next previous
Up: The strange case of


2 Observations and data reduction

We observed HD149404 over a period of 3 years with various instruments. A first set of spectra, covering roughly 45Å centred on the HeI $\lambda $4471, HeII $\lambda $4686 and NIII $\lambda \lambda $4634-41 lines, was obtained in June 1997 with the 1.4 m Coudé Auxiliary Telescope (CAT) at ESO, La Silla, feeding the Coudé Echelle Spectrometer (CES) and the Long Camera (LC). Additional data (covering 20Å around the HeI $\lambda $4471 and HeII $\lambda $4686 lines) were taken with the same spectrograph equipped with the Very Long Camera (VLC) in April-May 1998 and July 1998. In all cases, the detector was ESO CCD#38, a Loral 2688$\times$512 CCD with pixel size $15\,\mu$m $\times$ $15\,\mu$m. The nominal resolving power was 70000-80000. Typical exposure times were of the order of 15 min and the average signal-to-noise ratio was about 150.

In May 1999 and May 2000, we observed HD149404 with the Fiber-fed Extended Range Optical Spectrograph (FEROS, Kaufer et al. 1999) at the ESO 1.52m telescope at La Silla. The spectra cover the wavelength range 3650-9200Å with a resolving power of 48000. The detector was a 2k $\times$ 4k EEV CCD with pixel size $15\,\mu$m $\times$$15\,\mu$m. The typical exposure time was 2 min and the S/N ratio at 6000Å is about 160.

Five more spectra were taken in May-June 1999 with the Bench-Mounted Echelle spectrograph (BME) fed by the 1.5m Ritchey-Chrétien telescope at CTIO. These data covered the spectral range 3750-5800Å. The typical S/N ratio was 120 at 4620Å for exposure times of about 20min. The detector used was a TEK CCD with 2048 $\times$ 2048 pixels of $24\,\mu$m $\times$ $24\,\mu$m and the resolving power was 45000.

Finally, twelve echelle spectra were obtained in February 2000 with the 1.2 m Euler Swiss telescope at La Silla, equipped with the Coralie spectrograph and a 2k $\times$ 2k CCD with pixel size $15\,\mu$m $\times$ $15\,\mu$m. The Coralie instrument is an improved version of the Elodie spectrograph (Baranne et al. 1996). These observations covered the spectral range 3875-6800Å with a resolving power of 50000. The integration times varied between 30 and 45 min and the typical S/N ratio at 5000Å was 120-150.

The CAT data were reduced in a standard way with the MIDAS software developed at ESO. The spectra were normalized by fitting a low order polynomial to the continuum. The FEROS spectra were reduced using the appropriate context of the MIDAS environment together with several dedicated routines kindly provided by Dr. O. Stahl. The extracted FEROS spectra were rectified by fitting a polynomial of degree 4 or 5 to the continuum over a range of a few hundred Å at once.

The CTIO data were reduced using the IRAF[*] package, following the recommendations of the BME User's Manual. We used the smoothed projector flat field exposures as a first guess of the continuum, and the spectra were finally normalized using a low-order polynomial.

The Coralie data were reduced at the telescope, using the INTER-TACOS (INTERpreter for the Treatment, the Analysis and the COrrelation of Spectra) software developed by D. Queloz and L. Weber at the Geneva Observatory (Baranne et al. 1996). Next, the normalisation was done by dividing by the blaze fits and finally by fitting a polynomial to the continuum.

Whenever several exposures were available for the same night, we averaged them to increase the S/N ratio. Some preliminary results based on our data were already presented by Nazé et al. (2000).


  \begin{figure}
\par\includegraphics[width=8.4cm,clip]{MS10412f2.eps}
\end{figure} Figure 2: Variations of the equivalent width of the HeI $\lambda $4471 and HeII $\lambda $4542 lines and of the log(W') criterion for the primary star as a function of orbital phase


next previous
Up: The strange case of

Copyright ESO 2001