This combination provides OEMs with a scalable entry-level option that marries high AI performance with ultra-low power consumption in a familiar Arm environment.
Alif Semiconductor has launched the Ensemble E1C MCU. This new microcontroller integrates advanced digital and analog capabilities with a low-power on-chip neural processing unit (NPU) in a compact form factor, significantly enhancing machine learning in space- and power-constrained embedded applications and IoT endpoints. The E1C MCU features a 160MHz Arm Cortex-M55 CPU core with Helium vector processing extension, an Arm Ethos-U55 NPU generating up to 46 GOPs, and up to 2MB of tightly coupled SRAM, all within a tiny 3.9mm x 3.9mm WLCSP package.
The company’s aiPM power management system and the E1C’s ultra-efficient AI operations make it ideal for battery-powered wearables and similar devices. The E1C supports local ML workloads such as object recognition, speech recognition, sensor fusion, and adaptive audio noise cancellation, reducing reliance on the cloud and enhancing user experiences.
Key features include:
- Efficient AI Performance: The Ethos-U55 NPU with 128 MACs/cycle delivers 46 GOPS, enhancing inferencing speed and power efficiency.
- Power Management: aiPM technology dynamically powers only the necessary logic and memory, achieving minimal system power consumption. Four power modes, including a 700nA Stop mode, ensure optimal efficiency.
- Memory and Connectivity: Up to 2MB of MRAM and zero wait-state SRAM, high-speed OctalSPI interface, and a variety of analog and digital peripherals.
- System Integration: Includes 12-bit SAR ADCs, 24-bit sigma-delta ADC, 12-bit DAC, and MIPI DSI display interface. Serial interfaces include USB 2.0, SDIO, 2x CAN FD, and I3C.
- Security: Advanced secure enclave for hardware root-of-trust, secure boot, cryptographic accelerators, and certificate management.
The E1C has three package options: 90-bump WLCSP for maximum space savings, 120-bump FBGA for maximum I/O, and 64-lead TQFP for durability in harsh environments. The DK-E1 development kit supports the E1C, which is fully compatible with the Arm ecosystem. E1C devices and kits will be available to lead customers in August 2024, with full production ramping in Q4 2024.
Reza Kazerounian, President and Co-Founder of Alif Semiconductor, stated, “The E1C series pushes the boundaries of localized ML with greater power efficiency in a smaller form factor, opening new opportunities for applications like health monitoring, distributed sensing, and audio processing. Developers can now reduce size, power consumption, and cost even further.”
