DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Drone dropped from sky over ocean

... so this could have been magnetic interference ...
... which caused the drone to struggle with the GPS until just pushing you into Atti mode ...
... then it just got blown right down into the ocean ...

Magnetic interference have nothing to do with the GPS ... nothing. It will not make the AC "struggle" with the GPS neither when it comes to get sat lock nor the reception quality ... you've totally misunderstood this. And further more ATTI mode keep the AC height stabilized by means of the barometric sensor ... it will not send the AC from 28,4m height down into the sea, the uncommanded height loss started already 3sec before the 0,3sec long ATTI mode period which ended the log.

In the chart below (click on it to make it larger) you can for yourself see that the AC comes into the incident (where the markers are placed) with full elevator (right stick forward, light blue), the AC is making progress according to spec, both tilt angle (black) & speed (purple) wise for P-mode that was used (note that the heading speed (purple) is wrong, take the value times 10 to get it right ... perhaps @BudWalker can explain why it's wrong in CsvView)

... I am struggling with how the loss of prop or motor has anything to do with going into ATTI mode ?

Your struggle most probably have it's cause in you misunderstanding above.

Even though ATTI mode can sometimes be seen in yaw error cases due to a power up in a magnetic disturbed spot (if it's from there you got your wrong thinking ...) but it's not due to the initially set off compass or a loss of GPS reception ... it's due to the FC giving up. All DJI drones revert to ATTI mode when the FC can't cope with the situation, usually it will take some time & during that time most AC have crashed already.

ATTI mode is a kind of GPS mode, in which the AC no longer consider available position data trustworthy or lacking that data (or if your drone have an ATTI swith, just don't use the data for horizontal hold). So you can have ATTI mode because you've lost GPS reception ... or the quality isn't good enough. But you can get ATTI due to that the FC decides that the available position data no longer can be trusted as what ever it commands it doesn't get the desired effect ... like in a yaw error ... or like in this incident, when the AC have lost thrust from one of four sides. This will make the FC not succeed with it's intention to stabilize the AC & keep it airborne.

At the time the drone was upside down.
It wasn't able to recMeive any satellite signals.
No GPS = Atti mode

The AC was never upside down ... after the incident started, the tilt angle (black graph) went to a maximum of 88,9 degrees & a minimum of 34,6 degrees. The #sat lock was 18 (green) during the ATTI mode phase ... and most probably the .DAT log will show that the quality was considered good. But in this case the FC chose to ignore this & rate the NavHealth as 0 due to the explanation above.

1596316974562.png

... I'm slightly apprehensive going forward as obviously a drone randomly falling out of the sky isn't ideal.

Man made thing's will always fail in some point ... props & batteries simplified is what keeps your AC airborne, pay extra attention to them. Internal electronics can't be checked so they can suddenly fail & take your AC down. The only thing you can do to mitigate is to expect the worst & by that consider what you should fly over if it happens.
 
Last edited:
Magnetic interference have nothing to do with the GPS ... nothing. It will not make the AC "struggle" with the GPS neither when it comes to get sat lock nor the reception quality ... you've totally misunderstood this. And further more ATTI mode keep the AC height stabilized by means of the barometric sensor ... it will not send the AC from 28,4m height down into the sea, the uncommanded height loss started already 3sec before the 0,3sec long ATTI mode period which ended the log.

In the chart below (click on it to make it larger) you can for yourself see that the AC comes into the incident (where the markers are placed) with full elevator (right stick forward, light blue), the AC is making progress according to spec, both tilt angle (black) & speed (purple) wise for P-mode that was used (note that the heading speed (purple) is wrong, take the value times 10 to get it right ... perhaps @BudWalker can explain why it's wrong in CsvView)



Your struggle most probably have it's cause in you misunderstanding above.

Even though ATTI mode can sometimes be seen in yaw error cases due to a power up in a magnetic disturbed spot (if it's from there you got your wrong thinking ...) but it's not due to the initially set off compass or a loss of GPS reception ... it's due to the FC giving up. All DJI drones revert to ATTI mode when the FC can't cope with the situation, usually it will take some time & during that time most AC have crashed already.

ATTI mode is a kind of GPS mode, in which the AC no longer consider available position data trustworthy or lacking that data (or if your drone have an ATTI swith, just don't use the data for horizontal hold). So you can have ATTI mode because you've lost GPS reception ... or the quality isn't good enough. But you can get ATTI due to that the FC decides that the available position data no longer can be trusted as what ever it commands it doesn't get the desired effect ... like in a yaw error ... or like in this incident, when the AC have lost thrust from one of four sides. This will make the FC not succeed with it's intention to stabilize the AC & keep it airborne.



