How to protest a DJI crash analysis?

[GerdS]

Thanks for analyzing for me. I think the biggest thing I've learned is not to launch off of concrete. I'm not sure if it ever goes zooming off on it's own again in the future if I can leave my hands off the controls rather than trying to save it.

The other thing you should learn is how to fly your drone in Atti mode, without GPS support. This capability would have saved your Mavic.
Best training option is by buying a little toy drone for 20-50 bucks and training how to fly it with various orientations of the drone towards yourself. Don't use headless mode for training.
There is also no need for an onboard camera and video transmission for this purpose.

 

[sar104]

I should point out that the analysis of this event that you guys have been discussing is incorrect.

Firstly, the aircraft never switched to ATTI mode - it was in GPS ATTI (aka P-GPS) throughout the flight until it entered forced landing mode at the end. The satellite count wasn't great (9 - 12) but it was good enough, and GPS health never dropped below 4. It won't switch until it drops to 3.



Secondly, there were no obviously catastrophic compass issues. If you compare GPS location with the integration of the IMU north and east velocities, then the agreement is good enough. The aircraft behavior does not appear to have been caused by IMU yaw vs. compass yaw discrepancies.



If you look at the comparison of aircraft speed with stick inputs, it appears that the initial movement of the aircraft was caused by full backwards elevator at 6.5 s. The faster movements after that were due to the attempted CSC.



On that subject, the CSC failed because the right stick was not pulled fully inwards. Note that if your CSC attempt really was both sticks in and down, then it appears that you were flying in mode 3, with elevator and aileron on the left stick - is that correct?



The remaining puzzle regarding this flight is the yaw dataset, which shows very strange uncommanded periodic oscillations. That's going to require the mobile device DAT file to diagnose properly.



In summary, the log suggests that the behavior of the aircraft is largely accounted for by an attempted CSC. However, the compass data are strange and need further evaluation, requiring the mobile device DAT file.

 

[SteveH]

The other thing you should learn is how to fly your drone in Atti mode, without GPS support. This capability would have saved your Mavic.
Best training option is by buying a little toy drone for 20-50 bucks and training how to fly it with various orientations of the drone towards yourself. Don't use headless mode for training.
There is also no need for an onboard camera and video transmission for this purpose.

I like your suggestion but actually I started with toy drones before I ever purchased the Mavic. When the drone takes off flying in opposite directions from stick inputs you will be surprised at how quickly you loose control of it. Especially when it starts erratic flight straight from lift off.

 

[Simmo]

I should point out that the analysis of this event that you guys have been discussing is incorrect.

Firstly, the aircraft never switched to ATTI mode - it was in GPS ATTI (aka P-GPS) throughout the flight until it entered forced landing mode at the end. The satellite count wasn't great (9 - 12) but it was good enough, and GPS health never dropped below 4. It won't switch until it drops to 3.

View attachment 42472

Secondly, there were no obviously catastrophic compass issues. If you compare GPS location with the integration of the IMU north and east velocities, then the agreement is good enough. The aircraft behavior does not appear to have been caused by IMU yaw vs. compass yaw discrepancies.

View attachment 42474

If you look at the comparison of aircraft speed with stick inputs, it appears that the initial movement of the aircraft was caused by full backwards elevator at 6.5 s. The faster movements after that were due to the attempted CSC.

View attachment 42475

On that subject, the CSC failed because the right stick was not pulled fully inwards. Note that if your CSC attempt really was both sticks in and down, then it appears that you were flying in mode 3, with elevator and aileron on the left stick - is that correct?

View attachment 42482

The remaining puzzle regarding this flight is the yaw dataset, which shows very strange uncommanded periodic oscillations. That's going to require the mobile device DAT file to diagnose properly.

View attachment 42481

In summary, the log suggests that the behavior of the aircraft is largely accounted for by an attempted CSC. However, the compass data are strange and need further evaluation, requiring the mobile device DAT file.

I really want to have your babies!

 

[SteveH]

I should point out that the analysis of this event that you guys have been discussing is incorrect.

Firstly, the aircraft never switched to ATTI mode - it was in GPS ATTI (aka P-GPS) throughout the flight until it entered forced landing mode at the end. The satellite count wasn't great (9 - 12) but it was good enough, and GPS health never dropped below 4. It won't switch until it drops to 3.

View attachment 42472

Secondly, there were no obviously catastrophic compass issues. If you compare GPS location with the integration of the IMU north and east velocities, then the agreement is good enough. The aircraft behavior does not appear to have been caused by IMU yaw vs. compass yaw discrepancies.

