More for Less: Stable Future for 8-bit Microcontrollers

The venerable 8-bit microcontroller just keeps chugging along. Despite the performance advantages of its 16- and 32-bit siblings, 8-bit micros still accounted for a substantial percentage of the dollar volume of all worldwide microcontroller sales in 2003 and more than half of all unit shipments.

In a world that appears to dote on high-performance, wide-bandwidth, and an overabundance of I/O ports, 8-bit micros still find a lot to do—and do well.

In order for 8-bit micros to survive and even prosper for more than four decades, the chip designers who create them have evolved their silicon accordingly, most notably by adding a rich selection of peripherals and memory. The system designers who use them have responded with ever more innovative solutions to more and more control applications.

Low-cost chips and inexpensive development software is generally acknowledged to be the primary reason for 8-bit staying power. These attributes align perfectly with the system engineer's mission to design the most cost-effective solution. Ironically, the sophisticated systems that require high-performance chips to execute primary function often have quite a few mundane control functions to handle as well—and these are frequently best serviced by a distributed network of 8-bit microcontrollers.

But emerging applications are equally important. Today, 8-bit micros are small enough and inexpensive enough, for example, to penetrate the traditional market for 4-bit applications such as TV/audio remote controls, toys, and games. Packaging technology is a key determinant in how far the 8-bit parts can penetrate. The package costs more than the die and it has to be both small and sturdy.

For the simplest of all control applications, 8-bit micros can offer high-value "smart switch" functionality for applications that once used electromechanical devices or semiconductor logic—and this trend is just in its infancy.

Other applications that are using more 8-bit micros are automotive, networking of the industrial variety—CAN and LIN—and a host of others that emphasize control but do not call for much number crunching. But even some compute-intensive applications typically thought of as being in the 16- and 32-bit domain can be handled by an 8-bit micro if the function can be executed in hardware by a peripheral instead of being executed in software. In fact, peripheral integration is a key strategy in the 8-bit micro's longevity.

The two largest players in the microcontroller market—Renesas Technology, which leads the Japanese market, and Freescale Semiconductor, which leads in North America—both have carefully planned strategies in place to allow design engineers to move as effortlessly as possible from 8- to 16- to 32-bit microcontrollers. Most other microcontroller players have similar strategies. But the financial investment required to implement both a hardware and software continuum—compatible architectures for hardware and consistent user interface for development software—is an expensive proposition.

The dynamics of a platform strategy are more complex than simply moving designers along a path to higher functionality from 8- to 16- to 32-bit devices. As systems become more complex, distributed intelligence—particularly in relatively simple control functions—is the trend. A 32-bit microcontroller might "run the show" in such systems but there is room for 8- and 16-bit devices as well and a common platform makes design considerably easier.

Figure 1:  8-bit microcontrollers retain a substantial portion of the overall market. Source: Industry Analysts

Geographic regions will have an impact on the fortunes of 8-bit micros. Japan is still the largest market for 8-bit micros and it is dominated by Japanese suppliers, says Olaf Vogt, Supplier Marketing Manager for Philips Semiconductors, Eindhoven. Vogt sees a great deal of potential in the future for 8-bitters in the Asia-Pacific market, which includes China. Here, the traditional 4-bit applications such as toys and games are migrating to 8-bit designs, he says.

Relatively new applications such as remote metering and security—where security was never considered before in products such as electronically controlled window locks—are creating new, potentially high-volume applications as well. Battery control applications and small appliances also use 8-bit controllers, Vogt says.

China's emergence as a worldwide manufacturing power means it has a big role in 8-bit MCUs, says Ted Hwa, a Philips marketing manager who keeps a close watch on pan-Asian markets. New applications are still mostly being developed in North America and Europe, he says, but China is getting into the game of modifying older designs—many for products destined for sale inside China. Its burgeoning consumer market is good news for 8-bit micros.

Hwa believes the country with the second largest population worldwide is another potential market. Once the Indian economy starts growing, he says, it will become a major potential market for 8-bit micros. India also has very good software design facilities that can support the MCU applications.

Richard Sessions, Director of Product Marketing at Renesas Technology, has much the same view, citing a range of 35 to 50 cents in very high volume. Compare this pricing, however, to the low end of 32-bit microcontroller pricing—$5 to $7 in volume—and the playing field for 8-bit parts is a substantial niche. Renesas is, however, pushing its 16-bit solutions hard into applications traditionally dominated by 8-bit micros.

Low prices continue to make 8-bit designs viable in automotive networks such as CAN and LIN, Sessions says, and in the growing security market where a controller is needed to act on some sensing input such as temperature, or, to turn lights on and off.

Freescale Semiconductor (formerly Motorola SPS) dominates the automotive market in the U.S. and Kevin Kilbane, strategic marketing manger for 8- and 16-bit microcontrollers, says the silicon content in automobiles will continue its steady growth.

