Mirror Suspension Control VSR 1 learning E Majorana

Mirror Suspension Control VSR 1 learning E. Majorana INFN MSC workgroup LSC-Virgo joint meeting EM-MSC-251007 Hannover 25 -10 -2007

outline I. III. IV. EM-MSC-251007 MSC status at VSR 1 startup: quick reminder. MSC tuning during VSR 1. Noise issues. Post-VSR 1 considerations and perspective. 1

I. EM-MSC-251007 MSC status at VSR 1 startup: quick reminder 2

Mirror Suspension Control at VSR 1 startup: essentials Compensation of actuator non-linear recoil: for both Marionette and Super. Attenuator. Global-Inverted-Pendulum-Control (partial): to increase Pos/Acc sensor crossover frequency (up to 70 m. Hz) without significant Seism re-injection (0. 15 -0. 7 Hz). Pos/Acc prefiltering strategy tunable on-the-fly according to the wind (0. 02 -0. 7) or sea (0. 15 -0. 6) disturbance. Lock force hierarchically controlled through 4 marionette (two FP arms): to avoid saturations at low frequency. Local control roll-off (NI, WI, BSyaw) reduced to improve stability. IMC Suspensions (MC, IB) non-optimized and with no V-damp. EM-MSC-251007 3

I. MSC status at VSR 1 startup: quick reminder (2 -3). II. MSC tuning during VSR 1 EM-MSC-251007 4

MSC tuning and improvement during VSR 1 • NI-WI LVDT/ACC : anti-Wind => anti-Sea => anti-Wind (Jun 5) • MARIO lock re-allocation on NI-WI OFF (NE, WE only) (Jun 6) • BS LC tuning to improve regions with small phase margin (Jul 4) • Seism-Free Reconstruction of PR top stage Err Signal (Jul 6) • EQG Guardian to disable GIPC in case of Earth. Quake (Jul mid) • NI-WI mirror re-centring on beam (Jul mid) • GIPC and m. SFR complete configuration (Sep 15) Overall Mirror Suspension Control Noise: no significant limitation of present Virgo sensitivity. Open issues: worse sensitivity during bad weather days ( seism and wind). EM-MSC-251007 5

Initial Vs final: sensor prefiltering Example. . hybrid filters (on-the-fly tuning) old old new new ACC HPw LPw + mix HPLP mix HPs + cross IP seism LPs LVDT VSR 1: sensor prefiltering tuned, after few days, to reject wind diturbance injected by accelerometers. EM-MSC-251007 mix =0. 5 ‘medium’ attenuation of LVDT seism noise Compared to the starting config (crossover @ 50 m. Hz) mix =0 (wind-earthqukes, f <70 m. Hz): “aggressive” attenuation of accelerometer tilt noise. mix =1 ( seism, 150 -600 m. Hz) : “aggressive”, slightly worsened against tilt noise. 6

Initial Vs final: overall control strategies seism incoherent along the arm baseline Seism reduced at END sites by using position referred to INPUT; misfunction during EQ Also the Acceleration ! seism coherent in the central area SFR possible by referring PR and BS to NI and WI WE GIPC differential position deduced from GC reconstructed by local DSP PR NE BS Global. Control INPUT TOWERS AS GROUND REFERENCE EM-MSC-251007 7

MSC Vs robustness (1): earhquakes/HNSNS drops/EQG patch Coherent excitation M 6. 5 GIPC cannot be used under coherent excitation (far EQ*) Mpc m m fast recover An automated Guard to disable GIPC: EQG Top-stage corr. from locking V V V z. Corr Saturation 4 -min to recover H instead of 30 -40 min without EQG EM-MSC-251007 (*MSC talk at LSC-Virgo May 07, plenary): 8

MSC Vs rrobustness (2): earhquakes/stable GIPC Comparison of two events with similar local amplitude A “lucky” occurrence !* 40 m peak Previously we had saturation as the amplitude was less than 20 m MARIO: small residual correction (~7 m) The system is much more robust. Correction dynamics more than doubled. EM-MSC-251007 (*Indonesia M 6. 8, Sep-20 -08. 31): 9

