EDN logo


Design Feature: October 12, 1995

CCDs let you design vision into applications

John Gallant,
Technical Editor

The wide price range and versatility of CCDs make them well-suited for many applications. industrial and commercial uses for CCDs have flourished in the '90s

Since 1990, charge-coupled devices (CCDs) have become more sophisticated and smaller. Their evolution over the past decade has given them greater flexibility to reach more markets. For example, in commercial applications, CCDs run in camcorders. The devices work in fruit sorters in the industrial market. And, in scientific applications, the devices operate in mammography equipment.

Promising experimental results in the early to mid-1970s led to CCDs replacing vidicon tubes in most image-sensing devices in the 1980s. The 1990s has brought many applications for CCDs. In the cost-sensitive commercial market, CCDs are used extensively in video cameras for multimedia video-conferencing and in handheld camcorders. In the industrial market, CCDs can give machines "vision" to monitor moving parts in automatic-inspection systems. CCDs can accurately scan documents, labels, and complex images. The devices are also extensively used in factory-robotic systems.

In scientific-imaging applications, a CCD camera's capability ranges from sensing the smallest living cells to gathering light from distant stars and galaxies. In spectroscopy, CCDs can monitor the purity of chemicals and compounds. In medical imaging, CCDs in mammography equipment can capture sharp digital images. CCD technology even lets your dentist take soft X-ray images of teeth and bone structures.

Some equipment vendors integrate a CCD into a finished product instead of mounting a camera on the outside of the equipment. Examples of such CCD integration include placing a CCD inside a PC for video-conferencing, integrating a CCD into a door-opening system, and placing a CCD into a video phone. Equipment vendors choose this approach because off-the-shelf cameras lack the necessary features that the vendors want. Also, off-the-shelf cameras can be too bulky for certain applications.

In industrial applications, off-the-shelf cameras often can- not meet the stringent requirements of the industrial floor. Therefore, vendors often build their own CCD cameras into the equipment. The equipment vendor has to know the specifications of the chosen CCD to understand the CCD camera's full potential.


Know the architecture

There are two primary CCD architectures: linear and area arrays. Linear arrays have a single row of photosensitive MOS capacitors that form a single line of pixels for the CCD. Each pixel has an optical microlens that focuses the light onto the individual photosensitive capacitors. After exposure, the deposited charge packets in each capacitor are transferred initially to the parallel inputs of analog serial-shift registers. (The device usually has two registers—one on either side connected to alternate pixels.) The photosensitive capacitors are then reset and made ready to accumulate a new line of charge. The previous line, now in the analog shift registers, is transferred serially to the readout stage. At this stage, the charges are converted into signal voltages for the line. (See box, "The operation of a CCD".)

The operation of a CCD
A MOS capacitor forms the basis of a CCD. Virtually all CCDs manufactured today use a buried-channel MOS capacitor. The device consists of an n-type region that is formed on the surface of a p-substrate. A thin silicon-dioxide layer is grown on the n-type region and provides isolation for the deposition of a charge-transfer gate. A metal or heavily doped polycrystaline silicon layer is grown on the oxide layer to form the gate.

When entering the silicon structure, photons generate electron-hole pairs. The electrons are collected in the n-type region, which serves as a potential well. The holes pass onto the p-substrate. In an array, each potential well is isolated from a neighboring well by a field-oxide barrier. The CCD operates as an analog serial-shift register by transferring photocharges in one potential well to a neighboring well in the array. You accomplish this charge transfer by applying a positive voltage to the gate. By alternating the phases on successive gates to the potential wells, continuous-charge transfer can take place along the array to a final readout stage (Fig A).

As the charges arrive at the readout stage, they are converted to a proportional voltage level, a measure of the amount of light exposure. A readout diode adjacent to an output-amplifier stage makes this conversion. After readout, charges are drained in a reset diode via a MOS transistor. The responsiveness of the CCD is the output voltage per unit of light exposure for a specified radiation. The responsiveness is measured in V/µJcm2.

