Answers are grounded in this paper and cite their sources.
Conventional analog correlated multiple sampling (CMS) techniques become increasingly area-intensive when implemented in ping-pong column readout architectures because dedicated sampling capacitors are typically required for each interleaved readout path. This paper presents a compact analog ping-pong correlated multiple sampling (PP-CMS) architecture for cooled global-shutter image sensors that addresses this limitation by reusing storage capacitors across the two ping-pong paths. The proposed architecture enables 4-point CMS operation while requiring only two additional capacitors and three switches, without modifying the original ping-pong readout operation. The DDS-only column occupies approximately 550μm, and the additional PP-CMS circuitry increases the total column length by 50μm to approximately 600μm, corresponding to a 9.1% length overhead relative to the DDS-only implementation. A mathematical noise model is developed to analyze the thermal- and flicker-noise behavior of the proposed architecture and to clarify its trade-off relative to conventional switched-capacitor CMS. Since the proposed PP-CMS applies averaging directly to the double-delta sampling (DDS) output, it does not introduce the additional low-frequency attenuation associated with the reset–signal differencing operation of conventional SC-CMS; therefore, it is intended to improve the area–noise trade-off in ping-pong ROICs rather than to maximize intrinsic low-frequency flicker-noise suppression. A prototype readout integrated circuit (ROIC) fabricated in a 0.18-μm CMOS process was experimentally characterized at 80 K. The measured output noise was reduced from 93.5 μVrms for the DDS-only operation to 73.8 μVrms with 4-point PP-CMS, corresponding to a 21% reduction. These results demonstrate that the proposed PP-CMS architecture provides an effective trade-off between silicon area and readout-noise reduction for compact ping-pong ROIC implementations in cooled infrared imaging systems.
No takes yet. Share an insight, caveat, or question.
Şanli et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: