Mavic Mini uncommanded descent tests

[Ice_2k]

Ah - two different issues. What made little difference was applying backward elevator when moving forwards and in uncommanded descent - it didn't arrest the descent. Simply flying backwards doesn't lead to uncommanded descent because it puts much less demand on the rear props.

Got it, so I understood correctly. I was replying to Foxhall saying “Flying the drone backwards does seem to take the MM out of the uncommanded descent”.

 

[sar104]

Got it, so I understood correctly. I was replying to Foxhall saying “Flying the drone backwards does seem to take the MM out of the uncommanded descent”.

Well flying backwards prevents uncommanded descent. It was not high enough in my tests for the aircraft to stop forward motion and go backwards before it hit the ground, but perhaps if it had been higher then that might have been effective.

 

[Ice_2k]

Got it, I’ll keep it in mind if I come across this issue. Thanks again!

 

[FoxhallGH]

Got it, so I understood correctly. I was replying to Foxhall saying “Flying the drone backwards does seem to take the MM out of the uncommanded descent”.

I thought I had seen that mentioned somewhere - so I'm open to correction on that one ... My theory is that IF the uncommanded descent (UD) is caused by the rear motors rev'ing too high and putting deformed prop's into a blade-stall, then the way to recover from the blade-stall is to reduce the rpm on the rear motors. However, that wouldn't reduce the UD due to the Mavic Mini needing positive thrust to recover. It might seem sensible that the only way to get that positive thrust is to neutralise the up-stick and pull reverse-stick - which should reduce the rpm on the 'rear' motors and increase it on the front. The [desired] effect being to 'pop' the MM into level backward flight ...
But - if the Mavic's flight controller has an algorithm that wants to use full thrust all round to arrest the descent - that's going to just keep the rear motors at high rpm & rear prop's in the stall condition - and nothing will change until the MM bounces off the ground. If the MM does the UD from low enough that the ground-strike doesn't kill it, they do tend to bounce back into a low-level hover. That's telling me that the flight controller is seeing the ground impact as a positive change in direction and it reduces motor rpm back to a level that will sustain hovering flight and is before the onset of blade stall.

 

[Ice_2k]

Maybe I’m totally off here but, if the drone is high enough, maybe right stick backwards AND left stick DOWN would force it to drop the RPM of the engines, taking the blades out of stall?

 

[FoxhallGH]

Maybe I’m totally off here but, if the drone is high enough, maybe right stick backwards AND left stick DOWN would force it to drop the RPM of the engines, taking the blades out of stall?

It does depend on how the flight control system of the MM reacts ... If the algorithm 'says' that the descent must be arrested before the change in attitude command can be processed, then the MM will go all the way to the ground before it decides to do anything different. It's something that has to be tried - but in the heat of the moment, the natural reaction to stop the MM falling, is to bend the up-stick against the end-stop! If it's simply that reaction that's keeping the MM in blade-stall and descending, then there are probably several control inputs that could bring it out of it, as long as those inputs didn't involve max power being required on the rear motors.

 

[sar104]

It does depend on how the flight control system of the MM reacts ... If the algorithm 'says' that the descent must be arrested before the change in attitude command can be processed, then the MM will go all the way to the ground before it decides to do anything different. It's something that has to be tried - but in the heat of the moment, the natural reaction to stop the MM falling, is to bend the up-stick against the end-stop! If it's simply that reaction that's keeping the MM in blade-stall and descending, then there are probably several control inputs that could bring it out of it, as long as those inputs didn't involve max power being required on the rear motors.

You can see all these effects happening in the graphs - no speculation needed. The attitude change occurs during the uncommanded descent - it just doesn't arrest the descent, or at least not in time to prevent significant loss of height.

 

[FoxhallGH]

You can see all these effects happening in the graphs - no speculation needed. The attitude change occurs during the uncommanded descent - it just doesn't arrest the descent, or at least not in time to prevent significant loss of height.