Today's CCDs have features that virtually eliminate the problems that plagued earlier versions. For example, early CCDs were subject to blooming and smearing. Blooming occurs when incident light or an overlong integration period causes the number of photocharges to accumulate beyond a potential well's capacity. The charges spill into adjacent potential wells. Blooming is usually localized and can arise in scenes containing strong local illuminations. The spread of photocharges appears as a white blot or streak on a TV monitor.

There are two antiblooming approaches that CCD vendors use to reduce blooming effects. One technique involves implanting a deep, n-type diffusion layer that allows extra charges to be drawn towards the substrate. At the substrate, the charges can recombine. The other method is to implant an antiblooming diode at each photosite that has a control gate to separate the photosite from the potential well. The CCD biases the gate, so that all excess charge overflows into the antiblooming diode.

Smearing is similar to blooming. Smearing was a problem in early linear arrays. Photocharge overflow from one potential well into another can cause smearing. The smear appears as a bright streak as the CCD transfers charge packets during readout. Current linear-array CCDs combat smearing by using a light shield that lies close to the sensing surface. The shield prevents the light from piping from one well into another.

Area arrays arrange the photosensitive MOS capacitors into a 2-D matrix. You can think of an MxN array as M linear elements of N pixels. Each pixel has a microlens to focus the impinging photons. The M linear elements are aligned vertically. Channel-stop regions separate the M linear elements. The CCD places an additional independent analog serial-shift register at the bottom of the array. This register has a charge-transfer direction orthogonal to that of the array. A charge-detection readout stage and an output amplifier terminate the analog serial-shift register.

After an integration period, array readout involves clocking all rows of charge packets in the wells of the array toward the analog serial-shift register, one pixel at a time. This process causes the bottom row of charge in the array to transfer into the analog serial-shift register. The serial-shift register then transfers the charge packets toward the output amplifier. The device converts the charge packets to voltage. The resulting data stream is a pixel-by-pixel, row-by-row representation of the image falling on the area array.

Dark current limits all CCDs. The current determines the output signal that thermally generated electrons generate in the dark. The dark current is a linear function of the integration time and is highly sensitive to temperature. Manufacturers can reduce the amount of dark current by using multiphase-pinned (MPP) clocking. MPP holds the clocks for the vertical array at a negative potential as the CCD reads out the horizontal signal. Therefore, the chip's surface does not become depleted, and the device obtains dark current in the range of 10 to 50 pA/cm2. Without this design enhancement, dark current can be 1 nA/cm2 at room temperature. A drawback of MPP is reduced full-well capacity.

Linear-array CCDs are used in machine-vision applications to scan documents and labels as well as monitor parts in an automated assembly line. The devices are used in photocopiers, fax machines, and bar-code readers. The linear-scanning rate must be fast enough to keep up with the speed of the machine. Area-array CCDs are used in 2-D scan applications, such as cameras, telescopes, and microscopes. The number of pixels in the area array determines the resolution of the CCD.


As with many IC markets, Japanese companies dominate the commodity-quality commercial segment of the CCD market. US and European companies specialize in higher priced devices for the scientific, military, and industrial segments.

Panasonic, Sharp, and Sony are the primary suppliers of low-cost commercial linear and area CCDs. Toshiba supplies a line of linear CCDs that have from 128 to 5340 pixels. The devices range in price from $50 to $265 (100) and come in 12-pin CERDIPs or 22-pin DIPs. Many Japanese companies enjoy a captive market by supplying CCDs for their own fax machines, camcorders, and cameras.

The Japanese companies have developed a wide range of off-the-shelf CCDs for equipment vendors to design into their systems. The latest offering from Panasonic has VGA resolution (640x480 pixels). The area CCD has square pixels for equal horizontal and vertical resolutions. The MN3776RE uses a progressive scan technique that obtains signals from each pixel during a single exposure. The device sequentially scans each line of the image—unlike interlaced CCDs, which scan only part of an image's lines in each pass. Sample prices for the chip are about $200, but the price is expected to drop to about $30 (in production quantities) by next year.

