1 LTP: The LISA Technology Package aboard LISA Pathfinder Gerhard Heinzel, AEI Hannover 第 6 回 DECIGO ワークショップ 2008 年 4 月 16 日 国立天文台 三鷹 using material from Paul McNamara, Stefano Vitale and EADS Astrium
2 Purpose of LTP Test and verify the inertial sensor for LISA Verify operation of the proof mass as reference mirror in a pm interferometer Test drag-free operation and micro-newton thrusters side effect: Learn a lot about spacecraft design
3 Science Objectives
4 Terminology LISA Pathfinder (LPF) is the mission, managed by ESA LPF used two have two scientific payloads: The European LISA Technology Package LTP The US Disturbance Reduction System DRS DRS has been reduced to thrusters and a computer The payload LTP consists of: two inertial sensors with one test mass each the interferometer with laser, phasemeter etc. The LPF mission includes: micro-newton thrusters drag-free control all standard spacecraft things
5 LPF/LTP participants France: Laser modulator Germany: PI, LTP Architect (Astrium), Laser Italy: PI, Inertial Sensor (ISS), Caging Mechanism Netherlands: ISS SCOE Spain: Data Diagnostics System, Data Management Unit Switzerland: ISS Front End Electronics United Kingdom: Optical Bench, Phase-meter, Charge Management
6 LTP team LTP workshop in Trento (2005) showing maybe ½ of people working on LTP
7 Orbit Lagrange-Point L1, about 1.5 million km from Earth, limits downlink data rate constant orientation to Earth and Sun, stable thermal environment Separation from propulsion module after several months cruising phase stable without correction for the month mission
8 Launch planned 2010 Baseline is new European launcher VEGA from Kourou Alternative: using Rockot (former SS19 ICBM) from Plesetsk, Russia (latitude 63 ) Max lift-off weight of S/C: 1910 kg
9 Operation 8 hours per day contact via ESA 15m/35m ground stations science data downlink: average kbit/s program for 3 days in advance is uploaded and ready to run interaction mainly via parameters of procedures quasi real-time operation only in commissioning or emergency
10 Keeping the spacecraft with the proof-mass Thrusters Spacecraft Displacement sensor Test mass x High gain force feedback
12 Drag-free mode both test masses are optically sensed orientation is controlled to optimize interferometer contrast TM nominal position is unstable ( negative stiffness ) TM1 is drag-free reference, spacecraft follows TM1 with 65 mhz loop bandwidth TM2 has suspension controller (electrostatic) with 3mHz bandwidth
13 LPF mission goal LISA requires 3e-15 m/s2/sqrt(hz) at 0.1mHz LTP requires 3e-14 at 1mHz, but aims for LISA-like levels LTP carries many diagnostic items to analyze and correlate any noise that occurs
14 Comparison with other missions
15 Key components Inertial sensor: test mass: 2 kg of Au-Pt alloy, cubic form capacitive sensor with front-end electronics, also works as actuator vacuum enclosure with optical window charge management system: fiber-coupled UV light Drag-free system: micro-newton thrusters Software mass balancing Interferometer
16 Test mass 46mm cube of Gold-Platinum, 73% Au:27% Pt Mass = 1.96kg high density, low magnetic suscept., but: hardness and magnetic properties differ from small samples to large piece!
17 Capacitive Sensor/Actuator large gaps (2...4 mm) made from Mo/Al2O3/Au AC excitation (100 khz) front-end electronics challenging for noise and cabling redundancy
18 caging mechanism needs to hold test mass at launch, Force = 3000N release on orbit without sticking, velocity < 5µm/s! TM gold coating must not be damaged limited choice of materials difficulty was severely underestimated separation into caging (hydraulic) and release (piezos) qualification tests ongoing
19 Charge management Test mass charges due to cosmic radiation at ca e/s Charge is measured either continuously or periodically via electrodes discharge with UV light (Hg) shining from fibers on either test mass or housing but: Au work function depends on contamination. 254nm = 4.88eV, Aucontam. has up to 5.1eV!
20 Optical Window needed for optical access of interferometer beams like a flange of vacuum tank optical pathlength in transmission 12mm/24mm athermal glass S-PHM52 (Ohara) minimizes pathlength error dn/dt+(n-1)a extensive testing at AEI for radiation hardness, pressuredependent pathlength error and actual performance was successful.
21 Optical window electrostatics an isolating window may accumulate charges and disturb the test mass Solution: apply conductive ITO (In2O3/SnO2) layer to optical window
22 micro-newton thrusters three systems under development: Indium needle Cesium slit colloidal (organic ionic liquid) range about 100 µn, stepsize 0.1 µn, operated at a bias. LTP will test two or three of them. noise and frequency response (delay) are hard to predict. issues are reliability and lifetime.
23 Interferometer originally intended as a passive diagnostic tool with no feedback to test masses Now a central part of the experiment, controlling the test masses Using test masses as end mirrors has many complications and is a crucial experiment for LISA
24 Ifo requirements Pathlength noise 9pm/sqrt(Hz) with freq. dependence sufficient for LISA local readout prototype fully meets requirement
25 Ifo principle audio-frequency heterodyne Mach-Zehnder independent of operating point wide dynamic range (many fringes) no lock acquisition needed, immediately ready after power-on.
