The change of control of the IMC rampeauto (from 1/f to 1/f4) has improved strongly the locking performance of the cavities (even if they are not optimal yet, but it is better to check it when the TCS works are completed).

- purple : 7th of January with the control 1/f4

- blue : the 1/f control

- ligth brown: now (the West is not locked due to tcs activities)

A degradation of the lock has been observed in the last days. the power in the arms, locked on the IR, are fluctuating a lot. The corrections of the controls are saturated.

analysing the trends a matching can be found with the worsening of the RFC lock due to the change of filter (1/f instead of 1/f4).

Figure 1 shows the trends from datadisplay in which is visible that the spoiling of the cavity is more or less compatible with the worsening of the rfc accuracy

Figure 2 shows the matlab analysis:

in this analysis only the locked periods (for both arms are taken) averaging on 100sec. The first two plots shows the maximum power and the variation of the power for the two cavities. A worsening is visible in the recent days. The third plot shows the accuracy of the RFC lock, worsened after the change of filter and the last one shows the Sc_NI/WI_MIR_Z_CORR variations and it is visible that they starts to saturate when the RFC filter is changed.

this situation is visible also in Figure 3 when the two RFC control spectra are compared, showing also the cavity signals.

The control should turned back to the 1/f4 control.

From a more accurate evaluation of the etalon fringe periodicity showed that one fringe corresponds to 0.26K (instead of the previously stated 0.3K). The Finesse asymmetry accuracy are then lowered from the 0.05% in the Best case scenario to 0.06% and from 0.15%, in the worst case scenario, to 0.18%.

In order to quantitatively evaluate how much Finesse asymmetry can be reached given the control accuracy of Etalon obtained in O3 (total RMS of 6mK), the simulated Finesse behaviour vs Etalong, for the two cavities, has been simulated with Finesse (this study has been done in the framework of the discussion of having or not the Etalon for Adv+ phase II).

The optimal working point corresponds to the Best case scenario where the two mirror temperature is chosen to match the maximum of the West arm cavity finesse and the corresponding value for the North, to minimize the asymmetry (see Figure 1).
This should correspond to the best sensitivity wrt etalon, for this reason the Etalon working point chosen in O3 was the one that maximized the BNS range.

In that configuration the finesse asymmetry obtained, considering the control accuracy of 6mK, is of ~0.05%.

In the worst case scenario, Figure 2, where the slope of the Finesse curves is maximum the asymmetry achievable is about 0.15%.

This computation is done in the assumption that the wp has been chosen correctly (the one that minimizes the finesse asymmetry).

A study was conducted to determine if there is a need to wait during the lock acquisition (especially at Low Noise 1)  for the sideband power on B4 to build above a threshold prior to continuing the lock acquisition. The study shows that such a threshold does not exist and it is unnecessary to wait in low noise 1 during the lock acquisition.

The data set was built by searching for the point when the interferometer entered Acquire Low Noise 1 (METATRON state 122) during O3 (starting April 01). At each occurance the successive 80 mins were analyzed. The data was divided into three categories depending on whether the interferometer remained locked and reached low noise 3 squeezing at the end of the 80 mins (thus considering locks of minimum 80mins).  The lock category contains the points which reached low noise 3 squeezing without unlocking in the process. The unlock category contains points in which the interferometer unlocked (METATRON state dropped below 122). The third category contains points in which the Interferometer did not unlock but did not reach low noise 3 squeezing. Points in the third category were discarded as there are many reasons in which the ITF did not reach low noise 3 squeezing including manual holds.

The power of the sidebands were recorded at the initial point of acquire low noise 1, state 122, and at the point of low noise 1, state 123 (if reached in the case of the unlock category) as measured by LSC_B4_112MHz_MAG_mean.

Figures 1 and 2 show the initial power and the power at low noise 1 vs. time for the lock and unlock categories. There is no clear difference in the two data sets with many points of the lock and unlock categories nearly overlapping. It is particularly interesting when lock and unlock points overlap when plotted with time as the condition of the ITF would likely be similar in such a close time separation. Here it seems that the sideband power is not a good indication as to whether or not the ITF locks. The low power points of Figure 2 are attributed to when the ITF unlocks soon after or during low noise 1, particularly if the power drops before the ITF knows it has unlocked.

Figures 3 and 4 show a histogram of the initial power and the power at low noise 1. The histograms are very similar with the mean sideband power of the unlock category 5.6% lower and 5.4 % lower than the lock category for the initial and low noise 1 powers, respectively. The difference in mean values is not large enough to determine a threshold value of sideband power which will guarantee a lock with high confidence.

In conclusion. the sideband power is not a good indicator for a  successful lock. It is not recommended to wait in low noise 1 in order to build up power of the sidebands during the low acquisition.

We have been trying to calibrate the error signal, in order to know how precise is the lock. In order to do this, we calculated the velocity of each reasonance peak in m/s and we compare it with the slope of the PDH linear region at the same resonance.

In order to decide whether this velocity calculation was correct or not, we compared two different methods: one calculating the Doppler shift observed on the ringing (as described in entry 33856) and another one using the fact that between each resonance peak we have an FSR (as described in entry 33861). Figure 1 shows that they are in good agreement except for the low velocities. Notice that with the method of the ringing we can only fit velocities higher than th critical one (0.4 um/s) so we can not obtain any information of the low speeds. We can see that for the FSR method we have a disagreement at low speeds, but this is not surprising, because the velocity is not constant and the mirrors get some acceleration due to seismic noise, so the lower the speed, the higher the uncertainty.

So, we have used the Doppler shift method in order to obtian the value of the velocity on m/s, and we have repeated this process for all the peaks found on 20 data sets of 90s each. Figure 2 shows the relation between the speed in m/s and the derivative of the PDH on the linear region. From the fit we find the slope, which gives us the calibration factor. FIgure 3 shows an histogram of the calibration factor calculated for each dataset weighted by the quality of the linear fit. The calibration factor obtained is 3.6e8 - 0.1e8. The figure also shows the hsitograms of the lower and upper bound 95% confidence limits. For this calibration factor, we find that the acuracy of the cavity lock is around 140 pm.