Observing the changes happening in the world today, one might find significant insights just by looking around at the devices. Smart meters that allow data queries via smartphones, wearable devices that continuously monitor health data, white goods with various sizes of screens for interactive upgrades, and even some devices that have begun to leverage AI assistance.
The ubiquitous data, continuously upgraded interactions, and the emergence of AI at the edge are three major trends in this rapidly changing technological world. “In this fast-changing world, the automation level and connectivity of billions of IoT devices are constantly increasing, and they also need to process more data securely on-site. We call this the cloud-connected intelligent edge,” said Julian, Vice President of Microcontroller, Digital IC and RF Products Division (MDRF), IoT/AI Technology Innovation Center, and Digital Marketing at STMicroelectronics China. He indicated that the important driving force and core components behind all this are microcontrollers, and the recent release of a series of new products by ST will further unleash developers’ creativity and drive this change forward.
Continuous Data and Connectivity
With the refinement of scenarios, many small devices become primary sources for data monitoring, such as thermostats, smart meters, smoke detectors, and activity trackers with GPS that many people are currently using. These devices require batteries for power, necessitating an MCU that perfectly balances power consumption, functionality, and cost.
“In some portable insulin pumps, smart meters, and other products, we see increasing demands for functionality, resources, and security. The STM32U0 has performed excellently in some existing mass production projects,” said Ming Zhang, Marketing Manager of Microcontroller, Digital IC and RF Products Division at STMicroelectronics China.
The STM32U0 series, as a new member of the ultra-low-power MCU family, exhibits excellent low-power performance, achieving over 15 times power optimization in static power consumption and nearly 36% optimization in dynamic power consumption. In practical product verification, applications like water meters can achieve a 38% power optimization, while traditional industrial sensors can also see over 50% power optimization.
This is largely related to the internal structure of the STM32U0, which features a Cortex M0+ core and an ART accelerator for faster execution of related applications, allowing the entire system to enter low-power mode more quickly, thereby reducing overall power consumption. Additionally, the STM32U0 integrates functions such as an LCD driver and an MSI internal oscillator, which also lowers BOM costs and offers a more attractive price point.
Once data is available, the next demand is connectivity. For small and battery-sensitive devices, Bluetooth is the best choice for achieving connectivity. The STM32 WBA series wireless MCU products released simultaneously support various 2.4G protocols, including low-power Bluetooth, Open Thread, ZigBee, and the Thread-based Matter protocol, using a 40-nanometer process with an embedded Cortex M33 core that can reach a frequency of 100MHz. It supports temperatures of up to 105 degrees, making it better suited for industrial applications.
The STM32 WBA can effectively meet the scenario needs of industrial control, medical devices, consumer electronics, and smart home control. For example, in the health and medical field, products can support BLE 5.4, allowing for up to 20 simultaneous connections through multi-connection features, better meeting the connectivity needs between the main device and multiple sub-devices in portable medical devices. Additionally, its standby mode consumes less than 140nA, which is better suited for the battery life requirements of medical devices.
“For low-power Bluetooth, power consumption is a crucial indicator. Less than 140nA can provide better battery life for some devices when not in operation, while also offering various low-power operating modes during RF operation, allowing users to choose the most suitable low-power operation mode based on their device system requirements,” said Yin Yu, Marketing Manager of Microcontroller, Digital IC and RF Products Division (MDRF) at STMicroelectronics China.
Comprehensive Interaction Upgrades
Anyone who has updated white goods in recent years will notice that the most significant change is that the once-complex buttons and knobs are being replaced by a touch screen, and voice communication is even possible. The interactive upgrades that have become commonplace on mobile devices are gradually occurring in traditional white goods.
However, the newly added screen on white goods also requires computational support. “Our STM32H7R/S can fill this gap well, helping many customers achieve high real-time computing while also providing excellent graphical display,” said Huang Liu, Marketing Manager of Microcontroller, Digital IC and RF Products Division (MDRF) at STMicroelectronics China, regarding the new high-performance MCU STM32H7R/S, which can run MPU-level applications in real-time MCU and offers higher design freedom while inheriting the MCU ecosystem to accelerate engineers’ development.
