Follow up of [LHO ALOG 68050]

It was known that the peak at 32Hz appears in the MICH error signal (POP45) somewhere between the ISC guardian nodes TRANSITION_FROM_ETMX and NOMINAL_LOW_NOISE.

There was a chance to track this down today during the relocking around GPS time 1365720000.

Careful tracing of the MICH spectrum revealed that the peak appeared at GPS Time 1365720791. This time was in the middle of ISC Guardian node 560 "LOWNOISE_LENGTH_CONTROL". Attachment 1 shows the active steps around thhat time. The vertical lines indicate GPS times from 1365720789 to 1365720793 for each second. The switching of SRCL filter module corresponds to turning the SRCL OFFSET on (see Attachment 3 for the code).

The change in the MICH/PRCL/SRCL error and the REFL45I and REFL45Q for these five times are shown in Attachment 2. Since the ASD measurements were limited to a BW of 1Hz with AVG of 1, the measurement time is limited to 1 sec. Therefore it indicates that the SRCL offset contributes to the coupling of the peaks to the MICH. It is still unclear why the SRCL offset has this effect.

Attachment 2 also shows that REFL45Q already has the peak independently from the SRCL offset.

Many (but not all) LSC/ASC error signals have this 32Hz peak (in fact 32.2Hz). In the previous aLOG, the coupling to DARM was predominantly via MICH/PRCL. As the peak in PRCL via POP9 was not inherent but coupling from MICH feedback, implementing a 32Hz notch filter in MICH servo helped to remove the 32Hz noise in DARM. However, since the ASC and REFL45 signals have this peak, identifying and removing the noise source is ultimately desirable.