Efinix has introduced the Sapphire RV64 SoC, a configurable 64-bit RISC-V soft-core processor designed for its Titanium and Trion FPGAs. It is based on the VexiiRiscv core and can be configured with one to four RISC-V cores running at up to 400 MHz.

The Sapphire RV64 has a seven-stage pipeline, configurable L1 caches, optional 64–512 KB L2 cache, and 4–512 KB of on-chip RAM. It supports DDR3, LPDDR4x, and HyperRAM with up to 512-bit AXI interfaces and external memory from 4 MB to 8 GB. Other features include an optional FPU, branch predictor, prefetchers, SV39 MMU, up to 32 GPIOs, five I²C, three SPI, three UARTs, timers, watchdog, and up to eight user interrupts.

Efinix Sapphire RV64 SoC specifications:

• CPU
  • 1 to 4x VexiiRiscv 64-bit cores with a 7-stage pipeline (fetch, align-decompress, decode, issue, execute, memory, and writeback)
  • Configurable system clock frequency from 20 MHz to 400 MHz
  • Supports RV64IMACFD_Zba_Zbc_Zbs extensions (including single- and double-precision floating-point, atomic, compressed, bit manipulation, and cache block management)
  • Optional SV39 Memory Management Unit (MMU) for Linux support
  • Optional Physical Memory Protection (PMP) unit with 4 KB granularity
  • L1 Cache – Configurable multi-way instruction and data cache (up to 8 ways, 4 KB per way)
  • L2 Cache – Optional shared 8-way L2 cache ranging from 64 KB up to 512 KB
• Memory
  • On-chip RAM – 4 KB to 512 KB with a built-in boot loader for SPI flash
  • External Memory (via half-duplex AXI3 or full-duplex AXI4 interface)
    • DDR3 up to 400 MHz / 800 Mbps
    • LPDDR4x up to 1,600 MHz / 3,200 Mbps
    • HyperRAM up to 250 MHz / 500 Mbps
    • Supports from 4 MB to 8 GB
• I/O
  • Up to 32x GPIOs
  • Up to 5x I2C masters and up to 3x SPI masters
  • Up to 3x UARTs (running at 115,200 baud rate)
  • 1x AXI4 master channel and up to 5x APB3 slave interfaces for custom user peripherals
• Debug
  • Standard RISC-V debug
  • Up to 4x hardware breakpoints
  • Up to 4x performance counters (the user guide also lists 1 to 9)
  • Optional co-debug with the FPGA debugger
•  Misc
  • Up to 8x user interrupts
  • 3x user timers
  • Watchdog timer
  • Dynamic branch predictor
  • Software and hardware prefetchers
  • Custom instruction interface with 1,024 IDs
• Power – Varies depending on the FPGA, clock speed, and enabled features

The Efinity IP Manager lets you configure the SoC and generate a custom IP core. For firmware, Efinity RISC-V Embedded Software IDE provides an Eclipse-based environment for C/C++ development and debugging.

Depending on the configuration, the SoC can boot from an SPI flash device directly into the on-chip RAM or copy a larger user application into external memory before execution. Multi-core configurations utilize symmetric multi-processing (SMP), sharing the workload across all processors to improve overall system throughput.

We wrote about Efinix’s first RISC-V soft SoCs for Trion FPGAs, the Ruby, Jade, and Opal cores, 32-bit designs based on VexRiscv. After that, the company moved the idea into hard silicon: Titanium introduced a quad-core 32-bit RISC-V block based on the Sapphire soft core, and the Topaz family had the same kind of block on a lower-cost 16 nm FPGA. They also introduced the Topaz Tz170b devkit and the newer Titanium Edge, which were the boards and chips those cores run on. Sapphire RV64 is the soft-core follow-up: 64-bit, still built in the FPGA fabric, and aimed at designs that have outgrown those 32-bit cores.

The Sapphire RV64 SoC is available in the Efinity IP Manager for all Titanium and Topaz FPGAs and Trion devices from T20 and above. It requires Efinity 2026.1 or newer, and Efinix provides free Efinity software licenses. More details are available on the support page (with links to a datasheet and a user guide) and in the press release.

The post Efinix Sapphire RV64 SoC – Configurable RISC-V soft-cores with DDR3, LPDDR4x, and HyperRAM for Efinix FPGAs appeared first on CNX Software - Embedded Systems News.

DSPi is a full-featured audio DSP firmware for the Raspberry Pi Pico, Pico 2, and other Raspberry Pi RP2040/RP2350 boards, which turns them into a USB sound card with an integrated DSP.

It allows users to make use of essential tools like room correction, active crossovers, parametric EQ, time alignment, and more, and essentially works as a regular sound card on your computer.

DSPi firmware’s capabilities:

• Dual-Core DSP – EQ processing is split across both cores on both RP2040 and RP2350.
• USB Audio Interface – Supports 16-bit and 24-bit PCM input at 44.1, 48, and 96 kHz, working on macOS, Windows, Linux, and iOS.
• S/PDIF Input – 24-bit PCM stereo audio at 44.1 or 48kHz.
• 24-bit S/PDIF or I2S Outputs – Up to 4x independent stereo output slots (8x channels on RP2350, 4x channels on RP2040).
• Subwoofer Output – Dedicated mono PDM output channel with a high-performance 2nd-order delta-sigma modulator, enabling direct subwoofer output without the need for a second DAC.
• Per-Channel Preamp – Independent gain control for each USB input channel (L/R), applied as PASS 1 of the DSP pipeline before any other processing.
• Matrix Mixer – Routes either or both USB input channels to any output with independent gain and phase invert per crosspoint (2×9 on RP2350, 2×5 on RP2040).
• Parametric Equalization – Up to 10 PEQ bands per channel with 11 filter types (peaking, low/high shelf, low/high pass, notch, all-pass, plus gentle first-order shelf and all-pass variants). 110 total filter bands on RP2350, 70 on RP2040.
• Active Crossovers – Up to 4x dedicated crossover filters per output channel: Linkwitz-Riley, Butterworth, and Bessel low/high-pass up to 8th order (48 dB/oct), for multi-way speaker and subwoofer integration.
• Volume Leveller – RMS-based, stereo-linked, soft-knee upward compressor that lifts quieter content toward a target level without ever amplifying loud passages. Optional 10 ms lookahead, configurable speed and max-gain ceiling, with a -6 dBFS gain-reduction safety limiter.
• Loudness Compensation – Volume-dependent EQ based on the ISO 226:2003 equal-loudness contour standard.
• Headphone Crossfeed – BS2B-based crossfeed with interaural time delay (ITD) reduces unnatural stereo separation for headphone listening.
• Master Volume – Device-side output ceiling (-128 to 0 dB, with a true-mute sentinel) applied at the very end of the signal chain, independent of USB host volume and DSP processing.
• Per-Output Gain & Mute – Independent gain and mute controls for each output channel.
• Time Alignment – Per-output delay (up to 85ms) for speaker/subwoofer alignment with automatic latency compensation between S/PDIF/I2S and PDM output paths.
• Configurable Output Pins – All output GPIO pins (including I2S BCK/MCK) can be reassigned at runtime to suit custom PCB layouts.
• 10-Slot Preset System – Save, load, and manage up to 10 complete DSP configurations with user-defined names.
• Diagnostics – Per-channel peak/clip metering, USB PHY error counters (CRC, bit-stuff, timeout, overflow, sequence), buffer fill statistics, S/PDIF DMA starvation counters per output slot, and CPU load reporting per core.
• Firmware Update via USB – A vendor command reboots the device into the UF2 bootloader to push a new firmware via the DSPi Console application without a physical BOOTSEL press.

Both Raspberry Pi RP2040 and RP2350 implementations are similar, with the cores (over)clocked at 307.2 MHz, 44.1 / 48 / 96 kHz sample rates, 24-bit output bit depth, 10 user presets, S/PDIF input, and so on, but there are also some notable differences shown in the table below.

Besides the firmware, Weeb Labs also developed the DSPi Console GUI application to enable full control over the DSPi firmware on macOS, Windows, and Linux hosts. It enables real-time control, filter management, device configuration, and monitoring & visualization.

If you prefer working in the command line, DSPi Terminal is another option, basically offering the same capabilities as DSPi Console.

You’ll find the repositories for the DSPi firmware, DSP Console ports, DSPi Terminal, and more (STM32 port, ESP32-S3 front panel) on Weeb Labs’ GitHub account. He also posted an introduction on Audio Science Review in February 2026, and the thread is still active and popular with 121 pages, plus the DSPi firmware repo on GitHub already has 1.2k stars, so the project appears popular in the audio community. It was developed with the help of AI (Claude),  but an AI analysis of the code indicates the code is high quality for hobby software, although some C files appear to be fairly long (a lot of that due to comments).

Thanks to Hedda for the tip.

The post DSPi firmware turns Raspberry Pi RP2040/RP2350 into a USB sound card with an onboard DSP engine appeared first on CNX Software - Embedded Systems News.

Nuvoton has added MA35D05KH67C, MA35D05KI67C, and MA35D05KJ67C, three new LQFP128 MPUs, to the NuMicro MA35D0 series, along with the NuMaker-IoT-MA35D0-A2 (Chili Pro) evaluation board. Designed for industrial AI, smart gateways, and renewable energy systems, this heterogeneous MPU has two 64-bit Arm Cortex-A35 cores at up to 650 MHz, plus a dedicated 180 MHz Arm Cortex-M4 for real-time work.

Nuvoton’s press release calls Chili Pro the “industry’s smallest Linux platform,” but the board is 45 × 45 mm, which is bigger than other boards we wrote about previously, like the VoCore2 (25.4 × 25.4 × 2.8 mm), the Milk-V Duo 256M (51 × 21 mm), NanoPi NEO (40 × 40 mm), and even the Raspberry Pi Zero 2 W (65 × 30 mm). Still, considering that it fits full-sized RJ45 and USB ports along with the other interfaces, it is a pretty compact and capable board.

Nuvoton NuMaker-IoT-MA35D0-A2 evaluation board specifications:

• SoC – Nuvoton MA35D05KI67C
  • Application cores – Dual-core 64-bit Arm Cortex-A35 @ up to 650 MHz, 32KB I-cache and 32KB D-cache per core, 512KB shared L2 cache
  • Real-time core – Arm Cortex-M4 @ 180 MHz; IEEE 754 compliant FPU and DSP extensions
  • Memory – 256MB DDR3L stacked in the package (MA35D05K series up to 512MB)
• Storage
  • 256 MB Quad SPI NAND Flash (Winbond W25N02JWZEIF)
  • MicroSD 3.0 card slot
• Networking – Gigabit Ethernet RJ45 port with transformer via RTL8211FDI RGMII PHY
• USB
  • 1x USB 2.0 Type-A host port
  • 1x USB 2.0 micro USB port
  • 1x USB Type-C port for power only
• Expansion – Two GPIO extension connectors (CON2, CON3) with up to 4x CAN FD, 1x SPI, 2x I2C, 1x I2S, 2x ADC, and multiple UART interfaces (4-bit x 1, 2-bit x 5)
• Debugging – UART0 debug port and SWJ (JTAG+SWD) header
• Misc – User key button, PMIC reset key, user LED, and boot configuration DIP switch
• Power Supply – 5V/2A input via USB Type-C port with DA9062-3A PMIC
• Dimensions – 45 x 45 mm

The three MPUs use the same dual Cortex-A35 and Cortex-M4 architecture, and the only difference is memory configuration. The MA35D05KH67C has 128MB of memory, while the MA35D05KI67C and MA35D05KJ67C have 256MB and 512MB, respectively.  The CPU architecture and main peripherals remain the same across all three variants.

Nuvoton provides the board support package and NuWriter utility for programming the board through its OpenNuvoton GitHub repositories. The BSP includes Yocto 5.0 (Scarthgap), Yocto 3.1.3 (Dunfell), Buildroot 2024.02 and 2021.02, plus a bare-metal Non-OS package.

The software stack also includes U-Boot, TF-A, OP-TEE 4.2 and 3.9, and Linux 6.6.y and 5.10.y. Nuvoton also provides Linux applications and the libjpeg-turbo-vc8000 library for image and video processing. Buildroot and Yocto support lets you build customized Linux images, while the Non-OS BSP provides a bare-metal option for applications that don’t require Linux.

The new MA35D0 MPUs include Nuvoton’s Trusted Secure Island (TSI) and Arm TrustZone for isolating secure resources. They also support Secure Boot and hardware cryptographic accelerators for AES, RSA, and ECC. The security features are designed to support industrial security requirements such as IEC 62443.

The NuMaker-IoT-MA35D0-A2 (Chili Pro) is listed on Nuvoton’s TECHDesign store for $49. But the company doesn’t provide pricing details for the new MA35D05K MPUs on its store, Digi-Key, or Mouser. More information about the MPUs is available on the product page, and a few additional details are available in the press release.

The post NuMaker-IoT-MA35D0-A2 (Chili Pro) ultra-compact evaluation board features MA35D0K Cortex-A35/M4 MPU appeared first on CNX Software - Embedded Systems News.

The XIMEA MU003TG-SY-UC is a miniature industrial Time-of-Flight (ToF) camera with a 0.3MP Sony IMX556 CMOS sensor, global shutter, 640 × 480 resolution, and a 60 FPS frame rate. Its compact 26.4 × 26.4 × 28.67 mm enclosure and 31-gram weight make it suitable for space-limited embedded vision systems, robotics, UAVs, kiosks, 3D scanning, face recognition, and other machine-vision applications.

The module uses a USB 3.2 Gen 1 interface through a 14-pin I-PEX Cabline SS connector for data, power, and control. A separate 10-pin Cabline V connector provides synchronization with external VCSEL illumination. The camera does not include an illumination source or optical band-pass filter, so these components must be provided separately for depth sensing.

XIMEA MU003TG-SY-UC specifications:

• Sensor – Sony IMX556 CMOS Time-of-Flight sensor, global shutter, no on-sensor filter
• Resolution – 0.3 megapixels (640 × 480); active area 6.4 × 4.8 mm
• Pixel Size – 10 µm x 10 µm
• Optical Format – 1/2-inch (8.0 mm diagonal)
• Digitization – 12-bit
• Readout – 1, 2, or 4 ToF phases; 60 / 120 / 240 phases per second, tested at a 60 FPS transport limit
• Downsampling – x1, 2×2, and 4×4 binning (640×480, 320×240, 160×120)
• Exposure – 0.25 µs to 1000 µs
• Host Interface – USB 3.2 Gen 1 (5 Gbps) via Micro-Coaxial I-PEX Cabline SS 14-position receptacle
• Illumination Interface – Micro-Coaxial I-PEX Cabline V 10-position receptacle (5V power, High-speed LVDS switching, I2C telemetry)
• Lens Mount – C-mount (standard, pre-mounted); S-mount (optional, reduces depth to 23.67 mm)
• I/O – 2x Non-isolated digital inputs/outputs (3.3V logic) and 2x digital outputs
• Power
  • Supply – 5 V (4.5–5.5 V)
  • Consumption – 0.45 W (Idle), 1.80 W (Typical streaming), 2.00 W (Max peak)
• Dimensions – 26.4 x 26.4 x 28.67 mm (with C-mount)
• Weight – 31 grams (without adapters)
• Operating Temperature – 0 to 50 °C (ambient)

Each pixel has two charge-storage nodes, A and B. Incoming light is split between them in sync with the illumination modulation. The host can read A only, B only, A and B as separate images, or the combined results A+B and A−B, plus RAW grayscale. Depth is calculated on the host, or in the CamTool ToF plugin, rather than delivered as a finished point cloud. The sensor is also sensitive in the visible spectrum, but simultaneous high-quality depth and visible imaging need different exposure and illumination, and will reduce performance.

The diagram above shows the MU003TG-SY-UC, a modular vision system, and what you need to build a working depth-sensing setup around the camera. At the center (marked in orange) is the small ToF camera, which captures depth information. You can attach either a C-mount or S-mount lens and use mounting brackets to install it. The camera connects to a computer through its micro-coax USB 3 connection using the supplied cables and adapter. For actual ToF depth sensing, you also need the separate VCSEL-SLIM-W illumination board, which provides the infrared light used to measure distance. An optional USB hub can connect multiple cameras, while an additional optical camera can be added when a normal image is needed alongside the depth data.

The camera works with XIMEA’s xiAPI through C/C++, C#, and Python, with software packages available for Windows 10/11, Linux (Ubuntu), and macOS 10.8 or later. It also supports GenTL/GenICam, allowing integration with OpenCV, MATLAB, HALCON, and LabVIEW. Since the camera is currently an Engineering Sample, support is provided through the Beta software package, while LTS releases may not yet support the model.

The xiCOP utility handles camera configuration, diagnostics, and firmware updates. The same software package includes CamTool, which has a dedicated ToF plugin to calculate and display depth maps from the camera’s phase data. The camera also supports both free-run and triggered acquisition, including edge- and level-triggering through software or hardware, as well as burst acquisition. More information is available on the support page.

XIMEA has not announced a price for this miniature IMX556 ToF Sensor, as it’s currently listed as an Engineering Sample. A working ToF setup also requires the VCSEL-SLIM-W illumination board, control and data cables, and a Micro-B adapter, while the USB hub, mounting brackets, and S-mount adapter are optional accessories. You can find more information on the product page.

The post XIMEA MU003TG-SY-UC – A miniature, modular 0.3MP USB 3.0 ToF Sensor based on Sony IMX556 appeared first on CNX Software - Embedded Systems News.

