EKF Variance EKF_CHECK-2

I am getting EKF Variance warnings (Yellow/Red led, no buzzer) whenever I arm using 3.3rc5.

Rolled back to 3.2.1 no problems. I have two machines that have the same problem.

Can anyone confirm exactly what ERR>EKF_CHECK-2 is telling me. I can’t figure out if it is accell or gyro variance as both seem ok but the warning is always there. I have compared logs and nothing seems that different between the two.

I had a couple of flight with no problem then the message just appeared. No hardware changes.

Any ideas? Thanks

BTW. Generally I love the new firmware. I am guessing that this is not firmware but I am not sure.

I had a similar issue, I ended up out of frustration resetting the FMU & IO boards on my Pixhawk with the buttons on Pixhawk. Haven’t had the issue since, about 2 weeks. Hope this helps.

I have the same problem and resetting didn’t make any difference :cry:
What could be causing this “issue”?
I’m afraid to take it up in the air… :exclamation: :exclamation: :exclamation:

Hello all,
I am posting this in the hope that someone can offer assistance.
I am still getting EKF WARINING = EKF_CHECK-2 every time I run 3.3
It happens as soon as I arm, which is when the check routine starts running.
It also happens on the bench with no extra gear turned on to limit interference and in Stabilise mode.
Attached is a very small log file of it happening on the bench.
I would appreciate ANY HELP as I have three different 3DR Pixhawk v2 boards doing exactly the same thing.
Thanks in advance. I realise that this is not of concern to those who can run and fly with 3.3 but I am stuck on 3.2.1 at the moment. I am hoping that this is just a problem with my setup and not related to hardware.

Just an update, still not solution though. I am posting this in the hope that it helps the Developers identify any issues with the 3.3 code that may be causing this. I am flying fine with 3.2.1 and will continue to do so, I just thought that the Dev community would like to know if there are still issues running 3.3 on various boards.

I have done
1: complete recalibrations,
2: Disassembled my machines and tried running them with minimal electronics running,
3: loaded custom code to test while connected to USB (So no RF transmitter/receiver interference would be encountered)
4: Swapped out all GPS/Compass cabling
5: Changed GPS modules (although they are all the same 3DR model)
6: Reloaded GPS config from GITHUB 3DR GPS config file in the libraries folders

I have been hacking my way through the code and can narrow down the issue to the position estimates but I am not sure what is failing.

I notice that all of my logs show a flat line for EKF2>Ratio and the horizontal position estimates and readings also show flat lined values.

I had 3DR support look at my 3.2.1 logs and as far as they can see there is no hardware error. So that’s good!

I still get the EKF Variance (EKF_CHECK-2) on all of the following machines:
3DR Y6B running Pixhawk-v2, 3DR ESC & Motors, 3DR uBlox-6 series GPS/Compass.
Cinetank Mk-2 with Pixhawk-v2, T-Motor ESC’s and motors, 3DR GPS/Compass
FPV250, with Pixhawk-v2, Afro ESC’s, Multistar motors.

All of them do exactly the same thing which is to immediately switch into EKF Failsafe as soon as I arm. (With or without GPS lock).

Something has been bothering me though, with no lock my GPS will report it’s position as the last place it had a 3D Fix. Is that normal and will the non moving, fixed GPS position fight against the EKF even though it does not have a 3D fix?

Thanks and I know that this will be low on the priority list but I hope it helps. I can see that the work on 3.3 has some great stuff and I can’t wait to try the release version (as long as I can get it to work on my machines).

Here is a log from a short 3.3-rc7 hover flight. All flight control parameters are at default.
dropbox.com/s/4x3zdkxeo5yal … 4.bin?dl=0

I think this issue is now fixed. Changes to the code have stopped the EKF warning. I think it was due to the new code requireing a better GPS lock than previous versions. Initally 3.3 would give me an EKF failsafe and not allow me to arm in any mode but now it is working.

I have been enabling the fence to make sure that the EKF is happy with GPS lock as MP will report a 3D fix before the EKF is trully happy. Still a few issues with getting POSHOLD to lock properly but it seems to be working.

I also tried M8N and LEA-6 GPS modules and I have found it very hard to get the LEA=6 to get a good enough lock for 3.3 to work properly.

Thanks for the help.