MSC Vs roboustness (3): earhquakes overall Local seismometric SA response Vs ITF lock EQG action no longer needed. We use EQG as alarmed on-line seismometer m no unlock through GIPC disabling (EQG) unlock SFR_PR+EQG SFR_PRBS+stable. GIPC day EM-MSC-251007 10

Sub-conlcusion II Top-stage control strategies involving suspension operation as-a-whole improve disturbance rejection capability. EQ GUARD used in monitor mode. SFR (Central Area) GIPC (ARMs) higher cross-over (70 m. Hz) (lower wind noise without seism re-injection) & Coherent response to EQs EM-MSC-251007 Next ! ENV-TUNED GUARD 11

MSC Vs seism (3) : main path/payload motion attenuation at the suspension point Effective attenuation of seism at the payload The lock force applied to the marionette corrects the residual payload motion, whose rms above 100 m. Hz is ~ 1 order of magnitude smaller than the ground motion. EM-MSC-251007 12

MSC Vs seism (4) : rejection VSR 1 start-VSR 1 stop Monitoring of channels used for top-stage control at VSR 1 start setup. Combined channels used at VSR 1 stop: SFR ( Seism-Free Reconstruction) and GIPC (Global. Inverted-Pendulum Control) EM-MSC-251007 13

MSC Vs seism (5) : rejection VSR 1 start-VSR 1 stop Monitoring of channels used for top-stage control at VSR 1 start setup. Combined channels used at VSR 1 stop: SFR ( Seism-Free Reconstruction) and GIPC (Global. Inverted-Pendulum Control) EM-MSC-251007 14

MSC Vs seism (1): high seism day during VSR 1 seism decrease Horizon increase unaddressed noise bump seism > 1. 5 mrms Sensitivity stability seism < 1. 5 mrms Sensitivity seism < 1 mrms: No effect main misalignment indicators main stability indicators EM-MSC-251007 Diff pitch at DF DF before OMC IBpitch OMC transmission Sideband pwr probe Carrier pwr probe 15

Noise impact study using VSR 1 data and sub-conlcusion III Higher gain for injection bench angular control (pitch) necessary to improve stability during high sea activity. Even though the seism disturbance is the range of suspension resonant frequencies, the overall impact at the level of the mirror is relatively small: a factor 10 of rms at the ground worsens by a factor 2 the accuracy of controlled signals at the ITF level. What is the coupling path ? EM-MSC-251007 16

I. MSC status at VSR 1 startup: quick reminder (23). II. MSC tuning during VSR 1 (4 -10). III. Noise issues). EM-MSC-251007 17

I. III. IV. EM-MSC-251007 MSC status at VSR 1 startup: quick reminder. MSC tuning during VSR 1. Noise issues. Post-VSR 1 considerations and perspective. 18

OBO: attempts for systematics One-By-One injection to mimic seism disturbance occurred during VSR 1 seism OBOQuiet OBO Single multiple NE_top WE_top NI_top WI_top BS_top PR_top IB_top MC_top OB_top ………. Probes Locking feedback on end mirror payload (z. GC) Locking feedback on BS and PR (z. Corr) End Mirror Alignment (B 1 p_q 1_AC_pv) Input Mirror Alignment (B 1 p_DC) Injection beam jitter (IB_ty, IB_tz) Sideband power (B 1_DC) ……. . ? 4 entry-points: END, INPUT, BS, PR, ISYS 3 d. o. f: longitudinal (z), transversal (x), vertical (y) EM-MSC-251007 19

OBO: injection at NE-WE top-stage, a “positive” clear result. activated probe END mirror pitch SA TOP Payoad OBO ITF Actual seism GW channel yaw unperturbed z. GC pitch perturbed An efficient way to produce Horizon drops similar to actual one as seism rms is ~ 3 m is through mirror pitch noise (consistent with VSR 1 experience). EM-MSC-251007 20

