This is a follow up on measurements done in 51546  and modeling 50610  A few weeks ago I did a longer set of squeezing measurements using the full range of the CLF demod phase shifter and the sign flip in the CLF servo board.  I repeated the measurement for the nominal 100 count SRCL offset which we use to minimize the optical spring, and an offset of -100, which corresponds to roughly a 100 to 120 nm detunign of the SRC (extrapolating from 51592).  The first attachment shows these measurements with the correlated noise subtracted, you can see that we are reaching nearly .

The second and 3rd attachments shows the same data, I've taken the median in several frequency bands (avoiding lines) for each demod phase and used an MCMC fit for a model that includes loss, nonlinear gain, phase noise, and the ratio of the OMC 3MHz semi-major and semi-minor axsis.  I used Daniel's expression here to model the realtionship between squeezing angle and demod phase.  The last attachment is a collection of the corner plots for all of these fits.

The next two plots show the fit parmeters plotted as a function of frequency for the nominal SRCL offset  and with SRC detuning.  In these plots the blue stars show the median of the posterior dsitribution, while the triangles show the 16th and 84th percentile of the parmeter.  I've divided out the budgeted effecincy based on the known losses so that we should be able to compare these plots to similar plots produced by Lee's code.  I've plotted the normalized nonlinear interaction strength x, which is not frqeuency dependent, because there is some frequency dependence in the fit results, because of degeneracy with phase noise. 

The code used to make these plots is available here, inlcuding a script which will run the measurement by slowly varying the CLF demod phase and SRCL offset, and sqz_as_phase_rotates,py, the main script which produces the plots.