Powerful NXP System on Module SoM: Which Options Are There?

NXP I.MX6ULL SoM – Cost-Optimized for General-Purpose Applications

From the lineup, the NXP I.MX6ULL SoM is the cheapest option because it runs on the i.MX 6ULL applications processor, which is a power-efficient and cost-optimized SoC. This chip has a single-core ARM Cortex-A7 processor that can run at speeds of up to 900 MHz.

Although it might be a bit slow compared to the other SoMs we’ll look at later, it makes up for that with power efficiency. Additionally, it has a built-in power management module that reduces the complexity of the required external power supply. This module also simplifies power sequencing.

Memory-wise, this SoM supports various configurations to meet different customer needs. The standard is 8GB eMMC and 256MB NAND flash, but these can be increased to 256GB and 2GB, respectively.

On the connectivity front, the NXP I.MX6ULL has 2 high-speed USB OTG, 2 ethernet ports, and various serial interfaces that are handy for industrial applications.

The i.MX 6ULL applications processor falls under the i.MX 6 Series, which NXP builds for general-purpose solutions in automotive IoT, IIoT and industry 4.0 automation, and consumer applications. So you can use the NXP I.MX6ULL SoM to develop the following solutions.

• HMIs
• V2X communication
• Small-engine ECUs, such as motorcycle
• CAN bus gateways (for heavy machinery)
• Robotics
• 2D barcode scanners and printers
• Industrial edge computing gateways
• POS terminals
• Industrial-embedded Linux computers

The SoM has a small form factor (38mm x 38mm), so it can easily fit in these devices without requiring a large case or frame.

Case Study: DSGW-081 Industrial Modbus PLC Gateway with 2 SIM Card slots

Also Read: What is System on Module?

NXP i.MX 8M Mini SoM – For Consumer Electronics and Industrial IoT Applications

The NXP i.MX 8M Mini SoM is based on the i.MX 8M Mini applications processor, which falls in the i.MX 8 Series of chips that NXP builds for advanced neural network processing, graphics, video, audio, and machine vision applications.

But since this is a mini, it is on the lower scale of the performance aspect for these applications. However, it employs the advanced 4LPC FinFET process technology to deliver high speeds and improved power efficiency.

It features a heterogeneous system consisting of ARM’s big.LITTLE architecture, with a quad-core A53 that runs at speeds of up to 1.8GHz (per core) to do the heavy lifting, and a Cortex-M4 (400MHz) to run lightweight tasks while consuming little power.

The i.MX 8M Mini System on Module also has a VPU and GPU to enhance video and graphics processing applications, rich interfaces (MIPI-CSI, MIPI-DSI, PCIE 2.0, Gigabit Ethernet, I2C, UART, etc.), 2GB RAM, and 16GB eMMC.

But this SoM’s most significant visible difference from its sibling is that it leads out 200 pins, including almost all CPU pins, via two 100-pin board-to-board expansion connectors. These give access to various I/O signals and peripherals. It is also compatible with Linux, Android, FreeRTOS, and Windows 10 Enterprise.

The board is larger though, because it measures 60mm x 49mm, which you might consider a disadvantage if building an extremely compact device. But this isn’t a significant issue. Overall, this SoM is a cost-effective option for building edge computing IoT gateways, real-time industrial control systems, predictive maintenance (IoT), image analysis, and two-way video conferencing solutions.

Also read:

The Potential of i.MX 8M Mini SoM in Industrial and Medical Applications

A Typical IIoT Edge Computing Gateway

What Is an Industrial Gateway?

Case study: DSGW-089 IIOT Edge Computing Gateway for Elevators

NXP i.MX 8M Plus SoM – For Machine Learning, Vision, Multimedia, and Real-time Industrial Automation and Control

This NXP i.MX 8M Plus SoM falls in the same family as the Mini above but packs more power in its processing cores. It features a similar heterogeneous system with the big core being the 1.8GHz Cortex-A53 for multitasking and running complex computations.

