Uncontrolled Ascent -disparity between CTUN-Alt and Baro-Alt

4x Quad
Pixhawk autopilot
ArduCopter V.3.3-dev

Running an Auto mission in which the quad would Takeoff to an altitude of 15m and loiter unlimited. This mission had been tested multiple time with no issues.

Looking in the Data Flash Logs
the Baro-Alt shows data that looks correct. The data shows a 0m Alt until put into Auto Mode and the quad takes off and altitude raises to about 25m and then descends (rapidly).
The Baro-Press shows data that supports the Altitude numbers.

The GPS-Alt shows the same altitude track just smoothed out. This looks good.

The IMU-AccZ shows an correct initial measurement of 9.8, and then drops to -20 as the quad takes off. The AccZ is around 0 as it free falls to the ground. (the crash is another story that doesn’t need to be solved here.)

Here is my problem:
The CTUN-Alt shows an initial measurement of 0m after arming and slowly falls even as the quad takes off and climbs. The CTUN-Alt falls to -147m (that’s negative!!!)

How is the CTUN-Alt calculated? Why would the Baro-Alt, GPS-Alt, and Baro-Press show good reading, but the CTUN-Alt be so off? There is an ERROR 18-2 (Barro Glitch).

The CTUN-CRt (climb rate) was logging negative numbers between -20m/s - -500m/s.
How is the CTUN-CRt calculated?

Why would the Error 18-2 cause the quad to takeoff at max ascent rate (the motors were drawing 50A when they normally draw about 20A on takeoff), when all the raw altitude reading were correct???

Any help would be appreciated. I have attached some screen shots of some of the graphs, and the full log file.

Thanks,
Rich

CTUN.Alt is the inertial navigation system’s altitude estimate. It doesn’t look like it is being dragged towards the baro alt as it is supposed to.
Take a look at EKF1.PD. It has a good estimate.

There are vibration issues in your log (take a look at the difference between IMU.AccZ and IMU2.AccZ). I would try to resolve those, but they don’t seem to be the cause.

@Richard_M I am facing similar problem like you. Did you find any solution?

link to the similar topic we posted