Although the spectrogram of the drive-off mich segment shows a non-stationary noise in low frequency regime, we are wondering if this is indeed due to scatter light. Refering to the scatter light analysis noted in Elog 1122, I investigated roughly the coherence between the non-stationary noise and the z-motions of the blades. Using the standard equation f = 2/lambda*abs(dz/dt), we can predict for the corner frequency of scatter light noise caused by the moving optical elements. I overlapped the 200 seccond spectrogram (ww = Nfft = fs/2) taken from July 31 data starting from GPS time 1185610403 with a corner frequency prediction using the inline z1+z2 data.
We see that the non-stationary noise is barely correlated to the speed of light-scattering elements, so the low-frequency noise plateau is hardly caused by scatter light...
I used a shorter windowing (ww=np=fs/4) spectrogram (so we can see more instant variance) to choose a relatively quiet time (tvec = 114.3) vs. a noisy time (tt = 85.7) and compare the spectra. Difference can be observed in frequency regime 10 ~ 50 Hz:
 
| Quote: |
|
In order to study if the noise plateau is caused by scatter light, I first compared mich spectrum and spectrogram of half-hour data from recent quiet lock and the lock back in May:
The sensitivity is actually slightly improved in low-frequency range (<30 Hz). The noisier behavior at high frequency indicates a possible further improvement in michelson symmetry or alignment. The spectrogram shows that the noise below 50 Hz is non-stationary, which means that the noise bump we see at low frequency is likely caused by light scatter.
| Quote: |
|
To understand the noise in frequency range 15 ~ 100 Hz and therefore to be able to improve the sensitivity in that range, we would like to check if the major contribution to this noise plateau is: 1. Intensity noise, 2. Frequency noise, 3. Scatter light, 4. Calibration.
|
|
|