Both CCD and CMOS image sensors are manufactured in silicon foundries; base materials (silicon, silicon oxide and polysilicon) and equipment used are similar. CCD processers and imagers, however, have been optimized specifically for imaging applications for more than three decades. The main difference is architecture and design flexibility of CMOS sensors, especially for applications requiring dedicated signal or image processing that can be integrated on-chip, thus leading to a novel family of smart and compact imagers.
Image System Types
Image sensors are typically categorized into the following three types: CCD, CMOS APS and Pixim's digital-pixel-system technology. CCD sensors are an old technology, requiring complicated implementation systems and costly manufacturing processes that have limited expansion in many markets. CMOS APS (active-pixel-sensor) products were developed in the early 1990s as a result of emerging CMOS-manufacturing technology and have been the principal alternative to CCD sensors at the low end of the market. DPS (digital-pixel-system) products resulted from a technological breakthrough.
DPS technology, which was developed in the mid-1990s at Stanford University, provides enhanced capabilities over existing technology. Compared with CCD and CMOS APS, DPS platforms provide significant improvements in image quality and increased flexibility for camera designers and manufacturers.
CCD sensor technology has dominated image applications since the late 1960s. However, its analog design, is based on bucket-brigade architecture that requires a nonstandard fabrication process that is both complicated and costly to optimize.
It is difficult for commercially viable sensor arrays to integrate analog-to-digital conversion (ADC) for digital processing onto CCDs or to customize them by adding logic blocks to chips for specific applications. High voltage (+12 to +20 Volts) and power requirements shorten battery life and limit applicability in small-form-factor, portable devices. Moreover, these need external supporting components, including analog frontends, timing circuits, reference voltage, correlated double sampling, mechanical shutters and RAM, which results in large printed circuit-board real-estate requirements, contrary to miniaturization trends.
CMOS sensors were developed in the early 1980s. Passive-pixel-sensor (PPS) image sensors were the first products in this family to come to market. Large feature sizes available in existing CMOS technology allowed only single transistors and three interconnecting lines for each pixel. Speed and signal-to-noise-ratio was significantly lower than that of CCD sensors. In the 1990s, APS technology added amplifiers to each pixel, increasing sensor speed and improving signal-to-noise-ratio for a big advantage over PPS sensors.
When deep submicron CMOS technologies and microlenses appeared, APS became the alternative sensor technology. Low power consumption and near-standard manufacturing processes made it a competitor to CCD sensors for certain applications. APS technology, however, has inherent problems. Due to process variations that create non-uniformities in column-level ADCs and in-pixel amplifiers, large fixed-pattern noise (FPN) at high resolutions typically yields limited sensitivity--less than is required for many applications, including the security and film industries. Human eyes are particularly sensitive to image edges, and column-level ADCs amplify this noise.
Pixim chipsets apply DPS technology, which integrates ADCs into pixels; it can be manufactured using leading-edge semiconductor processes. Pixel arrays have significantly higher noise immunity than APS sensors because DPS technology employs digital readouts from each pixel. Additional image processing and camera functions implemented by Pixim provide a complete imaging solution in high-volume, commercially available chipsets.
DPS-image systems integrate sensing, memory and processing functions into two chips. This is especially important for imaging systems that require significant processing, where quality of output, small size, low power and portability are important. Until now, analog-to-digital conversion could be integrated only at the chip or column level.
Both approaches are common in APS-sensor solutions. For the chip-level approach, single conventional high-speed ADCs are integrated with sensors. For the column-level approach, one or more columns of pixel arrays has a dedicated ADC. ADCs are operated in parallel and, therefore, low-to-medium speed-conversion techniques must be used--single-slope, algorithmic, successive approximation or over-sampling.
By having separate ADCs for each pixel operating in parallel, ADCs operate at very low speed--a few thousand samples per second. This lessens noise and reduces power requirements.
The large bit stream used to read ADCs is supported by on-chip RAM. This feature enables much faster, more accurate readout characteristics. DPS sensors consist of "m x n" DPS arrays, ADCs and RAM. A separate chip incorporates digital signal processing (DSP) and I/O; sensor cores are powered by single low-voltage power supplies. Manufacturing is simpler than for CCDs and great miniaturization is realized.
Solid-state capture and display image systems are now integral components in an enormous variety of electronic solutions. They can be found in mass demand in markets as diverse as digital cameras, mobile phones, personal computers, factory automation, security, medicine and biotechnology. This reflects a convergence of technological advances in optics, semiconductors, electronics and embedded processing.
Next-generation advanced imaging systems delivering advantages of DPS-imaging have entered markets that are currently benefiting from DPS technology including the following:
- Security and Surveillance--Wide-dynamic range of DPS technology automatically adjusts to variable, real-world lighting conditions. High-speed video-capture rate of DPS enhances image detail. Integration of all sensor and processor functions in two chips allows compact cameras to be deployed in new security applications.
- Network and IP Cameras--In addition to security, IP-based cameras are being developed for broadband videoconferencing applications. Backlighting from windows makes it difficult to get high-quality images with typical cameras; DPS technology provides an ideal solution.
- Machine Vision--DPS technology gives manufacturers enhanced observation and detection capabilities along assembly lines and in other production settings where reflection causes problems. DPS technology holds great potential in manufacturing and industrial settings. The market for industrial imaging systems is pegged at more than US$500 million annually.
DPS technology is also ideal for additional applications where operation under extreme-lighting conditions is important, such as advanced automobile systems that help drivers avoid collisions and by determining whether to deploy vehicle airbags, depending on severity of impact.