Help me understand Mini2 crash.

[Mynah]

So after 300kms of happy Mini1 flying I upgraded to a Mini 2, only to crash it a week later.
The short story:
No wind.
Launched from exact spot I've launched from a hundred times without magnetic interference problems.
I launched, checked that home point was set while hovering.
I started ascending to gain height as always before going anywhere.
The Mini 2 went haywire, flew over the wall and crashed into my neighbor's house.
I have a friendly neighbor so retrieved my drone with gimbal cable ripped.
What I can gather from the flight record:
Everything went according to plan up to 5.1 sec into the flight, then 'stick input' went ballistic.
Now all I did different from all other successful flights from the same spot was that I tweaked the stick response in Fly app. I did however do an extensive flight after that without problem.
Link to Flight log viewer: DJI Flight Log Viewer - PhantomHelp.com
Link to CSV file: 2022-09-07_18-30-08_v2.csv
Thanks in advance for any input as to how I can prevent this from happening again. And if I can understand it I might prevent a recurrence.

 

[slup]

After powering on the drone, did you check that the drones direction on the app map was equal to how the drone was pointing in reality... was the drone pointing to south east as PhantomHelp log player shows when the flight starts?

Based on that you applied negative elevator (right stick backwards) to stop the uncommanded horizontal movement, I bet the drone didn't backed away from you as Phantomhelp shows, instead it was speeding away nose first... the deviation between the log player and reality then means that your drone had a yaw error of about 180 degrees... most probably due to a power on close to something ferromagnetic disturbing the compass which in turn made the IMU yaw to be wrongly initialized.

 

[Mynah]

Thanks for the insight Slup. The drone take-off position was as shown in the log player. I'm baffled since I was not wearing clothing that contain magnets or metal, not even a watch and it is my usual take-off spot and I powered op right there where I always do. After all the reading I did here on the forum it seems to be yet another typical 'yaw crash', only I have NO idea what confused the Mini 2.

 

[Meta4]

The drone take-off position was as shown in the log player.

But what direction was the drone facing,at launch time?
Was it toward the SE as shown on Phantomhelp?

I'm baffled since I was not wearing clothing that contain magnets or metal, not even a watch

Those things won't cause a problem.
Steel objects close to the launch point will.
What was the launch surface?
What steel might have been nearby (within a few feet)?

 

[Mynah]

But what direction was the drone facing,at launch time?
Was it toward the SE as shown on Phantomhelp?

Yes, it was pointing SE, always take off with nose pointing in that direction when flying at dusk because of my fixed backyard spotlight position. And my launch pad was my usual chipboard sheet with carpeting on top, positioned exactly where I always put it. I have done another flight few hours before that with exact same setup, settings and also Litchi as app, only difference with fatal flight was that I used a Mini battery, was about to test it's endurance versus the Mini 2 battery.

 

[Mynah]

I have done a lot of flights with the Mini batteries in the Mini 2 but this mishap has clearly nothing to do with battery. Nothing metal-wise has moved or changed in my backyard either, closest metal object to my launch pad is my washing line, 4 m away but has not interfered with any of my previous flights, Mini included.

 

[Mynah]

Things happen in a blur when it goes wrong, but I NEVER apply full stick forward OR backwards until I have well cleared my house and trees after take-off. The data shows full throttle being applied from 5.1 seconds into the flight. I might be wrong, but as far as I can recollect and because of habit I'd believe that was not me

 

[Mynah]

OK guys, my bad.
I've just out of desperation reviewed the security camera footage in my back yard. I powered the drone down to insert SD card after realizing it was still in my PC. I then powered the drone back on, holding the remote controller and phone right next to it, deviated from my own rule to put the controller down next to me while powering on the drone. I kept the controller in proximity of the drone while it booted because I for some reason decided to have another glance at the drone's back end, probably to check the SD card was properly seated or to double check the battery cover. Guess this was my magnetic interference. I then plonk it down facing NW since I'm now chuffed that I've got the SD card sorted and can't wait to take off, again breaking my own rule of taking off with it facing SE. I then start up motors and checking out of good habit that it was facing SE in my app and proceeded to take off to hover with the sweet 'home point updated' song in my ears . The rest is history and a broken Mini 2. My app said SE, the drone WAS in fact facing NW when I took off, can see it on the security camera. Expensive slip-up. Typical IMU confusion explained and yet again pilot error.
Thanks for the help guys, I'll do better next time.
I did not put the controller down next to me on the grass because of the subconscious realization that my dog was doing cartwheels around me for some funny reason while I was busy with the drone. Again, my fault, cannot even blame the dog

 

[Meta4]

The data shows full throttle being applied from 5.1 seconds into the flight. I might be wrong, but as far as I can recollect and because of habit I'd believe that was not me

I would think it was you making those joystick inputs, but they have nothing to do with the incident.
Throttle is pushing the left stick up to make the drone climb.
You pushed the left stick up and the drone climbed.
The issue was the drone already flying horizontally at 7 mph (12.1 sec) before you pushed the right stick forward.
From 12.1 sec you pulled the right stick back and hard to the left and held it there until the drone crashed at 16 seconds.
This was what caused the drone to accelerate backwards and crash.

