MYIR Launches i.MX 93 based SoM for Industrial Applications

MYIR has launched a new embedded System on Module (SoM), the MYC-LMX9X, based on the NXP i.MX 9352 processor. This processor belongs to the high-performance i.MX 93 family, featuring dual ARM Cortex-A55 cores clocked at up to 1.7GHz and Cortex-M33 cores running at 250MHz. It also incorporates a 0.5 TOPS Neural Processing Unit (NPU), enabling cost-effective and lightweight AI applications.

i.MX93 Processor Block Diagram

The MYC-LMX9X has a compact size, measuring merely 37mm by 39mm. In addition to the powerful i.MX 9352 CPU, it offers1GB/2GB LPDDR4 RAM, 8GB eMMCstorage, 32KB EEPROM, and a Power Management IC(PMIC), providing a robust foundation for your next embedded design. Leveraging the full capabilities of the i.MX 9352 SoC, the MYC-LMX9X supports a rich set of peripherals, including 2x USB 2.0, 3x SD/SDIO 3.01, 2x Gigabit Ethernet, 2x CAN-FD interfaces, as well as MIPI-DSI, LVDS and RGB display interfaces, and MIPI-CSI and Parallel CSI interfaces, through a 218-pin LGA expansion interface. This flexibility allows it to cater to a wide range of applications, including automotive, industrial, and consumer IoT solutions. The MYC-LMX9X is capable ofrunningthe Linux 6.1Operating Systemandincludes comprehensive software resources such as kernel and driver source code, along with detailed user manuals and documentation to expedite customer development, enabling you to bring your products to market faster.

MYC-LMX9X (Top-view and Bottom-view)

Features

Mechanical Parameters

  • Dimensions: 37mm x 39mm
  • PCB Layers: 10-layer design
  • Power supply: 5V/1A
  • Working temperature: -40~85 Celsius (industrial grade)

OS Support

  • Linux 6.1.36

Processor

  • NXP i.MX 9352processor
    – Up to 1.7GHz Dual-core ARM Cortex-A55 CPU
    – 250MHz Real-time ARM Cortex-M33 co-processor
    – Up to 0.5 TOPS Arm® Ethos™ U-65 microNPU

Memory

  • 1GB/2GB LPDDR4
  • 8GB eMMC (supports 4GB/16GB/32GB)

Peripherals and Signals Routed to Pins

  • Power Management IC (PCA9451AHNY)
  • 218-pin LGA Expansion Interface
    – 2x RGMII
    – 2x USB2.0
    – 8x UART
    – 2x CAN FD
    – 8x I2C
    – 2x I3C
    – 8x SPI
    – 1x MIPI-DSI
    – 1x LVDS
    – 1x RGB
    – 1x MIPI-CSI
    – 1x Parallel CSI
    – 3x SAI
    – 1x SPDIF
    – 4x ADC
    – 1x JTAG
    – Up to 87x GPIOs

Note: the peripheral signals brought out to the expansion interface are listed in maximum number. Some signals are reused. Please refer to the processor datasheet and SOM pin-out description file.

MYC-LMX9X Block Diagram

The MYD-LMX9X Development Board, built around the MYC-LMX9X SOM, offers a comprehensive range of communication interfaces on its base board. These interfaces include RS232, RS485, two USB2.0, one USB OTG, two Gigabit Ethernet, one CAN bus, a Micro SD card slot, and an M.2 Socket for USB-based 4G/5G LTE Module with dual SIM card holders. Additionally, it incorporates a USB2.0-based WiFi module, an Audio interface, a Mini-CSI interface, and multiple display interfaces including HDMI, LVDS, and RGB. The MYD-LMX9X Development Board comes with a Quick Start Guide, one USB to TTL serial cable and a 12V / 2A power adapter. MYIR also provides optional add-ons such as the MY-CAM003M MIPI Camera Module, MY-LVDS070C 7-inch LCD Module, and MY-LCD70TP-C 7-inch LCD Modulefor enhanced functionality.It serves as a robust and reliable reference design for development based on i.MX 93 solutions.