As a high-performance series, the STM32H7R/S features the strongest single core in the already released H7 series, with an Arm Cortex-M7 core running at a frequency of 600MHz, achieving a CoreMark score of 3174 and a DMIPS of 1284. The STM32H7R/S is also the first boot flash MCU in the STM32 series, equipped with a fast external memory interface, allowing developers more flexibility in choosing memory and architecture. In terms of interfaces, it has two 200MHz SPI interfaces of up to 16 bits, achieving high throughput. Additionally, if customers choose the latest NeoChrom GPU, it can support higher definition and smoother UI graphic design while reducing the load of graphical data on the CPU.
This means that the STM32H7R/S can support the display panels of refrigerators, washing machines, and microwave ovens. Its strong connectivity and security stability can also be effectively implemented in industrial control. Due to its 600MHz performance, it is also gaining widespread attention in some robotic applications.
ST also offers two development boards for customers, one of which can evaluate display screen performance with WVGA high resolution, aiding in the exploration and development of multifunctional prototypes.
AI at the Edge
When it comes to interaction, AI has undoubtedly become the key technology for achieving the next upgrade in interaction. The concept of edge AI has taken on a new mission in the “intelligent” aspect of cloud-connected intelligent terminals. Edge AI acts like a small brain for each IoT entity, providing a rapid response and instinctive intelligence.
“For any application related to intelligence, designers will consider whether there is AI involved?” expressed Ming Du, Marketing Manager of Microcontroller, Digital IC and RF Products Division (MDRF) at STMicroelectronics China, reflecting on the current thoughts about AI capabilities in the fields of smart furniture and smart cities.
The newly launched STM32MP2 microprocessor is specifically designed with an AI acceleration unit, with the NPU accelerator achieving a computing power of 1.35TOPS.
“I believe this is also a trend in MPU computing.” Ming Du stated that ST’s integration of NPU in high-performance MPU will allow for the sinking or decomposition of edge AI locally. At the same time, edge AI is more cost-effective, applications are more flexible, and system integration speeds are faster.
ST also provides two data model training methods for the STM32MP2. Engineers can either optimize and convert their own data models on the local MPU compiler for deployment or select a model from ST’s model library to establish edge AI computing capabilities, compiling optimized binary code through the MPU’s offline compiler or cloud compiler for the final deployment of edge AI applications.
ST’s model library offers four types of models, including motion pose estimation, image classification, semantic segmentation, and object detection. The MPU offline compiler is available in both desktop and cloud versions, allowing for model optimization, quantization, and conversion. Whether locally or in the cloud, the written code can run on the STM32MP2 development board for verification.
STM32’s Linux also provides an extension package called X-Linux-AI, which includes all software and tools within this Linux framework, including Python, Gstreamer, Open CV, etc., providing a complete AI ecosystem for developers with a one-stop solution.
Additionally, the STM32MP2 is a processor aimed at 64-bit applications, inheriting the excellent multimedia unit from the previous generation, supporting up to 1080p resolution with a 3D GPU, and supporting interfaces for RGB, LVDS, and MIPI DSI to meet various image output requirements. The built-in ISP function of the MIPI CSI two-wire camera interface has greatly enhanced machine vision and multimedia applications, allowing the AI ecosystem to run not only on the NPU but also on the CPU and GPU as needed.
Clearly, microcontrollers, as the driving force and core components of this world, are further advancing from uncovering more data and opportunities to enhancing connectivity, and constantly pushing the interaction upgrades of traditional devices, making AI ubiquitous. A series of new products from ST is unleashing new creativity.
Content Recommendations
Thirty Directions New: Beijing Unicom Builds an F5G-A All-Optical Quality Transport Network
The First Answer Sheet of the Comprehensive Deepening of the “Localization” Strategy: Aptiv Makes “Software-Defined Cars” a Reality
Apple’s Intent is Not on the iPad? Not Surprising, but Not Small Foreshadowing
