MPU6000 Calibration and Sampling

Hello all,

While working with the MPU6000 we have encountered a few questions:

1. Regarding the DLPF. What is the block diagram of the sampling process? We could not understand it from the datasheet. We understood:
[No analog pre-filtering]-->
[high rate sampling (what is the sampling rate?)] -->
[filtering using the DLPF parameter defined by the user]-->
[down sampling (taking every n-th sample) to the user-defined sampling rate using the SAMPLERATEDIV] ?

2. Regarding Calibration. We are trying to bound the total error of the gyro and accelerometer.
2.1 Regarding the gyroscope, the parameters relevant to the Sensititvity error are: over temperature, tolerance and cross-axis. For bias: initial value, and variation over temperature. With respect to what were those parameters calculated? Did you assume Normal distribution and taking 2 variances away? How can we use the cross-axis parameter? Finaly, How can we bound the total error assuming full scale temperature variation and worst-case criterion?
2.2 Same question as 2.1 regarding the accelerometer.
2.3 Does calibration improve the total error? How? Which parameter will zero-out?

Thanks,
John