1. (Some) noise hunting issues at Virgo
Gabriele Vajente
11th ILIAS WP1 meeting
Hannover, January

2. Summary Searching for a good operating point
ITF sensitivity and longitudinal offsets
Alignment noise projections
Implementing HACR at Virgo

3. Searching for the good operating point…

4. Virgo scheme
Beams
B1 = dark fringe B2 = ITF reflection B5 = BS pick-off B7 = north arm transmission B8 = west arm transmission Scanning F-P on B1
DOFs
DARM = differential long arms CARM = common long arms MICH = short michelson PRCL = recycling cavity

5. How to define a good O.P.
The ITF is at resonance, meaning powers are min/max at the locking point
In particular
Carrier resonates in PRC: B5_DC is maximum
Fringe is dark: B1_DC is minimum
Sidebands resonate in PRC: B5_2f_ACq is maximum
A consequence
Sidebands are well balanced at every port
BS pick-off beam demodulated at 2W gives an estimate of sidebands power into PRC

6. The old story
But changing only MICH offsets is not enough to reach a good operating point
Evidence that MICH offsets was not good (November 10th)
DF power
PRC power
DF ratio
PRC sidebands
ITF reflection
North transmission
X axis = offset on MICH

7. The idea MICH and PRCL loop are strongly related
Try to explore MICH-PRCL space by changing both offsets
Spiral experiment
Change simultaneously the two offsets
Error signals are reconstructed using Pr_B2_3f_ACp = ITF reflection demodulated at 3W Pr_B5_ACq = BS pick-off demodulated
Add electronic offsets to these two signals
Need to wait 2 hours after lock to let the ITF “warm up”
B5_d2_ACq
B2_3f_d1_ACp

8. Spiral experiments First “historical” one (November 24th 2006)
Proved the technique works
Second experiment centered around ”normal” locking point
Identified a new point, one night lock on it
Third experiment, centered around the new point

9. Sidebands imbalance at DF
First experiment
Dark fringe ratio
Sidebands inside PRC
Sidebands imbalance at DF
OFFSET ON B5_ACq [ADC counts]
Power inside PRC
OFFSET ON B2_3f_ACp [ADC counts]

10. Conclusions of the first experiment
Moving to lower right corner:
Fringe is darker
B5_2f_ACq is higher
Sidebands are more balanced
But B5_DC is lower
Left the ITF locked all the night (from 23.00LT 28th to 8.00LT 29th) with new offsets
B2_3f_d1_ACp = 800
B5_d2_ACq = -8000
Reached a bit more than 4 Mpc
Never obtained with B2_3f

11. Second experiment
Second experiment
(1100,-11000)

12. Second experiment
(1600,0)

13. Second experiment

14. Second experiment
(1250, -6000)

15. Conclusions of the second experiment
Three good candidates identified
Darkest fringe at (1100,-11000)
Maximum sidebands power in PRC at (1600,0)
Highest horizon at (1250, -6000)
The idea was to test all this points, measuring a sensitivity
But only the first one was briefly tested so far

16. Improvements in sensitivity
NSNS Horizon (opt)
PRC power
B2_3f offset
B5 offset
Loop UGFs
Some lock signals
DARM
PRCL
MICH

17. Improvements in sensitivity
Dark fringe
Reconstructed h

18. Conclusions on the operating point
The “normal” locking position is not the best operating point
Changing offsets it is possible to improve sensitivity quite a lot (horizon from 3 to 4.5 Mpc)
Still some tuning to do…
During WSR7 the ITF was run with an offset on B2_3f only
What to do next
Check repeatability (alignment issues?)
Offsets on CARM and DARM?
Find a new O.P. and tune parameters

19. Alignment noise projections

20. Measurements of alignment N.P.
Two methods:
Injection of noise (double 0 Hz, double 10 Hz, double 50 Hz)
Injection of lines 13, 72, 119 Hz
Estimation of transfer functions and noise projection for dark fringe

21. Some results / NE tx Coherence DF vs correction with noise
Correction signal with and without noise
Green stars are TF computed with the lines
Trans. Func. Correction to DF with noise
D.F. with and without noise

22. Some results / NI tx Coherence DF vs correction with noise
Correction signal with and without noise
Green stars are TF computed with the lines
Trans. Func. Correction to DF with noise
D.F. with and without noise

23. Some results / BS tx Coherence DF vs correction with noise
Correction signal with and without noise
Green stars are TF computed with the lines
D.F. with and without noise
Trans. Func. Correction to DF with noise

24. Transfer function estimation with lines
Coherence at lines (PR is not good)
Trans. Func. at lines (PR is not good)

25. Noise projection / tx (WSR7 sensitivity)

26. Noise projections / ty (WSR7 sensitivity)

28. Implementing HACR at Virgo

29. Status Started developing VirgoHACR Very good performances
C code based on HACR algorithm
Full interface with Virgo Data Distribution via FdIO library
Output is a text file with events for the moment
Will be a MySQL database, to better interface with GEO report system
Very good performances
1 s of data, 1 20 kHz analyzed in about 10 ms of machine time
Limited by data access
Some test peformed

30. Tests with simulated data…
Map of short FFTs, white noise with bursts

31. Tests with simulated data…
All events are found, parameters are reconstructed well

32. Test with simulated data…
Accuracy in time and frequency reconstruction

33. Starting tests on dark fringe signal…
Short FFTs map

34. Starting tests on dark fringe signal…
Mean FFT map

35. Starting tests on dark fringe signal…
Threshold crossing

36. Conclusions on VirgoHACR
Code seems to work fine
Need some time to tune parameters for dark fringe
Still to do
Optimize parameter reconstruction
Output to database
Integration with GEO reports builder to create automatic reports