Please select Into the mobile phone version | Continue to access the computer ver.
The dreaded compass / yaw / speed error and switch to ATTI mode
2412 18 2017-8-13
Uploading and Loding Picture ...(0/1)
o(^-^)o
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

And so the thing that I read much about, yet hoped would not befall me, happened to me today, on 32nd flight, after having spent 54km in the air in total.

The aircraft gave the dreaded "Compass error", "Yaw error", "Speed error" and switched to ATTI mode.

Here is the visualisation of the flight log at Phantomhelp: http://www.phantomhelp.com/LogViewer/LTPWIMXHH5JMHCH82160/#
And here are the logs from the app and the AC: https://drive.google.com/file/d/ ... Fk/view?usp=sharing

I was flying from a grassy field. I noticed the first anomaly during the pre-flight check - the AC arrow on the map was not pointing in the exactly same direction as the AC itself - the reading was a few degrees off. I opened the sensors menu to check the compass reading, and it showed both compasses as green, with compass #1 being in use (and having a value of about 180, while compass #2 had 6). I decided to carefully take off and everything seemed to be fine. Not entirely trusting the situation, I continued ascent from 10m hover while at the same time rotating the aircraft (taking an ascending panorama).

Then the errors started to roll. First a short burst of "Compass error" at 1m, while AC remained in GPS mode, then everythign seemed to be fine for a while, then again a short burst at 2:52, and then the final barrage of
"Speed error"
"Yaw error"
"Compass Error. Exit P-GPS Mode; Yaw Error", and finally
"In Flight,workingIMUencountersheading exception,please switch to atti mode if craft behave abnormally; Compass Error"  at 3:22.

The last error with "please switch to atti mode if craft behave abnormally" is strangely worded as there is no way to switch it to ATTI mode, and besides it was already in ATTI mode for a whole second by the time of this error.

The wind was at about 4ms and Mavic began drifting. I started bringing it down, while correcting for the drift.

At 3:46 the aircraft switched back to P-GPS mode as if nothing happened. I continued to land manually at 4:15.

During the whole of the encounter 20 GPS/GLONASS sats were visible, I was in an open area and no cloud cover. I took a look at the AC log, an it seems that as one point Mavic simply stopped listening to the GPS sats.

Would it not be best for Mavic to continue using GPS heading in the cases when it disagrees with the compass heading? Why did not Mavic switch to the second compass if the reading on the first one were off?

Maybe some of the resident log gurus or DJI techs can take a look at the logs for any anomalies?

After landing I did a compass calibration (compass #1 went from 180 green to 8 green), took off from almost the same spot, then flew for good 17 minutes over the park, training on coordinated turns and not wanting to fly too far away after the incident, but all went smoothly.


2017-8-13
Use props
hallmark007
Captain
Flight distance : 9812789 ft
  • >>>
Ireland
Offline

First just to give you an idea what happens to AC in this situation.

Why would a compass error disable GPS?
Orientation (compass data) and position (GPS data) are different but related.
When the Mavic is moving and getting position information from the GPS that tells the Mavic that it is travelling in a different direction from what it's getting from compass data, The Mavics programming can't work with the data conflict.
The solution is to drop data from one source and since the Mavic can't fly without a compass but can without GPS, it's the GPS data that gets dropped.

There have been some compass problems with Mavics  lately or your issue may have been related to something you did.
I'd be curious about when and where you calibrated it and where you launched from as these may be relevant factors.

2017-8-13
Use props
BudWalker
Captain
Flight distance : 5426050 ft
  • >>>
United States
Offline

I looked at this flight some. The Yaw and magYaw values look like what would be expected if the compass were not calibrated properly. Did you recently install equipment such as a tracker? Also, you stated that you calibrated the compass after this flight. Would it be possible to get the .DAT for the flight after the calibration?
BumblerBee.jpg
2017-8-13
Use props
DJI Thor
Administrator
Flight distance : 13602 ft
Offline

If you found that the red arrow's attitude is not pointing the same as the actual direction of the drone, we suggest that you calibrate the drone first.  
Had you contacted our support to report this case? If no, please send an email to our support team support@dji.com and start a ticket, we will help to make a data analysis.
2017-8-13
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

DJI Thor Posted at 2017-8-13 23:23
If you found that the red arrow's attitude is not pointing the same as the actual direction of the drone, we suggest that you calibrate the drone first.  
Had you contacted our support to report this case? If no, please send an email to our support team  and start a ticket, we will help to make a data analysis.

