A problem about EKF variance and crash

the drone flies in an uncontrolled direction, causing the crash
GPS:Glitch detected–>FS-EKF-INAV:Detected–>EKF variance–>Flight-Mode:RTL refused–>vibrarion compensation on–>Crash:Disarming.
HELP me find out problems in log please, Thank you friends.
Frame:QUAD Copter 4.0.0

Vibrations are very bad, you’ll need to fix whatever is causing that first.
Attitude control is extremely poor, possibly due to the crazy vibrations.

It seems like a new build with a lot of default values. Check this tuning guide:

And use this spreadsheet for some help:

download a copy of that spreadsheet and then you can plug in your prop size and LiPo battery cells.

thanks,YES I checked the log, and it seemed that the vibrations were bad after EKF variance, because drone were out of control.I wonder why EKF varance was detected,Could you help me analyze it.thank you for reply.

I see what you mean - looking in more detail everything goes bad, particularly vibrations, only after the GPS Glitch and then EKF errors.
I’ll keep looking…

But vibes shouldn’t be bad just because it’s out of control - vibes will be because of a physical issue.
Did the craft actually hit the ground or an obstacle and try to keep motors running for some time?

Please say what size props and frame you have, and also take off weight.

For any interested parties, vibes dont go bad till after “FlightMmode RTL refused”, around the same time as pitch and roll all diverge fro desired. I left vibes off this graph to reduce complexity.

The EKF warnings and the bad vibes might even be two separate issues.

frame type:QUAD
props size:1555
take off,4.1KG
motor :size:4108,380KV

You could use the MissionPlanner motor test to check one motor and prop at a time and see if any vibrate or behave strangely.
There could have even been a motor desync during decent causing all the vibrations. Most of the issues seem to be with Motor 2, it’s commanded to maximum a lot during the instability. Motor 1 also gets some maximum throttle but maybe only once M2 regains sync. Then Motor 2 loses sync (or thrust) again…

The fact remains, EKF thinks it should land, there’s no decent visible, in fact altitude climbs a little probably due to RTL interfering.
When there is any actual decent, Roll and Pitch goes bad and vibes go up.

Still no solution from me…

I will test it later, then give you a feedback,thank you bro. :handshake:

I want to know what the log analysis software is?

The first pic is APM Planner 2

I use a mix of APMPlanner2, MissionPlanner and https://plot.dron.ee/
Depends what you need:
MissionPlanner has some useful preset graphs
APMPlanner is great for changing scales and seeing a sequence of events
DroneePlotter is quick and easy, all the messages and parameters are easy to see

The map show GPS track and POS track are very different

The information displayed in the LOG curve is the same as that on the map.

From the information displayed by the attitude curve, the position should have moved a lot

So there may be an error in the EFK position calculation

It seems that the impact first occurred, and then the vibration value increased.

I had a very similar accident recently. I have a kakute-f7 flight controller, and while flying in position hold mode, I had an EKF variance msg and the drone started flying in a completely different direction on its own. Any ideas what could be the cause?


Hi, No logs in there

Pick a thread to post in so effort is not duplicated.

Sorry, I just uploaded the log now. I created the folder but forgot to upload the log :slight_smile:

Hi Dave, this post is unrelated to the other one. This is regarding another drone I have.

Another thing I forgot to mention is that I have seen the attitude flicking every once in a while through the GCS telemetry, I think this was also happening in the log. It quickly switches to nose down left roll and attitude and then back to level.

Just a quick look at that log, looks like you lost GPS (updates) before the end of the flight.

Check the 5vdc supply to the GPS unit.

I couldn’t identify the “attitude flicking” in that log, can you indicate a point in time?