I had the drone a fiew meters above earth and just looking aroun. All the sudden and without any input on the sticks it started flying fast to the left. I tried to push up on the left stick but I was not quick enough and it crashed into a tree. Here is a chached video of what happend. The impact caused the battery to come out and the full quality video got corrupted.

https://drive.google.com/file/d/1ttIlulhvnX-hniTuOFGL4IE-f8L34VmY/view?usp=sharing

I saved the log file to my pc and here is a link to it:

https://drive.google.com/file/d/15e4xdNSI_5gF7Pd3nxhWWe22_VULnhN9/view?usp=sharing

I was observing the drone the whole time and it was clear of any obstacles during flight.
I also had a look trough the log file, but could not see anything
Do you have any idea what caused this wierd behaviour? Makes me scared to fly now.
Thanks for your time

Do you have any idea what caused this wierd behaviour? Makes me scared to fly now.

Two questions
What was the surface you launched from?
What direction was the drone facing when it was on the ground for launching?

Labroides Posted at 8-30 14:23
Do you have any idea what caused this wierd behaviour? Makes me scared to fly now.

Two questions

Made the link to the video an actual link
https://drive.google.com/file/d/1ttIlulhvnX-hniTuOFGL4IE-f8L34VmY/view?usp=sharing

Labroides, Steel railings on both sides, what looks like a small piece of  rail track a couple of meters away over the steel railings to the left, tough call underfoot as it looks like small decorative slabs. Theres a video link in his post but its not linked, ironic lol it is google drive though, can you not access this?

Deffo a yaw error and looks like it could have been caused by the magnetic properties of the nearby steel, I'm getting flashbacks lol
Added a screenshot image for you.
=====================================================================

boga,  try following the instructions here to upload the proper logfile as the one you posted is encrypted.
Here, done it for you
https://www.phantomhelp.com/LogViewer/43KD0WSZZED3ZMCQKB1H

Would poss help others if you could upload to the likes of youtube as its a universal player

Hello there boga94. I am sorry to read and to know what happened to your DJI drone. Since this unfortunate incident happened to your DJI Mini 2. I would recommend for you to contact our DJI Support Team at https://www.dji.com/support?site=brandsite&from=nav for further assistance. We have a team that will do there best to help and will give out the best resolution for this issue. Again I am sorry for the trouble and thank you.

In the video, drone is pointed east at start. In the log file, drone is indicated as pointed north at the same time.
During the "flyaway" the drone is facing the wrong direction in the data again compared to the video.
Compass error.

https://app.airdata.com/share/DDbBbd

GaryDoug Posted at 8-30 19:31
In the video, drone is pointed east at start. In the log file, drone is indicated as pointed north at the same time.
During the "flyaway" the drone is facing the wrong direction in the data again compared to the video.
Compass error.

Agreed, certainly looks like a compass error, but WOW, is that a speed record for the Mini 2, 54mph and in normal mode lol,
I've managed 36.2mph in sports, that's restricted too
View the  detailed info here

Compass error info below

Bashy Posted at 8-30 17:17
Labroides, Steel railings on both sides, what looks like a small piece of  rail track a couple of meters away over the steel railings to the left, tough call underfoot as it looks like small decorative slabs. Theres a video link in his post but its not linked, ironic lol it is google drive though, can you not access this?

Deffo a yaw error and looks like it could have been caused by the magnetic properties of the nearby steel, I'm getting flashbacks lol

Deffo a yaw error and looks like it could have been caused by the magnetic properties of the nearby steel.
He is unlikely to have launched close enough to those fence panels to cause a problem.
It's much more likely to be what was right under the drone.

Labroides Posted at 8-30 21:23
Deffo a yaw error and looks like it could have been caused by the magnetic properties of the nearby steel.
He is unlikely to have launched close enough to those fence panels to cause a problem.
It's much more likely to be what was right under the drone.

Yep, some kind of rebar or something ferrous underneath left over from long ago.

