Nuvoton’s One-Arm M2L31 Microcontroller Uses ReRAM (Memristors) for On-Chip, Non-Volatile Storage

Where most of today’s microcontrollers use embedded Flash EEPROM for non-volatile storage of program and data, Nuvoton’s M2L31 family employs ReRAM. Depending on the family member, the size of the on-chip ReRAM ranges from 64 to 512 Kbytes. Nuvoton asserts that its ReRAM on-chip storage enjoys at least three advantages over Flash EEPROM: faster write speeds, superior durability, and lower power consumption. The faster write speed arises from ReRAM’s ability to update data without page-erase cycles.

Ceva Bluetooth Low Energy and 802.15.4 IPs Bring Ultra-Low Power Wireless Connectivity to Alif Semiconductor’s Balletto Family of MCUs

The Ceva-Waves Bluetooth Low Energy IP provides Balletto MCUs with the robust connectivity at ultra-low power consumption, and supports Bluetooth LE Audio and Auracast broadcast audio, for customers who wish to leverage Balletto to create highly differentiated wireless audio products. The Balletto family also relies on Ceva-Waves 802.15.4 IP for Thread, Zigbee and Matter support in smart home applications.

The dsPIC33A family features integrated analog peripherals, including 12-bit ADCs capable of conversion rates up to 40 Msps, high-speed comparators and operational amplifiers. These analog peripherals, in conjunction with Core Independent Peripherals (CIPs), allow for sophisticated sensing and high-performance control. In addition, the CIPs enable interaction among the peripherals without the need for CPU involvement, enhancing the efficiency of a single controller to manage multiple tasks. The result is more robust real-time control while reserving the CPU bandwidth for software stacks, functional safety diagnostics and security functions.

Did you know you can read full articles for FREE?

That's right - our free Starter Membership gives you access to 3 free articles a month.

Try this one:

Building a WebUSB or Web Serial Gadget

You Don't Even Need an App for That

Read Now

Editor's Desk by Kirsten Campbell

Beyond the Basics: The Rise of AI and Machine Learning on Microcontrollers

Tailoring AI for Microcontrollers

One of the biggest developments in this space is the creation of lightweight ML models specifically designed to run on microcontrollers. They are highly optimized for size and efficiency so they operate within the limited memory and processing power constraints of microcontrollers.

Techniques such as quantization, where model weights are reduced in precision, and pruning, which removes redundant parameters, make it possible to deploy complex AI algorithms on devices with as little as a few kilobytes of RAM.

TensorFlow Lite for Microcontrollers, a pared-down version of Google's TensorFlow, is a prime example of how the AI community is adapting to these constraints. It allows developers to train models on powerful machines and then deploy them on microcontrollers, enabling real-time AI inference without relying on cloud-based processing.

The Power of Processing Locally

By enabling AI at the edge, microcontrollers can reduce the need to send information back and forth to the cloud. This significantly reduces latency, making real-time decision-making possible. For example, in industrial automation, a microcontroller could instantly detect anomalies in machinery vibrations, triggering preventative maintenance without waiting for cloud-based analysis.

AI Pioneers

As AI continues to evolve, the role of microcontrollers will only grow. We can expect to see more sophisticated AI algorithms being adapted for these tiny devices, enabling even more complex decision-making at the edge. For engineers, this represents a new frontier in embedded systems design.

So, what's next for you?

What would you like to see, more editorial or less, more of a particular topic, or less? Let us know. Reach out to the Product Editor

Yuzuki Chameleon: A $25 Raspberry Pi-Like Board with Allwinner H616 SoC

The Yuzuki Chameleon is a single-board computer designed with the form-factor of the Raspberry Pi model A, offering an open-source and versatile platform based on the Allwinner H616 chipset. This SBC targets users looking for a compact yet powerful device capable of handling various applications, from media streaming to IoT projects.

Upcoming Events: