Air Crash.

[gupppy]

Hi, I had a MA crash a few day`s ago. I this is my favorite route and I have flow it 100`s of times. This is the recording of my last flight. A case has been opened @ DJI. Hopefully some some one could shed some light on what happened. The battery had been charged 175 times. Last MA flight.
A similar thing happened a few day`s ago, but luckily I got the MA back. I replaced the props and re calibrate everything, also refreshed the firmware.. It seem fine and very stable in flight.
It seem that the drone could easily be recovered, but the bush is so thick and after a couple of day`s I lost hope of finding it.
Thanks

 

[Meta4]

Hi, I had a MA crash a few day`s ago. I this is my favorite route and I have flow it 100`s of times. This is the recording of my last flight. A case has been opened @ DJI. Hopefully some some one could shed some light on what happened.

Here's the flight data: DJI Flight Log Viewer - PhantomHelp.com
Things looked normal until around 11:15.5.
You'd just descended from 80 metres to 17 metres and turned the drone when the speed picked up suddenly.
At 11:16 you still hadn't given it any right stick but the drone was making 18 metres/sec (faster than normal max speed in P-GPS).
You gave it some right stick and the speed increased to 24 metres/sec (53 mph) when the data stopped at 11:18.6.
The data stops before the drone crashed - indicated by no sudden changes in pitch, roll and yaw data.
From 11:15.5 until data stops at 11:18.6 the drone is descending although the left stick is in the centre position or pushed forward after 11:16.8.
The cause of the uncommanded descent and increased speed isn't obvious.
It was not due to losing a motor or prop, the drone did not spin at all.

 

[gupppy]

Thank you for your input very insightful, I now hope DJI could clear this up. This happened before and every time I was lucky it did not descent into people.

 

[sar104]

Thank you for your input very insightful, I now hope DJI could clear this up. This happened before and every time I was lucky it did not descent into people.

There are a few possible explanations for this behavior, but the txt log doesn't contain the necessary data to distinguish between them. The problem was a disagreement between the IMU velocity and position data, triggered by a CCW full rotation at 566 seconds:



Looking in position rather than velocity space:



It might have been a yaw error due to an incorrectly initialized IMU yaw value caused by taking off from a magnetically distorted location, a compass calibration issue, or simply a failure of the IMU inertial sensor package. The lack of any problems for most of the flight generally makes the first two options less likely, but IMU failure is much less common. The answer will likely be in the DAT file (FLY060.DAT) data if you want to retrieve and post that.

Mobile device DAT file: How to retrieve a V3.DAT from the tablet

 

[gupppy]

Thanks very helpfull!

 

[gupppy]

On the 2nd Aug 2019I the same happned on a spot some 300 meters away.I was lucky I happned to get the craft to land.I replaced the props calibrated the IMU, compas and re loaded the firmware. I had many flights after the insedent and flew up country no problems.
Thanks, it would be intresting to see what DJI would say.

 

[Skywalker]

Newbie here. I’m not following. Did it fall from the sky...lights out? Or was it shot down?

 

[sar104]

On the 2nd Aug 2019I the same happned on a spot some 300 meters away.I was lucky I happned to get the craft to land.I replaced the props calibrated the IMU, compas and re loaded the firmware. I had many flights after the insedent and flew up country no problems.
Thanks, it would be intresting to see what DJI would say.

Are you going to retrieve and post the DAT file?

 

[gupppy]

Newbie here. I’m not following. Did it fall from the sky...lights out? Or was it shot down?

No not at all, the craft started banking to the one side, increased speed (+80 K/h) and descended. On the 2nd Aug it did a similar thing over the sea and then the speed exceeded 100 K/h!

 

[AnzacJack]

Mavic air at 100km/hr would be impressive to see. I’ve only had my inspires at that speed.
Top speed I’ve had mavic was 96Km/hr

 

[gupppy]

Yes, it was uncontrolled! This seem to happened after a panno there is another clip that happened a month earlier.

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

 

[gupppy]

Are you going to retrieve and post the DAT file?

I have it but I cant`t post it.

 

[gupppy]

this is a screen shot of the Air top speed.

 

[Deleted member 94047]

this is a screen shot of the Air top speed.

At first, I was going to say that the Go app logs flights from the simulator as well and that this top speed may be one from a simulated flight. But then I watched the video you linked above and there it is


May be you have a super drone that DJI produced for internal testing only . Or a faulty one. As much as I want to believe in a Mavic Air that can do 100km/h, I doubt that speed reading is accurate.

 

[AnzacJack]

A big wind would do it. If it was repeatable I’d be impressed

 

[gupppy]

Thanks mate, I was in the bush where the drone fell, when I received you reply. Sorry still no drone. As I said I was lucky not to have hit someone.

 

[gupppy]

Are you going to retrieve and post the DAT file?

I hope this is the right file!

 

[BudWalker]

There are a few possible explanations for this behavior, but the txt log doesn't contain the necessary data to distinguish between them. The problem was a disagreement between the IMU velocity and position data, triggered by a CCW full rotation at 566 seconds:

View attachment 79897

Looking in position rather than velocity space:

View attachment 79898

It might have been a yaw error due to an incorrectly initialized IMU yaw value caused by taking off from a magnetically distorted location, a compass calibration issue, or simply a failure of the IMU inertial sensor package. The lack of any problems for most of the flight generally makes the first two options less likely, but IMU failure is much less common. The answer will likely be in the DAT file (FLY060.DAT) data if you want to retrieve and post that.

Mobile device DAT file: How to retrieve a V3.DAT from the tablet

From the .DAT it appears that the difference between the GPS derived velocity and the IMU velocity is related to pitch and Yaw.

The intervals where the velocity difference is small coincides with pitch close to 0.0. The velocity differences shift depending on Yaw. There must be another dependency besides pitch and yaw that accounts for the change at 250 secs.

Seems also that the compass isn't calibrated correctly as the Yaw/magYaw separation is dependent on Yaw.

 

[gupppy]

thank you sar104 said for the help.

 

[sar104]

From the .DAT it appears that the difference between the GPS derived velocity and the IMU velocity is related to pitch and Yaw.
View attachment 79949
The intervals where the velocity difference is small coincides with pitch close to 0.0. The velocity differences shift depending on Yaw. There must be another dependency besides pitch and yaw that accounts for the change at 250 secs.

Seems also that the compass isn't calibrated correctly as the Yaw/magYaw separation is dependent on Yaw.
View attachment 79952

I agree with both those observations. The uncontrolled flight is clearly triggered by yaw errors initially:




There is a compass calibration issue (or aircraft magnetization perhaps):



But, as you mention, the pitch and roll start to disagree badly as well:





Looking at the raw sensor data, the IMU0 and IMU1 accelerometers agree, as do the z-axis rate gyros. The problem appears to be in the x- and y-axis rate gyros, which is almost certainly what is causing the pitch and roll discrepancies:





When those start to go in opposite directions things don't go well.

Bottom line - I see at least those two problems which I think are independent, and either of which could be somewhat catastrophic.