Sony Semiconductor Co of America recently introduced the first 1/5-in. color- and black-and-white-area CCDs that conform to EIA and CCIR standards. (See "CCD sensor for color/black-and-white video measures 1/5 in. sq," EDN, Aug 3, 1995, pg 28.) The ICX076AL and ICX077AL are high-sensitivity devices (360 mV), have effectively 362x492 pixels, and come in a 14-pin DIP. The size and smaller lenses suit the CCDs for security applications, such as surveillance cameras and door-access cameras, two-way video applications for PC-based video-conferencing, and video-phone installations. The devices cost $19 (10,000).

Sony uses a process called Interline transfer to shift the photocharges stored in the vertical elements of Sony's area arrays. Interline transfer masks every other vertical element in the area array from exposure to light. The vertical elements that are exposed to light use the masked elements as vertical analog shift registers to transfer photocharges to the horizontal register at the bottom of the array. Interline transfer allows for cost-efficient data transfer but trades off on pixel resolution.

Sharp offers an area CCD that features 410,000 (811x507) pixels in a 1/4-in. area. The LZ2453 device comes in a 14-pin plastic DIP and conforms to color NTSC standards. The chip has an electronic shutter that ranges from 1/60 to 1/10,000 sec, and it operates from 20° to +70°C. The chip costs $80. Sharp's devices come in three formats: as a single chip, as a chip mounted on a module that contains the necessary integration electronics, or as a complete fully assembled camera. The company claims that almost all purchases of its CCDs are in the module format for easy integration into equipment.

All of the CCD companies offer a wide range of support chips that lets you complete a design. Examples of support chips are video drivers that generate the timing pulses for driving a CCD and produce the synchronous pulses for TV signals. Signal-processing ICs include clamping, S/H, AGC, and amplification functions. The ICs also provide composite video outputs. Hitachi, a third-party vendor, provides a four-chip set for interfacing CCDs to video cameras.

Looking ahead
You can expect more applications and sophisticated design in CCDs. Cost-sensitive applications require smaller CCDs. You'll see more 1/5-in. CCDs for cameras in the surveillance and video-conferencing markets. In addition, Sharp plans to offer CCDs that will fit on the end of a pen. Fiber-optic cable will be tied to the CCDs. These CCDs will allow the user to see into narrow pipes and around crevices.


Emphasizing quality

By emphasizing maximum quality performance and sophisticated technologies, several US and European companies have designed CCDs for applications in various fields, including astronomy, military surveillance, medical imaging, spectroscopy, and robotics. These companies do not compete with the Japanese companies in the low-cost consumer market. Prices for CCDs from the US and European companies range from $500 to over $100,000.

Scientific Imaging Technologies (SITe), formerly the Tektronix CCD Products Group, specializes in the research, design, and manufacture of CCDs and subassemblies containing CCDs. The company's standard off-the-shelf products are MPP devices, called SITe Imagers, that range in price from $500 to $160,000.

The 64x64-pixel SITe Imager finds widespread application in adaptive optics instrumentation. The 512x512-pixel SITe Imager can efficiently project scenes at low light levels from UV to near-IR wavelengths. The 1024x 1024-pixel SITe Imager is used in medical-imaging applications, including digital spot mammography. The family also includes a $160,000 2048x2048-pixel SITe Imager that allows simultaneous readout of each 1024x1024- pixel quadrant for mammography-equipment vendors. SITe does not produce cameras for its CCDs. The company engages in contracts for custom CCDs for military surveillance and astronomy.

