MCU Advances Focus on Neural Processing, RRAM, and Payment Security
MCU vendors continue to up their game. Latest chips serve up neural processing for the edge, innovations in on-chip non-volatile memory, and tech for secure biometric card payments.
This year has seen a steady roll out of microcontroller (MCU) technology advances. Gone are the days when MCUs were just a mix of processing, simple RAM memory, and peripherals. Today’s crop of MCUs pack in a rich set of technologies aimed at complex system design challenges.
In this article, we round up some recent near-end-of-the-year MCU news ranging from edge processing functionality to embedded MCU memory to payment security.
MCUs Serve up Neural Processing Technology
For its part, NXP Semiconductors recently announced its new MCX N MCUs. An expansion of the company’s MCX family announced back in June, the MCX N are the first devices to embedded an NXP NPU (neural processing unit).
At the Electronica trade fair last month, All About Circuits’ Editor-in-Chief Jeff Child (left) gets a demo of the MCX N MCU from Romain Ricci (right), NXP’s regional marketing manager for MCUs.
The first two members of NXP’s MCX N series include the MCX N94x and the MCX N54x. The two devices are general purpose MCUs and embed dual Arm Cortex-M33 cores with clock speeds operating up to 150 MHz. Other features include 2 MB of flash, with optional full ECC RAM, and a DSP co-processor for audio and voice processing.
As mentioned above, these MCUs are NXP’s first to introduce a new version of the NXP-designed Neural Processing Unit (NPU). The company says this integrated NPU provides up to 30x faster machine learning (ML) throughput versus a CPU-only core solution. This means it can spend less time awake, which shrinks overall power consumption. Engineers can make use of NXP’s eIQ ML software development tools to train and support ML models using the integrated NPU.
Rich Security and Peripheral Mixes
For security, the devices leverage NXP’s EdgeLock secure subsystem. Edgelock provides secured boot and crypto accelerators. The MCUs have on-the-fly encrypt and decrypt
for external serial flash/PSRAM that is connected via the FlexSPI interface. This ensures that code and externally stored data are protected, says NXP.
Block diagram of the N94x version of the MCX N MCU. Image used courtesy of NXP. (Click on image to enlarge)
While the N94x and N54x share many of the same features, the MCX N94x puts emphasis on advanced analog and motor control peripherals, while the MCX N54x includes peripherals ranging from high speed USB with PHY to SD or smart card interfaces suited to IoT and consumer applications.
The N94x boasts a CoolFlux DSP, three embedded op amps and two CAN-FD bus links. In contrast, the N54x lacks the DSP and op amps, and has a single CAN-FD interface. The MCX N94x and MCX N54x appear to share a product page. More information is available on the MCX N fact sheet. NXP plans to begin sampling the MCX N series in Q1 2023.
AURIX TC4x MCUs Get TSMC's RRAM On-chip
For its part, Infineon Technologies’ most recent MCU news revolves around memory. Infineon and TSMC recently announced that the two companies are working to add TSMC’s Resistive RAM (RRAM) non-volatile memory technology into Infineon’s next gen AURIX family MCUs.
The goal here is to serve the needs of automotive systems, says Infineon. MCUs with embedded flash have traditionally been key components in automotive electronic control units (ECUs). They are also critical for next gen advances such as electrification, new E/E (electrical/electronic) architectures, and autonomous driving.
Infineon and TSMC are working to embed TSMC’s RRAM non-volatile memory technology into Infineon’s AURIX TC4x MCUs. Image used courtesy of Infineon
While today’s MCUs are based on embedded flash memory technology, Infineon says that RRAM is a next step in embedded memory because it enables scaling to 28 nm and beyond. For Infineon’s AURIX TC4x MCUs, TSMC and Infineon say they have successfully created the basis for introduction of RRAM in the automotive domain.
The RRAM offers “high disturb immunity” while allowing bitwise write without the need of erase. Meanwhile, endurance and data retention performance is comparable to flash, says Infineon.
Infineon says it is now shipping samples of its AURIX TC4x family to lead customers that are based on TSMC’s 28 nm eFlash technology. First samples based on 28 nm RRAM technology are expected to be available to engineer customers by the end of 2023.
Certified for Biometric Card Payments
In other MCU news, STMicroelectronics (ST) recently announced that it has completed EMVCo1 certification for its STPay-Topaz-Bio biometric payment card platform. The platform is co-packaged with ST’s STM32L443 MCU in a module compliant with EMV specifications.
According to the company, the certification confirms the security of the platform and its interoperability with payment systems. EMVCo is the organization that manages the EMV specifications and test programs for card payments worldwide.
ST says the certification means that STPay-Topaz-Bio and its embedded secure hardware is now “the first one-stop-shop EMVCo-certified platform to comprise a biometric secure element and secure operating system (OS).”
ST’s MCUs are at the heart of its secure STPay-Topaz-Bio platform for biometric payment cards. Image used courtesy of STMicroelectronics
The secure element used in the platform is ST’s ST31N600 secure element IC, which is basically an MCU in its own right. The chip offers a variety of defenses for card-payment applications and sensitive processes. Among these are biometric template matching for cardholder authentication, which uses software libraries developed with Fingerprint Cards AB (Fingerprints).
Running on the ST31N600 is a secure OS. The device is built on the latest Arm SecurCore architecture for secure MCUs. In addition, designers can introduce value-added card features by securely connecting various types of peripherals.
In the STPay-Topaz-Bio platform, the STM32L443 MCU handles non-sensitive aspects, says ST. These include managing the card’s fingerprint sensor module and the user interface. The STPay-Topaz-Bio platform also boasts energy harvesting for batteryless operation and meets standards for contact and contactless cards: EMV ISO 7816, ISO 14443.
MCU Complexity Under the Hood
Embedded system design is all about complex functionality under the hood that enables powerful simplicity at the user level. To achieve this for today’s demanding applications, MCU vendors are providing new types of co-processing, memory technologies, and application-specific functionality such as secure payment capability.
Solutions such as described here will perhaps address emerging needs of design engineers, but the appetite for future advances will continue to grow.