Compass Error cause or correlation?

Joined
Jul 15, 2015
Messages
507
Reaction score
163
Location
Castle Rock, Colorado
I've been flying my P3A for 1-1/2 years and have never experienced an inflight compass error - until today. I'll tell you it's one scary feeling to watch your drone shoot off uncontrollably. Fortunately, after looking down and seeing the compass error warning flashing at me, I flipped the flight mode switch (P-A-F) to A (atti) and regained control. After moving the drone to a better position I carefully tried P- mode again and found everything to be normal.
I have flown this route most every day for a year now with this P3A and have never experienced anything like this.
Here is what I know:
  1. Everything was completely normal at takeoff
  2. Everything was normal after the event
  3. Before the error, I had been yawing while hovering. I rotated 360° to the right and immediately 2 360° rotations to the left. Both at full stick. I do have the controls slowed however for smoother videos.
Now you may be wondering why I would spin the P3 at full stick any way. Well, I was experimenting with CF vs. OE props. (I had not yet swapped the props, so this was before I changed them.)
BTW: I have decided the OE props are slightly less noisy (IMO) and have smoother response to the stick. (Again, IMO)


I did the flight again after changing the props, but this time using a slow yaw and not as many revolutions and it performed perfectly.

I can't say for sure what caused it, but let's just say I will never spin the P3 around and around at full stick. (Not without wearing a diaper anyway.)

2.jpg
3.jpg
 
To me it looks like a simple connection issue. Over period of time, somewhere compass connection got lose and that became really effective when you yawed your aircraft.


Sent from my iPhone using PhantomPilots
 
To me it looks like a simple connection issue. Over period of time, somewhere compass connection got lose and that became really effective when you yawed your aircraft.


Sent from my iPhone using PhantomPilots
Thanks for your input. Where would the connection be lost? I have not noticed any abnormalities before this nor have I physically disconnected the compass. Please tell me more.
 
Any recent upgrading?
What app?
How many satellites?
 
Any recent upgrading?
What app?
How many satellites?
Nothing new.
CF gimbal guard added several months ago to strengthen landing gear during hand catching.
Only thing different is rotating it in both directions at a rapid rate.

Good thing it was above the rooftops or I would be shopping for a new one tomorrow.
 
Any recent upgrading?
What app?
How many satellites?
Latest DJI Go app 3.0.1 I believe on iPad mini2
19 satellites. I believe it dropped to 12 during the event.
The playback in the Go app actually jumped position as though it had no record of it travelling away during the compass error.
Everything was normal again when I put it back into P mode. And in the next flight several minutes later.
Battery was still tight.
Beginning of the end?
 
Latest DJI Go app 3.0.1 I believe on iPad mini2
19 satellites. I believe it dropped to 12 during the event.
The playback in the Go app actually jumped position as though it had no record of it travelling away during the compass error.
Everything was normal again when I put it back into P mode. And in the next flight several minutes later.
Battery was still tight.
Beginning of the end?
When the P3 develops a yaw issue it gives a compass error. It most cases it can clear the error by doing a few 360 rotations.
 
  • Like
Reactions: Digdat0
I've been flying my P3A for 1-1/2 years and have never experienced an inflight compass error - until today. I'll tell you it's one scary feeling to watch your drone shoot off uncontrollably. Fortunately, after looking down and seeing the compass error warning flashing at me, I flipped the flight mode switch (P-A-F) to A (atti) and regained control. After moving the drone to a better position I carefully tried P- mode again and found everything to be normal.
I have flown this route most every day for a year now with this P3A and have never experienced anything like this.
Here is what I know:
  1. Everything was completely normal at takeoff
  2. Everything was normal after the event
  3. Before the error, I had been yawing while hovering. I rotated 360° to the right and immediately 2 360° rotations to the left. Both at full stick. I do have the controls slowed however for smoother videos.
Now you may be wondering why I would spin the P3 at full stick any way. Well, I was experimenting with CF vs. OE props. (I had not yet swapped the props, so this was before I changed them.)
BTW: I have decided the OE props are slightly less noisy (IMO) and have smoother response to the stick. (Again, IMO)


I did the flight again after changing the props, but this time using a slow yaw and not as many revolutions and it performed perfectly.

I can't say for sure what caused it, but let's just say I will never spin the P3 around and around at full stick. (Not without wearing a diaper anyway.)

View attachment 67833 View attachment 67834
Would it be possible to get the .DAT file for these flights? The .DAT file is on the P3 itself, not the tablet. To see how to retrieve the .DAT go here. The bottom half of that page has info on getting the correct .DAT.

It's possible that the error occurred because of a compass calibration problem. IMHO compass calibration errors are rare and over diagnosed. But, from your description it's at least a possibility in this case The rotations performed in these flights provide the hard data necessary to determine if the compass calibration was faulty. Could provide the .DAT for the flight with the error as well as the .DAT for the subsequent flight with slower rotations?

