Homodyne (DC) detection experiment at the 40 meter lab

1. Homodyne (DC) detection experiment at the 40 meter lab Alan Weinstein for the 40m group and Advanced Interferometer Configurations Working Group of the LSC January 14, 2005

2. Homodyne Detection for AdLIGO • LIGO-1-like sensing, based on beats between RF sidebands (LO) and carrier, is subject to considerable noise from the LO, which is not arm-filtered and which has additional oscillator phase noise. • Baseline optical configuration for AdLIGO is a Fabry-Perot Michelson with both power- and signal-recycling; the signal cavity will be detuned to produce enhanced sensitivity at some GW frequency > 0 • The detuned signal cavity means that the RF sidebands will be unbalanced; RF LO noise couplings will be worse • Using arm-filtered carrier light as the LO in AdLIGO (DC or homodyne detection) may result in improved noise due to less noisy LO (if implemented well). • Also slightly improved quantum noise (see next slide, from Buonanno and Chen). • AdLIGO will want an output mode cleaner (OMC) to reject junk light (not containing GW signal information; due to Michelson contrast defect, mode mismatching, etc) at the asymmetric port (AP). • This OMC can also reject the noisy RF sidebands, leaving only arm-filtered carrier and GW signal sidebands. An OMC which passes the RF sidebands is more complex. • DC detection requires seismic and acoustic isolation of the OMC and DC photodetector. • A pickoff before the OMC will permit standard RF detection as well, to aid in lock acquisition and to provide a backup for DARM control and GW signal extraction.

3. Quantum noise comparison Buonanno &Chen: Quantum noise + thermal noise • parameters optimized for NS-NS binaries • single optimal demod phase chosen for RF • SNR for homodyne higher than heterodyne by • 5% for spherical mirrors • 10% for mexican hat mirrors

4. Goals for 40m • The 40m is developing a full engineering prototype of the AdLIGO optical configuration. • We expect to lock the full DRFPMI in the next few months, and establish a reliable lock acquisition procedure for AdLIGO. • Then, establish expected response to large-amplitude GWs. Then, study noise. • This is a good time, and place, to test homodyne detection. • Initial goal is to implement a DC detection optical beamline, sensing and control electronics. • This can be done on existing seismic stack in (small) vacuum chamber, already in place for this purpose. • Then, establish expected response to large-amplitude GWs, and first look at noise. • Establishing low-noise sensing is another matter. • As everyone knows, finding and eliminating technical noise sources to the point where only fundamental noise sources remain, is a long, difficult, open-ended task. • Initial goal for 40m DRFPMI prototype with RF detection includes only an initial look at noise sources, not an exhaustive battle with them. • It is not clear how much can be learned about limits to noise in homodyne detection, in the 40 meter environment.

5. Elements of DC detection beamline • Currently, light exiting the signal mirror goes straight out of the vacuum chamber and into an RF detection beamline (AP1) on a nearby optical table (mech shutter, EOS, HFRFPD, DDRFPD, QPD, CCD camera; see next slide). • Plan is to pick off some/most of that light, and send it into DC detection system. • In-vac, seismically-isolated beamline (AP2): • Mode matching telescope (very short, off-axis mirrors, remote control focus) • Two steering mirrors (PZT, servo-controlled) • Output mode cleaner (3- or 4-mirror, fixed spacer, PZT-mounted mirror) • DC detection photodiode (in a pressurized can?) with baffling. • OMC reflected beamline (on optical table in air): • IF the OMC is to be locked using reflected sidebands, eg, 166 MHz, then we need the usual: Mech shutter, electro-optic shutter and power PD, LSC RFPD, QPD, WFSs, CCD camera… • If the OMC is to be locked using dither locking, need DC PD instead, either in reflection or transmission.

6. AP1 ISC System Mike Smith 2001 “straw man” design; implemented system is different! Only two LSC RFPDs; no WFSs. Power monitor Video camera Spectrum analyzer Mechanical shutter EO shutter

7. Parameter Requirement Actual Wavefront distortion < 1 wave @ 1064 nm < 0.3 wave @ 1064 nm Main beam sample fraction 0.01 0.01 WFS1, Guoy phase 1 Quad photodiode, 177.3 MHz QPD, 177.3 MHz WFS2, Guoy phase 2 Quad photodiode, 177.3 MHz QPD, 177.3 MHz LS, RF photodiode 177.3 MHz 177.3 MHz Fast beam shutter Yes EO shutter Mechanical beam block Yes Uniblitz Video camera Yes Watek Reflected power monitor photodiode yes OMC Reflected Beam Sensing System Mike Smith 2001 “straw man” design; not implemented. Mechanical shutter Power monitor Video camera EO shutter

8. Parameter Requirement Actual Spot size of OMC beam 0.37 mm Position steering +/-0.37 mm +/- 0.8 mm Divergence angle of OMC beam 0.00092 rad Angular steering 0.00092 rad +/- 0.004 rad Resonant frequency 3500 Hz Angle sensing Internal strain gage OMC Beam Steering Mike Smith 2001 “straw man” design; not implemented yet. Existing in-vac seismically isolated optical table Mike built a compact, monolithic MMT for our IMCIFO optical train, using expensive, long-lead-time off-axis-parabolic mirrors. Could use same design for IFOOMC. Ditto, for in-vac PZT steering mirrors. from SRM

9. Parameter Requirement Actual Clear aperture M1, mm 13 25 Clear aperture M2, mm 3 19 Input beam waist radius, mm 3.027 3.027 Output beam waist radius, mm 0.37 0.369 Power coupling error <0.05 < 0.0001 Wavefront distortion <0.2 Transmissivity across clear aperture > 99.8%, ion beam coating Magnification 0.23 0.23 OMC Mode Matching Telescope Mike Smith 2001 “straw man” design; not implemented.

