Cyclostationarity-based model for noise analysis of charge amplification with correlated double sampling

Abstract

A cyclostationarity-based analytical model for noise analysis of charge amplification with correlated double sampling (CDS) is proposed. It is established by signal description and transformation with considerations of input referred noise sources. It predicts that CDS operation has different noise effects: it suppresses reset noise and attenuates 1/f noise from the equivalent input noise voltage, but it has no noise-reduction effect on the input referred noise current. Parametrical analysis can also be performed by using this model. For example, by varying integration duration, the signal-to-noise ratio (SNR) will first be proportional to t int , then to t int , and finally limited by the 1/f noise component becoming predominant. This proposed model is validated by comparison of analysis results with those from temporal noise analysis using SpectreRF of Cadence TM CAD tool. Unlike time-hungry temporal noise analysis, this model provides rapid results.