Observed I2C inconsistencies & 5 questions

I have tried both the IMU3000 and the MPU6050 EVB with the ARM board. Occasionally the 3000 and 6050's respective PC demos will partially work and I can see the "oscilloscope" data so I know the chips are alive. Most of the time though the PC demo's just crash, though.

If I capture the initialization I2C commands sent from the ARM board to either EVB, and then reproduce them in a custom board with I2C, the IMU3000 and the MPU6050 both "ACK" the commands MOST of the time. I have verified the SDA/SCL lines and the transitions are very clean at both 100kHz and 400kHz I2C clock rates. Even if I repeat a single I2C sequence in a loop such as reading the WHO_AM_I register (reg0) the IMU3000 and MPU6050 both seem to occasionally "NACK" and not reply with the register value.

1) Is inconsistent I2C interface a known problem with these chips?
2) Will this be fixed in the next batch of MPU6050 coming in at the end of June?

I assume the I2C accessible registers in the MPU6050 are similar to the IMU3000 because the ARM board which is supposed to communicate with both EVB's sends the same I2C commands independent of which board is connected. So I have been referencing "https://invensense.tdk.com/mems/gyro/documents/ps-imu-3000a-00-01.1.pdf" as a description of the MPU's I2C registers.

3) Is this a fair assumption?
4) When will the I2C accessible registers be documented for the MPU6050?
5) Can the source code for the ARM board be made public?

Thanks.

Sam Waldbaum
Sam@TeamWaldbaum.com