The AC was never upside down ... after the incident started, the tilt angle (black graph) went to a maximum of 88,9 degrees & a minimum of 34,6 degrees. The #sat lock was 18 (green) during the ATTI mode phase ... and most probably the .DAT log will show that the quality was considered good. But in this case the FC chose to ignore this & rate the NavHealth as 0 due to the explanation above.

View attachment 109442



Man made thing's will always fail in some point ... props & batteries simplified is what keeps your AC airborne, pay extra attention to them. Internal electronics can't be checked so they can suddenly fail & take your AC down. The only thing you can do to mitigate is to expect the worst & by that consider what you should fly over if it happens.
here is my reccomendation. Props are cheap. I inspect them every flight and look for small cracks in the blade and at the pivot points. I have the props dated and recorded in log. I replace them every 20 hours of flight. I would also replace them if I hit something. At TO I hover at eye level and look at props and motors for trim and sound.

If you take off from the ground, blow the motors out gently. Debris can jump into the motors at TO and landing. I always use a pad to reduce the chances.

There are a lot of single points of failure in a drone. You prevent what you can.

Batteries, look at them cell by cell crequently and a close visual insprction at every charge.

Given that, things fail, that is why you don't fly over people.
 
.... (note that the heading speed (purple) is wrong, take the value times 10 to get it right ... perhaps @BudWalker can explain why it's wrong in CsvView)
....
CsvView just uses the TXTlogToCSVtool converter. It doesn't create the CUSTOM:hSpeed value.
 
  • Like
Reactions: slup
Aha... Have seen this before, always from M2 logs if I remember it correctly. Anyway to correct it?
 
The AC was never upside down ... after the incident started, the tilt angle (black graph) went to a maximum of 88,9 degrees & a minimum of 34,6 degrees.
I used the term upside-down to make the explanation simpler and save myself some typing but if you need to see a more detailed explanation ....
You seem to have forgotten the third axis of movement which is shown by the roll data (blue) which shows values significantly greater than 88° on multiple occasions.
i-t7nDgMS-L.jpg


Shortly before the data shows Atti Mode, the roll data shows the Mavic had rolled 169.6°
That alone is pretty close to being upside-down, but look closer at the data and you see that the drone was rolling and tumbling such that the data points 1/10th of a second apart show big changes between 1/10th second data points.
1/10th of a second after showing +169.6°, it has rolled to -18.2° and similar extremes before and after that.
This is at the same time as the Mavic is spinning and tumbling with extreme, rapid movements in three axes.
The result being that the drone, while tumbling, would have been literally upside-down a number of times and spent a significant part of the rest of the time in positions where the GPS antenna was not oriented towards the sky, and not able to receive clear satellite signal.

There is usually a lag between losing GPS signal and the flight controller before giving up and going to atti mode, as if it's briefly carrying on by dead reckoning in the hope that the signal interruption is temporary.
That will account for what was shown in the recorded data.
 
Last edited:
  • Love
Reactions: Phantomrain.org
May be something like this has happened to the OP's drone. I doubt if anyone can do anything to prevent it except for giving up flying. If the drone is still under warranty, DJI will probably give you a replacement .

 
I used the term upside-down to make the explanation simpler and save myself some typing but if you need to see a more detailed explanation ....
You seem to have forgotten the third axis of movement which is shown by the roll data (blue) which shows values significantly greater than 88° on multiple occasions.
i-t7nDgMS-L.jpg


Shortly before the data shows Atti Mode, the roll data shows the Mavic had rolled 169.6°
That alone is pretty close to being upside-down, but look closer at the data and you see that the drone was rolling and tumbling such that the data points 1/10th of a second apart show big changes between 1/10th second data points.
1/10th of a second after showing +169.6°, it has rolled to -18.2° and similar extremes before and after that.
This is at the same time as the Mavic is spinning and tumbling with extreme, rapid movements in three axes.
The result being that the drone, while tumbling, would have been literally upside-down a number of times and spent a significant part of the rest of the time in positions where the GPS antenna was not oriented towards the sky, and not able to receive clear satellite signal.

There is usually a lag between losing GPS signal and the flight controller before giving up and going to atti mode, as if it's briefly carrying on by dead reckoning in the hope that the signal interruption is temporary.
That will account for what was shown in the recorded data.
No, the AC wasn't nearly upside down.

It's easy to think that looking at only the roll axis which, as you say show +169,6 degrees at most. But this was combined with a pitch angle of +85,4 degrees. This combination of Roll & Pitch don't place the AC with it's back down ... it was most probably positioned with the front side straight up.