Bashy Posted at 8-30 20:16
Agreed, certainly looks like a compass error, but WOW, is that a speed record for the Mini 2, 54mph and in normal mode lol,
I've managed 36.2mph in sports, that's restricted too
View the  detailed info here

Good add. I did not know that there was a compass page in Airdata. Although those were after the collision with the tree.


ps: I know that some guys in these forums, not necessarily anyone in this discussion, refuse to believe that recorded videos can help to explain drone failures. This is one example that proves that wrong. I think that more information is always better.

Labroides Posted at 8-30 21:23
Deffo a yaw error and looks like it could have been caused by the magnetic properties of the nearby steel.
He is unlikely to have launched close enough to those fence panels to cause a problem.
It's much more likely to be what was right under the drone.

I could be wrong, like thats never happened before lol, but I didnt think it mattered so much as to where it took off from, but where it was actually turned on/initialised? If that was whilst he was closer to the railings then that would be an issue.

I would be surprised to see rebar below a pathway, we certainly don't reinforce pathways here in the UK and we are strict with regards to building codes etc. I know he's not in the UK, we don't have crossings like that nor do we drive on the wrong side of the road

Could be cables down there i spose, but with the mini, it's not like other DJI AC's, the button is in an awkward place, underneath *(rolls eyes), therefore it's in the air whilst turning on, also you would have to be quick to get it down on the floor as its quite quick to initialise too. Just to go 1 further, if you're holding the controller as well, that slows things down even more. Just my experience with it so far.

Hi,

Had a look at your log too.

As said, agree with Labroides, something under the takeoff spot caused the compass to go to an offset heading, not the correct heading.
Well, after take off compass changed back to correct heading, thus a conflict between compass data and stored data at powering up.

See compass line in my chart....line moves witthout any yaw input.

My tip : get used to powering the mini ON keeping it upright and don`t move it around or go to other position.
After that do a compass check.

How to prevent this from happening ? Always check the indicating heading of the drone in the map view in the fly app.
Compare this with the actual heading of the drone. If not the same : do not take-off!

cheers
JJB

JJB* Posted at 8-31 06:47
Hi,

Had a look at your log too.

"Well, after take off compass changed back to correct heading..."

When was that? I see it being off by about 180 degrees (data shows pointed south, video shows pointed north) at the end while it is going fast sideways toward the tree to the west.

GaryDoug Posted at 8-31 11:29
"Well, after take off compass changed back to correct heading..."

When was that? I see it being off by about 180 degrees (date shows pointed south, video shows pointed north) at the end while it is going fast sideways toward the tree to the west.

my text fault ; bad explantion!
after takeoff heading changed, into direction of where it shoud be...but did not reach that ofcourse.

Trying to say that after leaving the ground, the effect of the disturbance become less or none  so in the 100% oke world it would go back to where it should be. But yaw input will introduce a fly away as the inputs to the motors are to the wrong ones....

cheers
JJB

...........

Here is another view of the takeoff area:

One thing I would like to understand is why/how a compass problem caused the quad to simply fly off?

Are we suggesting that it wanted to go one way (lets say North), but the compass had initialized 90 degress off, so the quad paused East, trying to go North?

If so, why it not recognise that there was a problem and stop? Surely it would know if it was flying in the wrong direction using GPS, so why keep flying?

What would have happened if it didn't crash? Would it have just gone on forever?

jweaver Posted at 9-2 00:28
One thing I would like to understand is why/how a compass problem caused the quad to simply fly off?

Are we suggesting that it wanted to go one way (lets say North), but the compass had initialized 90 degress off, so the quad paused East, trying to go North?

This a bit over my head chap, I will leave that to Labroides to explain, I'm not sure why it induces yaw error that makes shoot off, had it happen to myself, I was lucky I had loads of room (100-200m), hence I no longer take off from my cars roof lol

jweaver Posted at 9-2 00:28
One thing I would like to understand is why/how a compass problem caused the quad to simply fly off?