The .DATs will be large and can't be posted here. Most pilots use Dropbox to provide a link. If you don't want to do that there is another method that I've developed. But, you'd be the first to try it.
 
When the P3 develops a yaw issue it gives a compass error. It most cases it can clear the error by doing a few 360 rotations.
It was after doing a few fast 360's that the compass error occurred. Maybe the compass wasn't calibrated as well as I thought. The sensor always showed within spec however.
I think I will not be spinning 360's again anytime soon. Out of control compass error events are no fun!
 
It was after doing a few fast 360's that the compass error occurred. Maybe the compass wasn't calibrated as well as I thought. The sensor always showed within spec however.
I think I will not be spinning 360's again anytime soon. Out of control compass error events are no fun!

BudWalker can tell us exactly what took place. I know he has diagnosed other flights that went through this same problem, and he was actually the 1st person on my mind as I read your details.

The fast 360 rotations would not have helped with what I had mentioned. I pretty much just wanted to point out that in certain compass errors the pilot might use the method I described and in turn help them get the aircraft home.

You also did the right thing which was to switch to A-mode and fly home. Not only that but you also did another really good thing which was stop and check if P-mode was again usable. I do want to point out that when flying home in A-mode to not forget about what altitude you're at verses any structure you might encounter along the way.

Don't forget the DAT file so BudWalker can take of you on that.
 
It was after doing a few fast 360's that the compass error occurred. Maybe the compass wasn't calibrated as well as I thought. The sensor always showed within spec however.
I think I will not be spinning 360's again anytime soon. Out of control compass error events are no fun!

I forgot to mention that in some cases even though it says a compass error, it's not always a compass error.
 
  • Like
Reactions: Dounin Front
Would it be possible to get the .DAT file for these flights? The .DAT file is on the P3 itself, not the tablet. To see how to retrieve the .DAT go here. The bottom half of that page has info on getting the correct .DAT.

It's possible that the error occurred because of a compass calibration problem. IMHO compass calibration errors are rare and over diagnosed. But, from your description it's at least a possibility in this case The rotations performed in these flights provide the hard data necessary to determine if the compass calibration was faulty. Could provide the .DAT for the flight with the error as well as the .DAT for the subsequent flight with slower rotations?

The .DATs will be large and can't be posted here. Most pilots use Dropbox to provide a link. If you don't want to do that there is another method that I've developed. But, you'd be the first to try it.

Hi Bud, thanks for your interest. Here are links to 3 log files I put on Google drive. You should not need to login to download.

FLY497 was a control flight done on 10/26: FLY497.csv

FLY498 was the flight with errors on 10/28: FLY498.csv

FLY499 was the flight done minutes later on 10/28: FLY499.csv

(FLY498 looks pretty crazy in Excel)
 
Hi Bud, thanks for your interest. Here are links to 3 log files I put on Google drive. You should not need to login to download.

FLY497 was a control flight done on 10/26: FLY497.csv

FLY498 was the flight with errors on 10/28: FLY498.csv

FLY499 was the flight done minutes later on 10/28: FLY499.csv

(FLY498 looks pretty crazy in Excel)
I looked primarily at FLY498. The rotations you were talking appear to be unrelated to the compass errors. The rotations finished at 47.72 and the compass errors started 13 secs later at 61 secs.
upload_2016-10-29_13-10-15.png

The compass errors are shown here with the colored background.

The compass errors started after full elevator had been applied and then abruptly removed. I've seen a lot compass errors happen where the P3 is subjected to abrupt movements like this. Probably the Flight Controller having trouble reconciling expected versus actual sensor values. Don't have a more compelling explanation why this happens except to say that the compass error detection scheme is so sensitive that it is subject to false positives. IMHO the compass error detection scheme is getting more sensitive with successive firmware updates. It's happened to me and I've ignored it.

I couldn't find any evidence of a bad compass calibration. If you're interested more about this take a look at this post

I noticed that you supplied the .csv after you ran them through DatCon. Maybe in response to my saying there is an alternate method of transmitting those files? Another way is to use CsvView. Load the .DAT into CsvView which will run DatCon internally to produce a temporary .csv. Then File->Export Workspace. This will produce a .zip containing the .csv. That .zip can then be attached to a PhantomPilots post directly. I.e. no Dropbox of Google Drive, etc.
 
  • Like
Reactions: Digdat0
@Dounin Front I looked a little more at this flight. DatCon computes a value, velGPS - velH(M/S), that is the difference between velocity reported by the P3 and the velocity that DatCon computes from successive GPS locations. The compass errors correlate with this value
upload_2016-10-29_14-50-58.png

This is consistent with the idea the Flight Controller sees values it doesn't expect.
 
