Mavic Mini uncommanded descent tests

[sar104]

After a little experimentation I was able to replicate the uncommanded descent problem under 0400 firmware by loading the rear props for a few hours as might happen in the flymore case. The props did appear slightly flattened compared to the front props. The problem quickly manifested itself in sport mode. As soon as full forward speed was commanded the aircraft very briefly pitched forwards to the expected 30°, but immediately reduced to less than 15°, with the rear motors spooling up to 17000 rpm. After several full elevator passes the problem became worse and the aircraft descended to the ground uncontrollably, bouncing back into the air from the soft grass fairway.

Note that the first few uncommanded descents occurred when the elevator was centered from full forwards, while the last two started under full elevator, with the aircraft slowing and descending. Releasing the elevator did not arrest the descent, and nor did the application of up throttle, but it would climb again after bouncing. It's clear from the log that the descent occurs while the back motors are still at full speed, and is actually caused by the front motors slowing, presumably to try to maintain forward pitch. It is eventually corrected by the rear motors slowing down and the front motors speeding up again, which allows it to climb. It really appears that the rear motors are not very effective at the highest motor speeds, but the details of that relationship are going to require some careful further testing to tease out.



I'll do some more tests tomorrow, then try replacing the props with new ones, and then upgrade and test the 0500 firmware.

No Mavic Minis were injured in the conduct of this experiment.

 

[MAvic_South_Oz]

Well done sar104, good to see something solid in evidence.
If this proves right and you can't replicate with new props, it leaves one to consider the best action for mini owners.

As the pros are screwed on, loctite thread, etc, they are obviously need to be left on in general for storage between flights, so the storage space will have to be assessed for keeping any stress off the props, particularly the rear.

Keep your mini safe tomorrow !
So after touch down and bouncing up, it was normal once more until sports mode / stick centred again ?
Did you try easing off the stick, rather than letting it go ?

 

[sar104]

Well done sar104, good to see something solid in evidence.
If this proves right and you can't replicate with new props, it leaves one to consider the best action for mini owners.

As the pros are screwed on, loctite thread, etc, they are obviously need to be left on in general for storage between flights, so the storage space will have to be assessed for keeping any stress off the props, particularly the rear.

Keep your mini safe tomorrow !
So after touch down and bouncing up, it was normal once more until sports mode / stick centred again ?
Did you try easing off the stick, rather than letting it go ?

As you can see from the graph, I always used somewhat gradual stick transitions. After the uncommanded descents it was normal in the sense that it would climb again, but it continued to be incapably of achieving more than around 15° of forward pitch. You probably also noticed that I forgot to test backwards pitch - I'll check that tomorrow.

 

[sli-woody]

That is awesome work SAR!
I made a video about this exact problem:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

A thought I had, one that I cannot prove, is that with the blades being so thin, that maybe when the blades have been flattened out, they invert or begin to raise upwards at the tips.
If that makes any sense at all^..
Basically as the motors spin faster the outer half of the rear props begin to fold upwards.
Would need a high speed camera to see.

 

[RayOZ]

Thanks for doing the experiment. I wonder at such high rev on the rear motors and how thin the props are, perhaps the high centripetal force deforms the props flatter, thus losing lift. Getting flatten while kept in the case doesn't help either.
Interesting that DJI might be aware of the issue, hence the label about storing the drone with props folded in a certain position, something probably most people didn't read, like manuals.

 

[droneloverair]

That is awesome work SAR!
I made a video about this exact problem:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

A thought I had, one that I cannot prove, is that with the blades being so thin, that maybe when the blades have been flattened out, they invert or begin to raise upwards at the tips.
If that makes any sense at all^..
Basically as the motors spin faster the outer half of the rear props begin to fold upwards.
Would need a high speed camera to see.

This i believe this theory.

because the mini has such a soft bendable blades, when it gets too high rev, it will lead to stress and fatigueing. Bending straight or bending upwards. Making the propellers pushing upwards rather than downwards.

 

[droneloverair]

Or.
Because the warpage of the propellers the motor tries to spin at higher speed, stressing the motors/over heats and goes into protection mode thus tries to descend to avoid damage.

so the fix is buy an aftermarket stiffer propellers.