The GEEKOM A5 Pro 2026 Edition is a pocket-sized 0.47L mini PC available at a $70 discount for the company’s Prime Day Sale until October 7. Powered by the 6-core Ryzen 5 7530U (up to 4.5GHz), it handles browsing, documents, video calls, and AI-agent tasks such as drafting, summarizing, and planning with ease. Dual HDMI 2.0 and dual USB-C ports support up to four displays (or one 8K display), enabling efficient multi-window workflows. Meanwhile, dual SO-DIMM slots (up to 64GB) and dual M.2 slots (up to 5TB total) provide plenty of room for future upgrades.

With 2.5GbE, Wi-Fi 6E, six USB ports, an SD card reader, and VESA mounting support, along with an aluminum body that has passed 339 reliability tests, the A5 Pro is a quiet, clutter-free desktop companion for offices and home workspaces running Windows 11 Pro or Linux. Backed by GEEKOM’s 3-year warranty, the A5 Pro is built for long-term reliability and can deliver stable performance for 5–10 years with proper care and maintenance.

GEEKOM A5 Pro 2026 Edition specifications:

• SoC – AMD Ryzen 5 7530U
  • CPU – 6-core/12-thread processor up to 4.5GHz with 16MB cache
  • GPU – Integrated AMD Radeon Graphics
  • TDP – 15W
• System Memory – 16GB dual-channel DDR4-3200 SODIMM memory (expandable up to 64GB)
• Storage
  • 512GB or 1TB M.2 2280 PCIe Gen 4 x4 SSD (upgradeable up to 2TB)
  • M.2 2242 SATA SSD slot (up to 1TB supported)
  • SD card reader slot
• Video Output
  • 2× HDMI 2.0 ports
  • 2x DisplayPort via USB Type-C ports
  • Up to 4x independent displays supported
• Audio
  • 3.5mm audio jack
  • Digital audio output via HDMI ports
• Networking
  • 2.5GbE RJ45 port
  • Wi-Fi 6 and Bluetooth 5.2
• USB Ports
  • 3x USB 3.2 Gen 2 Type-A ports
  • 2x USB 3.2 Gen 2 Type-C ports
  • 1x USB 2.0 Type-A port
• Misc – Power button, GEEKOM IceBlast 3.0 intelligent fan control, dual copper heat pipes
• Power supply – 19V/3.42A (65W) via DC jack
• Dimensions – 117 × 112 × 38.6 mm
• Weight – 652 grams

We reviewed the GEEKOM A5 Pro 2026 Edition earlier this year, first with the pre-installed Windows 11 Pro OS, and then Ubuntu 25.10 to evaluate Linux support and performance. We found it performed adequately with both operating systems, and it was especially interesting for people wanting a quiet mid-range mini PC with low power consumption.

To benefit from the $70 off offer, you’ll need to enter the TYA5PDFB coupon code when ordering on the GEEKOM US website to bring the price down to $499 (from $569). The company also launched a GEEKOM Prime Day Sale up to 30% OFF on its Amazon store for the entire product lineup, including the GEEKOM IT13 MAX (Intel Core Ultra 9 185H), which can also be purchased at a $140 discount using the TYT3XDFB discount code on the GEEKOM website.

The post Save $70 on GEEKOM A5 Pro 2026 Edition mini PC during Prime Day Sale (Sponsored) appeared first on CNX Software - Embedded Systems News.

The DEBIX M8391-01 is a compact industrial SBC built around the MediaTek MT8391 (Genio 720) processor, designed for edge AI, computer vision, robotics, and intelligent IoT applications. The SoC’s 9-TOPS NPU enables local AI inference without relying on cloud processing.

The board features up to 16GB LPDDR4 memory and up to 64GB eMMC storage, with M.2 PCIe 2.0 NVMe SSD and MicroSD card expansion. It also offers a 4-lane MIPI CSI camera input, MIPI DSI up to 2.5K and eDP 1.4 display interfaces, Gigabit Ethernet with PoE, Wi-Fi 6, Bluetooth 5.4, five USB ports, and a 40-pin expansion header. With an industrial operating temperature range (-40°C to 85°C), this credit card-sized, Raspberry Pi-inspired SBC can be used for edge AI vision, robotics, industrial IoT, and other embedded systems.

DEBIX M8391-01 specifications:

• MediaTeck Genio 720 (MT8391)
  • CPU
    • Commercial-grade (default) – 2x Arm Cortex-A78 cores @ up to 2.6 GHz, 6x Arm Cortex-A55 cores @ up to 2.0 GHz
    • Industrial-grade (optional) – 2x Arm Cortex-A78 cores @ up to 2.4 GHz, 6x Arm Cortex-A55 cores @ up to 2.0 GH
  • GPU – Arm Mali-G57 MC2 up to 1.1 GHz
  • VPU
    • Video Decode – Up to 1920 x 1200 @ 60fps
    • Video Encoder – Up to 1920 x 1200 @ 60fps
  • ISP – 2x ISP
  • NPU – NPU850 AI accelerator up to 9 TOPS
• Memory – 4GB LPDDR4 by default (8GB and 16GB optional)
• Storage
  • 32GB eMMC flash (8GB, 16GB, or 64GB optional)
  • M.2 KEY-M 2242 socket for PCIe 2.0 NVMe SSDs
  • MicroSD card slot
  • 2Kb EEPROM
• Display
  • eDP 1.4 (4-lane) via USB Type-C, supporting up to 4K @ 60fps
  • MIPI DSI (4-lane) connector, supporting up to 2.5K resolution
• Camera – MIPI CSI (4-lane) connector, supporting image sensors up to 32MP @ 30fps
• Audio
  • 3.5mm Headphone and Microphone combo jack
  • I2S audio via the expansion header
  • Speaker header (1W)
• Networking
  • Gigabit Ethernet RJ45 port with PoE support (requires separate PoE PD module)
  • Wi-Fi 6 (2.4/5GHz, 2T2R) and Bluetooth 5.4 with dual IPEX-1 antenna connectors (two external antennas included)
• USB
  • Dual-stacked USB Type-A combo with 2x USB 2.0
  • Dual-stacked USB Type-A combo with 1x USB 2.0 (upper) and 1x USB 3.0 (lower)
  • USB Type-C port with OTG and eDP 1.4 support
• Expansion – 40-pin (2×20) GPIO header with 2x UART, 2x I2C, I2S audio, 3x SPI, and 1x PWM; also routes 5V power supply, PoE power output, system reset, and ON/OFF signals
• Debugging – 1x UART Debug header, 1x JTAG header
• Security
  • Arm TrustZone
  • Secure boot
  • Hardware cryptographic engines
  • OTP (one-time programmable memory)
  • Random number generator
• Misc
  • Power (Red/Blue) and RUN (Green) LEDs
  • Power and Reset buttons
  • Hardware Watchdog
  • RTC
• Power Supply
  • Input – 5V/3A DC input via USB Type-C
  • Consumption – Maximum 1.74W (commercial-grade) and 1.57W (industrial-grade)
• Dimensions – 85.0 x 56.0 mm (±0.5mm)
• Weight – 45g (±0.5g)
• Temperature Range:
  • Commercial-grade – 0°C to 70°C
  • Industrial-grade – -20°C to 70°C, or -40°C to 85°C

DEBIX has been around for a few years, releasing several NXP-based SBCs like the original DEBIX Model A (i.MX 8M Plus) and, more recently, the DEBIX T62P-01 featuring a TI AM62P industrial SoC. The M8391-01 is DEBIX’s first compact industrial SBC based on a MediaTek processor.

In terms of software support, the company mentions that the M8391-01 supports Android, Yocto Linux, and Ubuntu. They also highlight the Mali-G57 MC2 GPU with OpenGL ES, Vulkan, and OpenCL support (implying 3D graphics acceleration and GPGPU support), and the 9 TOPS NPU for local AI workloads such as vision, transformer, and LLM inference. Information about software support is rather limited, and the documentation website appears to focus on the company’s NXP boards.

DEBIX has not announced a price for the M8391-01 yet, and the board is currently inquiry-only. However, the company’s boards eventually get sold online. More information about the board can be found on its product page.

The post DEBIX M8391-01 industrial SBC features MediaTek Genio 720 SoC with 9 TOPS NPU, dual-display support appeared first on CNX Software - Embedded Systems News.

Geekworm, in collaboration with Mcuzone, has launched the UPS Scap 5V5A, a new supercapacitor UPS HAT for the Raspberry Pi 5 designed to keep the Raspberry Pi running long enough (a few seconds to a few minutes) to save data and shut down safely, rather than provide hours of backup power.

Available in 100F, 60F, and 22F versions, the HAT has a built-in PD chip that negotiates a true 5V/5A output. This gives the Raspberry Pi 5 enough power to run properly without triggering the low-voltage warnings that can happen with some power supplies. The module connects to the Raspberry Pi 5 through pogo pins to detect power loss and supports I2C for checking voltage and current in real time. It also features dual independent power inputs, so you can power it with either a wide-voltage DC barrel jack or a standard USB-C PD charger.

Geekworm UPS Scap 5V5A supercapacitor UPS HAT specifications:

• Compatibility – Raspberry Pi 5 (Back-compatible with Raspberry Pi 4B, 3B, and Orange Pi via the standard USB output)
• Supercapacitor options – 100F, 60F, 22F
• Power input
  • 9V–24V DC via 5.5 × 2.1 mm barrel connector
  • 12V via USB Type-C PD input
  • Only one charging input (DC or USB-C) should be used at a time
• Power output
  • USB Type-C port with built-in PD handshake to bypass restricted peripheral “Low Power” warnings
  • Standard USB Type-A port
  • Total shared output of up to 5V / 5A across both ports
  • Unpopulated Pogo Pin output option
• Operating modes – Hardware solder pads to switch between:
  • Long-endurance Mode (Solder point L) with a 1.6V cutoff
  • Performance Mode (Solder point P) with a 2.4V cutoff
•  Misc
  • Main output power switch
  • Hardware discharge button paired with a discharge indicator LED (to safely drain the cap)
  • 5-level capacity LED indicators and a main power indicator LED
  • I2C pins for INA219-based supercapacitor voltage and current monitoring
  • Power loss detection pin (GPIO26) to signal when external power drops
  • Wake-up/sleep pin for power recovery signaling
• Dimensions – 85 × 56 mm

The supercapacitor UPS HAT supports software-based power management through GPIO26 and I2C. GPIO26 detects external power status, whereas the onboard INA219 provides real-time supercapacitor voltage and charge/discharge current readings over I2C. The company also provides reference scripts for power-loss detection, monitoring, automatic shutdown, and power recovery/wake-up on Raspberry Pi OS and Ubuntu, although they mention that the supplied shutdown example should be tested and adapted before unattended use. More information is available on the Wiki.

Unlike traditional 18650 lithium-ion UPS boards, like the SupTronics X1202 and SupTronics X1208, this module uses supercapacitors instead of batteries. The 100F version provides only about 9 seconds at 5V/5A, but can run for up to 3.5 minutes at 0.5A in Long Endurance Mode. This makes it better suited for short power cuts or giving the Raspberry Pi enough time to shut down safely and avoid SD card corruption. The trade-off is a much longer lifespan of over 500,000 charge cycles, faster charging in around 15 minutes, and no battery-related fire or chemical leakage concerns.

The Geekworm supercapacitor UPS HAT for Raspberry Pi 5 is available on AliExpress for $41.00 for the 100F version. On Geekworm’s official store, the 22F version costs $44.90, the 60F version costs $46.90, and the 100F version costs $48.90. The package includes the main UPS module, standoffs, and a U-shaped USB-C-to-C adapter board for routing power to the Raspberry Pi. Geekworm also has an Amazon store, so the Supercapacitor HAT may soon be available there as well.

The post Geekworm UPS Scap 5V5A – A supercapacitor UPS HAT for Raspberry Pi 5 appeared first on CNX Software - Embedded Systems News.

We have seen ESP32s used for everything from simple IoT devices to AI, audio, and vision projects. But while looking for a new project, I found out about ESP-SDR firmware, which turns the ESP32 into a native Software-Defined Radio (SDR) without additional hardware.

This open-source firmware bypasses the ESP32’s fixed-function Wi-Fi and Bluetooth modems and taps directly into the internal ADC signal chain to capture raw I/Q baseband samples. It then sends these samples to a computer over USB or UART, where they can be processed and viewed as a radio spectrum or waterfall display, turning supported ESP32 chips into low-cost software-defined radio receivers without requiring additional SDR hardware.

Key features and supported hardware:

• Supported chips – ESP32, ESP32-C3, ESP32-C5 (dual-band 2.4/5 GHz), ESP32-C6, ESP32-C61, ESP32-S2, ESP32-S3, and the preview ESP32-S31
• Tuning range – Nominally 2.4 GHz ISM (2.2–2.7 GHz on some chips) and 5 GHz (4.8–6.0 GHz on ESP32-C5); The C61 can tune up to 2.7 GHz and capture LTE Band 7 and 5G NR Band n7 signals
• Sample rate and bandwidth – Up to 80 MSa/s with ~13–54 MHz bandwidth, depending on the chip
• On-Chip FFT – The firmware supports on-chip FFT and sends compact spectrum data instead of raw I/Q samples

According to the developers, Espressif engineers likely left a debug I/Q sampling path in the silicon for factory wafer testing. By using LLMs to reverse-engineer the adctrig function within Espressif’s librftest library, the team figured out how to configure special debug registers.

Once enabled, the ESP32 writes raw I/Q samples directly into its internal SRAM. The firmware then uses a ring buffer to move the samples from SRAM and transfer them to a computer over USB or UART. To display the captured signals, you need ESP-WebSDR, a browser-based tool that connects to the ESP32 over USB and shows the radio spectrum and waterfall in real time. You can also use it to flash the ESP-SDR firmware directly from a supported web browser (like Chrome or Edge).

The ESP32 is fast and can capture data at up to 80 MSa/s, which produces around 2.56 Gbps of data, but getting all that data off the chip is the main problem. USB and UART cannot handle the full data rate, so ESP-WebSDR captures short I/Q snapshots to build the spectrum and waterfall. This means it can miss signals that arrive between capture windows. For continuous streaming, the team is working on SoapyESPSDR for the ESP32-S31, which can stream I/Q data over Gigabit Ethernet at 8 to 16 MSa/s to software such as GNU Radio and Gqrx.

Previously, we have written about several ESP32 radio and SDR projects, but all of them still relied on an external receiver or on processed Wi-Fi data. Just a few days ago, we wrote about how LakeShark firmware turns a LilyGO T-Display P4 into a handheld SDR scanner where the ESP32-P4 has no radio of its own and relies on an RTL-SDR Blog V3 or V4 dongle as the RF front end. We have also covered Chen Liang’s ESP32-C5 Wi-Fi analyzer, which monitors 2.4 GHz and 5 GHz Wi-Fi signals, while ESP-WIFI-CSI works with channel-state information. nRFBOX V2 and ESP32-DIV use an external nRF24 to scan 2.4 GHz. Compared to all of that, ESP-SDR relies on the ESP32’s built-in radio, so no extra SDR hardware is needed. This makes basic spectrum analysis possible on an ESP32 board that can cost less than $5.

Developers Florian Euchner and Zoltan Doczi of the ESPARGOS project found this feature while working on ESPARGOS One, an eight-antenna setup for Wi-Fi sensing. With ESP-SDR, it can also work as an eight-channel SDR and detect and locate signals such as Bluetooth LE and Zigbee in the 2.4 GHz band.

Both ESP-SDR and ESP-WebSDR are open-source projects released under the GPL-3.0 license. You can find the source code on GitHub and try the browser-based SDR demo with a compatible ESP32 development board. They also left a note about AI usage in the ESP-WebSDR repo:

Parts of this code are AI-generated. While we put a lot of manual effort into pyespargos and the firmware for our ESPARGOS One arrays, we don’t have the time to manually review all of the source code for ESP-WebSDR.

The post ESP-SDR firmware turns ESP32 into a 2.4/5 GHz Software-Defined Radio (SDR) appeared first on CNX Software - Embedded Systems News.

Hello, today I am going to review the UUNA TEK iAuto, an automatic handwriting machine designed to reproduce physical handwriting using a real pen on real paper. It also supports wireless connectivity, spreadsheet-based mail merge, custom fonts, and font variation features that can make the writing look more natural. The machine can handle paper sizes up to A3 and supports pens with a diameter of up to 18 mm. The UUNA TEK iAuto works with the manufacturer’s free iAuto software, which is available for Windows on both x64 and Arm-based systems, as well as for macOS on Apple M1, M2, M3, and Intel-based Macs. Most of the following review focuses on the machine’s writing capabilities, including tests with text, SVG graphics, different writing speeds, various pens, and different paper types.

Unboxing

The parcel was shipped from China by land transportation and took around six days to arrive at my office in Thailand. The machine was securely packed in a large, heavy, and sturdy plywood box. Several accessories were also included, such as a power adapter, USB cables, a Wi-Fi antenna, Teflon tape, and a protective machine cover.

Hardware and software setup

The overall setup is easy, and both the hardware and software installation are straightforward. The manufacturer provides all of the necessary software, drivers, and supporting files through a Google Drive folder.

First, I removed the unnecessary packing materials and accessories from the machine. For the initial setup, the machine needs to be connected to the computer using a USB cable. For this review, I used the supplied USB Type-B to Type-A cable and then connected the power cable. Next, I downloaded all the files from the manufacturer’s Google Drive folder and installed the iAuto Windows Software. The installer estimated the required disk space at around 390 MB and also installed the necessary CH340 USB-to-serial driver.

