until about 15:00 UTC, noisy spectral lines at abour 30 Hz, 60 Hz and 100 Hz
kept the the BNS range at a low value. Those lines, and others, are visible
in LSC_MICH and LSC_PRCL. They disappeared progressively like mechanical lines damped with
a high Q factor.
Some of those lines are visible in the spectrogram of h(t) but not in the
spectrogram of LSC_DARM.
In addition, LSC_PRCL shows 3 moving lines at about 80 Hz.
Plot 1 and 2: spectrograms of h(t)
Plot 3 and 4: spectrograms of LSC_DARM
Plot 5 and 6: spectrograms of LSC_MICH
Plot 7 and 8: spectrograms of LSC_PRCL
in the 0-60 Hz and 50-110 Hz frequency bands.
This is the spurious coupling between B4_56MHz_InPhase (frequency noise), and the quadrature (MICH error signal), which is present when IMC length is not well tuned on the modulation frequency (FmodErr too large). A loop is active since the beginning of ER11b, in order to control this error during the lock. A pre-tuning of IMC length is performed before the start of each lock acquisition, but unfortunately it is working with a poor accuracy, so it happens many times that MICH is huge at the beginning of the lock, and then slowly goes in the good state, if the duration of the lock is long enough.
An improvement of FmodErr pre-tuning is mandatory.