Abstract:
I measured the transfer functions from MN_SUMOUT_L_OUT to ALS_DARM signal.
Measurement was done only below 12 Hz because we cannot obtain high coherence above 12 Hz.
Measured TF follows f^-6 around 10 Hz, which is predicted from the theory.
So, I extrapolated TF and estimated noise spectrum from MN photosensor control, which was dominant below 50 Hz during O3GK.
Around 10 Hz, it is consistent with O3GK noise and simulation result of photosensor noise.

Detail:
I measured TF from ETMX_MN_SUMOUT_L_OUT to ALS DARM signals.
Figure 1 shows the result.
Since MN is two-stage upper from TM stage, TF should be a function of f^-6 theoretically, so I multiplied f^6 to the TF measurement results (fig2).
As you can see, TF is almost flat around 10 Hz, so there is no discrepancy from theory at this moment.

Since it is hard to measure above 12 Hz due to low coherence, I extrapolated the data with f^6 for estimating noise projection from MN photosensor control.
Figure 3 shows the result of noise projection (red: noise projection, black: best sensitivity).
In this plot, I assumed photosensor noise from each suspension equally contribute to DARM: in other words, just muliplied a factor of 2 to the noise projection from ETMX photosensor control.
As you can see, noise projection of current control noise is almost same as the noise level during O3.

According to my simulation of noise projection before (JGW-G2113593-v3), noise around 10 Hz is about ~10^-13, which is consistent with the estimation this time.

Note:
*Measured TF at 4.5 Hz is not on the line of f^-6 but it is not strange because there are many resonances around there and highest resonant frequency of L mode is about 5.1 Hz.
*As you can see, there are some bumps at 20-50 Hz in best sensitivity, which cannot be expected from the simulation; so this result doesn't mean suspension noise really decreases above 30 Hz. So, we need to update noise projection when we obtained better sensor for noise estimation like FPMI.
*Since damping control, especially rol-off frequencies are changed from the ones when I simulated photosensor noise, higher frequency noise is different from the simulation result.