Accelerometer calibration changed in the middle of a run

[gplracerx]

Between 18.75 and 18.95 seconds, the accelerometer calibration changed with the lateral and lineal axes rotating about 50 degrees counterclockwise. Looking at the data in the .csv file:

time lateral lineal
18.75 -0.78 -0.51
18.95 -0.09 -1.03

The acceleration source flag didn't change. It looks like the app decided that the orientation had changed in the middle of the run. Actually all three angles changed, but the roll and pitch change was small compared to the yaw change. Calculating the true orientation by comparing integrated yaw rate from lateral acceleration times velocity with GPS heading and integrated lineal acceleration with GPS velocity and calculating the coordinates necessary to minimize the difference of the sum of the squares, the initial calibration offsets were 0.5 degrees roll, -3 degrees pitch and -8 degrees yaw. After the shift it was 3.8 degrees roll, 2.9 degrees pitch and -60 degrees yaw. The iPhone 4 was solidly mounted in a RAM suction cup windshield mount and its physical orientation didn't change.

The calibration didn't change after that in later runs.

[Harry]

The yaw rotation is compensated during the last calibration step. This is done when LapTimer detects a rapid deceleration while driving straight (usually at the end of a longer straight when braking into the next corner). Sounds like in your case, the scenario has not been recognized correctly and LapTimer calculated the wrong yaw compensation.

- Harry

[gplracerx]

I renew my request for the ability to turn Laptimer accelerometer axis calibration off entirely. It looks like there's already a switch that could do that in Expert Settings. I can usually eyeball yaw orientation of the iPhone better than Laptimer can calibrate it. And I can do as well or better for roll and pitch by tweaking the position to zero out lateral and lineal acceleration while the car is sitting level.