Specifications
Manufacturer
Alif Semiconductor
Kit Models
Balletto B1 StartKit (SK-B1) · Ensemble E1C StartKit (SK-E1C)
MCU Family
Balletto B1 (BLE MCU) · Ensemble E1C
CPU
Arm Cortex-M55 @ 160 MHz with Helium-M vector extension
NPU
Arm Ethos-U55 (128 MAC/cycle, up to 46 GOPS)
On-Chip Memory
2 MB SRAM
Non-Volatile Memory
1.8 / 1.9 MB NVM
Wireless (SK-B1)
BLE 5.3 + IEEE 802.15.4, complete RF path to chip antenna
Wireless (SK-E1C)
None — no built-in wireless
Camera
Arducam camera header
Expansion
mikroBUS Click Board socket
Audio
2× PDM microphones
Debug
On-board Segger J-Link debugger
Power
5V DC via USB Type-C; on-board TLV75201 3.3V LDO
I/O Voltage
Flex I/O jumper (JP3) selectable 1.8V or 3.3V
Instrumentation
Current-measurement jumpers on VDD_MAIN and VDDIO_1V8
Indicators
Green power LED, green user LED
Dimensions
86.37 × 71.09 mm
Pricing
~$49.99 (Mouser) · SK-B1 $51.28 (DigiKey)
Toolchain
VS Code with Ensemble CMSIS Device Family Pack; Alif Zephyr SDK; Balletto iOS/Android app
Overview
The Alif Semiconductor Balletto B1 StartKit (SK-B1) and Ensemble E1C StartKit (SK-E1C) are two identically laid-out evaluation boards built around Alif's Cortex-M55 class microcontrollers. The choice between them comes down to radio: the Balletto B1 SK-B1 carries BLE 5.3 and IEEE 802.15.4 with a complete RF path to a chip antenna, while the Ensemble E1C SK-E1C omits wireless entirely for designs where connectivity is handled by a separate module or is simply not required.
Both kits pair a 160 MHz Arm Cortex-M55 CPU — which includes Arm's Helium-M vector extension for DSP-class signal processing — with an Arm Ethos-U55 neural processing unit rated at 128 MAC/cycle and up to 46 GOPS, backed by 2 MB of SRAM and roughly 1.8–1.9 MB of non-volatile memory. Alif quotes an order-of-magnitude improvement in ML inference performance and power efficiency per inference versus running the same model on a previous-generation Cortex-M core without neural acceleration, which is the whole argument for the Ethos-U55 pairing on tinyML workloads.
The boards are equipped for immediate experimentation rather than bare minimum bring-up: an Arducam camera header, a mikroBUS Click Board socket for the standard Click accessory ecosystem, two PDM microphones for audio and voice work, and an on-board Segger J-Link debugger so no external probe is needed. A jumper selects 1.8V or 3.3V I/O logic levels, and current-measurement jumpers on VDD_MAIN and VDDIO_1V8 let developers profile real power draw. Development is supported through Visual Studio Code with the Ensemble CMSIS Device Family Pack, the Alif Zephyr SDK for Bluetooth work, and Alif's Balletto mobile app for iOS and Android. Both kits list at around $50.
Key Benefits
Neural acceleration at microcontroller power: the Ethos-U55 NPU delivers up to 46 GOPS for vision and sensor ML on a Cortex-M55 platform, enabling always-on inference in battery-class devices. Pick your radio at purchase: identical boards in a wireless-free (SK-E1C) and BLE 5.3 / 802.15.4 (SK-B1) configuration let teams prototype both topologies on the same firmware base. Instrumented for power and audio work: current-measurement jumpers, dual PDM microphones and a camera header make the kit usable for genuine power-profiling and voice/camera ML development, not just GPIO demos.
Applications
TinyML and always-on vision prototyping, battery-powered smart sensors, voice and audio edge interfaces, BLE and 802.15.4 IoT product development, portable point-of-sale and access-control devices, and hardware-accelerated sensor fusion evaluation.
Request a Quote — ALIF SEMICONDUCTOR BALLETTO B1 & ENSEMBLE E1C STARTKITS — CORTEX-M55 + ETHOS-U55
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →