you experienced the problem with instability (along the red arrow in the sat pic.)

As seen here below your MA2 was fighting a strong headwind coming in from the front right ... the pitch angle is really close to the maximum 35 degrees the whole way back & rolling nearly 15 degrees to the right to arrest a course drift... so pitch & roll together is on & sometimes exceeding the max TILT angle specified.



This is the wind & tilt spec. for a MA2 ... meaning that your MA2 was utilizing full S mode tilt even though it was in N mode.



As seen from the graphs the flight was stable ... so craft movements wasn't the reason if the gimbal was moving uncommanded.

The strong wind + that your gimbal was on the HW stop (see pic below) was most probably the reason for any erratic gimbal movements.

With a gimbal on 0 degrees it will reach the HW stop on approx. a craft pitch of 24 degrees (probably a bit more as the 24 is max manual range).


So yeah ... that problem was most probably due to the wind, nothing else.


Then the landing ...

It was dark ... the VPS sensor couldn't work properly & you commanded the MA2 during the landing with your CSC command (which won't work when the craft is airborne or aren't in a (DJI defined) critical state. The CSC command instead made your MA2 yaw CW, descend, go sideways to the left & backwards ... so it's not strange that the landing became rough as the craft had a horizontal speed the whole way down. The landing sequence was actually initiated during your not applicable CSC by that you held the throttle down far enough & over the threshold needed for initiating the landing.

So bad lightning, & a craft still moving horizontally made you get problems during the landing & the rougher touch down made your gimbal overload.

 

[Korrd]

Thanks a lot for the detailed analysis! This answers the original question perfectly.

Ruled as wind-caused, and failed CSC during landing.

 

[Deleted member 103366]

The CSC did not fail. it simply is not permitted to work in normal flight when the relevant settings are the default.
THIS IS AN IMPORTANT POINT to understand and you should check the manual to see how you can change the settings to allow the CSC to work if needed.
I deliberately tested this with a Mavic Mini. I sent it up to 100m+ and WITH THE SETTING AT THE DEFAULT put the sticks in the CSC position and held them there. The drone descended in a perfectly controlled helix until I released the stick at around 2m, the drone then hovered. The drone had even slowed it's descent as the ground came within the range of the VPS. NOTE I used the slowest fight mode for that test and the following.
I did a similar test with the CSC response set to "anytime" but started the CSC at a height of around 2m. The drone again descended under perfect control for the delay period, it dropped once the motors had stopped. The drop was about 18 inches onto grass.
Personally I think that this arrangement is daft, when the response mode to the CSC position is "Anytime" the delay period is too short to leave the response setting at "Anytime" and it would be to easy to trigger a mid air shut down accidentally. I once caught myself holding a Phantom 3's joysticks in the CSC position, fortunately the Phantom 3's delay period is 3 or 5 seconds and a shut down was not triggered.

 

[Korrd]

I've been wondering about that. While I've been testing my drone's limits in order to build trust on it, I am yet to do CSC experiments.

I wonder what is considered an emergency by its onboard software as to have it accept a CSC command. The manual is not super clear on that other than prop failures and related.

 

[tlswift58]

I saw the drone shaking with my own eyes, as I was looking at the 90% mark, and also saw the gimbal on screen do weird stuff too. Can't explain why, unfortunately.

I will conduct a test flight tomorrow and see if I can replicate its behaviour under similar conditions, as it'll be windy during the morning. I will do the same under no wind conditions later in the day, as it will be windless at about 1800, this time recording screen too.
If I can't make it fail, I'll just archive it as a fluke and leave it at that. Otherwise, we should get some more data to analyze.

One of those - who knows things. Would def watch the drone for next few flights to make sure nothing crazy happens and "do not" fly over water to be safe.

But, that was an awesome photo no less. A one in a million shot.

 

[Korrd]

Thanks!

Yeah. Will do a test flight once I'm out of Meetings Tuesday hell.

 

[Deleted member 103366]

I wonder what is considered an emergency by its onboard software as to have it accept a CSC command. The manual is not super clear on that other than prop failures and related.