In many instances, 8-bit micros will be able to do the job adequately, he says, and their inherently lower power consumption can be an advantage. Recent legislation requiring inflation pressure monitors in automobile tires provides still another niche for 8-bit devices. Here, the pressure monitor, microcontroller, and the low-cost RF device that will send status warnings to the instrument panel are all battery powered. The control module must be small, inexpensive, and use little energy—all attributes of 8-bit systems.

Mike Gershowitz, director of marketing for micro-devices at ZiLOG, points out that 8-bit microcontroller pricing is driven by the number of peripherals and the amount and type of memory that comes with the microcontroller core. While Flash memory has become a critical requirement for some applications, micros with ROM or OTP (one-time-programmable) memory of 4 Kbytes or less accounts for the low end pricing.

The average selling price in that end of the market will probably continue to drop, he says, but so will margins—and there is also a limit to how low prices can go. The selection of memory types and peripherals such as LCD drivers for applications such as TV remote controls, Ethernet MACs, CAN and USB bus interfaces, analog-digital converters, and pulse-width modulators (PWMs) raise prices and margins.

ZiLOG, a company once almost synonymous with 8-bit micros through the widespread popularity of the Z8, is headed into that higher-end arena but will make room for the lower end as well through strategic moves. It recently decided to follow a fabless semiconductor strategy and will continue to pursue business relationships in China.

Admittedly, the role of technologically emerging countries such as China and India make setting a definitive crossover date hard to predict. Regardless of when this might happen, and of how inventive design engineers can be in optimizing systems built around 8-bit micros, the silicon vendors are acting on the reality of falling prices and profit margins.

Virtually all microcontroller companies are responding to this challenge and the two largest—Freescale and Renesas—are leading the way.

The microcontroller core itself tends to occupy a very small portion of the device's die, says Freescale's Kilbane, and therefore contributes little to its cost. As shown in Figure 2, peripherals and memory, in particular, occupy most of the die. This allows for a considerable amount of customization—matching chip functionality to the application with peripherals and memory. In control applications, in particular, retaining the same microcontroller core but adding memory or peripherals can prolong a design's useful life as succeeding generations of the application become more complex.

Figure 2:  Memory and peripherals occupy most of an 8-bit microcontroller's die

Freescale also provides overlapping solutions between architectures. Memory options for 8-bit micros range from 1 to 60 Kbits of Flash; 32 to 512 Kbits for 16-bit micros and over 1 Mbit for 32-bit micros, Kilbane says. Freescale calls its strategy a controller continuum. Development software is another important aspect of a continuum. Freescale and its tool partners maintain a consistent user interface across architectures.

The platform strategy at Renesas emphasizes 16-bit cores which can be accompanied on the die by 8-bit peripherals and an 8-bit bus to derive the benefits of smaller die size and power consumption. In terms of die space "you almost don't care what the CPU is," says Sessions.

Another aspect of projecting a platform strategy into the market is to appreciate the coding preference of many engineers. Renesas built its 16-bit cores to be C-code efficient, he says, because companies want to be able to reuse the applications programs they develop. In addition to the trend away from assembly language programming, using C and C++ means that code reuse is easily accomplished.

Other companies are on the same general track but with variations often dictated by their product portfolio. Philips, for example, advocates a leap from 8-bit micros to 32-bit, which leaves 16-bit parts more or less out of the migration equation. But the fundamental strategy remains the same: having its software partners provide very similar tool environments for both architectures to reduce the learning curve for design engineers. And no one is predicting the imminent demise of 8-bit micros—the idea of mixed architecture environments with efficient migration paths prevails.

The familiar plot of chip complexity vs. time illustrates the classic exponentially rising curve of Moore's Law and Wallace contends the area underneath that curve is proportional to the cumulative number of electronic applications that have been created over the years.

As the Pentium/Cray bleeding edge shoots skyward, he says, the cost of designs closer to the X-axis becomes less and less expensive. That's the bailiwick of 8-bit micros and it could be called the real technology revolution because it is happening where it affects people the most, at the low end.

"I've been making single-chip guitar tuners out of some few-dollar micros," he says, "and 10 years ago that could not be done. With every tick up in the curve, more low-end applications become possible, and thousands of engineers live there, earning a living making last decade's impossible ideas come true."

"Every Pentium price drop enables dozens of new applications in the scum on the bottom of the pond. That's where the action is. 8-bitters will never die. Neither will RS-232.," he says.

With advocates like Wallace, the future of 8-bit micros seems secure.

Contributing writer Jack Shandle is a former chief editor of both Electronic Design magazine and ChipCenter.com. He holds a BSEE degree and has written hundreds of articles on all aspects of the electronics OEM industry. Jack is president of eContentWorks, a consultancy that creates high-value content for publishers, eOEM corporations, and industry associations. His email address is jshandle@earthlink.net.