1. Caltech, February 12th1 Virgo central interferometer: commissioning and engineering runs Matteo Barsuglia Laboratoire de l’Accelerateur Lineaire, Orsay

2. Caltech, February 12th2 Summary Introduction The central interferometer Operation with a simple Michelson Operation with a recycled Michelson Operation with the full injection system E-run programs Conclusions

3. Caltech, February 12th3 Pisa Virgo aerial view

4. Caltech, February 12th4 Virgo sensitivity Seismic noise Thermal noise Shot noise

5. Caltech, February 12th5 Virgo optical scheme

6. Caltech, February 12th6 The central interferometer (CITF) BS

7. Caltech, February 12th7 CITF: goals Test all the tehcnical choice during arm construction: Suspensions Fully digital control chain Output mode-cleaner Local controls

8. Caltech, February 12th8 Suspensions Top stage Last stage Seismic filters

9. Caltech, February 12th9 Suspension Control Top stage Lower suspension stages: “marionetta” (from upper suspension stage) mirror (from “reference mass”)

10. Caltech, February 12th10 Control Architecture Completely digital LinxOS C or C++ photodiode-read-out 20 kHz Control 10 kHz Global control (PowerPC platform): read phd signals algorithm for lock acquisition linear locking and alignment For each suspension DSP correction sharing Resonance compensation local controls DAC 20 bits Photodiodes (powerPC platform): ADC 16 bits compression dynamics filters GPS Timing DOL’s

11. Caltech, February 12th11 Local Controls Output mode-cleaner Completely out of vacuum CCD camera Coarse system, markers (50 mrad) Fine system (laser beam, optical lever) Control from marionetta (noise filtering)

12. Caltech, February 12th12 Detection system Suspended detection bench Output mode-cleaner

13. Caltech, February 12th13 Operation with a simple Michelson Superattenuator controllability Hierarchical control Digital control chain Output mode-cleaner Control robustness

14. Caltech, February 12th14 Suspension performances No excitation of unwanted degrees of freedom High robustness to non stationnary noises Passive filtering experiment (see Braccini’s talk)

15. Caltech, February 12th15 Michelson locking with top stage Fast corrections (f > 70 mHz) Slow corrections (f < 70 mHz) 3.5 mN Force applied to mirror No feedback to top stage with feedback to top stage

16. Caltech, February 12th16 Control robustness Results from E0 run (72 hours) : ITF continuously locked on dark fringe for more than 51h 1 unexpected loss of locking, duty cycle > 0.98 % 51 hours unlocked (bright fringe) locked (dark fringe)

17. Caltech, February 12th17 OMC locking on dark fringe Transmitted power  signal TEM 00 Contrast improvement ~ 10

18. Caltech, February 12th18 Operation with a recycled Michelson Lock acquisition Frequency stabilization Linear alignment

19. Caltech, February 12th19 Lock acquisition

20. Caltech, February 12th20 The lock acquisition problem modules storage North tunnel Force needed to stop the mirror (finesse = 250) maximum force 40 mN (limited by EM noise)

21. Caltech, February 12th21 Strategy (I) - enlarge the acting time modules storage North tunnel Use of an antisymetric trigger > 50 % open few % close Widening the error signal Pr_B5_ACq (Pr_B5_DC) p

22. Caltech, February 12th22 A simulated lock acquisition trigger ITF internal power Dark fringe speed recycling speed PR correction WI correction

23. Caltech, February 12th23 A real lock acquisition Correction PR Correction WI ITF internal powerDark fringe power

24. Caltech, February 12th24 Frequency stabilization crossover ~ 3 Hz very aggressive filtering above 13 Hz

25. Caltech, February 12th25 Linear alignment

26. Caltech, February 12th26 Linear alignment - results ugf ~ 5-10 Hz

27. Caltech, February 12th27 Operation with injection system

28. Caltech, February 12th28 Acquisition detection switch Dark fringe control switched from B1p to B1 Offset between B1p and B1 dark fringe on B1p  dark fringe on B1 Need offset compensation and smooth transition

29. Caltech, February 12th29 Optical characterization Input power ~ 2 - 2.5 Watts Recycled power (maximum) ~ 240 Watts Not coupled light ~ 30 % Interferometer contrast: ~ 5 10 -4 (before OMC), ~ 5 10 -5 (after OMC)

30. Caltech, February 12th30 E4 sensitivity Alignment control noise Laser frequency noise

31. Caltech, February 12th31 High frequency noise Laser frequency noise Peaks: mirrors + holders

32. Caltech, February 12th32 Intermediate range noise mode-cleaner mass TF no common mode loop

33. Caltech, February 12th33 CITF e-run program 5 e-runs (september 2001-july 2002) 72 hours each 8 hours shift 4 people in shift (1 ITF, 1 laser/injection, DAQ, 1 learner) 12 on call sub-system experts central building closed, remote control

34. Caltech, February 12th34 Sensitivity evolution during e-runs

35. Caltech, February 12th35 Lock robustness during e-runs E0 1 (local ctrl fail) 98% ~ 51 h E1 1 (local ctrl fail) 85% ~ 27 h E2 3 (2 ctrl software, 1 vacuum) 98% ~ 41 h E3 4 (1 ctrl software, 3 ctrl tuning) 98% ~ 40 h E4 4 (2 ctrl software, 2 injection) 73% ~ 14 h Run #losses (in ‘normal’ operation) duty cycle longest lock Normal operation = no experiments, no special conditions, no calibration

36. Caltech, February 12th36 Data acquisition during e-runs 20 kHz 2 writing processes in paralles ~ 4 Mbytes/sec 1 Tbyes/e-run 3 kind of data streams: 20 kHz frames 50 Hz Trend (1 Hz)

37. Caltech, February 12th37 Run overview – E4 (July 2002) calibration and other special investigations ~ 7 hours « stable » operation ~ 61 h 30’ calibration ~ 3 h 30’

38. Caltech, February 12th38 Duty Cycle – E4 Normal operation ~ 61h 30’ Locked ~ 42h 20’  Duty cycle ~ 73 % 6 streams with CITF locked longest (5) ~ 14h 30’ shortest (3) ~ 55’ 6 5 2 3 1 4 Duty cycle limited by lock acquisition problems of retroreflected light from ITF to injection system

39. Caltech, February 12th39 Investigation groups Sources of lock losses Suspension motions Angular drifts Output mode-cleaner Calibration Angular noise Seismic noise Acoustic noise Noise gaussianity/stationarity Glitches Lines identification Injection system noise

40. Caltech, February 12th40 Lock losses study - example burst in the local controls of IB 1 sec

41. Caltech, February 12th41 Offset = 1 ·10 -11 Rms = 9 ·10 -12 Offset = 4 ·10 -14 Rms = 1 ·10 -12 Locking accuracy

42. Caltech, February 12th42 Conclusions - sensitivity Solutions for frequency noise Replace MC suspension Add “common mode” loop Solution for alignment noise automatic alignment filtering of high frequency noise

43. Caltech, February 12th43 Conclusions - I Technical choices validated superattenuators “out of vacuum” local controls with CCD cameras digital control chain output mode-cleaner and detection system Lot of experience E-runs program very useful for detetector characterisation

44. Caltech, February 12th44 Virgo Planning Now: large mirror installation vacuum leak tests new MC suspension local control improvements

45. Caltech, February 12th45 Mirror installation I

46. Caltech, February 12th46 Mirror installation II