EKF2 Yaw Inconsistent

cyberpilot360,
im flying 4.03 with pixhawk1, pixhawk2 (Black Cube) and pixhawk4 (holybro)
On all Hardware the same symptoms:
not permanent; mostly after a while when they stand still on the ground; reboot Ctrl+F helps; when the message is their arming is not possible. etc…
It started with software 4.0

But if it consistently, their must be another problem as well. Calibrate all the sensors again would be my next step.

@rmackay9 randy, you are so quite in this issue, can we do something to help?

There’s a global pandemic going on, let’s be a little patient please. I’m sure the team will get to this when things calm down.

I’ve started to look at this in the codebase. My EKF knowledge is limited but I’ll post here if I find anything.

Nikhil, I know, and I didn’t want to put stress on anyone. It was more an offer to test or help.

What you mean reboot ctrl+f? I don’t see a reboot option on the pop-up screen. In my case it goes away after 2-3 tries so it’s not always there.

EDIT: Found the command ! Really helpful actually thanks:)

One workaround is to disable the INS checking in arming_check (set to 65518) until the problem gets fixed. Use this at your own risk!

Funny, today I flow a new build kopter, cube black, firmware 4.0.3.
This time I start at zero. A new controller, new calibration, writing all the params new in manual (and not from an old saved file) etc.
First Time for long that I didn’t here any “yaw inconsistenc at xxx deg”
Maybe that is the key… To reset everything and start from the beginning

That’s interesting anybody else willing to test ?

rmackay : Jan 13

@Harald, @mboland, @sergbokh,

Thanks for the report on the “DCM Yaw inconsistent” pre-arm message. I think this was added to Copter-4.0 as an additional safety check of the attitude before takeoff but I’ll bring it up at our weekly dev call tomorrow and consider removing it. Copter never falls back to DCM so it might be OK to remove it especially if we are finding there are more false positives rather than real issues.

By the way “DCM” stands for “Direct Cosign Matrix” and it is attitude estimation system that pre-dates the EKF (mentioned on the wiki here )

Ok so it will be repaired in next firmware releases ?

I have the same behavior on my drone, 4.02, now and then.

I had this for the first time today - Pixhawk with ArduCopter V4.0.3 the inconsistency was between 20 and 28 degrees, but only when it was on the ground.

We have this problem with 2 drones, both cube black and RM3100 compass. It comes 1-2 minutes after connecting the battery. Only solution so far, we let the cube warm up and restart.

It does not happen with another drone, cube orange and here2.

The question is, where does this problem come from?

I have some problems on a drone with RM3100 too. The more i use and try to fix it the more i think there may be something wrong with the driver. Being relatively new it could be some little bugs that will be solved in next releases.

I get this problem sometimes too. Reboot seems to solve it though.

I don’t have anything to add other than I fly with a different flight controller and have a few rm3100s and I am using 4.0.3 with zero issues…

I don’t think its an AC4.x and rm3100 interaction

I have a MRO Kitcan coming in Wednesday with a RM3100, I will test that as well

I solved by changing the CAN cable to STANDARD(i2C), and using the i2c external compass. I had many problems using the CAN system, many times the board (Pixhawk 2.1 cube black) does not recognize a compass.

I have always solved the can compass not being recognized by changing the brd_boot_delay to 5000 or 7000 On my Here2

Little update from us, we used the Drotek RM3100 and this GPS Module with integrated Compass:

We connected the Compass from the GPS module, calibrated it, same problem. Because the black cube had some gyro drift on the IMU2, we had to exchanged it with an orange one.

Did a new calibration with the lis3mdl, the problem did not occur so far and we will test it tomorrow with the RM3100 again.

Anyone come up with any work around for this issue? I am also getting this “EKF2 Yaw Inconsistent by XX degrees” message after flight that is more than about 2min long. Consistently about 15sec after I land I get this message and cannot arm. Rebooting makes the issue go away (until I land again) or I can wait about 3min and the message will go away.
Ive tried recalibrating many times and disabling compasses with no success. (V4.0.3)

There are two workarounds that I know of:

  1. Completely switch over to EKF3 and disable the EKF2 cores.
  2. Set ARMING_CHECK to 65518, which disables the IMU checks and thus ignores the EKF2 yaw incosistent messages.

I don´t know which one is less risky, but they work.