Thanks, DJI Thor. In retrospect, I should have done that, but to my defence, I saw recommendation to not do calibration unless the app actively asks for it. And in this case the sensor reading was in the green with no prompt to calibrate.

I'll be sending a support e-mail...
2017-8-14
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

BudWalker Posted at 2017-8-13 15:56
I looked at this flight some. The Yaw and magYaw values look like what would be expected if the compass were not calibrated properly. Did you recently install equipment such as a tracker? Also, you stated that you calibrated the compass after this flight. Would it be possible to get the .DAT for the flight after the calibration?

Hi, thanks for the feed-back. I do indeed have a passive tracker (Loc8tor) attached, but I've done over 15 longish flights with it with no problems.

Before this calibration, I only did compass calibration once, when Mavic first arrived and the app displayed a message prompting me to calibrate.

Here is the log of the flight that followed right after the calibration:
https://drive.google.com/file/d/ ... Ws/view?usp=sharing
2017-8-14
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

hallmark007 Posted at 2017-8-13 15:18
First just to give you an idea what happens to AC in this situation.

Why would a compass error disable GPS?

This is what baffles me - I would consider GPS readings to be more reliable than compass if you have 20 or thereabout sats locked (unless you are in a conflict zone, where civilian GPS accuracy is reduced or garbled).

See my above answer re last calibration.

What I find strange, is that Mavic eventually switched back to P-GPS, though according to the logs it landed 2 meters below and a bit off to its starting point, whereas in rality it took off and landed in exactly the same spot.


ATTI.JPG
2017-8-14
Use props
Bekaru Tree
Captain
Flight distance : 15513100 ft
South Africa
Offline

forgive if this appears to be an irrelevant question - Can you let me know what the weather conditions were. I am also trying to understand what causes IMU vs GPS conflict and i wonder if quick changes in temperature or humidity might be something to do with this. my angle is that if the weather (or some relevant condition changes quickly that it might temporarily confuse the barometer which then gets very confused when arguing with gps - then kicks out gps when it disagrees - could be other reasons , i do not know
2017-8-14
Use props
hallmark007
Captain
Flight distance : 9812789 ft
  • >>>
Ireland
Offline

BumblerBee Posted at 2017-8-14 10:54
This is what baffles me - I would consider GPS readings to be more reliable than compass if you have 20 or thereabout sats locked (unless you are in a conflict zone, where civilian GPS accuracy is reduced or garbled).

See my above answer re last calibration.

The problem with only having gps in dji craft it doesn't know which direction to go unless it has compass and Imu directing AC , there is some thinking that maybe having gps loose compass and maybe give option to land, but I just don't know much about how exact this might work, I do however know how these craft work now and most of the time it's safe and works pretty well.
2017-8-14
Use props
BudWalker
Captain
Flight distance : 5426050 ft
  • >>>
United States
Offline

BumblerBee Posted at 2017-8-14 10:51
Hi, thanks for the feed-back. I do indeed have a passive tracker (Loc8tor) attached, but I've done over 15 longish flights with it with no problems.

Before this calibration, I only did compass calibration once, when Mavic first arrived and the app displayed a message prompting me to calibrate.

Much better. Do you have a .DAT from before you attached the Loc8tor? And, then one right after the Loc8tor was attached?
BumblerBee1.jpg
2017-8-14
Use props
DJI Thor
Administrator
Flight distance : 13602 ft
Offline

BumblerBee Posted at 2017-8-14 10:46
Thanks, DJI Thor. In retrospect, I should have done that, but to my defence, I saw recommendation to not do calibration unless the app actively asks for it. And in this case the sensor reading was in the green with no prompt to calibrate.

I'll be sending a support e-mail...

Sometimes it would be better to calibrate the compass if you change another place to fly. Also, you also need to calibrate your mobile device's compass. That will help to correct the attitude indicator.
2017-8-14
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

Bekaru Tree Posted at 2017-8-14 14:27
forgive if this appears to be an irrelevant question - Can you let me know what the weather conditions were. I am also trying to understand what causes IMU vs GPS conflict and i wonder if quick changes in temperature or humidity might be something to do with this. my angle is that if the weather (or some relevant condition changes quickly that it might temporarily confuse the barometer which then gets very confused when arguing with gps - then kicks out gps when it disagrees - could be other reasons , i do not know

The weather was pretty good - mostly clear sky with scattered light clouds. Wind of 4-5m/s at 120m according to UAV Forcast (3-4m/s on the ground).
DJI_0021_mini.jpg
2017-8-15
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

BudWalker Posted at 2017-8-14 17:22
Much better. Do you have a .DAT from before you attached the Loc8tor? And, then one right after the Loc8tor was attached?

