Echidna Performance of Echidna fiber positioner for FMOS
Echidna: Performance of Echidna fiber positioner for FMOS on Subaru M. Akiyama (Tohoku Univ. ), S. Smedley, P. Gillingham, J. Brzeski, T. Farrell, R. Muller (AAO), M. Kimura, N. Takato (Subaru) SPIE Marseille 7018 -106
Echidna: novel technology fiber positioner for Subaru FMOS, named after Australian spiny ant eater, developed in Anglo-Australian Observatory. 1) introduction of the fiber positioner 2) results of lab testing of stand-alone performances at sea-level (Hilo commissioning) 3) short summary of on-sky commissioning at Mauna Kea conducted Dec. 2007, Jan. , May. , and Jun. 2008.
Introduction: what is FMOS ? Fibre Multi-Object Spectrograph • 2 nd generation open-use instrument for Subaru. • Simultaneously obtain spectra of 400 targets in d=30’ Fo. V of the primefocus. • Fibers are fed to two OH-airglow suppression spectrographs with wavelength coverage of 0. 91. 8 micron. • Spectral resolutions of the spectrographs are R=500 and R=2200. For spectrographs status see, Iwamuro et al. [7014 -27] Dalton et al. [7014 -137]
Introduction: new method required ! • Why do we need new method for fiber positioning ? – The problem is the physical size of the focal plane • Subaru Prime-focus 8. 2 m F/2. 0 FL/16 m = 135 mm/0. 5 deg. • If we distribute 400 fibers, the spacing will be 7 mm, • Then if we configure 400 fiber with a robot + magnets, like 2 d. F, 150 mm 100 micron=1. 2” 1. 0”=83 micron
Introduction: Echidna principle Quadrant tube piezo actuator Tilt magnified, one step will be 14 -40 micron at 160 mm tip. The maximum tilt will be 7 mm, to the home positions of the neighboring fibers. • Slow bend and quick bend back of quadrant tube piezo actuator with the sawtooth signal make the “stick and slip” movement of the pivot ball of the “spine”. • For principle and prototype developments, see Gillingham et al. 2000 SPIE 4008, 1395, Moore et al. 2003 SPIE 4841, 1429, Gillingham et al. 2003 SPIE 4841, 985.
Introduction: Echidna module • The 30’ Fo. V is covered with the 400 fibers divided into 12 modules. One module consists of 40 science fibers + 2 guide fiber bandles. Tiled 12 modules cover the 30’ Fo. V. Out of 480 fibers, 400 fibers are connected to the two spectrographs (for the details of the mechanical structure see Brzeski et al. 2004, SPIE 5492, 1228).
Introduction: entire unit
Introduction: installation to the telescope • The fiber positioner unit is attached to the prime-focus of the telescope enclosed in the prime-focus unit. For the prime-focus unit, see Kimura et al. [7014 -199]
Introduction: “configure fibres”
Introduction: “configure fibres” zoom-up
Introduction: requirements for the positioner • The optimal core size of the science-fibers was determined to be 1. 2”, which corresponds to 100 micron, maximizing the expected SN ratios of faint galaxies. Coupling efficiency as a function of the misalignment of the fiber to the target. Stellar object is assumed • For misalignment less than 20 micron, the reduction of the coupling efficiency is less than 10%, even under a good seeing condition with FWHM=0. 5”. • Due to a number of physical effects, the repeatability of spine movement is not precise, thus iterations are required to achieve the required accuracy. Position the 400 fibers with the 20 micron accuracy to targets within 15 minutes.
Performance: Spine camera fiber position measurements Sky camera Accurate measurements of the fiber positions are crucial to the evaluations of the mechanical and dynamical properties of the positioner. The positions are measured with the spine camera of the Focal Plane Imager. The camera covers the entire focal plane by moving with an XY gantry. The positions of the 400 fibers can be sequentially measured in 100 seconds. The spine camera uses a telecentric lens to image the spine tips. The distortion pattern of the spine camera and the non-orthogonality of the XY gantry were calibrated by measuring a gridpattern light-source attached at the focal plane in stead of the fiber modules (shown in the picture). The residuals of the calibrations are 2. 0 micron RMS in the entire Fo. V, including both residuals of the distortion pattern and the non-orthogonality models.
