Optical sensors are becoming ubiquitous, but application and performance must match.
Optical sensors once seemed destined solely for high-end roles in applications such as biotechnology, astronomy, industrial inspection, or telecommunications. Instead, low-cost optical sensors have become pervasive in digital cameras, storage devices, mobile phones, toys, optical mice, and scanners. And the niche markets in research, manufacturing, and communications have remained just that, limited—if lucrative—market segments.
One star of this story is the complementary-metal-oxide-semiconductor (CMOS) image sensor. In a two-horse race for overall market share, CMOS sensors are beginning to gallop ahead of charge-coupled devices (CCDs), which have been the dominant technology for the past two decades. Fortunately for manufacturers and end users of both sensor technologies, the size and diversity of the markets keep expanding.
In unit sales, CMOS image sensors actually overtook CCDs in 2002, but will not overtake them in revenue until 2007. Comparing the two is somewhat confusing according to a recently released study of the image-sensor market by Strategies Unlimited (Mountain View, CA).1 Tom Hausken, author of the study, says it's a bit like comparing DRAM to flash memory, or optical to magnetic storage. These pairs of technologies compete in some applications but mainly coexist by thriving in their own niches.
Hausken reckons that the total image-sensor market in 2003 was for 280 million units, comprising 113 million CCD and 167 million CMOS units. Through 2008, CMOS units are predicted to see a compound annual growth rate (CAGR) of 29%, while CCDs will grow at a CAGR of only 1%. Although both sensor types should see a slight decline in unit price, the average price per CCD unit will remain around $14, while CMOS units will remain about $4.50 per unit.
Given the buying power of consumers, applications that cater to the consumer market should generate the most revenue for manufacturers of both types of image sensors, particularly those applications that target digital cameras, mobile phones, and camcorders, which demand sensors in high volumes yet command moderate prices. Hausken sees these applications as the leaders, with the large-scale introduction of image sensors for intelligent vehicles and biometrics being delayed.
For now, the critical test for CMOS sensors is whether they can displace CCDs in mobile phones with embedded cameras. Mobile-phone makers demand very low prices to ensure commercial success, but producing sensors with small size, high quality, and low price is very difficult. Hausken questions whether CMOS can meet these expectations, or whether demand for camera phones will be as great as advocates currently believe.
CMOS sensors tend to capture new markets, and even create them, following the path taken by the optical mouse. The computer mouse was invented in 1968, but Agilent Technologies (Palo Alto, CA) released an optical version based on an LED and a CMOS sensor in 1999, and optical mice from Agilent and STMicroelectronics (Geneva, Switzerland) now completely dominate the personal computer market.
So the market advantage goes to CMOS sensors in such new applications or in applications with fast production cycles such as mobile phones. And although the imaging quality of CMOS sensors generally trails that of CCDs, advantages in size, programmability, and power dissipation go to CMOS (see "High-speed CMOS imagers are flexible," p. 83). Finally, unlike CCDs, numerous independent CMOS foundries have sprung up giving fabless CMOS-sensor suppliers and integrators resources to produce relatively low-cost, customized sensors in limited production runs.
It will be interesting to see how markets develop. One downside of having optical sensors permeate the consumer market is that users who need sophisticated vision systems may come to think they can meet their needs at unrealistically low prices. Optical mice and consumer-grade digital cameras lie at the opposite end of the spectrum from the high-performance products these applications require. Consumers certainly benefit from domesticated optical sensors, but high-end users need a more powerful beast.
Image sensor market review and forecast: 2003, Strategies Unlimited, Report OM-28 (November 2003).
CONARD HOLTON is editor in chief of Vision Systems Design; e-mail: [email protected].