5.2 Control of the proof-of-concept experiment
5.2.1 Degrees of freedom
The arm cavities of the Sagnac speed meter, like those of a Fabry-Perot Michelson interfer- ometer, must be held resonant in order to maintain the light power required for the design sensitivity, and so these cavities represent a degree of freedom that must be controlled with active feedback. Meanwhile, the error signal is insensitive to the motion of the inter-cavity mode matching mirror, M9, since this is situated at half the total round trip distance and is sensed by the counter-propagating modes at almost the same time. Other mirrors are potentially significant: the beam splitter M6and steering mirror M7, as shown in figure 4.9. As these mirrors are situated near the start of one and the end of the other modesโ round trips, a velocity dependent signal is created at the balanced homodyne detector (BHD, see section 4.2.1). We neglect all other auxiliary optics.
To assess the importance of the optics to the interferometerโs sensitivity to differential arm cavity length ๐ฟ(โ), transfer functions from individual mirrors to the BHD port, where the ๐ฟ(โ)signal should by design couple most strongly, were calculated using Optickle (see ap- pendix C.1.2). The results in figure 5.1 show that the cavity mirrors are the most important positions to control, with the arm cavity finesse enhancing the sensitivity of the BHD to the arm cavity mirrors such that they dominate the signals from M6and M7. These results have been confirmed both with Finesse and analytically [141].
The common mode motion of the arm cavities, ๐ฟ(+), will also need to be controlled by means of a photodetector placed at the input port to sense the light returning from the interferometer back towards the laser. This motion will be suppressed by applying strain and heat to the laserโs crystal to change its geometry and therefore lasing frequency. This solution involves the creation of a wide bandwidth controller able to provide large correc- tions within the audio band. While the control of ๐ฟ(+)is crucial to maintain the light power within the arm cavities, we focus on ๐ฟ(โ) given that it represents the main signal appear- ing at the output port and the one which will primarily contribute to the sensitivity of the interferometer in the context of gravitational wave detectors.
While the motion of M9 can be suppressed at the main readout given suitable mirror po- sitioning in order to cancel the signal from each of the counter-propagating modes, the effect of M6 and M7 is less clear cut. To assess the impact the motion from these mirrors has on ๐ฟ(โ) sensitivity, a calculation of the effect of seismic noise from M7to the BHD can be made. M6need not be considered separately here: the transfer function is almost iden- tical to that of M7 and so we need only calculate one, and the suspension designโa work in progress at the time of writingโis intended to have better isolation than that of M7โs auxiliary suspension.
104 106 108 1010 1012 Resp onse ( W m ) ๐ฟ(โ) ๐6 ๐7 101 102 103 104 105 Frequency (Hz) โ180 โ135 โ90 โ45 0 45 90 135 180 Phase (ยฐ )
Figure 5.1: Transfer functions from important mirrors or combinations of mirrors in the Sagnac speed meter experiment to the balanced homodyne detector. The ๐ฟ(โ) degree of freedom has the
strongest response by design. The main beam splitter, M6, and the steering mirror for cavity A, M7, have response a factor of 10โ3that of ๐ฟ
(โ). Other mirrors have significantly lower coupling. Measurements of the seismic motion present upon the ground outside the vacuum system can be propagated through a model of the passive seismic isolation within the vacuum sys- tem to obtain the effective seismic-induced motion of the tables upon which the suspensions sit. The seismic motion of M7can then be calculated by multiplying this spectrum with the transfer function of the auxiliary suspension from the table to the test mass, taken from a state-space model. This seismic noise can be projected into an effective differential arm cavity motion displacement spectral density by multiplying it by the ratio of the transfer functions of M7 and ๐ฟ(โ) to the BHD port1, taken from figure 5.1. This can be compared with the requirement for sensitivity of the BHD to ๐ฟ(โ). Figure 5.2 shows that M7โs motion, projected into ๐ฟ(โ), will meet the requirement above 100 Hz, and the result is similar for M6.
The results in figures 5.1 and 5.2 show that control of ๐ฟ(โ) will be required to meet the sensitivity requirement at the BHD port above 100 Hz, where the measurement of reduced
1This is the same as multiplying the motion of M
7by its transfer function to the BHD port to yield a signal
in WโHzโ1, and dividing by the transfer function from ๐ฟ
(โ)to the BHD to yield an effective motion in terms
101 102 103 Frequency (Hz) 10โ22 10โ21 10โ20 10โ19 10โ18 10โ17 10โ16 10โ15 10โ14 10โ13 10โ12 Effe ctiv e ๐ฟ(โ) displacement noise ( m โ Hz )
๐7effective ๐ฟ(โ)noise from seismic Sagnac speed meter requirement
Figure 5.2: Effective ๐ฟ(โ)seismic noise contribution from M7. This is calculated by first propagating
a seismic noise spectral density for the laboratory near the vacuum system through damping and suspension models to obtain the motion of the M7 test mass. With this figure, the response at the BHD can be calculated from the transfer function shown in figure 5.1, and this in turn can be expressed in units of differential arm cavity motion by dividing it by the response of ๐ฟ(โ) to the
BHD port. The requirement is given only for frequencies above 100 Hz where the measurement of reduced radiation pressure noise will be made, and this figure shows that seismic motion of M7 will not represent a significant problem to the sensitivity of the experiment in the desired band. This conclusion applies also to the main beam splitter, M6, which is expected to have even greater isolation from seismic noise.
radiation pressure noise will be made. It should be noted, however, that the desired BHD homodyne angle depends on the relative path lengths of M11 to M16 and M6 to M16. This length will be controlled by an auxiliary loop not considered part of the longitudinal control strategy, and will be the subject of future work alongside a strategy for the control of ๐ฟ(+).