MYD-LMX9X Development Board Top-view
MYD-LMX9X Development Board Bottom-view
MYD-LMX9X Development Board Block Diagram

MYIR provides two standard configurations for the MYC-LMX9X, offering either 1GB or 2GB of LPDDR4 RAM to accommodate different memory requirements. Additionally, customization is available to meet specific quantity demands. The MYC-LMX9X SOMs are competitively priced, and bulk discounts are offered for cost-effective purchasing.

More information about the MYC-LMX9X can be found at: https://www.myirtech.com/list.asp?id=757

New DART-MX93 System on Module by Variscite Brings Machine Learning to Compact, Cost-Optimized, Rugged Edge Devices

Based on NXP’s i.MX93 processor with Ethos™ neural processing unit

Variscite, a leading worldwide System on Module (SoM) designer, developer and manufacturer, today announced the newest member of the DART Pin2Pin family for machine learning on edge devices at a competitive price for markets like industrial, IoT, smart devices, and wearables.

At only 55 x 30 millimeters—about half the size of a credit card—the DART-MX93 enables machine learning, even in compact devices. Its energy-flex architecture provides efficient processing; built-in security ensures data and applications are protected.

Similar to the successful VAR-SOM-MX93 which was already launched by Variscite in 2023, the DART-MX93 is based on the same NXP i.MX93 processor with up to 1.7GHz Dual Cortex™-A55, and 250MHz Cortex-M33 real-time co-processor, equipped with the Arm® neural processing unit Ethos™-U65 microNPU. The new SoM includes robust features like 2x CAN bus, 2x GbE, and industrial temperature grade plus a wide range of features and connectivity options: camera inputs, audio in/out, ADC, 2x USB, certified Wi-Fi 6 dual-band 802.11 ax/ac/a/b/g/n with optional 802.15.4, BT/BLE 5.3, and LVDS display outputs.

“Variscite’s DART-MX93 is another breakthrough that offers advanced processing power and energy-efficiency with industrial-grade features at an extremely attractive price,” said Ofer Austerlitz, VP Business Development and Sales of Variscite. “The DART-MX93 is positioned to enable a new class of smart edge devices that are rugged, energy-wise, performant, and affordable in volume.”

Variscite offers the broadest range of SoMs based on NXP’s i.MX 9 series. The upcoming VAR-SOM-MX91, to be released in parallel with NXP’s iMX91, is ideal for price-sensitive devices. The DART-MX95, powered by NXP i.MX95, was created for powerful edge platforms that demand high performance, high-end graphics, AI/ML capabilities, advanced security and safety functions, and a rich set of high-speed connectivity options.

Longevity and Upgrade Path

Variscite’s DART-MX93 is included in the company’s long-term longevity plan and is expected to be available to customers for the next 15 years.

The DART-MX93 is part of the DART Pin2Pin family that gives Variscite customers a future-proof way to scale their device capabilities with new generations of DART SoMs. This ensures extended longevity for customer products, as well as reduces development time, costs, and risks for future product iterations. From the i.MX8M Mini, through the i.MX8M and iMX8M Plus, to the i.MX 95.

Availability

Variscite’s alpha customers can now place orders in production quantities for DART-MX93 SoMs starting at $39 per unit. DART-MX93 evaluation kits, which include the scalable VAR-DT8M custom board and an optional LVDS display with a touch panel, are also available.

For more information, please contact Variscite’s sales team at https://www.variscite.com/contact-us/

NanoCell V2.1 is A ESP32-C3 IoT dev board for Home Assistant

NanoCell V2.1 a ESP32-C3-based dev board by Frapais' lab in Greece. Preloaded with ESPHome, and includes BMS IC for low-power apps.

The NanoCell V2.1 is an ESP32-C3 powred dev board designed by Frapais’ lab in Greece. The board comes preloaded with ESPHome and features a battery management IC for battery-based low-power applications.

The board has A buck-boost converter that has been developed to minimize standby current consumption to 66uA, with the exclusion of the current drawn by the ESP32 module. Furthermore, the integrated battery management system (BMS) offers precise capacity measurement and safeguards attached Lithium batteries from potential hazards like overcharging. Additionally, the board features two LEDs functioning as power and charging indicators to communicate the board’s status effectively.