Apologies sar104 - My head starts to spin if I look at those graphs - but I do know they deserve study & respect. Thanks.
One question for you - and again, this is probably shown in your data - Is it possible to confirm that there is an rpm point on the rear motors where above that point the prop's go from being able to produce lift, to where the Mavic starts to descend? In other words, I'm wondering if it is possible to determine the rpm at which the rear props become unresponsive/stalled and lose lift?

 

[sar104]

Apologies sar104 - My head starts to spin if I look at those graphs - but I do know they deserve study & respect. Thanks.
One question for you - and again, this is probably shown in your data - Is it possible to confirm that there is an rpm point on the rear motors where above that point the prop's go from being able to produce lift, to where the Mavic starts to descend? In other words, I'm wondering if it is possible to determine the rpm at which the rear props become unresponsive/stalled and lose lift?

It clearly isn't just due to exceeding a certain rpm. Many flights, including those above, show level flight at around max rpm on the rear motors for some period, followed by loss of altitude at a similar rpm, suggesting that further flattening of the props is occurring dynamically in flight. But note that what directly triggers the descent is the resulting reduction in front motor rpm - apparently to try to maintain the forward pitch when it runs out of headroom on the rear motors.

 

[FoxhallGH]

It clearly isn't just due to exceeding a certain rpm. Many flights, including those above, show level flight at around max rpm on the rear motors for some period, followed by loss of altitude at a similar rpm, suggesting that further flattening of the props is occurring dynamically in flight. But note that what directly triggers the descent is the resulting reduction in front motor rpm - apparently to try to maintain the forward pitch when it runs out of headroom on the rear motors.

So it's like you input high-to-max stick-forward, but the rear motors have no more to give, and if the front motors reduce rpm, then rather than going forward, the Mavic goes down! If you are seeing max rpm when the MM is in level flight then that has got to be an alarm signal. Most DJI Mavic drones appear to have a longer flight duration when they are moving [14kph for the max fly time of the MM] rather than hovering, so that would imply there is some load taken off the prop's when the Mavic is flying. So the question is "Why is the MM needing max rpm at the rear motors just to fly level?"
I don't have a set of MM prop's to play with, but if they are made of the same material as the Mav' Pro and Mav' 2 then they will twist longitudinally. However, I would expect that since they are smaller diameter, they should be a bit 'stiffer' than the DJI props and high rpm airflow should not be able to feather much more than the tips of the blades. Even if the tips are feathering, and if they are at all like other DJI props, they will still have an aerodynamic 'wing' cross-section and therefore, should provide some lift.

Does your data show an rpm difference between a 'squashed' set of props and a new set for hover or for a prescribed forward speed?

 

[sar104]

So it's like you input high-to-max stick-forward, but the rear motors have no more to give, and if the front motors reduce rpm, then rather than going forward, the Mavic goes down! If you are seeing max rpm when the MM is in level flight then that has got to be an alarm signal. Most DJI Mavic drones appear to have a longer flight duration when they are moving [14kph for the max fly time of the MM] rather than hovering, so that would imply there is some load taken off the prop's when the Mavic is flying. So the question is "Why is the MM needing max rpm at the rear motors just to fly level?"
I don't have a set of MM prop's to play with, but if they are made of the same material as the Mav' Pro and Mav' 2 then they will twist longitudinally. However, I would expect that since they are smaller diameter, they should be a bit 'stiffer' than the DJI props and high rpm airflow should not be able to feather much more than the tips of the blades. Even if the tips are feathering, and if they are at all like other DJI props, they will still have an aerodynamic 'wing' cross-section and therefore, should provide some lift.

Does your data show an rpm difference between a 'squashed' set of props and a new set for hover or for a prescribed forward speed?

Forward flight typically leads to some aerodynamic lift overall, but always requires more thrust from the rear motors to keep forward pitch - more than in a hover. That's offset by less thrust from the front motors. And just to avoid any potential confusion - level flight is not zero pitch, of course.