Performance: length of the spines • Due to the fast F/2 ratio of the prime- 150 micron 100 micron Best-focus 100 micron focus, careful control of the parfocality of the fibers is important. If a spine is shorter (or longer) and is defocused by 150 micron, the reduction of the coupling efficiency is estimated to be 5% under typical seeing condition. • The parfocality of the tips of the fibers is checked with a CCD camera with a high-magnification lens attached to the Focal Plane Imager. 150 micron Science fibres in black, and guide fibres in light blue. The best focus position is shown with red line. • All but two of the science fibers are within 100 micron of the best focus position. The RMS of the deviation is 33 micron. For the guide-bundles, all but two spines have length within 150 micron of the focus position.
Performance: flexure of the spines The direction of gravity in X Deflection of the spines are measured by tilting the unit. The positions of the fiber tips are measured with the spine camera. The average deflection of the spines is 50 microns between tilt angle of 0 deg. and 60 deg. But, it should be noted that the absolute deflection is not an issue and the relative deflection needs to be examined, because “science” targets are acquired and tracked in relative to “guide” stars.
Performance: flexure of the spines The size of the relative deflection is smaller than 15 micron (0. 18”) for most of the fibres from ZD=60 deg to ZD=0 deg. The scatter of the relative deflection is sufficiently small for long integration time. The size of the relative deflection from ZD=60 deg to ZD=0 deg measured at 4 different position angles (shown with different line type).
Performance: cleanliness of the fiber tips Cleanliness of the fiber tip is examined attaching microscope on the FPI. There are some dirty fibers (samples are shown above), but we found they can be cleaned easily.
Performance: Coarse mode X+ step size spine movement The movement of the tips of the spines happen in “steps” following the input saw -tooth signal to the quadrant tube piezo. There are two operation modes; the coarse (140 V 70 Hz) and the fine modes (55 V 15 Hz). The size of the “step” is measured with the spine camera at three tilt angles. The measured step sizes are shown for X+ movement. The horizontal axis is the measured step size at tilt angle of 0 degree. The step sizes have factor of 2 scatter around the average. Fine mode X+ step size The vertical axis is the difference of the step size for each spine between different tilt angles. In order to achieve ~40% accuracy for the movement, we need to calculate the required step number for each spine using the step size of each spine, but we do not need to consider the variation of the step size of each spine due to the change of the elevation angle.
Performance: Fiber positioning tests Intensive positioning tests have been done inside tilted prime-focus unit at various rotation angles. At tilt angle<60 deg. , 95% of the spines reach target position within 12 micron with 7 iterations. About 10 out of 395 fibres cannot reach the target positions. At tilt angle=60 deg. with some rotation angles, the performance degrades. ~20 out of 395 fibres cannot reach the target positions with 7 iterations. With 7 iterations, the fibre configuration takes 13 minutes currently. The bad positioning accuracy at ZD=60 deg is some spines have difficulty with moving against the direction of the gravity in the current mode. This may be solved with different frequency and voltage signals to the fibre movement. The fine tuning of the frequency and voltage is still underway.
Performance: Long-term position stability No systematic movements of the spine tips were observed at tilt angle of 0 and 30 degrees. Small systematic shifts were observed at tilt angle of 60 degree as shown below, but still negligible.
On-sky: distortion pattern measurements The first step of commissioning is the evaluation of the telescope distortion pattern using the sky-camera mounted on the FPI. We observed fields with many bright stars, i. e. open clusters and galactic plane, with the sky-camera.
On-sky: distortion pattern measurements The residual of the current model is 0. 3” RMS.
On-sky: spectra of astrometry calibration stars
On-sky: stars acquired with guide-bundles
Merci: Thanks for your attention. Echidna is working well with sufficient accuracy (12 micron accuracy within 13 minutes). is under on-sky commissioning. Currently, the error of the fiber positioning is dominated by the uncertainties of the distortion pattern model (0. 3” RMS, i. e. p-p = 1. 0”). The model will be updated further with the data taken Jun. 2008 commissioning (last week !) and the new model will be tested in the next observing run in Aug. 2008.
