IMU3000 noise characteristic

I have been trying to compute the noise characteristics of the imu3000 with some imu modeling tests.

I have noticed that my imu3000's outputs are significantly corrupted by additive sinusoidal components.
Here is the (rectified) correlation values of the stationary gyroscope outputs:
people.ucalgary.ca/~yyuksel/fig16.eps

The data for these figures were collected at approximately 1210Hz on a stationary platform (essentially it was part of the Allan variance tests). The internal BW was set to 256Hz (DLPF_CFG=0, FS_SEL=1, SMPLRT_DIV=0).

There was no other system near the sensors which may cause vibration during the tests. The ultimate source of the power was the usb port of the computer which I use to communicate with the mcu recording the data. (Furthermore, I don't think that the additive sinusoidal is due to the power supply, as the amplitudes are different on each axis. Besides I use Atmel's inertial board which also has an internal 2.5V regulator)

My question is that can you verify (confirm) that your gyroscope does not have any structural problem causing this? Can you provide me some 2-3 hours Allan variance data (the data itself not the Allan variance values) so that I can personally check it by myself? (I probably need more than 1 such data set, because as I far as I can see it has also an interesting repeatability characteristic.)