The LTP interferometer and Phasemeter

Transcription

1 The LTP interferometer and Phasemeter Gerhard Heinzel, Albert-Einstein-Institut Hannover, for the TAMA Symposium, Osaka 2005/02/18. 1

2 2

3 3

4 4

5 Heterodyne Mach-Zehnder interferometer Laser f0 AOM f1 AOM f0+f1 f0+f2 φ PD1 PD2 f1 f2 = f het f2 It has constant sensitivity over a range of > ±100 µm. The heterodyne frequency f het is a few khz (1.6 khz in the EM). Photocurrent φ time 5

6 Interferometer budget optical pathlength noise [m/ Hz] pm/ Hz SMART2 mission goal 9 pm/ Hz interferometer budget 1 pm/ Hz each interferometer contrib. µrad/ Hz frequency [Hz] interferometer phase noise [rad/ Hz] Note the new interpretation. As compared to some earlier versions, nothing has changed for the interferometer budget, which is, however, now a factor of nearly 20 below the mission goal. The frequency dependence of all interferometer-related budgets is y(f) = y(30 mhz) 1 + ( 3 mhz f and all budgets in the following are given at 30 mhz (such as 9 pm/ Hz for the interferometer). ) 4, 6

7 4 interferometers: Frequency Fiber Inputs x 1 x 2 provides the main measurement: the distance between the two test masses and their differential alignment. x1 Reference x1 x2 x 1 provides as auxiliary measurement the distance between one test mass and the optical bench and the alignment of that test mass. Reference provides the reference phase for x 1 x 2 and x 1. Frequency measures laser frequency fluctuations with intentionally unequal pathlengths. 7

8 x 1 x 2 Testmass 1 WIN1 PDA1 PDFA M12 M11 BS1 PDFB BS2 BS7 BS16 BS9 BS4 BS11 M8 M6 BS6 M4 M10 BS8 BS3 PD1B M1 BS5 M14 PDRB M5 PD1A BS10 PD12B M15 PDA2 PDRA PD12A WIN2 GHH AEI Hannover <> :20:54< with an extra pathlength of mm in the reference fiber, the pathlength difference is mm. 8

9 x 1 Testmass 1 WIN1 PDA1 PDFA M12 M11 BS1 PDFB BS2 BS7 BS16 BS9 BS4 BS11 M8 M6 BS6 M4 M10 BS8 BS3 PD1B M1 BS5 M14 PDRB M5 PD1A BS10 PD12B M15 PDA2 PDRA PD12A WIN2 GHH AEI Hannover <> :20:54< with an extra pathlength of mm in the reference fiber, the pathlength difference is 0.02 mm. 9

10 Reference Testmass 1 WIN1 PDA1 PDFA M12 M11 BS1 PDFB BS2 BS7 BS16 BS9 BS4 BS11 M8 M6 BS6 M4 M10 BS8 BS3 PD1B M1 BS5 M14 PDRB M5 PD1A BS10 PD12B M15 PDA2 PDRA PD12A WIN2 GHH AEI Hannover <> :20:54< with an extra pathlength of mm in the reference fiber, the pathlength difference is mm. 10

11 Frequency Testmass 1 WIN1 PDA1 PDFA M12 M11 BS1 PDFB BS2 BS7 BS16 BS9 BS4 BS11 M8 M6 BS6 M4 M10 BS8 BS3 PD1B M1 BS5 M14 PDRB M5 PD1A BS10 PD12B M15 PDA2 PDRA PD12A WIN2 GHH AEI Hannover <> :20:54< with an extra pathlength of mm in the reference fiber, the pathlength difference is 380 mm. 11

12 Testmass 1 WIN1 PDA1 PDFA M12 M11 BS1 BS2 BS7 PDFB BS16 BS11 BS4 BS9 M8 M6 BS6 M4 M10 BS8 BS3 PD1B M1 BS5 M14 PDRB M5 PD1A BS10 PD12B M15 PDA2 PDRA PD12A WIN2 GHH AEI Hannov

13 Optical bench manufacturing The OB was manufactured by RAL from a Zerodur baseplate and fused silica optical components, using hydroxycatalysis bonding from U Glasgow and the optical design from AEI. 13

14 Recovery from accident A handling mistake caused 4 components to break at a late assembly stage. They could be repaired with interface plates ( bridges ). 14

15 Remaining assembly problems One bond is incomplete: Refinement of the bonding procedure is needed. The alignment procedure of the fiber injectors needs some refinement ( 100 µrad vertical error). More components need to be aligned with the jig (horizontal template accuracy insufficient for long lever arms). 15

