Wed  7 Mar 2012 14:46:47 EST

A data log of a car drive with APM2 fixed to the dashboard running 3 solutions at once
The samples were collected at 50hz, micros() timestamps each line

The '.plot' file is for OSX "Plot" a brilliant and fast (free) Plot utility.
http://plot.micw.eu 
Just open the file.

DCM and Madgwick are running at 100hz, so two update cycles per printed line
MPU6000 is running at chip speed (200hz)
GPS is running at 10hz
Barometer is running at 100hz
Compass is running at 10hz

Trip6.csv -- Calibrate at a slight downward tilted location, drive & park at a flat location
Trip7.csv -- Calibrate at flat location, drive & park at a downard tilted location

NOTES:

The accels are adjusted for GPS acceleration and GPS Speed X Gyro rates, the amount of adjustment is in the 3 fields, to recover the original accels you must subtract them. Adjustments for linear acceleration should work to reduce the X axis acceleration to zero when there is no actual attitude change. Since this is a car trip, the true pitch attitude of the car mostly matched the roads (which are hilly). The roll attitude of the car should be even less extreme except during faster turns where there is some suspension roll.

The MPU6000 is not using the adjusted accels, its attitude solution is therefore more impacted by acceleration.

The GPS acceleration field is calculated from a smoothed GPS 2d speed, and is significantly lagged, when the car moves from a stop, the acceleration shows up in the sensing first, then the GPS speed starts to show and finally there is an adjusting acceleration in the opposite direction showing up in "adj_x".

The compass was calibrated in-situ and gives a reasonably match the GPS Course
