Crash - Pilot error or drone issue?

[steve495]

sar104 provided excellent information concerning the uncontrolled descent and lack of tilt angle. I'd get back to DJI with that information and let them know you have the drone.

Since you posted and I replied originally, I think you're in better shape for your case. You even have the drone back and it may be able to be fixed.

 

[Wild Drone Pilot]

View media item 3206Check the Flight Log Retrieval and Analysis Guide by @sar104 to make sure you've collected all the available data for analysis. Best of luck!

Fox not ox

 

[Deleted member 93449]

Post #4 really is a warning that winds aloft (speed and direction) can be quite a bit stronger and in a different direction than winds at the surface. It seems the best course if you find the drone drifting at altitude and not responding as expect to stick inputs is to descent as quickly as possible. Getting lower may get you to an altitude where you can regain control of the drone.

 

[sar104]

No .. The problem is that the drone doesn't tilt enough.
It ignores throttle inputs to climb.

It doesn't appear to be tilting as much as it should to combat the wind, but I don't think that's the direct cause of the descent. If you look at the "not enough force" status, tilt and height during the flight, there is a clear transition in behavior from a 10° tilt in RTH until 234 seconds when cancelling RTH led to the pitch excursion and then pitch/roll instability and uncontrolled descent. I'm not sure that we can say definitively that it is ignoring throttle inputs without seeing the DAT file.

@theburninator - can you see if you can find the mobile device DAT file ending FLY017.DAT?

 

[steve495]

.... I did not push the drone back home because I was unsure of the orientation of the drone (still my fault obviously). ...

Just remembered I saw this and wanted to comment. Some might mention line of sight here, but even when you can see the Mini, it is difficult to see its orientation.

The application can certainly help here when everything is working properly (good GPS signal, compass calibrated properly...). We have an Aircraft Orientation indicator at the bottom center of the screen. It can be used in conjunction with the map - which can be expanded - to fly the drone back home manually.

See Page 32 and 33 of the Mavic Mini manual

 

[Prismatic]

Fox not ox

You think that was a mistake? Really?
?

 

[RayOZ]

I think the motor speed is limited in the software to prevent overload. If motor power is reduced, the max tilt needs to be reduced. If it tilted more, the lift would reduce and it'll descend a lot quicker.

 

[MavicDelight]

Sorry about your lose.

I use UAV Forecast app and Open Sky app, are there any other apps I should be using to make my flights as safe as possible and not have the above losses happen?

 

[sar104]

I think the motor speed is limited in the software to prevent overload. If motor power is reduced, the max tilt needs to be reduced. If it tilted more, the lift would reduce and it'll descend a lot quicker.

Possibly, but that doesn't explain why it was in uncontrolled descent at the end of the flight while tilted at only 6°, when it had previously been holding altitude while tilted at 10°.

 

[Peio64270]

Sorry about your lose.

I use UAV Forecast app and Open Sky app, are there any other apps I should be using to make my flights as safe as possible and not have the above losses happen?

This is a big problem at the altitudes where we fly (usually under 200m). Weather forecast models use grids of various sizes. Larger grids (like ECWMF or GFS) relates mainly synoptic winds at 10m above sea level but do not take into account local elevation, relief etc. This is an example of ECWMF forecast fot my area this night :

You can sea that the observed wind in different locations (airports, weather stations....) can differ by a large extent even if they are roughly at the same elevation above sea level. In Biarritz airport, for example, the mean wind is measured as being from south 6 knots while less than 20km away, in the Saint-Jean-de-Luz station it is measured 13 knots (more than the double). Just south from the latter (about 3km), there is no wind but i guess that if I launch my drone to 100m altitude at this place, I'll be facing at least the 13 knots south wind measured in St-Jean-de-Luz station. The wind (almost synoptic) given by ECWMF for this whole area is 10 knots, from South (in black square).
The thinest grid we have in France is the AROME model of forescast (grid of 1X1km). This is the same map given by Arome :

You see that it is rather different from the former. Local wind speeds, catabatic winds and so on are better taken into account and reflects more what is actually measured locally than ECMF and its wide grid.
But AROME is limited to the next 36h.
What I would recommand is thus to use an app like Windy which allows to compare rapidly several models of forecast with different grid sizes and then to rely mostly to the model with the smaller grid.
Further, Windy gives also a pretty good forecast for local gusts force and for us, this is even more important than mean wind speed :

It also gives a lot of other previsions such as thunderstorms, clouds (at high, mean and low altitudes), snow, rain, mist and much more.
I recommand warmly this app (for all platforms) but after, do not forget to calculate that the wind at 80m altitude can be twice the wind measured at 10m (and more if you are in hills or mountains).

 

[Johnsrog]

Thanks for all of the analysis on this incident. It has helped me, as a new PIC, to take it easy going up to get a feel for sudden changes in wind speed and direction. I’m also glad the wife said get the Zoom instead of the Mini. There are just too many issues with the Mini.