Previously we have written about many ESP32-C3-Based development boards like the SparkFun Pro Micro – ESP32-C3, Olimex ESP32-C3-DevKit-LiPo, and many others feel free to check those out if you are interested in the topic.

NanoCell V2.1 specifications:

  • Microcontroller – ESP32-C3 RISC-V microcontroller @ 160MHz, with Wi-Fi and Bluetooth 5 (LE)
  • Battery Management
    • battery capacity measurement IC
    • Li-ion/Li-po battery charging & protection ICs
    • Accurate battery capacity measurement IC (MAX17048), accessible via I2C on pins 2 and 3.
  • USB – USB Type-C for charging and uploading firmware
  • Buttons – Reset and Boot
  • LEDs – Charging and USB power LED indicators
  • Breadboard-compatible pin headers break out all the pins of ESP32-C3, USB, battery, and VCC voltage.

The NanoCell V2.1 is designed to work seamlessly with Home Assistant and ESPHome automation systems. A setup guide is available in the NanoCell-C3 GitHub repository.

NanoCell V2.1 integrates smoothly with Home Assistant and ESPHome automation systems. You can find a setup guide in the NanoCell-C3 GitHub repository. Priced at $14.49 on Tindie and $14.90 on Elecrow with bulk discounts available, it’s fully open-source, offering hardware schematics and initialization firmware in the GitHub repository.

Raspberry Pi Compute Module 4S Gets 2GB, 4GB, and 8GB Memory Upgrade

Raspberry Pi Compute Module 4S with expanded RAM options (2GB, 4GB, 8GB) featuring BCM2711 SoC, starts at $25, but minimum order is 200 units.

Raspberry Pi has recently announced the release of three new Raspberry Pi Compute Module 4S(CM4S) with expanded  RAM options now offering 2GB, 4GB, and 8GB variants. These new modules feature the same Broadcom BCM2711 quad-core Cortex-A72 SoC used in Raspberry Pi 4 and Raspberry Pi CM4 and have a starting price of only $25. The only odd thing about this is that Raspberry Pi Foundation did not make this for a single purchase, and you have to buy a minimum quantity of 200 to get your hand on one of these modules.

The CM4S stands out from other Compute Modules in a few ways. It keeps the same size and shape as the CM3+ but has a faster processor called the BCM2711 chip. It doesn’t have some features found in the CM4, like Wi-Fi, Bluetooth, USB 3.0, and PCI Express.

The reason for creating the CM4S is to help industrial users who are using CM3-based designs but can’t find the parts they need. The CM4S is designed to be an easy upgrade for these users because it fits into the same slot as the CM3+ while offering improved performance.

Raspberry Pi Compute Module 4S Specifications:

  • SoC:
    • Broadcom BCM2711
      • CPU: Quad-core 64-bit Cortex-A72 processor @ 1.5 GHz
      • GPU: VideoCore VI GPU
      • 3D Graphics: Supports OpenGL ES 3.0 and Vulkan 1.1
  • Video:
    • H.265 (HEVC) decoding up to 4Kp60
    • H.264 decoding up to 1080p60 and encoding up to 1080p30
  • Memory:
    • LPDDR4-3200 SDRAM with ECC, available in 1GB, 2GB, 4GB, and 8GB configurations
  • Storage:
    • eMMC flash options: 8GB, 16GB, or 32GB
    • CM4S Lite variant offers an option for 0GB eMMC Flash
  • Ports and Interfaces:
    • Display interfaces:
      • HDMI 2.0 port supporting up to 4Kp60
      • 2-lane and 4-lane MIPI DSI display interfaces
      • Composite TV out (PAL or NTSC)
    • Camera interfaces:
      • 2-lane and 4-lane MIPI CSI camera interfaces
    • USB:
      • 1x USB 2.0 port (high speed)
    • Other I/Os:
      • 46x GPIO signals
      • 1x SDIO 2.0 (available on CM4S Lite variant)
  • Supply Voltage:
    • Requires VBAT (2.5V to 5V) and +3.3V supplies
    • +1.8V is no longer used but can be supplied for backward compatibility
  • Dimensions:
    • 67.6mm × 31.0mm, compatible with JEDEC MO-224 mechanical specification for 200-pin DDR2, but not electrically compatible with DDR2 SODIMM modules
  • Production Lifetime:
    • Raspberry Pi Compute Module 4S will remain in production until at least January 2034.