 

[sli-woody]

Or.
Because the warpage of the propellers the motor tries to spin at higher speed, stressing the motors/over heats and goes into protection mode thus tries to descend to avoid damage.

so the fix is buy an aftermarket stiffer propellers.

We need someone with the problem and a high speed camera LOL

 

[sar104]

We need someone with the problem and a high speed camera LOL

I can do that.

 

[droneloverair]

Who was to the first to encounter this problem anyways?

iv been eyeing a new mavic mini combo but everytime i think of it i could buy a used mavic air combo for closeat the same price.

 

[sar104]

Who was to the first to encounter this problem anyways?

iv been eyeing a new mavic mini combo but everytime i think of it i could buy a used mavic air combo for closeat the same price.

Are you looking for patient zero?

 

[sar104]

I repeated the tests this morning, with some additional maneuvers. Firstly, the rear props clearly recovered slightly overnight, and it took a few minutes of sport mode flying to get back to uncommanded descent, even though slow forward speeds were present from the start.

Below is the same basic graph that I posted above for the previous test:



Note the addition of a full backwards elevator run - it generated constant pitch and achieved the full rated sport mode speed of around 13 m/s, with front and rear motors running at similar rpm. That occurred with less than the maximum rated pitch of 30°, possibly due to decreased drag at the flight elevation of around 7200 ft MSL.

The forward runs all started with full pitch, but it was immediately scaled back to 12 - 15°, achieving less than even the rated P-mode forward speed. It was also accompanied by noticeable pitch oscillation that became progressively worse. The final forward run in that log shows almost immediate speed loss and rapid descent. I was unable to arrest the descent - the aircraft crash landed and the FC killed the motors:

205.068 : 11831 [L-FDI][CTRL]: fault on , impact_in_air​

205.085 : 11832 [L-FMU/MOTOR]impact near ground req reset craft​

205.085 : 11832 [L-FMU/MOTOR] Stop. reason:impact near ground​

Once again it is clear that the uncommanded descent is caused by the FC decreasing the front motor speeds in an attempt to maintain pitch when it cannot increase the rear motor speed any further. The variation in the amount of decrease is clear indication of variation in the thrust produced by the rear motors - they are definitely deforming in flight when run at high rpm - a condition almost certainly itself initially caused by deformation of the rear props in storage. Whether high rpms alone can deform initially undeformed props is still an open question.

While the problem is clearly caused by a mechanical issue with the props, the firmware response is not optimal. The obvious way for the FC to compensate is to reduce the rear motor speeds rather than attempting to maintain pitch by reducing the front motor speeds. That, of course, will cause the aircraft to slow down, but not descend. I suspect that the new firmware is aggressively using that strategy - hence the low forward speeds being seen by a number of users. If that's correct then speed flying backwards should be fine. That indicates that I need to run another test to apply backward elevator when the uncommanded descent starts, which may override the FC motor control.

More analysis to follow...

 

[sli-woody]

I repeated the tests this morning, with some additional maneuvers. Firstly, the rear props clearly recovered slightly overnight, and it took a few minutes of sport mode flying to get back to uncommanded descent, even though slow forward speeds were present from the start.

Below is the same basic graph that I posted above for the previous test:

View attachment 99688

Note the addition of a full backwards elevator run - it generated constant pitch and achieved the full rated sport mode speed of around 13 m/s, with front and rear motors running at similar rpm. That occurred with less than the maximum rated pitch of 30°, possibly due to decreased drag at the flight elevation of around 7200 ft MSL.

The forward runs all started with full pitch, but it was immediately scaled back to 12 - 15°, achieving less than even the rated P-mode forward speed. It was also accompanied by noticeable pitch oscillation that became progressively worse. The final forward run in that log shows almost immediate speed loss and rapid descent. I was unable to arrest the descent - the aircraft crash landed and the FC killed the motors:

205.068 : 11831 [L-FDI][CTRL]: fault on , impact_in_air​

205.085 : 11832 [L-FMU/MOTOR]impact near ground req reset craft​

205.085 : 11832 [L-FMU/MOTOR] Stop. reason:impact near ground​

Once again it is clear that the uncommanded descent is caused by the FC decreasing the front motor speeds in an attempt to maintain pitch when it cannot increase the rear motor speed any further. The variation in the amount of decrease is clear indication of variation in the thrust produced by the rear motors - they are definitely deforming in flight when run at high rpm - a condition almost certainly itself initially caused by deformation of the rear props in storage. Whether high rpms alone can deform initially undeformed props is still an open question.

