Modeling and Mitigation of Supply-Induced Row-wise Fixed-Pattern Noise (RFPN) for Low-Noise QDPD-Based SWIR Imaging

Authors:

Abstract
Power supply noise in thin-film photodiode (TFPD)-based short-wave infrared (SWIR) image sensors induces row-wise fixed-pattern noise (RFPN) through the capacitive coupling of ITO bias voltage fluctuations to the floating diffusion (FD) node. While correlated double sampling (CDS) provides partial suppression of this noise, its frequency-dependent response leaves significant residual noise at minimum rejection frequencies. In this work, we develop an analytical model of the supply-induced noise coupling mechanism and characterize the power supply rejection ratio (PSRR) frequency response. To mitigate this noise, an optical black (OB) column-based compensation approach compatible with TFPD image sensors is demonstrated. By utilizing physically shielded OB columns as a row-wise reference, static subtraction effectively suppresses the supply-induced RFPN. Validated by silicon measurements, the proposed scheme improves the peak PSRR significantly from -0.82 dB to -30.3 dB at 350 kHz under a 1.25 μs CDS time in dark imaging conditions. In addition, we analyze the compensation degradation observed under bright conditions, attributing it to an illumination-induced increase in photodiode capacitance; this insight underscores the necessity of engineering signal-independent photodiodes or transitioning to a four-transistor (4T) pixel architecture for future TFPD SWIR sensors.