J. Kissel

For future reference, I attach the differences between 
    - the O1 H1SUSETMY L2/L3 crossover design (critiqued in G1501372) which had been accidentally running for the first month of O2 (see LHO aLOG 32540) and 
    - the improved design (created by E. Goetz in July 2016; see LHO aLOG 28746) now used since the restart of the run post-winter break (see LHO aLOG 32933).

Recall the goal of the improvement:
    - To high-pass the TST/L3 drive at low frequency more aggressively
    - To low-pass the PUM/L2 drive at high frequency in a more simple, sensible way to get isolation from the TST/L3 stage down to much lower frequency
    - To improve the interaction between PUM/L2 and TST/L3 such that the super-actuator transfer function was not a wiggly mess

The .pdf attachments show the old vs. new design in the form of actuator authority, in test-mass displacement in [m] per LSC input [ct] (where the hybrid offloaded / distributed hierarchy filters have been converted to entirely distributed for sanity's sake). In the new design, the extremely aggressive, high-Q, many-order elliptic 900 Hz low-pass filter in the PUM stage has been removed, and replaced with a much softer Q'd low-pass at lower frequency. Similarly, on the low-frequency end, an additional high-pass has been added to the L3 stage filtering.

The .png attachments show the requested DAC outputs for all globally driven stages of H1 SUS ETMY, for the new design (taken on 2017-01-19) and the old design (taken on 2016-12-21). The spectra were gathered under similar environmental conditions (0.5 - 0.7 [um/s] RMS microseism, ~5 [mph] winds), and when the IFO was performing roughly the same (at 65-70 [Mpc] BNS range). Although we can't see everything from the data taken in the past (the 2016-12-21, old design), we can see that in the new (2017-01-19, new design) data the PUM/L2 stage requested actuation is consistently ~2 orders of magnitude below the TST/L3 stage, where it used to be only ~1 order of magnitude about a few hundred [Hz]. We can also see that, although the difference is not hugh, the TST/L3 DAC request is now slightly below the PUM/L2 request as one might intuitively hope a from reasonable design.

In summary: all goals have been achieved, the design is now sensible, and the balance of requested output from each stage is better.