PreArm: "Internal errors 0x100000 l:921 flow_of_ctrl"

I get the “Internal errors 0x100000 l:921 flow_of_ctrl” after some minutes of flight.

Is this message important?
Here is my log:
https://drive.google.com/file/d/1CFgpPS6J3m6fcS-KmnmhfRpPV-0IhbRd/view?usp=sharing

There’s some discussions and work being done on this already.
I’ve found if you get that error during or after flight, you cant arm again without a power-cycle.

You can try some mitigations to reduce the CPU load and long loops:

EK3_IMU_MASK,3 (was 7)
INS_FAST_SAMPLE,1 (was 5)
INS_USE3,0 (was 1, it may even reset to 1)

Separately, the quad is flying OK but you’ve got a bunch of default PID values and things - we could make that quad fly better. Yell out if you want help with that.
Current monitoring would be nice if you can set that up.
It’s also a bit overweight, just, so consider ways to reduce weight or make serious changes if you ever need to add payloads.

1 Like

Hello Shawn.
I have a question about the EKF lanes.
What amount of difference between the values of the EKF lanes is acceptable?
according to the figure below, there are about 4 degrees of differences between the first and second lanes for the pitch angle. It seems that the first and third lanes have close values.

Isn’t it better to set EK3_IMU_MASK = 5?

I would be grateful if you could guide me in this matter.
I really appreciate your help.

Just to keep discussions linked we’ve got two other discussions of the Internal Error 921 here and here.

1 Like

The third IMU is not vibration-isolated in your Cube Black, making it the one least likely to be relied upon. Best to use the first two IMUs making the mask 1+2 = 3 , instead of 7 for use everything.
You definitely only want to fast sample the first IMU too.

Do a test flight and see if you still get the errors.

As Randy mentioned there’s other discussions, and there’s work going on in the background.

1 Like

I’ve placed links to some alpha testing binaries in this thread that may help resolve these issues.

No pressure of course but if people are happy to help us test these binaries that would be a great help!

1 Like