The new Raspberry Pi CM4S, now with built-in eMMC storage, is available for purchase through Raspberry Pi Resellers (you can find a list on their website). If you’d prefer a version without the eMMC storage, the CM4S Lite is also an option. For more details and the latest information, visit the Raspberry Pi press release page.

Waveshare ESP32-S3-Tiny Board Cost $5 and Measures Only 23.50 x 18 mm

Waveshare ESP32-S3-Tiny Board is an ESP32-S3-powered development board in a 23.50 x 18 mm from factor. What makes this board different from the Waveshare RP2040-Tiny is that it features 34 multi-function GPIO pins compared to 23 found on the RP2040-Tiny board.

On the board, you have an ESP32 chip, an LDO, an FPC Connector, a WS2812B RGB LED, and a 3D chip antenna. You will not find any USB to UART convert on the board because it gets connected to the board via the onboard FPC connector, this unique approach saves cost and time. Additionally, it has Wi-Fi, BLE, SPI, I2C, UART, ADC, PWM, and more.

You can buy either the ESP32-S3-Tiny or the ESP32-S3-Tiny-Kit with the USB-C board directly from Waveshare’s website. If you’re developing a product, you might start with a kit and later order individual MCU boards as needed.

The board is compact, and the inclusion of the ESP32-S3 makes it an ideal choice for IIoT applications. For those looking to get started with the board, the company provides detailed onboard parts lists to help you get started.

Waveshare ESP32-S3-Tiny Board Specifications

  • MCU: Espressif Systems ESP32-S3FH4R2
  • CPU: Dual-core Tensilica LX7 @ up to 240 MHz with vector instructions for AI acceleration
  • Memory: 512KB RAM, 2MB PSRAM
  • Storage: 4MB QSPI flash
  • Connectivity: 2.4 GHz WiFi 4 and Bluetooth 5.0 LE with support for long-range and mesh network
  • GPIO Pins: 34 multi-function GPIO pins via 20x through and castellated holes, and 14 pads
  • Peripheral Interfaces: SPI, I2C, UART, ADC, PWM
  • FPC Connector: 8-pin FPC connector for adapting to USB Type-C port via optional adapter board
  • Antenna: 3D PCB Antenna
  • Power Management: ME6217C33M5G low dropout LDO capable of delivering 800mA (Max)
  • Additional Features: WS2812 RGB LED, power indicator, RESET button, and BOOT button
  • Physical Dimensions: 23.5 x 18 mm
Adapting USB Type-C Port Via Adapter Board with FPC Cable

This compact board measures a mere 23.5 x 18mm with a slim 2.45mm thickness. Its versatile 2.54mm pin pitch wraps around three sides, making it easy to integrate into your projects. You have excellent programming flexibility with support for ESP-IDF, Arduino IDE, and MicroPython. To get started, you’ll find all the necessary documentation (including pinout diagrams, installation guides, and resource links) conveniently located on the board’s wiki page.

 

The board is cleverly designed to add more I/O within its compact form factor. Waveshare has added solderable pads just beside the pins of the dev board, increasing the I/O pin count to 34.

Waveshare’s ESP32-S3-Tiny board is available for purchase on a few different platforms. You’ll find both the standalone board and the ESP32-S3 Tiny-Kit on Amazon for $11.49 and $12.49 (with shipping included). If you’re looking to save a little, try checking Aliexpress or shopping directly from the Waveshare store, where prices begin at a budget-friendly $4.99 that of course does not include shipping.

