Answers are grounded in this paper and cite their sources.
Correlated multiple sampling (CMS) is a popular oversampling technique that is widely integrated into recent low-noise CMOS image sensors (CISs). In its common implementation equal weights on both reset and signal samples are applied to compute a low-noise signal sample. In this article, we analyze the effectiveness of four oversampled techniques on CIS read noise performance. We compare CMS, noise optimized correlated multiple sampling (NOCMS), skipper multiple sampling (SMS), and noise optimized skipper multiple sampling (NOSMS). We verify our model on a sensor offering CMS functionality. We derive noise parameters based on regular correlated double sampling (CDS) measurements, and then compute CMS noise estimates using our model. We obtain good agreement between model and measurements. We then use model parameters of two sensors to compare the four oversampling techniques. We point out that floating diffusion (FD) leakage current becomes a dominant noise factor in CMS, which NOCMS can greatly reduce. Finally, we conclude that SMS and NOSMS have more potential to further reduce read noise than CMS or NOCMS.
No takes yet. Share an insight, caveat, or question.
Suess et al. (2022) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: