Mavic 2 Pro Wild Ride....

[sar104]

As for the flight with the problem - things are not always what they seem at first glance. The FC reported a magnetic heading error at 138 seconds:

138.649 : 13503 [L-FDI]NS(0) FUSION(1): fault on , magn_heading_err_large​

and then attitude control failure at 140 seconds:

140.049 : 13585 [L-FDI][CTRL]: fault on , horiz_ctrl_fail​

All typical of a heading problem caused by disagreement between the compass and IMU, and that disagreement clearly occurred. From the IMU data in the txt logs we see this:



and also the expected divergence in predicted position:



Ugly, but not due to a simple compass problem. This was failure of the active IMU - in this case IMU1. Looking at the aircraft quaternions the problem becomes more apparent:



It's not just the quatZ (describing yaw) that goes wrong - X, Y and W from IMU1 all diverge from IMU0. In other words IMU1 loses track of pitch, roll and yaw. The pitch and roll errors cause both the yaw error, because IMU1 now incorrectly interprets the magnetometer data, and is likely the primary cause of the control failure.

To be more specific the IMU1 rate gyro data go bad in all three axes. It looks like a completely out-of-calibration gyro package, but could also perhaps be a processor error. We've seen this a few times now on this forum with the Mavic 2, and it always seems to be IMU1 that goes bad. Or maybe IMU1 is the default IMU and so we don't usual notice when IMU0 loses it.

Either way, I'm not sure I'd trust IMU1, at least not without a couple of successful calibrations and some careful test flights. I suspect there are more clues hidden in the DAT file, but that's what I've found so far.

 

[GadgetGuy]

It will takeoff with the distance/time calibration request active. It looks like it won't but it will accept a motor start CSC and will then desist with the calibration request.

How do you distinguish between the distance/time calibration request and a real compass interference at the launch site? You certainly don't want to fly with a truly interfered compass. Will the CSC motor start not work when there is a true compass interference?

 

[Deleted member 94047]

How do you distinguish between the distance/time calibration request and a real compass interference at the launch site? You certainly don't want to fly with a truly interfered compass. Will the CSC motor start not work when there is a true compass interference?

At the time of the flight, you can't. The only way to tell them apart is by looking at the log files after the fact which isn't really that useful. If only DJI had chosen to use more sensible phrasing like "you have moved to a new location" or "it has been this many days since your last calibration" instead of throwing up the same "strong magnetic interference" for everything compass related. They make these incredibly complex, sophisticated machines and yet they can't seem to get such simple things right. Very odd.

 

[GadgetGuy]

At the time of the flight, you can't. The only way to tell them apart is by looking at the log files after the fact which isn't really that useful. If only DJI had chosen to use more sensible phrasing like "you have moved to a new location" or "it has been this many days since your last calibration" instead of throwing up the same "strong magnetic interference" for everything compass related. They make these incredibly complex, sophisticated machines and yet they can't seem to get such simple things right. Very odd.

@sar104 seemed to imply that the M2 might not respond to a CSC motor start if it detected true magnetic interference. Is that not the case? Will it start anyway under both scenarios?

 

[Deleted member 94047]

@sar104 seemed to imply that the M2 might not respond to a CSC motor start if it detected true magnetic interference. Is that not the case? Will it start anyway under both scenarios?

I have the Air, and I really don't know the answer to that. Also, I have never tried to take off with an active compass calibration request on the screen. An interesting experiement for when I move next.

 

[jeplane]

How do you distinguish between the distance/time calibration request and a real compass interference at the launch site?

I do 2 things to differentiate:

I go to sensors on the app, look at compass. If the interference numbers are yellow or red, that answers that.

I then look at the heading of the aircraft on the map. If it is not moving, and if it is correct with its position, I am clear to go.

I have taken off multiple times with a compass calibration request on a MP2, without doing it, multiple times.

 

[CrouchyUK]

I am learning so much here but still scrabbling to keep up with the in-depth analysis- thank you so much for all your involvement and time!

From the graphs presented can I say then that at some point one of the IMU's went crazy- causing the drone to basically chase its own *** around the sky trying to get everything to agree? It seems the ONLY effect I had as a pilot was to get it to ascend- which if the flight had been over land would have been more crucial to try and prevent a crash. Still this was a miracle I got it back- which leads me to the posts about unexplained fly-aways and crashes with the Mavic 2 and the possibility DJI have a problem with these drones? The last thing I wanted to do was to be the guy that posted up a dodgy flight and immediately blame everything but the pilot, but....

I note also that on this flight all the craziness stopped and for the final return home and landing it flew normally- so the issue self-resolves?

