Mavic3 Cine crash opinions wanted.

[Meta4]

I was told that the drone is initialised on a table, then transferred to hand, and hand launched.

They cover all the bottom sensors with tape

That explains the VPS sensor reading issue I was wondering about.

They say ATTI mode is a bit faster than sport mode, but it traveled faster than expected.

It was very fast and faster than I'd expect for Atti mode (unless there was a strong tailwind).

They had done a hundred ATTI mode flights before this one, and have done a hundred since with a new cine and had no issues.

Flying in atti mode shouldn't cause any problems

IMO the Mavic 3 doesn't have any "boat mode" algorithm like other DJI drones in the past, and it probably had a dodgy IMU going bad around the moment of takeoff.

I've never heard of a "boat mode" and am not sure what you mean, but there was an issue with the IMU.
It looks to me to have been initialised with incorrect directional data.

It launched normally, but at 4.3 seconds the heading data jumps from 193° to 160° without any rudder input.
This strongly indicates that the compass was deflected by a magnetic field at startup.
The IMU took its initial heading data from the compass which was incorrect.
As the drone moves away from the magnetic interference, it returns to normal, but the gyro sensor can't.
This causes a data conflict between compass and gyro sensor.
When the drone senses that it has moved off course and tries to get back on course, it's using incorrect heading data from the gyro sensor which puts the drone further off course.
It tries to correct and gets further off course etc, etc.
It ends up out of control and going much faster than normal, even with no joystick input.
This is called a yaw error and they are often deadly for the drone.

If I was powered up on a table, I'd be very careful to make sure there are no steel or steel fittings or live electric cables close to the drone when it's powered up.

 

[slup]

This case should have needed the DAT log... unfortunately that is encrypted for the M3.

I don't strongly rule out yaw error effects due to a power on near magnetic interference... but honestly doubt it as the first signs of that something was wrong showed up when the M3 still was in Atti mode (which shouldn't assist with positional hold = no automatic motor control = no uncontrolled flight due too that wrong motors are commanded).

Below the 2 major oddities in the flight...

The AC is in Atti mode (through Drone Hacks = still have positional data) & all good from the beginning, the AC symbol is correct directed (yaw angle) in the sat.picture relative both the flight direction & the stick inputs made there, it's apparent that the IMU knows what's happening & responds correctly to the given commands. But when the AC enters the red circled area (still in Atti) it starts to yaw uncommanded CCW even though a command for a slight CW rotation is applied. The winds there was pretty much a direct tail wind from the start point & weak, so it's not likely that the wind turned the AC around by force.

A bit later (in the green circled area) the craft is turned to Sport (with a brief period of Normal mode)... & it's here the AC really goes haywire in a way that is similar to a yaw error due to a power on near magnetic interference.

(Click on the charts below to make them larger)


If we below look at the velocity differences between the GPS & the IMU (which during normal circumstances should be very near 0mph) we see that the first oddity during the Atti mode indeed indicate large deviations... these deviations grows then when the AC is turned to Sport & definitely makes the M3 uncontrollable. (The deviation can also be seen in the different flight paths on the sat.picture, green path is IMU & red is GPS).



My guess to this is a IMU failure... & the reason for that is that the disaster start off when the M3 was in Atti mode, if the problem had been isolated to the Normal & Sport mode I had put my bet on a ordinary pilot induced yaw error due to powering on near magnetic interference.