Nordic Semiconductor has announced a new member of that family, the Nordic Thingy:53

Nordic Thingy:53 Embedded Machine Learning and Internet of Things Rapid Prototyping

With Edge Impulse integration, you can create tinyML models from start to finish.

Thingy is a series of fast prototyping platforms developed by Nordic Semiconductor. By integrating sensors and interfaces to their popular system-on-chip (SoC) solutions, these gadgets go beyond standard development boards. The Nordic Thingy:53, a new member of the family, was introduced today.

Nordic designed this new Thingy around their flagship nRF5340 SoC, as the name says. It also comes with a wide range of sensors, over-the-air update capabilities, and out-of-the-box interaction with the Edge Impulse embedded machine learning platform.

An inbuilt nPM1100 power management IC (PMIC) and a nRF21540 front-end RF module round out the nRF5340 (FEM.) A slew of sensors is also included on the board. While Thingy:53 is mostly wireless, it also has physical connectors such as USB-C, a Qwiic/Stemma/Grove type connector, a programming header, a power switch, and a debug board port. Additional connectors on that board allow you to measure current usage using tools like the Power Profiler Kit.

Let's have a look at the SoCs first.

The nRF5340 was the first SoC featuring two distinct Arm processors and wireless connection when it was released. This notion is similar to dual "core," in that it involves two somewhat different processors splitting edge processing work. For high performance, one Arm Cortex-M33 operates at 128 MHz with 1 megabyte of Flash and 512 kilobytes of RAM. It also features a floating-point unit (FPU) and can do digital signal processing (DSP) operations. Another Arm Cortex-M33 processor is designed for network processing. It has 256 kilobytes of Flash and 64 kilobytes of RAM and works at up to 64 MHz.

The nRF5340 features a 2.4 GHz radio that supports Bluetooth LE, Bluetooth mesh, and proprietary protocols, as well as Bluetooth LE and Bluetooth mesh. It also supports Zigbee, Thread, Matter, and even NFC protocols. Thingy:53 also features a nRF21540 FEM for improved wireless communication range and reliability. A power amplifier (PA) and a low-noise amplifier are included in this RF front-end IC (LNA).

If you're unfamiliar with Matter, it was previously known as Project Connected Home over IP. The Connectivity Standards Alliance (CSA), formerly known as the Zigbee Alliance, hosts this working group. Smart home technologies like Apple HomeKit, Amazon Alexa, Google Assistant, Samsung SmartThings, and others may connect with Matter-enabled products. Matter, while not a wireless technology, is an application layer that sits on top of a physical layer like Bluetooth.

When the USB-C connection is used to charge the internal 1350 mAh LiPo battery and the buck converter, a nPM1100 PMIC is used.

Thingy:53 Sensors provides a large selection of sensors for embedded machine learning applications. The following sensors are among them:

• Temperature
• Humidity
• Air quality
• Air pressure
• Color/light sensor
• 6-axis IMU (accelerometer + gyroscope)
• Magnetometer
• PDM (pulse density modulation) microphone

When activities occur, the microphone and accelerometer might wake up the SoC. The microphone, for example, may detect sound, and the accelerometer, when motion is detected, may activate. Low-power activities benefit from these advantages.

An RGB LED, user-definable push buttons, and an SWF-style connector for RF measurements with a spectrum analyzer are also included.

If you require a sensor that the Thingy:53 does not currently have, you may use the Grove, Stemma, or Qwiic connection to expand its capabilities. This connection connects to a variety of sensors, actuators, and display ecosystems that are simple to connect.

It's amazing to collect data from all of these sensors, but that data needs to go somewhere in order to be useful.

Firmware development and machine learning

Thingy:53 is clearly aimed at embedded machine learning (ML) applications. Nordic worked with Edge Impulse on the default firmware first. Second, the firmware is ready to gather sensor data and send it to a mobile device running the Edge Impulse Mobile software over Bluetooth LE.

This combination cuts down on the time it takes to transfer data into Edge Impulse Studio, where you may build (and test) an embedded machine learning model. You can then deploy the model back to the Thingy:53 when it's ready to run on the nRF5340's high-performance application processor.

Edge Impulse is a machine learning development platform for embedded (or edge) applications. Accepting input from practically any sensor type, generating a model, testing it (in simulation or real), and distributing it to edge targets like the Thingy:53, speeds up the installation of a machine learning solution.

Nordic's nRF Connect SDK supports the nRF53 and Thingy:53 for more extensive firmware development. Other Nordic families, like the nRF52 and nRF91, use the same Zephyr RTOS-based SDK. You may also load fresh firmware over the air with the nRF Programming mobile app.

For more information, go here.

Nordic claims that wholesalers such as Farnell will receive supply as soon as today. Visit the Thingy:53 product website to discover more about this innovative IoT prototype gadget.