Gps glitch and compass error in flight

ArduCopter Copter 4.5
, ,
1

hello.

I have a 5 inch build, with Kakute H7 as the fc and matek sam-m10q gps module with compass.
The fw i use is 4.5.0-beta1. When airborned, depending on the manuvers i make, i get errors like:
ekf imv0 emergency yaw reset, gps glitch or compass error and after that i start getting oscilations.
I usually have 20 sattelites detected, i have everything calibrated multiple times, but still get this issue.
I have attached my current settings and log file. The flight from the log is in Stabilize mode.
Another issue is if i fly in loiter or alt hold, when the glitches appear the drone flies away.

Any help for fixing this?
Thanks
config.param (20.7 KB)

And the .bin file - 1980-01-06 17-14-17.bin - Google Drive

2

Have you Methodically tuned ArduCopter ?

Especially the section about the position controller?

3

Same issue with 5" inch build and speedybeef4v3 fc.
Please note that on another 5" build and a 10" build problem does not show.
The only difference is pwm protocol does show fhe problem (old escs) instead of dshot.
Dshot does not show the problem.
No autotune at all, only basic missionplanner configuration for all three builds.
I just bought a 4 in 1 esc with blheli-s and dshot, just to see if the problem goes away upgrading motor output and response.

Alfredo

4

You have magnetic field from battery fluenced on magnetometer.
IMX IMY IMZ in XKF3 more +/- 100. Should be less 50.
If you’ll make Compass-Motor test you will see magnetic fluence more 70%.

You need move compass far from battery and make Compass-Motor test to calibrate magfield.
or just fly without real compass, like usual FPV.

5

Thanks,

this could be also my problem, but my compass is on opposite battery side, quad head, and at top, battery is bottom of quad.
I do not trust my current power board (old zmr250 frame with built in power management and old simonK 12 escs). This is why I am moving to 4 in 1 esc and upgrade motor protocol.
I will check these log values too in old logs.

Alfredo

6

Comparing param history.
I discovered that MOT_PWM_TYPE disappeared … was 0.
This is vary strange …I will update firmware and reconfigure everything …

Alfredo

7

Then make sure to use the ArduPilot methodic configurator software this time.
It will prevent you from doing most errors

8

Do You mean MissionPlanner or something else ?

9
10

In order to get again alive MOT_PWM_TYPE I had to completely copy all params from another 5 inch quad quite same configuration.
No way to go to defaults again with reset options found googling around.

I guess this is a sw bug …

I read in forums that something similar happened in the past.

Alfredo

11

No, that kind of software bug has a very low probability due to the extensive automated regression test suite.

The higher probability is that you simply did not followed instructions while configuring.

No vehicle can run on default parameters, you always need to configure it.

And the better you do it, the better it works.

Just read this forum to see how often configuration causes issues.

12

I completely agree with You,
but I am asserting something slighty different :
I got in a "stall’ config issue and I could not go back to “first installation steps”. Only way to proceed has been param copy. I did not reflash and erase all data and code, but probably it would have worked too.

Alfredo

13

I am asserting that erasing the parameters and reconfiguring it methodically to not miss any steps and ensuring the correct order would have fixed the issue in a safer way.