Precisely.

It seems to me that it would be after the event, which seems daft.

If for example you lost positional control and it was heading for people I would MUCH rather drop it into the ground than stop the motors after it had sliced someone.
In my early days I actually did, or attempted to, deliberately crash a 'run away' Phantom 3 into a wall rather than have it fly, at windscreen height, out over the road it was heading for.
Whether it responded to the closed throttle I can't now remember but it did, thankfully, catch the wall and crash on the safe side of the wall.

With regards to CSC experiments, I initially hand held the Mavic Mini (MM), started the motors and then wobbled my hand. Due to the wobbling, the MM thinks it is flying and fights to hold position, then, whilst still wobbling my hand, I executed the CSC. In the "Emergency only " mode stopping the motors was a rare occurrence and may actually have been due to excessive tilt. (try a search for snatch an twist hand catches)
In "Anytime" mode the motors always stopped......after the delay.
It would be riskier to try this with the Air 2 due to the bigger props etc.
I think I tried it with the Mavic 2 but if I did I had a very firm grip on the Mavic 2 and, asince I remember no difference, the results presumably similar.

 

[Korrd]

Yeah. I always saw a CSC as something you would use in order to avoid causing harm with the drone. Like moving off the path of an aircraft, or prevent it doing damage or hurting someone.

 

[sar104]

Yeah. I always saw a CSC as something you would use in order to avoid causing harm with the drone. Like moving off the path of an aircraft, or prevent it doing damage or hurting someone.

That may count as a legitimate emergency from a safe flight point of view, but it's not an operational emergency situation. The emergency CSC is for use in the event of flight control failure where other failsafe shutdowns don't happen automatically.

 

[Deleted member 103366]

To Sar104, meaning no offense, do you know for a fact that, with the CSC response set to "Emergency Only" etc., a CSC will shut down the motors in the event of a "flight control failure" and what precisely do you mean by "flight control failure"?

 

[sar104]

To Sar104, meaning no offense, do you know for a fact that, with the CSC response set to "Emergency Only" etc., a CSC will shut down the motors in the event of a "flight control failure" and what precisely do you mean by "flight control failure"?

The situations that I'm aware of are blocked motors, IMU errors and compass errors. Inverted flight may also be one, but that's hard to test.

 

[BSauce955]

It became unstable at the point marked 90% battery. That's what made me go back.


Near landing in the bay was not so for landing, but to capture this beautiful pic:

No way?! That picture has got to be fake!

 

[BSauce955]

The situations that I'm aware of are blocked motors, IMU errors and compass errors. Inverted flight may also be one, but that's hard to test.

Inverted flight is possible?!

 

[sar104]

Inverted flight is possible?!

If a motor fails or it loses a prop - yes.

 

[BSauce955]

If a motor fails or it loses a prop - yes.

Oh, I thought you meant like inverted flight but like normally. Would it be possible if the motor cap things (the parts that have white and black rings to show which props go where) were switched or would the drone do automatic shutdown since it's literally upside down?

 

[sar104]

Oh, I thought you meant like inverted flight but like normally. Would it be possible if the motor cap things (the parts that have white and black rings to show which props go where) were switched or would the drone do automatic shutdown since it's literally upside down?

It won't auto-shutdown in flight just because it goes inverted, but it won't arm the motors if its attitude exceeds certain limits (70° on most firmware I think).

 

[Capt KO]

If in doubt, do an imu calibration. Can’t hurt and often solves unusual behavior.
Good luck and keep up the water pics. Enjoyed it and the story behind it. Another learning experience for a fellow A2 pilot.

 

[Korrd]

Will do. I'm on a world tour, so plenty opportunities to take great pics!

I do post them to my instagram. space.indiana.jones

 

[BSauce955]

It won't auto-shutdown in flight just because it goes inverted, but it won't arm the motors if its attitude exceeds certain limits (70° on most firmware I think).

So there's no way to take-off upside down?