Electronic Design

CMOS And CCD Image Sensor Breakthroughs Promise A "Bright" Future

Improved design and fabrication methods teamed up with high-speed processors are spawning low-cost, high-performance CMOS and CCD imagers across a wide range of applications.

The latest generation of CMOS and charge-coupled device (CCD) image sensors features wider spectral bandwidths, higher sensitivity levels, lower noise operation, and smaller form factors. Better fabrication processes help lower costs. And, novel architectures are injecting greater flexibility and versatility into circuit designs.

As a result, imaging sensors now find homes in mobile phones, notebook and laptop PCs, digital still cameras, video games, toys, medical devices, automobiles, security, industrial, and many other applications. According to IC Insights, CMOS and CCD imagers will see a compound annual growth rate (CAGR) of 14% over the next five years. Both types of sensors are finding wide use, but prognostications for CMOS imagers are particularly strong. Forecasts show that they will garner a 73% market share by 2012, up from 58% last year.

Like most electronic devices, performance and cost will continue to be the main issues for CMOS and CCD imagers. Although CMOS imagers are predicted to garner more applications, there’s still a need for CCD imagers in applications that require high performance levels. It isn’t simply a question of which type of imager is better. Depending on the application, a CMOS or CCD sensor may be the best choice based on performance and cost parameters.

CMOS imagers, which are generally less expensive than CCD imagers, no doubt can be found in many consumer electronics items that stress low cost. Their performance is on the rise— they’ve been making inroads into automotive safety applications while encroaching into the CCD imaging arena, where performance levels are acceptable but at a lower cost.

CCDs feature the higher performance needed in industrial and machine-vision inspection applications, as well in security systems and scientific and military aerospace applications. They can also be found in niche apps like astronomy and the medical realm. Cost is dropping, too, while still making impressive leaps in performance that’s outstripping CMOS imagers (see “CCDs: Performance That Can’t Be Beat”).

CMOS imagers are meeting many system requirements determined by multiple application parameters, such as wider bandwidths and global-shuttering capability. And new design and manufacturing proposals and implementations will drive their performance even higher.

One novel idea from NASA’s Jet Propulsion Laboratory (JPL) substantially reduces imager diffusion crosstalk. The researchers propose adding two implants in each CMOS pixel that would affect vertical isolation between the MOSFETs and the pixel photodiodes used in their imager (Fig. 1). They argue that this separation makes it possible to optimize both the MOSFET and the photodiode performance, eliminating or vastly reducing crosstalk and noise, while increasing sensitivity, spatial resolution, and color fidelity.

Image synchronization and operation under often difficult and unfavorable conditions, particularly in machine-vision automated inspection applications, is a big challenge facing CMOS imager designers. The industry has traditionally relied on CCD imagers using interline-transfer techniques to deliver high-speed shuttering for crisp images.

Recent CMOS imager advances have enabled these sensors in machine-vision applications. With parallel outputs, windowing, and on-chip integration, some CMOS image sensors now offer capabilities that rival those of CCD imagers for some machinevision applications.

For instance, CMOS sensors from Cmosis feature globalshuttering capability. Thanks to its pipelined global-shutter pixel technology, imaging systems can capture the next frame during the readout process. Cmosis achieves this by incorporating a storage node in each of the image sensor’s pixels, to which the signal is transferred after the image capture step.

The storage node has an extremely low parasitic light sensitivity. Each pixel can be read out with low noise and with a wide dynamic range. The firm developed fast analog-to-digital converters (ADCs) located in the sensor’s pixel columns.

Dalsa Corp. has come up with interline-transfer CMOS imagers that can also deliver high-speed shuttering capability. These sensors provide the sensitivity, signal capacity, noise performance, and dynamic range that’s required in many machinevision applications.

Photonfocus employs its patented LinLog technology in the A1312 CMOS imager for fast shuttering capability and a wide dynamic range of up to 120 dB. The sensor features 8- by 8-µm pixels in a 1312- by 1028-pixel format and operates at 110 frames/s with full resolution.

