DCDS weighted averaging theory and development for improved noise filtering in scientific CCD applications

Readout noise is a key factor in the performance of optical systems based on charge coupled devices (CCDs). Recent developments have shown that digital correlated double sampling (DCDS) using weighted averaging may provide a further reduction in the system readout noise. This paper describes recent advances in noise filtering using DCDS. Particular emphasis is placed on optimising weighted averaging filters to reduce 1/f noise and the characterisation of system performance when using the unsettled samples within the pixel period. Experimental results are presented and compared with theoretical predictions based on the extracted noise spectrum. The analysis provides a detailed study of the relationship between the 1/f corner frequency, the pixel frequency and weighted averaging technique in comparison with the theory of matched filters. Furthermore, the results include a comparison of the noise profile with measured and simulated noise patterns. Key system metrics, including linearity and gain stability, have been characterised and are presented to confirm the suitability of this technique for high-performance scientific applications.