For a few years, new kinds of setups for residual stress analysis by X-ray diffraction have been commercialised by manufacturers with two linear Position Sensitive Scintillation Detectors (PSSD) including Charge-Coupled Device (CCD) sensors. Although these equipments allow an important reduction of acquisition time, some questions subsist on their measurement reliability, especially on the raw profile corrections and on the associated statistical uncertainty. One of them concerns the required gain correction because of the weak spatial homogeneity of these detectors. In fact, a bad knowledge of sensor noise does not permit a good correction of the raw patterns which can significantly affect the results of stress measurements. In this study, an original statistical analysis is proposed to visualize and analyse the various kinds of intrinsic noise of PSSD detectors, and especially the most important one: the dark noise. Based on the results of this investigation, a method for gain correction is then proposed. The method, easy to apply, permits a better correction of the sensors defects without increasing acquisition time. This analysis also allows a better understanding of the sensors behaviour and thus an optimisation of the acquisition.