Noise Projections: S5 Night and Weekend

1
Noise ProjectionsS5 Night and Weekend

• Joshua Smith
• for the GEO 600 team

2
Sensitivity now vs. last GEO mtng.
3
Noise projections reminder
Noise projections show how noise N shows up in
detector output P.
Noise Meas. Point
Det. Output
0) Identify a noise N that could couple to P. 1)
Dominate noise present in both N and P. Measure
transfer function (TF) from N to P. 2) Measure
ASDs of N and P. 3) Multiply N ASD by TF to form
projection
4
Current noise projections status

• 15 noises routinely projected
• Feedbacks
• Laser intensity (RIN)
• Oscillator noises
• Each involves 100s injection of random noise to
  measure TFs
• In S4 the TFs were tracked using calibration
  lines
• Currently commissioning, repeated measurement
  necessary 1/month
• Could do tracking in S5 if thought necessary

5
New projections to h

• Final detector output, h, has best SNR for strain
  signal.
• Its produced by correcting for optical gain,
  subtracting Michelson long. FB, and optimally
  combining the two output quadratures, P and Q.
• See Martins talk
• Projections to P, Q not simple to interpret with
  respect to h.
• In past, noise injections meant no h.
• Can now fix calibration parameters, allows direct
  measurement of TFs to h.

6
Transfer functions to h Feedbacks
1/f4 15
1/f2
Signal-Recycling Long FB to h, hP, hQ
MI Autoalign. Tilt FB to h
7
Transfer functions to h RIN
E-arm RIN to h, hP, hQ
Laser input RIN to DRMI ports
8
Projections February 16th
9
Current sensitivity vs. Feb projs
10
Other examples beam pointing

• Vary tilt of output beam
• measure tilt using Position Sensor
• measure TF from Pos. sensor to h

11
Other examples beam pointing 2
Proj.
Beam pointing to h
??
SNR of pos. det. too low above 100Hz
12
Looking forward noise

• Further reduction of noise
• Shot noise end
• Remove power-up glitches, increase laser power
• Feedback end
• More aggressive filtering, diagonalize feedbacks,
  cleaner error signals
• Intermediate range
• Reduce environmental couplings, scattered light,
  etc. See Stefans talk(s).
• Keep projecting new things
• For what cant be reduced
• Veto (see Ajiths talk) or subtract known noises.

13
The end !
14
Shot Noise Model

• Further plans
• Another DC current estimate
• Modulate power and fit scatter plot (noise vs P)
  
• Include Optical Spring in curve