It's not easy being a 16-bit MCU vendor these days, but things are not as bad as some want you to believe. Choosing the right microcontroller for an application still places 16-bit solutions at the forefront, and there is no shortage of products. In fact, the 16-bit market is showing growth along with the 8- and 32-bit MCU market. The challenge to 16-bit MCUs arises from 8-bit MCUs becoming more powerful and the recent influx of low-cost, low-power 32-bit MCUs that may cap the upward growth of 16-bit MCUs. Arm's (www.arm.com) recent success licensing its 32-bit ARM architecture to major MCU vendors has resulted in standard parts designed to take on the 16-bit MCU. The 16-bit Thumb instruction set helps keep program memory size down.
Many 16-bit MCU vendors like Motorola (www.motorola.com) counter that their well established architectures compete well with 32-bit MCUs. The 16-bit architectures support high-level languages like C and C++ that are the norm on 32-bit platforms. This means the MCU can be chosen based on the merits of the chip rather than the architecture. Upward compatibility is another area where 16-bit parts continue to shine. Kevin Kilbane, strategic marketing manager for Motorola, notes that the 8-bit HC11 and 16-bit HC12 have a similar architecture and are code compatible. As with many MCU vendors, they have common peripherals.
Two primary reasons for choosing a 16-bit architecture over an 8-bit alternative have been memory addressing and data manipulation. The 8-bit MCUs overcome the memory addressing issue by incorporating 24-bit registers as in Zilog's eZ80Acclaim! (www.zilog.com) and Rabbit Semiconductor's Rabbit 3000 (www.rabbitsemiconductor.com). Michael Gershowitz, Zilog's director of product marketing, notes that its 8-bit processor handles 16-bit data and incorporates a 32-bit multiply-accumulator logic. This type of acceleration allows the 8-bit products to handle chores normally requiring a more powerful processor architecture.
The digital signal processing arena is one place where 16-bit solutions are increasing in number. Notable additions like Analog Devices' Blackfin (www.analog.com) and Texas Instruments' C5000 (www.ti.com) provide performance necessary for motor control and low-power mobile applications like cell phones. Steve Marsh, director of strategic marketing for Microchip (www.microchip.com) points out that for many systems, the cost of flash memory is a significant factor. Microchip's 16-bit digital signal controller, the dsPIC, mixes flash with a high-performance DSP design,
Many 8-bit vendors have skipped the 16-bit market completely and instead added a 32-bit solution to their top end. A low-end, 32-bit part may not always be the right choice. Kevin Klien, standard products marketing manager for Motorola's 32-bit Embedded Controller Division, points out that often a high-performance 32-bit part running at a slower speed may be a better choice and that a 16-bit MCU may run rings around a 32-bit part for a particular application.
The bottom line is that 16-bit MCUs are here to stay. In fact, the 8- and 32-bit parts may be getting more competition for the MCU in the middle.