Mavic Air behaved weird and crashed

Slipurson · Sep 10, 2021

Here is my logfile: DJI Flight Log Viewer - PhantomHelp.com

I had control up until 12,5 seconds, after that it started drifting, and from the logfile it seems it had some little input on the left stick, but no matter how i tried it would just go away from me.

I then had to go down before it would hit the trees to the west of me.. also started to spin it to make it not go so fast into the ground.

A couple of days before this I had a quick popup, that went away in a few seconds, about calibrating the IMU, I then restarted and the error was gone, I checked the IMU and there was no need for calibration.. I then made 5-6 more flights before the one linked above.

My question:
I cant see any errors other then the tiny input on the left stick there at 12,5 seconds, could I have saved it? Did i just make it worse with the inputs I did to "save it" from the trees? From 23,8 second the left stick is centered and it keeps going"backwards and tho the right" (north on the map)

Sorry if I am rambling, but I just cant figure this one out..

(the drone is fine and I have made a few flights after this without problems, but i did calibrate Sticks, IMU and Compass)

slup · Sep 10, 2021

Did the craft drift away out over the field sideways, like shown in the pic below... or was the craft directed in another direction?

Slipurson · Sep 11, 2021

slup said:
Did the craft drift away out over the field sideways, like shown in the pic below... or was the craft directed in another direction?

View attachment 134709

I am pretty sure it was not facing that direction, as I tried to get it back to me by pulling down and left on the left stick, and that made it slow down. At that point i realized that I would be fly it back to the homepoint and started the spining/descending.

Also now that you mention it I remember that the drone did almost a 180 after it had set the homepoint, usualy it is just a slight correction a few degrees.

I guess it had north/south etc messed up... I usualy check everything each time I fly, but the lest few days I have done many several flights/day and my routines got sloppy :/

slup · Sep 11, 2021

Yeah ... this really looks like a yaw-error of about 90 degrees which explains the initial rather straight flight (drift) path out over the fields. But as you used Litchi we don't have any means to investigate if it was due to a power on of the craft in a magnetic interfered spot (IMU wrongly initiated due to a deflected compass) or if it was IMU problems (...this as no mobile device .DAT log is created with that app).

All rotating and the slightly curved path the craft took later into the flight was due to your stick inputs ...

From 16,3sec to 21,8sec you had these inputs...

...which would have made your craft rotate, descend & have a horizontal heading backwards to the left (while rotating). After that all stick inputs is rather confused ... but all together made your craft take the path it did.

Slipurson · Sep 12, 2021

slup said:
Yeah ... this really looks like a yaw-error of about 90 degrees which explains the initial rather straight flight (drift) path out over the fields. But as you used Litchi we don't have any means to investigate if it was due to a power on of the craft in a magnetic interfered spot (IMU wrongly initiated due to a deflected compass) or if it was IMU problems (...this as no mobile device .DAT log is created with that app).

All rotating and the slightly curved path the craft took later into the flight was due to your stick inputs ...

From 16,3sec to 21,8sec you had these inputs...

View attachment 134716

...which would have made your craft rotate, descend & have a horizontal heading backwards to the left (while rotating). After that all stick inputs is rather confused ... but all together made your craft take the path it did.

I have started from that same spot several times so there should not be anything there, but there are some powerlines 30 meters away.

Anyway thanks for looking into this. It helped me see that it was probably an error from the beginning and that i need to check everything before i lift off.

The bit at the end where it is spinning and the sticks pointing down/left was just me trying to get it down on the ground before it hit the treeline or the house that is there ?

slup · Sep 12, 2021

Slipurson said:
...but there are some powerlines 30 meters away.

Power lines hardly never have any effect on your drone ... especially if they are 30m away. In order to get a yaw error your drones compass needs to be centimetres away from magnetic interference when the drone is POWERING ON. A nail in a wooden deck unfortunately just below the drone is a typical error ... pilots think's all is wooden but doesn't think about that nail.

Slipurson said:
...i need to check everything before i lift off.

As the yaw error is created during the initial phase during powering on the drone ... & the compass then needs to be away from magnetic interference, I have a special drill when I'm using my own Mavic Air 1.

Step 1:
I'm powering on the drone keeping it horizontal, camera facing away in my out stretched hand (without watches, rings or other magnetic objects) … once the live view have turned up in the mobile device I put it on ground for take-off. In this way I have done every effort possible to keep the drone away from possible hidden magnetic objects under the ground ... & the IMU should have been initialized by a clean, undisturbed compass.

Step 2:
After the powering on of the drone is complete & I've placed it in the take-off spot on ground, waiting for it to acquire enough satellites & get a HP recorded ... I check that the drone icon on the map in the app is pointing equal in relation to other objects in the map as the drone does in reality ... if it doesn't, I abort the launch attempt, POWER THE DRONE DOWN & move away, power up again and repeat.

Lastly, just for you knowledge ... it's quite common that the Mavic Air 1 in particular, sometimes turn slightly uncommanded (approx 15 degrees) during the ascent when you take-off, weather you take-off manually or automated through the app. This is a bit of a mystery ... but we have studied the DAT logs from events like this & it seems that a probable reason for this could be that when the motors start & rev up to take-off they disturb the compass slightly. This shouldn't be a concern ... no incidents have been reported in this or in the DJI forum where this little oddity have been the cause.

