Pixracer second accelerometer vibrations can cause altitude rise?

very interesting log part


Vibration on the second accelerometer is much stronger then on second.
Can this vibration cause uncommanded altitude drift?
Where i can see which accelerometer commanded “rise up”?
set accel filtering to 10hz. This settings apply to all accelerometers?

2017-06-04 20-03-15.bin (2.3 MB)
also this setting
“INS_ACC_BODYFIX”


Set it to less noiser accelerometer can help?

add dampers
before


after

vibrations are smaller, but still clipping.
why?
vibes ±1.5 on noiser accel

but why clipping rises?
2017-06-05 18-36-01.bin (2.2 MB)
with dapmers.bin (2.4 MB)

also


PI show nothing

I had this pattern when it was mounted with self sticking foam and since I installed it on a damping platform its flying smoothly.

1 Like

i agree, FC need to be soft mountd. I use O-rings, but vibrations still not good. Try soft mount motors - still not good.


last flight
2017-06-08 20-23-20.bin (795.2 KB)
i not have altitude rise in this flight, but strange
PI - Index showing which instance of EKF2 has been selected for flight control
But my logs have blank graph

Alexey,

Thanks for testing and reporting back.

The vibe levels in your last log look totally fine. All below 20m/s/s which is within tolerances (less than 30m/s/s is “acceptable”, 30~60m/s/s is the “grey zone”, above 60m/s/s is definitely bad). You shouldn’t see altitude hold problems at this level.

There are just 3 clipping events which is also nothing to worry about. Maybe it got a momentary whack as it hit the ground or something but in any case, 3 is nothing to worry about. 300, that’s trouble but 3… no problem I think.

1 Like

thanks!
There is nothing suspicious that the graph PI is empty?

I must admit I rarely look at the EKF/NKF messages. I only dig into them if something else looks odd in the main vehicle logging. We’ve done some updating of the EKF description and I think @priseborough plans to update the EKF wiki pages more. … perhaps we can add some descriptions of those log messages.