The goal of todays shift was to perform some measurements analyzing the influence of the online subtraction of MICH and SRCL on Hrec and a test of an improved MICH subtraction filter.

To test the influence of online MICH & SRCL subtraction on Hrec, we let the interferometer run in LN3 with the subtraction methods off and on. We let the interferometer converge to a steady-state for 5 minutes after which we took 10 minutes of clean data in every situation to analyze their influence on the sensitivity. The GPS times are:

14:08:00 UTC + 10 min: MICH on & SRCL on

14:33:00 UTC + 10 min: MICH off & SRCL on

14:48:00 UTC + 10 min: MICH off & SRCL off

15:05:00 UTC + 10 min: MICH on & SRCL off

Interestingly we noticed that the 50 Hz line subtraction performed significantly worse after disabling the MICH subtraction. At 14:31:00 UTC we attempted to reset the 50 Hz subtraction to see if this had any influence but we didn't see a noticable difference.

We will post a more detailed analysis tomorrow on the results of this test.

After that we moved on to testing a new subtraction filter for MICH which should improve the subtraction between 10 and 20 Hz.

We first performed noise injections for MICH and DARM using the current filter (GPS file 1414250738) and without alpha filter enabled (GPS file 1414251289).

At 15:51:10 UTC we enabled the new filter and at 15:57:55 we optimized the gain of the filter by minimizing the LSC_DARM_MICH_COUPLING channel which computes the coupling on the line frequency of MICH (which we enabled at 15:57:55 UTC). We then performed noise injections using this new alpha filter (GPS file 1414253778). 

While analyzing the data we let the interferometer run for 30 minutes starting from 16:30:00 UTC with the DRMI lines enabled to gather long stretch of data to see how much the coupling fluctuates over time.

During the analysis we noticed a mistake in the filter (we added a pole at 1000 Hz for roll-off but this double pole has a significant influence on the performance of the alpha filter around 20 Hz, i.e., reduces the maximum attenaible attenuation by a factor 4-10 roughly. We removed this pole and started tuning the gain of the filter by injecting noise in MICH.

We first measured the coupling by injection noise in MICH using the current filter at 17:22:30 UTC + 30s. 

We then moved to tuning the new filter without the additional pole from 17:24:30 onwards. After that we again ran the injection script for MICH and DARM (GPS file 1414259325). 

Finally at 18:16 UTC we reloaded the filter because we made a mistake setting the gain and gain frequency in ACL. After that we quicly retuned the gain and found an optimum of 1.032. 

We are in the process of analyzing the data and will add another entry once we finished analyzing the data.