First thing we did in the morning was to switch from PD1 to PD2 (which has a notch at 6MHz). It worked straight away without the need of big modifications. Once this change was done, we have spent some time fine tuning the gain of the SSFS during the lock acquisition and in Low Noise 3 to be as high as possible (~10 kHz in LN3).
Then we have tried to characterize a bit the SSFS loop by making several noise injections at different frequency regions. We had particular troubles to inject noise at high frequencies, this needs to be better understood by making more injections over 1 kHz both in recombined and in DF. Here are the GPS times (we took at least 2min per injection):
- @1kHz - 9.21 UTC
- @200-2000 Hz - 9.33 UTC
- @60-500 Hz - 9.48 UTC
- @10-200 Hz - 9.58 UTC
Once in LN3 we took some data at different gains of the SSFS in order to spot the frequency noise coupling to the sensitivity:
- B4_GAIN = -590 -> 1210510542
- B4_GAIN = -400 -> 1210510911
- B4_GAIN = -650 -> 1210511220
- B4_GAIN = -700 -> 1210511410
Then we increased the modulation depth of the 6 MHz in steps until we unlocked the ITF (@ 9dB). It seemed that the offender this time was the 12 MHz but an offline analysis will provide more details on when does the PD2 saturates and why. This activity starts at 13.14 UTC (1210511660) and the steps can be seen at the channel DAQ_LNFS_6MHz_mag. We start from -15 dB (17.5e-3V) and we unlock at 9 dB (0.28V). The steps are each 3 dB.
Finally the last action was to engage the 3rd order boost on the SSFS instead od the second order one we had at present. The engagement of the filter itself was succesfull but we encountered some unexpected digital noise. We removed the reset of the digital noise correction to check whether this level of noise remains constant or diverges.
At the end of the shift we realized that the phase of the B4 error signal had changed from the beginning of the shift by 0.3 rad, so a further retuning of the phases has been done. As in the past we have observed that the phase of B4 in recombined is not the same as in LN3, we have recovered the code in metatron that adds this delta to the demodulation phase, 0.4 rad.
The detailed analysis of these activities will be posted soon.