The polarizationbased collimated beam combiner and the proposed
The polarization-based collimated beam combiner and the proposed NOVA fringe tracker (NFT) for the VLTI Jeffrey A. Meisner, Sterrewacht Leiden PRESENTER Walter J. Jaffe, Sterrewacht Leiden Rudolf S. Le Poole, Sterrewacht Leiden & TNO Silvania F. Pereira, Technische Universiteit Delft Andreas Quirrenbach, Landessternwarte Heidelberg David Raban, TNO Science and Industry Amir Vosteen, TNO Science and Industry Outline of paper Interferometric beam combination and the problem of “photometric crosstalk” The polarization-based collimated beam combiner topology The NOVA Fringe Tracker, designed using that concept Laboratory demonstration and results implementing that topology
Basic on-axis beam combiner using partially transmissive reflector (beamsplitter) nominally T=R=50% Subtraction of complementary detected outputs from beamcombiner yields estimate of visibility However unbalanced combiner (T =/= R) with unequal photometric levels (|EA|2 =/= |EB|2) leads to photometric crosstalk in interferometric determination! EA EB E 1 Detector I 1 E 2 Detector I 2 + - |V| cos( ) + photometric crosstalk []=s[] E 1 E 2 EA EB
The photometric asymmetry coefficient Electric Fields: []=[ ][] E 1 E 2 s 1 A s 1 B s 2 A s 2 B Intensities: [ ] =[ I 1 I 2 EA EB ][] |s 1 A|2 |s 1 B|2 |s 2 A|2 |s 2 B|2 IA IB + Interferometry If lossless beamsplitter: =[ ][] R T T R IA IB + Interferometry where R = ½ (1 + ) and T = ½ (1 - ) Or in general: = ( |s 1 A| / |s 2 A| - |s 1 B| / |s 2 B| ) / ( |s 1 A| / |s 2 A| + |s 1 B| / |s 2 B| ) Visibility estimator I 1 – I 2 inevitably includes a photometric crosstalk term = (IA – IB)
Combatting photometric crosstalk - I Naive approach: Force =0 by making the beamsplitter's T = R = ½ but: It will not generally be constant over wavelength It will almost always be different between the polarizations Moreover this is difficult to do in the first place! +. 6. 4. 2 0 -. 2 -. 4 -. 6 For instance, the VINCI (fiber) beam combiner, regularly adjusted for maximum fringe contrast, had a very unpredictable (almost never near zero!) as plotted over 3 years:
Combatting photometric crosstalk -II Using photometric pick-offs: Split off a fraction m of the incoming optical powers, and add them in the correct proportions to the interferometric outputs so that the effects of IA and IB are cancelled. EA EB m IA E 1 m IB E 2 Detector I 1 + -
Using photometric pick-offs -II Drawbacks: • Photometric monitoring beams rob power from the interferometric channels • Added photometric corrections IA and IB contain detector noise, added to resulting visibility determination • Want cancellation in both polarizations and at all wavelengths Even using the optimum m (left graph) there is a substantial increase (right graph) in the noise of the visibility estimate due to (1) and (2) Optimum m Noise increase (power) 6 4 2 1 =. 2 . 4
Combatting photometric crosstalk -III -- III Modulation of the OPD (The most common solution!) ABCD phase stepping, or: Scanning of fringe packet, etc Then the visibility appears as an AC signal on the photodetector. Just ignore the DC (and low frequencies) then. So measure |E 1+ E 2 exp(j )|2 as a function of (t) OPD Modulator A B C D (for instance) Coherent Detection Estimate of complex V
Modulation of the OPD -II Drawbacks: • Requires more than one detector readout to measure a visibility • Since the photometry (and OPD!) is changing due to the atmosphere, this fluctuating component leaks into the result. • In order to reduce that effect, a faster readout speed is required, reducing sensitivity in a NIR instrument. Example: the incoherent power spectrum (top) and coherently integrated power spectrum (bottom) of 4 consecutive VINCI observations (14 Aug. 2001) of eps sco at different framerates: Good! LF Noise More White Noisier 590 Hz: too slow. Spectrum is broadened by atmospheric OPD Just right! (For this VERY bright star!) 3384 Hz: too fast. Detector noise enhanced due to short exposure
The Second Generation Fringe Tracker for the VLTI, planned by ESO to supercede PRIMA Requirements: Accept beams from 4 or 6 telescopes (not just 2) either from science target itself or from off-axis reference star Measure phase for control of VLTI delay lines (fringe locking) for long coherent exposures & phase referenced imaging Sensitive to group delay for dispersion control and fringe-jump detection Tolerant of wavefront and photometric fluctuations Tolerant of different (possibly small) visibilities on some baselines Possibility of combining AT (1. 8 meter telescope) with UT (8 meter telescope), 20 x brighter! Rapid update rate possible (up to 2 KHz) with best possible limiting sensitivity (of course!) Last 4 requirements are challenged by concerns arising from photometric crosstalk and the inadequate solutions to it.
