GPS and compass failure in Sport mode

[beach_Dr]

Hi
I posted this as a reply to another topic, was suggested to start own thread.

New here and new to drones in general. I have about 2 1/2 hours flying time on my Mavic. In the last week (first time using sport mode), dropped to Atti mode 3x in 1 flight, GPS loss. Contacted DJI,suggested firmware update and re-callibrate compass, which I carried out.
On Saturday past, dropped GPS 8 times in first flight following firmware update, plus several compass errors, and a general error on landing. All events have been in sport mode. Most drops are less than 5 seconds, but last drop on the most recent problem flight lasted about 30sec, with drone drifting badly in Atti until I dropped altitude enough and brought it in from over the ocean onto land for vision sensors to kick in.

I've obviously been scouring the forum for posts on the GPS subject.

DJI have since just suggested flying from another location, or sending in for service if problem continues (How many times must it happen before it is considered service worthy?). Have been flying from beach.

Has anyone with these problems sent their MP in for service and had the GPS issue successfully fixed?

I again recalibrated compass have done 2 short flights today (very windy), did not enter sport mode, and the MP behaved as well as could be expected in the wind.

 

[beach_Dr]

Here is list of errors received on the MP, starting about 10min into flight.

06:08:24 Compass Redundancy Switch
06:08:24 Satellite Positioning Off. Fly with caution
06:08:26 Satellite Positioning Off. Fly with caution
06:08:29 [GPS]data exception (In Flight) (Using)
06:08:30 Satellite Positioning Off. Fly with caution
06:08:31 [GPS]data exception (In Flight) (Using)
06:08:40 Satellite Positioning Off. Fly with caution
06:08:41 [GPS]data exception (In Flight) (Using)
06:09:03 Satellite Positioning Off. Fly with caution
06:09:04 Satellite Positioning Off. Fly with caution
06:09:06 [GPS]data exception (In Flight) (Using)
06:09:22 Abnormal compass function or GPS signal detected. Aircraft switched to ATTI mode.

On landing, had a general aircraft status error as follows:

[GPS]data exception please ask technical assistance for help

I assume the compass redundancy switch is just the new firmware trying to compensate for sport mode.

link to DAT file (copied straight from internal memory)
Dropbox - FLY034.DAT

link to larger DAT file (created from DJI Assist, incorporating vision data)
Dropbox - DJI_ASSISTANT_EXPORT_FILE[2017-03-05 20-43-33].DAT

thanks
Really enjoying the MP, but getting wary of using sport mode!

 

[BD0G]

This thread may prove useful at diagnosing your issue.

DIY rear compass cabling fix

 

[Mavicsailor]

Hi
I posted this as a reply to another topic, was suggested to start own thread.

New here and new to drones in general. I have about 2 1/2 hours flying time on my Mavic. In the last week (first time using sport mode), dropped to Atti mode 3x in 1 flight, GPS loss. Contacted DJI,suggested firmware update and re-callibrate compass, which I carried out.
On Saturday past, dropped GPS 8 times in first flight following firmware update, plus several compass errors, and a general error on landing. All events have been in sport mode. Most drops are less than 5 seconds, but last drop on the most recent problem flight lasted about 30sec, with drone drifting badly in Atti until I dropped altitude enough and brought it in from over the ocean onto land for vision sensors to kick in.

I've obviously been scouring the forum for posts on the GPS subject.

DJI have since just suggested flying from another location, or sending in for service if problem continues (How many times must it happen before it is considered service worthy?). Have been flying from beach.

Has anyone with these problems sent their MP in for service and had the GPS issue successfully fixed?

I again recalibrated compass have done 2 short flights today (very windy), did not enter sport mode, and the MP behaved as well as could be expected in the wind.

I'm curious as to when this Mavic was purchased. Please let us know. I took a look at the DAT file and there appears to be a lot of electromagnetic noise after switching to sports mode at high currents. This is consistent with the problem others have seen before the factory and field fix where the back motor wires were twisted as BD0G points out in the previous post. You also show messages that indicate the compasses may have switched from back to front, as you indicate attempting to compensate for this noise. (this was a FW upgrade) Interesting that I couldn't look at MagZ front and back, or RC control. I thought I read that others may have also had problems with these values after a FW upgrade. BudWalker, could this possibly have something to do with the values you used to calculate these MagZ values that may have changed after a FW change? Maybe it was just a one off.

