Eoxys Xeno+ Nano ML is a wireless machine learning (ML) board with either Nuvoton NuMicro M2354 or STMicro STM32L4 microcontroller, InnoPhase IoT’s Talaria TWO ultra-low power Wi-Fi and BLE 5.0 module, and the Syntiant Core 2 NDP120 neural decision processor we first noticed in the Arduino Nicla Voice module a few weeks ago.
The boards/modules are designed for intelligent and secure IoT devices for smart home, industrial, and medical automation applications, and the company claims it can be used in Wi-Fi IoT sensors with up to 10+ years thanks to the low-power chips and circuitry used in the design.
Eoxys Xeno+ Nano ML specifications:
- General purpose MCU (one or the other)
- Wireless module
- Innophase Talaria TWO ultra-low-power 2.4GHz 802.11b/n/g WiFi 4 and Bluetooth LE 5.0 module with either PCB or u.FL antenna
- WiFi Active Mode Power – 31mA (Rx Mode @ 1Mbps 802.11b), 178mA (Tx Mode @ 1Mbps 802.11b +17.5dBm)
- Wi-Fi Power Save Mode Power at 802.11b, 1Mbps (Clean Environment, @ 3.3V) – 57uA (DTIM10)
- Deep Sleep Mode Power (@3.3V, Memory Retained) – 11-19uA (RTC, memory retained, depends on the amount of SRAM retained)
- Module Standby Power – 50uA (STM32L4 in standby and Wi-Fi module is in shutdown mode)
- Machine learning DSP – Syntiant NDP120 Neural Decision Processor (NDP) with one Syntiant Core 2 ultra-low-power deep neural network inference engine, 1x HiFi 3 Audio DSP, 1x Arm Cortex-M0 core up to 48 MHz, 48KB SRAM
- Storage – 8MB serial Flash on SPI bus
- Audio – On-board Stereo PDM Microphones
- USB – 1x USB Type-C port power and serial debug
- I/Os via 2x 20 castellated holes
- NDP120 – I2S, PDM, TDM, and I2C.
- GP MCU – UART, SPI, I2C, ADC, DAC, PWM, and GPIOs
- Debugging – USB-C port as above, 3-pin SWD pins for software programming and debugging
- Power Supply
- 5V via USB-C port
- 2.5V to 5V battery power from a non-rechargeable (or) rechargeable battery.
- Dimensions – 68 x 38 mm
- Weight – 5 grams
The STM32L4 board came first, and the NuMicro M2354 board was just announced yesterday. They probably launched the N2354 as the Arm Cortex-M23 should be more power efficient and it also includes Arm TrustZone support for security. I can’t see any other major differences between the two variants of the Xeno+ Nano ML board. The pinout diagrams are slightly different and while the STM32L4 is programmed with the STM32Cube IDE, the M2354 relies on the Arm Keil IDE. Both boards run FreeRTOS real-time operating systems.
It’s quite surprising or disappointing that they could not find a single-chip solution with microcontroller, WiFi, and AI instructions that could match the requirements, and they had to go with a three-chip solution. I’d assume it’s because NuMicro M2354, Talaria TWO (also found in a WiFi AI camera development kit with one+ year of battery life), and Syntiant NDP120 all excel at low power consumption, and an all-in-one chip like the ESP32-S3 would not meet their power consumption requirements. The solution will mostly be interesting for smart voice applications that require a very long battery life. You can watch a demonstration of the STM32 board with Amazon Alexa in the video below.
The XENO+ Nano ML module (NuMicro N2354, part number: XNO-W102N-M1-V10) is sold for 6,000 Indian Rupees (~$72 US) on the company’s store. More details may be found on the product page and the recent press release about the NuMicro N2354 module.
Thanks to TLS for the tip.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.