New bench layout

Today I changed the layout of the bench. I installed the 50% beam splitter. To have the correct polarization on the BS, I placed a polarizer after the AOM. I also used a single lens, before the BS, to focus the beam on the PDs.
The figure below shows the new setup.



I also performed some closed loop measurements in the low frequency band (1Hz to 101Hz). I managed to put around 700mW on the In-Loop PD and only 470mW on the Out-of-Loop PD. I cannot say why the power (measured with the power meter) is so different between the two PDs. The voltage out of the PDs is almost the same (25mW difference). The responsivity of PD12599 is 0.050V/W*100 (ampli gain), while that of PD12595 is 0.056V/W*100. The difference between the two powers should be only 40mW, considering that the voltage out of the PDs is of the order of 1.9V. Anyway, I'm reporting the comparison of the noise spectra as measured by the out-of-loop PD when the loop is open and closed. The filter parameters are: AC-1s, G=300, LF=30Hz, HF=100Hz.



I have a small comment: 1Hz lines are clearly visible in the open-loop spectrum. During the measurement, I kept the PAR off. We have to look somewhere else for the culprit...
The picture below, shows the ratio between the open-loop and closed-loop spectra, that is the suppression factor.

The reason why the reflected and transmitted powers were not the same, is that the beam was not impinging at 45°. I rotated the BS, so now the powers are 586mW and 570mW for the reflected and transmitted beams respectively (much better).

I also checked what voltage we should have with these powers considering the PD responsivities reported in the VIGO data sheets.

PD12595 (transmission): 0.57W*0.056V/W*100 (G_ampli)= 3.2V (measured 2.0V)

PD12599 (reflection): 0.586W*0.050V/W*100 (G_ampli)= 2.93V (measured 1.75V)

For both PDs, the measured voltage is lower than expected, this could be either due to saturation (power is not so high and spot size does not seem to be so small to justify this) or to the fact that the responsivities are actually lower than those reported in the data sheets. The second idea would be in agreement with the measurement done by Maurizio a long ago (see Calibration of PDs even if there is NO comment to the measurements...).