As CMOS imager pixel sizes shrink, maintaining imager performance and image quality becomes a tougher task. One option has been backside illumination. Working with Taiwan Semiconductor Manufacturing Corp. (TSMC), OmniVision believes it has found the key with the OmniBSI approach (Fig. 2). The company is able to produce 8-Mpixel devices from a 1.4-µm process for mobile phones.

Sony has also seen success with backside illumination. The company has produced a 5-Mpixel device on a 1.75-µm process for mobile phones, digital cameras, and camcorders. And STMicroelectronics, working with France’s CEA Leti and Tracit Technologies, has demonstrated the feasibility of manufacturing 3-Mpixel CMOS imagers on a 1.45-µm process using backside illumination.

Continue to page 2

Penetration of higher-performance CMOS imagers into the consumer electronics market is evidenced by STMicroelectronics’ first 0.25-in., optical-format 3.15-Mpixel imagers. These devices feature raw Bayer sensors with integrated depth-of-field capabilities. According to the company, the VD6853 and VD6803 imagers are made on a 1.75-µm process and provide excellent image quality at focus distances down to 15 cm (Fig. 3).

Embedded image-enhancement filters, including fourchannel anti-vignette circuitry, balance uneven illumination or defect correction on the fly. The imagers can be used in mobile phones, laptop cameras, toys, and even machine-vision applications. The VD683 features a 10-bit parallel interface. The VD6803 has a CCP2 interface.

Recently, Samsung introduced a 0.25-in. optical format 1.2-Mpixel system-on-a-chip (SoC) imager in a 1280- by 960-pixel format. The S5K4AW CMOS imager for notebook and desktop computers addresses the specific needs of high-definition real-time video applications by binning images in 2-by-2 groups. It also can display images in a standard VGA format without the need for cropping.

At this year’s IEEE International Solid State Circuits Conference (ISSCC), Canon discussed a 3.3-Mpixel CMOS image sensor that promises higher-quality video and imaging for mobile devices. It achieves this performance by using new column readout circuits to lower noise by 30%.

At the 2009 Mobile World Congress in February, OmniVision Technologies demonstrated an 8-Mpixel CMOS imager for mobile phones on a platform that combines the sensor’s capabilities with Fujitsu Microelectronics’ mobile Milbeaut M-5M0 image signal processor. The unit is based on OmniVision’s OmniBSI architecture and is manufactured on a 1.4-µm process.

Earlier this year, OmniVision introduced the OV7740, its latest product for notebook Web cams. This small-profile, low-power CMOS imager delivers better sensitivity at 6800 mV/lux-s. It can operate at 60 frames/s with VGA resolution and at 120 frames/s with QVGA resolution. The sensor can be used in automotive safety applications as well.

The company additionally introduced its CameraCube imager for ultra-slim mobile phones. This 3D reflowable, total camera solution combines the full functionality of a single-chip image sensor, embedded image processor, and wafer-level optics in a compact small-footprint package that’s as small as 2.5 by 2.9 by 2.5 mm. Two such devices are being offered. The OVM6680 offers 400- by 400-pixel resolution, and the OVM7690 provides VGA resolution.

The latest generation of digital single-lens reflex (SLR) cameras is proving to be an enticing target for CMOS imagers. The Aptina imager from Micron Technology is one of the more notable image sensor products. The company is reportedly making its MT9V113M02STC 9-Mpixel VGA wafer-level CMOS camera module available for use in digital SLRs.

Sony has developed a 35-mm full-size CMOS imager for digital SLRs with 24.81 effective Mpixels. It employs the company’s column-parallel analog-to-digital conversion technique. With this technique, each pixel column has its own analog-to-digital converter (ADC) to minimize image degradation caused by the noise that arises during analog processing, while simultaneously delivering extremely high signal-conversion speed.

An important CMOS-imager advance from Eastman Kodak, aimed at mobile phones and digital SLRs, involves next-generation color filter patterns that more than double the sensitivity of both CMOS and CCD imagers. The method departs from the widely used Bayer filter pattern with an arrangement of red, green, and blue pixels (which was also developed by Kodak) by adding a fourth pixel that has no pigment on top.

CMOS imagers are also paving the way to new applications for consumers and professionals. Consider the MicroExplorer digital inspection camera from RIDGID (see this issue’s cover). This device employs a CMOS sensor on the end of a flexible cable attached to a handheld device with a color LCD.