SITe develops thinned, back-illuminated CCDs that dramatically improve quantum efficiency—the measurement of the incident-light conversion to electronic-charge efficiency. Consumer products, such as video cameras, scanners, and fax machines, use front-illuminated CCDs. In applications measuring both visible and short-wavelength radiation, such as blue, UV, and soft X-rays, images from front-illuminated devices are nonuniform, and the gate structure degrades the spectral response of the images. A thinned back-illuminated device typically has a 50 to 150% higher quantum-efficiency rating than does a front-illuminated device.

Thomson-CSF also makes thinned backside-illuminated CCDs for high response in the UV and soft-X-ray spectrum. The process develops silicon wafers as thin as 10 µm. In addition, the company uses a proprietary organic-phosphor coating to improve the sensitivity of CCDs in the blue and UV wavelengths. The company offers a line of black-and-white-only linear and area arrays. The linear arrays have pixel sizes as small as 6.5 µm, a dynamic range as high as 15,000:1, and a readout frequency as fast as 25 MHz.

A CCD's specifications determine which device an equipment manufacturer selects and, ultimately, which camera it decides to use in a particular application. System designers choose CCDs based on specifications that they want in the final camera. These specifications for camera design include the number of pixels; the dimension of the pixels, which determines the resolution; the dynamic range; the maximum data rate; the responsiveness; the dark current and MPP operation; and the antiblooming features. The equipment manufacturer can build the camera into its equipment. The company can also purchase the chosen CCD and license the device free to a camera designer for final design.

Thomson-CSF also offers a line of area CCDs for industrial and scientific applications and short-wave IR detectors made of InGaAs photodiodes. The IR detectors work in machines such as fruit sorters in assembly lines. The IR detector can sense the water vapor in different types of fruit. The company provides evaluation kits for many of its array CCDs. The kits generally include a camera. The company's line of area CCDs start at $600, and the evaluation kits start at $3000.

EEV is a European vendor that is developing sophisticated CCDs. EEV offers scientific and TV CCDs for high-resolution cameras. EEV CCDs have as high as 2186x1152 pixels in an active area of 49.2x26 mm. The company has designed a high-performance device with 1024x256 pixels that are 26 µm2 for spectroscopic applications.

Both EEV and Thomson-CSF use frame transfer (FT) to transfer photocharges in an area array. In FT, an unexposed area is the memory zone. The memory zone is as large as the illuminated area, the image zone. The image zone receives the photocharges en masse. The charges in the memory zone are then transferred pixel by pixel to the readout stage, and a new image is sensed in the image zone. The FT operation is continuous, so it reduces blurring in fast-moving images.


You can reach Technical Editor John Gallant at (617) 558-4666; fax (617) 558-4770.


For free information...
For free information on the CCDs discussed in this article, circle the appropriate numbers on the postage-paid Information Retrieval Service card or use EDN's Express Request service. When you contact any of the following manufacturers directly, please let them know you read about their products in EDN.
EG&G Reticon
Sunnyvale, CA
(408) 738-4266
EEV Chelmsford
Essex, UK
(0245) 493493
In the United States, (914) 592-6050
Hitachi America Ltd
Brisbane, CA
(800) 285-1601, EXT 05
Loral Fairchild Imaging Sensors
Milpitas, CA
(800) 325-6975
Panasonic Industrial Co
Secaucus, NJ
(201) 348-7000
Sharp Electronics Corp
Camas, WA
(206) 834-2500
Sony Semiconductor Co of America
San Jose, CA
(800) 288-7669
Scientific Imaging Technologies Inc (SITe)
Beaverton, OR
(503) 644-0688
Thomson-CSF Semiconductors
Cedex, France
(33 1) 69 33 00 00
In the United States, (201) 812-9000
Toshiba America Electronic Components Inc
Irvine, CA
(800) 879-4963



| EDN Access | feedback | subscribe to EDN! |
| design features | design ideas | columnist |


Copyright © 1995 EDN Magazine. EDN is a registered trademark of Reed Properties Inc, used under license.