The NOVA Fringe Tracker (NFT) Result of one of 3 phase A studies to propose to ESO a concept for a second generation fringe tracker for the VLTI B e ams Local Polariz a Revers tion ers Switch yard Polariz Recom ation bining Stage Spec Anal tromet er Ca yzer m Dete era ctor
The NOVA Fringe Tracker (NFT) Main design features Wideband interferometry over 1. 2 - 2. 4 microns simultaneously Spectrally resolved detection over 4 - 7 pixels (reconfigurable) Always at least 2 pixels over K band for dispersion detection Single beam combiner, all wavelengths fixed to same reference plane Combines up to 4 (6) telescopes pairwise over 4 (6) baselines No spatial filtering (is optional) for highest limiting sensitivity No fiber injection loss, hassle Wide (effective) visibility fluctuations tolerated Two-phase interferometric detection with fringe-locking OPD corrections based on Im{V} No need to use ½ the photons for measuring Re{V}, doubles sensitivity. Correction to OPD supplied every frame, no OPD modulation required Based on the Polarization-Based Collimated Beam Combiner topology Balanced beam combination, photometric crosstalk rejected st Can track on low |V| sources, including stars past 1 visibility null Combining AT with UT (20 x brighter!): no problem
The Polarization-Based Collimated Beam Combiner: a solution to the problem of photometric asymmetry • Combines beams pairwise. • Each telescope's light is split by polarization, to be combined with 2 other telescopes. • Each combination produces 2 (or more) interferometric outputs based on balanced combination: visibility estimate is immune to photometric crosstalk Requires 3 essential stages: 1 Beam 2 Beam 3 Beam 4 H V Polarization Reversers (even channels only) p s p 2 Polarization Recombined Beam Polarization Recombinaton Stage 3 Detectors Polarization Analyzer @ 45 o s 2 -phase detection configuration shown Visibility Estimate + -
1 In NFT, Polarization Reversers implemented using (almost) achromatic ½ wave plates. o Even channels: at 45 to reverse polarization o Odd channels: at 0 , no change in polarization (but compensates for material dispersion) All are rotatable so their roles can be reversed VLTI Beams from pick-off mirrors Local Switchyard Half wave plates (polarization reversing in even beams) 1 2 3 4 M 2 mirrors include very short stroke OPD adjustment + piezo, and motors for beam alignment Polarization Recombining Stage (PRS)
Polarization Recombining Stage (PRS) 2 Consists of multiple Polarizing Beamsplitters (PBS) s polarization from telescope M is paired with p polarization of telescope M+1 into same spatial mode but a different polarization (thus a different mode) Thus the “polarization recombined beam” can proceed through various (non polarized) optical elements and both waves are affected identically Only when they finally reach the polarization analyzer are the two waves actually interfered and directed onto 2 (or more) photodetectors Original Polarization H V 2 s 2 p V H 3 s H V 4 s 4 p Telescope 3 p Beams Polarizing Beam Splitters Original Polarization Recombined Beams 1 s 2 p V V 2 s 3 p H H 3 s 4 p V V Important point: After the PRS, the “polarization recombined beams” no longer need to be treated according to interferometric standards. OPD variations/instability, wavefront degradation do not affect the visibility or rejection of photometric crosstalk!
