This article is part of the TechXchange: Powering the Smart Home with Matter, TechXchange: The Internet of Things, and TechXchange: TechXchange Talks.
The Overview: Matter Protocol Now Serves More Apps
In a bid to provide IoT and IIoT connectivity in a greater swath of applications, Silicon Labs’s xG26 lineup of connectivity and MCU modules picks up where its earlier xG24 devices left off, offering designers more flash and RAM memory as well as double the GPIO pins. As a result, the XG26 modules bring powerful capabilities to smart-home, smart-city, and industrial use cases.
Who Needs It & Why: Bringing Memory and Processing Power to IoT/IIoT Apps
Designers seeking to endow their projects with extensive connectivity capabilities are turning increasingly to the Matter protocol, as it affords the ability to improve interoperability between devices from varied makers. This is especially critical in end products intended for “smart anything” applications, which are addressed by manufacturers of all stripes, many of whom wish to lock end users into their respective “walled gardens.”
When you factor in the growing complexity of today’s “smart anything” applications, designers quickly realize that their projects demand a lot more memory and GPIO capabilities. Many of their end products will have lifespans of 10 years or more. Thus, “future-proofing” these designs calls for connectivity and MCU modules that can accommodate a long lifetime of firmware and software updates even as code sizes continue to expand.
The modules’ high GPIO counts serve the trend toward integration of greater functionality on single devices. Consider something like a smart lock, for example: Rather than having RF transceiver functions on one chip and a central MCU on another, these devices are able to consolidate both on one module. This can, in many cases, lead to some space savings on printed-circuit boards (PCBs).
Under the Hood: Versatility of Silicon Labs's xG26 Modules
- The PG26 is a non-connected MCU powered by a 78-MHz Arm Cortex-M33 core. It carries up to 512 kB of RAM and up to 3 MB of flash. The low-power MCU consumes just 33.4 µA/MHz and sports several peripherals from serial interfaces to LCD drivers, ADC/DAC, an AI/ML accelerator, and a temperature sensor. The PG26 is software-compatible with the xG26 wireless SoCs, so it’s easily upgradable to add wireless support in the future.
- For connectivity applications, the BG26 (Bluetooth LE) and MG26 (Matter) SoCs address high-end, low-power IoT/IIoT mesh devices. Each carry high-performance radios with up to +19.5-dBm transmit capabilities and highly sensitive receivers (–97.6 dBm for Bluetooth LE at 1 Mb/s and –105.4 dBm for 802.15.4). They carry the same Arm Cortex-M33 core as does the PG26 and are loaded with a bevy of peripheral devices.
Importantly, both the BG26 and MG26 provide dynamic multi-protocol support for Zigbee, Thread, Matter over Thread, Bluetooth 5.3, Bluetooth Mesh, and proprietary variants. They operate from 1.71 to 3.8 V DC and offer up to 3200 kB of flash and 512 kB of RAM for future-proofing end devices. There’s also up to 64 GPIOs and four analog inputs.
Across the xG26 lineup, one will find overlap from an application standpoint in the home-automation and industrial segments, but the differentiators are in whether the application demands connectivity or not, or if it’s running Zigbee, Matter, or Bluetooth.
Cybersecurity: Robust Protection in IoT with xG26 Modules
Cybersecurity in IoT devices is a bigger concern than ever. The xG26 family addresses those concerns with Silicon Labs’s Secure Vault, a dedicated security core that meets all requirements of Matter today. It not only addresses the obligatory items like true random-number generation, a crypto engine, and secure application boot, but also optional and/or recommended security features. As security needs evolve over time, the core’s firmware can be updated in the future to provide security in the long run.
Offloading Machine-Learning Inferencing
The AI/ML hardware accelerator included on the xG26 devices is embodied in what SiLabs calls its Matrix Vector Processor (MVP). In IoT applications at the edge, the MVP uses a hardware core that offloads machine-learning inferencing from the MCU. This results in inferencing at speeds up to 8X over the Arm Cortex-M33 MCU.
Furthermore, because this computing is taking place locally without needing to transmit the data to the cloud, it does the processing at up to 6X lower power. Data transmission is a major consumer of battery power, so this ends up providing applications with longer battery life.
As usual, Silicon Labs provides designers with a broad variety of resources in terms of documentation. There are also hardware tools and software to help with application development, along with SiLabs’s Simplicity Studio development tool.
Key customers are now receiving samples of the xG26 SoCs. Production begins in the third quarter of 2024.
Related links:
Read more articles in the TechXchange: Powering the Smart Home with Matter, TechXchange: The Internet of Things, and TechXchange: TechXchange Talks.