16 Laser source The laser (by Tesat) is already space qualified and delivers 25 mw at the end of an optical fiber. It will be included in a larger box together with the Acousto optical modulators and associated electronics. 16

17 Functional overview Laser head Laser Box 1064 nm light Modulation bench 1064 nm light Optical Bench Frequency stabilization OPD stabilization PD signals Phasemeter PD Front end A/D converter FPGA preprocessing DMU processing pump light Pump module Light power Control f het Amp. Stab. sampling clock Power Control Output Master Oscillator Spacecraft 17

18 fast serial digital TBD laser unit 2*fast analog frequency 2 DAC analog slow power serial digital RS422 UART TBC 2*digital control 2 * analog monitor laser interface electronics power diag. 2*digital status laser assembly 1 fiber power supply BS AOM AOM 2*RF OPD act. OPD act. 2*analog dual frequency synthesizer, 2*RF power amp, 2* OPD actuator driver, laser amplitude servos, f_het, master clock digital serial interfaces(s), photodiode bias master clock AOM opto mechanics AOM electronics: fast serial digital TBD serial digital RS422 UART TBC DMU C&C 2 fibers OPD servo data 2 clocks frequency divider thermal shield 2*laser IN optical bench 8 quadrant photodiodes: 8*5 = 40 2 SE photodiodes: 2*2 = 4 thermometers? coils? heaters? phasemeter back end LTP optical metrology avionics DRAFT 1.0 master clock 40 wires plus screens 4 wires plus screens PCU phasemeter front end bias pre amp antialiasing A/D converters FPGA preprocessing clock 2 * 32 kbyte/s serial (4 * RS 422) serial digital RS422 UART TBC 2*16 quadrant channels phasedata S2 AEI TN 3018 C&C digital signal analog signal 1064nm light clock signal 28V power DC power C&C laser frequency servo data PCU n * ADC plus interface? high power DAC high power DAC n sensor outputs sensor control and power? RF clock 1 PPS signal MIL STD 1553 TBD power analog? power analog? S/C master oscillator S/C data bus magnetic coil heater 28V S/C power magnetometer solar power meter TBD particle counter thermometers some on optical bench? 18

19 OPD actuation EMI shield RF to AOM EMI shield RF to AOM LTP AOM electronics: one possible implementation Version S2 AEI TN 3019 RF power amp ON/OFF amplitude control amplitude monitor RF power amp ON/OFF amplitude control amplitude monitor stable bias VCXO VCXO PDA1 PDA2 OPD actuator drivers preamp + preamp + beam power + adder + + adder + error signal servo servo PLL dual frequency synthseizer f_het 2 DAC 2 ADC Monitor DAC 2*setpoint DAC lowpass lowpass 4 ADC Monitor 4 ON/OFF switch 2 ADC Monitor digital interface master clock fast OPD servo data command/control 19

20 Laser power stabilization 10-2 relative power fluctuations [1/ Hz] requirement from radiation pressure filtered voltage reference (AD587/OP177) req. from phase measurement shot noise (0.5mA photocurrent) radiation pressure: δp P < m c ω2 2P δy phase measurement: δp P < c 2 2 δϕ at 1 mhz at f het frequency [Hz] Stabilization via split feedback to Laser pump module (common mode) and AOM RF power (differential mode, BW>50 khz). 20

21 AOM driver A laboratory prototype of the AOM driver was built and characterized. It consists of two independent TCVCXO s, which are frequency-locked by a PLL to give a constant difference frequency (e.g. 1.6 khz). LF 10 MHz Input 10 MHz Output φ f Σ 1 TCVCXO 80 MHz Hz DBM TCVCXO 80 MHz 250 Hz ΑΟΜ1 ΑΟΜ2 500 Hz LP 500 Hz Input φ f LF Comp Sign Comp φ f Div /16 Div /16 5 MHz ε 5 MHz +ε 5 MHz +ε BP DBM 5 MHz ε BP 10 MHz BP Comp rad rms / Hz TCVCXO = temperature compensated voltage controlled crystal oscillator LP = lowpass filter BP = bandpass filter LF = loop filter Comp = comparator to generate logic level signals Div = digital frequency divider DBM = double balanced mixer φ f = digital phase/frequency detector Both the difference frequency ( 1.6 khz) and the average frequency (80 MHz) are controlled by phase-locked loops (PLL) k 10k 100k Fourier frequency [Hz] The phase noise of each oscillator is < 10 6 rad/ Hz at 1 khz. 21