Are we suggesting that it wanted to go one way (lets say North), but the compass had initialized 90 degress off, so the quad paused East, trying to go North?

When the drone is powered on, the gyro sensor has no directional reference.
It gets that from the compass.
But if the compass is reading incorrectly because it's affected by nearby steel objects, the gyro sensor gets an incorrect reading to start with.
When the drone gets away from the source of the problem, the compass reads correctly, but the gyro still reads incorrectly and the data conflict causes bug problems.
When the drone moves, the flight controller thinks the drone is drifting off course and tries to correct, but the correction will be in the wrong direction.
That happens many times per second, the drone accelerates when it's just trying to hold position.

Labroides Posted at 9-2 04:15
When the drone is powered on, the gyro sensor has no directional reference.
It gets that from the compass.
But if the compass is reading incorrectly because it's affected by nearby steel objects, the gyro sensor gets an incorrect reading to start with.

Thanks.. I was pretty sure that was what people were saying, but wanted to check.

I am surprised it can't detect when its flying away.. Would it go on forever like this? Or at some point would its GPS location give it a clue that something wasn't right?

At point F in the log file, does that warning seem to imply that the drone is in Sport mode, when it's clearly shown to be in Normal mode? After all, why would a warning in relation to Sport mode be displayed when in Normal mode. Also, even if the compass is messed up, why has the drone suddenly decided to fly off by itself if there was no stick inputs?

jweaver Posted at 9-2 05:00
Thanks.. I was pretty sure that was what people were saying, but wanted to check.

I am surprised it can't detect when its flying away.. Would it go on forever like this? Or at some point would its GPS location give it a clue that something wasn't right?

I am surprised it can't detect when its flying away.. Would it go on forever like this?
Only until something gets in its path.
Or at some point would its GPS location give it a clue that something wasn't right?
It's getting perfect GPS data the whole time, but the gyro has been initialised wrongly so every time the flight controller tries to correct (many times per second), it's sending the drone further off course.

steamship Posted at 9-2 05:37
At point F in the log file, does that warning seem to imply that the drone is in Sport mode, when it's clearly shown to be in Normal mode? After all, why would a warning in relation to Sport mode be displayed when in Normal mode. Also, even if the compass is messed up, why has the drone suddenly decided to fly off by itself if there was no stick inputs?

At point F in the log file, does that warning seem to imply that the drone is in Sport mode, when it's clearly shown to be in Normal mode?
It's a weird message that I've not seen before.
But the drone was never in Sport Mode.

Also, even if the compass is messed up, why has the drone suddenly decided to fly off by itself if there was no stick inputs?
The compass isn't messed up, it's been reading properly from just after start.
It's the Gyro sensor that's the problem and I've explained that in a couple of posts above.

jweaver Posted at 9-2 05:00
Thanks.. I was pretty sure that was what people were saying, but wanted to check.

I am surprised it can't detect when its flying away.. Would it go on forever like this? Or at some point would its GPS location give it a clue that something wasn't right?

"... at some point would its GPS location give it a clue that something wasn't right?"

I think it does correct, however that takes a while. I seem to remember some flights described here with compass errors at start that were resolved later in the flight. But that takes some time to "compute". In this case, the crash occurred in just seconds. Just my opinion, no parametric evidence to confirm that ;-)

jweaver Posted at 9-2 05:00
Thanks.. I was pretty sure that was what people were saying, but wanted to check.

I am surprised it can't detect when its flying away.. Would it go on forever like this? Or at some point would its GPS location give it a clue that something wasn't right?

For some of these that I've seen, there's a perfect arc on the flightpath.

If you have nothing in the way for around 200m distance from the take-off point, with limited stick input  you can be lucky enough that it will come back around and once its close enough to the take-off point it will enter that magnetic field again and full control will be acquired again if that happens land immediately, turn it off, if you wish to fly the same area again, then move it away from that spot or if there is no steel etc nearby then its under foot so, turn it on above head height and hold till gimbal stops dancing, then take off

GaryDoug Posted at 9-2 19:14
"... at some point would its GPS location give it a clue that something wasn't right?"

