Quad-Photosite Pixel Architecture for Phase Detection Autofocus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing phase detection autofocus systems using dual pixels for electronic masking are less efficient than single pixels in terms of Charge-Voltage conversion Factor (CVF), leading to reduced image quality and dynamic range, especially due to increased capacitance and reduced useful surface area.
Innovation Solution
An image sensor pixel array with four photosites separated by transverse and longitudinal isolator walls, featuring independent transfer gates and shared conversion nodes, optimized for minimal capacitance and electron collection efficiency, maintains a high Charge-Voltage conversion factor while minimizing surface area loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If dual pixels with two photosites are used for phase detection autofocus, then focusing capability is improved, but Charge-Voltage conversion Factor (CVF) decreases leading to reduced image quality and dynamic range
Solution Approach 1:
The patent combines phase detection autofocus functionality and main image sensing functionality into a single integrated pixel structure. The four photosites within each pixel work together to provide both phase shift measurement for autofocus and high-quality image capture, eliminating the need for separate dedicated focusing sensors and maintaining full pixel efficiency for both purposes simultaneously.
2Measurement precision
If photosensitive pixels are physically masked to enable phase detection, then focusing measurement is improved, but image quality deteriorates and corrective processing is required
Solution Approach 1:
The pixel is segmented into four distinct photosites arranged in a 2x2 configuration, with each photosite having independent charge-to-voltage conversion nodes. This segmentation allows selective reading of specific photosite combinations for phase detection while maintaining the ability to use all photosites for high-quality image capture, avoiding the quality degradation associated with physical masking.
3Adaptability or versatility
If dual pixels are integrated into the main image sensor, then focusing and imaging are unified, but surface area efficiency decreases and manufacturing complexity increases
Solution Approach 1:
The patent transitions from a linear dual-pixel arrangement to a two-dimensional 2x2 array of photosites within each pixel. This dimensional change allows four photosites to be packed into a compact square configuration, improving surface area efficiency while maintaining the phase detection capability through selective reading of adjacent photosite pairs.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This configuration reduces surface area loss by 26% and capacitance, resulting in a 17% decrease in the Charge-Voltage conversion factor, enhancing image quality and dynamic range compared to dual pixel systems.
Implementation Method 1
an array of four photosites... Each photosite has its own independent charge-to-voltage conversion node
Implementation Method 2
Each photosite has its own independent charge-to-voltage conversion node... the conversion factor CVF contributes to a good dynamic range of the image
Data Source
AI summary
An image sensor pixel may include an array of four photosites, a transverse isolator wall separating the array in two rows of two photosites, and a longitudinal isolator wall separating the array in two columns of two photosites. Both ends of the longitudinal wall may be set back relative to the edges of the array. First and second conversion nodes may be arranged in the spaces between the longitudinal wall and the edges of the matrix. Each conversion node may be common to two adjacent photosites, and an independent transfer gate may be between each photosite and the corresponding conversion node.