Flight Logs: https://www.phantomhelp.com/LogViewer/0GBHW9K3VR57VZ4TGOSL

So I took off of a bridge yesterday, and flew the mini about 20' out over the water.  It was sinking and I had to raise its altitude a few times and was just waiting for it to stabilize.  I turn it around and all of the sudden it just takes off at full speed sideways.  I had maybe 4-5 seconds before it smashed right into the side of the bridge and ended up in the river below.  During these 4 seconds I first tried to give it opposite input, then tried to get it to climb to clear the bridge.  It didn't respond at all to anything.  So it went swimming, and so did I.

Looking for advice on a possible refund from DJI.  It was maybe the 10th flight with this drone, but I've had a dozen before this.

First, you were impatient and launched without waiting for GPS or the drone to record its homepoint.
Without GPS, proper analysis of your flight data is tricky but your description and location give enough information to see what happened..

I took off of a bridge yesterday
This is an immediate red flag.
The bridge is made of steel and it's difficult (impossible) to launch from the middle of a steel bridge without your compass being deflected by all that steel so close.
Your drone uses gyro sensors as well as a compass and the gyro needs to be initialised at power up when it gets it's initial values from the compass.
If the compass is reading wrong because of the effect of all that steel, the gyro sensor is incorrectly initialised.

I turn it around and all of the sudden it just takes off at full speed sideways.
This is the classic indication of a yaw error, flying far enough from the steel for teh compass to now read the earth's normal magnetic field, but the grro sensor is giving incorrect values because it was initialised incorrectly.
When the drone is turned to face a different direction, the directional information from the gyro and compass are conflicting and things go downhill quickly.

The problem was entirely due you where you launched your drone.
Launching from steel and reinforced concrete surfaces is just asking for trouble.

Labroides Posted at 9-25 16:39
First, you were impatient and launched without waiting for GPS or the drone to record its homepoint.
Without GPS, proper analysis of your flight data is tricky but your description and location give enough information to see what happened..

I calibrated the compass immediately before taking off and hand launched it, the same way I have 100 times before near metal.  This drone has had a compass problem that I've never had with any other, in that I need to calibrate it before literally every single flight.  Also, that doesn't explain the drone not responding to commands to rise to clear the bridge

fansd8e94fa7 Posted at 9-25 16:43
I calibrated the compass immediately before taking off and hand launched it, the same way I have 100 times before near metal.  This drone has had a compass problem that I've never had with any other, in that I need to calibrate it before literally every single flight.  Also, that doesn't explain the drone not responding to commands to rise to clear the bridge

I calibrated the compass immediately before taking off and hand launched it, the same way I have 100 times before near metal.
Calibrating the compass won't make a scrap of difference to what I explained in my initial post.

Also, that doesn't explain the drone not responding to commands to rise to clear the bridge.
Your Mini shows the common issue of dodgy altitude stability, possibly because of the state of the propellers.
It did climb in response to your throttle input from 0:21.7 until the end of recorded data.

The problem was entirely due you where you launched your drone.
Launching from steel and reinforced concrete surfaces is just asking for trouble.

Labroides Posted at 9-25 19:03
I calibrated the compass immediately before taking off and hand launched it, the same way I have 100 times before near metal.
Calibrating the compass won't make a scrap of difference to what I explained in my initial post.

Thanks for your opinion

Hi,

Had a look at your flight too.

Flying away fromn the bridge in OPTI mode, so no LAT LON data. looking at the map your flight direction, away from the bridge, was approx. heading North.

In the log the compass data is heading 175 when flying away fromn the brigde to North.
Left yaw and MM starts to fly away, towards the bridge but in the log heading direction is 050.

So obvious a compass induced fly away. (correction made to keep craft GPS stabilized are sent to the wrong motors as the software use the incorrect compass data)

Best is to always check if compass is aligned to the actual direction of the drone, us the map view to check.
If arrow is pointing not as the real heading ; don`t take-off.

At the end, drone was flying fast towards the bridge. It did react oke to the up command (vertical speed UP), but too late to climb enough.
MM is bit underpowerd, UP and FULL yaw (as you did) slow down the UP movement a bit.


PS starting in OPTI mode is never a good idea, as without GPS reception Homepoint is not set.
So if the flight was succesfull, and after more flight time HP was set (GPS sats+reception oke) than in case of a disconnect your MM had flown to a HP wich was not set at your take-off location.

Another reason why flying in OPTI mode is risky in combination with a wrong compass, the moment GPS is getting there (in number and reception) the first LatLon positions can vary a little, thus in GPS mode it starts to correct to the positions to stabilize. With a wrong compass the wrong corrections are made.See the MinSecs Secs colonm where [-HP] is written (so LL positions are received) ; speed, without Pitch/Roll RC inputs,  start to increase....


