We announce the discovery of a new eclipsing hot subdwarf B + M dwarf binary, EC 10246-2707, and present multi-colour photometric and spectroscopic observations of this system. Similar to other HW Vir-type binaries, the light curve shows both primary and secondary eclipses, along with a strong reflection effect from the M dwarf; no intrinsic light contribution is detected from the cool companion. The orbital period is 0.1185079936 +/- 0.0000000009 days, or about three hours. Analysis of our time-series spectroscopy reveals a velocity semi-amplitude of K_1 = 71.6 +/- 1.7 km/s for the sdB and best-fitting atmospheric parameters of Teff = 28900 +/- 500 K, log g = 5.64 +/- 0.06, and log[N(He)/N(H)] = -2.5 +/- 0.2. Although we cannot claim a unique solution from modeling the light curve, the best-fitting model has an sdB mass of 0.45 Msun and a cool companion mass of 0.12 Msun. These results are roughly consistent with a canonical-mass sdB and M dwarf separated by a ~ 0.84 Rsun. We find no evidence of pulsations in the light curve and limit the amplitude of rapid photometric oscillations to < 0.08%. Using 15 years of eclipse timings, we construct an O-C diagram but find no statistically significant period changes; we rule out |P-dot| > 7.2 x 10^(-12). If EC 10246-2707 evolves into a cataclysmic variable, its period should fall below the famous CV period gap.
A spectroscopic analysis of SDSS J160043.6+074802.9, a binary system containing a pulsating subdwarf-O (sdO) star with a late-type companion, yields Teff = 70 000 +/- 5000 K and log g = 5.25 +/- 0.30, together with a most likely type of K3V for the secondary star. We compare our results with atmospheric parameters derived by Fontaine et al. (2008) and in the context of existing evolution models for sdO stars. New and more extensive photometry is also presented which recovers most, but not all, frequencies found in an earlier paper. It therefore seems probable that some pulsation modes have variable amplitudes. A non-adiabatic pulsation analysis of uniform metallicity sdO models show those having log g > 5.3 to be more likely to be unstable and capable of driving pulsation in the observed frequency range.
