Comparison of tuned and detuned Signal-Recycling. Stefan Hild for the GEO-team

Transcription

1 Comparison of tuned and detuned Signal-Recycling Stefan Hild for the GEO-team Stefan Hild 1 ILIAS WG1 meeting, Cascina, November 2006

2 Stefan Hild 2 ILIAS WG1 meeting, Cascina, November 2006 Signal-Recycling in short An additional recycling mirror (MSR) at the dark port allows: enhancing the GW signal shaping the detector response MSR Two main parameters: Bandwidth (of the SR resonance) broadband narrowband Tuning (Fourier frequency of the SR resonance) tuned detuned

3 Stefan Hild 3 ILIAS WG1 meeting, Cascina, November 2006 SR-Tuning in GEO so far For various reason we are not able aquire lock for tuned SR. Lock takes place at a detuning of a few khz. Afterwards the aquisition detector is tuned to ist operation point. So far tunings between 5kHz and 250 Hz had been realized. Downtuning in steps of 25 Hz, 6 Parameters need to be adjusted: SR frequency SR gain SR phase MI gain MI phase MI autoalignment gain

4 Motivation (1): Detuned SR complicates noise couplings and TFs Stefan Hild 4 ILIAS WG1 meeting, Cascina, November 2006 Frequency noise coupling to h(t) Laser intensity noise coupling to h(t) In a detuned detector TF may become complicated due to interaction and different resonance conditions of various sidebands.

5 Motivation (2): Sideband picture for tuned and detuned SR Stefan Hild 5 ILIAS WG1 meeting, Cascina, November 2006 Frequency of the light

6 Motivation (2): Sideband picture for tuned and detuned SR Stefan Hild 6 ILIAS WG1 meeting, Cascina, November 2006 Carrier (GW-sidebands) Frequency of the light

7 Motivation (2): Sideband picture for tuned and detuned SR Stefan Hild 7 ILIAS WG1 meeting, Cascina, November 2006 Michelson CTRL SB Michelson CTRL SB Frequency of the light

8 Motivation (2): Sideband picture for tuned and detuned SR Stefan Hild 8 ILIAS WG1 meeting, Cascina, November 2006 Signal recycling CTRL SB Signal recycling CTRL SB Frequency of the light

9 Motivation (2): Sideband picture for tuned and detuned SR Stefan Hild 9 ILIAS WG1 meeting, Cascina, November 2006 Comb of Equidistant SR resonances Frequency of the light

10 Motivation (3): Sideband picture for detuned SR (510 Hz) Stefan Hild 10 ILIAS WG1 meeting, Cascina, November 2006 MSR is locked onto upper SB => each SR resonances is 510 Hz left of a FSR of SRC. Upper SR-SB resonant, lower SR-SB off resonance. Lower MI-SB nearly resonant, upper MI-SB far off resonance. Upper and lower GW signal see different completely different resonance condition.

11 Motivation (4): Sideband picture for tuned SR Stefan Hild 11 ILIAS WG1 meeting, Cascina, November 2006 Going form detuned (510Hz) to tuned: comb of SR resonances is shifted 510 Hz to the right. Upper and lower SR-SB see the same resonance condition (nearly resonant). Upper and lower MI-SB see the same resonance condition. Upper and lower GW signal SB see the same resonance condition.

12 Stefan Hild 12 ILIAS WG1 meeting, Cascina, November 2006 Motivation for tuned SR There is the possibility: Get better noise performance due to symmetric sidebands.

13 Stefan Hild 13 ILIAS WG1 meeting, Cascina, November 2006 Phase noise in P originates from big DC in Q We believe that a large contribution of the phase noise in P originates from a large DC signal in the Q quadrature. Therefore it would be nice to reduce the DC signal in Q. (see Josh s talk)

14 Stefan Hild 14 ILIAS WG1 meeting, Cascina, November 2006 Motivation for tuned SR There is the possibility: Get better noise performance due to symmetric sidebands. Less RF amplitude modulation on main photodiode (reduction of potential saturation / nonlinear effects)

15 Stefan Hild 15 ILIAS WG1 meeting, Cascina, November 2006 Saturation in our main photo diode We believe we saw some saturation in the main photodiode due to the large RF amplitude modulation (Q-DC). The signal voltage modulated the bias voltage (5V) by about 20%. Problem was for the moment (higher bias, different circuit). For long term it would be desirable to reduce Q-DC!!

16 Stefan Hild 16 ILIAS WG1 meeting, Cascina, November 2006 Motivation for tuned SR There is the possibility: Get better noise performance due to symmetric sidebands. Less RF amplitude modulation on main photodiode (reduction of potential saturation / nonlinear effects) Get simpler noise couplings and transferfunctions. Get a better understanding of the detector Nice research / demonstration project

17 Stefan Hild 17 ILIAS WG1 meeting, Cascina, November 2006 Signal-Recycling errorsignal SR detuned, lower sideband tuned SR Nominal operating point, SR detuned, upper sideband Derived from an RF modulation-demodulation technique.

18 Stefan Hild 18 ILIAS WG1 meeting, Cascina, November 2006 Why can t we tune down to tuned SR? Tuning is done by changing the SR RF-modulation frequency. Signal amplitude [a.u.] 0 SR Mod = Hz, tuning = 1000 Hz SR Mod = Hz, tuning = 330 Hz Position of MSR [nm] Signal amplitude [a.u.] 0 SR Mod = Hz, tuning = 330 Hz SR Mod = Hz, tuning = 200 Hz SR Mod = Hz, tuning = 0 Hz Position of MSR [nm] When we tune below 250 Hz the controlsignal vanishes / goes cracy.

