PWM Pixel Sensor Architecture for High Dynamic Range Imaging

The patent transitions from a single-pixel 2D array to a 3D stacked architecture where multiple pixel layers are vertically integrated. Each layer captures light at different intensities, effectively adding a vertical dimension to light intensity measurement. This enables simultaneous capture of both dark and bright areas by distributing pixel responses across multiple layers, thereby extending dynamic range without requiring each individual pixel to have variable capacity.

The image sensor is divided into multiple pixel layers, each with its own photodiode and readout circuitry. The pixel array is segmented across vertical stacks, where different layers can be optimized for different light intensity ranges. This segmentation allows the system to capture a broader dynamic range by combining responses from multiple specialized layers rather than relying on a single pixel type.

The patent employs periodic scanning of pixel rows combined with a rolling shutter mechanism that sequentially activates different pixel layers at different times. This periodic activation allows the system to capture light over extended effective exposure times while maintaining temporal resolution. By staggering the exposure timing across multiple layers and scanning them sequentially, the system achieves high sensitivity without requiring any single layer to be exposed for a prolonged duration, thereby reducing motion blur.

The sensor implements dynamic control of pixel activation and readout timing through a rolling shutter mechanism. Different pixel layers can be activated and read out at different times, allowing the system to adapt exposure timing to scene conditions. This dynamic approach enables the sensor to capture sufficient light for low-light conditions while minimizing the effective exposure time for any given scene, thereby reducing motion artifacts.

The patent resolves this contradiction by moving from a single-layer 2D pixel array to a multi-layer 3D stacked architecture. High pixel density is achieved in the horizontal plane while additional vertical layers provide increased light storage capacity. The full-well capacity is effectively distributed across multiple layers, so each pixel maintains high spatial resolution while the stack as a whole captures a broader dynamic range by summing electrons across layers.

The patent implements a nested structure where multiple pixel layers are vertically integrated within a compact footprint. Each layer contains complete pixel functionality (photodiode, transfer gate, floating diffusion, readout circuitry), and layers are nested one above another. This nesting allows the system to pack high pixel density horizontally while utilizing vertical space for additional light storage capacity, effectively combining high resolution with extended dynamic range.

The patent segments the pixel array into multiple independently scannable rows and layers. Instead of scanning the entire array sequentially in a single rolling shutter pass, the system can scan different rows or layers in parallel or use overlapping scan patterns. This segmentation of the scanning process reduces the effective scan time per frame, allowing the system to maintain motion artifact reduction benefits while increasing frame rate.

The patent implements preliminary charge transfer and accumulation in floating diffusion nodes before the actual readout process. Electrons are transferred from photodiodes to floating diffusions during the exposure period, and this preliminary action prepares the signal for rapid sequential readout. By pre-positioning charges in ready-to-read locations, the system reduces the time required for the rolling shutter scan, thereby improving frame rate without sacrificing motion artifact reduction.