If you like, put a link one here to the DAT [FLY024.DAT] file for this flight, more info in there about your compass.

cheers
JJB

Dear Colleague MM Pilot,

Although I started to fly the MM only a month ago, I flew the P3 Pro for 2 years and real planes for a long time.
From the basics of NAVSTAR system, you need 3 GPS satellites for a 2D fix (i.e. no height) or 4 satellites for a 3D fix.  However, the accuracy with a reduced number of satellites can vary by one hundred meters, or much more.
From the flight log, your MM GPS transitioned from ATTI to GPS at 2020/09/24 17:32:38.588, but
was tracking only 5 satellites, which means that MM position's accuracy calculation would be very inaccurate. However, in this case the calculation was not possible, probably due to the satellite geometry in the sky (the GPS receiver could be getting almost aligned satellites which, sometimes, can not be computed).
Probably, the geometry of the only 5 satellites  received and also in conjunction with the magnetic properties of the bridge influenced the MM not to get a definition for your home point (you can check that on the latitude and longitude not defined on the flight data log). It could only compute the height, 15.93 meters.
As a conclusion, if you waited until your MM acquired more satellites for certain you would not have that problem.
I hope it could help and that you were able to recover your MM, wishing you great and magnificent future flights.

Cients Posted at 9-27 08:12
Dear Colleague MM Pilot,

Although I started to fly the MM only a month ago, I flew the P3 Pro for 2 years and real planes for a long time.

From the basics of NAVSTAR system, you need 3 GPS satellites for a 2D fix (i.e. no height) or 4 satellites for a 3D fix.  However, the accuracy with a reduced number of satellites can vary by one hundred meters, or much more.
Things are a bit different with DJI.
Your drone won't be in P-GPS mode until you have a minimum of 6 sats and even then the flight controller assesses the spread of sats etc and rates the GPS Health 0-5.
If GPS Health is 0 or 1, the flight controller doesn't use the GPS data and the drone is effectively in atti mode.

From the flight log, your MM GPS transitioned from ATTI to GPS at 2020/09/24 17:32:38.588, but was tracking only 5 satellites, which means that MM position's accuracy calculation would be very inaccurate.
In the OP's incident, GPS Health only reached 4/5 for the last 1/2 second of the recorded flight data.
He did not have position holding and with the yaw error, his drone was racing away, chasing its tail, just trying to hover.

As a conclusion, if you waited until your MM acquired more satellites for certain you would not have that problem.
This would be true except that his Mini had a yaw error as described earlier.
Because of the yaw error, this flight was doomed from power on.

Labroides Posted at 9-27 14:29
From the basics of NAVSTAR system, you need 3 GPS satellites for a 2D fix (i.e. no height) or 4 satellites for a 3D fix.  However, the accuracy with a reduced number of satellites can vary by one hundred meters, or much more.
Things are a bit different with DJI.
Your drone won't be in P-GPS mode until you have a minimum of 6 sats and even then the flight controller assesses the spread of sats etc and rates the GPS Health 0-5.

Thank you for the very detailed explanation!

Hello there. I am sorry to read and to know what happened to your DJI Mavic Mini. Since this unfortunate incident happened on the said DJI Drone. It would be best to contact our DJI support team at https://www.dji.com/support for further assistance. We have a team that will do there best to assess the said drone and give out the best resolution for this issue. Again, I am sorry for this trouble and thank you for your understanding.

Cients Posted at 9-27 08:12
Dear Colleague MM Pilot,

Although I started to fly the MM only a month ago, I flew the P3 Pro for 2 years and real planes for a long time.

Hi,

With not enough sats a DJI MM will fly in ATTI mode OR, if there is enough light for visual reference and  within range sensor range ; drone will fly in OPTI mode. Check this (in the OP flight)  with the use of RC input and the drone pitch and roll values, not a drone in ATTI as in ATTI it will not brake.

This flight was started in OPTI mode, not in ATTI. OP could fly his MM for 20 seconds away from the bridge, various inputs and craft was stable.So this MM flight stabilty was done by sensors in OPTI mode, so the drone did fly actually stable before the sh*t hits the fan.

See the chart, see the VPS height lines ; GREEN line = VPS height measuring, BLUE line = Sensor positioning active.
DAT file will tell more, the very moment that this MM started to record (8 sats) GPS positions it started to accelerate.
And ofcourse compass was indicating the wrong headings, so or the compass caused the flyaway, or the change from OPTI to P-GPS (no stable data ) or combi of both...


But think its a compass yaw error,  a fly away due to yaw error can only happen in P-GPS mode, as the needed correction needs compass data and GPS data.
In ATTI no fly-away, only a drift-away (wind and/or wrong or overcompensating RC inputs). That`s why with severe compass/yaw errors a DJI changes his mode from a stabilized mode into ATTI; to give the OP the chance of controlling it again.

That`s why i like to see the DAT file for this flight.


PS easy to check OPTI mode (and see the values change in the CSV file) ; fly from inside your house (only with enough light, and bad GPS reception) to outside your house.In the csv file you can check the values, compare it with PhantomHelp and notice that the flightmode in PH is not correct for OPTI.

cheers
JJB

JJB* Posted at 9-27 22:46
Hi,

With not enough sats a DJI MM will fly in ATTI mode OR, if there is enough light for visual reference and  within range sensor range ; drone will fly in OPTI mode. Check this (in the OP flight)  with the use of RC input and the drone pitch and roll values, not a drone in ATTI as in ATTI it will not brake.

Thank you for your detailed technical vision.