26 Reference subtraction Each photodiode measures Pathlength difference starting from first BS External contributions must be subtracted via stable reference interferometer subtraction is imperfect, hence phase of Ref. Ifo ( OPD ) must be stabilized
27 Interferometer noise The usual suspects are below 1 pm each: laser frequency is stabilized via auxiliary interferometer with unequal pathlength. laser power is stabilized both at mhz for radiation pressure and at khz for the beatnote phase measurement. The OPD is stabilized via a Piezo in one of the arms. Phasemeter electronic / digitization noise is below 1 pm.
28 Angle measurements uses Differential Wavefront Sensing (DWS) large amplification TM angle to audio phase difference (about 5000) one quadrant diode is enough, no reference needed immune to several noise sources excellent sensitivity
29 Laser Nd:YAG NPRO 35 mw output power frequency and power actuators flight heritage on Terra-SAR X also useable as seed laser for high-power fiber amp (LISA)
30 Laser Modulator contains beamsplitter, 2 AOMs and Piezo OPD actuator fiber coupled output: 2 x 5mW frequencies generated by 2 crystals in PLL arrangement AOM also used for power control challenge: spectral purity of output
31 Optical bench 20*20cm Zerodur baseplate hydroxycatalysis bonding about 30 components 4 interferometers challenges: vertical alignment fiber launchers absolute alignment w.r.t. test mass housing
32 4 interferometers TM1 position and orientation w.r.t. optical bench TM2 position w.r.t. TM1, TM2/TM1 orientation w.r.t. optical bench Reference phase Frequency noise via unequal pathlengths
33 coupling of unwanted d.o.f. A perfect interferometer would sense only x An offset d will couple testmass (or spacecraft) rotation into x An angle phi will couple y/z motion into x. The coupling depends on the precise interferometer layout and the beam parameters
34 Optical bench alignment bonding accuracy is roughly 10um or 50urad per component extensive Monte-Carlo simulation of many possible misalignment combinations predict length error of about 10pm/sqrt(Hz) mitigation: fitting coupling coefficient using natural fluctuations and subtraction of predicted contribution subtraction has been experimentally verified
35 data flow On-board computer (OBC) does not deliver its clock LTP runs on nominally same frequency but not synchronized drag-free requires continuous intimate interaction between them non-synchronuous operation creates many problems
36 Phasemeter One ADC in each channel immediate single-bin DFT in FPGA hardware AEI breadboard: 18bit/800kHz, 20 channels FM: 16bit/100kHz 2*16 channels, fully redundant breadboard noise <0.1µrad
37 interferometer postprocessing Phasemeter delivers DC, real and imaginary at 100Hz longitudinal and alignment signals are derived by simple operation like arctan() raw longitudinal signals are periodic with the wavelength phasetracking algorithm removes jumps by 2π phasetracking needs to be reset at known test mass position in order to provide absolute measurements nominal signal handling is straightforward, 80% of the effort goes into proper handling of non-nominal situations (loss of one quadrant, temporary loss of contrast etc.)
38 interferometer data position and orientation of both test masses summary status information for debugging, many other channels exist: contrast, power levels, intermediate results,... several menus of data packets depending on application
39 Data analysis Software must be verified and delivered to ESA before launch, since results are used to upload new parameters Supplied by PI Institutes (Hannover and Trento) Based on MATLAB with extensive own programming Must have GUI interface for non-matlab experts The package LTPDA (LTP Data Analysis) contains Time series tools (segmentation filtering, coloured noise generation,...) Frequency domain tools (spectra, cross-spectra, time-frequency analysis...) Arithmetic functions, data handling tools and auxiliary functions Access to MATLAB internal functions via wrappers LTPDA is also useful for other work,is freeware open-source freeware, and we welcome all new users:
40 Analysis objects A useful result is not a graph or a file full of ASCII data Each result must know how it was produced with all details Each result can be reproduced by any user with access to the raw data, also with modified processing All intermediate steps and final results are stored as MATLAB structures called Analysis objects (AO) Each processing step appends its name, version and parameters to the history of the resulting AO
41 Mock data challenge Test of the data analysis pipeline: One team generates downlink data with a spectrum kept secret Second team uses LTPDA to recover the spectrum First round (simple model) successfully concluded
42 Lessons learned (my personal view) Drag-free spacecraft require unusual architectures, the textbook-style clear separation between spacecraft/payload does not work well. Asynchronous clocks are a bad idea. Industry does not know or guess what scientists need, they work strictly on requirement documents. You'll need more data channels for debugging than you might think now. Start early to think about on-board data flow, software and computer tasks. Handling of errors and non-nominal situations is a major part of the software. Resources on a spacecraft may be unbelievably limited (e.g. 20 MHz CPU with 256 kbyte array for user data)! Requirement documents must be over-complete. Even self-evident things must be spelled out in detail. It is very hard to add things later.
43 We are looking forward to an exciting time! Wishing you good luck with DPF, Thank you for your attention!