There was a small yaw error situation which caused the drone to move slowly without joystick input, but the high speed and crash was contributed to by by the pilot.

 

[Donnie Frank]

What is the WiFi situation like in that neighborhood? Because WiFi is now so pervasive, I run WiFi forensics before every flight. It's mind blowing how much interference there is out there compared to just 5 years ago.

D

 

[Mynah]

There are a few spots around me with high interference, like at a filling station with an outside ATM not far from here, guess they use an uber wi-fi signal to stay connected but I'm lucky to live on the edge of town with unpopulated space next to me. This specific incident was definately not RF signal related, even the old Mini does fine here without occusync.
Most of my intricate flying gets done in the bush on hiking trials where there's not even cellphone signal but agreed, radio wavelength pollution in general has snowballed during the past 10 years.

The Mini 2 is boxed and on its way to have the gimbal ribbon cable replaced. I'm still in total disbelief as for the extent of lack of logical control during the fatal 15 sec flight and the uncontrolled acelleration that took place. I have practised ATTI mode flying in an open field to get a feel for it but even then did the Mini listen to my stick commands.

 

[slup]

What is the WiFi situation like in that neighborhood?...

Wifi interference will never make a drone fly in any direction uncommanded...at most it will swamp the connection to the degree that you loose it and the drone failsafes and RTH (if the default setting is used).

... I'm still in total disbelief as for the extent of lack of logical control during the fatal 15 sec flight and the uncontrolled acelleration that took place. I have practised ATTI mode flying in an open field to get a feel for it but even then did the Mini listen to my stick commands.

You can't draw a parallell between ATTI mode & a yaw error...

During ATTI nothing helps the drone hold position, it will easily be affected by outside forces like wind & will drift with it if you don't actively brake and resist the drift. During ATTI the drone also lack the possibility to brake after you release the sticks, this as the drone doesn't have any positional information at all.

If you instead have a yaw error the drone have positional information... but the flight controller have wrong information about how the drone points, in reality the drone points north for example, but the flight controller thinks it points south (180 degree wrong, as in your incident). All will be good as long as the flight controller doesn't need to resist a drift & hold position (no positional error have occurred). But as soon as the drone starts to get pushed out of position by the wind the flight controller will react & rev up certain motors to get back to position... but as there is a yaw error the wrong motors will be used & the positional error will be larger instead of smaller... & the flight controller will again try the same wrong motors but this time even more forceful but the positional error just grows and all repeats over & over again ... and in a second the drone speeds away, often exceeding both specified max speed & tilt angle. During happenings like this stick commands comes second and the effect from them comes with a delay.

All that happened during your incident are very typical for a yaw error...

 

[Mynah]

Thanks Slup, makes sense now, other than preventing it to start with there's no saving the situation then, maybe emergency motor stop before it gets too far away to make wreckage recovery easier, in my case my neighbour's thatched barbeque roof stopped it for me, luckily with minimal damage save the ribbon cable which can be replaced here in South Africa. Slight upside is that at least its last known position was very accurate on the map, it knew where it was, just not how it got there in such a hurry

 

[Mynah]

Follow-up, simply as a matter of interest. I decided to buy another Mini 2, not that I really can afford it but with the idea of mothballing one as soon as the damaged one is fixed. We don't have the luxury of 'refresh' here in South Africa so I view it as long term insurance.
So, of course, even before its maiden flight I simply had to try recreate the mishap that started this thread.
This Mini 2 comes from the same batch of factory refurb fly-more combos currently for sale here, in fact, the serial numbers only differ by two digits. (They've marked the price down somewhat, more motivation to me buying another one while waiting for the damaged one to be fixed). So same firmware, same app version, same phone. I also used the same Mini battery, in case the Mini battery contains more ferrous metal than the Mini 2 battery.
I could simply, after 10 efforts, not recreate the IMU error. Same power-up spot, same controller next to drone at power-up. I could not even trigger a compass calibration request. Only difference to the original incident was the time of day. I even whipped out the metal detector to sweep my launch spot in case of a small object under ground that might have made a miniscule difference in exact launch area. I also reviewed my security cam footage to ensure I was in the same spot and held the drone and RC the same way.
So, conspiracy theorists, fire away....