22 TCVCXO 80 MHz DBM 2 W PA 20 db DC 2 W out BP Det LP LF 10VRef AM input TCVCXO = Temp. compens. VCXO DBM = double balanced mixer (used as attenuator) PA = Power Amplifier DC = Directional Coupler BP = 80 MHz Bandpass Det = Schottky Detector LP = 10 MHz Lowpass LF = Loop Filter 1/sqrt(Hz) 1e-05 free 1e-06 outloop 1e-07 1e-08 inloop 1e-09 preamp 1e Hz The RF amplitude of each oscillator is stabilized to 10 8 / Hz at 1 khz and has a fast input (BW > 100 khz) to compensate light power fluctuations that are measured at the fiber end. 22

23 Laser Frequency noise Laser frequency fluctuations δν = δω/(2 π) cause spurious phase fluctuations δϕ via a pathlength difference l between the arms. Conversion factor δω [rad/s] δϕ : τ = l/c, the differential time delay. Budget: δϕ < 6 µrad/ Hz between 3 mhz and 30 mhz Frequency stability requirement: δν = c 2π l δϕ = 28 khz Hz /[ l 1 cm ] Frequency noise [Hz/ Hz] measured 10 5 requirement 10 4 noise of aux. ifo Frequency [Hz] 23

24 Frequency stabilization We use the extra interferometer with L = 38 cm as sensor with sufficiently low noise. Two options are: Use that signal in a feedback loop to actively stabilize the laser (the baseline): Required loop gain : 100 at 30 mhz. With a 1/f simple integrator as loop filter we need unity gain frequency > 3 Hz. Allowing an extra phase delay of 45 in the loop gain at 3 Hz, the permissible processing time delay is 40 ms (achievable). Small complication with DC feedback: laser is forced to follow drifts of auxiliary interferometer (solvable). Do not stabilize the laser but use that signal to correct the main output signals for the frequency flucuations thus measured (fallback option). The actual pathlength differences l must be known to relatively high precision: δl = 0.1 mm and δl = 4 mm. Manufacturing to such accuracy is difficult, but measurement during operation is possible. 24

25 Phasemeter using SBDFT (Single-Bin Discrete Fourier Transform) Inputs from one quadrant diode: x i = U A (t i ), same for U B (t), U C (t), U D (t). First step: SBDFT achieves data reduction by a factor of 100: DC components: DC A, DC B, DC C, DC D (real) : DC A = n 1 i=0 x i, f het components: F A, F B, F C, F D : R(F A ) = The constants s i and c i are pre-computed: c i = cos ( 2π i k n n 1 i=0 x i c i, I(F A ) = ), si = sin ( 2π i k n ). n 1 i=0 x i s i. This operation is done in dedicated hardware (FPGA). 25

26 PM3 preliminary first results (DDS synthesizer, one channel - average of all channels) optical pathlength noise [m/ Hz] pm/ Hz LTP mission goal 9 pm/ Hz interferometer budget 1 pm/ Hz each interferometer contrib. µrad/ Hz OB EM Delft PM interferometer phase noise [rad/ Hz] frequency [Hz] 26

27 Further processing in DMU Longitudinal Signal: = F A + F B + F C + F D the total f het amplitude on the first quadrant diode, and F (1) Σ F (2) Σ for the second (reference) quadrant diode equivalently. ϕ long = arg(f (1) Σ ) arg(f(2) Σ ) + n 2π, (integer n from phasetracking algorithm). Alignment signals, independently on each diode: F Left = F A + F D : amplitude in left half, DC Left = DC A + DC D : average in left half, F Right, F Upper, F Lower, DC Right, DC Upper, DC Lower equivalently. The DC (center of gravity) signals: x = DC Left DC Right DC Σ, y = DC Upper DC Lower DC Σ, The DWS (differential wavefront sensing) signals: Φ x = arg ( FLeft F Right ), Φ y = arg ( ) FUpper F Lower Alignment signals are obtained from each quadrant diode individually (no reference needed) Rejection of several common mode noise sources., 27

28 Optical windows There will be 4 transmissions through an optical window of approx. 5 mm thickness in the main x 1 x 2 measurement path: 28

29 Pathlength effects Four major disturbing effects on the optical pathlength are expected: Thermal variation of optical pathlength. Stress-induced change in refractive index. mechanical motion of the window in z-direction. mechanical tilt fluctuations of the window. The sum of all noise contributions of one window (in double pass) is counted as one interferometer noise source and allocated a bufget of 1 pm/ Hz. Hence the window effects contribute no more than 1 pm/ Hz in the x 1 measurement and no more than 2 pm/ Hz in the x 1 x 2 measurement. Each effect is allocated 0.33 pm/ Hz (for one window double pass). 29

