[Lucia, Luca, Jenne]

Today we diagonalized the ETMY optical levers using the free swinging data after Masayuki kicked the suspensions this morning (klog 10581).  The first attachment shows the original signals from the oplevs on the left, and the result after the diagonalization of the input matrix on the right.  Note that the yaw resonance is at about 0.35 Hz, and the length is about 0.64 Hz. The plot on the right after diagonalization is just the input matrix applied to the original data, so certainly our actual decoupling won't be as good as this plot implies.  To do the diagonalization we found the peak heights for each resonance in the two QPD outputs, and inverted the resulting matrix.  We then made the plot on the right to confirm that our inverted matrix made sense.  When we implemented the matrix, we normalized the yaw row of the matrix by the tilt_yaw -> yaw element, and normalized the length row of the matrix by the len_yaw -> length element.

After that, we measured the transfer functions driving each of the lowest 3 stages in length, and measured the test mass oplev length signal.  The 2nd, 3rd, and 4th attachments are these measurements.  In each of the measurement plots I have also included the TypeA transfer function, exported from foton from the relevant stages' TM oplev damping filter bank.  For the test mass stage, I took it from ETMX since ETMY didn't have a susmodel filter, but it should basically be the same.  The models assume no damping, but we did indeed have damping on during our measurements.  The blue traces are the measured data points.  There aren't any error bars plotted, but the measurements match the model fairly well, which means that we can trust the models when using them to design offloading of the DARM control to the upper stages of ETMY.   For each of these measurements, the drive went through the LOCK_L filter bank of the relevant stage.  To do this, I turned off the input of the Lock_L filter bank (just in case), and made sure that the gain was 1, the output was on, and I turned off any filters that were on.  A trick that we used here, to avoid ringing up suspension resonances, was to individually chose what frequencies to measure, and what amplitude of drive to use.

Rana is working on the model of the ETMY suspension so that we can design offloading filters to pull the low frequency DARM control signal to the MN stage of ETMY, leaving only the higher frequency components on the test mass.