An imager circuit includes an array of pixels, each pixel including a sensor (photodiode) connected to an input terminal of a comparator. The comparators of each pixel row have output terminals connected to a latch. A counter generates a sequence of digital values that are transmitted to a digital-to-analog converter (DAC) and to the latch of each row. The DAC generates a ramp voltage that is transmitted to a second input terminal of each pixel's comparator. The comparators of a selected pixel column are enabled to generate output signals when the ramp voltage equals each pixel's voltage, causing the associated latches to capture the current digital values. The comparators are formed such that each pixel row shares a cascode mirror circuit that detects differential currents in data line pairs connected to each pixel in that row.
A semiconductor image pickup device's pixel circuits each include a photodetection element, and first and second current mirror circuits provided as current generation circuit. The photodetection element generates at a node a photocurrent corresponding to a quantity of light received. The first current mirror circuit passes first and second currents corresponding to the photocurrent to an internal node and an output node, respectively. The second current mirror circuit is connected to generate a fourth current corresponding to a third current passing through the internal node and also allow a differential current between the second and fourth currents to flow through the output node. Each pixel circuit has its internal node electrically connected by a resistance component, which implements an inter-pixel connect, to the internal node of at least one of adjacent pixel circuits. An output current will be a current corresponding to a relative difference between the photocurrent of the pixel circuit of interest and the third current corresponding to an average quantity of light received at adjacent pixels. Thus a high precision, miniaturized semiconductor image pickup device can be provided that can detect bright and dark portions in sufficient contrast even if there is a large brightness distribution within a single field of view.
