It’s Time for a New Generation of Microcontrollers

December 06, 2022

Blog

Applications deployed in today’s smart homes, smart factories, and smart cities demand smart microcontrollers.

Spoiled for Choice

Have you ever visited a restaurant, only to find yourself overwhelmed by the number of dishes on offer? If so, have you also occasionally found yourself wishing for a simpler selection, perhaps comprising only four appetizers, four main dishes, and four deserts?

Designers of embedded systems know this feeling well when it comes to selecting the most appropriate microcontroller unit (MCU) for a particular project because there is a mindboggling array of options available. Each vendor may have multiple families—oftentimes based on completely different processor architectures—with each family offering myriad variants (pins, packages, peripherals, memory, etc.). Apart from anything else, migrating an application from one of these families to another can be so time-consuming and prone to problems that such a relocation is never even attempted, even if migrating to a less sophisticated device could save money, or migrating to a more powerful device could support more features. Based on all this, it’s no surprise that designers sometimes feel “spoiled for choice” (and not in a good way).

On the other hand, it’s also true that it’s necessary to cover a wide range of requirements and deployments. Some applications require ultra-low-cost MCUs, while others necessitate ultra-low-power, and many modern IoT devices demand MCUs that include on-chip wireless connectivity. Furthermore, one technology area that is currently increasing at an exponential rate is that of smart MCUs capable of performing extremely high levels of computation while also offering high levels of security. These MCUs are also often required to perform sophisticated artificial intelligence (AI) and machine learning (ML) tasks. And all of this is required to take place at “the edge,” which is where the internet meets the real world (or where “the rubber meets the road”).

Smart MCUs for Smart Homes, Cities, and Factories

Target deployments for this latter class of MCUs include smart homes, smart cities, and smart factories, including emerging industrial and IoT edge uses. We have already grown familiar with having things like voice assistants and intelligent thermostats that learn to accommodate our unique requirements in our homes. The next step will be to have intelligent devices, like washing machines and driers that can detect potential problems before they happen and alert their owners that it would be a good idea to have them serviced. It won’t be long before some devices, like electric toasters, for example, are both vision and speech enabled so they can recognize multiple users and present their toasted products (bread, bagels, croissants, waffles…) accordingly.

Welcome to the world of smart homes, cities, and factories.

Have you ever sat at a red traffic light twiddling your thumbs in frustration because there is no traffic whatsoever on the crossroad that has the green light? This should not happen in the case of a smart city. Furthermore, using cameras located at the junction, a smart MCU should be able to see your car approaching, realize that there is no other traffic, and change the lights to green before you even reach the junction.

In the case of smart factories, it’s no longer feasible to perform reactive maintenance—which means waiting for something to break and then fixing it—because the factory’s entire operation may be disrupted. Similarly, scheduled maintenance—which involves checking things on an aggressing schedule before anything has a chance to fail and replacing parts that may yet have life left in them—can quickly become a very expensive option in terms of personnel and materials. A substantially more efficient and cost-effective approach is that of predictive maintenance whereby smart MCUs employ AI and ML techniques to monitor machines, detect anomalies, identify potential problems, and issue appropriate alerts along the lines of: “It is highly recommended that the main bearing is replaced in Pump #3 in the next 24 hours, otherwise catastrophic failure is predicted in 72 hours +/- 6 hours.”

Multiple Processor and Accelerator Cores

If you wished to create the specification for a smart MCU, a good starting point would be a 32-bit Arm Cortex-M33 processor core equipped with a powerful floating-point unit (FPU) and incorporating TrustZone security that programmers can use to more easily achieve software isolation and create the security foundation required for modern IoT devices.

Many of today’s high-end applications can benefit from having two processor cores—perhaps a second “bare bones” Arm Cortex-M33 core that can be placed in charge of handling any low-level background tasks, thereby freeing up the main processor to concentrate on high-priority activities.

Another consideration is that many applications involve a large amount of complex and compute-intensive digital signal processing (DSP). While this could be performed by either of the Cortex-M33 cores, it can be performed more efficiently (higher performance while consuming less power) using a special DSP core or coprocessor. Similarly, many applications require data streams to be encoded and decoded. Once again, these tasks could be performed using the Cortex-M33 cores or the DSP core or coprocessor, but they can be performed more efficiently using a dedicated Codec accelerator core.

Last, but certainly not least, we come to AI and ML tasks on the edge. Use cases with AI in general and ML capabilities, in particular, are changing how we interact with devices and machines. The Cortex-M33 processor is scalable and flexible enough to run any type of AI/ML workload, but a dedicated artificial neural network accelerator in the form of a neural processing unit (NPU) core can do things tens of times faster while consuming a fraction of the power, thereby leaving the other cores free to perform the tasks for which they are best suited.

A Representative Example of a Smart MCU

One representative example of a smart MCU would be the new MCX N series from NXP. Different members of the MCX N series offer various combinations or one or two Cortex-M33 processors coupled with DSP and other accelerator cores and peripherals. Of particular interest in the context of these discussions is the fact that some members offer a dedicated NPU core with a high-efficiency compute architecture for real-time inference, thereby supporting local detection and decision-making without any need to communicate with the cloud.

The MCX N Series also features the EdgeLock® secure subsystem with immutable core security functionality. Part of the silicon root of trust, this and features side-channel protected cryptographic acceleration engines, key management, and device attestation. Pre-configured security and key management policies enable device makers to simplify the path to certification, avoid costly mistakes, and reduce development time.

 

Meet the MCX portfolio of MCUs.

In fact, the MCX portfolio includes four MCU series. In addition to the MCX N Advanced series, which is designed for secure, intelligent applications, the MCX A Essential series is optimized to provide critical functionality for a broad range of applications such as motor control, where cost constraints, advanced analog capabilities such as high-precision data converters and fast time to market are key considerations.

Meanwhile, the MCX W Wireless series offers low-power narrowband connectivity, including Bluetooth® Low Energy. Designed to simplify adding wireless connectivity to IoT devices, its energy-efficient radio helps extend the battery life of small connected systems. Last, but certainly not least, the MCX L Ultra-Low Power series is designed for power-critical applications. With one of the industry’s lowest static and dynamic power consumption, these devices will help extend battery life significantly compared to traditional MCUs.

The Advent of The Smart Future

We started this column talking about the problem of being spoiled for choice. In the case of NXP, customers had the choice of two industry-leading microcontroller offerings in the form of the LPC and Kinetis MCU families.

The MCX portfolio combines the best features of the LPC and Kinetis MCU families to redefine the future of general-purpose MCUs for intelligent connected edge and industrial applications. In addition to providing seamless migration from the LPC and Kinetis MCU families to the MCX domain, it’s also easy to migrate applications across devices in the MCX portfolio.

Now the choice is simple. The smart future is here and now, and smart MCUs like the MCX N series are going to be key when it comes to powering our increasingly smart homes, smart cities, and smart factories.