I've found another defect in the P2 firmware that causes problems in the H3-3D gimbal. If your gimbal has problems with a drifting horizon, this issue probably effects you.


Some important things to note:

- The Phantom 2 supplies inertial movement data to the H3-3D to help it stay level.
- The Phantom 2 had a change to fix declination offset in firmware 3.06 and up.
- The 3.06 firmware adjusts compass data for declination offset and remembers it between flights.
- The H3-3D will exhibit loss of level horizon when the compass data is poor. Videos of TBE show significant loss of horizon in the H3-3D.
- The H3-3D works normally if used in the same location. Note: other compass / H3-3D errors can cause horizon drift as well.

- The H3-3D will exhibit horizon drift if you take it between areas of different magnetic declination - e.g. from LA to Chicago to NYC.

Conclusion:

- Compass adjustments stored in memory by the Phantom 2 are causing horizon drift when the Phantom is used in different areas.
  
This has been tested and confirmed on two different Phantom 2 / H3-3D systems. Both exhibit exactly the same behavior. Used in one location, the horizon is consistently level. Once used in an area with a different magnetic declination, the H3-3D will exhibit issues with horzion drift.  


I will experiment with reseting the FC on the Phantom 2 to see if that removes the memory effect on the gimbal.

I've always thought that was a hardware issue.  The shaft that the camera is mounted on is only held on by friction and sometimes little bumps can get the shaft and sensor out of alignment.  There are a number of different videos on YouTube and I would suggest viewing a couple of different ones for ideas, this is just the first one I came across http://youtu.be/5iTz3Dd2Ecs