Injection @ NE-WE As expected, due to GIPC, injections @ NI-WI produce very similar effects. EM-MSC-251007 21

OBO: injection at IMC top-stages Very difficult to excite the IB simply injecting pseudo Seismic noise at the top-stage. Power fluctuations close to the actual ones only by using a strong line, tuned on IB main pitch mode. EM-MSC-251007 22

Injection @ Input. MC a surprisingly “negative” clinical result, Some hidden path… EM-MSC-251007 23

Sub-conlcusion IV and summary In spite of suspension resonances, normally there is no environmental seismic effect on the sensitivity. High seism means that the ground shake in the range 0. 2 -0. 7 Hz can increase up to 10 times, worsening mirror control signals by a factor 2. Two main effects ITF Power fluctuations Specific coupling with sensitivity due to diff. pitch Improvement of IB angular control (coming soon) In order to cope with a noise bump at 200 -300 Hz (that should be removed) we can: - decrease the gain demand to alignment control (by setting input mirrors under AA) - improve the strategy (higher LVDT/Acc crossover and tuned reallocation techniques). Quite reasonably achievable to gain a factor 2 EM-MSC-251007 24

Not used EM-MSC-251007 25

STANDARD CONFIGURATION FOR LONG SUSPENSIONS stage variable actuator ref SA TOP Pos/Accel 3 D+yaw Ground(~)/stars SA BOTTOM Position yaw Ground PAYLOAD Position 2 D+pitch/yaw SA BOTTOM Basic requirements: sensing and actuation diagonalization + EM-MSC-251007 26 hiearchical control

Virgo “standard-super-attenuator”suspension … Soft isolator concept: 1. very efficient passive attenuation 2. active controls for normal mode damping EM-MSC-251007 27

seism Vs VSR 1 sensitivity: quicklook to the data/ seism > 1. 5 mrms Seism level: 3 m rms Seism level: 2. 5 m rms Seism level: 2 m rms Seism level: 1. 5 m rms Seism level: 1 m Seism level: 0. 5 m rms 100 -500 Hz: power fluctuations due to injection misalignment driven + specific bump coupled to ITF pitch misalignments Other features: 10 -100 Hz: glitches uncorrelated to power fluctuations EM-MSC-251007 small noise floor fluctuations well correlated to power fluctuations 28 500 -10000 Hz:

MSC Vs seism (2): main path/ground excitation LVDT sensor measures the position of top-stage suspension point (top of the IP) with respect to a grounded rigid mechanical frame. Top-stage LVDT channels provide the best measurement of ground noise EM-MSC-251007 29

MSC Vs seism (6) : residual impact Impact on the ITF longitudinal error signals EM-MSC-251007 30

MSC Vs seism (7) : residual impact Impact on the ITF angular error signals, in loop full bandwidth EM-MSC-251007 31

MSC Vs seism (8) : residual impact Impact on the ITF angular error signals, DC controlled in loop full bandwidth EM-MSC-251007 32

MSC Vs seism (9) : residual impact Impact on the IMC angular error signals, in loop longitudinal EM-MSC-251007 33

Injection @ BS Not enough to explain the noise in actual condition, considering that the applied disturbance was much larger in OBO tests. EM-MSC-251007 34

Noise injection at BS suspension top-stage With a similar excitation of the top stage, the longitudinal accuracy is much worse. The angular motion is larger in tx, smaller in ty. EM-MSC-251007 35

Noise budget now + what next? (E. Tournefier) Control noises: further reduction - sensing/driving matrices improvements / 8 MHz? - angular control filters/ better signals with new end benches telescope Actuator noise is not far at low frequency => new coil driver (more filtering) Where is the Eddy current noise? Remaining mystery noise => Brewster removal + diffused light mitigation 700 Hz monster? ? Cleaning of the mirrors + TCS will the error signals get cleaner? EM-MSC-251007 36