Performance: fiber position measurements • The distortion of the fibre camera and the distortion of the gantry XY movement (non-orthogonality etc. ) are calibrated by measurering a grid-pattern illumination source on the focal plane of the echidna unit in stead of the fibre modules. The measurements were done at various ZD (tilting PIR) and rotator angle. The distortions of the fibre camera optics and the XY gantry movement can be described well with the “mean” distortion model determined by averaging the models measured at various positions. This means we do not need to consider ZD/rotator angle dependence of the model, and this simplifies the software. Still need to check temperature dependence of the FPI distortion model up at summit. • Residuals in one Fo. V of the fibre camera after removing third order distortion is 1. 27 micron R. M. S. with maximum error of 3. 14 micron. No systematic residual. • Residuals in the entire focal plane after removing the camera distortion model and the gantry XY movement distortion model. 1. 98 micron RMS using “mean” distortion model.
Echidna: fibre “spine” properties : patrol area The “spines” are designed to reach the center of the neighboring spines, i. e. 7 mm 0. 5 mm (tip thickness) = 6. 5 mm. The size of the patrol area for each fibre “spines” are measured by moving each spines to the edges. Patrol area of a spines is defined with the mean radius from the center of the patrol area. All of the spines have the patrol area larger than 6. 5 mm radius.
Echidna: control s/w and GUI Echidna is controlled by the ICS s/w running on the “echidna” PC in the electronics enclosure of the unit. “echidna” PC is a disk -less PC booting using linux image on “fmos 01”. Observation catalog with targets’ RA, DEC, priority etc. is fed to the fibre allocation s/w. Optimizing fibre assignment (observe higher priority targets, minimizing fibre tilt, etc. ) the s/w determines the fibre allocation and store the data in. s 2 o file. Based on the allocation, ICS s/w automatically configure fibres with several times (~7) iterations. The results of the configuration will be stored in. efai file in the “fmos 01” and fetched by the FITS creation software. The target information, configuration results are stored as ASCII extension header of the FITS images file. Engineering menus (for example calibrating fibre movement automatically) are also available from the GUI.
Echidna: spares • Spare boards are available for most of the electronics boards. • Two spare modules (with 40 fibres, there are 12 modules in the focal plane) will be made. They can be replaced with any two of the 12 modules currently on the focal plane. Module replacement is still a big task, but can be done by Subaru staff. We HOPE the two spare modules will cover the life time of the instrument (~10 yrs).
Echidna: FMOS astrometric standards for eng. obs. FMOS astrometry fields in Feb. FMOS astrometry fields in Aug. We select several fields as the FMOS astrometric standard fields. In order to check the corrector distortion pattern and the positioning accuracy we need relatively low Galactic latitude fields with accurate astrometric calibration. We are proposing observations of 14 selected fields with >100 astrometrically calibrated stars (with UCAC 2 catalog) with Scam in this Oct. The Scam data ensure better astrometric accuracy and covers fainter stars. The 14 fields will provide engineering targets at various ZD during whole night throughout year. For open use observations, we need some sort of astrometry check s/w for observer’s catalogs with UCAC 3 catalog (not leave observer’s own risk…like VIMOS, DEIMOS…)
Echidna: Issues (not mentioned so far) Before on-sky commissioning • Transfer to the summit ! With BSIT maximum acceleration is expected to be ~1 G and spines are tested fine up to 8 G, but… • Piezo tube used to control fibre spines are fragile ceramic, and careful/gentle handling is necessary. • Re-termination (replace carbon-fibre tube with new one) of fibre spines needs to be done for ~10 more spines. Will be completed in the next two weeks. • Echidna properties up at summit need to be examined just after the summit transfer on Nov. 1. On-sky commissioning • Accuracy of fibre bundle guiding. We have a problem of even illumination of the bundles of 7 fibres. Better diffusers will be installed to the back-illumination unit for the guide bundles in the next two weeks. Guide fibre image on “fibre” camera
- Slides: 30