In the test of SR misalignment on November 1st the main effect of SR angle on ITF sensitivity was due to optical gain change. The same effect was much less visible during similar tests done before reducing the input laser power from 25W to 12W.
The attached plots summarise the relation of Hrec noise around 120 Hz with optical gain and double cavity pole frequency. The plots concern the period from 7/10 to 1/11, with data at 25 W input power in blue, and data with 12 W input power in red.
The legend was wrong on two of the plots (thanks Michal for pointing this out). In attachment the same plots with correct legend
We might be mislead by issues with calibration that were fixed during that same time period. Using V1:Hrec_ORfpole_meancavity for the pole cavity and V1:Hrec_ORgain_V1:Sc_NE_MIR_Z_CORR_62.5 for the optical gain at 62Hz and plotting all data since October 1 I get figure 1 & 2 (figure 2 has the x axis in log scale instead of linear scale). The point to the right of an optical gain of ~2.2e9 W/m are at 25W input power and to the left at 12W of input power. In both cases when there was some experiments that reduced the SR+arm pole cavity frequency the optical gain increased. The slope is not exactly the same, but it is similar. To compare this more precisely one would need to select only the data when the experiments of SR misalignment were performed.
The attached plots are now limited to the experiments with SR misalignment on 09/10 (25W input power) and on 01/11 (12W input power), using V1:Hrec_ORgain_V1:Sc_NE_MIR_Z_CORR_62.5 for the optical gain and V1:Hrec_ORfpole_meancavity for the DCP frequency.
The effect of cavity pole on in-band optical gain is actually similar. The effect on Hrec noise in the bucket is much more evident in the data with 12W input power, likely due to issues with Hrec in the data of 09/10.
For the misalignment test of Oct 9 there was a recalibration done after fixing the issues in Hrec. The corresponding data are in /data/dev/hrec/hrectest/hrec_srhinj9-1380880808-3608.gwf
Figure 1 shows the recalibrated spectra during 1h of that test. An aligned SR configuration correspond to the last two curves when there is ~700mW on B1p. There was during that actually a clear improvement at 100Hz in sensitivity, and noise getting higher at 1kHz due to the lower bandwidth of SR+arm cavity pole.
Figure 2 is the same data zoomed around 100-145Hz showing that the calibration line at 106Hz is constant, and the 137Hz calibration lines moving by less than 5% in amplitude.
Figure 3 zooming vertically further in, the noise floor level changes by a factor 1.45, a bit less than during the Nov 1 test at 12W input power when it changed by a factor 1.65, for a factor 2 decrease in higher order modes (as a measure of the SR misalignment). But in general it is very similar.
Figure 4 are the corresponding spectra of B1, as for the 12W test the noise floor is not moving in level, and what is changing is the height of the calibration lines (the interferometer optical gain).