In the graph below the purple background is ATTi. As the current increased, so did the noise as shown by magmod. I used this value to show overall noise as I was unable to display MagZ front and back. BTW, your top speed was about 45MPH. You probably had a bit of a tailwind.

Maybe you could follow the instructions in the above post shown by FD0G (thanks!) and take a look to see if you have twisted wires. I believe the factory twisted wire mod was done sometime around mid December. Without being able to look at MagZ front and back, at least for me, its difficult to tell if your Mavic has the twisted wires. I can only tell that the drop into atti occured when their was non-specific electromagnetic noise.

 

[beach_Dr]

Mavic was ordered mid october, received 1st week January. Part of first second batch to reach South Africa. So as to when it was manufactured... anywhere I can look?

Should I post pre-firmware update flight DAT for comparison? Only had GPS fall out then, no compass issues.

I realised the effect of tail wind the first time I used sport mode and was wondering why it was only flying up the beach at 42kph (26mph), when I looked at flight log saw it had reached 70.2kph (43.6mph) flying down the beach... That was on previous flight though

 

[beach_Dr]

This thread may prove useful at diagnosing your issue.

DIY rear compass cabling fix

Thanks for the link - I'll definitely give the leg twisting a try. Not brave enough to remove the whole top - may still need a warranty!

 

[Mavicsailor]

Mavic was ordered mid october, received 1st week January. Part of first second batch to reach South Africa. So as to when it was manufactured... anywhere I can look?

Should I post pre-firmware update flight DAT for comparison? Only had GPS fall out then, no compass issues.

I realised the effect of tail wind the first time I used sport mode and was wondering why it was only flying up the beach at 42kph (26mph), when I looked at flight log saw it had reached 70.2kph (43.6mph) flying down the beach... That was on previous flight though

I believe others are still trying to crack the date of manufacture code. Yours very well may not have the twisted wires depending on how long it sat on a shelf. I'm not sure why I can't view magZ front and back. Not sure if firmware related, but BudWalker our true expert would know. If you'd like, go ahead and post some older data and I'll take a look when time. Was the latest flight done after upgrading to the latest FW a few days ago?

 

[beach_Dr]

I believe others are still trying to crack the date of manufacture code. Yours very well may not have the twisted wires depending on how long it sat on a shelf. I'm not sure why I can't view magZ front and back. Not sure if firmware related, but BudWalker our true expert would know. If you'd like, go ahead and post some older data and I'll take a look when time. Was the latest flight done after upgrading to the latest FW a few days ago?

The DAT I posted was after the newest firmware. Will post an older pre-firmware file tonight, may be interesting to compare

 

[BudWalker]

.....Interesting that I couldn't look at MagZ front and back, or RC control. I thought I read that others may have also had problems with these values after a FW upgrade. BudWalker, could this possibly have something to do with the values you used to calculate these MagZ values that may have changed after a FW change? Maybe it was just a one off.
......View attachment 7876

There is a new CsvView (version 1.0.6) available that works with .DAT files generated by Firmware v01.03.0500. It also has limited support for Inspire 2 and Matrice M600. Version 1.0.6 is labelled as a Beta because it was an unanticipated release. To obtain this version 1.0.6 go here.

 

[Logger]

Mavic was ordered mid october, received 1st week January. Part of first second batch to reach South Africa. So as to when it was manufactured... anywhere I can look?

Here We can tell your build date it you post pass us the first 8 character of your serial no. such as 08QDDAF0

 

[Randy]

Thanks for the link - I'll definitely give the leg twisting a try. Not brave enough to remove the whole top - may still need a warranty!

Removing the top is the best way, better and safer than trying to remove just the boom and twisting, I had the same problem, it's a very easy fix, remove about 12 screws and you can make the twists much easier.

 

[Mavicsailor]

Thanks to BudWalker, we now have the latest CsvViewer, and I am able to view your MagZ back and front.

Unfortunately, it appears your Mavic does not have twisted wires to the rear motors. Those with twisted wires typically have a max PP (peak to peak) reading of 60-70. Yours is far higher than that. The first one below is yours. The second graph is from a Mavic that has twisted wires. The green background is sports mode on both. Guess it's time to get your tools out. The guys on this forum made it sounds fairly easy to fix.

 

[Mavicsailor]

Sorry, Posted wrong image first time. Here's what a twisted wire fix looks like

 

[BudWalker]

Thanks to BudWalker, we now have the latest CsvViewer, and I am able to view your MagZ back and front.

