3. Observations and reductions
Crt has been observed in 6 out of 9 nights in May 1997 at the National Astronomical Observatory, Rozhen, of the Bulgarian Academy of Sciences, using the Coudé spectrograph at the 2m RCC-telescope. The spectra were recorded with the ASTRO-550 system manufactured by Ista Ltd. (Berezin et al. 1991) with a virtual phase CCD developed and produced by Electron Corp., St. Petersburg, Russia) (see also Shcherbakov et al. 1995). The camera was provided by the Helsinki University Observatory. The pixel size is . The slit-width was set to 300 m, so that the FWHM of a typical comparison line is about 2 pixels. The grating 632/ was used, providing a dispersion of 0.01 nm per pixel. The total length of the recorded spectra is thus about 6 nm. The observed wavelength region is centered at 490 nm; in some nights additional spectra centered at 505 nm were taken. The integration times were between 5 and 10 minutes. The signal-to-noise ratio of the spectra is typically around 150. Due to the position of Crt, observations were only possible very early in the night at very high zenith-distances. The first of our spectra was taken too early, and is heavily contaminated by a superposed solar spectrum.
During observations and reductions the 3A-software package (Ilyin 1993) has been used. The reductions contain the usual steps of bias subtraction, flat-fielding, spectrum extraction and wavelength calibration. The spectrum extraction was done by an algorithm similar to the optimal extraction described by Horne (1986), getting rid of the cosmics at the same time. The wavelength calibration is based on typically 25 Fe and Ar lines, whose wavelengths are taken from the catalogue by Hirata & Horaguchi (1995). The typical error of the dispersion curve (being a polynomial of degree 1) is 0.26 pm, corresponding to 0.16 km s-1 at 490 nm.
The external accuracy of the spectra and the reductions was checked using spectra of the Sun (evening sky). One spectrum in each of the two spectral regions was available. The positions of solar absorption lines were measured by approximating them by Gaussians. The line shifts then were computed by comparing with the wavelengths given by Pierce & Breckinridge (1973). The mean shifts from many lines are: km s-1 and km s-1 for the spectra at 490 nm and 505 nm, respectively. There are thus no significant systematics, and the main source of RV errors for sharp lined spectra is the error of the dispersion curve.
© European Southern Observatory (ESO) 1997
Online publication: March 26, 1998