The limitation above 10 Hz in oplev spectrum was found to be the electric noise (which is the dark noise). We took the data of each QPD segments without oplev laser (dark noise), and square-sum the four segment data (=sqrt(SEG1^2+SEG2^2+SEG3^2+SEG4^2)) so to simulate the dark noise level in PIT/YAW spectra, in which the four segments are summed and subtracted. ADC noise was estimated as well and it was found not limiting the noise. See the attached plot.

In the plot, the curve labeled high power is the oplev PIT spectrum without any power normalization or calibration when the full power (SUM=19000). Same for ''low power'' when SUM=1700. 10Hz peak is a calibration peak. You can see the signal to noise ratio degrades when low power because the signal goes smaller whereas the dark noise level does't change.