Non-linearity in roll angle readings

[Shrest]

Hello,

         To begin with Thanks to Jeff for the wonderful library. I could interface and get data from the sensor.However, there are few issues I can see when I did some checks.

1.         I have a GY521 breakout board, which I fixed on a cuboid made of metal for rigidity and I have measurements at different angles(deg.) - 0,5,10,15,etc.

        The readings I got are as follows:

0  -> 0.37

5  -> 4.98

10 ->9.57

15->14.31

30-> 28.10

45->42.17

 

These are averaged readings over a period of about 1 minute each. The readings are stable - there is no drift. But, as you can see, the error is increasing with angle. Im not sure where the correction is needed.Any ideas for the same are welcome.

 

2. Im logging data at 10 Hz on my laptop over TCP connection, I have given a delay of about 95000 us in the loop on the processor. And I see FIFO overflow sometimes when I observe for few minutes. 

 

I need advise on resolve these two issues. Kindly advise.

 

Thanks.