The MM props feel extremely flimsy both in torsion and bending compared to MP and M2 props. Most of the lift appears to be generated by 3-D shape, not thickness, and the leading edges are barely thicker than the trailing edges.

 

[FoxhallGH]

Forward flight typically leads to some aerodynamic lift overall, but always requires more thrust from the rear motors to keep forward pitch - more than in a hover. That's offset by less thrust from the front motors. And just to avoid any potential confusion - level flight is not zero pitch, of course.

Yes - definitely understood and on the same page there ... It's just that I'd be surprised that full power would be required from the rear motors to 'fly' the MM level in any other condition than S mode, and full forward-stick. From what you have observed, I'm revising my thinking from the blade stall theory, to blade distortion. It does look like the blades just no longer have the pitch to be fully effective! I'm still mystified though as to what the trigger is that makes the drone go from 'normal' to 'descending' ...

The MM props feel extremely flimsy both in torsion and bending compared to MP and M2 props. Most of the lift appears to be generated by 3-D shape, not thickness, and the leading edges are barely thicker than the trailing edges.

Hmmm - further indication that they could become deformed while warm and compressed. Toy prop's rather than professional!
Since the DJI Stock would form the templates for 3rd party copies, there is little hope of using alternate stock to get different results - except perhaps 'more-reputable' [??] carbon fibre copies which would tend not to flex like the DJI stock.

 

[sar104]

Yes - definitely understood and on the same page there ... It's just that I'd be surprised that full power would be required from the rear motors to 'fly' the MM level in any other condition than S mode, and full forward-stick. From what you have observed, I'm revising my thinking from the blade stall theory, to blade distortion. It does look like the blades just no longer have the pitch to be fully effective! I'm still mystified though as to what the trigger is that makes the drone go from 'normal' to 'descending' ..

I don't think that there is much mystery about the trigger. The props deform and lose thrust, leading the FC to resort to reducing front motor speed to maintain forward pitch.

 

[FoxhallGH]

I don't think that there is many mystery about the trigger. The props deform and lose thrust, leading the FC to resort to reducing front motor speed to maintain forward pitch.

Agreed - but if the rpm isn't changing, what is making the prop's deform?

 

[sar104]

Agreed - but if the rpm isn't changing, what is making the prop's deform?

Creep, I assume. The process I mean - that was not a derogatory salutation.

 

[FoxhallGH]

OK - Is there enough evidence to conclude that the theory of the prop's being compressed and subsequently deformed under the body of the drone, is likely to be the cause of the problem?

 

[sar104]

OK - Is there enough evidence to conclude that the theory of the prop's being compressed and subsequently deformed under the body of the drone, is likely to be the cause of the problem?

The only firm evidence that I have is from my own tests. With initially undeformed props I didn't see any uncommanded descent behavior. After deliberately compressing them against the body of the aircraft for 12 hours it happened almost immediately. So my unproven hypothesis is that compression starts the process, causing the rear motors to spin faster which may further deform the props to the point that the rear motors max out.

 

[FoxhallGH]

The only firm evidence that I have is from my own tests. With initially undeformed props I didn't see any uncommanded descent behavior. After deliberately compressing them against the body of the aircraft for 12 hours it happened almost immediately. So my unproven hypothesis is that compression starts the process, causing the rear motors to spin faster which may further deform the props to the point that the rear motors max out.

Sounds logical and fits the testing. Are you able to post some photographs showing the difference between the deformed and undeformed prop's? e.g. from the side comparing the proper pitch 'twist' to the deformed pitch twist? Thanks

 

[sar104]

Sounds logical and fits the testing. Are you able to post some photographs showing the difference between the deformed and undeformed prop's? e.g. from the side comparing the proper pitch 'twist' to the deformed pitch twist? Thanks

I tried to get some photos, but the difference really doesn't show up well. I'll see if I can find a way to illustrate the difference.

 

[JAW]

@sar104 ——Would you be willing to do the test on a set of the new carbon fiber props for the MM? If so, I’ll have a set sent to you from China.