View attachment 42474

If you look at the comparison of aircraft speed with stick inputs, it appears that the initial movement of the aircraft was caused by full backwards elevator at 6.5 s. The faster movements after that were due to the attempted CSC.

View attachment 42475

On that subject, the CSC failed because the right stick was not pulled fully inwards. Note that if your CSC attempt really was both sticks in and down, then it appears that you were flying in mode 3, with elevator and aileron on the left stick - is that correct?

View attachment 42482

The remaining puzzle regarding this flight is the yaw dataset, which shows very strange uncommanded periodic oscillations. That's going to require the mobile device DAT file to diagnose properly.

View attachment 42481

In summary, the log suggests that the behavior of the aircraft is largely accounted for by an attempted CSC. However, the compass data are strange and need further evaluation, requiring the mobile device DAT file.

Thank you, thank you, thank you for checking this out. I've went over this event so many times in my head wondering what happened or what did I do. So it appears I did at first try to land it by full down throttle and when it didn't seem to respond I attempted forced motor shut down. Maybe I didn't get the stick fully there as I watched in horror as my bird headed for a crash. I'm surprised I didn't break the sticks off trying to force it down faster. Where do I find my DAT file to upload?

 

[SteveH]

I should point out that the analysis of this event that you guys have been discussing is incorrect.


View attachment 42475

On that subject, the CSC failed because the right stick was not pulled fully inwards. Note that if your CSC attempt really was both sticks in and down, then it appears that you were flying in mode 3, with elevator and aileron on the left stick - is that correct?

No, I have never switched those settings, throttle and yaw on left stick.

 

[sar104]

Thank you, thank you, thank you for checking this out. I've went over this event so many times in my head wondering what happened or what did I do. So it appears I did at first try to land it by full down throttle and when it didn't seem to respond I attempted forced motor shut down. Maybe I didn't get the stick fully there as I watched in horror as my bird headed for a crash. I'm surprised I didn't break the sticks off trying to force it down faster. Where do I find my DAT file to upload?

Instructions to retrieve the DAT file are here. You won't be able to upload it to this site with the .DAT extension - change that to .txt first.

On the RC mode, your attempted CSC was recorded as full negative input on the elevator, throttle and rudder (nearly), and full positive on the aileron. That's not consistent with mode 1 down and inwards, which would be negative elevator throttle and aileron, with positive rudder. Please check the mode setting on the RC too.

 

[SteveH]

Instructions to retrieve the DAT file are here. You won't be able to upload it to this site with the .DAT extension - change that to .txt first.

On the RC mode, your attempted CSC was recorded as full negative input on the elevator, throttle and rudder (nearly), and full positive on the aileron. That's not consistent with mode 1 down and inwards, which would be negative elevator throttle and aileron, with positive rudder. Please check the mode setting on the RC too.

 

[SteveH]

I should point out that the analysis of this event that you guys have been discussing is incorrect.

Firstly, the aircraft never switched to ATTI mode - it was in GPS ATTI (aka P-GPS) throughout the flight until it entered forced landing mode at the end. The satellite count wasn't great (9 - 12) but it was good enough, and GPS health never dropped below 4. It won't switch until it drops to 3.

View attachment 42472

Secondly, there were no obviously catastrophic compass issues. If you compare GPS location with the integration of the IMU north and east velocities, then the agreement is good enough. The aircraft behavior does not appear to have been caused by IMU yaw vs. compass yaw discrepancies.

View attachment 42474

If you look at the comparison of aircraft speed with stick inputs, it appears that the initial movement of the aircraft was caused by full backwards elevator at 6.5 s. The faster movements after that were due to the attempted CSC.

View attachment 42475

On that subject, the CSC failed because the right stick was not pulled fully inwards. Note that if your CSC attempt really was both sticks in and down, then it appears that you were flying in mode 3, with elevator and aileron on the left stick - is that correct?

View attachment 42482

The remaining puzzle regarding this flight is the yaw dataset, which shows very strange uncommanded periodic oscillations. That's going to require the mobile device DAT file to diagnose properly.

View attachment 42481

In summary, the log suggests that the behavior of the aircraft is largely accounted for by an attempted CSC. However, the compass data are strange and need further evaluation, requiring the mobile device DAT file.