2 6 NFT implementation of Polarization Recombining Stage for 6 telescopes, including end-around channel Based on giant prism block with 2 polarizing beamsplitting surfaces per channel s=H p=V 5 s=V p=H 4 s=H p=V 3 s=V p=H 2 s=H p=V 1 s=V p=H 6 p Constant OPD plane at 45 o 6 s+5 p 5 s+4 p 4 s+3 p 3 s+2 p 2 s+1 p 1 s End-around recombiner (set for M=4 telescopes) 1 s+(M)p
3 Polarization Analyzer produces output beams implementing a balanced beamcombiner. When their powers are detected and differenced, photometric crosstalk is suppressed! Telescope 1 Telescope 2 1 s 2 p Detector A Detector B 45 o Rotation of Coordinate System A = 1 s + 2 p Polarization Analyzer in rotated system IA + - B = 1 s - 2 p Detectors Differential Amplifier IB Implementation in NFT: Wollaston Prism Camera Lens Detector Array A = 1 s + 2 p IB B = 1 s - 2 p IA
3 NFT Backend Implementation M 5 Channel 1 2 3 4 5 6 blue PRS M 4: off-axis paraboloids red Possible layout of 2 spectra from each channel on 40 micron pixels of PICNIC detector. All 6 polarization recombined beams pass through the same optics after diverging from their foci at the mask Mask with ~. 5 mm holes at intermediate focus (Not a spatial filter) Wollaston prism (polarization analyzer) Spectral Prism (zero-deviation) Spectral resolution variable by shifting Array Detector
3 Additional options for the polarization analyzer/detectors (The NFT just uses 2 -phase detection --> sin( ). ) But for a visibility measuring interferometer. . . Quadrature detection of interference: Polarization Recombined Beam s PRS Splitter p ¼ wave plate Polarization Analyzer @ 45 o 0 o detection 180 o detection 90 o detection 270 o detection 3 -phase detection (a little less detector noise): Special beamsplitting coating: B polarization: T=100% A polarization: R=2/3 T=1/3 plate at 45 o w/r/t A&B A = 1 s + 2 p 240 o detection B = 1 s - 2 p Detectors Detector 0 o detection 120 o detection
Laboratory setup of the polarization-based collimated beam combiner at TU Delft to demonstrate concept and measure performance achieved
Coherent source supplied by He-Ne laser (polarized at 45 o), split into 2 beams (“ 2 telescopes”) to be polarization recombined and interfered using PBS and 2 photodiode detectors (hidden inside rotating assembly) He-Ne Laser Mirror on piezo actuator Polarization Recombining PBS Analyzer: PBS with 2 photodiodes in rotatable assembly
Added “photometric noise” (incoherent) into each beam path from 2 red laser diodes modulated at 250 Hz and 1000 Hz respectively. Laser diodes (pulsing) Beamsplitters (non polarizing) to inject laser diode beams into beam paths Scope trace from a single photodetector (unbalanced) showing “photometric noise” from pulsating laser diodes Polarization Recombining PBS Analyzer & detectors
Laser Diodes (source of “photometric interference”) He-Ne Laser LD Polarizing Beamsplitters ATMOSPHERE Detector assembly rotates about beam axis, nominally at 45 o Analyzer PBS Photodiode ND Filter T = 3% LD P St iez ac o k CORR Hot turbulent air from hair dryer Scan (for testing) + Control loop filter integrator Lock HV Amp PRS Detectors Photodiode preamps & differencing circuit
1. Experimental Results. . . As expected, the “photometric crosstalk” contribution from the pulsing laser diodes, seen in both of the photodetector outputs (left) are rejected when subtracted, leaving only the actual interferometry (small sine wave, right trace) due to scanning of the piezo (visibility was reduced by intentionally misaligning interferometric beams) A+B (photometry only) in red A-B (photometry rejected) in white • Routinely achieve >>100: 1 photometric rejection (need to block one interferometric beam to measure!) thus <<. 01 Rejection stable over time (weeks, if not touched)
2. Experimental Results. . . In fringe tracking mode, run error signal (~ Im(V) ) from 2 -phase interferometric detector into filter-integrator driving the piezo amplifier. With no OPD disturbances, residual noise from interferometry is too small to measure, <<. 1 radian (= 10 nm) 30 Hz square wave added to error signal, causes ~20 nm change in equilibrium tracking point. Is cancelled by an equal and opposite interferometric phase detection (below). OPD residuals from interferometric output (top trace) <. 1 radians with hair dryer running (plus induced square wave) With hair dryer blowing hot turbulent air in 15 cm beam path, piezo voltage tracks the induced OPD well. Range ~ 5 wavelengths ( =633 nm). Normally no loss of tracking. A fringe jump is easily noticed after turning off hair dryer since the mechanical stability <<
3. Experimental Results. . . Inserted neutral density filter (T=. 03) in one interferometric beam, simulating interference between a VLTI UT (8 meter telescope) and an AT (1. 8 meter telescope). Still with laser diode mixed with each beam, pulsing at 250 Hz and 1000 Hz and hair dryer creating “atmospheric turbulence. ” No noticeable change in tracking performance Demonstration of insensitivity of polarization combined beam to modest optical disturbances which would be impossible if applied to one beam before the PRS Inserted a wine glass in the 15 cm space between the PRS and the analyzer (beam passing through both sides of the glass) and wiggled it around. No noticeable effect on tracking, no fringe jump during entire period.
The End
End of presentation Extra slides follow. .
2 Alternative PRS implementations: Original Polarization H V 2 s 2 p V V V H 3 s H H H V 4 s 4 p 3 p “Polarization Recombined Beams” V V Dispersion compensation for equal path length in glass Using Wollaston Prisms s=H p=V s=V p=H s=H p=V “Polarization Recombined Beams”
- Slides: 28