30 Thermal variation of optical pathlength: s = T L ( dn dt + (n 1)α). dn/dt + (n 1)α is 5 ppm/k for most glasses (e.g. BK7). Athermal glasses (e.g. Ohara S-PHM52, Schott N-FK51 and Schott N-FK56) have ppm/k. The Schott glasses have a high α 15ppm/K, not well matched to Ti. The best candidate that we identifed so far is Ohara S-PHM52. At 1064 nm, dn/dt + (n 1)α = 0.59 ppm/k. The linear thermal expansion coefficient α is 10.1 ppm/k, well matched to Ti. All these athermal glasses are difficult to polish and very brittle, which may limit the mounting options. With glueing, care must be taken to avoid high static stresses that might cause the glass to break in thermal cycling. 30

31 From δt = δs L ( dn ), dt +(n 1)α a pathlength error of δs = 0.33 pm/ Hz and L = 12 mm, the required thermal stability at the window is: δt < K/ Hz Thermometer noise Ohara S-PHM52 Linear thermal expansion α = 10.1ppm/K. LSD (K/ Hz) AD-590: out-of-loop PT : out-of-loop NTC: out-of-loop req. at 1064 nm: n = , dn/dt + (n 1)α = ppm/k Frequency (Hz) 31

32 Stress-induced change in refractive index: While in some materials the stress-induced birefringence can be made small, we have here the absolute variation in refractive index, which is never small. The relevant material constant is: Photoelastic constant β = 1.0 nm/cm/10 5 Pa. For a pathlength error of 0.33 pm/ Hz and L = 12 mm, the required stability of mechanical stress in the window is: δσ < 30 Pa/ Hz. We have no knowledge of the real stress fluctuation. This error might be big. Mounting of the optical window will be critical (In seal? Au-Sn seal?). Measuring the thermally induced pathlength fluctuation is essential. 32

33 Mechanical motion: If there is a deviation γ from parallelism and the window moves in z direction by z this yields a pathlength error (double-pass): s = 2γ(n 1) z. For γ = 30, n 1 = 0.6 and a pathlength error of 0.33 pm/ Hz one gets δz < 2 nm/ Hz. If this is difficult, the obvious remedy is to improve the parallelism. 33

34 Mechanical tilt fluctuations: l = d cos β, α β α β d a l x s = d a = l cos(α β), sin α sin β = n, s = nl a cos 2α + 2n cos α d With α = 2.5, d = 6 mm, n = 1.6: d( s) dα = m/rad For a pathlength error of 0.16 pm/ Hz (0.33 pm/2 because of double-pass) one gets δα < 1.6 nrad/ Hz. If too difficult, α might be reduced at the expense of stray light problems. 34

35 Functional, environmental and performance tests The EM was tested during March and April, 2004 at TNO/TPD, Delft. The tests included: Functional tests before and after each other test, Thermal vacuum test: several cycles C, Vibrational test (with dummy masses): 8 g rms sine and random, 25 g at the struts. Performance tests: Full stroke test: each mirror moved by ±100 µm, Noise test: mirrors not actuated Tilt test: each mirror tilted by ±1000 µrad, All tests were successful! The AEI test team at TNO. 35

36 EM Test results: high velocity full stroke test (2.9 samples/cycle) 0 LPF OB Full Stroke Test 1/2: end mirror 1 actuated TNO/AEI 2004/03/ LPF OB Full Stroke Test 2/2: end mirror 2 actuated TNO/AEI 2004/03/04 1 optical pathlength [um] measured x 1 measured x 1 -x 2 contrast Ref contrast x 1 contrast x 1 -x contrast optical pathlength [um] measured x 1 measured x 1 -x 2 contrast Ref contrast x 1 contrast x 1 -x contrast time [sec] time [sec] optical pathlength [um] mirror 1 PZT motion [um] optical pathlength [um] mirror 2 PZT motion [um] 0 PZT stroke 0 0 PZT stroke time [sec] time [sec] 36

37 EM Test results: Noise LPF OB performance AEI/TNO 2004/03/ LPF EM (TNO/AEI) no PZT-stab, quad diodes Glasgow/AEI best with PZT-stab, quad diodes x 1 - x 2 LPF EM (TNO/AEI) with PZT-stab, quad diodes Glasgow/AEI best with PZT-stab, single-elt. diodes phase [rad/ Hz] LPF mission goal interferometer goal optical pathlength [pm/ Hz] frequency [Hz] 37

