Altitude drop when changing from Stabilize to Loiter when flying master

[Thorsten]

Hi all,

I wanted to upgrade a copter from 3.2.1 to 3.3rc5. It went fine, I made an autotune and the flight characteristics were perfect. The only thing I realized were small jumps in altitude (about 30cm) when changing flight modes (a logs at: PostAutotuneAdjusted.zip). 

Today, I compiled master (latest commit: 814edfa457e5490ea892723a8fa644e1a39cfe25) and ran all calibrations again. 

Take off in Stabilize was ok. There was quite some wind so it was drifting a bit. When switching to Loiter the copter lost altitude. The same happened when switching from Loiter to PosHold. This time it almost crashed.

I had a look at the logs. You can see that the Baro and the GPS Alt are ok. But the estimated altitude is not following them:

The other thing I found is that there are inconsistencies in the AccZ values when switching the modes which level out quickly. 

The red line is IMU 1. 

AccZ vibration of IMU 2 is a bit too high for this copter. I have to check that.

This is a very reliable copter (4S, 15'', hexa) with more than 100 mapping missions all over the world. And apart from updating the firmware and replacing the GPS mast no changes have been made.

Interestingly, I had a similar issue some months ago http://ardupilot.com/forum/viewtopic.php?f=80&t=10847 when flying master but with the opposite effect (high throttle out). Back then it was a Pixhawk and on this copter it is DroPix. I thought that the issue with the Pixhawk was that the baro could not breath because I mounted it without the housing and foam under the baro might have been to too stiff. But this seems to be only part of the story since this DroPix copter was flying perfect since installing master.

A difference between the two logs linked here is that today I was flying with a payload of 400g (stabilized gimbal and camera). This might explain the higher vibrations.

Any idea what might cause this IMU mismatch?

Thanks and best regards,

Thorsten

PS: I posted this "user help request" here because I was flying the lastest master and it seems somehow related. Hence I hope it is of interest for you...

[Thorsten]

Hi all,

I solved it.

I first tried another vibration damping system, lowered STAB P a little and rearranged some parts. Switching flight modes now resulted in a small jump - which I thought is related to a little too high throttle. But it wasn't. 

I found the cause due to an uncexpected battery failsave (my setting of the capacity was much too low). The buzzer started. Interesting log! Vibrations followed the beeps - as did the mag. Then it was pretty clear.

The problem is that I placed the buzzer pretty close (1.5cm) to the autopilot. And when you switch the flight mode we now have a tone (which btw. from a user's perspective is one of the best features added recently) and a magnetic field also influencing the accelerometer.

After relocating the buzzer all is fine now.

Summary: It is Jonathan's fault!!! ;-)

Joking aside, be carefull where you place your buzzer...

Cheers,

Thorsten