10. Parameter Requirement Actual AP2 power ratio 0.005 0.005 GWS photodetector frequency response DC - 10 KHz DC - 10 KHz Seismic velocity of GWS photodetector TBD TBD AP2 ISC Optical Train Mike Smith 2001 “straw man” design; not implemented. Video camera GW DCPD Will be in-vac! power sensor EO shutter from OMC

11. Many interesting questions • First and foremost: given the 40m environment, is the entire task worth pursuing? What will we learn? • Assuming it is worth pursuing and that we will learn a lot… • Is the overall scheme sensible? Other/better ways? • Predict expected noise sources / couplings: • laser frequency and intensity noise • Mach-Zehnder phase and amplitude noise • MICH/PRC/SRC/DARM servo offset/noise • OMC servo noise Using methods of Camp et al, Mason, Adhikari and Ballmer,… Kentaro Somiya accounts for full quantum noise effect, ala Buonanno&Chen. Must pursue in parallel with design and construction of engineering prototype of homodyne detection. • With DC detection, laser intensity noise feeds directly in to the GW signal;it must be suppressed sufficiently well. Need DC photodiode on input light for intensity stabilization to AdLIGO spec. • OMC: 3-mirror or 4-mirror?

12. OMC – 3-mirror or 4-mirror? 3-mirror PMC-like? (incident angle ~42º,small astigmatism) 4-mirror Keita-design? (incident angle ~30º,larger astigmatism, half as many HOM accidental resonances)

13. Many interesting questions / 2 • Locking the OMC: PDH reflection locking? Dither-lock? • Will using the RF sideband (+166 MHz) to reflection-lock imprint the very noise we are trying to get away from, onto the transmitted carrier light, thus defeating the primary purpose of homodyne detection? • Is dither-locking simple and straightforward? At 8192 Hz or 4096? (We routinely ditherlock our Michelson at 120 Hz). • To get the arm-filtered DC light out the AP: • offset lock the arms? • offset-lock the MICH? From seminal paper by Camp/Yamamoto/Whitcomb/McClelland (J. Opt. Soc Am 17, 120)

14. Better stop here • Following slides are Peter Fritschel’s talk from LSC meeting at Hannover, Aug 2003

15. DC Readout for Advanced LIGO P Fritschel LSC meeting Hannover, 21 August 2003

16. Heterodyne & homodyne readouts • Heterodyne: traditional RF modulation/demodulation • RF phase modulation of input beam • Lengths chosen to transmit first-order RF sideband(s) to anti-symmetric output port with high efficiency • Initial LIGO: RF sidebands are in principal balanced at AS port • AdLIGO: with detuned signal recycling, one RF sideband is stronger than the other • RF sideband(s) serve as local oscillator to beat with GW-produced field • Signal: amplitude modulation of RF photocurrent • Homodyne: DC readout • Main laser field (carrier) serves as local oscillator • Signal: amplitude modulation of GW-band photocurrent • Two components of local oscillator: • Field arising from loss differences in the arms • Field from intentional offset from dark fringe

17. Why DC readout now? • Requires an output mode cleaner • Technical amplitude noise on the total output power would be much too high • GW-band laser power noise has to be extremely low in any case • Radiation pressure noise from unbalanced arms • Anticipated technical noise difficulties with unbalanced RF sidebands • Anticipated sensitivity advantage: lack of non-stationary noise increase

18. Technical noise sensitivity

19. Implementation comparison • Output mode cleaner • RF: • long mode cleaner, like the input MC • individually isolation mirrors • DC: • Short, monolithic: 0.5-1 m (similar to a pre-mode cleaner) • Requires a bit more beam size reduction • Photodetection • Both • High quantum efficiency, low back-scatter • In-Vacuum • Low power, < 100 mW • RF • Low capacitance, 180 MHz ?

20. Quantum noise comparison Buonanno &Chen: Quantum noise + thermal noise • parameters optimized for NS-NS binaries • single optimal demod phase chosen for RF • SNR for homodyne higher than heterodyne by • 5% for spherical mirrors • 10% for mexican hat mirrors

21. Making the DC local oscillator • Two components • Carrier field due to loss differences (not controllable?) • Carrier field due to dark fringe offset (controllable) • Loss mismatch component • Average arm round trip loss: 75 ppm • Difference between arms: 30 ppm • Output power due to mismatch: 1.6 mW • Detection angle, β • Tuned by adjusting fringe offset • Broadband (NS-NS) optmimum: • Fringe offset power: approx. 0.3 mW • Differential arm offset: approx. 1 pm • Can tune from 0 to 80 deg with 0-100 mW of fringe offset power fringe offset β Loss mismatch

22. Summary • Sensitivity of a DC readout is a ‘little better’ than that of an RF readout: 5-10% • Technical noise sensitivity favors DC readout • Technical implementation favors DC readout • Proceeding with DC readout as baseline design • Output mode cleaner design: • Vibration isolation • Sensing and control: length and alignment • Could we switch later to RF readout? Hardest part would be changing the output mode cleaner (to pass the RF sidebands)