The reported survey was conducted as a supplementary
project during a long observing run devoted mainly to a search for eclipsing
detached binaries in the globular cluster M 4 (Kaluzny 1996).
The central part of NGC 6397 was monitored with the CTIO 0.9-m telescope
and Tektronix 2048 No. 3 CCD. The field of view of the camera was
with scale of 0.396 arcsec/pixel.
Observations were performed using the Johnson B and V filters.
The exposure time ranged from 150 to 300 s for the B-band, and
from 90 to 200 s for the V-band.
To search for variable stars we used 83 frames in B and 142 frames in V.
All the B-band exposures and most V-band exposures were collected on
the nights of July 8/9 and July 9/10, 1995. A few additional exposures
in the V-band were collected on three subsequent nights.
A condensed log of observations is given in Table 1 (click here). A more detailed log
was submitted to the editors of A&A (see Appendix A). Frame numbers
quoted occasionally below refer to that log.
Table: Log of observations of NGC 6397. The third column gives a number
of frames collected in both filters
The preliminary processing of the raw data was made with IRAF
.
The flat-field frames were prepared by combining sets of 10-15
frames obtained by observing an illuminated screen in the dome.
The reduction procedures reduced total instrumental systematics to
below 1% for the central 
area
of the images. Some systematic residual pattern at the 
level was left near borders of the images.
Stellar profile photometry was extracted using DoPHOT (Schechter
et al. 1993). We used DoPHOT in the fixed-position mode.
The stellar positions were provided based on a list obtained by
reduction of a ``template" image. Two consecutive images of
a good quality (frames #1703 and #1704; both with 
and 
)
were combined to produce template for the V filter observations.
An individual image (frame #1705, 
,
) served as a template for observations in the B
filter.
The images collected with the CTIO 0.9-m telescope
show a significant positional dependence of the point spread
function. To cope with this effect we applied a procedure similar to that
described in details in Kaluzny et al. (1996). In short, each
analyzed frame was divided into a 
grid of overlapping
sub-frames. An instrumental photometry derived for a given sub-frame of the
frame was transformed to the common instrumental system of
the ``template" image.
The data bases for the V and B filers contained light curves
for 12259 and 17654 stars, respectively.
Photometry derived from ``template" frames was used to
construct the color-magnitude diagram (CMD) for the monitored field.
The instrumental photometry was extracted using DAOPHOT/ALLSTAR
(Stetson 1987). Following Walker (1994) we
selected a Moffat-function point spread function, quadratically varying with
X and Y coordinates.
The instrumental photometry was transformed to the standard BV
system using relations:
where lower case letters correspond to the instrumental magnitudes.
The color terms of the transformation were determined based on
observations of standard stars from the Landolt (1992) fields.
The zero points were calculated using photometry
of 9 local standards from NGC 6397 field (Alcaino & Liller
1986). The derived CMD of the cluster is shown in Fig. 1 (click here).
Poor measurements were flagged in the
photometry files generated by DAOPHOT/ALLSTAR. A given measurement was
considered to be poor when the formal error of photometry was 2.5 times
or more larger than the average error of photometry for other
stars of comparable magnitudes. Stars for which either V or B photometry
was flagged as poor, were not plotted in Fig. 1 (click here).
Stars with V<12.0 were in fact badly overexposed on the template
image for the V filter. Their V photometry was extracted from a
single, shortly exposed image (frame #1911, FWHM = 1.85, 
).
Figure 1: The color-magnitude diagram for the monitored field.
The presented photometry was extracted from a pair of BV frames
with the help of DAOPHOT. The V-band photometry for the brightest stars
was extracted from an additional shortly exposed image. Stars
with unreliable photometry and V>14.0 were not plotted
Another CMD of the monitored field was constructed based on average
magnitudes of stars included in the data bases for both filters.
To calculate average magnitudes we selected only relatively good
frames (
, clear sky during observation).
A total of 49 and 97 frames were selected for B and V bands,
respectively. The average magnitudes were then calculated for stars
with at least 15 and 35 measurements in the B and V bands,
respectively. The 3 most strongly deviating measurements were rejected while
calculating average magnitudes. Moreover, we used only measurements
to which DoPHOT assigned type equal 1 (stellar objects of relatively
high S/N). The resultant CMD is presented in Fig. 2 (click here).
This photometry covers a smaller range of magnitudes than that
presented in Fig. 1 (click here). However, the CMD based on average magnitudes
has a better internal accuracy and shows the better defined
main-sequence of the cluster.
The tables with photometry presented in Figs. 1 (click here) and 2 (click here) are available in the electronic form from the CDS (see Appendix A). We emphasize that the presented CMDs were obtained as a by-product of a survey for variable stars. These observations were not aimed of getting a deep and accurate photometry suitable for study of the cluster properties. All images of the cluster were obtained under a rather unfavorable conditions (see Table 1 (click here)). The zero points of the presented BV photometry are based entirely on the local standards (Alcaino & Liller 1986). These local standards were set up using a photoelectric photometer.
Figure 2: The color-magnitude diagram based on the averaged photometry derived
with the help of DoPHOT