@Dounin Front I looked a little more at this flight. DatCon computes a value, velGPS - velH(M/S), that is the difference between velocity reported by the P3 and the velocity that DatCon computes from successive GPS locations. The compass errors correlate with this value
View attachment 67884
This is consistent with the idea the Flight Controller sees values it doesn't expect.
Hey Bud,
Maybe the flt controller sensed the drone go off to the NW with no input from me.
Odd the little flight record in the Go app showed the drone instantly jumping to a new location during the error message.
Does the internal log show me taking control and switching back to P-mode before landing?
Do you have any advice other than going easy on the controls in the future?
I normally am very smooth with the controls so I don't stress the drone. I'm hoping to help it last a good long time.
Could any adjustments I've made to gains, etc have caused this under full stick maneuvers? I haven't changed anything for a long time, but I did smooth things things out some time ago.
BTW: thanks again for looking at this. I really appreciate it.
 
Hey Bud,
Maybe the flt controller sensed the drone go off to the NW with no input from me.
Odd the little flight record in the Go app showed the drone instantly jumping to a new location during the error message.
Does the internal log show me taking control and switching back to P-mode before landing?
Do you have any advice other than going easy on the controls in the future?
I normally am very smooth with the controls so I don't stress the drone. I'm hoping to help it last a good long time.
Could any adjustments I've made to gains, etc have caused this under full stick maneuvers? I haven't changed anything for a long time, but I did smooth things things out some time ago.
BTW: thanks again for looking at this. I really appreciate it.
This is kinda creepy. For reference here is the compass error.
upload_2016-10-29_16-17-6.png

Here are the controls with the mode switch in the background
upload_2016-10-29_16-17-57.png

Looks like you switched into F mode for a second and then corrected that by switching to A mode.
Here is where the P3 was at 61.3689
upload_2016-10-29_16-19-28.png

It moved from here to the southwest. The P3 is facing about 153 degrees. At first the sticks were neutral. But then the aileron was pushed all the way left for about 9 secs. The P3 should've moved to it's left but moved to it's right instead.

You really should install CsvView and see this for yourself.

Now that I see all this I take it back. If it were my P3 I would be concerned. Why don't you send the .DAT to DJI and ask them to explain it.
 
This is kinda creepy. For reference here is the compass error.
View attachment 67890
Here are the controls with the mode switch in the background
View attachment 67891
Looks like you switched into F mode for a second and then corrected that by switching to A mode.
Here is where the P3 was at 61.3689
View attachment 67892
It moved from here to the southwest. The P3 is facing about 153 degrees. At first the sticks were neutral. But then the aileron was pushed all the way left for about 9 secs. The P3 should've moved to it's left but moved to it's right instead.

You really should install CsvView and see this for yourself.

Now that I see all this I take it back. If it were my P3 I would be concerned. Why don't you send the .DAT to DJI and ask them to explain it.

For sure I won't be flying anymore Lichi missions till I find out more. Don't know what I would have done if it was 3/4 mi out when it happened.
I will see if DJI will help also, if they will even talk to me about a drone that's 1-1/2 yrs old.
 
Hi All
I thank you, especially Bud Walker for helping me with this issue. I took your advice and called DJI this morning. I'm actually still on with them. Here is the result.
They are very nice.
I sent them a link to the .dat log
They sent me a new .bin file. (same one I have installed) The P3 ignored it and said it was the same.
They told me to calibrate the IMU (which I did last night along with the compass)
Finally, they said they could only look at the log if the was a fly-away or a crash. I said it did flyaway, but I recovered it.
Then,
  • I do not have customer care and it is beyond warranty. Here is a link to my discussion about the issue. If someone could look at it and the log it would be very helpful before I take off again. Compass Error cause or correlation?
  • staff.png

    Oh you are beyond the warranty coverage? I do apologize but if that is the case, I am really sorry however we cannot be able to provide a data analysis service for your drone :(
Guess I'll take my chances, but no more sudden movements.
 
Considering the situation....in that the aircraft is 1.5yrs and out of warranty, and the fact that I completely disagree with directing dji to this website for various reasons.

If this were my P3A, I would research and locate others who have had the same or similar issue and find out what the cure was and go from that point. I have read recently where someone with the same issue sent their a/c to dji for repair. I can't remember 100% what they said, but I think the person said dji replaced the IMU. This took place during the time we were without a data converter which tied our hands from being able to analyze data the way it needed to be.

Now one thing is clear, and that is when someone receives their repaired a/c back they are given almost NO details whatsoever about the a/c. This is really sad. I sort of understand that when the a/c is repaired under warranty that it was broke going in, and fixed coming back seems to satisfy most. But I personally want to know everything that took place with my a/c. That tends to be valuable info later on especially after a warranty expires.

At any rate, what has happened with Dounin Front's a/c is not rare. There are others out there that has experienced the same issue. I don't think it will take much networking to find the golden info.

Btw DF, I totally agree with BudWalker in that you should install CsvView. It is not a matter of not knowing how to use it. That is one of the things about CsvView is, you can learn from it just by comparing a known area with a issue in the .txt file and pull that same point up in CsvView and see what wrong and right look like in that particular situation.
 

Members online

No members online now.

Forum statistics

Threads
143,066
Messages
1,467,358
Members
104,935
Latest member
Pauos31