This is why I chose to look at the OSD:tiltInclination:C signal ... which is the inclination from the vertical axis combining both Roll & Pitch, so with the AC completely leveled with no roll or pitch gives 0 degrees in tilt inclination.


1596361868022.png
 
I believe the drone entered ATTI mode at the end of the flight because the attitude of the drone was too large for GPS to be used in any meaningful ways, e.g. to hold the position. I can't imagine how the drone can hold position when it's back is facing the ground unless the motors can spin in reverse direction like those drones that can do 3D maneuvers . I am not sure about it but taking GPS out of the equation may remove some algorithmic constraints and allow the flight controller to correct the attitude more quickly.

I dont think not being able the receive GPS signals is the reason because the number of satellites seen was still18 when the drone hit the ground
 
Last edited:
No, the AC wasn't nearly upside down.
It doesn't matter that it wasn't exactly lying on its back with it's feet in the air.
It was tumbling and spinning all over the place as it fell.
Optimum GPS reception is achieved with the top of the drone steady and facing upwards.
This one was in many other positions where it was not able to properly receive satellite signals.
With the signals being rapidly lost and re-established repeatedly, proper GPS reception was eventually lost.
I previously explained how the GPS receiver carries on after losing signal in case the interruption is temporary.
 
Last edited:
It doesn't matter that it wasn't exactly lying on its back with it's feet in the air.
It was tumbling and spinning all over the place as it fell.
Optimum GPS reception is achieved with the top of the drone steady and facing upwards.
This one was in many other positions where it was not able to properly receive satellite signals.
With the signals being rapidly lost and re-established repeatedly, proper GPS reception was eventually lost.
I previously explained how the GPS receiver carries on after losing signal in case the interruption is temporary.
Right ... the true reasons behind the ATTI switch will be unknown as we don't have access to the DAT log. But it's clear that the 0,3sec long ATTI phase didn't cause this incident.
 
  • Like
Reactions: Phantomrain.org
Hi all.Thanks again for looking into this so forensically.

I went diving today and found the drone at a depth of 12.1m pretty close to where I lost GPS. The memory card still works although obviously drone looks in very bad condition (pics attached). Lots of corrosion? on the underside. Two props attached, two missing.
 

Attachments

  • 20200802_194831.jpg
    20200802_194831.jpg
    476.6 KB · Views: 62
  • 20200802_194918.jpg
    20200802_194918.jpg
    851.9 KB · Views: 62
  • 20200802_194843.jpg
    20200802_194843.jpg
    762.8 KB · Views: 62
  • 20200802_194926.jpg
    20200802_194926.jpg
    818.3 KB · Views: 58
  • Like
Reactions: slup
Hi all.Thanks again for looking into this so forensically.

I went diving today and found the drone at a depth of 12.1m pretty close to where I lost GPS. The memory card still works although obviously drone looks in very bad condition (pics attached). Lots of corrosion? on the underside. Two props attached, two missing.

Are you sure that is the right drone, that thing looks like its been down there for years . WOW !
 
Hi all.Thanks again for looking into this so forensically.

I went diving today and found the drone at a depth of 12.1m pretty close to where I lost GPS. The memory card still works although obviously drone looks in very bad condition (pics attached). Lots of corrosion? on the underside. Two props attached, two missing.
glad you got it back , do you have refresh? what is the final thinking on what brought it down?
 
glad you got it back , do you have refresh? what is the final thinking on what brought it down?
You have the most probable (due to that we lack access to the DAT log) cause in post #8
 
Aha... Have seen this before, always from M2 logs if I remember it correctly. Anyway to correct it?
Check out Version 4.0.4 IMUCalcs:VelH
 
  • Like
Reactions: slup
Hi "peterc4". I am doing some research on Mavic 2 drones falling out of the sky for no apparent reas
on.


Tilt2K posted on the sudden loss of his drone over the ocean. "Slup" reviewed the data and determined that it was most likely prop or motor failure.

I documented a case of potential prop failure with "knock off" third party props. The pin that holds the prop to the prop hub fell out while I was handling the drone.

See the link to the post and photos:

If this happened in flight, the loss of a prop could explain the end of flight of the "Tilt2K" drone.

The other cause of sudden disappearance is in-flight battery separation. I recently posted my experience on the General Mavic 2 Discussion forum. I experienced a crash due to the battery ejecting from the drone. Battery with less than 35 charge cycles in good condition, no swelling.

Unless you secure your battery to the drone so that it cannot pop out, you are taking the risk that your drone will crash without warning due to this apparent design flaw.
 
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,593
Messages
1,554,204
Members
159,598
Latest member
fast54