 

[VegasFlyer]

This is a big problem at the altitudes where we fly (usually under 200m). Weather forecast models use grids of various sizes. Larger grids (like ECWMF or GFS) relates mainly synoptic winds at 10m above sea level but do not take into account local elevation, relief etc. This is an example of ECWMF forecast fot my area this night :

@MavicDelight

I prefer Ventusky over Windy. Windy's screen is a bit too cluttered for me. With Ventusky, I can check the winds at a variety of altitudes in just a couple of clicks. Whichever you choose, the point is you need to rely on a product other than UAV Forecast.

 

[RayOZ]

Possibly, but that doesn't explain why it was in uncontrolled descent at the end of the flight while tilted at only 6°, when it had previously been holding altitude while tilted at 10°.

Not sure. Maybe need to have a look at the .dat file and calculate the trend of output to the motors. If the power to the motors are some how maxed limited, a tilt of 10° would make it descend faster than 6°. Perhaps, even at hovering, power to the motors is not enough to hold altitude, much less if it's trying to move horizontally. What's the roll angle between those 2 scenarios? It was fighting against the wind too.

 

[sar104]

Not sure. Maybe need to have a look at the .dat file and calculate the trend of output to the motors. If the power to the motors are some how maxed limited, a tilt of 10° would make it descend faster than 6°. Perhaps, even at hovering, power to the motors is not enough to hold altitude, much less if it's trying to move horizontally. What's the roll angle between those 2 scenarios? It was fighting against the wind too.

You would expect that if it were power limited then a tilt of 10° would cause it to descent faster than 6°. But the reverse happened. Tilt includes both pitch and roll - it's the combined angle. I requested the DAT file in post #24.

 

[m80116]

DJI Issue
The uncommanded descent is a big problem for the Mini that DJI don't seem to be doing anything about.
I'm seeing it a few times each week and when it starts to happen, it doesn't seem that anything fixes it.
I would make a point of mentioning this to DJI because they don't seem to even be aware of it.

What do you think about a theory pointing at a barometric pressure issue ? The airspeed around the drone body causes the IMU to gather inaccurate data to which the FC responds accordingly. If you abruptly halt the drone, like in many cases of ud, the drop commences but ends shortly after, in this case being the airspeed around the drone higher than its capability to hold against the wind (in P mode or RTH) the drop is continued. I am basically suggesting fast airspeed around the drone body creates vacuum inside the chamber where the atmospheric pressure sensor is located.

 

[Meta4]

What do you think about a theory pointing at a barometric pressure issue ? The airspeed around the drone body causes the IMU to gather inaccurate data to which the FC responds accordingly. If you abruptly halt the drone, like in many cases of ud, the drop commences but ends shortly after, in this case being the airspeed around the drone higher than its capability to hold against the wind (in P mode or RTH) the drop is continued. I am basically suggesting fast airspeed around the drone body creates vacuum inside the chamber where the atmospheric pressure sensor is located.

I think that DJI has made millions of drones that have no problem with any of this, ... including lots of Mavic Minis.

 

[RayOZ]

You would expect that if it were power limited then a tilt of 10° would cause it to descent faster than 6°. But the reverse happened. Tilt includes both pitch and roll - it's the combined angle. I requested the DAT file in post #24.

Something happened around 234 secs. But power (battery voltage * current) only dropped a little after that point when it starts to descend sharply. It would be interesting to see the motor rpms. Probably it didn't help with "Not Enough Force" for most of the flight. Maybe something gave way, like one of the motors overheated or something, the sudden pitch/roll at 234 secs. After that, it's just descending and not enough power to maintain altitude.

 

[m80116]

I think that DJI has made millions of drones that have no problem with any of this, ... including lots of Mavic Minis.

I tend to believe the exact opposite, all Mavic Minis demonstrate the uncommanded descent, some just happen to be in less favourable situations than others. This forum has taught me a lot, it's been because of you I learned of this problem and conceived a way to replicate it with almost 100% hit rate every time... but I am glad I know now, so I can avoid flying strategies that bring to that outcome when it could be critical.

 

[xspwhite]

Thanks for all of the analysis on this incident. It has helped me, as a new PIC, to take it easy going up to get a feel for sudden changes in wind speed and direction. I’m also glad the wife said get the Zoom instead of the Mini. There are just too many issues with the Mini.

Spend a 100-bucks or so on a cheap drone to practice with. Dont experiment with the zoom when you get it.

 

[Peio64270]

What do you think about a theory pointing at a barometric pressure issue ? (...)

With my first M1P, I had barometric pressure issues (it was exchanged under warranty two months after my purchase).
Altitude was not measured accurately with a severe drift all along flights. But this was reflected in telemetry (and therefore in flight logs). In the present case as in others dealing with the same issue, altitude measurements look accurate.