But the little chip is slightly more powerful than in the Mini(single 800MHz Cortex-M7), and it handles real-time control tasks efficiently while ensuring high reliability and low latency. This core is ideal for running control networks, and the board has dual CAN FD and dual Gigabit Ethernet (with TSN) to support this industrial control application.

On top of that, this i.MX 8M Plus System on Module features a 2.3 TOPS NPU for accelerating machine learning inferencing in facial detection, object detection/recognition, and natural language processing. This NPU is critical for the SoM because it can process neural networks about 30X faster than ARM’s cores.

But the SoM is particularly suited for machine vision applications because it features two camera interfaces (MIPI-CSI) and dual image signal processors (12MP) with up to 375MPixels/s input rates to provide the video feeds into the intelligence computing system.

Its multimedia capabilities are also top-notch because it has a GPU for 3D and 2D graphics acceleration, a VPU for 1080p (60 frames/s) encoding and decoding, and low-power voice acceleration (powered by a Cadence Tensilica HiFi 4 digital signal processor). The SoM also has various audio outputs and input interfaces (eARC, 8-channel PDM mic, etc.), and three display interfaces.

But its main selling point is its machine vision capabilities, making the SoM ideal for powering AI cameras and AMR/AGV robots.

Also read: Why NXP i.MX8M Plus SoM is A Powerful and Versatile Platform for Machine Learning & Vision?

Differences Between i.MX 8M Mini, Nano, and Plus

These three SoCs are quite popular, and since they are in the same i.MX 8 Series family, we have to compare them to see their differences performance-wise.

The other features not included in this comparison are similar in all three.

NXP I.MX 93 SoM – For High-performance Automotive, ML Acceleration, and Real-time IoT Edge Computing

NXP builds i.MX 9 Series SoCs for high-performance applications because they feature independent MCU-like, low-power, real-time domains to handle critical functions with ease. They also feature ML acceleration and EdgeLock Secure Enclave for maximum security.

The NXP I.MX 93 SoM borrows these features because it runs on the i.MX 93 SoC, which is part of that series family. Its multicore, heterogeneous system features two Cortex-A55 cores (1.7GHz) paired to a single 250MHz (Cortex-M33).

The SoM’s Cortex-M33 provides real-time responses in the low-power, real-time domain functions, while its Arm Ethos-U65 microNPU enables it to run energy-efficient ML inferencing for applications like speech/audio recognition.

In a nutshell, this SoM’s selling points are its energy-efficient, high-performance ML acceleration and security. Since it’s built on the Energy Flex Architecture, the i.MX 93 SoC in the module provides flexible performance and power tuning with dedicated clocks and supply voltages for its heterogeneous cores and NPU.

On matters security, EdgeLock provides:

• Key management
• Platform security and trust
• Crypto services
• Remote attestation
• Trust provisioning

So if you’re looking for efficient, secure, high-performance ML processing solutions with rich interfaces (CAN bus, camera interface for machine vision, multiple display interfaces, etc.), this module is the one to go for.

Final Words

All these SoM modules are supported by a wide range of development documents and software resources to ease application development. Dusun IoT’s task is to handle hardware design, BSP development, and testing, but we also provide app development support to make your work easier and reduce your time to market.

But you have to pick the right SoM for the task, which is why I’ve done this detailed comparison. For instance, I recommend the i.MX 8M Plus module for robotics applications and smart shopping carts because it has two camera inputs and ISPs. Two is better than one, so its vision system will be highly accurate.

However, you can integrate sensor vision using LiDAR by connecting the sensors to the other interfaces to improve vision. So the NXP I.MX 93 SoM can also work accurately. Plus it introduces efficiency and security into the mix, making it the best option in the lineup.

They are not limited to AI and ML though. Contact us with your requirements, and we’ll help you pick the most suitable SoM for your project and budget.

You May Also Be Interested In