Where does that leave me as to future flights- do I do a complete IMU and compass calibration and fly conservatively (which I do anyway)? Would it be irresponsible of me to fly a drone which may at any time head off uncontrollably and crash into the ground, buildings- God forbid even people? Is this behaviour grounds for contacting DJI and getting a replacement?

 

[sar104]

How do you distinguish between the distance/time calibration request and a real compass interference at the launch site? You certainly don't want to fly with a truly interfered compass. Will the CSC motor start not work when there is a true compass interference?

At the time of the flight, you can't. The only way to tell them apart is by looking at the log files after the fact which isn't really that useful. If only DJI had chosen to use more sensible phrasing like "you have moved to a new location" or "it has been this many days since your last calibration" instead of throwing up the same "strong magnetic interference" for everything compass related. They make these incredibly complex, sophisticated machines and yet they can't seem to get such simple things right. Very odd.

Agreed. The distinction is noted in the log file in the form:

10.932 : 454 [L-COMPASS][mag_cali_pt] date_from_last 33|​

or

-72.595 : 1623 [L-COMPASS][mag_cali_pt] distance_from_last 9258717.0 m|​

I do 2 things to differentiate:

I go to sensors on the app, look at compass. If the interference numbers are yellow or red, that answers that.

I then look at the heading of the aircraft on the map. If it is not moving, and if it is correct with its position, I am clear to go.

I have taken off multiple times with a compass calibration request on a MP2, without doing it, multiple times.

That's what I do too.

@sar104 seemed to imply that the M2 might not respond to a CSC motor start if it detected true magnetic interference. Is that not the case? Will it start anyway under both scenarios?

Good question. I haven't actually tried a motor start with real interference that was detected by the aircraft, or seen a log file that reflects that situation. Note that real interference is only detected if it is strong enough to exceed some fairly generous bounds in the firmware. Easy to test though - just bring a magnet close enough to the compass to cause the warning, and then try a motor start.

 

[sar104]

I am learning so much here but still scrabbling to keep up with the in-depth analysis- thank you so much for all your involvement and time!

From the graphs presented can I say then that at some point one of the IMU's went crazy- causing the drone to basically chase its own *** around the sky trying to get everything to agree? It seems the ONLY effect I had as a pilot was to get it to ascend- which if the flight had been over land would have been more crucial to try and prevent a crash. Still this was a miracle I got it back- which leads me to the posts about unexplained fly-aways and crashes with the Mavic 2 and the possibility DJI have a problem with these drones? The last thing I wanted to do was to be the guy that posted up a dodgy flight and immediately blame everything but the pilot, but....

I note also that on this flight all the craziness stopped and for the final return home and landing it flew normally- so the issue self-resolves?

Where does that leave me as to future flights- do I do a complete IMU and compass calibration and fly conservatively (which I do anyway)? Would it be irresponsible of me to fly a drone which may at any time head off uncontrollably and crash into the ground, buildings- God forbid even people? Is this behaviour grounds for contacting DJI and getting a replacement?

That's a fair description. If the FC doesn't know the orientation or attitude of the aircraft then obviously it finds it difficult to keep it in stable flight. It's possible that the FC switched to IMU0 towards the end of the flight to resolve the problem, but I didn't find the DAT file for the period of flight after the reboot amongst the files that you posted, so I couldn't tell. That should be another DAT labelled FLY074, but it was not either of the two that you included. You could check for another one.

 

[CrouchyUK]

That's a fair description. If the FC doesn't know the orientation or attitude of the aircraft then obviously it finds it difficult to keep it in stable flight. It's possible that the FC switched to IMU0 towards the end of the flight to resolve the problem, but I didn't find the DAT file for the period of flight after the reboot amongst the files that you posted, so I couldn't tell. That should be another DAT labelled FLY074, but it was not either of the two that you included. You could check for another one.

Sorry- I attach all of the FLY074 files again I must have missed one. The flight log (TXT) upload shows the flight stabilising somewhat towards the end, but the drone still trying to RTH- pointing the right direction but making no headway as if against a headwind. But the wind, if any- should have been behind it. I should not believe the 15mph speed reported here- it seems as though that was what it should have been doing?

 

[sar104]

Sorry- I attach all of the FLY074 files again I must have missed one. The flight log (TXT) upload shows the flight stabilising somewhat towards the end, but the drone still trying to RTH- pointing the right direction but making no headway as if against a headwind. But the wind, if any- should have been behind it. I should not believe the 15mph speed reported here- it seems as though that was what it should have been doing?

Yes - that's the missing file. I'll add that to the previous DAT and take a look this evening.

