Run Away Drone in the Mountains

[bwildes]

Hi I am hoping this community can help me figure out why my drone started randomly flying away from me earlier today. I had nine satellites connected to the drone and when I started the drone up and launched the drone using the automatic take-off, no warnings or issues were displayed. I started walking a few feet away and didn't see the two compass errors display on the screen. The next thing I knew the drone started flying away without having been promoted by me to start flying in any direction and without me having touched the controller. I tried a few joystick commands, but the drone kept flying away. Thankfully my luck or by pushing down on the elevation joystick I was able to crash the drone into a few trees which allowed the drone to fall into a soft pile of leaves (thankfully the drone still works). Attached is the flight data from the phone and a link to some other data. Any help understanding what went wrong would be appreciated. (Link to phantom help DJI Flight Log Viewer - PhantomHelp.com).

 

[sar104]

That was almost certainly caused by magnetic interference at the location where you powered up the drone. While you didn't input any stick commands, when it started to drift a little then its attempted correction was in the wrong direction, leading to uncontrolled flight. The signature in the comparison between inertial and positional data are very characteristic:



If you can retrieve the mobile device DAT file ending FLY042.DAT then it will have the definitive raw magnetometer data to confirm the issue.

 

[bwildes]

How do I make that graph. I was playing around with the software earlier but can't make heads or tails of what the systems all mean?

 

[slup]

Yup ... confirmed, this was a power on in a magnetic disturbed area when the IMU Yaw was initialized by the magYaw. At about 31sec drift speed is picking up (light blue) Soon followed by a major disagreement between IMU Yaw (green) & magYaw (blue) ... the disagreement was in the worst point 136 degrees. This left the AC without correct information about heading direction leading to position hold corrections in the wrong direction making the position error rapidly larger & larger ... & the AC speeds away.

Lessons learned ... After powering on your AC, connected to your RC/app & placed it in the take off spot , ALWAYS check that the AC icon on the map in your app is pointing equal to reality ... if not, abort launch attempt, power down & move away, power up again and make a new attempt.

A good procedure when powering on the AC can be to have it horizontally in your hand (with no watch, rings or bracelets) with your arm fully stretched out ... in this way you put as much distance between your self (with perhaps metal buttons, buckles) & the ground (with possible hidden magnetic objects).

 

[sar104]

Yup ... confirmed, this was a power on in a magnetic disturbed area when the IMU Yaw was initialized by the magYaw. At about 31sec drift speed is picking up (light blue) Soon followed by a major disagreement between IMU Yaw (green) & magYaw (blue) ... the disagreement was in the worst point 136 degrees. This left the AC without correct information about heading direction leading to position hold corrections in the wrong direction making the position error rapidly larger & larger ... & the AC speeds away.

Lessons learned ... After powering on your AC, connected to your RC/app & placed it in the take off spot , ALWAYS check that the AC icon on the map in your app is pointing equal to reality ... if not, abort launch attempt, power down & move away, power up again and make a new attempt.

A good procedure when powering on the AC can be to have it horizontally in your hand (with no watch, rings or bracelets) with your arm fully stretched out ... in this way you put as much distance between your self (with perhaps metal buttons, buckles) & the ground (with possible hidden magnetic objects).

View attachment 102155

I don't think so - look at the magnetic yaw and magnetic modulus on take off, and then on the first yaw maneuver at 35 seconds. It's obviously not magnetic distortion at the takeoff location because as the aircraft ascends the magnetic yaw doesn't change much. The small change is in fact due to the aircraft pitch changing on takeoff (it wasn't level on the ground) and that's another clue itself.

If it were a magnetically distorted power up location and subsequently moved to the takeoff location then the IMU yaw and magnetic yaw would disagree on the ground, which they don't.

Instead what happens is that the IMU yaw and magnetic yaw values both agree until the first yaw maneuver, which is CCW as shown by the stick input and the rate gyros, but the magnetic yaw goes CW. That's a strong indication that the compass is being affected by an uncompensated magnetic field on the aircraft itself, which is supported by the wild variation of magnetic modulus with attitude.





The compass certainly needs to be recalibrated, and the aircraft may need to be demagnetized.

 

[slup]

I don't think so - look at the magnetic yaw and magnetic modulus on take off, and then on the first yaw maneuver at 35 seconds. It's obviously not magnetic distortion at the takeoff location because as the aircraft ascends the magnetic yaw doesn't change much. The small change is in fact due to the aircraft pitch changing on takeoff (it wasn't level on the ground) and that's another clue itself.

If it were a magnetically distorted power up location and subsequently moved to the takeoff location then the IMU yaw and magnetic yaw would disagree on the ground, which they don't.

