 |
|
|
 |
||
 |
|
|
|
|
|
|
|
|
|
|||||||||||||||||
Since the last edition of Challenge was published, ST has completed its acquisition of Vision Group plc, a world leader in the rapidly emerging field of CMOS imaging technology. A major reason for ST's acquisition of Vision was the high degree of synergy between the complementary strengths of the two companies, a synergy that opens new perspectives in systems on silicon solutions for the consumer, computer, communications, automotive and industrial markets.  The 356 x 292 pixel device shown here facilitates the design-in of imaging applications - enabling the use of an imaging solution where previously it may not have been praticable on cost grounds. These products are ideally suited to applications such as biometrics, bar code reading, automatic meter reading, security, inspection systems and automotive applications. During its ten-year history, Vision has achieved worldwide recognition as a leading designer and supplier of CMOS image sensors. It's products have been successfully used in a wide variety of applications, including desktop video conferencing, digital still cameras, security, biometrics, automotive systems and toys. Now, this large pool of imaging know-how can be combined with ST's rich portfolio of Intellectual Property, cores, libraries and technologies. The key to this synergy is the integration of high performance image capture and image processing on the same silicon chip. This is possible because Vision's CMOS image sensing technology is based on standard CMOS processes rather than the traditional charge-coupled device (CCD) technique, allowing the sensor array to be combined with powerful digital signal processors, memory, data converters and I/O interfaces. Vision already produces successful products that combine image capture and image processing but now has access to one of the world's most formidable embedded processor and memory capabilities.  ST digital camers deliver frame rates of up to 30 frames per second while setting impressively high standards of picture quality. With a choice of two easy-to-use PC interfaces, USB and Parallel Port cameras are designed to provide perfectly formatted CIF video. Customers for videoconferencing solutions include Creative Labs, Xirlink, Zoom Telephonics and Cubic VideoComm. CMOS versus CCD Solid-state image sensors based on charge-coupling have been around for 25 years and offer excellent image quality with low noise. However, they also have a number of built-in disadvantages that have made them vulnerable to the enormous progress that has been made in the last ten years in CMOS imaging technology. Both approaches use an array of photodetectors, typically photodiodes, one for each pixel in the image. Each photodetector generates an electric charge in response to incident light. In a CCD device, a specialized VLSI process is used to form a closely packed array of polysilicon electrodes, one for each photodetector. These electrodes are so closely coupled that the charge on one electrode can be transferred, with the help of a suitable clocking scheme, to the neighbouring electrode. To capture the entire image, a complex sequence of clock signals is used to shift the electric charges across the two-dimensional chip surface to a one-dimensional array of sense amplifiers that actually measure the charge.  ST now offers the most comprehensive family of CMOS video camera reference designs and chipset solutions for both analog and digital video applications. This breakthrough has been achieved by integrating the key elements of a complete video camera on just two silicon devices. This approach has two major drawbacks. First, the CCD process does not easily allow other functions such as clock generation and signal processing to be integrated onto the same chip; it can be done but the cost is very high. Even worse, to ensure that charge packets are shifted across the chip without degradation, the amplitude and shape of the clock pulses must be accurately controlled. Generally, this requires a specialized clock chip with as many as five or six different power rails.  CMOS image sensors of the type developed by Vision also use an array of photodiodes but contain in addition an array of integrated CMOS sense amplifiers that eliminate the need to clock electron packets out of the sensor array for subsequent measurement. The simplest CMOS image sensors, known as passive pixel sensors, operate rather like DRAMs in that one sense amplifier is provided for every column in the array, with each pixel containing a transistor that acts as a Column Select gate. Unlike DRAMs, of course, the data stored in each pixel is an analog charge value. Active pixel sensors take this concept to its extreme, integrating a separate charge amplifier at each pixel site in the array. This gives the best performance, with the lowest noise levels, but uses more silicon area as each pixel needs at least three transistors. A problem with early CMOS sensors was fixed-pattern noise caused by variations in the sense amplifiers across the array but this problem has now been completely overcome, reducing fixed-pattern noise to insignificant levels. The major benefit of CMOS sensing compared to CCDs is that it uses standard CMOS technology. Not only does this allow full integration of clocks, peripherals and signal processing circuitry, but it also greatly reduces the power consumption - Vision's CMOS sensors typically consume just one-third of the power required by comparable CCD cameras at the system level. The addition of advanced imaging technology to ST's system integration skills will help take system-on-chip technology a stage further towards its ultimate goal where the entire system, including sensors and actuators as well as all of the processing and memory functions are integrated onto a single chip.  |
|
||||||||||||||||||
|
|
 |
|
|
|
|
||