I think it does correct, however that takes a while. I seem to remember some flights described here with compass errors at start that were resolved later in the flight. But that takes some time to "compute". In this case, the crash occurred in just seconds. Just my opinion, no parametric evidence to confirm that ;-)

The problem is that it's trying to correct, but because of the incorrect gyro heading data, each correction only makes things worse.
Attempted correction using bad gyro data is what causes the problem.
The drone is unable to self-correct for that.

Labroides Posted at 9-2 21:32
The problem is that it's trying to correct, but because of the incorrect gyro heading data, each correction only makes things worse.
Attempted correction using bad gyro data is what causes the problem.
The drone is unable to self-correct for that.

I wonder if a fail-safe could be added to halt the ac as soon as stick inputs do not match what the ac is doing, perhaps a halt and return to the hoping using GPS only, just tossing thoughts is all...

So I guess the question is "how do we know this is going to happen"? And what do we do if it does? Would you see any alarms or warnings before takeoff? Or are we saying tha its just one of those things?

Did the Pilot in this case do anything wrong? Anything that could have been avoided?

What if we you at a spot which has a magnetic field and you didn't know.. Do you just accept it when the Quad flys away or crashes as 'normal'? Is there anything we could/should be doing before flying to ensure this could never happen?

Bashy Posted at 9-2 22:16
I wonder if a fail-safe could be added to halt the ac as soon as stick inputs do not match what the ac is doing, perhaps a halt and return to the hoping using GPS only, just tossing thoughts is all...

Hi Bashy,

Interesting idea, but DJI drones need GPS and Compass data to calculate the correct power to the correct motors.

If you stop the video at 14s : drone actual heading is approx 090 degrees. Compass data in the log that time = 330 degrees.
If you stop the video at 44s : drone actual heading approx North. Compass data in the log that time = 160 degrees.
During the first seconds after takeoff (only UP and no yaw input) the compass heading changed from 330 to 240 ! Indicates that the compass was influenced by metal at ground start position.

If the compass heading is not the actual heading at takeoff, than a fly away is 'initiated' the moment the drone is moved away from its original take off position.
Moved away by pitch and/or roll input or bc of the wind. GPS position 'knows' that is has moved from that position, to correct the changed position to the before position compass data is used to give to power signals to the correct motors.
If compass data is not correct it simply powers the wrong motors, resulting in fly away further from position, need more correction, again further away etc etc.
At the end max power is sent to the motors thus max speed is far beyond normal speed limits in N or S mode.

If the offset (actual heading and compass 'wrong'data) is not that large than, flying in free airspace, drone will fly a toilet bowl pattern....
BTW sometimes in that situation the RTH will stop this and drone will correct its compass and fly back home as normal.


cheers
JJB

jweaver Posted at 9-3 02:35
So I guess the question is "how do we know this is going to happen"? And what do we do if it does? Would you see any alarms or warnings before takeoff? Or are we saying tha its just one of those things?

Did the Pilot in this case do anything wrong? Anything that could have been avoided?

Hi,

To prevent this from happening ; always check before takeoff the compass heading in the flyapp map view and compare this heading with the actual heading!

If not the same : DO NOT FLY.

IMO the most important check before starting a flight.
Personally i
1) start my flights with a hand-take-off ; so no risk of metal in ground ect to hamper my compass
2) powering up my Mini2 (stupid place for the power ON switch) always upright and don`t move craft during powering up sequence

cheers
JJB

JJB* Posted at 9-3 03:08
Hi,

To prevent this from happening ; always check before takeoff the compass heading in the flyapp map view and compare this heading with the actual heading!

And when you say "check the heading with the actual heading", you mean just rotate it around and make sure its actually pointing to objects on the map? Or literally compare it to another compass?

