Mavic Pro Platinum strange crash

[eEridani]

It's not like the old days where an ESC will brute force current to motors. If the ESC has current limiting, which most current/modern ESC's used in commercial aircraft do have, the motor current may well be limited during unusual load events like low RPM and high torque loads as shown in the log. Modern electronics prevent failures by understanding failure conditions and compensating. That's why I am leaning towards something impeding motor rotation, like a flagging propeller blade.

@BudWalker - yeah -after looking at the direction of travel, winds aloft, and the points of acceleration and decel, the blade showing up in the video fits the gimbal pitching up. But it also highlights the moment of extreme acceleration just before the drone starts spiraling out of control. A high instantaneous load on a propeller is when most failures occur. Heck, the drone was out of sight when the event happened, it could even have been a bird impact.

​

 

[eEridani]

I guess you were going to say ... "aren't correct"?

Have never, during several years of incident investigation, seen a log with either telemetry or sensor data that are just "a bit" wrong ... like showing roll to the left instead of right. When data in the logs deviate from reality, it's off by a huge amount ... like in a recent case where the reported height suddenly said negative height with several hundred meters when the flight was conducted on a parking lot.

It's rarely any doubt if the data is reliable or not.


That's all good, but usually a luxury when it comes to a drone that have fallen to ground from 500m height, especially if you don't have the corpse on hand. The logs contain several hundreds of data signals & many of them can be put up against each other in order to determine which is likely & can be trusted. If we, as in this case also have a video capturing the incident ... that of course goes in as one piece of the puzzle & the total analysis.

Already in post #15 I explained what a analysis is ... "you break all available data down into its various elements and then asking critical thinking questions such as WHY and HOW in order to reach some conclusion"


If you look at post #17 you can see a simulation done with a MP moving according to the recorded telemetry ... It's not the prop you see in the video clip from the OP ... it's likely the fuselage.


Without going into what thing's that can block a motor ... a blocked motor always have a certain characteristic influence on the recorded log data. If something by force blocks a electrical motor that is commanded to turn with maximum rpm's that will create a very clear & high amp draw from that motor. This burns the motor eventually as the energy put in doesn't generate rotation ... it instead generate heat in the motor coils.

So if we can't see this amp increase for the stopped motor ... it usually means that the cause isn't anything mechanically blocking the motor, the cause is something else.

And if we there also see log information about ESC failures ... the cause is rather electrical than mechanical.


I'm fully aware of that ... that's why I & several others make analyzes, we don't just throw out opinions as soon as we see a shadow coming into the video frame ... & immediately draw conclusions that's not supported by other data also on hand.

Opinions is actually rarely relevant when it comes to incident analysis ... this as opinions tends to be "you form a view or judgement about something, not necessarily based on facts or knowledge"

Ideas on the other hand ... that needs to be investigated further & either proven or discarded ... yes, all analysis activity start with that.

It still boils down to YOUR interpretation of the data @slup. And your interpretation is not FACT.

From a local dictionary:

... an explanation or way of explaining.
plural noun: interpretations
"this action is open to a number of interpretations"

 

[slozukimc]

I guess you were going to say ... "aren't correct"?

Have never, during several years of incident investigation, seen a log with either telemetry or sensor data that are just "a bit" wrong ... like showing roll to the left instead of right. When data in the logs deviate from reality, it's off by a huge amount ... like in a recent case where the reported height suddenly said negative height with several hundred meters when the flight was conducted on a parking lot.

It's rarely any doubt if the data is reliable or not.


That's all good, but usually a luxury when it comes to a drone that have fallen to ground from 500m height, especially if you don't have the corpse on hand. The logs contain several hundreds of data signals & many of them can be put up against each other in order to determine which is likely & can be trusted. If we, as in this case also have a video capturing the incident ... that of course goes in as one piece of the puzzle & the total analysis.

Already in post #15 I explained what a analysis is ... "you break all available data down into its various elements and then asking critical thinking questions such as WHY and HOW in order to reach some conclusion"


If you look at post #17 you can see a simulation done with a MP moving according to the recorded telemetry ... It's not the prop you see in the video clip from the OP ... it's likely the fuselage.


