Took another data set of FIS with different SRCL offsets, to try to set the SRCL detuning for calibration measurement, similar to 79903.
Here are some plots of this code, made by borrowing heavily from Vicky's repo here and from the Noise budget repo. I will put this into a repo soon, perhaps here.
The first plot shows the spectra with different SRCL offsets in, always with the squeezing angle optimized for kHz squeezing. The no squeezing model isn't well verified, I've used SRCL detuning of 0 which we know isn't correct. We use this no squeezing model to subtract from the no squeezing measurement to estimate the non-quatum noise, shown in gray here. The SRC detuning doesn't change this estimate much without squeezing injected.
The next plot is a re-creation of Vicky's brontosaurus plot, as in 79951. The non-quantum noise estimate is subtracted from each of the FIS curves, which are then plotted in dB relative to the no squeezing model. Each of those shows a squeezing data set with a model, where I by hand adjusted the SRCL offset in the model based on this plot. The subtraction is needed to make the impact of the SRCL offset clear.
The final plot shows the linear fit of the SRC detuning to SRCL offset, which gives us the SRCL offset we should use to go toward 0 detuning, (-191 counts).