Weekly Roundup #52 – New Maker Products // News


Embedded World 2018

WR-NMP52p7This is something to get excited about. If you’re in Nuremburg, Germany next week I’d suggest checking out Embedded World 2018, where you’ll see the latest in cool embedded stuff, like the STMicro discovery kit.
Hopefully there won’t be any power outages.

Technologic Systems TS-4100

WR-NMP52p8Technologic Systems have created a new low powered SBC based on the i.MX6 UltraLite SoC with 1G RAM, 4G eMMC, SD slot, dual 100MbE and USB. All the usual stuff you see on an SBC, but this one also packs in a Lattice Semiconductor MachX02 FPGA with 62 accessible GPIOs.
They have also added in an open-source based ZPU soft core allowing you to easily offload tasks to the FPGA using a typical GCC based build chain.
The board also has a header designed to connect into this base-board giving you GbE, M.2 SATA and a whole lot more GPIOs.

Emitron RZ/N1D SBC

WR-NMP52p9Emtrion is jumping on the SBC band-wagon with a new board based on the dual-core Renesas SoC. Comes with 2G RAM, 64G eMMC, CAN transceiver, RTC and unusually 5 GbE ports.

Neutis N5

WR-NMP52p10The unusually named Neutis company have a small SBC aimed at Pro-Makers to be shipped in April. The tiny 42mm by 30mm board contains the quad-core AllWinner H5 with 512MB RAM, 8GB eMMC, HDMI, WiFi, Bluetooth, on-board crypto chip and 38 GPIOs pushed out running off a 3.3v supply.
What makes it a Pro-Maker SBC is the fact that it’s a module that plugs into a baseboard. Still pretty good for a US$49 price tag.


WR-NMP52p11STMicro will be releasing a new MCU targetting wireless communications in competition with Espressif’s ESP32.
The new chip will have 256KB RAM, 1MB flash and support for USB, audio, LCD, up to 72 GPIOs, embedded crypto-engine, Bluetooth 5 with mesh support and also an 802.15.5 radio that provides OpenThread, ZigBee and other protocols.
Looks like a promising chip.


WR-NMP52p12The SAMA5D2 from MicroChip is a System-in-Package running a 500MHz Cortex-A5 with on-board 128M RAM and other cool features allowing you to run mainline Linux in the same way as the MediaTek SiPs.
MicroChip have now released a cheap US$39 module with this SiP at the core as well as a PMIC, 64Mb SPI flash, 2Kb EEPROM, 100MbE transceiver and all the GPIOs broken out.


WR-NMP52p13After being scoffed at for years, the Raspberry Pi is finally being seen as a platform of choice in areas where, traditionally, you’d have seen products selling for 10x the price.
A good example of this is the BASpi from a Building Automation company. It’s a Pi hat with BACNet comms allowing control over A/C equipment, lighting, elevators and other things you’d find in a building. It has 12 BACnet points with another 24 virtual points and a Sedona Virtual Machine.
If you’re in to cheap building management devices this one will probably excite you.


WR-NMP52p14There’s a number of embedded O/S around. Most of you would have heard about FreeRTOS, Contiki, mbed, NuttX and at least a couple of others on this list.
One that isn’t on here is TockOS. This one claims to be Extensible, Reliable and low powered.
The core is written in Rust and contains all the usual RTOS line-up, but there’s also the concept of Capsules, which are sort of like Kernel modules but run co-operatively, as well as user space processes which run pre-emptively.
MCU support is minimal at the moment with only the Atmel SAM4L and Nordic nRF52 MCUs supported.

Particle Mesh

WR-NMP52p15This next one will interest a couple of my Patreons.
Particle have released three new modules designed for mesh networks.
The Particle Xenon runs an nRF52840 MCU supporting Bluetooth 5 mesh networks,
The Particle Boron with an nRF52840 as well as a uBlox LTE-M module.
and the Particle Argon with nRF52 as well as an ESP32.
All the boards are otherwise identical with 20 GPIOs, 2MB SPI flash, USB, JTAG and LiPo battery management.
If you look at them closer, you’ll see they are also identical to AdaFruit’s Feather.

The Trillium Project

WR-NMP52p16ARM is jumping on the AI bandwagon with an announcement of their ARM ML and OD processors. The ML is short for Machine Learning and OD for Object Detection.
These processors will allow you to detect objects in complicated orientations at 1080p at up to 60fps.
They are also claiming to play nice with other neural network software like TensorFlow and Caffe by providing a translation stack called ARM NN without any code modification.
Looks promising.