High-precision time-resolved CCD photometry

High-precision time-resolved photometry and detection of low-amplitude stellar oscillations are limited in precision by many factors. The noise term for a typical time-series observation contain observational noise and errors as well as instrumental noise and reduction noise. This paper describes some of the problems related to high-precision photometry, esecially focusing on CCD photometry both on focused stars (in semi- and non-crowded fields) and defocused field stars. We show that atmospheric transparency variations and instrumental drift are the main noise sources at low frequencies (below 1-2 mHz), and that CCD photometry is the best way of doing differential measurements in order to detect oscillations at low frequencies. We also describe a special reduction program, Multi Object Multi Frame photometric package (MOMF), which can reduce time-resolved photometry in half-crowded fields. It has been developed to limit the photometric noise and runs fully automatically, with only a few interactive steps. We compare the theoretical description of noise and problems in high-precision photometry with CCD time-series observations in a non-crowded test field observed with the Danish 1.5-m at ESO La Silla, Chile in 1990. In addition we describe observations of defocused double stars observed with the Nordic Optical Telescope in 1991.