I'm very perplexed as to why it appears I was flying in mode 3. I flew just minutes before this crash but landed because the gimbal apparently had not calibrated correctly and was tilted in the horizontal axis. So I landed, only shut down the Mavic, left the controller and iPhone on, restarted the Mavic let it connect, did a compass calibration, set it down on the concrete driveway to wait for GPS lock. Never changed any other settings in the DJI Go app.

 

[SteveH]

I'm very perplexed as to why it appears I was flying in mode 3. I flew just minutes before this crash but landed because the gimbal apparently had not calibrated correctly and was tilted in the horizontal axis. So I landed, only shut down the Mavic, left the controller and iPhone on, restarted the Mavic let it connect, did a compass calibration, set it down on the concrete driveway to wait for GPS lock. Never changed any other settings in the DJI Go app.

I have never switched from mode 2 to any other mode.

 

[7thstring]

Did you CSC bottom inwards or bottom outwards? I guess that's the question. How did you start it in the first place?

 

[7thstring]

Assuming you were in mode2 (default), by what sar104 is saying you might have gone for the first image, rather than the second.

 

[sar104]

Assuming you were in mode2 (default), by what sar104 is saying you might have gone for the first image, rather than the second.

That would explain the values, but the OP clearly said down and inwards so I had discounted that possibility.

 

[SteveH]

Did you CSC bottom inwards or bottom outwards? I guess that's the question. How did you start it in the first place?

Sadly I can't remember which way I did either at this point. Normally I hand launch with the auto take off method but this time I launched from the concrete because of the gimbal issue I wanted to launch from a flat surface. I don't remember if I did auto take off or manual.

 

[7thstring]

Based on the readings you attempted all the way outward, but might have failed on positioning the left stick (the rudder).
So in turn you might have told the AC to lose altitude, go backwards, yaw left (gently ) and move sideways to the right, all at the same time.
You said the AC went sideways on its own before the attempted CSC, was it the left or right (relative to his heading)?

(Anyway one things is calibrating on a flat surface and another is how you launch. If the gimbal is level it will keep regardless of the hand launch).

 

[SteveH]

Based on the readings you attempted all the way outward, but might have failed on positioning the left stick (the rudder).
So in turn you might have told the AC to lose altitude, go backwards, yaw left and move sideways to the left, all at the same time.
You said the AC went sideways on its own before the attempted CSC, was it the left or right (relative to his heading)?

(Anyway one things is calibrating on a flat surface and another is how you launch. If the gimbal is level it will keep regardless of the hand launch).

It went left and aft initially. If I remember correctly it was a clockwise yaw at the same time.

 

[7thstring]

Actually, it look like you:
1) pushed both sticks all the way down for a split second
2) pushed again all the way down and halfway outward (sideways right, yaw left)
3) while keeping down and backwards you pushed halfawy inwards (sideways left, yaw right)
4) and ultimately tried the outward move, but possibly failing on the left stick by a millimeter

This reading screams panic all the way, for at least 6 goods seconds you commanded full backwards descent meanwhile. Never occurred to me, however if I were losing control I would try to gain height hoping he fixes itself far from danger, not to have it land immediately.

 

[sar104]

Sadly I can't remember which way I did either at this point. Normally I hand launch with the auto take off method but this time I launched from the concrete because of the gimbal issue I wanted to launch from a flat surface. I don't remember if I did auto take off or manual.

If you look at the first graph in post #22 you will see that you used auto takeoff.

 

[Kilrah]

A bad stick calibration could prevent you from reaching the CSC point.
If you start the motors with a CSC it won't work so you notice it quick enough, but if you use the app button to take off you may never realise sticks are badly calibrated.

Also unless you switch the setting to always allow CSC emergency shutdown it only lets you do it "in case of flight controller malfunction" which nobody knows what actually means. Very likely this scenario is not considered a "malfunction" and thus the CSC won't do a thing. Which is why I set it to "always".

 

[SteveH]

View attachment 42492

Actually, it look like you:
1) pushed both sticks all the way down for a split second
2) pushed again all the way down and halfway outward (sideways right, yaw left)
3) while keeping down and backwards you pushed halfawy inwards (sideways left, yaw right)
4) and ultimately tried the outward move, but possibly failing on the left stick by a millimeter

This reading screams panic all the way, for at least 6 goods seconds you commanded full backwards descent meanwhile. Never occurred to me, however if I were losing control I would try to gain height hoping he fixes itself far from danger, not to have it land immediately.

I will whole heartily agree with panic! It was dusk on July 4th. I thought it was going to go fly away from into the night!