Slipurson · Sep 12, 2021

slup said:
Power lines hardly never have any effect on your drone ... especially if they are 30m away. In order to get a yaw error your drones compass needs to be centimetres away from magnetic interference when the drone is POWERING ON. A nail in a wooden deck unfortunately just below the drone is a typical error ... pilots think's all is wooden but doesn't think about that nail.

As the yaw error is created during the initial phase during powering on the drone ... & the compass then needs to be away from magnetic interference, I have a special drill when I'm using my own Mavic Air 1.

Step 1:
I'm powering on the drone keeping it horizontal, camera facing away in my out stretched hand (without watches, rings or other magnetic objects) … once the live view have turned up in the mobile device I put it on ground for take-off. In this way I have done every effort possible to keep the drone away from possible hidden magnetic objects under the ground ... & the IMU should have been initialized by a clean, undisturbed compass.

Step 2:
After the powering on of the drone is complete & I've placed it in the take-off spot on ground, waiting for it to acquire enough satellites & get a HP recorded ... I check that the drone icon on the map in the app is pointing equal in relation to other objects in the map as the drone does in reality ... if it doesn't, I abort the launch attempt, POWER THE DRONE DOWN & move away, power up again and repeat.

Lastly, just for you knowledge ... it's quite common that the Mavic Air 1 in particular, sometimes turn slightly uncommanded (approx 15 degrees) during the ascent when you take-off, weather you take-off manually or automated through the app. This is a bit of a mystery ... but we have studied the DAT logs from events like this & it seems that a probable reason for this could be that when the motors start & rev up to take-off they disturb the compass slightly. This shouldn't be a concern ... no incidents have been reported in this or in the DJI forum where this little oddity have

slup said:
Power lines hardly never have any effect on your drone ... especially if they are 30m away. In order to get a yaw error your drones compass needs to be centimetres away from magnetic interference when the drone is POWERING ON. A nail in a wooden deck unfortunately just below the drone is a typical error ... pilots think's all is wooden but doesn't think about that nail.

As the yaw error is created during the initial phase during powering on the drone ... & the compass then needs to be away from magnetic interference, I have a special drill when I'm using my own Mavic Air 1.

Step 1:
I'm powering on the drone keeping it horizontal, camera facing away in my out stretched hand (without watches, rings or other magnetic objects) … once the live view have turned up in the mobile device I put it on ground for take-off. In this way I have done every effort possible to keep the drone away from possible hidden magnetic objects under the ground ... & the IMU should have been initialized by a clean, undisturbed compass.

Step 2:
After the powering on of the drone is complete & I've placed it in the take-off spot on ground, waiting for it to acquire enough satellites & get a HP recorded ... I check that the drone icon on the map in the app is pointing equal in relation to other objects in the map as the drone does in reality ... if it doesn't, I abort the launch attempt, POWER THE DRONE DOWN & move away, power up again and repeat.

Lastly, just for you knowledge ... it's quite common that the Mavic Air 1 in particular, sometimes turn slightly uncommanded (approx 15 degrees) during the ascent when you take-off, weather you take-off manually or automated through the app. This is a bit of a mystery ... but we have studied the DAT logs from events like this & it seems that a probable reason for this could be that when the motors start & rev up to take-off they disturb the compass slightly. This shouldn't be a concern ... no incidents have been reported in this or in the DJI forum where this little oddity have been the cause.

I think I will start using your method when taking of also, as I do now I power on and place it where I am gonna take of, and then wait for it to get enough satelites (green statusbar and homepoint set. I have actually never checked, before liftoff, that the icon on the map is facing the right way tho as I use litchi alot for waypoint flying I would quite soon notice if it was pointing the wrong way (but at that time it would probably be to late, like this time).

The slight turn I have noticed in almost every, if not even every, liftoff so that was something I just have gotten used to, thats why, this time, I reacted to the almost 180 it did.

I quite often fly over or under the powerlines I mentioned and with the Air I have never noticed anything, but the Spark I owned before showed errors now and then as I flew close to the lines (close as in several meters away, dont want to hit them and crash

)

Thank you so much for your time and the explanations in this matter, it was greatly appreciated.

slup · Sep 12, 2021

Slipurson said:
I think I will start using your method when taking of ...

You do that ... experience a yaw error in a confined space isn't pleasant at all.

This with the flyaway is due to that the IMU (which keep track of the flight direction during flight) think's it pointing in another direction than it does in reality. All will be fine to start with ... but as soon as the flight controller notice that it have to command the motors to hold position it will command the wrong motors --> the position error just grows instead of getting smaller -->it commands the motors again, and again ... but always the wrong ones. This happens in a lightening speed & the drone speeds away ... often exceeding both max tilt angels & max speeds. If the IMU was initialized close to 90 degrees wrong the flyaway path usually are pretty straight .. if closer to 180 degrees the path will be curved. If you can avoid crashing, the drone will eventually go into ATTI mode as the flight controller gives up ... but usually the drone have crashed long before that.

This is a classic one with a Spark ... he's starting from a balcony, which most probably have reinforced concrete which deflected the compass.