Camera-based driver-assistance systems using CMOS imagers deliver huge safety benefits. Together with the latest powerful processors, DSP chips, and software algorithms, these imaging systems are set to become the hub for operational video, radar, and lidar data streams that will act as the brains of future cars.

They can sense driver fatigue through images of the driver’s face and recognize road signs. Also, they provide warnings while backing up and parking in addition to lane-departure warnings, blind-spot detection, front and rear vision, pedestrian detection, and night-vision assistance.

Examples of automotive CMOS imagers abound. Some of the most recent and impressive devices have come from Melexis, OmniVision Technologies, Sensata Technologies, and STMicroelectronics.

The Melexis MLX75307 CMOS imager specifically targets automotive front-vision applications like advanced driver assistance systems (ADAS), high-beam assist, and night vision. It improves safety by proactively alerting the driver of potential dangers.

Continue to page 3

Furthermore, the MLX75307 offers a wide dynamic range of 190 dB for multi-scene conditions (110 dB for intra-scene conditions) and resolution of 750 by 400 pixels. Its signal-to-noise ratio is 102 dB and operating voltage is 3.3 V. Operating temperature ranges from –40°C to 125°C. Melexis characterized the device according to the European Machine Vision Association (EMVA) 1288 standard.

STMicroelectronics’ VL5510 automotive CMOS imager also suits ADAS applications. It features a 1024- by 512-pixel format, 7.14-V/lux sensitivity, 5.6- by 5.6-µm pixel size, 33-aA/pixel low dark current (at 25°C), 34-frames/s frame rate, and high quantum efficiency at near-infrared wavelengths. The processor complements a dedicated vision processor developed by the company in collaboration with Mobileye.

Sensata Technologies offers a CMOS imager, the Avocet, for ADAS applications with an enhanced dynamic range and excellent sensitivity for low-light and night-to-bright daylight driving conditions. It features a 154-dB dynamic range and comes in RGB or RGBi versions that feature the company’s Autobrite widedynamic- range technology, which was acquired from Cypress Semiconductor.

After nearly a decade of research, the Swiss Center for Electronics and Microtechnology (CSEM) discussed a smart CMOS image sensor design at this year’s ISSCC. Aimed at lowering the cost of automotive, industrial, and consumer electronics applications, the Icycam imager integrates a DSP chip with a CMOS sensor on a single die. It features QVGA resolution (320 by 240 pixels) and a digital logarithmic image compressor.

CMOS imagers together with powerful image processors have made significant contributions to the medical field. These combined devices are finding their way into imaging applications, diagnostic probes, swallowable pills, and a host of other applications (see “The Pulse Quickens For Cutting-Edge Medical Electronics Advances”).

For example, they’re enabling disposable diagnostic instruments such as endoscopes, like those from Micro-Imaging Solutions (Fig. 4). The company offers a complete CMOS-based camera system on a postage-stamp-sized circuit board. The camera-on-a-chip design is manufactured by several CMOS camera companies.

Micro-Imaging Solutions is concentrating on placing the video processing as well as most of the timing and control circuitry away from the pixel array plane. The whole idea is to minimize the size of the imager array. The circuitry can be stacked directly behind the imager plane or placed several meters away and connected to the imager via an RF or a disconnectable cable link.

Analog Devices offers a pair of powerful, highly integrated, 14-bit image processors for CMOS and CCD sensors. The dual-channel ADD9978A and quad-channel ADD17004 deliver a high degree of clarity, visualization, and accuracy in 75-MHz digital still and videocamera designs for medical and industrial machine-vision applications.

Beyond the realm of CMOS and CCD image sensors, SiOnyx Inc. is developing a new material called “black silicon.” The company believes the material will lead to a new class of image sensors that are 100 times more sensitive than conventional silicon, detect energy from the ultraviolet to the short-wave infrared bands, operate at very low voltage levels, and can be made in extremely thin 0.5-µm forms (Fig. 5). Most importantly, the material is compatible with existing CMOS processing methods.

“This is a brand new material that is compatible with the largest manufacturing infrastructure of the world,” says Stephen Saylor, SiOnyx’s president and CEO.

TAGS: Automotive
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.