 

[GadgetGuy]

I do 2 things to differentiate:

I go to sensors on the app, look at compass. If the interference numbers are yellow or red, that answers that.

I then look at the heading of the aircraft on the map. If it is not moving, and if it is correct with its position, I am clear to go.

I have taken off multiple times with a compass calibration request on a MP2, without doing it, multiple times.

Thanks for providing a good verification of the cause of the compass calibration request. Should be interesting if it becomes a demand, and then refuses to take off with CSC before a recalibration or moving the aircraft.

 

[Deleted member 94047]

Thanks for providing a good verification of the cause of the compass calibration request. Should be interesting if it becomes a demand, and then refuses to take off with CSC before a recalibration or moving the aircraft.

I have the Air, and I really don't know the answer to that. Also, I have never tried to take off with an active compass calibration request on the screen. An interesting experiement for when I move next.

What a coincidence, I had a compass calibration request this morning since it has been, apparently, more than 30 days since my last one.

78.396 : 3783 [L-COMPASS][mag_cali_pt] date_from_last 31|
78.955 : 3811 [L-FMU/LED]action changed. compass out of range:need_cali(1)

I knew it was not actual magnetic interference because this is my usual launch site and there is nothing around that can possibly make me suspect a real interference. I checked the compass readings, it was around 73. So I tried to take off, tried CSC first and got an on screen message about there being too much magnetic interference (yeah right!). Tried the auto take off button on the screen, that didn't work either. So, at least for the Air, a compass calibration request seems to disable takeoff.

107.441 : 5239 [L-RC]rc cmd:STD_ASSI_TAKEOFF
107.461 : 5240 [L-FMU/MOTOR]start req: MR_REQ_RC_NORMAL, fail for:Compass error

 

[GadgetGuy]

What a coincidence, I had a compass calibration request this morning since it has been, apparently, more than 30 days since my last one.



I knew it was not actual magnetic interference because this is my usual launch site and there is nothing around that can possibly make me suspect a real interference. I checked the compass readings, it was around 73. So I tried to take off, tried CSC first and got an on screen message about there being too much magnetic interference (yeah right!). Tried the auto take off button on the screen, that didn't work either. So, at least for the Air, a compass calibration request seems to disable takeoff.

Hmmm...I'll check on my M2 next time I get the message to see if CSC will bypass the compass calibration mandate, after checking the compass values to see which type compass calibration demand it is!

 

[sar104]

Hmmm...I'll check on my M2 next time I get the message to see if CSC will bypass the compass calibration mandate, after checking the compass values to see which type compass calibration demand it is!

As I mentioned earlier - the M2 will takeoff when requesting a calibration due to time or distance. I confirmed that for the latest firmware just a few days ago.

 

[sar104]

Sorry- I attach all of the FLY074 files again I must have missed one. The flight log (TXT) upload shows the flight stabilising somewhat towards the end, but the drone still trying to RTH- pointing the right direction but making no headway as if against a headwind. But the wind, if any- should have been behind it. I should not believe the 15mph speed reported here- it seems as though that was what it should have been doing?

Unfortunately that's inconclusive. IMU1 appears to continue as the active IMU until the end of the flight, despite repeated fusion failure with its data. As an aside, I don't see the point of redundant IMUs if it's not going to switch in these situations. Comparison of the pure GPS and IMU0 data seem to indicate that IMU0 was performing nominally.

Anyway - the discrepancies are gradually resolved as the flight continues:



As a result, the magnetic heading error is also reduced, which is why the aircraft is able to navigate home:



None of which hints, to me at least, at a specific causal explanation. I still cannot tell whether this is a computational or hardware problem.

@BudWalker may be able to offer some further insights when he returns from his road trip.

 

[GadgetGuy]

As I mentioned earlier - the M2 will takeoff when requesting a calibration due to time or distance. I confirmed that for the latest firmware just a few days ago.

The $64,000 question is will the M2 also still take off when it encounters real compass interference, with a demand for compass calibration? Still unsure about the answer to this one.

 

[sar104]

The $64,000 question is will the M2 also still take off when it encounters real compass interference, with a demand for compass calibration? Still unsure about the answer to this one.

Very easy to test.

 

[GadgetGuy]

Very easy to test.

Yes. Hoping someone else already knew the answer, before messing with my already fussy compass, which has cost me precious flying time to keep happy, just when the lighting was perfect!

 

[Gryphon962]

Yes. Hoping someone else already knew the answer, before messing with my already fussy compass, which has cost me precious flying time to keep happy, just when the lighting was perfect!

This must be a MP2 thing. My MP1 never asks for a compass cal no matter how far I travel, unless I try to start it up near some metal; in which case, the warning goes away when I move it.