Is it OK, to look at the map, point the quad to various objects (buildings, roads etc) and check it alligns? Or do you mean something more scientific?

jweaver Posted at 9-3 03:20
And when you say "check the heading with the actual heading", you mean just rotate it around and make sure its actually pointing to objects on the map? Or literally compare it to another compass?

Is it OK, to look at the map, point the quad to various objects (buildings, roads etc) and check it alligns? Or do you mean something more scientific?

Compare the compass direction your real drone is facing with the compass direction that the drone icon on your map view is facing.

If the drone is facing east and the icon on the map also is facing east, you are good to go.
If the drone faces a different direction from the icon, that indicates that the drone's compass is affected by nearby steel and giving a false reading.

Avoiding launching from reinforced concrete surfaces is a good idea.
It's the most common source of the problem.

jweaver Posted at 9-3 03:20
And when you say "check the heading with the actual heading", you mean just rotate it around and make sure its actually pointing to objects on the map? Or literally compare it to another compass?

Is it OK, to look at the map, point the quad to various objects (buildings, roads etc) and check it alligns? Or do you mean something more scientific?

In addition to Labroides text,

Not scientific at all...

Sometimes is it not neccessary to know the actual heading in degrees....I have a small hand held compass to check, but rarely use it.
If you have your drone at the ground, aligned with a road, waterway, building etc.
Than in the map view you see the arrow pointing aligned with that road etc.
No need to move your drone in various heading, oke  = oke

cheers
JJB

A small appreciation and a great thanks to Labroids and JJB for your detailed input here, as well as to all others that has contributed here. I have read everything with great interest and copied the important things to my list of important things to know about drones. It has really taught me a lot about the drones compass and Gyro, which I did not know about and probably never would have found out otherwise. My sincere thank you for your explanations and hope that every one would read this. Cheers and many thanks again, Tony the old Swedish Viking.

JJB* Posted at 9-3 03:05
Hi Bashy,

Interesting idea, but DJI drones need GPS and Compass data to calculate the correct power to the correct motors.

"BTW sometimes in that situation the RTH will stop this and drone will correct its compass and fly back home as normal."

Just what I noticed in some cases.

OK. I must be doing something wrong...or very right!

I just tried repeatedly to duplicate the OP's issue, using my Mini. I put a steel tool under the middle of my landing mat and tried takeoff. I did this several times, first starting the drone away from the mat and moving it to the mat, then launching. I also tried several times to do the power up and launch both from the middle of the mat where the steel tool was placed. In all cases, the initial indication for the compass was off by 90 to 180 degrees, depending upon the placement, and I saw the compass warning. Also in all cases, very shortly after launch, I saw the compass indication correct to the proper direction. I could see the indicator slowly rotating even though I was not yet doing any rotation (yaw) with the controls. In no cases did the Mini do anything unusual; it worked normally in all directions and heights. How do I duplicate the problem?

GaryDoug Posted at 9-3 07:11
"BTW sometimes in that situation the RTH will stop this and drone will correct its compass and fly back home as normal."

Just what I noticed in some cases.

Good test. That's interesting. So the RC indicator slowly corrected itself from the drones compass once up, but nothing happened to the flight. Has something changed in a FW upgrade perhaps from previous FW????

GaryDoug Posted at 9-3 07:11
"BTW sometimes in that situation the RTH will stop this and drone will correct its compass and fly back home as normal."

Just what I noticed in some cases.

i do not want to test this out the way you do this....

But as are a real dare devil...do the same, takeoff   and push the drone manuall away from its position....
take care not to get hurt while you mini performs a fly-awy!  ;-)

cheers
JJB

JJB* Posted at 9-3 10:29
i do not want to test this out the way you do this....

But as are a real dare devil...do the same, takeoff   and push the drone manuall away from its position....

I did that several times, nothing went wrong. I will try a larger piece of steel, a 3' long crowbar next.