38 EM Test results: Contrast 1 LPF OB performance: Contrast TNO/AEI 2004/03/19 contrast Ref contrast x 1 -x contrast time[sec] 38

39 EM Test results: Noise sources 1 At frequencies < 3 mhz, real motion of the test mirrors is dominant: phase x 1 - x 2 [rad] optical pathlength fluctuation between mirrors [nm] phase x 1 - x 2 [rad] nm/h drift subtracted optical pathlength fluctuation between mirrors [nm] time[sec] time[sec] 39

40 EM Test results: Noise sources 2 An attempt to glue the Zerodur mirrors to the Zerodur baseplate failed: Curing of the glue caused 200 µrad misalignment and a contrast drop to < 0.5. This is mainly a testing problem. 40

41 EM Test results: Noise sources 3 F AOM L1 Fiber L1M f2 L2 L2M M Laser BS f1 L1R L2R R PDR PDM f het ϕ el ϕ R ϕ M ultra stable Fluctuations of the fibers Optical Pathlength Difference (OPD, F ) should ideally completely cancel, but in reality some error remains. 41

42 4 1.2 ϕ = ϕ R - ϕ M [mrad], drift subtracted ϕ = ϕ R - ϕ M [mrad], drift subtracted time [seconds] time [seconds] The measured pathlength x 1 x 2 signal has an erroneous component of mrad magnitude which is quasi-periodic with F. 42

43 ϕ R, ϕ M [rad] time [seconds] Unless the origin of the noise will be understood, a remedy is to actively stabilize F. This was done using an analog phasemeter, analog servo and long-range PZT at TNO. Further investigations are under way in Hannover and Glasgow. 43

44 Alignment measurement with quadrant photodiodes 3 ways to use a quadrant diode: A C B D Σ = A + B + C + D is used as before for the longitudinal readout. The DC signals y = A + B C D and x = A + C B D measure the average lateral displacement of both beams. Differential wavefront sensing measures the angle between interfering wavefronts: split photodiode Phase meter avg pathlength change testmass longitudinal Phase meter diff wavefront angle testmass tilt 44

45 4 tilt test: mirror 1 x (horizontal) 4 tilt test: mirror 2 x (horizontal) DWS signal [rad] DWS signal [rad] x1 Phix -3 x1 Phiy x12 Phix -4 x12 Phiy mirror tilt [urad] -2 x1 Phix -3 x1 Phiy x12 Phix -4 x12 Phiy mirror tilt [urad] DC signal x1 DCx -0.3 x1 DCy x12 DCx -0.4 x12 DCy mirror tilt [urad] DC signal mirror tilt [urad] x1 DCx x1 DCy x12 DCx x12 DCy x1 contrast x12 contrast contrast contrast x1 contrast x12 contrast mirror tilt [urad] mirror tilt [urad] 45

46 EM Test results: Differential wavefront sensing (DWS) The conversion factor of test-mass angle α to (differential) phase readout ϕ is analytically: ϕ/α = 2 2πw(z)/λ 5000 rad/rad. Rot. TM1 Rot. TM2 units x 1 ifo predicted (numerical) rad/rad x 1 ifo measured (x) rad/rad x 1 ifo measured (y) rad/rad x 1 x 2 ifo predicted (numerical) rad/rad x 1 x 2 ifo measured (x) rad/rad x 1 x 2 ifo measured (y) rad/rad conversion factor depends on beam parameters; calibration is necessary. Better than the angular readout capability of the capacitive sensors; will be used to stabilize the alignment of the test masses. DWS works only when there are fringes (test mass absolute alignment better than 300 µrad). Otherwise, DC alignment signals are used for rough alignment of test mass. 46

47 EM Test results: Differential wavefront sensing (DWS) TM1-TM2 differential misalignment (nrad/ Hz) Frequency (Hz) PDR x PDR y PD12 x PD12 y 10 nrad/sqrt(hz) 47

48 Summary interferometry and phase measurement for LTP work as predicted. some minor refinements are needed in the construction procedure. environmental and performance testing was successful. further investigations will concentrate on the small vector noise. 48

49 Future work for LISA Caging mechanism! actuators (PZT?) for cavity length and point-ahead angle. optical sensing of test mass in all d.o.f. clock synchronization and distance measurement ( ranging ). data transfer using sidebands with 1 pw power. extending frequency range downto Hz. 49