The challenge will be to recall when I attached it

In the meantime, I remembered one detail: the flight preceding the faulty flight happened in a completely different place, but on the same day. When I ended that flight, I landed on a concrete path (taking off was done from a safe grassy area). I wonder if landing on concrete could have affected the AC. Here is the log of that preceding flight:

https://drive.google.com/file/d/ ... 0k/view?usp=sharing
2017-8-15
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

DJI Thor Posted at 2017-8-14 23:47
Sometimes it would be better to calibrate the compass if you change another place to fly. Also, you also need to calibrate your mobile device's compass. That will help to correct the attitude indicator.

Thank you. I'll take a note of that. I often fly in different locations - will be keeping an eye on any compass anomalies pre-flight.
2017-8-15
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

BudWalker Posted at 2017-8-14 17:22
Much better. Do you have a .DAT from before you attached the Loc8tor? And, then one right after the Loc8tor was attached?

I think this is the before and after loc8tor attacking flights

https://drive.google.com/file/d/ ... jQ/view?usp=sharing
2017-8-15
Use props
BudWalker
Captain
Flight distance : 5426050 ft
  • >>>
United States
Offline

BumblerBee Posted at 2017-8-15 10:39
I think this is the before and after loc8tor attacking flights

https://drive.google.com/file/d/0B8KNePIdzwBsb3BrUng5OTRYYjQ/view?usp=sharing

With the exception of FLY073.DAT all the .DATs that you have submitted look normal. In particular, there was no change when the Loc8tor was attached.

Did you move or relocate the Loc8tor prior to FLY073?

FLY073 has what would be expected if a magnetic anomaly were located on the AC and rotated with the AC. FLY073 does not have any of the characteristics of a magnetic anomaly external to the AC. In particular, there is no magYaw/Yaw separation after launch as the AC ascends. At 10 meters height the magYaw/Yaw separation begins but that is coincident with the AC starting to rotate. I'd show you one of those plots but I'm on a friends computer and can't manange it.

I'm gonna have to give up on this one. I'll think about it some more though. I did also take a look at the eventLog streams for FLY072, 073, 074 and didn't see anything. Usually, the eventLog stream doesn't help much; the content and format changes with every firmware release.

You had asked if the geomagnetic environment from a previous flight could have affected this one. The answer is no. Every flight is independent.

2017-8-16
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

BudWalker Posted at 2017-8-16 06:21
With the exception of FLY073.DAT all the .DATs that you have submitted look normal. In particular, there was no change when the Loc8tor was attached.

Did you move or relocate the Loc8tor prior to FLY073?

Thank you for the in-depth analysis.

No, the loc8tor tag is staying in the same spot all the time since I attached it.

Once my time-off coincides with good weather, I'll try to repeat the same rotating ascent over a field nearby, to see if the problem would occur again. It would be interesting to verify if rotating ascent could have been the compass-confusing factor.
2017-8-16
Use props
BumblerBee
First Officer
Flight distance : 639764 ft
Norway
Offline

I flew today in similar conditions: clear sky, windy (5m/s), and tried the same rapid rotating ascent to 120m. After that flew a few round practising linear POI maneuver. No anomalies.
Here is the log for the curious:
https://drive.google.com/file/d/ ... Dg/view?usp=sharing
2017-8-16
Use props
DroneFlying
Captain
Flight distance : 10774613 ft
United States
Offline

BumblerBee Posted at 2017-8-16 11:47
I flew today in similar conditions: clear sky, windy (5m/s), and tried the same rapid rotating ascent to 120m. After that flew a few round practising linear POI maneuver. No anomalies.
Here is the log for the curious:
https://drive.google.com/file/d/0B8KNePIdzwBsYjBqQWdiMlgzZDg/view?usp=sharing

That one (FL079.DAT) looks fine too:

BumblerBee.png
2017-8-16
Use props
Advanced
You need to log in before you can reply Login | Register now

Credit Rules