While the problem is clearly caused by a mechanical issue with the props, the firmware response is not optimal. The obvious way for the FC to compensate is to reduce the rear motor speeds rather than attempting to maintain pitch by reducing the front motor speeds. That, of course, will cause the aircraft to slow down, but not descend. I suspect that the new firmware is aggressively using that strategy - hence the low forward speeds being seen by a number of users. If that's correct then speed flying backwards should be fine. That indicates that I need to run another test to apply backward elevator when the uncommanded descent starts, which may override the FC motor control.

More analysis to follow...

That sir is some awesome testing! You deserve a medal.
I have noticed another thing with my Mini that could be relate but I am not 100% sure, see what u think:

When I used the drone for the first month, I noticed that it was very stable and able to hold pretty much a straight line.
Slowly over time I have noticed that the drone is becoming less stable, especially when flying directly forward. It sways left and right (even with zero wind speed)
I don't know whether this is just me getting used to the drone or a related problem to the props or me storing the drone in the case.
The drone appears to fly more stable if I fly backwards.
Has anyone else noticed this, or am I just losing my marbles?

 

[sar104]

That sir is some awesome testing! You deserve a medal.
I have noticed another thing with my Mini that could be relate but I am not 100% sure, see what u think:

When I used the drone for the first month, I noticed that it was very stable and able to hold pretty much a straight line.
Slowly over time I have noticed that the drone is becoming less stable, especially when flying directly forward. It sways left and right (even with zero wind speed)
I don't know whether this is just me getting used to the drone or a related problem to the props or me storing the drone in the case.
The drone appears to fly more stable if I fly backwards.
Has anyone else noticed this, or am I just losing my marbles?

That could be due to the props - you would need to take a look at the flight logs to be certain.

 

[rfc]

So is a possible workaround until the problem is sorted out, to immediately turn 180 degrees and start flying backwards to where you want to go at full speed, lol?

 

[sar104]

So is a possible workaround until the problem is sorted out, to immediately turn 180 degrees and start flying backwards to where you want to go at full speed, lol?

Yes - provided you don't need to see a forwards camera view then that will work.

 

[rumitcu]

While the problem is clearly caused by a mechanical issue with the props, the firmware response is not optimal. The obvious way for the FC to compensate is to reduce the rear motor speeds rather than attempting to maintain pitch by reducing the front motor speeds. That, of course, will cause the aircraft to slow down, but not descend. I suspect that the new firmware is aggressively using that strategy - hence the low forward speeds being seen by a number of users. If that's correct then speed flying backwards should be fine. That indicates that I need to run another test to apply backward elevator when the uncommanded descent starts, which may override the FC motor control.

More analysis to follow...

Looking forward to the additional analysis and any testing you do with the new firmware.

On behalf of all the Mini owners who have experienced uncommanded descents, thank you for performing this testing and sharing the results.

 

[RichR]

Many thanks for carrying out the research to confirm some of the theories that have been around. Whilst I've not experienced the issue myself (only had about 15 flights since new, due to weather and only being able to fly in/above my garden), it's something I'd been monitoring on the forums even before I bought my Mini.

Would it be fair to say that the issue is less likely to occur at slower speeds? If so, people who spend most of their time in C or P mode in light winds may potentially never see the issue?

 

[Xanadu]

Thank you very much for taking the time and thoroughly testing out this issue. Your assistance on this is invaluable.

Reading through the posts in the forum about this issue that has plagued many MMs, was my concern that I had when purchased mine.

 

[ALan_m10]

Would it be fair to say that the issue is less likely to occur at slower speeds? If so, people who spend most of their time in C or P mode in light winds may potentially never see the issue?

I wish this was the case. I had an uncommanded descent last week while in C taking some video of a local harbor. Unfortunately being at a harbor you can guess what happened next.

Luckily DJI have concluded that it is a warranty issue and are sending me a new MM but it is certainly something that needs to be fixed soon given the amount of posts about it!