[Laurence, Jenne]
We've recently imported V2 of Lee's IIRrational, and have been updating the feedforward fitting scripts to work with it (with lots of great support from Lee), and have applied it to feedforward measurements taken on Monday, after we'd thermalized at the new PSL power of 37W.
Today we took a few minutes of pre-approved commissioning time to switch to the new MICH and SRCL feedforward filters. The attached screenshot shows that the MICH filter is doing some good, however the SRCL filter is doing more harm than good. I'd like to leave it on for a little while, but since we've got another pre-approved commissioning time once the Evening operator arrives to help the TCS CO2 laser, I'll likely revert the SRCL filter.
In the attached plot, upper left shows LSC-DARM_OUT and the MICHFF and SRCLFF outputs that get sent to the ETMs to sum in with DARM. References are all on the old 35W filters, current traces are the new filters. You can see that there isn't much difference with the MICH output, but the SRCL output is much larger at low frequencies, and slightly larger at high frequencies.
Upper right shows the difference in the calibrated Darm spectrum, that we've improved a bit between 10Hz - 15Hz, and above that any changes are quite subtle. We've definitely degraded the spectrum below 9 Hz (the SRCL fit really, really wants to give a highish-Q peak there, to get the fit better above 10 Hz), which will contribute to upconversion around lines in the bucket.
Lower left shows the coherence between SRCL and DARM, and you can see that between 10-15 Hz, we've made an improvement, but both below and above that we have significantly hurt things, which is why I will revert the SRCL FF in 30 min or so.
Lower right, however, shows that the coherence between MICH and DARM is improved between 10-25 Hz, so that seems good.
