This is an extract from an article in the latest eTech Magazine
The general-purpose microcontroller, or MCU, is now a commodity product. Perhaps this is an overstatement. But in one sense, at least, it can be strongly supported by the growing trend in the use of processor cores that share a ‘common’ or ‘open’ architecture. Today, while there are still plenty of MCU vendors, it appears that 32-bit MCU roadmaps are increasingly based on processor cores developed by one company: ARM.
Since their invention and first production in the early 1970s, MCUs have been designed and manufactured by a large number of semiconductor companies, which have historically developed their MCU product lines primarily on their own proprietary architectures. Especially in the 8-bit market with devices such as the H8 from Renesas, the ST6/7 from STMicroelectronics, the PIC from Microchip, but also in the 32-bit market, such as the R32 and SuperH from Renesas or the Power Architecture from Freescale or C28x from Texas Instruments, to name just a few.
However, over the past ten years or so, there has been a growing trend among silicon vendors not to develop their own proprietary microprocessor cores, but to license cores and architecture from IP companies such as ARM and MIPS. These two companies, in particular, have fought head-on to win key designs for advanced microprocessors, ASICs and ASSPs for a range of embedded markets and applications, but outside of the PC microprocessor socket.
Moving to a Common Architecture
Certainly there are multiple challenges and major issues to consider for MCU vendors to even deviate from their proprietary architectures. All the years of investment, in terms of hardware, software and development tools, are not so easy to give up. And are they willing to put doubts in customers’ minds about future roadmaps and all their legacy code developed specifically for a particular MCU architecture or family?
Naturally, an open architecture can bring huge benefits for customers, such as being able to source from a large number of vendors or migrate from one to another for reasons of cost or performance or peripheral selection. Although Cortex-M3 based MCUs, for example, are not going to be identical, the migration process from one vendor to another, while not being trivial, should not pose the same level of challenge to change between substantially different proprietary architectures. Also, given these perilous times, the market has seen some significant consolidation of MCU vendors over the past few years. In the industrial market especially, customers need to rely on a processor being around for 10 to 15 years or more. So the common architecture may bring a significant degree of obsolescence safety.
Another advantage is the wide availability of software component libraries, enabling benefits in time-to-market. The move to a common ARM architecture also means the wide availability of ARM-based development/debugging tools from all the leading third-party tool vendors.
Selection by Peripheral
If the trend for MCU vendors is to base their product portfolios around the ‘nonproprietary’ or ‘open architecture’ of ARM, then they will need to find other ways of differentiating their products. A number of factors will be included in the process for selecting an MCU, including the microprocessor core, speed, memory, peripheral selection, price, development tools, operating system and software support. The choice of the processor core should determine, or at least influence, some of the other options. But, it appears that the industry is moving into the phase where it is the selection of peripherals that is paramount, and less a choice based on the MCU core functionality. Certainly peripherals are a key factor in choosing for specific applications. So it could be that following the ‘convergence’ around the ARM Cortex family (i.e. the processor core becomes the commodity), there will be a ‘divergence’ of solutions that offer many different combinations of peripherals for diverse and niche markets, relatively speaking. So, we might find two or three key suppliers offering product for each major MCU market, for example in high-quality audio where an I2S output is obligatory. And then a whole host of suppliers positioned somewhere in the mainstream offering similar memory, I/O and wireless options, fundamentally fighting it out based on price. Is this what the MCU market could look like in a few years time?
Read this article in full