Unfortunately, it appears your Mavic does not have twisted wires to the rear motors. Those with twisted wires typically have a max PP (peak to peak) reading of 60-70. Yours is far higher than that. The first one below is yours. The second graph is from a Mavic that has twisted wires. The green background is sports mode on both. Guess it's time to get your tools out. The guys on this forum made it sounds fairly easy to fix.

View attachment 7892

With Firmware v01.03.0500 the back and front compass data are scaled and offset the same as magX, magY, magZ data. Before Firmware v01.03.0500 the back and front had a scale of roughly 3.5. What this means is

1) a 60-70 PP before Firmware v01.03.0500 is equivalent to 227 PP after Firmware v01.03.0500.
2) comparisons between front, back and magX, magY, magZ don't require assumptions. This was my main concern about making these comparisons and drawing conclusions.

 

[Mavicsailor]

With Firmware v01.03.0500 the back and front compass data are scaled and offset the same as magX, magY, magZ data. Before Firmware v01.03.0500 the back and front had a scale of roughly 3.5. What this means is

1) a 60-70 PP before Firmware v01.03.0500 is equivalent to 227 PP after Firmware v01.03.0500.
2) comparisons between front, back and magX, magY, magZ don't require assumptions. This was my main concern about making these comparisons and drawing conclusions.

 

[Mavicsailor]

Thanks! Good to know when making comparisons between old and new CsvViewers. With this the first plot will be divided by 3.5 as a comparison? That brings the first graph above lower but still not sure if as low as the second, but close. Working off iPad now and just eyeballing it. Maybe it does have twisted wires. Maybe take it on a vacation flight and ask security if you can take a peak at the X-ray?

 

[beach_Dr]

Just checked my Mavic re twisted wiring:
Serial no is DCHA (according to interpretation means built 17 December), still has the dustbin picture, so not that recent, but popped the top a little to inspect and it does have twisted wiring already.
So that rues that out as cause of the problem.

 

[Mavicsailor]

Good to know. With that it may be best to let DJI take a look. For some reason EM Noise is causing issues on your Mavic. Could be something as simple as a bad compass or GPS. Maybe soldering connections making your Mavic more sensitive to Noise. Too bad DjI will never communicate the root cause.

 

[BudWalker]

Thanks! Good to know when making comparisons between old and new CsvViewers. With this the first plot will be divided by 3.5 as a comparison? That brings the first graph above lower but still not sure if as low as the second, but close. Working off iPad now and just eyeballing it. Maybe it does have twisted wires. Maybe take it on a vacation flight and ask security if you can take a peak at the X-ray?

It's not a comparison between old and new CsvView. It's a comparison between .DAT files coming from before and after Firmware v01.03.0500. The noise pk2pk magFront and magBack values in a pre Firmware v01.03.0500 .DAT should be multiplied by 3.5 in order to compare them to the noise values in a post Firmware v01.03.0500 .DAT.

Before Firmware v01.03.0500 the best way to evaluate back compass noise was to compare it to the front compass noise on the same Mavic in the same flight. This eliminates most of the assumptions that would otherwise have to be made. Doing that for this flight shows that the back compass is 3 to 6 times noisier. It also shows some really interesting behavior that I've not seen before. Here is magX,Y,Z at the moment the switch from the back to the front occurred.

Both magZ and magX are noisier before the switch. Here are the entries from the eventLog
521.902 : 2724876816 : 29994 [L-FDI]COMPASS(0): disconnect
521.902 : 2724877641 : 29994 [L-FMU/DM]Busy Device Changed. Type:compass0, <ID:7 idx:0-->ID:8 idx:1>, change times
522.060 : 2725587270 : 30001 [L-DBG1]forearm_led_ctrl 255, cmd 0, type 0
522.420 : 2727207596 : 30019 [L-DBG1]forearm_led_ctrl 255, cmd 0, type 0
522.445 : 2727320540 : 30021 [L-FMU/LED]action changed. compass out of range(0)
:
What's even more interesting is the back compass which flat lined for about 2 seconds



The gpsHealth-drops-from-5-to-1 occurred 0.1 secs before the back compass resumed reporting

 

[Mavicsailor]

BudWalker

Thanks for the clarifications. You never stop amazing me with the detail you can pull from flight logs. This indeed is an interesting one.