The iAuto Windows Software itself is free, but registration is required to unlock its full functionality. The Machine ID needed for registration can be found under the Help/Machine Information menu. After entering the required registration information, the iAuto Windows software was ready to use.

The next setup step was installing the pen. Clicking the Change Pen button in the Pen Refill Statistics panel moves the pen holder down into the installation position. I then loosened the screw and inserted the pen at the desired position. The machine supports both diagonal and vertical pen orientations, and for this review I mainly used the vertical position. I adjusted the pen until its tip lightly touched the support bed, tightened the screw to secure it, and then clicked OK. The pen holder moved back to its upper position, and the machine was ready to use.

In addition to the USB connection, the machine also supports Wi-Fi connectivity. To configure the wireless connection, the machine must first be connected via USB so that the Wi-Fi SSID and password can be entered through the Network Settings menu. Once the setup is complete, the assigned IP address can be used to connect to the machine wirelessly.

First test with the UUNA TEK iAuto handwriting machine

My iAuto Windows Software version is 3.9.9, and it provides five drawing modes: Table, Simple Note, Character, SVG, and Drawing Board. For this review, I mainly used the Table mode to evaluate the machine’s writing performance. I began by clicking the Characters button in the Add panel to create a text box. I set the encoding to UTF-8, entered some song lyrics as the test text, and then clicked the Add button.

Each text box is displayed with a green outline and small red control icons at its corners for moving, scaling, and rotating the object. Without carefully reading the user guide, it took me a few minutes to become familiar with the software interface and its operation. In brief, before performing any action on a text box, it must first be selected or activated by clicking one of these red control icons. Other selection methods, such as dragging the mouse over the text box or clicking directly on the green outline, do not work. We can also click or right-click the control icon to access additional operations, such as Lock, Delete, Edit, and Copy.

Font deformations

In addition to functioning as a general pen plotter, the iAuto also includes a useful feature for deforming fonts and strokes to make the text look more like natural handwriting. Character variation can be controlled through the Font Deformation settings, which include options for size, spacing, tilt, vertical stretch, and horizontal stretch. The software also provides Stroke Deformation settings, including angular stretch, length stretch, and performance, allowing further adjustment of the writing style.

Writing test with default settings

I left all settings unchanged and performed a quick writing test by clicking Current, followed by the Write button in the Write panel. The machine immediately started writing, and the result is shown in the image below. For this test, I used a blue ballpoint pen on generic 80 gsm paper. I could already see some variation applied to the character shapes and spacing, which made the writing look less mechanical. Overall, I was very satisfied with the output.

I then tested longer passages copied from Wikipedia using multiple fonts while keeping the other settings at their default values. The results were still very impressive.

Drawing an SVG clipart

Although the machine is mainly advertised as a handwriting machine, it also supports vector graphics. SVG objects can be moved, scaled, and rotated in the same way as the text objects described earlier, and the software also supports placing multiple SVG graphics on a single page.

To test this feature, I asked ChatGPT to generate a dummy floor plan as an SVG file, and the plotted output is shown below. Overall, the result was acceptable, but I noticed some missing lines and a few inconsistencies in the plotted geometry. For example, some dimension lines were not positioned correctly relative to the walls.

I then downloaded an additional SVG file from Wikimedia Commons and repeated the test, but several attempts produced similar results. As shown in the image below, some misalignment between drawing elements was still present, such as the shifted cockpit canopy frames in the top view. Regarding the missing-line issue I encountered earlier, I tested several other generated SVG files and did not experience the same problem again. This suggests that the missing lines may have been caused by an improperly generated SVG file rather than by the machine itself.

However, since the machine is primarily designed and advertised as a handwriting machine rather than a precision drawing plotter, I did not consider the remaining misalignment a major issue. Another small issue I noticed is that the paper size is reset or changed whenever a new SVG file is imported, so the paper size may need to be checked and set again each time.

Creating custom fonts

The manufacturer provides several fonts through the supplied Google Drive folder. Installing them is straightforward, as the font files can simply be copied into the /gcodefonts folder inside the software installation directory. Additional fonts can also be downloaded directly from the provider’s cloud through the iAuto Windows software.

The manufacturer also provides a Font Creator tool for creating custom fonts. During this review, only the Windows version of Font Creator was available, and I did not find versions for other operating systems in the provided Google Drive folder.

To create a test font, I clicked the New Font button, which opened a pop-up window asking for basic font information. The Font Name and Author fields are required, and a drop-down list is also provided for selecting a predefined character template. I selected English and then clicked OK.

The Font Creator then enters drawing mode. The software supports several character drawing methods, including brush, pen, and geometric drawing tools. For this review, I used the Brush mode. The screenshot below shows the software in Brush drawing mode, with the available character list displayed on the right side. Each character must be selected and drawn manually, one by one. After completing the characters, the font can be saved or exported as a .gfont file. I then returned to the main iAuto Software application, opened the Font Manager menu, and imported the newly created font using the Import button.

Next, I tested creating a Thai font. In this case, I selected Empty from the Type drop-down list and manually added the required characters through the Add Character menu. This opens a pop-up window where characters can be entered into a text area. For the Thai font test, I selected UTF-8 encoding, entered a short Thai phrase, “ทดสอบ”, and clicked the Add button to insert the characters into the main window. All of the five characters from the Thai phrase then appeared in the character list on the right side, allowing me to draw them one by one in the same way as in the previous test. After completing the font, I exported it as a .gfont file and imported it back into the iAuto software.

In general, the software was able to load, display, and write Thai characters. However, as expected, there were some issues with the positioning of Thai vowels and other combining marks. Apart from this limitation, the custom Thai font worked as expected. Editor (CNXSoft) note: still not good enough for Thai students to cheat on their home work.

Testing mail merge and integration with Spreadsheet

One of the UUNA TEK iAuto’s most useful features is its spreadsheet-based batch writing function, which works somewhat like a mail merge. In brief, we can create text cells in the software, bind them to spreadsheet cells, and then attach the spreadsheet file to the batch function. This makes it possible to generate multiple personalized outputs automatically.

I started testing this feature by preparing a sample spreadsheet. I asked ChatGPT to generate five random people, each with five fields of information, and then exported the data as an .xls file.

In the iAuto software, I created a new table cell with a 1 × 1 layout and resized it to fit the required content. I then selected the text box and used the Binding function to assign the corresponding spreadsheet table and cell information. After that, I copied and resized the text box until I had all five required fields arranged on the page. I then opened the Batch function and selected the target spreadsheet file. Finally, I tested the batch output by using the page up and page down controls in the Batch panel. This allowed me to preview each record and confirm that the spreadsheet data was being inserted into the correct fields. Then, I tested writing on 10.8 × 23.4 cm envelopes.

As expected, the actual writing performance was good. However, I found that the envelopes were sometimes creased or wrinkled, especially when they were fed backward through the machine, as shown in the image below.

Other tests

Pens and papers

Next, I tested the machine with several types of pens and different paper thicknesses. For dry or semi-dry writing tools, I used a ballpoint pen and a gel pen. I also tested wetter writing tools, including an acrylic brush pen and an acrylic paint pen. For the paper tests, I used standard white Double A 80 gsm paper, black 110 gsm paper, orange 160 gsm paper, and white 200 gsm paper. Overall, the machine handled most of these materials very well. I did not experience any problems with the 200 gsm paper.

The following video demonstrates writing with a white acrylic brush pen on 110 gsm paper.

One issue I noticed with wet pens is that the paper may move back and forth during plotting, especially when drawing graphics. This can cause wet ink to transfer onto the machine’s rollers and may also smear or damage the output. The problem was much less noticeable when writing normal text, since the machine usually writes line by line progressively. Therefore, dry or fast-drying pens are generally more suitable for larger graphics or jobs that require frequent paper movement.

Print speed and output quality

I also compared the writing speed and output quality using the different speed settings available in the iAuto software. The software provides four preset modes: High Quality (Slow), Medium Quality (Medium), Low Quality (Fast), and Ultrafast. There is also a Custom mode, which allows the writing speed to be adjusted manually, typically within a range of around 8,000 to 10,000 rpm.

To compare writing speed and output quality, I started with the fastest setting, Ultrafast, and then gradually reduced the speed. I manually moved each test text box downward so that the results from the different speed settings could be compared on the same sheet. The following video demonstrates the writing speed comparison. Briefly, I could not see any noticeable difference in output quality between the different speed settings.

Testing writing small font

Next, I tested how small the machine could write by reducing the font size from 70 down to 10. For this test, I selected the slowest writing mode to give the machine the best chance of reproducing fine details accurately. Font sizes from 70 down to 30 remained clearly readable with good character shapes. At size 20, the text was quite small but still readable. At size 10, however, many of the lines began to merge together, making the characters difficult to distinguish, and the sentence was no longer readable at a normal viewing distance.

At size 20, the text was quite small but still readable. At size 10, however, many of the lines began to merge together, making the characters difficult to distinguish, and the sentence was no longer readable at a normal viewing distance.

Testing the provided API

The manufacturer mentions API functionality several times on its website, so I wanted to see what could actually be done with it. With the machine connected and using iAuto software version 3.9.9, I selected the Tool / API menu. This opened a web browser with some basic API documentation, while additional details for each API could also be found by viewing the source code of the corresponding page.

I first tested the login API directly through a web browser to confirm that the API was accessible. Since the API functions are called through HTTP requests, I then tested them using Postman. I tried the login request, machine status request, and SVG writing function. All of these tests worked successfully. I was able to retrieve the current machine status and also send an SVG file to the machine using an HTTP POST request.

The API itself appears useful, especially for developers who want to integrate the machine into their own software or automated workflows. However, the available documentation is still quite limited. In the version I tested, only a small number of functions were clearly documented, with very little information about more advanced writing and machine-control features.

Conclusion

Overall, I am very satisfied with the machine’s build quality and, more importantly, with the quality of its writing output. Throughout my testing, the UUNA TEK iAuto performed reliably with different fonts, writing modes, pens, paper types, SVG graphics, spreadsheet-based batch data, and even API-based control. Its handwriting-oriented features also make it more interesting than a conventional pen plotter.

One minor issue for me is the software’s UI/UX design. As mentioned earlier, I found the interface somewhat difficult to understand during my first few minutes of use, and it took some time to become familiar with the workflow.

I would like to thank UUNA TEK Co. for providing the UUNA TEK iAuto for this review. The machine is available for purchase from the UUNA TEK website for $2,999.00 (Standard version) and $3,149 (Advanced version). You can also get 15% OFF with the coupon code CNXVIP15. Alternatively, you’ll find the iAuto handwriting machine on Amazon for under $3,000.

The post UUNA TEK iAuto review – A professional handwriting and drawing machine appeared first on CNX Software - Embedded Systems News.

AAEON has introduced a new version of its UP Xtreme PTL Edge Panther Lake mini PC with the “Air” variant featuring Frore Systems AirJet solid-state active cooling solution.

The company says the AirJet passive cooling technology makes the Panther Lake mini PC 35% thinner, 43% lighter, and capable of operating silently at under 21 dBA. This is mostly achieved by removing the thick (and heavy) heatsink found on the original design. The mini PC still benefits from up to an Intel Core Ultra X7 358H with 180 TOPS of AI performance, a 40-pin GPIO header, two RS-232/422/485 COM ports, and a wide 19 to 36V DC input range.

UP Xtreme PTL Edge Air specifications:

• Intel Core Ultra Series 3 Panther Lake SoC (one or the other)
  • Intel Core Ultra X7 Processor 358H
    • CPU – 16-core (4P + 8E + 4LPE) @ 1.9/1.5/1.5 GHz Base frequency, 4.8/3.5/3.5 GHz Turbo frequency
    • Cache – 18 MB Intel Smart Cache
    • GPU – 12-core Intel Arc B390 GPU @ 2.5 GHz up to 122 TOPS (INT8)
    • AI accelerator – 50 TOPS NPU; CPU + GPU + NPU combined: 180 TOPS
    • Processor Base Power (PBP) – 25 Watts
  • Intel Core Ultra 7 Processor 356H
    • CPU – 16-core (4P + 8E + 4LPE) @ 1.9/1.5/1.5 GHz Base frequency, 4.7/3.5/3.3 GHz Turbo frequency
    • Cache – 18 MB Intel Smart Cache
    • GPU – 4-core Intel Graphics @ 2.45 GHz up to 40 TOPS (INT8)
    • AI accelerator – 50 TOPS NPU
    • Processor Base Power (PBP) – 25 Watts
  • Intel Core Ultra 5 Processor 325
    • CPU – 8-core (4P + 0E + 4LPE) @ 2.1/1.6 GHz Base frequency, 4.5/3.4 GHz Turbo frequency
    • Cache – 12 MB Intel Smart Cache
    • GPU – 4-core Intel Graphics @ 2.45 GHz up to 40 TOPS (INT8)
    • AI accelerator – 47 TOPS NPU
    • Processor Base Power (PBP) – 25 Watts
• System Memory – Up to 128GB dual-channel DDR5-7200 via two SODIMM slots
• Storage – 2x M.2 2280 M-Key sockets for NVMe SSDs
• Video Output
  • 2x HDMI 2.1 ports
  • 2x DP 2.1 via USB4 Type-C ports
• Audio – 3.5mm audio jack (mic+line-out)
• Networking
  • 2x 2.5GbE RJ45 ports via Intel Ethernet I226-IT controllers
  • Optional WiFi/Bluetooth via  M.2 2230 E-key module
• USB
  • 2x USB 3.2 Gen 2 Type-A ports
  • 2x USB4 Type-C ports with DisplayPort Alt mode
• Serial – 2x RS-232/422/485 DB9 connectors
• Expansion
  • 40-pin GPIO header
  • M.2 2230 E-Key (PCIe Gen 4 [x1], USB 2.0) socket for WiFi and Bluetooth or other expansion
  • 2x M.2 2280 M-Key (PCIe Gen 4 [x4], PCIe Gen 5 [x4]) sockets for storage or other expansion
• Security – Onboard TPM 2.0
• Misc – Optional VESA mount/Wallmount
• Power Supply – 19 to 36V DC input; AT/ATX type
• Dimensions – 236 x 129 x 35.8mm (original was 218 x 129 x 49.5mm; note: longer design might be including the wall mount option)
• Weight – 710 grams (original was 1.42 kg)
• Temperature Range
  • Operating: -10 – 45˚C; (note: the original was rated 0°C ~ 60°C, 0.7m/s airflow)
  • Storage: -40˚C ~ 80˚C
• Humidity – 0% ~ 90% relative humidity, non-condensing
• Certifications – CE/FCC Class A, RoHS Compliant, REACH
• MTBF – 971,676 hours
• Certifications – CE/FCC Class A, RoHS Compliant, REACH

AAEON provides support for Windows 11 IoT Enterprise and Ubuntu 24.04 LTS for the mini PC. Everything is the same as the original model, except for the size, weight, and operating range. The choice of processors was also updated since the initial release, but I can see that’s now the same for the original model with a thick heatsink.

Frore Systems AirJet solid-state active cooling modules looked promising when they were introduced in January 2023, but we hadn’t heard much about them after ZOTAC added some to the PI430AJ Pico mini PC later that year. I understand AirJet modules add quite a bit to the system cost ($100+), and AAEON’s industrial/embedded customers may not be as cost-sensitive as retail customers.

You should soon be able to get samples of the UP Xtreme PTL Edge Air on AAEON’s e-shop, where the Core Ultra 5 325 model is listed for $1366, but currently out of stock. Additional information may be found on the product page.

The post UP Xtreme PTL Edge Air Panther Lake Mini PC gets AirJet solid-state active cooling solution appeared first on CNX Software - Embedded Systems News.

Avaota F2 is the first SBC based on an Allwinner V861 dual-core 64-bit RISC-V SoC with 128MB on-chip DDR3 memory, support for 4K cameras, H.265 video codec, and a 1 TOPS AI accelerator.

It’s an update to the earlier Avaota F1 camera board based on an Allwinner V821 SoC. The new open-source hardware F2 SBC offers several benefits, including support for both Full HD and 4K camera sensors, motor control for the PTZ (Pan-Tilt-Zoom) feature, and improved audio support through speaker (one) and microphone (two) connectors. However, the F1 also included WiFi, which the Avaota F2 lacks.

Avaota F2 specifications:

• SoC – Allwinner V861M2-XXX
  • CPU
    • Dual-Core RISC-V XuanTie C907 (RV64GCBV/RV32GGCBV) clocked up to 1.4GHz with RVV 1.0 extensions
    • Single-core RISC-V XuanTie E907 (RV32IMAFC) clocked up to 800MHz
  • VPU
    • Video Encoder
      • H.264/H.265 up to  4K @ 25fps
      • (M)JPEG up to 8192×8192
    • Video Decoder – (M)JPEG up to 1080p60
  • AI accelerator – 1 TOPS (INT8) NPU dubbed “AI-ISP 2.0”
  • Memory – 128MB DDR3
• Storage
  • 16MB NOR flash (PY25Q128)
  • MicroSD card slot
• Display – Support for SPI screens via headers, including a 3.5-inch 320×480 display and a 1.54-inch 240×240 display
• Camera
  • 4-lane MIPI CSI connector
  • 2-lane MIPI CSI connector
  • Support for GC2083 (1920×1080 @ 30fps) and GC8613 (3840×2160 @ 15fps)
• Audio
  • Speaker connector
  • Two microphone connectors
• USB – USB Type-C port for power, data, and serial output
• Expansion
  • 2x 17-pin headers (through and castellated holes) with up to 28x GPIO
  • 8x castellated holes for motor control (useful for PTZ camera)
• Debugging – UART serial and ADB USB debugging are supported
• Misc
  • Power and FEL (Firmware flashing) buttons
  • User LED
• Power Supply
  • 5V via USB-C port
  • Integrated AXP333 power management chip (also used for PTZ)
• Dimensions – About 40 x 22mm (6-layer PCB)

The Avaota F2 is open-source hardware with PDF schematics, EasyEDA Pro PCB files, Gerber files, a 3D STEP file, a bill of materials, and pick-and-place instructions available on GitHub under a Creative Commons CC0 license.

