In 82097 we got some constraint on the circulating power in the arm from the squeezer data set.  I'm working on looking at the data that Camilla took in 83660, and realize that with the higher nonlinear gain the frequency independent squeezing gives us a better constraint on the arm power.

Using known injection losses of 8.2%(google sheet),   and  6% extra HAM7 losses from 83070 gives 13.7% minimum injection losses for squeezing, and our known readout losses are 8.4%.  Based on 16.1dB of anti squeezing (for NLG of 19 as reported by Camilla) and -4.97 dB squeezing, the infered NLG is 18.7 and the total sqz efficiency is 0.694 (with 0 phase noise). This means that in addition to the 6% extra losses measured on the homodyne, we have another 12% unknown squeezer losses.  We can assign those 12.2% losses to either injection losses (unlikely since there isn't much between homodyne pick off point and injection into the OFI), readout losses that also impact the IFO readout, or squeezer readout losses that don't impact the IFO (SQZ to OMC mode matching). 

If the only IFO readout losses are the known losses, the IFO readout efficiency is 91.6%, and we can use the shot noise level without squeezing injected to estimate that the minimum arm power is 310kW.  Using the total squeezer efficiency and the known losses, we can place bounds on the readout efficiency between 80.4% and 91.6%.  The level of shot noise measured without squeezing mostly depends on arm power and readout efficiency, so this maximum readout efficiency gives us a minimum arm power of 310kW (all squeezer losses are injection losses, readout efficiency is 91.6% and arm power is minimum), while the minimum readout efficiency gives us 353kW (all unknown squeezer losses are common to IFO, which probably wouldn't be the case for mode mismatches).