Waveshare UGV Rover is Powered By a Raspberry Pi

The Waveshare UGV Rover is an open-source educational and commercial platform that can be used for remote exploration, object recognition, autonomous navigation, and more. The Rover is built around a Raspberry Pi and can be configured with A Pi5 or Pi4B. There are also two additional boards within the rover, one is an ESP32 power driver and control board that controls the wheel the Pan-Tilt module, and many other things. There is also an audio board that handles all the Audio processing tasks. The Pi sits on top of these two boards, sends out the control commands, and does the AI and Image Recognition tasks if configured.

The rover has a 2mm thick aluminum body and six 80mm shock-absorbing tires, but it gets a four-wheel drive system controlled by the ESP32 sub-controller. The sub-controller also handles sensors, LiDAR, cameras, and more.

Previously we have written about other rover modules like the Turtle Rover, the Makeblock MBot Neo Rover, the open-source Tele-Robotics platform and more feel free to check those out if robots like this interest you.

ESP32 powred sub-controller

ESP32 powred sub-controller

The ESP32 powred sub-controller supports different communication protocols, including serial port, HTTP request, and ESP-NOW. The module supports ESP-NOW meaning it can communicate with other nearby rovers while allowing 4G/5G module expansion for communication. other than that the sub-module performs basic tasks like high-frequency PID controller, high-frequency inverse kinematics calculations, position interpolation, Pan-Tilt angle control, OLED screen control, read data from IMU, and battery voltage sensor.  Additionally, it can automatically achieve the camera vertical stabilization function.

Waveshare UGV Rover Audio Controlee and Processing Board
Audio Controlee and Processing Board

The audio driver board is specifically designed for a robot’s main controller, with a built-in USB interface that works well with various motherboards including Raspberry Pi 4B and Raspberry Pi 5, among others. It utilizes the SSS1629A5 audio control chip, that ensures a hassle-free plug-and-play interface. Additionally, it has an APA2068 audio amplifier chip to drive the speakers on the rover. Moreover, it integrates FE1.1S USB 2.0 HUB and CH340 chips, facilitating seamless data transfer from serial peripherals like Lidar to the main controller via a USB cable, without burdening the resources of the main controller excessively.

Now we know about all the features but we don’t know about the console mechanism of the rover for that the company provides a web interface through which you can control the rover with a PC, phone, or tablet.

Waveshare UGV Rover Specifications:

  • Core Control & Processing
    • Raspberry Pi 4B or 5 for computer vision and machine learning
    • ESP32 sub-controller for dedicated tasks
    • High-frequency PID controller for precise movement
    • Closed-loop speed control for consistent wheel speeds
  • Movement and Navigation
    • 6-wheel, 4-wheel drive for superior terrain handling
    • High-torque motors with encoders for accurate motion
    • IMU for orientation and balance sensing
  • Sensing & Feedback
    • Pan/tilt servo system with feedback for camera control
    • OLED screen for system status
    • Battery voltage monitoring for power management
    • Audio system: speaker, microphone, audio jack, text-to-speech
  • Connectivity
    • Fast networking on Raspberry Pi (Gigabit Ethernet, WiFi 5, Bluetooth 5)
    • ESP32 with 2.4GHz WiFi, Bluetooth, and ESP-NOW support
    • Optional 4G/5G for expanded communication range
  • Camera & Expansion
    • Pan-tilt module with 360°/120° motion, 5MP camera, stabilization
    • Mounting plate for LiDAR, battery, or custom devices
    • Optional rail and servo for tactical upgrades
  • Control & Power
    • Wireless gamepad for remote operation
    • Hotspot auto-creation for network flexibility
    • 3x 18650 UPS module for reliable power
    • Optional battery set
  • Construction
    • Sturdy 2mm aluminum chassis
    • Shock-absorbing 80mm tires
    • High-brightness LED for low-light vision

The UGV Rover boasts additional AI capabilities such as object, gesture, and face detection, motion tracking, vision line tracking, color recognition, and auto-targeting, all powered by multi-threaded computer vision. Additionally, the company plans to offer Ngrok tutorials, although they won’t provide accounts or servers. Compatibility extends to Debian Bookworm, ROS2-HUMBLE-LTS, and JupyterLab, with forthcoming guides and tutorials for all software features.