GitHub lacks software information, but the developer shared a screenshot showing the board runs OpenWrt 21.04 (fork) with Linux 6.6, a few months ago, as part of Allwinner’s Tina Linux SDK. Instructions to download the SDK can be found on the documentation website, along with tutorials to use Allwinner V861 features; sadly, in Chinese only at this time.

The Avaota F2 is now sold for $51.59 on AliExpress, however it’s from “Chip Board House Store” which often sells hardware at twice the price it should be…. The AliExpress page also lists the STM32-based “Avaota Hypercard CKLink” JTAG/Serial debugging board for $27.89, which could be required for low-level programming/debugging.

Via GLGH_ on X

Updated: This article was initially published on March 19, 2026, and updated following availability on AliExpress

The post Avaota F2 – Allwinner V861 RISC-V SBC targets AI cameras with PTZ and audio support appeared first on CNX Software - Embedded Systems News.

ASRock Industrial has introduced the NUC Core 300 BOX series of mini PCs and the NUC Core 300 motherboards built around Intel Core Series 3 Wildcat Lake-U processors. The lineup includes the NUC BOX-320, with the Intel Core 5 320 SoC, and the NUC BOX-304, which uses the Core 3 304 SoC. The company also offers the NUC-320 and NUC-304 as standalone motherboards for custom systems.

The systems support up to 64GB DDR5-6400 with In-Band ECC, dual M.2 PCIe Gen4 storage, dual 2.5GbE, Wi-Fi 6E, Bluetooth 5.3, triple-display output, USB 3.2 connectivity, and 12–24V DC input. These features make them suitable for industrial PCs, edge computing systems, and other embedded applications.

ASRock NUC Core 300 series specifications:

• Wildcat Lake SoC
  • Intel Core 3 304 (NUC BOX-304, NUC-304)
    • 5-core CPU – 1x P-cores @ 1.5/4.3 GHz (Turbo) + 4x LPE-cores @ 1.4/3.3 GHz (Turbo)
    • GPU – 1-core Intel Xe3 Graphics @ 2.3 GHz (9 TOPS)
    • NPU – 15 TOPS
  • Intel Core 5 320 (NUC BOX-320, NUC-320)
    • 6-core CPU – 2x P-cores @ 1.5/4.6 GHz (Turbo) + 4x LPE-cores @ 1.4/3.4 GHz (Turbo)
    • GPU – 2-core Intel Xe3 Graphics @ 2.5 GHz (20 TOPS)
    • NPU – 16 TOPS
• Memory – Up to 64GB single-channel DDR5-6400 via 1x 262-pin SO-DIMM; In-Band ECC on all SKUs
• Storage
  • M.2 Key M 2242 (PCIe Gen4 x1) for NVMe SSD
  • M.2 Key M 2242/2280 (PCIe Gen4 x2) for NVMe SSD
  • 1x SATA 3.0 (NUC-320, NUC-304 only)
  • RAID 0/1 across the two M.2 Key M slots (NUC BOX models)
• Display
  • 2x HDMI 2.1 up to 4Kp60
  • 2x USB-C ports with DisplayPort Alt Mode (DP 1.4a; some BOX sheets list DP 2.1/1.4a) up to 4Kp60
  • Support for up to 3 independent displays
• Audio – 1x front-panel 3.5mm headphone and microphone combo jack; Realtek ALC256 audio codec
• Networking
  • 1x 2.5GbE RJ45 port via Intel I226-V controller
  • 1x 2.5GbE RJ45 port via Realtek RTL8125BG controller
  • M.2 Key-E 2230 socket (PCIe x1 + USB 2.0 + CNVi) for wireless module (dual-band WiFi 6E and Bluetooth 5.3 module pre-installed on NUC BOX-320, BOX-304)
• USB
  • 2x USB 3.2 Gen2x2 Type-C ports with DisplayPort Alt Mode and 5V/3A power output (front)
  • 3x USB 3.2 Type-A ports
  • 2x USB 2.0 via 2.0mm-pitch header (internal; NUC-320, NUC-304)
• Serial – COM port header (RS-232/RS-422/RS-485) (NUC-320, NUC-304)
• Misc
  • Intel PTT (TPM)
  • Watchdog Timer (256 segments, 0 to 255 seconds)
  • 256-Mbit SPI flash for AMI BIOS/UEFI
  • Cooling fan for NUC BOX
• Power
  • 12V to 24V DC-in jack
  • AT/ATX power-on supported
  • 19V/90W Power Adapter included (NUC BOX-320, NUC BOX-304)
• Dimensions
  • 117.5 x 110.0 x 49 mm (NUC BOX-320, NUC BOX-304)
  • 104 x 102 x 38 mm (NUC-320, NUC-304)
• Temperature:
  • Operating – 0°C to 40°C (NUC BOX-320, NUC BOX-304); -20°C to 70°C (NUC-320, NUC-304)
  • Storage – -40°C to 85°C
• Humidity – 5% to 90% non-condensing
• Weight – 1.0 kg net weight (NUC BOX-320, NUC BOX-304)
• Mounting – VESA mount bracket included (NUC BOX-320, NUC BOX-304)

ASRock only lists Windows 10 and Windows 11 for the NUC BOX models, while the standalone NUC-320 and NUC-304 boards do not specify an operating system. Linux is not officially certified, but Wildcat Lake already has upstream support in newer Linux kernels, along with Mesa and Intel’s NPU driver, so a recent Linux distro should support the platform, including its dual 2.5GbE ports and NPU. We’ve also tested Ubuntu 26.04 on the Beelink EQi Wildcat Lake mini PC.

We’ve previously written about ASRock Industrial’s higher-end NUC(S) Ultra 300 BOX Series based on Intel Core Ultra Series 3 “Panther Lake” SoCs, as well as the earlier NUC Ultra 200 (Arrow Lake-H) and NUC Ultra 100 (Meteor Lake-H) platforms. Intel has since launched the more affordable Core Series 3 “Wildcat Lake” family, which we’ve already seen in the Advantech MIO-5356 3.5-inch SBC, the Avalue EPC-WCL fanless Edge AI mini PC, and the consumer Beelink EQi Core 3 304 mini PC.

ASRock Industrial is now bringing the same silicon to its compact NUC motherboards and fanned NUC BOX systems. Compared to the NUC Ultra series, the NUC Core 300 series uses Intel’s more affordable Core Series 3 “Wildcat Lake” processors and targets compact industrial and edge systems.

ASRock Industrial has not announced pricing yet. The BOX models are fan-cooled barebones that include a 19V/90W power adapter and VESA bracket, but RAM and storage are not included. More details are available on the product pages for fanned BOX PCs and NUC single board computers.

The post ASRock Industrial NUC 300 series – Intel Core 3 304/Core 5 320 mini PCs and motherboards appeared first on CNX Software - Embedded Systems News.

LakeShark is open-source firmware by Samuel Reynolds (SAMS0N1TE) that turns the ESP32-P4-based LilyGO T-Display P4 into a handheld SDR scanner. Paired with an RTL-SDR Blog V3 or V4 dongle, it supports P25 Phase I trunking, analog FM/AM, POCSAG pager decoding, and 1090 MHz ADS-B reception directly on the ESP32-P4, without a PC, Raspberry Pi, or smartphone. It also includes an onboard SX1262 that supports LoRa and MeshCore applications.

SDR devices like LimeSDR Micro, xSDR, UUGear’s VU GPSDR, and many others usually rely on a PC, Raspberry Pi, or smartphone for processing, but LakeShark shows that a powerful microcontroller can also handle many SDR tasks. We previously covered the LilyGO T-Display P4 in March 2026, and LakeShark makes good use of its built-in hardware, including the SX1262 for LoRa and MeshCore applications, L76K GPS receiver for location tracking and offline maps, and ESP32-C6 for Wi-Fi and Bluetooth connectivity.

LakeShark hardware requirements:

• Primary board – LilyGO T-Display P4 (specifically the SKU with the 4.1-inch 568×1232 AMOLED display and 16MB flash)
• Radio dongle – RTL-SDR Blog V3 or V4 connected via the board’s USB-A port. (HackRF USB IQ at 2 MSPS is currently experimental).
• Optional Keyboard – T-Display P4 Keyboard (MIX-RF), which adds CC1101, nRF24, and NFC (ST25R3916) capabilities.

The interface can be controlled using the touchscreen or optional hardware keyboard, with automatic screen rotation and a simple ASCII-style layout.

Key applications include:

• P25 and FM: Supports P25 Phase I conventional and trunked voice with AUTO/manual demodulation, analog FM/AM, band scanning, and POCSAG pager decoding.
• ADS-B map MAP: Tracks aircraft at 1090 MHz and displays them on offline PMTiles maps, with flight history and spoken callouts.
• FALLS: Provides a real-time waterfall display for P25, FM, and LoRa, with tap-to-tune support.
• MESH and LORA LABS: Uses the onboard SX1262 for MeshCore messaging, packet plots, direct radio controls, and RSSI history.
• MIX-RF and NFC: With the optional keyboard, adds CC1101 energy monitoring, nRF24 scanning, and Classic 4K NFC tag reading.
• SUB-GHZ and REC: Supports sub-GHz scanning, learning, capture, and experimental OOK/FSK replay, with .sub file support and recording.
• COMPASS: Includes a tilt-compensated true-north compass and a FIND function for locating the direction of a transmitter.
• NOTES: A Markdown-based field journal that saves notes to the SD card along with live GPS, heading, and radio data.
• RADIOS and DIAG: Provides radio controls, Wi-Fi/Bluetooth settings, and system and signal information.
• CELL WATCH: Provides RF baseline measurements for comparing cellular signals; it is not an IMSI-catcher detector.

On top of that, the latest v2.5.0 release, launched on September 27, 2026, adds several mapping improvements. The offline MAP app can now show aircraft trails, MeshCore nodes, and user marks, along with FOLLOW, GO TO, range rings, and eight map palettes. It also adds a Night theme, spoken callouts for aircraft and mesh messages, and RAM optimizations that allow MeshCore to keep receiving while ADS-B is running.

LakeShark also supports an optional Flipper Zero companion app, which turns the Flipper into a remote control for the ESP32-P4 receiver. It connects to the P4 over Bluetooth or UART to control P25, FM, ADS-B, POCSAG, and recording while the P4 handles the actual radio processing.

The developer notes a few experimental features to keep in mind. P25 Phase II, HackRF support, and some Sub-GHz replay functions are still under development and have not been fully tested. The developer also recommends powering the P4 through its USB-C port, as feeding 5V back through the keyboard can cause reboot loops. A battery setup to power both the P4 and RTL-SDR is still in development. Users should also check their local laws before listening to RF signals.

LakeShark is open-source and released under the GPL-3.0 license. The easiest way to install it is through Chrome or Edge using the TerminalBay web flasher; you can also compile the firmware with ESP-IDF 5.4.3. TerminalBay also provides detailed installation guides, hardware-specific setup instructions, troubleshooting information, a Wiki, Flipper Zero setup guides, firmware downloads, and other tools for configuring LakeShark. The project’s GitHub repository contains the source code, releases, build instructions, experimental features, and technical documentation.

This is essentially a DIY project, so you will need to purchase the hardware separately. The LilyGO T-Display P4 with an AMOLED display costs around $145, while the optional MIX-RF keyboard costs about $74. A genuine RTL-SDR Blog V3 or V4 dongle adds another $30–$40 to the setup, bringing the basic hardware cost to around $245–$255, including the optional keyboard.

The post LakeShark firmware turns LilyGO T-Display P4 into a handheld ESP32-P4 SDR scanner appeared first on CNX Software - Embedded Systems News.

Raspberry Pi Smart Display Module is a CM5 carrier board compatible with the Intel Smart Display Module (SDM) specification, more specifically the SDM-L form factor (as opposed to the smaller SDM-S option).

The Intel SDM standard has been around for years, and we first wrote it in 2018, covering the Axiomtek SDM300s (SDM-S) module. These cards are designed to be plugged into compatible digital signage displays, kiosks, point-of-sale systems, and so on. They feature a standard 98-pin PCIe edge connector with USB 3.0, HDMI 1.4/2.0, I2C/UART/SPI signalling, as well as Ethernet, HDMI, USB ports, and an M.2 expansion socket.

Raspberry Pi Smart Display Module specifications:

• Compatible with Raspberry Pi Compute Module 5
• Video Output – Full-size HDMI 2.0 port
• Camera – 22-way MIPI DSI/CSI-2 connector (internal)
• Networking
  • Gigabit Ethernet RJ45 port
  • WiFi 5 and Bluetooth 5.2 (on CM5) with U.FL antenna
• USB
  • USB 3.0 Type-A port
  • USB 2.0 Type-C port for programming Compute Module 5 only
• Standard Intel SDM edge connector (96-pin PCIe x8)
  • HDMI 2.0
  • USB 3.0
  • 2× I2C
  • General SPI
  • UART
• Expansion – M.2 M-key PCIe socket compatible with the 2230, 2242, 2260, and 2280 modules (typically storage or AI accelerator)
• Misc
  • Boot mode switch
  • External power button
  • LED indicator
  • RTC battery socket (CR2032)
  • Mounting for Raspberry Pi Antenna Kit

• Power Supply – +12V & +3.3V via SDM connector (TBC)
• Dimensions – 193 x 115 x 25 mm
• Weight – 225 grams fully assembled; 146 grams as a standalone module
• Life cycle – Available until at least January 2036

The Raspberry Pi Smart Display Module ships with a passive aluminium heatsink, four M3 screws to secure the heatsink, two M2 screws to secure the Compute Module 5, and a drive attachment screw to secure an M.2 card.

There’s no specific software for the CM5 SDM carrier board, so most will probably install Raspberry Pi OS with their preferred digital signage CMS solution. While Raspberry Pi doesn’t list any recommended displays, the module was developed in collaboration with Sharp Display Solutions Europe, so I’d expect their product to have been tested with it. Some products include Sharp MESSAGE-line large-format displays, M-series 2nd generation, and P-series 6th generation.

The Raspberry Pi Smart Display Module sells for $30 plus taxes and shipping from approved distributors. A few more details may be found on the product page.

The post $30 Raspberry Pi Smart Display Module for CM5 targets digital signage displays and kiosks appeared first on CNX Software - Embedded Systems News.

Tokyo-based Enactic OpenArm 2.0 is an open-source hardware 7-DOF robot arm designed for physical AI research, teleoperation, and contact-rich data collection. It uses quasi-direct-drive (QDD) backdrivable joints to support compliant motion and bilateral force-feedback control.

The arm supports 4.1 kg nominal and 6 kg peak payload, features 1 kHz CAN-FD control, and a compact parallel gripper with an in-hand camera and replaceable fingers. It supports bilateral force-feedback teleoperation, gravity compensation, and compliant interaction, with an aluminum and stainless-steel structure and MISUMI-frame base. The ecosystem also includes the OpenArm Cell for standardized evaluation and OpenArm KER, a motorless replica for teleoperation and data collection. Open-source CAD, BOM, firmware, control software, ROS 2, MuJoCo, Isaac Lab, and dataset tools support applications such as imitation learning, reinforcement learning, sim-to-real, and robotic manipulation.