Instead what happens is that the IMU yaw and magnetic yaw values both agree until the first yaw maneuver, which is CCW as shown by the stick input and the rate gyros, but the magnetic yaw goes CW. That's a strong indication that the compass is being affected by an uncompensated magnetic field on the aircraft itself, which is supported by the wild variation of magnetic modulus with attitude.

View attachment 102167

View attachment 102168

The compass certainly needs to be recalibrated, and the aircraft may need to be demagnetized.

Interesting, how come that the onboard uncompensated magnetic field manifest itself just there when a rudder input is given ... anything magnetic attached to the AC itself that shift position?

 

[sar104]

Interesting, how come that the onboard uncompensated magnetic field manifest itself just there when a rudder input is given ... anything magnetic attached to the AC itself that shift position?

It doesn't just manifest on rudder - it was likely incorrect all along, but until the aircraft rotated there was no disagreement between the compass and the gyros, because the magnetic field hadn't changed and the gyros hadn't detected any rotation from the original (probably incorrect) IMU yaw.

 

[DaveS]

My experience has been that mountains can produce magnetic interference in many ways. In the Rockies I actually had magnetic interference that effected the video from my professional video camera while shooting hand-held. So, I'm sure it could effect a small drone. Mountain are magnificent and weird things!

 

[sar104]

My experience has been that mountains can produce magnetic interference in many ways. In the Rockies I actually had magnetic interference that effected the video from my professional video camera while shooting hand-held. So, I'm sure it could effect a small drone. Mountain are magnificent and weird things!

They certainly are, but that's not what happened here. And "mountains" is rather overstating it in this case - this location was in the low hills near the California coast.

 

[slup]

It doesn't just manifest on rudder - it was likely incorrect all along, but until the aircraft rotated there was no disagreement between the compass and the gyros, because the magnetic field hadn't changed and the gyros hadn't detected any rotation from the original (probably incorrect) IMU yaw.

Sorry to nag you sar just want fully grip what you're saying as this conclusion is new to me ...

So if the magnetic interference was constantly present (AC magnetized) should mean that the magYaw was wrong all the time ... not just at ground at power on, instead at any height ... meaning that the drone icon direction below at start likely was wrong?

If this is the case the AC should have wrong info about the real AC heading constantly & ... can that lead to that wrong motors get's applied when trying to hold position with following uncontrolled flight ... even though no disagreement exists between IMUYaw & magYaw ... or would the AC just start to have a crabbing flight?

Then ... what made the magmod to change?

 

[sar104]

Sorry to nag you sar just want fully grip what you're saying as this conclusion is new to me ...

So if the magnetic interference was constantly present (AC magnetized) should mean that the magYaw was wrong all the time ... not just at ground at power on, instead at any height ... meaning that the drone icon direction below at start likely was wrong?

If this is the case the AC should have wrong info about the real AC heading constantly & ... can that lead to that wrong motors get's applied when trying to hold position with following uncontrolled flight ... even though no disagreement exists between IMUYaw & magYaw ... or would the AC just start to have a crabbing flight?

Correct.

Then ... what made the magmod to change?

The magnetic modulus as recorded is the total magnetic field strength after correction by the compass calibration - i.e. after the expected internal magnetic field has been subtracted. As such it should just be the earth's magnetic field strength at that location, and be constant in magnitude even when the measured Mx, My, Mz components in the aircraft FOR change as the aircraft rotates, pitches and rolls. In other words, √(Mx² + My² + Mz²) should be constant.

However, if the measured field after subtraction is not the the external field, but contains internal components as well, then those will variably increase or decrease the total field as they add or subtract to the earth's field, depending on orientation. You can set that up yourself as a simple vector problem, or just do a thought experiment: consider a simple bar magnet with locally the same field strength as the earth's magnetic field, and then rotate that in the earth's field. In one orientation (N-S aligned with earth N-S) it will add to the earth's magnetic field, and the measured field would be doubled, while in the opposite orientation they would cancel, and the measured field would be approximately zero.

 

[slup]

Correct.


The magnetic modulus as recorded is the total magnetic field strength after correction by the compass calibration - i.e. after the expected internal magnetic field has been subtracted. As such it should just be the earth's magnetic field strength at that location, and be constant in magnitude even when the measured Mx, My, Mz components in the aircraft FOR change as the aircraft rotates, pitches and rolls. In other words, √(Mx² + My² + Mz²) should be constant.

However, if the measured field after subtraction is not the the external field, but contains internal components as well, then those will variably increase or decrease the total field as they add or subtract to the earth's field, depending on orientation. You can set that up yourself as a simple vector problem, or just do a thought experiment: consider a simple bar magnet with locally the same field strength as the earth's magnetic field, and then rotate that in the earth's field. In one orientation (N-S aligned with earth N-S) it will add to the earth's magnetic field, and the measured field would be doubled, while in the opposite orientation they would cancel, and the measured field would be approximately zero.

