2 The data

 

2.1 Description of the data set

The data have been collected at La Silla ESO in Chile with a 40 cm telescope (f/10) equipped with a thick CCD camera composed of $8\times2$ CCD chips of $400\times579$ pixels with scale of $1.21\hbox{$^{\prime\prime}$}$/pixel (Arnaud et al. 1994b; Queinnec 1994 and Aubourg et al. 1995). The gain of the camera was 6.8 e^-/ADU with a read-out noise of 12 photo-electrons. For the 1991-92 campaign only 11 chips out of 16 were active. Due to technical problems, we only analyse 10 of them. The monitoring has been performed in two wide colour bands (Arnaud et al. 1994a). Exposure times were set to 8 min in red ($\langle{\lambda}\rangle = 670$ nm) and 15 min in blue ($\langle{\lambda}\rangle =$ 490 nm). As the initial goal was to study microlensing events with a short-time scale (Aubourg et al. 1995), up to 20 images per night in both colours are available. A total of 2000 blue and red images were collected during 95 nights spread over a 120 days period (18 December 1991 - 11 April 1992). The combined CCD and filter efficiency curves as shown in Grison et al. (1995) lie below 15% in blue and below 35% in red. Bias subtraction and flat-fielding have been performed on-line by the EROS group.

The seeing varies between 1.6 and 3.6 arcsec with a mean value of 2.9 arcsec (typical dispersion 0.5 arcsec). It should be emphasised that the observational strategy (exposure time) has been optimised for star monitoring. In other words, this means that the photon noise associated with the mean flux (typically 280 ADU per pixel in red and 100 ADU in blue) is relatively large: 6.6 ADU in red and 3.8 ADU in blue. To apply the Pixel Method to this data set, we take advantage of the large number of images available per night, increasing the signal-to-noise ratio with an averaging procedure.

2.2 Absolute calibration

The procedures described below are performed with respect to a reference image. The correspondence between the flux measured on these images and the magnitude, deduced from Grison et al. (1995), is as follows:

$$m_B = -2.5 \log \phi_B + 24.8$$ (3)

$$m_R = -2.5 \log \phi_R + 24.9$$ (4)

where $\phi_B$ and $\phi_R$ are the flux of a star in ADU in the blue and red respectively. Note that the zero point is about the same in the two colours, whereas the background flux is much larger in red than in blue. The correspondence with the Johnson-Cousins system can be found in Grison et al. (1995).

The aim of the whole treatment presented below is to obtain pixel light curves properly corrected for variations of the observational conditions. The PEIDA package used by the EROS group was adapted for pixel monitoring. This treatment is applied to the first CCD campaign (1991-92) of the EROS group on the LMC bar, i.e. 10% of the whole data set analysed in Renault (1996).