Attached are few figures relative to the measurements.

Figure 1 - SSFS noise injections. In particular, the SSFS TF is shown, and the coherence with DARM.

Figure 2 - Time series of the lock/unlock measurements. What is worth noticing is that the beam size and position change between lock and unlock, and this variation is seen by cameras and QDs of both B4 and B1p. It's not clear why it happens. There are two hypothesis:

- the beam reflected by the ITM has a different alignment between the lock and unlock state, and this translates in a misalignment through the telescope up to the camera, which induces an astigmatism (most probable)

- the beam really has a different shape between the locked and unlocked state

In the next figures, the beam shape is also shown by the B4 and B1p cameras. In particular:

Figure 3 - North arm locked

Figure 4 - North arm unlocked, only NI mirror aligned

Figure 5 - West arm locked

Figure 6 - West arm unlocked, only WI aligned.

I tried to extrapolate the arm RTL from the photdiodes signals on B1p, B4 and B5 during the lock/unlock. The signals are shown in the first attached picture. Only B1p_PD2 seems to give reliable values for the losses. They are 61 +/-5 ppm for the Narm and 56 +/- 5 for the Warm.

B5 is measuring losses higher than 150 ppm which are not compatible with the upper limit set by the recycling gain.

The behaviour of B4 is particulary strange as it observes an increase of the reflected power when the cavity is locked. This is probably explained by the observed change in the beam position and shape between the locked and unlocked state. To be investigated...

In the second attachment there are data I used to check PD offests (I took a IMC unlock just before the measurement).

I remark that when the cavity is unlock the end mirror is also misaligned. It means that the input power used for the computation is actualy the reflected power divided by (1-Tin).