Waveshare offers customizable options for the UGV rover, allowing users to order without a Raspberry Pi if they already own one and to include the Pan-Tilt Module. Power plug options include US, EU, and UK variants. This Raspberry Pi 4/5-based robot promises enjoyable experiences, although documentation might be more comprehensive for other models like the SunFounder PiCar-X 2.0.

The UGV Rover PT PI5 AI Kit or UGV Rover PT PI4B AI Kit, including the Pan-Tilt Module, can be found on Amazon for $292.99 (including shipping). Waveshare also offers a UGV Rover base kit (without Pi or Pan-Tilt Module) for $244.99 in their store. More details can be found on there Wiki page.

SBC Case Builder v3.0: Design Your Custom Case for SBCs and Standard Motherboards

SBC Case Builder v3.0 is the latest iteration of the versatile case design utility, offering users the ability to create thousands of cases for popular single-board computers (SBCs) and standard motherboards. This new version boasts an impressive library of over 1,000 standard cases, with the option for further customization to suit individual needs.

The utility supports a wide range of SBCs, including those from Raspberry Pi, Hardkernel, Orange Pi, Radxa, and others. Additionally, it accommodates standard motherboards following form factors such as Mini-ITX, Pico-ITX, NUC, Nano-ITX, and more. Notably, SBC adapters adhering to these standards enable users to seamlessly integrate SBCs into cases designed for larger form factors, allowing for unique configurations such as installing a Raspberry Pi 5 into a mini-ITX case.

Originally starting as a command-line utility relying on OpenSDAD for DIY case design in April 2022, SBC Case Builder has evolved significantly. Edward Kisiel, also known as hominoids, swiftly introduced a GUI with the release of version 2 in October of the same year. Now, with the launch of SBC Case Builder v3.0, the utility has undergone further enhancements to provide users with a more intuitive and feature-rich experience.

ODROID-M1S and UPS Kit housed in a standard mini-STX

Specifications:

  • Hardkernel ODROID-C1+, ODROID-C2, ODROID-C4, ODROID-XU4, ODROID-XU4Q, ODROID-MC1, ODROID-HC1, ODROID-HC4, ODROID-N1, ODROID-N2, ODROID-N2+, ODROID-N2L, ODROID-N2LQ, ODROID-M1, ODROID-M1S, ODROID-H2, ODROID-H2+, ODROID-H3, ODROID-H3+, and ODROID-Show2
  • Raspberry Pi Pico, Pi Pico W, Pi Zero, Pi Zero 2 W, Pi A+/B+, Pi 3A+/3B/3B+, Pi 4B, Pi 5, CM1, CM3, CM3L, CM3+, CM4, CM4L, and CM4 IO Board
  • Pine64 Rock64, RockPro64, Quartz64 Model A and B, H64 model B, and Star64
  • Radxa ROCK4A, ROCK4A+, ROCK4B, ROCK4B+, ROCK4C, ROCK4C+, ROCK5B-v1.3(pre-release), ROCK5B, and the new Radxa NIO 12L
  • Khadas VIM1, VIM2, VIM3L , VIM3, VIM4
  • ASUS Tinker board, Tinker board-S Tinker board 2, Tinker board 2S, Tinker board R2, Tinker board S R2.0
  • Orange Pi 5, Orange Pi Zero, Orange Pi Zero2, Orange Pi R1/R1+ LTS
  • Libre Computer LePotato, SweetPotato, Tirtium-H2+, Tritium-H3, Tritium-H5, Solitude, Alta
  • NVIDIA Jetson Nano
  • Sipeed Lichee RV and Dock
  • StarFive VisionFive2 SBC
  • Digital Loggers, Inc Atomic Pi, an ultra-cheap x86 SBC from a defunct project. Probably out of stock forever now.
  • RAKwireless WisBlock RAK19007 base board for IoT prototyping
  • Standard motherboard form factor – SSI-EEB, SSI-CEB, ATX, Micro-ATX, DTX, Flex-ATX, Mini-DTX, Mini-ITX, thin Mini-ITX, Mini-STX, thin Mini-STX, Nano-ITX, NUC, Pico-ITX