Yep ... think I got the magmod change sar, thank's for your patience

Just wish for one small clarification to your answer "Correct" ... was it only the disagreement Yaw-magYaw coming from the AC rotation that made it ... how about the constant unknown heading, could that also have started up this without the disagreement, making the AC failing position hold? Or would that just result in crabbing?

 

[sar104]

Yep ... think I got the magmod change sar, thank's for your patience

Just wish for one small clarification to your answer "Correct" ... was it only the disagreement Yaw-magYaw coming from the AC rotation that made it ... how about the constant unknown heading, could that also have started up this without the disagreement, making the AC failing position hold? Or would that just result in crabbing?

It's not the disagreement between the magnetic and IMU yaw values that causes uncontrolled flight, it's simply the IMU yaw, which is what the FC uses to navigate, being incorrect. It doesn't matter whether the magnetic yaw is different or not.

When the IMU yaw is wrong because it was incorrectly initialized due to an interfering external magnetic field at power up then the magnetic yaw will disagree after takeoff, and will be correct. In a case like this, however, the magnetic yaw was wrong all along, because the interference was on board and never went away - it continued to be wrong and the IMU yaw continued to be wrong.

 

[bwildes]

The compass certainly needs to be recalibrated, and the aircraft may need to be demagnetized.

Is there a way then to demagnetize the aircraft safely then, and what do you think of the probability being likely the aircraft will encounter the same set of problems if I were to fly it again in that area? You mentioned non-leveled ground when it was started up, which is possible since this is a backdoor trail and the ground isn't the most level in some places, but I thought I had the drone take off from a pretty level spot? Would launching it from my hand be smarter choice then the ground if the ground itself isn't perfectly level.

 

[sar104]

Is there a way then to demagnetize the aircraft safely then, and what do you think of the probability being likely the aircraft will encounter the same set of problems if I were to fly it again in that area? You mentioned non-leveled ground when it was started up, which is possible since this is a backdoor trail and the ground isn't the most level in some places, but I thought I had the drone take off from a pretty level spot? Would launching it from my hand be smarter choice then the ground if the ground itself isn't perfectly level.

I don't think the problem had anything to do with the location - I think that the drone or a component near the compass has become magnetized. Have you stored or transported it near any strong magnets, or added any components to it?

Demagnetizing is pretty simple, using a CFixer or other less expensive demagnetizer.

The aircraft was pitched up at 11.5° on the ground, but that's not a problem. My comment was in reference to the magnetic modulus changing when it transitioned to level once airborne - that indicates pitch-dependent modulus which should not be the case.

 

[slup]

It's not the disagreement between the magnetic and IMU yaw values that causes uncontrolled flight, it's simply the IMU yaw, which is what the FC uses to navigate, being incorrect. It doesn't matter whether the magnetic yaw is different or not.

When the IMU yaw is wrong because it was incorrectly initialized due to an interfering external magnetic field at power up then the magnetic yaw will disagree after takeoff, and will be correct. In a case like this, however, the magnetic yaw was wrong all along, because the interference was on board and never went away - it continued to be wrong and the IMU yaw continued to be wrong.

Splendid sar ... good learning today also ... just a pity that it comes through others unfortunate events

 

[bwildes]

I don't think the problem had anything to do with the location - I think that the drone or a component near the compass has become magnetized. Have you stored or transported it near any strong magnets, or added any components to it?

Not that I can think of. No components have been added to it. And I don't have any magnets around me. I flew over a power line the other day, but it didn't affect the drone during that flight. I have an ID badge used for clocking in at my work station next to the drone, but that's a reach in my opinion. Otherwise nothing else I can think of being magnetized has been near the drone. Are there any common place everyday objects that I'm not thinking of which could have affected the drone?

 

[bwildes]

Would this Demagnetizer work?

 

[sar104]

Not that I can think of. No components have been added to it. And I don't have any magnets around me. I flew over a power line the other day, but it didn't affect the drone during that flight. I have an ID badge used for clocking in at my work station next to the drone, but that's a reach in my opinion. Otherwise nothing else I can think of being magnetized has been near the drone. Are there any common place everyday objects that I'm not thinking of which could have affected the drone?

Well there are a couple of things to check:

1. Power up the aircraft and set it on the ground facing a known direction - e.g. north - and then check the aircraft orientation arrow on the map to see if it is pointing in the same direction. You can also rotate the aircraft by hand and check if it reads correctly in all directions.
2. Go into the sensor section of the DJI GO 4 app and check the magnetic interference level.

 

[sar104]

Would this Demagnetizer work?

Yes - I have one of those and it works fine.