Kara Merfeld, Jenne Driggers, Keita Kawabe

We have created a PRCL feedforward filter such that it reduces the coherence between PRCL and DARM at frequencies between about 20-200Hz.  

We isolated the frequencies that had a coherence with DARM higher than .9, and fit a function to the necessary transfer function at those frequencies.  In the plots of the fit and the residuals below, we only care about the quality of the fit within the frequency band that is highlighted green.  Outside of this band, we only care that the fit has a small magnitude well below the transfer function, so as to avoid injecting into DARM.  

The residuals of the magnitude in the attachment are a ratio of the transfer function to the fit, so an optimal fit has magnitude residuals =1.  

We can see in the coherence spectrum that we have increased coherence around 10Hz from what we had with no filter.  This is a result of the filter fitting the transfer function poorly around those frequencies, and we are actually making an injection around there.  We can also see this at lower frequencies in the DARM spectrum.  But more importantly, we reduced the coherence between PRCL and DARM between ~20 and ~200 Hz.  We did an on/off test to confirm that the reduction of coherence above a few tens of Hz is really from the new FF. 

After applying this filter, our range increased by 2-2.3Mpc, giving us about 5% more sky volume.  Here is a url to the range plot: https://ldas-jobs.ligo-wa.caltech.edu/~detchar/summary/day/20190827/ to show the increased range, after proper calibration is applied.

The new settings for the filter are included as an attachment.  This filter has been added to the ISC_LOCK guardian.  We implemented this filter on August 8th, 2019, 22:30:00 UTC.