Once you have generated a case design, you can use the resulting OpenSCAD file with a 3D printer or CNC machine to build the enclosure. More information is available on GitHub.

image: www.cnx-software.com

Radxa Unveils AICore SG2300x Module: Powering Edge AI with 32 TOPS Performance

Radxa’s AICore SG2300x Module Brings 32 TOPS to Bear on Edge AI, On-Device Generative AI

Radxa, a leading name in embedded computing, is set to revolutionize edge artificial intelligence (AI) with its latest offering: the AICore SG2300x system-on-module (SOM). Promising an impressive 32 tera-operations per second (TOPS) of compute performance, this module is poised to drive on-device AI to new heights.

At the heart of the AICore SG2300x lies the SOPHON SG2300x system-on-chip, boasting eight Arm Cortex-A53 processor cores clocked at up to 2.3GHz. Augmenting this processing power is a dedicated tensor processing unit (TPU) coprocessor, delivering a claimed 32 TOPS at INT8 precision, with capabilities extending to 16 TOPS in FP16/BF16 precision and 2 TOPS at FP32 precision. Complementing this computational prowess is a generous 16GB of LPDDR4X RAM and a 64GB eMMC storage module, expandable via an SDMMC interface.

Radxa’s AICore SG2300x Module

The AICore SG2300x is engineered to handle the demands of on-device AI, including large language models (LLMs) and generative AI models, with ease. Radxa highlights its compatibility with popular generative AI models such as generative pre-trained transformers (GPTs), the Stable Diffusion image generation model, and the ChatDOC model. For scenarios requiring even greater processing power, the module supports the “cascading” of two units to achieve up to 64 TOPS.

Radxa ensures seamless integration with leading machine-learning frameworks, including TensorFlow and PyTorch, through its BMNNSDK software development kit. The toolkit facilitates model optimization, and efficient runtime support, and provides access to a model zoo featuring pre-trained models like YoloV8, ResNet, PP-OCR, DeepSORT, and OpenPose.

In terms of connectivity, the module offers robust peripheral support, including one PCI Express 3.0 four-lane root complex and endpoint, up to three UART and three I2C buses, and 32 general-purpose input/output (GPIO) pins with two pulse-width modulation (PWM) channels. Additionally, it features dual gigabit Ethernet PHYs and hardware decoding for 32 channels of H.264/H.265 video at 1080p25, alongside encoding for 12 channels of H.264/H.265 at the same resolution and refresh rate.

While pricing and availability details are yet to be disclosed on the Radxa website, plans to release an open-source carrier board design and a design guide document for enthusiasts keen on building their systems.

Introducing the Banana Pi BPI-M5 Pro: A Next-Generation AIOT Platform

Banana Pi BPI-M5 Pro

The Banana Pi BPI-M5 Pro presents a cutting-edge solution for AIOT (AI + IoT) applications, boasting the advanced features of the second-generation 8nm high-performance Rockchip RK3576 chip design. This powerhouse is equipped with a 6 TOPS computing power NPU (Neural Processing Unit) and supports up to 32GB of expansive memory, ensuring seamless performance for a wide range of tasks.

One of the standout features of the BPI-M5 Pro is its support for 8K video encoding and decoding, pushing the boundaries of multimedia capabilities. Moreover, it offers a plethora of connectivity options with dual gigabit Ethernet ports, WiFi 6, and BT5, enabling high-speed data transfer and seamless networking. Additionally, it provides various video outputs, ensuring versatility in display options for different applications.

