IMU mismatch and high throttle out when switching modes

Hi all,

I am having problems since some days:
When switching from Stabilize to PosHold or from PosHold to Auto there is

  • an IMU mismatch
  • and a sudden high throttle out for some seconds.
    I am flying Jonathan’s devbuild. It occurred first after installing the latest devbuild version but after switching back it is still there. Hence, I assume it is not related but some other software or hardware failure.
    I changed the system to mount the Pixhawk to the frame as well as EKF_ALT_NOISE to 2. Both should not have that impact.

Any help is greatly appreciated!
Thanks and best regards,

Wow, this is a weird one for sure.

Can you post a picture of your frame setup? I notice that generally your vibrations are very low, but then sometimes it is very high. The Z-axis looks odd like that, but the X and Y also do it. I wonder if you have a really soft mounting, but then sometimes something is contacting the Pixhawk, transfering high vibrations into it. It might be aliasing really badly on those sensors.

Hi Rob,

Thanks for looking into this.
Yes this is really weird.
And your observation is correct! The vibrations are sometimes higher and sometimes lower. (I hope) this is mostly due to different setups (frame, motors, Pixhawk mount). They are all similar but it is a continuous development.
I also have the feeling that sometimes something is transferring high vibrations. But I have not really figured it out even though I was looking specifically for anything that might be the cause.
I changed the damping a few times between the 3DR foam, moon gel, the small blue silicone ones from hobbyking and the rubber ones you can find at mikrocopter. I also tried different weights of the platform the Pixhawk is mounted on. If the weight is low the soft rubber ones seem to be best. They are all more or less comparable with respect to softness.
What is strange is that in most cases the vibrations are well below the recommendation but only if it is really low the copter is perfectly stable, i.e. not twitchy.
One reason I can think of why the vibrations are sometimes higher or lower is temperature. The first flight is often with low vibration. If the components are getting colder over time vibrations increase. I reckon it is the cables becoming more stiff. But not really sure.
Any comments and suggestions are welcome!

The other thing is that vibrations do not seems to be related to the strange behavior we see here. I made a short test with another Pixhawk and everything seems to be ok. So I assume it is an hardware issue. Although I have no idea how a hardware issue should be sensitive to changes in flight mode and why the IMUs are suddenly uncorrelated. I guess without EKF these are perfect conditions for flyaways.

Thanks again!

Interestingly the time period until the system calms down again looks pretty constant…

I made another test with the other pixhawk and the result is the same!

Three possibile causes:
i) some wrong parameter settings
ii ) the Pixhawk is not correctly mounted
iii) the baro is not correctly covered

I have mounted the Pixhawk without its case. I have done this before without any problems. Hence it was nothing special to me. But anyway it might be the cause. I either destroyed the boards (some parts are very close to the mounting holes) or I covered the baro in a wrong way (I use the original foam in the same place).

In the second flight with the other pixhawk all was fine in the beginning but got continuously worse over the flight. Vibrations are comparable to the log posted above and consistent over the fight. The strange thing is that once it settled after some seconds all is fine no matter what flight mode you are in. And, any of the possible causes doesn’t really explain that weird behavior. If it would be a hardware issue it should not settle. The only thing I can imagine is that the foam is pressed on the baro a little to tight. But this should not influence IMU values that way…
I have currently very little time for flight tests but hope to find some time end of next week. I’ll report.

Any ideas and suggestions are greatly appreciated!

Hi all,

I solved the issue.

The problem was that I placed the buzzer pretty close (1.5cm) to the autopilot. And when switching the flight mode we now have a tone and thus a magnetic field influencing the accelerometer.

After relocating the buzzer all is fine now.