[Yuhang, Aritomi, Eleonora, some remote suggestions from Matteo B.]
We succeded in feeding back a part of the filter cavity PZT correction to the end mirror, so that we could reduce the PLL phase noise.
What we have done after the preliminary results (entry #1506)
1) We amplified PZTmon with a standford. Gain 50. (so it is half of the real PZT correction). This also removed the 50 Hz oscillation we saw when the signal was sent directly to DGS.
2) We measured again the TF between Length excitation and PZT mon (pIc1, top right). The blue curve is seen by PZT corr, the red one by END length OPLEV.
Driving matrix for length:
| coil 1 | -1 |
| coil 2 | 1.2 |
| coil 3 | -1 |
| coil 4 | -1 |
3) We measured the PZT correction when the cavity is locked witouth any feedback on the test mass (pic 2). The spectrum is calibrated taking into account the ADC gain (6e-4 V/cout), the piezo gain (2e6 V/Hz), the gain of stanford (50) and attenuation of pzt mon (1/100). Maybe there is a factor 2 missing from SHG. Anyway, it shows a good agreement with the expected free running laser noise (1e4/f Hz/sqrt(Hz)). Except for the region from 0.1 Hz to 5 Hz where the cavity seems to move more than the laser.
4) We tested two different filters for the test mass feedback. One with a derivator witch only damps the length resonance in the ~1Hz region and one with also a low frequency pole (0.01Hz) .The performances of the different filters are shown in pic3 (red line: no feedback, bue line: damp, green: damp+DC, purple: damp+DC (with double gain)). The time signal in the first three cases is shown in the following video: https://drive.google.com/file/d/1yqvl5w8y_eeEE88MZ77aKqHNJG7Un1Ap/view?usp=sharing
5) Yuhang measured the PLL CC2 phase noise in these configurations. The results are shown in pic 4. I'm not sure about the RMS but it seems that, as expected, the noise is lower when we engaged the feedback on the mirrors. The loop can be optimized to damp also the peak at about 3.5 Hz.
6) We also checked the low frequency phase noise of PLL CC2 in the 'damp' and the 'DCdamp' cases (pic4). We wanted to see weather the main laser noise is increased in the 'DCdamp' case, due to the fact that at low frequency the laser is maybe less stable than the cavity. However it seems that the noise is the same in the two cases. Probabily the correction to the laser from the rampeauto is anyway much stronger.
We will check in the future if one of the two configurations is better in terms of squeezing performances.
