Evaluation of correlated double sampling used with solid state imagers

Abstract

Correlated double sampling (CDS) is a widely used signal processing technique for removal of the Nyquist (reset) noise which is associated with charge sensing circuits employed in a solid state imager. In this thesis work, the power spectral density at the output of a correlated double sampling circuit with first-order low-pass filtered white noise at the input is calculated. A circuit constructed with discrete elements is made to simulate the output stage of a charge-coupled device (CCD). A low-pass filtered wide-band noise from a noise generator is added to the reset reference level when the output signal from this simulator is sampled by the correlated double sampling technique. The experiment measurements show that only about 10% of the noise power measured by simple sampling is obtained when CDS is employed. An autoregressive (AR) model is assumed to fit the sampled data and a recursive algorithm, based on least-squares solutions for the AR parameters using forward and backward linear prediction, is adopted for spectrum estimation. Some conclusions on choosing the bandwidth of the low pass filter for optimum operation is also included.