Intermittent log recording, Unhealthy GPS

ArduCopter
1

Hello,
We’re having some trouble with a couple of things that might be related.

  1. We see data recording stops in the dataflash log, for different log messages including RCIN and RCOUT. This happens mostly in Loiter (as far as I could see).

    image
    image1597×871 102 KB

  2. We occasionally receive Unhealthy GPS and GPS Glitch messages.

Is it possible this have something to do with computational load on the CPU?
I couldn’t understand how to use the MP log parameters to diagnose performance.

The data gaps in the log - meaning data was actually missing (i.e. GPS data wasn’t received), or log write was halted for some reason?

Adding a flight log and the firmware (‘custom’ firmware created with the ardupilot webtool).
FW
Log

*The Mission Planner log viewer does not show gaps, it is best seen at plot.ardupilot.com

2

What build options did you choose for the firmware? And why a Beta version?
Why disable one of the IMUs?

You would be better off using the current Stable version firmware, maybe remove a few options in the custom firmware builder in case you need to add a couple of features.

I do see what you mean about the breaks in data.

  • In the web UAV log viewer you see those gaps.
  • In MissionPlanner there is a straight line drawn from one point to the next, so it looks as if there is no missing data until you look closely.

This would be a part of the log where web UAV log viewer would show missing data

image
image1481×887 82.9 KB

and there are similar “straight lines” in the corresponding RCin and Vibration section too.

I couldnt see any missing data in many other important categories, so only some things are affected.

You have long loops and relatively high load especially during Loiter.

image
image1685×881 74.4 KB

I didnt see a compass detected or calibrated, that’s going to make things tricky.

For the GPS glitches and similar, I think you’ll still be better off with:
GPS_GNSS_MODE,65

EDIT I see GPS is showing a large amount of interference in your area - you will struggle.

I find Beidou gives me a poor position despite a large number of sats, and Galileo is GREAT if you have a very new GNSS unit (maybe with the extra bands too) - so I stick with GPS and GLONASS unless there’s a special reason not to.
You can see on the map that GPS path and IMU-calculated path sometimes deviate a lot. - that gives you the Glitch messages.
image

I would set the battery failsafe actions:

BATT_FS_CRT_ACT,1
BATT_FS_LOW_ACT,2 or 3

And this didnt make much sense, I’d expect it to be 3 at least.
INS_HNTCH_HMNCS,2

3

Why disable one of the IMUs?
I didnt see a compass detected or calibrated, that’s going to make things tricky.

I should’ve elaborate more about our build…

  • We cannot use a compass, so we’re flying GSF.
  • We also currently can use only a single IMU due to hardware limitations…

What build options did you choose for the firmware? And why a Beta version?

Seems dumb but I cannot recall exactly what options were selected, I now see an option to view the build log after it is completed and will save this next time.
As for Beta - probably a mistake.

And this didnt make much sense, I’d expect it to be 3 at least.
INS_HNTCH_HMNCS,2

My intention was to reduce CPU but now I see it’s a bitmask rather than a count. will change to 3 (or 1, if it indeed reduces CPU load).

Follow up questions:
Is it possible that Ardupilot priorities flight critical computations and avoids handling serial communication for some periods of time?
That might explain some of the GPS glitches and some telemetry issues we struggle with. The OSD also sometimes is not stable.

Can you think of other parameters that might affect CPU load?

I’ll have a read on this post here to see if there’s anything I can implement in our parameters.

4

We continue to struggle with GPS issues.
The most worrying thing is it actually affecting flight - we notice the drone flying a bit unstable when having these error.

I’m adding another flight log from a recent flight that might help someone help us.
LOG

5

I’m guessing you tried other values for GPS_GNSS_MODE since it’s on 4 now (in this log).
Did you try values of 5 or 65?
You will definitely continue to struggle in your region unless you get a very new multiband GNSS unit like a ZED-F9P and maybe an external helical antenna. Holybro have some options.

Set these:
INS_HNTCH_HMNCS,3
INS_HNTCH_REF,1

CPU Load and long loops looks reasonable for that flight controller.

6

Thanks. We’re using MATEKSYS M10-5883. Due to its smaller size - I’m guessing it’s not one of the higher end units you mentioned.

Yes we tried multiple configurations, mostly with the big 3 (GPS, GLONASS and Galileo).

As for the harmonic notch - haven’t implemented that yet, thanks for the reminder.

7

Yes I gathered it was an M10 from the logs. They do have quite a small antenna.
I think all the HNOTCH setting were in place and you just had to adjust those two params I listed and it should work.