19 Stefan Hild 19 ILIAS WG1 meeting, Cascina, November 2006 Kicking MSR Kicking MSR in a controlled way: Fast enough that all other loops can t recognize. 4 ms of acceleration and 4 ms of deceleration. Works fine: Jumping to tuned and to the lower SR sideband

20 Stefan Hild 20 ILIAS WG1 meeting, Cascina, November 2006 Sideband picture for tuned and detuned SR From detuned to tuned: CTRL-signals that are generated from carrier need to be adjusted for the different resonance condition of the carrier.

21 Stefan Hild 21 ILIAS WG1 meeting, Cascina, November 2006 Parameter adjustment When jumping to tuned SR you need to change a few parameters / compensate for the pole of the SR cavity: MI long needs an additional differentiator switched in. MIAA gains need to be adjusted. Swap sign of SR_EP. Adjust SR long gain. Simulations done with FINESSE

22 Determine the optical gain for P- and Q-quadrature (TF from diff armlength change to detector output) Stefan Hild 22 ILIAS WG1 meeting, Cascina, November 2006

23 Optical gain for tuned and detuned SR Phase [deg] Optical Optical Gain Gain [V/um] [] Frequency [Hz] P Optical Gain, SR tuned Q Optical Gain, SR tuned P Optical Gain, SR tuning = 550 Hz Q Optical Gain, SR tuning = 550 Hz Frequency [Hz] Matches our expectations: Bandwidth of the TF goes down by a factor of 2. Stefan Hild 23 ILIAS WG1 meeting, Cascina, November 2006

24 Stefan Hild 24 ILIAS WG1 meeting, Cascina, November 2006 Significantly recuded RF AM. The signal from the Q-quadrature is in the tuned case reduced to nearly zero (red). RMS of Q-signal is reduced by a factor of 12!! 1 Time-series plot Amplitude [V] : G1:LSC_MID_EP-Q_HP-debug tuned SR 04: G1:LSC_MID_EP-Q_HP-debug SR detuned (550Hz) Amplitude [V] Time [s] 01: G1:LSC_MID_EP-P_HP-debug tuned SR : G1:LSC_MID_EP-P_HP-debug SR detuned (550Hz) Time [s] Nice for the photodiode: Less potential saturation / nonlinearitis!! What is about phase noise in P-quadrature??

25 Stefan Hild 25 ILIAS WG1 meeting, Cascina, November 2006 MI Oscillator phase noise (OPN) OPN tuned SR: P dramatically reduced, Q significantly reduced.

26 Stefan Hild 26 ILIAS WG1 meeting, Cascina, November 2006 MI Oscillator amplitude noise (OAN) OAN tuned SR: P and Q significantly reduced.

27 Stefan Hild 27 ILIAS WG1 meeting, Cascina, November 2006 Laser intensity noise (LIN) LIN tuned SR: P and Q significantly reduced.

28 Stefan Hild 28 ILIAS WG1 meeting, Cascina, November 2006 Laser frequency noise Laser freq noise tuned SR: P and Q worse (but structure slightly simpler).

29 Stefan Hild 29 ILIAS WG1 meeting, Cascina, November 2006 Summary of noise couplings OPN tuned SR: P dramaticcaly reduced, Q significantly reduced. OAN tuned SR: in both quadratures clearly reduced. Laser frequency noise tuned SR: worse in both quadratures. Laser intensity noise tuned SR: P and Q significantly reduced. In general: The size of TFs for the noise couplings are reduced on average and the structure of the TFs look a bit simpler.

30 Calibration to starin sensitivity (frequency domain) Stefan Hild 30 ILIAS WG1 meeting, Cascina, November 2006

31 Stefan Hild 31 ILIAS WG1 meeting, Cascina, November 2006 Sensitivity: tuned vs detuned (550Hz) SR : G1:DER_DATA_HP fs = 8192 : 100s from :00:00 - nfft=8192, nolap=0.50, enbw=1.5, navs=199 Tuned SR, Pwr = 75 deg, HP : G1:DER_DATA_HQ fs = 8192 : 100s from :00:00 - nfft=8192, nolap=0.50, enbw=1.5, navs=199 Tuned SR, Pwr = 75 deg, HQ STRAIN [1/sqrt(Hz)] STRAIN [1/sqrt(Hz)] Frequency [Hz] Frequency [Hz] No improvement in sensitivity (not surprising, as we are not limited by shotnoise at low frequencies and we did not optimize the detector / parameters for tuned SR)

32 Stefan Hild 32 ILIAS WG1 meeting, Cascina, November 2006 Noise projections for a time of S5 So far we don t understand why Q got worse at high frequencies for tuned SR. (?)

33 Stefan Hild 33 ILIAS WG1 meeting, Cascina, November 2006 Summary Demonstrated controlled jumping to tuned SR. No sensitivity improvement for tuned SR at the moment (spend no time for optimization). Some of the noise couplings are significantly reduced for tuned, but their structure is still complicated. In tuned case the RF AM is reduced by a factor of 12 Outlook: To gain understanding of the noise couplings: seems important in order to speed up comissioning of GEO is essential for good design of next generation of instruments

34 E n d Stefan Hild 34 ILIAS WG1 meeting, Cascina, November 2006

35 Stefan Hild 35 ILIAS WG1 meeting, Cascina, November 2006 Acquisition of Dual Recycling MHz MHz