Block Diagram
Block Diagram

Hardware Specifications:

  • SoC: Rockchip RK3576
  • CPU: Integrated with four Cortex-A72 cores @ 2.2GHz and four Cortex-A53 cores @ 1.8GHz, along with a separate NEON co-processor.
  • GPU: ARM Mali G52 MC3 GPU
  • NPU: Up to 6 TOPs computing power (INT8), supports INT4/INT8/INT16 mixed operations.
  • VPU/Encoding & Decoding:
    • Hardware Decoding: Supports H.264, H.265, VP9, AV1, and AVS2 up to 8K@30fps or 4K@120fps.
    • Hardware Encoding: Supports H.264 and H.265 up to 4K@60fps, high-quality JPEG encoder/decoder supports up to 4K@60fps.
  • RAM: 8/16GB 32-bit LPDDR4x, default is 16GB. RK3576 supports a maximum of 16GB.
  • Flash: 32/128GB eMMC, default is 128GB eMMC.
  • Operating Voltage: Wide input voltage range, from 4.5V to 23V (voltage error ±5%).
  • Operating Temperature: 0°C to 80°C
  • Operating Systems:
    • Official Rockchip Support: Android 14, Debian 11, Buildroot
    • Third-party Support: Armbian
  • PCB: 8-layer PCB board design
  • Weight: 43g
  • Dimensions: 92mm × 62mm

The Banana Pi BPI-M5 Pro is not for sale on SinoVoIP’s Aliexpress store just yet possibly because the OS images are not available yet, but you’ll find the ArmSom Sige5 Pro Max for pre-order for $148 with 16GB RAM and 128eMMC flash. It does not seem particularly good value as the Orange Pi 5 Pro with similar features (but only one GbE port) and a more powerful Rockchip RK3588S processor goes for $128 on Amazon in the same 16GB/128GB configuration. Maybe it is still worth it for use cases where the low-profile design is important.

Louder Raspberry Pi – Home Media Center powered by Raspberry Pi and TAS5805M DAC

Louder Raspberry Pi

Louder Raspberry Pi is an open-source home media center, developed by Andriy Malyshenko of Sonocotta, built on the Raspberry Pi Zero W or Zero 2 W and Texas Instruments TAS5805M DAC. It’s a compact audio entertainment platform designed for ease of use.

Louder Raspberry Pi combines the computing power of the Raspberry Pi Zero and the audio processing capabilities of the TAS5805M DAC in a sturdy aluminum case. It delivers up to 25W per channel stereo output and is powered by a 65W+ USB-C PD3.0 adapter, suitable for medium-to-large speaker systems. It supports both Wi-Fi and Ethernet connectivity.

The Raspberry Pi Zero was chosen for its simplicity in development and compact size, making it ideal for a home media center.

Louder Raspberry Pi incorporates the computing power of the Raspberry Pi Zero and the Hi-Fi audio processing capabilities of TI’s TAS5805M DAC in a compact, aluminum case.
Louder Raspberry Pi incorporates the computing power of the Raspberry Pi Zero and the Hi-Fi audio processing capabilities of TI’s TAS5805M DAC in a compact, aluminum case.

Key specifications of Louder Raspberry Pi include:

  • Single-board computer (SBC): Raspberry Pi Zero W or Raspberry Pi Zero 2W
  • Digital-to-analog converter (DAC): Texas Instruments TAS5805M  with integrated D-Class amplifier
  • Ethernet: Wiznet W5500 SPI Ethernet
  • USB: 1x USB-C PD3.0 for power delivery and serial port
  • Audio Output: 2x 22W at 20V input over USB-PD
  • Additional features: 1x IR reader, 2-pin speaker terminal
  • Power: 65W+ USB-C power adapter
  • Dimensions: 88 x 38 x 100mm

Louder Raspberry Pi is part of Sonocotta’s line of Raspberry Pi-based media center devices, including Loud Raspberry Pi and Hi-Fi Raspberry Pi. It can be configured with Volumio, Mopidy, or other music player software, with instructions available in the Sonocotta media center repository on GitHub repository.

The device costs $35 for the base board and DAC on Tindie. The Raspberry Pi Zero W version costs $55, and the Zero 2 W version is $60. Adding a Lenovo 32GB Class 10 SD card costs an additional $10. Detailed information, including board schematics and PCB designs, is available in the Sonocotta media center repository for those interested in building their own Louder Raspberry Pi.

TOP PCB Companies