Integration strategies transform microcontroller design
By Kevin Klein, Standard Products Marketing Manager ,32-Bit Embedded Controller Division, Semiconductor Products Sector (SPS), Motorola, Inc., Austin, Texas, EE Times
August 30, 2002 (11:52 a.m. EST)
In the 32-bit embedded control space we are seeing major changes in the types of devices being developed. The tremendous performance that these devices provide at the attractive costs that modern semiconductor technology allow is driving their use in a more varied mix of applications.
As this happens, the types of control devices being developed are changing to reflect this new usage. The simple 32-bit computation engines of the past are becoming rarer, and instead the power of 32-bit CPUs is being harnessed in new ways for new types of systems. New application demands and semiconductor economics are driving two related MCU and MPU product trends. These trends are the ever-increasing levels of integration demanded of these products and the blurring of the traditional boundaries between MCUs and MPUs.
First, let's look at the trend towards ever-increasing levels of on-chip integration. This increase in integration is being driven by s everal semiconductor technology factors.
The ability to put increasing numbers of transistors on a given amount of silicon area at a relatively constant cost provides the ability to achieve previously impossible or uneconomic integration levels. This means that in many instances it is often now more economical to integrate more functions of the overall system into a single silicon device.
More interestingly, though, is the fact that the cost of bringing I/O signals off the chip has been decreasing more slowly than the cost of putting additional circuitry on the chip. This often results in situations where it is not only less expensive from an overall end user system perspective to integrate additional functions onto a single silicon device, but it is actually less expensive to perform relatively complex functions on-chip rather than incur the necessary cost to bring out the signals necessary to communicate with an external device.
On top of basic semiconductor economics, electronics technology trends also continue to drive increasing integration. The need for increasing levels of functionality in applications, the decreasing amount of physical space within applications, the need to maintain EMI/EMC within constant limits while performing more functions at higher frequencies, and reduced power budgets all drive customer demand for increased integration.
Typical of the new generation of connected control oriented processors, the MCF5249, for example, combines a wide range of functionality on a single silicon device, including a 140 MHz 32 bit CPU core, 96 kbytes of SRAM, audio processing, IDE and flash media interfaces, timers, DMAs, and an SDRAM controller.
Devices like these often enable specific small footprint iappliance applications, such as a highly integrated MP3 player. However, their ability to provide a powerful combination of features at low costs means these devices can find widespread use in a variety of control applications and sometimes can even enable to tally new applications, further increasing the size of overall end market and reinforcing the basic trends driving the need for increased integration. Blurring lines
Additionally, as the level of integration increases and SoC design techniques are more widely used, we are seeing a blurring of the traditional boundaries between MPUs and MCUs.
The increasing use of operating systems as well as the increasing use of on- and off-chip memory for security and performance needs is driving MCUs to incorporate features like cache memory and memory controllers more typical of an MPU.
In devices such as the MCF5249, traditional MPUs are developing into devices with increasing levels of integration as well as incorporating features traditionally found in MCU-like devices at the low costs demanded by traditional MCU embedded applications. The classic demarcation between an MCU and MPU -- single-chip processor with embedded memory and dedicated peripheral set vs. CPU-type processor, is be coming increasingly hard to apply, and in fact decreasingly relevant. The right feature set at the right price for the application will determine which processor gets developed and designed in.
As established 16-bit MCU customers migrate to 32-bit MCUs, and 32-bit MPU applications require higher levels of integration to decrease system cost while increasing system performance, new types of 32-bit embedded controllers are being developed.
Advances in semiconductor manufacturing and design technologies enable us to integrate tremendous levels of functionality at lower costs than provided by previous discrete solutions. More importantly, though, the widespread adoption of these devices in novel applications is driving the industry towards a blurring of the traditional borders between what is an MCU and what is an MPU, such that future 32-bit embedded control devices wi ll adopt the best of both sides to meet the cost and performance needs of future applications.