OpenArm 2.0 specifications:

• Communication
  • CAN-FD (recommended): nominal 1 Mbps, data 5 Mbps, control loop 1 kHz
  • Classic CAN 2.0 fallback @ 1 Mbps
  • Motor ID/parameter flashing via Damiao debug tool over UART @ 921600 bps
  • Host path: Linux SocketCAN (Ubuntu 22.04 / 24.04); no onboard computer
• Degrees of Freedom (DOF) – 7-axis arm + 1 parallel pinch gripper (8 actuators). Bimanual system = 14 arm DOF + 2 grippers
• Motors / Actuators – 8× Damiao motors per arm (QDD / low-ratio planetary; Damiao protocol)
  • 2× DAMIAO DM-J8009P-2EC — J1, J2 (shoulder / high load); Rated 20 N·m, peak 40 N·m, 9:1 torque
  • 1× DAMIAO DM-J4340P-2EC — J3 (upper arm / high load); Rated 9 N·m, peak 27 N·m torque
  • 1× DAMIAO DM-J4340-2EC — J4 (elbow / mid load; 4340 series is not QDD);  Rated 9 N·m, peak 27 N·m torque
  • 4× DAMIAO DM-J4310-2EC V1.1 — J5, J6, J7 + gripper (wrist / hand); Rated 3 N·m, peak 7 N·m torque
• Payload (includes the end-effector)
  • Nominal 4.1 kg — held for 1 minute in worst posture (arm fully extended)
  • Peak 6.0 kg — 3 s move into worst posture + 1 s hold
  • If a 1.5 kg tool is fitted: remaining payload ≈ 2.6 kg nominal / 4.5 kg peak
• Reach – 606 mm (OpenArm 2.0 partner / CAD-cited figure).
• Gripper
  • Compact 2-finger pinch / parallel gripper
  • Max jaw opening 88 mm; fully closed = jaws touching
  • Rotor travel closed → open 60°; motor zero = fully closed
  • Ball-bearing sliders + linkage bearings (meant for force feedback / bilateral teleop)
  • In-hand RGB camera on the 2.0 gripper
  •  Supports Intel RealSense D435 for chest/overhead views and D405 for wrist-mounted close-up views
  • Custom tools mount by replacing interface part J8_B
• Encoders – 14-bit single-turn magnetic, dual encoder per motor (output + rotor)
• Control modes – Damiao MIT mode (kp, kd, q, dq, tau), plus Damiao position and speed modes
• Safety
  • Mechanical limits on every axis
  • QDD/low-ratio joints are backdrivable (arm yields on contact)
  • Damiao driver protections: over-temperature, over/under-voltage, over-current
  • Bus-off/comms-loss handling on SocketCAN (interface can be left down after bus-off)
  • Cell add-on: light curtains/area sensors, e-stop, calibration jig
• Power
  • Motors – 24 V DC (8009P also supports 24–48 V)
  • Currents – 8009P rated 20 A/peak 50 A each; 4340/4310 rated 2.5 A
  • OpenArm Cell reference draw ≈ 480W + PC
• Dimensions
  • Human-scale, proportioned for a person ~160–165 cm tall
  • Base plate 190 × 250 mm, 8 mm thick, 30 mm hole pitch, 48 × M6 taps
• Weight – ~5.5 kg per arm
• Material
  • Aluminum + stainless steel on structural / high-load parts
  • 3D-printed casings

The OpenArm project also includes the OpenArm Cell, a standardized evaluation enclosure built using MISUMI aluminum profiles. It weighs around 100 kg and includes a motorized Z-axis lift that allows the height of the work area to be adjusted. A dedicated calibration jig also sets the robot to a fixed zero position, helping keep the arm’s starting position consistent between setups. The Cell also includes safety features such as light curtains that stop the robot when someone enters the operating area.

The third part of the project is the OpenArm KER (Kinematic Equivalent Replica), a motorless leader arm designed for teleoperation. It uses the same joint layout and link dimensions as the OpenArm, providing a 1:1 motion mapping to the robot arm. The KER uses 16 magnetic encoders and bearing joints instead of motors, so the operator can move it with little resistance during long data-collection sessions. While KER is part of the OpenArm 2.0 lineup, its full release timeline is still to be announced.

OpenArm is controlled by software running on a Linux PC (Ubuntu 22.04/24.04) and communicating using SocketCAN. The open-source software stack includes C++ and Python CAN-FD libraries, ROS 2, MoveIt2 configurations for bimanual operation, and Dora nodes for data collection and teleoperation. It also supports MuJoCo and Isaac Lab, as well as Hugging Face’s LeRobot and the mink library for kinematic solving. The platform includes OpenArm Online, where researchers can submit trained policies as Docker containers and evaluate them on simulated tasks through a web browser. The hardware CAD files in STEP, STL, and Fusion 360 formats are licensed under CERN-OHL-S-2.0, while most software repositories use the Apache 2.0 license. The hardware files, software, CAD designs, and other project resources are available on GitHub and their docs page.

We have previously written about other open-source robotic arms, including the Waveshare RoArm-M2, a 4-DOF ESP32-based arm, the Yahboom DOFBOT, a 6-DOF Jetson Nano-based vision arm, and Seeed Studio reBot Arm B601-DM, a 6-axis robotic arm (plus a parallel gripper) designed to lower the barrier to entry for embodied AI learning. The OpenArm 2.0 is a high-end design with QDD motors, an aluminum and stainless-steel structure, 4.1 kg nominal and 6.0 kg peak payloads.

Enactic works with certified manufacturers rather than selling OpenArm directly. WowRobo is one of the certified manufacturers, with the OpenArm 2.0 starting at $6,500, the OpenArm Cell at $6,200, and the OpenArm KER at $2,599. WowRobo lists a 20–40 day lead time and worldwide shipping. RT Corporation in Japan also supplies assembled and tested OpenArm units, with pricing available by quotation. More information, including full technical specifications, tutorials, and CAD files, is available on the OpenArm website.

The post OpenArm 2.0 – An open-source 7-DOF robot arm with QDD joints, bilateral force feedback, in-hand camera  appeared first on CNX Software - Embedded Systems News.

Waveshare ESP32-C5-Touch-LCD-3.5 is another ESP32-C5 devkit with a built-in display, but equipped with a larger 3.5-inch color touch LCD and an optional BF3901 low-resolution (0.08 MP) camera and/or a 500 mAh battery.

It also features a speaker connector and a built-in microphone for voice interaction, a 6-axis IMU  and a temperature and humidity sensor, a microSD card slot for additional storage, and a 32-pin GPIO connector for expansion.

ESP32-C5-Touch-LCD-3.5 specifications:

• Wireless module – Espressif Systems ESP32-C5-WROOM-1U-N32R8
  • SoC – ESP32-C5HR8
    • CPU
      • Single-core 32-bit RISC-V processor @ up to 240 MHz
      • Low-power RISC-V core @ 40 MHz acting as the main processor for power-sensitive applications
    • Memory – 384 KB SRAM on-chip, 8MB PSRAM
    • Storage – 320 KB ROM
    • Wireless Connectivity
      • Dual-band (2.4GHz/5 GHz) 802.11ax WiFi 6, with 802.11b/g/n WiFi 4 standard fallback
      • Bluetooth 5.0 Low Energy (LE)
      • 802.15.4 radio for Zigbee 3.0 and Thread 1.3
  • Storage – 32MB flash
  • Antenna – IPEX antenna connector
• Storage – MicroSD card slot
• Display
  • 2.8-inch 262K color IPS display with 480 x 320, capactivity touchscreen function
  • ST7796 SPI display controller
  • FT6336 I2C capacitive touch controller
• Camera – Optional BF3901 camera with 320 x 240 resolution
• Audio
  • Onboard MEMS microphone
  • 2-pin speaker connector
  • ES8311 low-power audio codec
  • NS4150B audio power amplifier
• USB – USB Type-C port for power, programming, and data
• Sensors
  • SHTC3 temperature and humidity sensor
  • QMI8658 6-axis IMU (3-axis gyroscope and 3-axis accelerometer)
• Expansion – 32-pin (2.54mm pitch) header with GPIO, UART, I2C, etc…
• Misc
  • Power, Reset, and Boot buttons
  • Power and Charging LEDs
  • PCF85063 RTC chip
• Power Supply
  • 5V via USB Type-C port
  • 2-pin header for 3.7V Lithium battery; supported via AXP2101 power management chip
• Dimensions – 95.11 x 63.67 x 14.10 mm

The specifications are quite similar to the ESP32-C5-Touch-LCD-2.8, except the display is larger and of higher resolution, and it features an IPEX antenna (instead of a PCB antenna), an optional camera, a 32-pin GPIO header, and a more advanced AXP2101 PMU.

Waveshare provides factory firmware, apparently based on ESPBrookesia, basic instructions to get started with the ESP-IDF framework or the Arduino IDE, and an AI tutorial section with XiaoZhi AI voice assistant (and soon OpenClaw and ESP-Claw?) on the wiki.

It’s the largest display we’ve seen on an ESP32-C5 devkit so far. Other kits include the LILYGO T-DONGLE-C5 with a tiny 0.96-inch no-touch OLED, the ESP32-C5-LCD-1.47 devkit with a 1.47-inch non-touch LCD, and LILYGO T-Display C5 with a 1.9-inch IPS LCD and optional capacitive touch. All these are SPI displays, and rendering may reach limits (become sluggish) if the UI is too complex on the 3.5-inch model.

For higher resolution and/or smoother GUI,  and dual-band WiFi 6, you’d need an ESP32-P4 + ESP32-C5 solution like GUITION JC-ESP32P4-M3-DEV or Wireless Tag WT99P4C5-S1 and a compatible MIPI display. If you don’t care about dual-band WiFi 6, you’d find a wider range of ESP32-P4 display solutions.

Waveshare sells the ESP32-C5-Touch-LCD-3.5 development kit with or without camera/battery for $37.99 – $43.99 on AliExpress, $43.99 – $47.99 on Amazon, or $31.99 – $35.99 on its own online store.

The post ESP32-C5 devkit features 3.5-inch touch LCD, optional camera and battery appeared first on CNX Software - Embedded Systems News.

NXP has just launched the FRDM-IMXRT1186, a development board based on the i.MX RT1186 crossover MCU for industrial automation, robotics, PLCs, and edge computing applications. The MCU uses a heterogeneous architecture with an 800 MHz Arm Cortex-M7 core for application processing and a 300 MHz Arm Cortex-M33 core for real-time control and communications.

The board includes 64 Mbit HyperRAM, 512 Mbit HyperFlash, and 128 Mbit QSPI Flash, along with two Gigabit Ethernet ports for TSN switching and two Fast Ethernet ports for EtherCAT or TSN switching. It also features a USB Type-C OTG port and a CAN transceiver, while debugging is supported through an on-board MCU-Link debugger and JTAG connector. For expansion, the board includes Arduino UNO R3 and FRDM headers, a FlexSPI follower header, an SRAMC header, and motor-control headers.

NXP FRDM-IMXRT1186 development board specifications:

• SoC – NXP MIMXRT1186CVJ8C (i.MX RT1180 family, LFBGA196)
  • CPU
    • Arm Cortex-M7 application core @ up to 800 MHz with FPU, 32 KB I-cache, 32 KB D-cache, and up to 512 KB TCM with ECC
    • Arm Cortex-M33 safety/communications core @ up to 300 MHz with TrustZone-M, 16 KB I/D caches, and 256 KB TCM with ECC (TÜV-certified self-test library)
  • On-chip memory – up to 1.5 MB ECC-protected SRAM (512 KB M7 TCM + 256 KB M33 TCM + OCRAM)
• Memory – 64 Mbit HyperRAM memory.
• Storage
  • 512 Mbit HyperFlash
  • 128 Mbit QSPI Flash
  • FlexSPI follower header for extra HyperFlash/HyperRAM/QSPI devices
• Networking
  • 2x YT8531SH-CA PHYs for RGMII Gigabit Ethernet (TSN switch)
  • 2x RTL8201FI PHYs for MII 10/100Mbps Ethernet for EtherCAT (or TSN, jumper-routed)
• USB
  • 1x i.MX RT1186 USB 2.0 Type-C OTG port
  • 1x MCU-link USB Type-C port
  • Either Type-C port can supply the board’s 5 V rail
• Industrial I/O – NXP TJA1057BT high-speed CAN transceiver
• Expansion
  • Arduino UNO R3 + FRDM headers (J1, J2, J51, J52) — UART, SPI, I2C, ADC, PWM, GPIO
  • Dual motor-control headers compatible with FRDM-MC-LVPMSM and similar shields
  • 50-pin SRAMC header for asynchronous SRAM / FPGA
• Debug
  • Integrated MCU-link debugger
  • MCU-link SWD interface
  • JTAG/SWD header
• Misc
  • Power slider switch
  • 5x user buttons (3x user; 2x reset according to block diagram)
  • RGB LED
  • DIP switch for boot selection
• Power
  • 5V via DC barrel jack
  • 5 V from either USB-C port
  • On-board current-measurement points for MCU power domains
• Dimensions – TBD

In terms of software support, the FRDM-IMXRT1186 is supported by the MCUXpresso SDK, with examples for FreeRTOS, lwIP, EtherCAT, multicore operation, USB, security, and peripheral drivers. Development is supported through MCUXpresso for VS Code, MCUXpresso IDE, IAR Embedded Workbench, Keil MDK, and Arm GCC with CMake.

The board also supports Zephyr RTOS, with separate CM33 and CM7 targets and drivers for Ethernet, FlexCAN-FD, LPSPI, LPUART, ADC, PWM, and GPIO. Industrial networking support includes EtherCAT, PROFINET, EtherNet/IP, Modbus TCP, TSN, and OPC UA, while additional software covers TrustZone-M, EdgeLock Enclave, MCUBoot, OTA updates, eIQ, FreeMASTER, and secure provisioning.

We first wrote about the i.MX RT1180 in 2022 when NXP introduced Gigabit TSN and EtherCAT support to its crossover MCU lineup. We have covered several other FRDM boards, including the Cortex-M33-based FRDM-MCXA series and the Linux-based FRDM-IMX95-PRO, but the FRDM-IMXRT1186 is the first FRDM board based on the i.MX RT Series.

The NXP FRDM-IMXRT1186 development board is currently listed as active and is priced at $85.00 USD. More information, including design files, software downloads, and user manuals, is available on the product page; the press release also includes additional details.

The post NXP FRDM-IMXRT1186 development board features i.MX RT1186 crossover MCU, four Ethernet ports appeared first on CNX Software - Embedded Systems News.

Walnut Pi CM2 is a low-cost Raspberry Pi CM5-compatible Compute Module powered by an Allwinner T527 octa-core Cortex-A55 processor paired with up to 4GB LPDDR4 RAM, 16GB eMMC flash, and an optional wireless module.

It’s basically the company’s Walnut Pi 2B credit card-sized SBC in Pi Compute Module form factor, exposing I/Os via two 100-pin high-density connectors. Besides the price, benefits include a WiFi 6 and Bluetooth 5.0 module, a Gigabit Ethernet PHY on the module, and a built-in 2 TOPS NPU in the T527 SoC. Downsides are that the Allwinner T527 is not quite as powerful as the Broadcom BCM2712 (4x Cortex-A76 cores) found in the CM5, and as with other such “Raspberry Pi CM5-compatible” modules, it’s only partially electrically compatible.

Walnut Pi CM2 specifications:

• SoC – Allwinner T527
  • CPU
    • Octa-core Arm Cortex-A55 processor with four cores @ 1.80 GHz and four cores @ 1.42GHz
    • E906 RISC-V core up to 200 MHz
  • DSP – 600MHz HIFI4 Audio DSP
  • GPU – Arm Mali-G57 MC1 GPU with support for OpenGL ES 3.2, OpenCL 2.2, Vulkan 1.3
  • AI accelerator –  2 TOPS NPU
• System Memory – 1 GB, 2 GB, or 4 GB LPDDR4 RAM
• Storage
  • Optional 16 GB eMMC flash; note: options for 32GB, 64GB, and 128GB capacity based on resistor selection on silkscreen.
  • 8MB SPI flash
• Networking
  • On-module Gigabit Ethernet PHY
  • Dual-band 802.11ax WiFi 6 and Bluetooth 5.0 via AW869 module and ceramic antenna
• 2x 100-pin B2B connectors
  • Video Output
    • HDMI up to 4Kp60
    • 4-lane MIPI DSI up to 1080p60
  • Camera Interface – 4-lane MIPI CSI
  • Audio
    • HPOUT (speaker output)
    • 2x microphone interfaces
  • Networking – Gigabit Ethernet
  • USB – 1x USB 3.0, 3x USB 2.0
  • PCIe – PCIe Gen2 x1 (multiplexed with USB 3.0)
  • Other I/Os – GPIOs, infrared Rx, user key, BOOT…
  • Debugging – UART debug
• Power Supply – 5V/2A
• Dimensions – 55 x 40 mm
• Weight – 9.6 grams

You obviously won’t be able to install Raspberry Pi OS on this module, and the company provides support for the Debian 12-based Walnut Pi OS, Ubuntu, Android, and Home Assistant. You’ll find the OS image and instructions to get started on the wiki shared for the Allwinner T527 SBC and Compute Module.

As noted in the introduction, electrical compatibility will be limited, as only one HDMI 2.0 is supported (not two), MIPI interfaces won’t usually work with Raspberry Pi cameras and displays, and PCIe 2.1 is muxed with USB 3.0. So you have to carefully check out the pinout diagram below to evaluate whether it’s a potential fit for your carrier board.

The best way to fully evaluate the Allwinner T527 Compute Module is to connect it to the Walnut Pi CM2 IO board with the following key features:

• Compatible SoM – Walnut Pi CM2 described above
• Storage – MicroSD card slot
• Display I/F – HDMI, MIPI DSI connectors
• Camera I/F – MIPI CSI connector
• Audio
  • 3.5mm audio jack
  • 2x microphones (TBC, the photo below only shows one)
• Networking – Gigabit Ethernet RJ45 port
• USB – USB 3.0 Type-A port, USB 2.0 Type-A port
• Expansion
  • Color-coded 40-pin GPIO header
  • M.2 Key-M (PCIe Gen2.1 x1) socket for 2230, 2242, 2260, or 2280 NVMe SSD or other compatible module
• Debugging – 3-pin Debug UART header
• Misc
  • Power key, Boot button, user key
  • Power LED, user LED
  • Coin cell battery holder for RTC
  • 4-pin PWM fan connector
• Power Supply – 5V via USB-C port
• Dimensions – TBD

We have seen a few other Allwinner T527 system-on-modules from MYiR Tech and Forlinx, but the Walnut Pi CM2 is the first to follow the Raspberry Pi CM5 form factor. Otherwise, you’ll find plenty of Rockchip-based CM4/CM5 alternatives like Radxa CM5, Geniatech XPI-3576-CM5, or Banana Pi BPI-CM5 Pro, among others.