By the way, the time for the compass to correct it's indication was just a few seconds, under 20 for sure.
Here is a sample. In all takeoffs, the Mini was facing NNW. In this flight, the compass starts off pointing south. Note the compass warning. At first, it is just a few feet above ground but later in the flight, I take it higher.

https://app.airdata.com/share/pLwQbG

GaryDoug Posted at 9-3 18:15
I did that several times, nothing went wrong. I will try a larger piece of steel, a 3' long crowbar next.

By the way, the time for the compass to correct it's indication was just a few seconds, under 20 for sure.

By the way, the time for the compass to correct it's indication was just a few seconds, under 20 for sure.That's completely normal.
The compass reads normally when it gets far enough from the steel object that affected it initially.

The compass isn't the issue, which is the conflicting compass and gyro data once the compass is reading normally.

There have been reports of recent Mavics being more resistant (but not completely immune) to yaw errors.
I'm not sure what exactly has been done in their programming to achieve this.

Earlier you mentioned that:  I saw the compass warning.
Usually a compass error warning will prevent takeoff.
The warning appears when the magnetic field strength is substantially more than the earth's magnetic fiels and it's obvious that there's something wrong.
Most yaw errors are a result of a weak magnetic field that's not strong enough to bring up a warning, but still deflects the compass sensor.

GaryDoug Posted at 9-3 18:15
I did that several times, nothing went wrong. I will try a larger piece of steel, a 3' long crowbar next.

By the way, the time for the compass to correct it's indication was just a few seconds, under 20 for sure.

Hmm, i think there are 2 ways of getting a incorrect reading of the compass

1) powering up a drone away form interference (so compass and gyro have the same value after initialisation), than put  drone on the ground with rebar etc ; compass will change heading and after take off : compass will move the the correct value and no conflict comnpass-gyro. No fly away.

2) powering up in a place with metal interference, compass and gyro have the same value, after take-off compass will change heading and there is a conflict gyro-compass. Fly away.

Mayby you can do such a test.....

I like my hand-takeoff more and more for my drones!

MyChartYourFlight:  After takeoff your the compass value from 185 to 330 (all approx), but in this time no real change in hover position.
With the compass at incorrect heading nothing will happen when drone yaws or up.down at same position ; bc no position correction is needed wich if executes will go to the wrong motors. At x-axis ID 190 you see a pitch fwd and aft input, that time compass aligned....or almost, see the rapid change from 320 to 330.
cheers
JJB

JJB* Posted at 9-4 00:55
Hmm, i think there are 2 ways of getting a incorrect reading of the compass

1) powering up a drone away form interference (so compass and gyro have the same value after initialisation), than put  drone on the ground with rebar etc ; compass will change heading and after take off : compass will move the the correct value and no conflict comnpass-gyro. No fly away.

"Maybe you can do such a test....."

I did that today several times. I was unable to create the problem. If I started the drone first away from the pad and then put it onto the pad with the steel tool on it. There was no change in the indicated compass direction even as i moved the tool all around the Mini. but the compass error appeared anyway.

This is a link to the closest I came to a bad indication. It is just like the others yesterday. Mini was facing south and compass showed north. It was about 2 feet from a steel fence at startup. Note the sudden change of the indication from north to south just after takeoff. No change in operation; all motions shown in the flight were commanded.
https://app.airdata.com/share/lQEZPS

Note: Once the compass error appeared at startup, it did not go away by moving. I had to perform a calibration or restart both drone and controller. It was quicker to do the re-cal.

Funny thing: The last flight was launched from the steel hood of my car, no issues.
https://app.airdata.com/share/GxGssP


Labroides said this fault often happens without the compass message, like in the case of the OP (no error). I wonder if the Mini does not update the IMU, if there is a compass error, until you launch a feet feet upward and the compass corrects. It's like some kind of conditional failsafe. I am thinking that something else happened to the OP in addition to the compass anomoly.... or that the error flag did not get set for some reason, and the IMU got updated imediately instead of being delayed.