Evan, Matt, Lisa

While IFO realignment work was in progress yesterday, we spent sometime looking into the H1 quantum noise.

There are a couple of known issues when trying to explain the measured H1 shot noise level:

1) the H1 measured cavity pole (~340 Hz, see  the calibration companion paper ) is actually lower than expected from the nominal IFO parameters, so you can't get the right shape with the current GWINC parameter model (nominal parameters give a higher pole frequency);

2) beside that, higher losses and/or lower power are needed to explain the actual measured shot noise level.

Evan calculated the H1 shot noise contribution in the H1 noise budget (fig 2 in the detector companion paper) from the shot noise on the DCPD sum (8e-8 mA/rtHz), and then referred to displacement noise using the GDS channels for optical gain and cavity pole (optical gain = 3.24 mA/pm). This measurement well explains the H1 noise level at high frequency.

When trying to match the measured shot noise level with GWINC, we realized that an easy way to match the frequency dependence caused by the low H1 cavity pole frequency is to increase the transmission of the SRM to 39% (nominally 37%). We believe the real mechanism is actually a mismatch of the SRC cavity wrt to the arms, translating in a lower "effective" reflectivity of the SRM mirror (i.e., less efficient signal extraction). 

Still, even after fixing the frequency dependence with the higher SRM transmission, if one tries to reproduce the shot noise level with GWINC by using parameters inferred from measurements of loss/power at the various IFO ports, it is evident that extra losses (or lower powers) are actually needed. 

With power entering the IFO ~19W, recycling gain ~37, intra cavity power ~95kW, we need ~15% extra losses in the readout chain wrt to the measured ones (25%) in order to reproduce the observed shot noise level. One obvious way of getting that is if the OMC throughput and mode matching are not as good as inferred from  previous measurements .

A 10 degree mistuning of the homdoyne phase wrt to the "nominal" (pi/2) could explain up to ~5% of the observed loss. 

Bottom line: while one can see many ways to improve this analysis, either powers / recycling gain are not what we think they are, or, most likely, we have higher loss in the readout chain that we think we have (and an imperfect homodyne phase). 

The plot shows the best H1 noise, the H1 measured shot noise, and the GWINC quantum noise model including radiation pressure noise corresponding to 95kW in the arms.


% GWINC parameters
ifo.Laser.Power = 21 * 0.88;
ifo.Optics.Loss = 45e-6;  %Arm loss = 90 ppm
ifo.Optics.BSLoss  = 0.5e-3;  
ifo.Optics.PhotoDetectorEfficiency  = 0.65; %Readout loss tuned to match measured shot noise level
ifo.Optics.SRM.Transmittance  = 0.39;    % "Effective" Transmittance of SRM    
ifo.Optics.SRM.Tunephase = 0;            % SRM tuning
ifo.Optics.Quadrature.dc = 90 * pi/180;  % demod/detection/homodyne phase pi/2  

%GWINC Output
Laser Power:              18.48 Watt
SRM transmission:          0.3900
Finesse:                  445.93
Power Recycling Factor:   36.83
Arm power:                95.26 kW