To be able to actively follow and control the etalon effect, we implemented a new server that tracks the frequency of the butterfly-modes of the input mirrors. This was done in the past using Michele's line monitor, but this hasn't been updated for the new mirrors and it is running on Michele's desktop PC.

The server is a small program in C that wraps a compiled matlab algorithm, similar as was done for the analysis of the phase-camera images, but much more flexible. Changing the algorithm or adding new inputs and outputs only requires changing and compiling the matlab code and updating a configuration file. The server takes data from FbmFE and sends its outputs to FbmMain, so the outputs should be available to the various Alp servers for implementing servos.

The algorithm to compute the frequencies first demodulates the B1_ACp around the frequencies of interest and down-samples the data to 4 Hz. Every 25 seconds, a FFT is calculated using 50 seconds of data, which is zeropadded by a factor 4. An average spectrum is then calculated using the last 5 half-overlapping windows. The frequency of the peak is finally found as the center-of-mass of all points that are above a certain fraction of the maximum amplitude. This should yield a frequency resolution of a few mHz and thus a temperature resolution of a few mK. For the moment, only the frequency of the butterfly modes is calculated since, at least for the WI, they appear to give to cleanest signals. Two new channels are available: ML_NI_Butterfly_Freq and ML_WI_Butterfly_Freq.