SummaryELODIE is a cross-dispersed échelle spectrograph permanently located in a temperature-controlled room in the first floor of the 1.93-m telescope building. The first observations with this instrument were made on June 1, 1993. Light from the Cassegrain focus is fed into the spectrograph through a pair of optical fibers. Two focal-plane apertures are available (both 2 arc-sec wide), one of which is used for starlight and the other can be used for either the sky background or the wavelength calibration lamp, but can also be masked. The spectra cover a 3000 Å wavelength range (3850-6800 Å) with a spectral resolution of 42000. The instrument is entirely computer-controlled and a standard data reduction pipeline automatically processes the data upon CCD readout up a point where a choice must be made between continuing with very accurate (better than 10 m/s) radial velocity measurements by numerical cross-correlation techniques or proceeding with no further analysis.
A publication is available describing in detail the instrument, the data reduction procedures and some of its characteristics.
Pointing and guidingA special adapter carrying the optical fiber heads is mounted at the Cassegrain focus of the 1.93-m telescope. It also includes an autoguider CCD camera, the calibration lamps and atmospheric refraction correctors. The field of view is 4.6 arc-min. The pointing system features digital encoders in both declination and hour angle with an accuracy of 5 arc-sec. A pointing model corrects for mechanical flexure and allows for accurate setting of the telescope. The coordinates are displayed in the observing room and in the dome. There are limits on the pointing of the telescope due to its asymmetrical mounting.
SoftwareAll spectrograph functions are controled via dedicated software (called LIDO) running on a Sun workstation (called "alix") which also hosts the data reduction software (called TACOS). The status of the spectrograph, of the exposure in progress and of the reduction pipeline are displayed on the terminals located in the observing room (the old Coudé room). An exposure can be started either by filling in a form where the star is identification is entered, using a special catalog code, together with its coordinates and the observing program number, or by reading from a previously formatted observing catalog, for which detailed intructions are available.
Optical LayoutThe following illustrations show the principal optical elements of ELODIE :
Entrance apertureThere is no slit. The two circular entrance apertures (2 arc-sec wide) are separated by 1.8 arc-min in the focal plane. They feed light into the spectrograph through a pair of optical fibers. There are actually two pairs of fibers, one of which includes an optical scrambler system designed to render the measured radial velocities largely insensitive to the exact position of the star in the entrance aperture. This special pair (called "fibres brouillées") is used by the Geneva group, in combination with a simultaneous wavelength calibration exposure on the "sky" aperture, for high-accuracy (▒ 6 m/s) velocity measurements. Other observers are recommended to use the other fiber pair (called "fibres directes") for most observations, since it has higher throughput.
GratingThe Milton-Roy échelle grating blazed at 76° has dimensions of 40.8x10.2 cm and is mounted in a fixed configuration. The spectrum as projected onto the TK 1024x1024 CCD detector yields 67 spectral orders between 3895 and 6815 Å. The measured resolution is 45000. A list of orders is available.
Calibration lampsTwo lamps are mounted in the Cassegrain adapter : a tungsten lamp for relative flux calibration ("flat field") and a Thorium lamp for wavelegth calibration. Exposure times for both lamps are automatically set for the observer.
DetectorThe CCD is a thinned, back-illuminated, anti-reflection coated Tektronix (now SITe) 1024x1024 with 24 microns square pixels. The CCD gain normally used is 2.65 e/ADU. The read-out noise is below 7 e. The bias level is measured once over the full image at the beginning of the night, but accurate values are measured for every exposure using the overscan zone in the CCD frame.
Signal-to-NoiseA plot shows the expected S/N ratio (at 5500 ┼) as a function of the magnitude of the star (for a standard exposure time of 1 hour) for two different situations : maximum and standard efficiency (which combines factors such as seeing and transparency). Also shown in the plot is the maximum safe S/N ratio recommended before non-linearities set in (a value of ~400).
Basic data reductionA special feature of the ELODIE spectrograph is its associated automatic data-reduction pipeline, designed and implemented by the Geneva Observatory team. This software performs localization of all orders on the frame, optimal order extraction, cosmic-ray rejection and wavelength calibration. Also computed for each order are the blaze function and a "one-dimensional" flat field, both derived from the Tungsten exposure. The latter is derived for each order from the average of the rows perpendicular to the dispersion. The narrowness of the orders (3-4 orders) and the high quality of the CCD makes any further second-order flat-field correction unnecessary. The extracted spectra are saved as 1024x67 images (so-called "s2d" files). All TACOS identification and calibration information is written in the file headers.
Cross-correlation analysisThe automatic data reduction pipeline halts, waiting for user input, at the stage where the user can decide whether to go ahead with cross-correlation analysis or to forego any further processing. Several numerical cross-correlation masks are available, the most useful are those for F-G type stars and for K-M stars. The user then interactively fits a single or a double gaussian to the cross-correlation profile to derive the radial velocity. The velocities are corrected to the center of the solar system using the coordinates in the catalog or from the telescope if none have been entered. The accuracy of the radial velocities, which depends on parameters such as spectral type, rotation, duplicity, and signal-to-noise ratio can vary from ▒ 50 m/s in the worst cases, to ▒ 6 m/s in the case of simultaneous Thorium calibration and use of the scrambled fibers.
Introduction and tutoringEach new observer is given a complete introduction by local staff to recommended observing procedures and tutored on software operation and data reduction as needed for succesful observing. This introduction normally takes place in the afternoon (14h local time) preceeding the first night of the run.
Beginning of the nightObservers desiring to obtain standard échelle spectra are advised to adhere to the following procedure for starting an observing night. In the afternoon (between 14h and 16h local time) proceed as follows :
End of the night/Data archivingIt is recommended that a short (a few seconds) exposure on the sky be secured at the end of the night. After acquisition and reduction of this "dummy image" is over, the observed normally clicks on the button ("Fin de nuit") to end the night. At this time the user must make sure that final data reduction is selected by answering the query "Traitement complet: OUI" so that calibrated FITS files are computed for all science frames (67 files for each spectrum). The "dummy image" specified above is needed so that the last science frame is fully reduced. Otherwise reduction halts at order 64 (software bug). The files are then archived on CD-ROMs by the day staff.
Night assistantA night assistant is on duty at the 1.93-m telescope at all times (except for meal time around midnight) and is responsible for telescope safety. He will open, close and turn the dome, point the telescope, start and set up the auto-guider. Observing is the responsibility of the astronomer. The night assistant is normally able to deal with most problems in the course of the night, but in very difficult cases he might call up resident staff for help. Phone numbers are listed in the observing room.
Should minor problems ariseDetailed instructions on how to overcome possible minor problems ("bugs") with the software and hardware are available in the observing room. Most of them are well-known and can be solved by logging out, rebooting the CCD controller and logging in again.
English version : 18 March 2003
revised : 18.08.04