Without going into what thing's that can block a motor ... a blocked motor always have a certain characteristic influence on the recorded log data. If something by force blocks a electrical motor that is commanded to turn with maximum rpm's that will create a very clear & high amp draw from that motor. This burns the motor eventually as the energy put in doesn't generate rotation ... it instead generate heat in the motor coils.

So if we can't see this amp increase for the stopped motor ... it usually means that the cause isn't anything mechanically blocking the motor, the cause is something else.

And if we there also see log information about ESC failures ... the cause is rather electrical than mechanical.


I'm fully aware of that ... that's why I & several others make analyzes, we don't just throw out opinions as soon as we see a shadow coming into the video frame ... & immediately draw conclusions that's not supported by other data also on hand.

Opinions is actually rarely relevant when it comes to incident analysis ... this as opinions tends to be "you form a view or judgement about something, not necessarily based on facts or knowledge"

Ideas on the other hand ... that needs to be investigated further & either proven or discarded ... yes, all analysis activity start with that.

No sir. I said exactly what I meant to say.

Mike

 

[slup]

No sir. I said exactly what I meant to say.

Mike

In that case we are in agreement ... but I would tend to say that the logs always "aren't incorrect"

 

[slup]

It still boils down to YOUR interpretation of the data...

Sure ... that's why I explain & present the data findings which I base the interpretation on ... & write "is most probably electrical"

...If the ESC has current limiting, which most current/modern ESC's used in commercial aircraft do have, the motor current may well be limited...

...Heck, the drone was out of sight when the event happened, it could even have been a bird impact.

​

You have the same possibility regarding these ideas you present ...

We have seen several proven blocked motor cases here at the forum earlier ... several from newer crafts than a MPP, it's just for you to compare & present what happened to the current draw in those incidents to strengthen your thesis.

The same regarding the bird impact ... as we have the DAT log on hand it would be totally possible for you to analyze the accelerometer data, confirm your idea & present your findings here ...

 

[slozukimc]

In that case we are in agreement ... but I would tend to say that the logs always "aren't incorrect"

Read the whole darned sentence instead of just the parts that make you right.

Mike

 

[slup]

Read the whole darned sentence instead of just the parts that make you right.

Mike

In that case you might need to clarify what point I'm missing ... can't see neither of us being either right or wrong.

 

[slozukimc]

In that case you might need to clarify what point I'm missing ... can't see neither of us being either right or wrong.

I think I got confused with who was saying what when I replied. I just read back through and I don’t see what I thought you were saying so please disregard my comment there.

Mike

 

[orion900]

Hello, I was trying to extract dat file from drone itself, but for some reason it does not have March 19th file (when crash happened) in it, only before and after (after fixing) dates.

 

[BudWalker]

Hello, I was trying to extract dat file from drone itself, but for some reason it does not have March 19th file (when crash happened) in it, only before and after (after fixing) dates.
View attachment 146096

Sometimes the dates are incorrect. Can you select these flights

and then provide the created .DAT via a public sharing site like DropBox or GoogleDrive?

Edit: The Flight Index numbers will be different (shifted up) since you will have powered up the MP since this screen was created. Select the flights that match the dates shown in the red box.

 

[eEridani]

Sometimes the dates are incorrect. Can you select these flights
View attachment 146099
and then provide the created .DAT via a public sharing site like DropBox or GoogleDrive?

Edit: The Flight Index numbers will be different (shifted up) since you will have powered up the MP since this screen was created. Select the flights that match the dates shown in the red box.

Wait, are you saying DJI logging software contains errors as significant as the date of recording?

Sorry - that was sarcasm. But thanks for letting others know that defect is in there.

 

[Jamesz]

Flying at 500m altitude and out of sight is not a responsible way to be flying a drone. Please consider the safety of other aircraft.

 

[stevenws]

It's very clear why the drone start to rotate & lose height at 438,6sec into the flight ... the rear right motor stops rotating & doesn't generate any thrust.

The root cause isn't as simple as a blocked motor or a failing prop though...

If looking at the moment where the main incident starts, at 438,6sec into the flight and check how the problematic motor is behaving where the drone rotation/tumbling starts.

