Specific factors of a baseline Rn-222 detector installed at Cape Point, South Africa, were studied with the aim of improving its performance. Direct sunlight caused air turbulence within the instrument, resulting in 13.6% variability of the calibration factor. Shading the instrument eliminated this effect. A residual temperature dependence of the calibration factor was reduced to negligible levels with an improved photomultiplier tube. A superior detector head permits field servicing of the instrument, and has reduced one component of the instrumental background by a factor of 2. The other component probably constitutes thoron emissions from the stainless steel walls. The detection limit of the instrument could be reduced from its current 33 to 20 mBq m(-3) if the thoron were to be eliminated.