ST introduces second generation Industry 4.0-ready Edge AI powered MPUs

2 mins read

STMicroelectronics has introduced a second generation of its STM32 MPUs (microprocessors), along with a new architecture.

Intended to deliver improved levels of performance and security for applications at the industrial and IoT edge the STM32MP2 Series combines 64-bit cores with Edge AI acceleration, advanced multimedia features, graphics processing, and digital connectivity.

According to Ricardo De Sa Earp, Executive Vice-President and General-Purpose Microcontrollers Sub-Group GM, STMicroelectronics, “Integrating advanced security features in hardware, the new MPUs are ready for emerging opportunities in secure Industry 4.0, IoT, and rich user-interface applications.”

The first product line, the STM32MP25, is available with single or dual 64-bit Arm Cortex-A35 cores running efficiently at 1.5GHz, complemented with a 400MHz Cortex-M33 embedded core that handles real-time processing.

A dedicated neural processing unit (NPU) adds up to 1.35 TOPS (tera-operations per second) of computing power, suitable for Edge AI acceleration in applications such as advanced machine vision and predictive maintenance. With support for 32-bit DDR4 and LPDDR4 memories, the STM32MP25 ensures long-term support for cost-optimised designs.

Also featuring Gigabit time-sensitive networking (TSN) support and a two-port Gigabit Ethernet TSN switch, with PCIe, USB 3.0, and CAN-FD peripherals, the product line provides intensive connectivity for real-time industrial applications, data concentrators and gateways, and communication equipment. 

The device’s processing and networking capabilities enhance both detection and feature recognition for security applications and industrial automation. For example, it can acquire video from a 5Mpixel sensor at 30 frames per second (fps), perform analytics with the Edge AI accelerator, and send relevant video (encoding with an HW encoder) with detection metadata leveraging Gigabit Ethernet TSN, all in real-time streaming mode.

A 3D graphics processing unit (GPU) with 1080p resolution for graphics and video capabilities permits rich user interfaces, with support including Vulkan real-time graphics for Android applications. A1080p encoder/decoder and multiple display connections including LVDS, 4-lane MIPI DSI, and a MIPI CSI-2 camera interface simplify connecting displays and digital cameras including raw-Bayer image sensors.

Security features ensure SESIP Level 3 certification and include Arm’s TrustZone architecture and resource isolation framework (RIF). This is complemented with secure key storage, secure boot, a unique device ID in one-time programmable (OTP) memory, hardware cryptographic engines, and on-the-fly DDR encryption/decryption.

The devices are specified over the extended temperature range from -40°C to 125°C, easing thermal management and enhancing reliability in industrial environments, according to ST. In addition, like other STM32 MPUs targeting industrial applications, the series comes with ST’s 10-year longevity commitment.

A selection of package options is offered including a 0.8mm-pitch chip-scale package (TFBGA). This helps ease PCB design routing and allows economical designs down to four layers avoiding costly laser vias.

Developers using these new devices can leverage the extensive STM32MPU ecosystem, which contains the OpenSTLinux distribution that includes the complete AI framework (X-Linux-AI), as well as the STM32Cube development tools. The STM32Cube firmware will run bare metal or RTOS on the Cortex-M33 embedded core.

ST is now delivering samples now and volume production of chips and boards is scheduled to begin in the first half of 2024.