Walnut Pi sells the Compute Module 2 (CM2) on AliExpress for $28 to $49 depending on configuration. For reference, the 2GB/16GB/wireless variant sells for $49, which compares to $97.50 for the equivalent Raspberry Pi CM5 module (CM5102016). The aforelinked AliExpress page also features bundles for  $37.80 to $58.80 for the CM2 + IO board bundle, and you can get a third bundle adding $7.70 for an enclosure.

The post Walnut Pi CM2 – A cheaper Raspberry Pi CM5 alternative based on Allwinner T527 octa-core SoC appeared first on CNX Software - Embedded Systems News.

Critical Link MitySOM-AM62x/A/P is a family of TI Sitara AM62-based industrial SO-DIMM SoMs designed for embedded Linux systems, industrial automation, HMI, and other embedded applications. The family includes the general-purpose MitySOM-AM62, AI-focused MitySOM-AM62A, and multimedia-oriented MitySOM-AM62P system-on-modules.

The standard AM62 modules feature up to a quad-core Arm Cortex-A53 processor at 1.4 GHz, along with a 400 MHz Cortex-M4F MCU and dual programmable real-time units (PRUs). The AM62A adds a C7x DSP with Matrix Multiply Accelerator supporting up to 2 TOPS and an image signal processor (ISP), while the AM62P includes a 3D GPU rated above 50 GFLOPS and a four-lane MIPI DSI output.

MitySOM-AM62x / AM62Ax / AM62Px specifications:

• SoM options
  • MitySOM-AM62x – Texas Instruments Sitara AM62x
    • Application CPU – Up to Quad-core Arm Cortex-A53 up to 1.4 GHz
    • Real-time cores
      • Arm Cortex-M4F up to 400 MHz
      • Dual-core Programmable Real-Time Unit (PRU-SS) up to 333 MHz
    • GPU – 3D Graphics Processing Unit (>8 GFLOPS) supporting OpenGL ES and Vulkan (available on AM625x models)
    • NPU – None
  • MitySOM-AM62Ax – Texas Instruments Sitara AM62Ax
    • Application CPU – Quad-core Arm Cortex-A53 up to 1.4 GHz
    • Real-time cores  – 1x Arm Cortex-R5F up to 800 MHz
    • GPU – None
    • NPU / Accelerators – C7x DSP with Matrix Multiplication supporting up to 2 TOPS; includes hardware video encode/decode and ISP acceleration
  • MitySOM-AM62Px – Texas Instruments AM62P (AM62P54CVMHIAMHR)
    • Application CPU – Quad-core Arm Cortex-A53 up to 1.4 GHz
    • Real-time cores
      • 1x Arm Cortex-R5F up to 800MHz (integrated with MCU channel)
      • 1x Arm Cortex-R5F up to 800MHz (for device management)
    • GPU – Imagination IMG BXS-4-64 GPU up to 50 GFLOPS; supports OpenGL ES 3.2, Vulkan 1.2, OpenCL 1.2 EP
    • NPU – None (AI inference handled by A53 cores or GPU)
• System memory
  • MitySOM-AM62x – Up to 4 GB DDR4 x16 @ 800 MHz with 3.2 GB/sec burst bandwidth
  • MitySOM-AM62Ax – Up to 16 GB LPDDR4 x32 @ 1866 MHz with 14.9 GB/sec burst bandwidth
  • MitySOM-AM62Px – Up to 8 GB LPDDR4 x32 @ 1600 MHz with 13.2 GB/sec burst bandwidth
• Storage
  • MitySOM-AM62x/AM62A – Up to 128 GB eMMC flash (8-bit, HS200 speed); up to 256 MB Octal or Quad SPI NOR flash
  • MitySOM-AM62P –  Up to 256 GB eMMC flash (8-bit, up to HS400 speed); up to 256 MB Octal or Quad SPI NOR flash
  • EEPROM – 2 Kbit (256 x 8-bit) I2C configuration EEPROM for module identification
• 260-pin SO-DIMM edge connector with
  • Memory – 16-bit GPMC (Connects external memory, FPGAs, and memory-mapped peripherals)
  • Storage – 2x 4-bit MMC/SD/SDIO (MMC1 & MMC2)
  • Display interfaces
    • MitySOM-AM62x – Dual display support up to 1080p each
      • 1x 24-bit RGB output
      • 2x 4-lane OLDI/LVDS
    • MitySOM-AM62A – Single display support up to 2048×1080 @ 60 fps via 1x 24-bit RGB output
    • MitySOM-AM62P – Multi-display support up to 3840×1080
      • 1x 24-bit RGB output
      • 1x 4-lane OLDI/LVDS
      • 1x 4-lane MIPI DSI
  • Camera – 1x 4-lane MIPI CSI-2 / CSI 1.3 camera input with D-PHY (up to 2.5 Gbps per lane)
  • Audio – 3x Multi-channel Audio Serial Ports (McASP)
  • Networking – 2x Gigabit Ethernet MACs (10/100/1000 Mbps) with RGMII/RMII and MDIO support.
  • USB – 2x USB 2.0 High-Speed Dual-Role Device (DRD/OTG) ports
  • Low-speed I/Os
    • 3x CAN interfaces supporting CAN-FD
    • Up to 7x UARTs, 3x SPI
    • 3x I2C interfaces (MitySOM-AM62x and MitySOM-AM62A) or 4x I2C interfaces (MitySOM-AM62P)
    • 4x general-purpose Timers
    • 3x Enhanced Capture (eCAP) modules
    • 3x Enhanced PWM (eHRPWM) modules
    • 3x 32-bit Enhanced Quadrature Pulse Encoder (eQEP) modules
    • Multiple multiplexed GPIOs with interrupt support
  • Debugging – JTAG interface
• Security
  • Secure Boot support
  • Hardware cryptographic accelerators (AES, SHA, DRBG, PKA)
  • High Security Field Securable (HS-FS) support
  • Integrated Watchdog Timer and Real-Time Clock (RTC)
• Power
  • Input voltage – Single +3.3V DC input supply (3.2V to 3.5V range)
  • PMIC
    • TIs TPS65219 PMIC on AM62x and AM62A modules
    • TIs TPS6522430 PMIC on AM62P module
  • Power outputs – 1.8V I/O supply output (up to 1500 mA)
• Dimensions – 69.60 x 38.10 (260-pin DDR4 SO-DIMM edge connector form factor)
• Temperature – Commercial: 0°C to +70°C; Industrial: -40°C to +85°C; Storage: -65°C to +80°C

The modules are supported by TI’s Processor SDKs and Critical Link’s software stack, which provides Yocto-based Linux, prebuilt images, Linux kernel and U-Boot sources, toolchains, and development documentation. All three modules use Linux as the primary operating system. The AM62 also supports MCU+ SDK development for its Cortex-M4F and PRU-ICSS. The AM62A and AM62P use the MCU+ SDK for their Cortex-R5F cores. The AM62A adds an Edge AI software stack for its C7x DSP, ISP, and vision processing. The AM62P is supported by TI’s Android SDK, including its GPU, video codec, and multi-display features. Critical Link provides separate Yocto configurations and prebuilt images for each module, so development can start without building the full stack from scratch.

The company also offers the MitySOM-AM62x/A/P Development Kit, which exposes the SoM’s main interfaces, including dual Gigabit Ethernet, dual 10/100 Ethernet, dual CAN, USB 2.0, HDMI, OLDI/LCD, audio I/O, microSD, and a Wi-Fi/Bluetooth module adapter. It also includes onboard pushbuttons, boot configuration switches, and debug interfaces.

Other Texas Instruments AM62x/A/P system-on-modules and single board computers include the Variscite VAR-SOM-AM62, Toradex Verdin AM62, and MYiR MYC-YM62X, as well as the iWave Systems iW-RainboW-G55M (AM62A),  VAR-SOM-AM62P, and the DEBIX T62P-01 industrial SBC. Compared to these products, Critical Link’s MitySOM-AM62x/A/P modules are similar, but put the AM62x, AM62A, and AM62P variants in one pin-compatible 260-pin SODIMM family with a shared development kit.

The MitySOM-AM62x/A/P Development Kit is available on the Copant store for $499 to $668.75, with the price depending on the selected configuration. The kit includes the SoM, base board, power supply, USB and Ethernet cables, a USB drive with development software and documentation, and a quick-start guide. Online purchases are limited to US customers and one kit per order; for larger quantities or other purchase options, Critical Link asks customers to contact its sales team. More information about the new modules and the development kit can be found on the Critical Link website.

The post Critical Link MitySOM-AM62x/A/P – TI Sitara AM62 SO-DIMM SoMs for edge AI and industrial automation appeared first on CNX Software - Embedded Systems News.

Up2Stream started as Arylic’s DIY audio board line and has now been spun off as its own brand. The split is not a rebadge or a clone of Arylic’s finished consumer streamers and amplifiers. It is a deliberate move to put the maker community first: boards, accessories, documentation, and support aimed at people who build speakers, restore vintage gear, or design custom multiroom systems.

Many long-time users still treat “Up2Stream” and “Arylic” as the same storefront. Others assume the boards are a copy of Arylic’s boxed products. They are related in history, but they serve different jobs. Arylic’s consumer line is complete hardware you plug in. Up2Stream is the modular stack you drop into a cabinet, a plate amp, a passive pair, or a 2.1 build and wire yourself. Same streaming DNA, different product mission.

The lineup is built around a few core modules and a set of add-ons that share the same app and multiroom layer. Compact streamer boards such as Up2Stream Mini and Up2Stream Pro add Wi-Fi, Bluetooth, AirPlay, Spotify Connect, DLNA/UPnP, USB playback, and analog or digital I/O to an existing amp or active speaker. Up2Stream HD DAC is the high-res option, with an ESS Sabre DAC, dual-band Wi-Fi, aptX HD Bluetooth, AirPlay 2, Tidal Connect, and analog/digital in and out up to 24-bit/192 kHz.

Amplifier boards cover the next step: Up2Stream Amp (about 2×50 W stereo or bridged mono) plus Amp 2.1 and Plate Amp 2.1 for builds that need a dedicated subwoofer channel.

ACPWorkbench is the paid DSP/EQ tool for people who want more than a preset curve. Control is through the 4STREAM / Go Control apps, plus HTTP/UART/TCP APIs for home-automation hooks.

That mix is meant to fix the usual DIY wireless-audio headaches: stuffing streaming into a tight enclosure, matching a streamer to an existing amp, adding a wireless sub without a second stack of boxes, grouping rooms without locking into one finished brand, and tuning crossovers instead of living with a fixed EQ. Boards stay small enough for furniture and speaker cabinets; modules can be mixed in one network so a Mini in the workshop and an Amp 2.1 in the living room play in sync.

Up2Stream says more DIY lines are planned for next year. Specs are not public yet. The near-term ask is simpler: treat Up2Stream as its own brand, shop the boards at up2stream.net, and keep it in mind the next time a project needs wireless audio without a finished consumer chassis.

The post Up2Stream spins off from Arylic as a dedicated DIY audio brand (Sponsored) appeared first on CNX Software - Embedded Systems News.

LILYGO T-Dongle-C5 is an ESP32-C5 dual-band WiFi 6, BLE, and 802.15.4 USB dongle that builds upon the earlier ESP32-S3-based T-Dongle-S3 board.

The new board still features an optional 0.96-inch color display, a microSD card slot, and a USB-A male port. The main differences are support for dual-band WiFi 6, 16MB flash, 8MB PSRAM, and the addition of a Qwiic expansion connector.

T-Dongle-C5 specifications:

• Wireless SoC – Espressif Systems ESP32-C5HR8
  • CPU
    • Single-core 32-bit RISC-V processor @ up to 240 MHz
    • Low-power RISC-V core @ 40 MHz acting as the main processor for power-sensitive applications
  • Memory – 384 KB SRAM on-chip, 8MB PSRAM
  • Storage – 320 KB ROM
  • Wireless Connectivity
    • Dual-band (2.4GHz/5 GHz) 802.11ax WiFi 6, with 802.11b/g/n WiFi 4 standard fallback
    • Bluetooth 5.0 Low Energy (LE)
    • 802.15.4 radio for Zigbee 3.0 and Thread 1.3
• Storage
  • 16MB flash
  • MicroSD card socket integrated in the USB connector
    • Fitted with a “virtual TF card”, a blank microSD card-shaped PCB to protect the USB-A connector. A better name would probably be a “filler card”
    • The virtual TF card can be replaced with a standard microSD card for storage
• Display – Optional 0.96-inch 65K color IPS LCD (ST7735 SPI controller) with 160 x 80 resolution
• USB – USB Type-A male port for power and programming
• Expansion – STEMMA QT/Qwiic connector
• Misc
  • Boot button, RGB LED
  • Antennas
    • Built-in ceramic antennas
    • 2x IPEX connectors for external antennas (requires rotating two resistors on the board using a soldering iron)
• Power Supply – 5V via USB male port
• Dimensions –  58.6 x 17.9 x 8.7 mm

LILYGO has released firmware and code samples for PlatformIO and the Arduino IDE, including the factory sketch, SD card, LCD, LED, and WiFi serial demos, along with PDF schematics on GitHub. You’ll also find instructions there for switching between built-in antennas and ceramic antennas. The wiki may also have a few more details.

The company first introduced the T-Dongle-C5 about eight months ago. Since then, it’s more broadly available, and some community projects/ports have surfaced, notably ESP32 Marauder WiFi/Bluetooth penetration testing firmware, DIY biscuit wireless audit/wardriving platform, and VordC5 BLE skimmer detector firmware.

LILYGO sells the T-Dongle-C5 with or without display on Amazon for $24/$17, on AliExpress for $20.97/$19.97, and on its own store for  $13.98/$13.02. As usual, the final price will depend on taxes and shipping.

The post LILYGO T-Dongle-C5 – An ESP32-C5 USB dongle with microSD card slot, optional 0.96-inch OLED appeared first on CNX Software - Embedded Systems News.

KryonOS is a small JavaScript-powered OS that turns a supported ESP32 development board with a touchscreen into a standalone computer with its own graphical interface, application launcher, and file storage. It provides the basic software environment needed to use the board without writing and flashing a separate firmware program for every application.

Unlike traditional ESP32 firmware, KryonOS keeps applications separate from the main system and runs them through its built-in JavaScript runtime. It also includes an App Store that allows users to download and install applications and updates over Wi-Fi without reflashing the device.

KryonOS key features:

• Supported MCUs – ESP32 (WROOM-32), ESP32-S2, ESP32-S3, and ESP32-C3
• Storage – MicroSD card support via HSPI bus
• Display
  • ILI9341 display controller with XPT2046 resistive touch controller connected via VSPI
  • Experimental support for some other display controllers, e.g., ST7789V2
• JavaScript Engine – Duktape 2.x (ES5/ES5.1 compliant).
• Graphics Engine – Direct-to-glass 16-bit RGB565 rendering with hardware double-buffering (Mini-Sprites) to eliminate 3D screen flickering
• System APIs – Extensive C++ bindings to JS for hardware GPIO, ADC, PWM, touch polling, unified file system (/local/ and /sd/), and sub-millisecond delays

In simple terms, you flash the ESP32 one time. From then on, you control physical hardware (sensors, motors, screens) entirely by writing JavaScript text files, eliminating the need to touch C++ or the Arduino IDE again.

KryonOS applications are simple to write because they’re written in JavaScript and run on the Duktape 2.x engine. The runtime supports ECMAScript 5.1, along with some ES6 features, but does not support arrow functions, let/const, or ES6 classes. Each application has around 90 KB of usable RAM, so memory usage needs to be kept low. KryonOS also performs garbage collection when System.delay(ms) is called, which helps reduce memory fragmentation during long-running applications.

Hello world JavaScript code sample:

// Clear the screen
Graphics.fillScreen(Graphics.COLOR_BLUE);

// Draw some text in the center
Graphics.setTextColor(Graphics.COLOR_WHITE);
Graphics.drawString("Hello KryonOS!", 120, 160, 2);

// Wait for 3 seconds
System.delay(3000);

// Close the app and return to the OS Launcher
System.exit();

To build an app, you create a folder containing an app.json file for metadata (like the package name, version, and OTA update URL) and a main.js file for the application logic. The OS abstracts away the complex C++ setup, letting you draw shapes, read from the SD card, or toggle hardware pins with simple JavaScript commands.

The OS uses the same display and touch pinout as the ESP32 Marauder v4, v6, and v6.1, so it can run on these devices without changing the default wiring. The project also has experimental support for several ESP32-based devices, including the Cheap Yellow Display (CYD), M5Stack Cardputer, and LilyGO T-HMI.

Previously, we have written about Espruino, another JavaScript-based platform for microcontrollers that provides a JavaScript interpreter rather than a standalone operating system. It has been used on platforms such as the ESP32 and Espruino’s own Pixl.js board, where JavaScript can be used to control GPIOs, displays, Bluetooth LE, NFC, and other hardware peripherals. It also supports applications such as games, clocks, weather stations, and projects using Arduino shields. In contrast, KryonOS combines a desktop-like interface, file explorer, and JavaScript App Store into a single operating system for ESP32 boards.

If you have a compatible ESP32 board, an ILI9341 screen, and an SD card module lying around, you can flash KryonOS today. Precompiled .bin files are available to flash directly via esptool.py, or you can build the OS from source using PlatformIO. You can find the source code, App Development Guide, and the full JavaScript API documentation on the project’s GitHub repository.

Via Hackaday

The post KryonOS turns ESP32 boards into JavaScript-powered mini computer appeared first on CNX Software - Embedded Systems News.

8devices 8Sight T100 is a compact Long-Wave Infrared (thermal imaging) camera for UAS, robotics, and industrial applications. It combines a Lynred ATI320 LWIR sensor with an STM32N6 Arm Cortex-M55 microcontroller featuring a Neural-ART NPU and a hardware ISP for edge AI inference and real-time thermal image processing.

The camera delivers 320×240 thermal video at 60 fps and operates completely offline, allowing you to deploy custom AI models for object detection, anomaly detection, or temperature-triggered events without relying on cloud resources.

8devices 8Sight T100 specifications:

• MCU – STMicro STM32N657X0H3Q with an Arm Cortex-M55 core up to 800MHz with a 600 GOPS Neural-ART Accelerator (NPU) and hardware ISP
• System Memory – 256 Mbit (32 MB) RAM @ 200 MHz (16-bit)
• Storage – 512 Mbit (64 MB) NOR flash
• Thermal Camera – Lynred ATI320 LWIR (Long-Wave Infrared) sensor
  • Spectral range – 8 to 14 µm
  • Resolution – 320 x 240 pixels
  • Frame rate – 60 FPS
  • Sensitivity (NETD) – < 50 mK
  • Latency – Image processing latency < 50 ms; AI inference time < 100 ms (model-dependent)
  • AI Inference Time – < 100 ms (model-dependent)
• Networking
  • 100 Mbps Ethernet PHY via JST connector
  • Supports MJPEG and H.264 streams over RTP, RTSP, and RTCP protocols
• USB
  • USB 2.0 high-speed interface via JST connector
  • Supports UVC-compliant (MJPEG, H.264, YUV422) and non-UVC (RAW8, RAW16) camera streams
• Misc – 4x M2 mounting holes on four sides
• Power – 5V via USB JST connector
• Dimensions – 28.5 x 30.5 x 29.5 mm (UAS-optimized housing)
• Weight – 38 grams
• Compliance – NDAA and TAA compliant

The 8Sight T100 thermnal camera comes with with proprietary firmware with a failsafe recovery bootloader. The firmware supports object detection, classification, AI-assisted upscaling, and temperature-triggered event detection, with preliminary image-processing latency below 50 ms and AI inference below 100 ms, depending on the model.

For host integration, the camera provides USB 2.0 connectivity with MJPEG, H.264, YUV422, RAW8, and RAW16 formats, while its 100 Mbps Ethernet interface supports MJPEG and H.264 streams. However, 8devices does not currently publish an SDK, API reference, sample code, model deployment workflow, host drivers, or UVC, RTSP, ONVIF, or other Ethernet streaming protocols. 8devices also offers custom firmware and software development for the T100. For UAS video transmission, the camera can be integrated with the company’s RobonodeVision system (early access in Q4 2026) and Robosoft Linux platform.

Other STM32N6 platforms that support thermal cameras/sensors include the OpenMV N6 and Ohm Lab Neuro N6, but these are development board, while the 8Sight T100 is a complete product with a higher-resolution 320 × 240 LWIR sensor capturing data at up to 60 fps.

8devices has just opened the 8Sight T100 early access program, initial engineering samples are expected to ship by the end of 2026, with the first production run currently planned for Q1 2027. More information is available on the product page and in the press release.

The post Compact Long-Wave Infrared camera pairs STM32N6 MCU with Lynred ATI320 thermal sensor appeared first on CNX Software - Embedded Systems News.

MemBrowse tool allows embedded software developers to track the RAM and flash footprint of their firmware in continuous integration, with support for GitHub Actions, GitLab CI/CD, and other CI solutions.

Engineers can monitor footprint changes in a granular way each time code is submitted to their project, instead of finding out when the firmware image stops fitting into the target’s flash, or can’t run due to excessive RAM usage. MemBrowse also provides the full history showing how memory usage progressed over time, and how much is left for extra features.

MemBrowse key features:

• Integrates with CI/CD – It works with GitHub Actions, GitLab CI, Jenkins, and other CI systems.
• Automatic memory tracking on every commit to your main branch. The open-source MemBrowse CLI parses the project’s ELF file and linker script, extracts memory usage data, and uploads a report to the MemBrowse Portal.
• PR comments show how much memory each change adds or removes
• Historical trends to visualize memory usage over time
• Target comparison to understand how different boards and build configurations differ
• Budget alerts that fail CI when memory exceeds user-defined limits

MemBrowse has several pricing options. It’s “free forever” for open-source projects hosted on public repositories with unlimited tracked targets supported. As I understand it, projects such as NuttX, TinyUSB, wolfSSL, Flipper One (RP2350 firmware), RT-Thread, and Adafruit are using it at no cost. The fee starts at $99 per month for closed-source projects with up to 10 tracked targets. The price goes up if you need to track more targets.

You can even access the live dashboard of some of the open-source projects to quickly understand what the portal has to offer. For more details on how to get started, check out the documentation instead.

The post MemBrowse continuously tracks firmware memory footprint, is free for open-source projects appeared first on CNX Software - Embedded Systems News.

Qualcomm has released an “Early Developer Preview” of Linux for the Snapdragon X2 Armv9 family of laptop SoCs. This release is meant for distribution maintainers, toolchain developers, kernel contributors, and hardware enablement engineers.

New Qualcomm processors for mobile PCs (that’s what Qualcomm calls laptops) usually target Windows only. We get comments asking about Linux support on most of these posts, but Qualcomm’s history with maintaining Linux is rather mixed. For instance, they demonstrated Debian 12 and Linux upstreaming for the Qualcomm Snapdragon X Elite SoC in collaboration with Linaro in 2024, but based on comments I’ve read, support is incomplete and not to the level you’d get on x86 targets. Maybe it will be different for the Snapdragon X2 Series.

The announcement mentions that Qualcomm engineers have been working on upstream Linux since Snapdragon Summit 2025 (in September, one year ago). So far, the work includes low-level SoC enablement and porting the Snapdragon X2 Series reference hardware.

They plan to roll out Linux support in stages as follows:

1. Initial Enablement (early bring-up)
   • Boot and system bring-up – Qualcomm Linux supports systemd-boot as the UEFI boot manager to load and boot the Linux kernel.
   • Core I/O – Support for  PCIe, Qualcomm universal peripheral (UP) serial engine (UART/I2C/SPI)
   • Graphics and display – Freedreno, Turnip, and Rusticl open-source Mesa drivers to enable a desktop UI, browser acceleration, graphics workloads, and GPU compute.
   • AI acceleration – FastRPC driver upstreaming to enable Hexagon NPU support for local inference workloads.
   • Power, thermal, and reliability
2. Expanded Enablement (suitable for application development and device evaluation)
   • Broader hardware enablement – more peripherals and subsystems brought online on Snapdragon X2 Series
   • Better reliability
   • Expanded upstream kernel support with more patches landing in mainline
3. Production Readiness (planned for the end of November)
   • More features tested on real hardware
   • Expanded support across laptops with Snapdragon X2 and hardware variants
   • Improved stability, performance, and completeness

An important point is that Qualcomm has no plan to support a specific distribution, and it will be up to the distribution maintainers to provide an ISO image for Snapdragon X2 hardware:

…this effort is less about supporting specific Linux distributions and more about empowering the developers who make Linux available on new hardware.
…
For end users looking for a turnkey “install and go” experience, that will come later as distributions adopt what lands upstream. We’re encouraging the community to build on this work and help shape what comes next.

[Update: Canonical has announced it will support Ubuntu on the Snapdragon X2 with a target release in 2027 for a “fully optimized, certified Ubuntu image”]

You can play around with Qualcomm Linux on Snapdragon X2 hardware platforms today by following the build instructions. Two build paths are available:

• The prebuilt firmware path builds only the Debian operating system layer while firmware ships as binaries.
• The full firmware path builds firmware from source and then the Debian layer. This requires a registered Qualcomm developer account.

Most people will use the first path. You can get the required code from GitHub and build the Debian 13 image using debos. There are quite a few steps, just make sure to follow the build instructions.

I could not find screenshots or video demos for the current image, but Qualcomm has been demonstrating Linux on X2 at the Snapdragon Summit in Maui, Hawaii (September 22-24).  The Snapdragon X2 Linux announcement follows a promise that Qualcomm Linux 2.0 for Dragonwing IoT platforms would be “upstream first”. We’ll see how this all works out over time.

The post Qualcomm releases Linux developer preview for the Snapdragon X2 Series appeared first on CNX Software - Embedded Systems News.

CyboPal ONE is a 6-axis desktop robotic terminal designed to move a 27-inch 4K display using voice and gesture control, spatial sensing, and automatic display positioning based on the user’s position.

It includes a camera, a binocular 3D sensor, and a motion sensor (IMU), and supports joint force sensing, obstacle avoidance, collision detection, and power-loss soft landing. The 6-DOF arm has a 747 mm reach and supports face tracking, manual adjustment, and saved positions. A built-in Pauli agent also provides voice control for screen positioning, application control, text input, and workflow tasks across Windows, macOS, and Linux.

CyboPal ONE specifications:

• SoC – Rockchip RK3588 octa-core processor with
  • CPU – 4x Cortex‑A76  cores @ up to 2.4 GHz, 4x Cortex‑A55 cores @ 1.8 GHz
  • GPU – Arm Mali-G610 MP4 GPU
  • Video decoder – 8Kp60 H.265, VP9, AVS2, 8Kp30 H.264 AVC/MVC, 4Kp60 AV1, 1080p60 MPEG-2/-1, VC-1, VP8
  • Video encoder – 8Kp30 H.265/H.264 video encoder
  • AI accelerator – 6 TOPS NPU
• System Memory – 8 GB RAM
  
• Storage – 128GB on-device storage
• Robotics and motion system
  • 6-DOF (degrees of freedom) robotic system with a 747 mm arm reach
  • Face tracking that responds in under 4 seconds
  • Zero-force manual adjustment
  • Collision sensing with a rapid emergency stop
• Display
  • 27-inch 4K (3840 × 2160) main display with a 160 Hz refresh rate
  • 3.5-inch Pauli companion display on the bottom side
• Video Inputs
  • 1x USB Type-C port for display input (also works as a 65 W PD charger)
  • 1x HDMI 2.1 port
  • Supports direct connection from computers, compatible phones, and tablets.
• Cameras
  • Equipped with a high-quality HDR camera
  • Binocular 3D sensor for spatial awareness, path planning, and obstacle avoidance
• Audio – Built-in speaker and microphones, with Privacy Mode that physically cuts power to the A/V sensors
• Sensors
  • IMU for vibration sensing
  • Joint force sensors measure applied push, pull, or twisting force on them
• USB – USB-C port for display input and 65 W PD outbound to the host, not inbound to the arm
• Misc
  • Mode button for toggling Privacy Mode
  • Hardware LED (verifies the power state of the cameras and microphone)
• Power – 240 W peak power with a backup battery for controlled landing during power loss
• Dimensions – TBD

CyboPal ONE works on Windows, macOS, and Linux through dedicated desktop clients. Its Pauli AI agent handles voice commands and can control supported applications such as Gmail, Slack, Zoom, Chrome, Gemini, ChatGPT, Claude Code, and Codex. Face tracking, gesture recognition, and 3D spatial sensing run locally on the RK3588-based Linux AI Compute Base.

You can control the system with voice commands through the Pauli AI agent, which can operate supported applications such as Slack, Gmail, Chrome, ChatGPT, and coding agents. Hand gestures can be used for active tracking, screen rotation, and scrolling.

We’ve previously written about various desktop robotic arms and AI-assisted platforms, but most are designed for education, prototyping, or light industrial tasks. For example, the M5Stack Stackchan is a compact ESP32-S3-based desktop robot, while the SO-ARM101, Seeed Studio’s Amazing Hand, and Elephant Robotics ultraArm P340 are robotic arms designed for tasks such as drawing and object manipulation. Compared to all of that, the CyboPal ONE instead combines a 27-inch display with a 6-axis robotic arm for computer use and adaptive screen positioning.

CyboPal ONE is available on Kickstarter, with the Super Early Bird package priced at HK$12,536 (about $1,599). The planned retail price is $2,499, and multi-unit packages are also available. Shipping is free in the United States, while shipping to other regions costs $50 per unit, with import duties and taxes covered by CyboPal. Delivery is expected to start in December 2026.

The post CyboPal ONE – An AI-powered, 6-DOF robotic terminal with 27-inch 4K display that adjusts to your posture (Crowdfunding) appeared first on CNX Software - Embedded Systems News.

Quectel has launched the SE200ZC-AP, a compact (40 x 40 mm) smart module based on the Rockchip RV1126B/RV1126BJ quad-core Arm Cortex-A53 SoC with a 12MP ISP, an 8MP AI-ISP, and a 3 TOPS NPU.

The module supports up to five 12MP cameras through two 4-lane MIPI CSI interfaces and one DVP interface. Other interfaces include Gigabit Ethernet, USB 3.0/2.0, two CAN FD, up to eight UART, six I2C, two SPI, two SDIO, three SAI, 22 ADC, 28 PWM, and up to 115 GPIOs. It ships with 1GB, 2GB, or 4GB LPDDR4x and 64GB eMMC flash, and is available in commercial and industrial grades, with operating temperatures of -25°C to +75°C and -35°C to +80°C, respectively.  These features make it suitable for smart security, industrial automation, autonomous vehicles, robotic vision, and Smart Home applications.

Quectel SE200ZC-AP specifications:

• SoC – Rockchip RV1126B or RV1126BJ (industrial-grade version)
  • CPU – Quad-core Arm Cortex-A53 up to 1.6 GHz (commercial) or 1.3 GHz (industrial) with 32KB L1 I-Cache and 32KB L1 D-Cache, unified 512KB L2 Cache
  • GPU – 2D Graphics Engine
  • VPU
    • Video Decoder – H.265/H.264 up to 3840×2160 @ 30 FPS
    • Video Encoder  – H.265, H.264, JPEG up to 3840×2160 @ 45 FPS
    • JPEG Decoder
  • AI accelerator – Rockchip NPU engine up to 3 TOPS (INT8); supports INT4, INT8, INT16, and FP16 models; TensorFlow, ONNX, PyTorch, and Caffe frameworks.
• System Memory – Up to 4GB DDR4 options
• Storage – 64 GB eMMC flash (default)
• LLC + LGA design with
  • Storage – 2x SDIO for SD or microSD card support
  • Display I/F
    • 4-lane MIPI DSI up to 1920×1080 @ 60fps
    • LCDC multiplexed interface
  • Camera I/F – Up to 5x cameras via
    • 2x 4-lane MIPI CSI D-PHY v1.2 (can be split into 2x 2-lane + 2x 2-lane)
    • 1x DVP (Max 12 MP @ 30fps per camera)
  • Audio – 3x SAI (Serial Audio Interface), 1x PDM (up to 8 channels), 2x DSM, 2x MIC
  • Networking
    • 1x GMAC supporting Gigabit Ethernet or Fast Ethernet PHY
    • Supports external Quectel Cat 4/1, 4G, Voice, Wi-Fi 6, and GNSS modules.
  • USB –  USB 3.0 (OTG), USB 2.0 (Host)
  • Serial – Up to 8x UART
  • Other I/O – Up to 115x GPIOs, 6x I2C, 22x ADC, 2x SPI, 3x PWM, 2x CAN FD, , Keypad interface
  • Misc – PWRKEY and RESET_N Pins
• Power Supply – 4.5V to 5.5V (typ. 5V)
• Dimensions – 40.0 x 40.0 x 2.9 mm (LCC + LGA package)
• Temperature Range – Extended Commercial Grade: -25°C to +75°C; Industrial Grade: -35°C to +80°C

On the software side, the SE200ZC-AP comes preloaded with a Debian (Linux Kernel 6.1) image currently under development. The NPU supports mixed INT4, INT8, INT16, and FP16 operations, while the multimedia stack handles 4K video encoding at 45 FPS and decoding at 30 FPS using H.265 or H.264, including multi-stream video processing. Other than this, not much information is available for the module.

This is not the first time we have covered a product based on the Rockchip RV1126B or RV1126BJ, as we have already seen it in fanless edge AI systems such as the Forlinx FCU3101, FET1126Bx-S low-power SoM, and SBCs such as the Luckfox Aura and Boardcon Tiny1126B.

At the time of writing, there is no development board for the SE200ZC-AP. The product page already includes an EVB Kit section, meaning an evaluation board is planned, although details are not publicly available yet. Based on its previous smart modules, Quectel generally ships dedicated EVBs for products in this category.

The Quectel SE200ZC-AP smart module will soon be available for customer sampling and evaluation. More details are available on Quectel’s product page and press release.

The post Quectel SE200ZC-AP smart module supports up to five cameras with Rockchip RV1126B/BJ SoC appeared first on CNX Software - Embedded Systems News.

MediaTek Dimensity CX C10 Max is a 3nm “All Big Core” octa-core Armv9 CPU with an 11-core Immortalis-G925 MC11 GPU, and MediaTek’s  890 NPU delivering up to 55 TOPS for an “upcoming premium, ultra-portable Googlebook device” with all-day battery life.

Googlebook is Google’s new premium laptop category built around the company’s Gemini Intelligence and running Googlebook OS, codenamed Aluminium OS, which merges parts of Android and ChromeOS platforms. If you visit the Googlebook website, you’ll find pre-orders for Intel Core Ultra and Qualcomm Snapdragon X Elite models, but Google will soon launch at least one model equipped with the new MediaTek SoC.

