IoT and Mobile SoCs: Processing Power for Next-gen Devices

Ever since the first smartphones hit the market, the race has been on to pack more functionality into a single device. This doesn’t just happen at the software level, it also requires the right hardware to provide the necessary processing power. With more functionality being packed into a smaller space, newer smartphones include ever-more-powerful system-on-chip (SoC) devices that provide data processing and interfacing with other subsystems in mobile devices.

Mobile SoCs for use in newer smartphones and IoT include an LTE modem, a graphics processor, digital signal processor to support artificial intelligence (AI) functionality, cache memory, device security, and other functions, all integrated into a single chip. New products will require powerful mobile SoCs as the IoT revolution continues and functionality demands in these devices continue expanding.

IoT and Mobile SoC Controllers

With newer IoT devices requiring integration of more software-level functions that have been relegated to computers for some time, these devices will need greater processing power for use in a variety of new applications. Some envisioned functionality includes machine learning and artificial intelligence, both of which are processing and memory intensive applications. The most advanced chips currently used in commercially available smartphones are Apple’s A12 Bionic, Qualcomm’s Snapdragon 855, and Huawei’s Kirin 980, which are fabricated using a 7 nm lithography process. The table below shows a feature comparison between the three.

The Kirin 980 and Snapdragon 855 SoC controllers have CPUs based on the ARM architecture, which is an acronym for Advanced Reduced Instruction Set Computing (RISC) Machine. This architecture is licensed to micro-controller chip manufacturers for inclusion in controllers for mobile devices. The RISC architecture in ARM-based SoC controllers requires fewer transistors, hence these controllers cost less and dissipate less power than controllers found in most personal computers. This makes ARM-based mobile SoCs well suited for smartphones, IoT devices, and other embedded systems.

The newest generations of ARM Cortex-M processors target IoT device applications by providing security and machine learning solutions, in addition to high performance embedded-system requirements like real-time deterministic interrupt response, low power consumption, and 32-bit or 64-bit word size.

The Cortex-M23 and Cortex-M33 processors are available with a security technology named TrustZone^TM, which provides system-wide hardware isolation for trusted software. The Cortex-M7 and Cortex-M33 cores provide support for digital signal processing (DSP) and single-precision (32-bit) floating point processing. These capabilities will enable on-device machine learning for use in applications such as computer vision and edge computing for IoT devices.

ARM Cortex Mobile SoC Microcontrollers

ST Microelectronics, STM32L552RC

The STM32L552xx series of devices are an ultra-low-power family of microcontrollers (STM32L5 Series) built on top of the ARM Cortex-M33 32-bit RISC core in a 64-pin LQFP package. These devices include embedded high-speed memories (256 KB SRAM/512 KB Flash), an extensive range of enhanced I/Os and peripherals on two APB buses, and two AHB buses with a 32-bit multi-AHB bus matrix:

The Cortex-M33 core features a single-precision floating-point unit (FPU), which supports all the Arm^® single-precision data-processing instructions and all the data types. The Cortex-M33 core also implements a full set of DSP (digital signal processing) instructions, TrustZone aware support and a memory protection unit (MPU) which enhances the application’s security. [From the product brief]

In addition, these devices include two 5 Msps 12-bit ADCs, two DAC channels, two comparators, two operational amplifiers, an internal voltage reference buffer, a low-power RTC, two general-purpose 32-bit timers, two 16-bit PWM timers dedicated to motor control, seven general-purpose 16-bit timers, and two 16-bit low-power timers. The devices support four digital filters for external sigma delta modulators (DFSDM). Up to 22 capacitive sensing channels are available for HMI integration.

STM32L552RC block diagram from ST Microelectronics.

ST Microelectronics, STM32F746NG

The STM32F745xx and STM32F746xx series of devices have a low price point while still offering comparable or better capabilities than the previous controller. These devices are based on the ARM Cortex-M7 32-bit RISC core. It also implements a full set of DSP instructions and memory protection unit (MPU) for enhanced IoT application security. This series of devices also incorporates high-speed embedded memories (320 KB SRAM/1 MB Flash), including 64 KB of TCM RAM for real-time processing of critical data.

In addition to the bus architecture found in the previous product, this product offers similar signal processing/conversion functionality with advanced communication features:

All the devices offer three 12-bit ADCs, two DACs, a low-power RTC, thirteen general-purpose 16-bit timers including two PWM timers for motor control and one low-power timer available in Stop mode, two general-purpose 32-bit timers, a true random number generator (RNG). They also feature standard and advanced communication interfaces. [From the product brief]

NXP Semiconductors, MKL16Z256VLH4

The MKL16Z256VLH4 is an ultra-affordable mobile SoC built on the ARM Cortex-M0+ core running at 48 MHz. Although it offers slower processing speed, it still provides 32-bit processing, ultra-low power consumption with sleep mode, and embedded memories (32 KB SRAM/256 KB Flash). Given its lower price point, multiple standard communication interfaces, and analog modules (16-bit SAR ADC and 12-bit DAC), one application for this product is in small IoT devices that will acquire and process signals from sensors. This product comes in a 64-pin LQFP package, although there is a variant that comes in a 64-pin MAPBGA package (MKL16Z256VMP4).

MKL16Z256VLH4 block diagram from NXP Semiconductor.

Embedded computing in IoT and other application areas will continue to advance, and you can maximize the performance of your next system with the right microcontroller or other programmable logic device.

We hope you found this post useful! Stay up-to-date with our latest articles by signing up for our newsletter.