-The green graph is the yaw movement on a 360 degree axis
-The black graph is the motor rpm
-The blue graph is the Flight controllers command percentage for the motor
-And the dashed purple is the rudder stick command

Clearly the flight controller doesn't get what it requires, the rpm falls quickly to near 0rpm's while the FC want's between 92-97% of full rpm. This makes the drone to start rotate due to the torque imbalance & lose height faster then before.

(Click on the chart's below to make them larger)
View attachment 145910

At 462,5sec the rear right motor start to rev up again, but there the drone is down at 15m above the HP & have probably damaged a prop blade against those trees seen in the video clip ... the damage isn't worse than it can produce enough thrust for gaining height ... but it can't stop the rotation due to still having a thrust generated torque imbalance.

View attachment 145911

In the end at 634,7sec, there up on approx 40m height all motors turn off & the drone free falls to the ground.

View attachment 145912

Looking again at the first initial incident we see that this doesn't fully look like a blocked motor ... usually that means that the motor current sky rockets ... but that doesn't happen here, even though the motor doesn't move & the FC wants nearly max revs.

The dashed yellow graph is the rear right current draw ...

View attachment 145913

If looking into the DAT log event stream we see that the problems started off much earlier than when the main incident started ...

Already at 240,3sec into the flight the errors start to build up regarding propulsion troubles ...

View attachment 145914
When the main incident starts this is seen ...

View attachment 145915
And when the motor start to work again ...

View attachment 145916
So the root cause for this whole incident is most probably electrical ... affecting the ESC for the rear right.

That the drone shuts off all motors in the end is due to the wild tumbling which turns the drone upside down ...

View attachment 145917

What a

 

[stevenws]

It's very clear why the drone start to rotate & lose height at 438,6sec into the flight ... the rear right motor stops rotating & doesn't generate any thrust.

The root cause isn't as simple as a blocked motor or a failing prop though...

If looking at the moment where the main incident starts, at 438,6sec into the flight and check how the problematic motor is behaving where the drone rotation/tumbling starts.

-The green graph is the yaw movement on a 360 degree axis
-The black graph is the motor rpm
-The blue graph is the Flight controllers command percentage for the motor
-And the dashed purple is the rudder stick command

Clearly the flight controller doesn't get what it requires, the rpm falls quickly to near 0rpm's while the FC want's between 92-97% of full rpm. This makes the drone to start rotate due to the torque imbalance & lose height faster then before.

(Click on the chart's below to make them larger)
View attachment 145910

At 462,5sec the rear right motor start to rev up again, but there the drone is down at 15m above the HP & have probably damaged a prop blade against those trees seen in the video clip ... the damage isn't worse than it can produce enough thrust for gaining height ... but it can't stop the rotation due to still having a thrust generated torque imbalance.

View attachment 145911

In the end at 634,7sec, there up on approx 40m height all motors turn off & the drone free falls to the ground.

View attachment 145912

Looking again at the first initial incident we see that this doesn't fully look like a blocked motor ... usually that means that the motor current sky rockets ... but that doesn't happen here, even though the motor doesn't move & the FC wants nearly max revs.

The dashed yellow graph is the rear right current draw ...

View attachment 145913

If looking into the DAT log event stream we see that the problems started off much earlier than when the main incident started ...

Already at 240,3sec into the flight the errors start to build up regarding propulsion troubles ...

View attachment 145914
When the main incident starts this is seen ...

View attachment 145915
And when the motor start to work again ...

View attachment 145916
So the root cause for this whole incident is most probably electrical ... affecting the ESC for the rear right.

That the drone shuts off all motors in the end is due to the wild tumbling which turns the drone upside down ...

View attachment 145917

What a thorough Explanation. That's great!

 

[EpicFlight]

It still boils down to YOUR interpretation of the data @slup. And your interpretation is not FACT.

From a local dictionary:

... an explanation or way of explaining.
plural noun: interpretations
"this action is open to a number of interpretations"

No his interpretation is not fact. But the data is.

And his reasoning is sound, and logical, with little assumption. Yours, on the other hand, is largely speculative, with the key point resting on visual interpretation of a recorded video. Already there is plenty of dispute over what you claim you saw.

There's no argument over what the log says.