MediaTek Dimensity CX C10 Max specifications:

• Octa-core CPU
  • 1x Arm Cortex-X925 core clocked at up to 3.73 GHz with 2MB L2 cache
  • 3x Arm Cortex-X4 cores with 1MB L2 cache
  • 4x Arm Cortex-A720 cores with 512KB L2 cache
  • Shared cache – 12MB L3 cache, 10MB SLC (System Level Cache)
• GPU – Arm Immortalis-G925 MC11
• VPU
  • 4Kp60 10-bit decode (HEVC/AVC/VP9/AV1)
  • 4Kp60 10-bit encode (HEVC/AVC)
  • 4Kp60 8-bit encode (HEVC)
• AI – MediaTek NPU 890 accelerator
  • Up to 55 TOPS (dense)
  • Generative AI, Agentic AI
  • Supports INT8, INT16, FP16, BF16
• Memory – LPDDR5X up to 9600 Mbps
• Storage – UFS 4.0
• Display
  • DP/MIPI up to 4Kp60 for internal display
  • DP/MST up to dual 4K for external display
• Camera
  • Up to  32MP main camera with Imagiq 1090 ISP
  • Features – Electronic Image Stabilization (EIS), HDR video, auto zoom
• Wireless connectivity (via external chip/module?) – WiFi 7 up to 7.3 Gbps, Bluetooth 6.0
• Process – TSMC 3nm

The Dimensity CX C10 Max is supported by Googlebook OS, and no other operating system is listed for the new SoC. Googlebook OS is mostly based on Android 17 under the hood, with some Chrome OS features such as a desktop shell and windowing, up to 10 years of OS updates from Google, support for the Titan security chip and verified boot, and more. The OS also integrates “Gemini Intelligence”, which enables it to be context-aware and provide options using what’s on screen, your (Android) phone, Gmail/Calendar, etc. One example is the Magic Pointer, which offers specific actions when users hover/wiggle the pointer on text or an image.

A few more details are available on the product page and in the press release.

Thanks to TLS for the tip.

The post MediaTek Dimensity CX C10 Max octa-core Cortex-X925/X4/A720 SoC targets Googlebooks appeared first on CNX Software - Embedded Systems News.

South Korean startup Soulart has launched the Soulart Station, a markerless full-body motion tracker for VR, VTubing, and motion capture. The device uses a Time-of-Flight (ToF) sensor and an onboard Neural Processing Unit (NPU) to capture 3D point cloud data and process motion tracking on the device.

By running machine learning models locally, it avoids expensive PC-side GPU computation or cloud processing, delivering 23-joint 6DoF motion data with a claimed ML latency under 10 milliseconds.

Soulart Station specifications:

• SoC – TBD; They only mention
  • Onboard dual-core NPU and built-in graphics processor
  • Fully on-device processing (no PC GPU or cloud required; only motion data is transmitted for privacy)
• Connectivity – Wi-Fi
• Sensors
  • Custom Time-of-Flight (ToF) module (940 nm IR)
  • 4K RGB + NIR cameras
• Tracking
  • Method – Markerless 3D scan + on-device ML pose (no wearables, suits, or pucks required)
  • Point cloud – ~300,000 points (millimeter-scale) with sub-1% error
  • Scan Rate – Up to 120 Hz / 120 FPS
  • Pose accuracy – < 20 mm error per station (improves with multi-unit setups)
  • Output data – 23-joint 6DoF motion data
  • ML latency – < 10 ms (on-device model only)
  • Supports single or multiple Stations with ToF interference cancellation and up to 120 Hz capture across 50 m²
• USB – 1x USB Type-C port for power and data
• Misc
  • 1/4″ camera threads on back and bottom for mounting
  • Rear power button
• Power – 12V DC, 2A via a rear charging port
• Dimensions (W x D x H) – 75 x 75 x 70 mm
• Weight – 280g
• Mounting – 2 x 1/4″-20 UNC threaded mounting holes (1 rear, 1 bottom)
• Certification – IEC 60825-1 Class 1 Eye-Safe (underway)

On the software side, the cross-platform Soulart App manages the Stations and routes motion-tracking data to supported applications across Windows, Meta Quest, and Android XR, with a SteamVR driver for SteamVR. It supports features like Live Preview and Automatic Avatar Retargeting, as well as VMC, OSC, and BVH streaming, along with BVH/FBX export and integrations with VTube Studio, Warudo, VBridger, Unity, Unreal Engine, Blender, Maya, MotionBuilder, and ROS2.

The system also features a “Hub Mode,” where one Station can combine data from other Stations and handle additional ML processing locally, reducing the need for PC-side processing. Soulart plans to open-source the app and plugins in early 2027, but the Station firmware will remain closed source. The team also plans to roll out free OTA updates to add facial tracking in H1 2027 and hand tracking in H2 2027.

Previously, we covered full-body tracking systems like the open-source SlimeVR Butterfly Trackers, which use wearable IMU sensors but can experience drift and require sensors to be attached to the body. In contrast, the Soulart Station uses ToF-based markerless tracking with interference cancellation for multi-Station setups and an onboard NPU for motion processing.

Soulart Station is currently on Kickstarter, where it has raised over $450,000 so far, surpassing its $300,000 goal. Pledges start at $419 for a single Station, while the two-Station and eight-Station packages cost $799 and $2,850, respectively. Shipping is extra and is estimated at around $50 for the US. The first units are expected to ship in June 2027.

The post Soulart Station – A markerless full-body motion tracker for VR and VTubers (Crowdfunding) appeared first on CNX Software - Embedded Systems News.

Gearotons M17 open-source hardware servomotor integrates the four parts of a motion axis into one standard NEMA-17 body with no protrusions: the motor, the driver, the motion controller, and the encoder.

This level of integration enables it to be low-cost ($20 and up) and easier to use. The user has nothing to source and match, nothing to wire between boards, and nothing to configure. Just connect power and one data cable, and you’re good to go.

Gearotons M17 specifications:

• MCU – ST STM32G0 Arm Cortex-M0+ microcontroller
• Single device with motor, motor driver, motion control system, and encoder
• Motor control highlights
  • High-precision closed-loop control with built-in encoder and PID control loop that runs at 32 kHz
  • Higher power efficiency in closed-loop control vs. conventional stepper
  • Up to 560 RPM
  • Torque-to-weight ratio is the same as an equivalent stepper motor
  • 64-bit unique ID and a one-byte alias per motor
• Four models
  • M17‑60 – 0.65 N·m rated torque, 59.7 mm height, 470 grams, 26.4 W rated power
  • M17‑48 – 0.55 N·m rated torque, 48.7 mm height, 360 grams, 26.4 W rated power
  • M17‑40 – 0.42 N·m rated torque, 40.1 mm height, 285 grams, 26.4 W rated power
  • M17‑34 – 0.28 N·m rated torque, 33.5 mm height, 210 grams, 24.0 W rated power
• Host interface – RS-485 at 230400 baud; also used for daisy-chaining multiple motors
• Power Supply
  • 12V to 24V DC input
  • Integrated over-current, over-voltage, and over-temperature protection
• Dimensions
  • About the same size as a NEMA 17 stepper motor
  • Standardized NEMA 17 mounting dimensions
• IP Rating – IP20

The Gearotons M17 servomotors are compatible with a variety of interfaces and hardware, including Raspberry Pi, Arduino, ESP32, Macs, and PCs. The company (Move the Needle Limited) says they are suitable for robotics, CNC, automation, scientific instruments, testing jigs, 3D printers, and more.

The STM32G0 firmware handling a 32 kHz closed loop and RS-485 communication, Python and Arduino libraries, the KiCad hardware design files, and 3D files can all be found on GitHub.

The project’s documentation provides tutorials, Python and Arduino code samples, a prompt to use AI to generate code more easily for the motors, and more. An MCP server also allows AI assistants to drive the motors directly.

Gearotons M17 servomotors sell for $20 to $35 depending on the size, and a USB-to-RS-485 adapter is also offered for $10. The Electromaker video below shows Tom Rodinger, the company’s founder, demonstrate the solution at Maker Faire Shenzhen 2025.

The post Gearotons M17 – A low-cost NEMA-17 4-in-1 servomotor with driver, motion controller, and encoder appeared first on CNX Software - Embedded Systems News.

Nordic Semiconductor recently expanded its nRF54L Series with the nRF54LC10A, a new SoC with less memory and a smaller peripheral set compared to the higher-end nRF54L15, while supporting Bluetooth LE with Channel Sounding and IEEE 802.15.4 for Thread, Zigbee, and Matter.

Built on a 22nm process, the SoC combines a 128 MHz Arm Cortex-M33 processor with a multiprotocol 2.4 GHz radio. It also features a sub-50 nA hibernation mode for devices shipped with (3V) batteries installed, reducing battery drain during storage and transportation.

Nordic Semiconductor nRF54LC10A specifications

• CPU
  • Arm Cortex-M33 @ 128 MHz with DSP, FPU, MPU, and TrustZone
  • 128 MHz RISC-V coprocessor (VPR) with I/O acceleration for SoftPeripherals and time-critical offload
  • Performance – 239 CoreMark/mA, 503 CoreMark (preliminary)
• Memory – 192 KB RAM
• Storage – Up to 1012 KB NVM
• Wireless
  • Bluetooth 5.4 LE with direction-finding, Bluetooth Mesh, etc. Ready for future Bluetooth releases
  • Bluetooth Channel Sounding support for distance measurement/location tracking
  • 802.15.4 radio for Thread, Matter
  • Proprietary 2.4 GHz up to 4 Mbps
  • -97 dBm RX sensitivity @ 1 Mbps for Bluetooth LE, −101 dBm (802.15.4)
  • Up to +4 dBm TX power
  • Single-ended antenna output (on-chip balun)
• Peripherals
  • Up to 31x GPIO (P0/P1 only; no high-speed P2 port; 8 MHz ports)
  • 3x SPI/TWI/UART with EasyDMA
  • 8-channel SAADC – 10-bit at up to 1 MS/s, 14-bit with oversampling
  • Global RTC (GRTC)
  • 7x 32-bit timers, watchdog timers
  • Temperature sensor
  • Comparator and low-power comparator (wake from System OFF)
• Security
  • Arm TrustZone isolation
  • Secure boot and secure storage
  • Tamper detection / physical-attack protection
  • Cryptographic engine with side-channel leakage protection
  • Authenticated debug/debug access port protection
• Power
  • Supply and GPIO voltage – 1.7 V to 3.6 V
  • Consumption
    • Radio @ 3 V – 4.0 mA Rx, 5.8 mA Tx @ 0 dBm
    • Sleep modes @ 3 V – 0.5 μA to 1.6 μA
    • Hibernation mode – <50 nA for shipping and storage
    • 22 nm process; Nordic quotes 30–50% lower Bluetooth LE power than nRF52 Series
• Packages
  • CSP29 – 1.93 × 2.29 mm, 12 GPIOs (0.35 mm pitch)
  • CSP39 – 1.93 × 2.29 mm, 20 GPIOs (0.3 mm pitch)
  • QFN48 – 6 × 6 mm, 31 GPIOs (0.4 mm pitch)
• Temperature range – −40°C to +85°C

The new nRF54LC10A SoC is directly comparable to the nRF54L15 and nRF54L10, but it has fewer peripherals, leaving out NFC, PDM, I2S, PWM, and QDEC. It still includes the RISC-V coprocessor and 802.15.4 radio, so it can handle applications such as object trackers, Smart Home sensors, and networked sensor nodes.

The nRF54LC10A is supported in the nRF Connect SDK built on Zephyr RTOS. Each wireless standard is supported by a dedicated wireless stack:

• Bluetooth LE uses Nordic’s SoftDevice Controller with the Zephyr Bluetooth Host
• Thread and IEEE 802.15.4 use Nordic’s radio driver plus OpenThread
• Matter is available through the Matter add-on
• Zigbee is available through the separate ZBOSS R23 add-on

The SDK also supports the SoC’s RISC-V FLPR coprocessor for software-defined peripherals. Development tools include nRF Connect for VS Code, nRF Connect for Desktop, nRF Cloud, and Nordic Developer Academy.

Nordic also offers the nRF54LC10A DK development board featuring a 2.4 GHz PCB antenna, a SWF RF connector, an onboard SEGGER J-Link OB programmer/debugger with USB-C, four programmable LEDs, four buttons, and current-measurement pins. An nPM1300 PMIC handles power management, while software development is supported through the nRF Connect SDK, Zephyr RTOS, and VS Code.

Over the years, we have written about various devices built around Nordic’s expanding nRF54L ecosystem, including the Seeed Studio XIAO nRF54L15 and the Icy Electronics nRF54L15 Discovery with its snappable CMSIS-DAP debugger. Nordic also previously announced the nRF54LV10A, a specialized low-voltage variant designed for medical biosensors running on 1.5V silver-oxide batteries. The new nRF54LC10A shares a similar physical footprint but is designed for standard 3V coin cells.

The nRF54LC10A is currently available through a project-based early access program. Development kits are expected to become generally available in Q1 2027, with mass production planned for Q2 2027. More details are available on the product page and a related technical blog post.

The post Nordic nRF54LC10A – A tiny Bluetooth LE and 802.15.4 SoC with sub-50 nA hibernation mode appeared first on CNX Software - Embedded Systems News.

Realtek RTL8773J Series is the first Bluetooth 7-ready microcontroller family we’ve seen so far. It features a dual-core design with a 200 MHz Cortex-M55 application core and a Cortex-M33 sensor and network core, along with DSPs for high-fidelity codec processing and an accelerator for ANC, Talk Clear, and Keyword Spotting.

Other features include up to 2MB SRAM, 8MB PSRAM, 16MB flash on-chip, Bluetooth 6.2/7 and 2.4 GHz proprietary, audio features, and a range of I/Os. The main feature of the upcoming Bluetooth 7 release will be High Data Throughput (HDT), delivering up to 7.5 Mbps data link over Bluetooth. Realtek has developed a custom version (Realtek HDT+) with claims of up to 15 Mbps data rates.

Realtek RTL8773J specifications:

• MCU cores
  • Arm Cortex-M55 application core with MVE up to 200MHz
  • Arm Cortex-M33 sensor and network core
• DSP
  • Dual-core DSP HiFi 1s up to 400MHz
  • HW FFT engine
• Memory
  • Up to 2MB SRAM total (including core-dedicated RAM + 1024KB shared configurable SRAM)
  • 8MB embedded PSRAM (part number dependent)
• Storage
  • 4MB, 8MB, or 16MB embedded flash
  • SPIC (QSPI) interface for SPI flash
• Audio
  • Speaker out
  • Analog (Line in, L/R channel) and Digital (PDM) mic inputs
  • I2S interface
  • Codecs
    • SBC and mSBC codecs
    • Optional LC3, LC3Plus, and/or OPUS codec (depends on part number)
• Wireless
  • Bluetooth
    • Bluetooth 6.2 core specification
    • Bluetooth 7 features (pre-ratification)
    • BR/EDR/LE-1M/LE-2M/LE Coded (Long Range)
    • AoA/AoD
    • LE audio
    • HDT/Realtek-HDT+
    • Channel Sounding
    • 1.25ms Short connection interval (SCI)
    • Frame Space Update supports a negotiable frame space shorter or longer than 150us
  • 2.4GHz proprietary protocol
  • RF Performance
    • Tx Power
      • Up to 15 dBm @ BLE
      • Up to 12 dBm @ EDR 2Mbps
    • Rx sensitivity
      • -101 dBm (1M BLE)
      • -97 dBm (1M BR)
      • -97 dBm (2M EDR)
    • RS-FEC – Forward Error Correction technology
    • Fast AGC control to improve receiving dynamic range
  • OTA (Over-the-Air) firmware upgrade support
• Peripherals
  • Optional USB 2.0
  • I2C, UART, SPI, PWM
  • 13x to 30x GPIO pins depending on SKU
  • Optional SDIO
  • Up to 8-channel 12-bit ADC
• Crypto engine – AES 128/256, SHA256/384, PKE (RSA-4096/ECC P-384)
• Power Management
  • Built-in LDO for the I/O and flash memory
  • Built-in Li-ion battery charger with up to 500mA current
• Package – QFN52, 6.4 x 4.1mm or BGA108, 5.9 x 4.2mm

Four SKUS are available in the RTL8773J Series, all with Cortex-M55 and Cortex-M33 cores, BLE audio support, etc., but with some notable differences shown in the table below.

Basically, you’d use an RTL8763JKU-based dongle to add Bluetooth 7 HDT+ to your laptop, computer, or SBC, RTL8773JKE for regular OWS or TWS Bluetooth 6.2 earbuds, RTL8773JBM for speakers and gaming headsets, and RTL8773JBS for more advanced versions of these devices, plus “acoustic devices”, basically AI-enhanced hearing aids.

The new RTL8773J Series Bluetooth 7-read microcontroller family was initially unveiled at Computex 2026, but with minimal details. We now have pretty detailed specifications, but still no publicly available developer documentation, firmware, or development board.  A few more details may be found on the product page on the RealMCU website (MCU-focused website from Realtek)

Thanks to TLS for the tip.

The post Realtek RTL8773J – A Bluetooth 7-ready Cortex-M55/M33 microcontroller with HDT support appeared first on CNX Software - Embedded Systems News.