In my inbox this morning was some article from one of the trade rags. On a lark, I decided to send it to my microcontroller students, in hopes of stimulating some discussion. I am particularly hoping they'll notice the following:
What say you folks? My original email including the Article, and a link to the original article follows.
- The author's employer
- Some pros and cons of various low-end uC's
- ??? who knows what else
What say you folks? My original email including the Article, and a link to the original article follows.
- PHB - Pointy Haired Boss
- uC - microcontroller
Here is an article from one of the trade rags concerning microcontrollers. It is followed by a discussion/trollfest of some Engineering types and PHB types with too much free time. Take a few minutes to read the article and some of the comments. As always, analyze the article and comments with a critical eye. What is he trying to say? Why is he saying it? What is his reasoning? Why should students such as yourself care? Be prepared to discuss this for a few minutes tomorrow.
-sbombs
Link to original article: http://www.embedded.com/223500065?cid=NL_embedded
The 8-bit MCUs won't be going away anytime soon
Despite claims that the 8-bit MCU will soon be obsolete by 32-bit devices, that may not exactly be the case.
By Steve Drehobl
Embedded.com
(03/11/10, 09:34:00 AM EST)
For the past 20 years, a debate over the longevity and legitimacy of the 8-bit microcontroller (MCU) periodically erupts. The debate is usually sparked by the introduction of a higher-end processor or architecture and is almost always accompanied by overstated claims of a market moving away from 8-bit MCUs or transitioning to higher-end devices. It wasn't too long ago that the 16-bit market was doomed to disappear, due to pressures from 8-bit on the low end and 32-bit at the high end. The view of the microcontroller market in strict terms of data bus width, whether it is 8, 16, or 32 bits, is overly simplistic. The reality is that the microcontroller market is a more continuous spectrum, with a multitude of attributes and requirements that balance the system-level integration, power consumption, computational efficiency, robustness and, ultimately, cost. Even for 32-bit MCUs, there's a vast array of different architectures, manufacturers, and products that support the premise that there is no such thing as "one size fits all" when it comes microcontrollers. Some of the attributes don't involve the product itself and, in Microchip's experience, a different microcontroller is most often used for production than what was originally identified at the start of a project. Therefore, the ability to easily move from one device to another in a vendor's product family is a much more important selection criteria than the specific architecture used.
Comparisons of microcontrollers usually center on floating-point calculations, DFTs, DMIPS, and associated code-size and performance benchmarks. However, 8-bit MCUs are typically used at lower clock-frequencies and computation applications such as coffee machines, smoke/CO detectors, toasters, key fobs, security tokens, security-system sensors, toothbrushes, PC-fan controllers, point-of-load power supplies, battery chargers, 3D glasses, lighting, disposable medical, water/gas metering, and thousands of other applications, where math throughput is not critical. For the 8-bit applications, the ability to operate at 5 V in a noisy environment or to run for 10 years on a Lithium coin cell is much more important. The innovations in 8-bit MCUs focus on reducing power, physical size, and cost, while maintaining the benefits of 5-V operation, ease of use, and simplicity. These innovations enable the use of microcontrollers in new applications where electronics was not even an option a few years ago.
The analogy is to compare the performance and road-holding capabilities of a sports car with that of an off-road vehicle or super compact. Certainly, the sports car will perform better in standardized tests, but this doesn't mean that it is better than the off-road vehicle on a dirt road, or that it can achieve the fuel economy or cost of the super compact. In fact, cost is exactly where the newest claims of 32-bit dominance stem from. While some of the newer 32-bit microcontrollers are sold for prices previously reserved for 8-bit MCUs, the 8-bit parts are also constantly lowering costs and increasing capabilities. The evolution and innovation in the 8-bit market space has not slowed down at all as professional engineers find new ways to solve problems. The 32-bit devices can certainly execute 32-bit floating point multiplications faster than the 8-bit microcontroller, but that is only important when it's needed. An 8-bit MCU can operate at 5 V, directly drive a segmented LCD, has build-in high endurance EEPROM and has standby currents of a few nanoamps, all in single, miniature package a few square millimeters in size.
The physics that enable process technologies and, consequently, 32-bit microcontrollers to achieve new cost points are directly opposed to some of the features required for 8-bit applications. Smaller geometry processes inherently leak more and can't operate at higher voltages. Silicon designers have significant challenges to develop robust analog peripherals and high endurance non-volatile memory on these technologies. Not that these challenges can't be overcome, but often, the simpler answer is the better answer. For this reason, the 8-bit microcontroller will continue to play a large and important role in embedded design.
Steve Drehobl currently serves as the vice-president of the Security, Microcontroller, and Technology Development (SMTD) Division at Microchip Technology. He holds a Bachelor of Science in Technology degree from the University of Dayton.