Using Open Standard Modules in the IoT age
Demand from consumer, industrial and medical markets is creating the pull for accelerated embedded systems design. By reducing dependence on low-level hardware design, project resources can focus on differentiating features at the application level.
Standardized Computer-on-Modules (COMs) and Single-Board Computers (SBCs) are designed to be engineer-friendly. Recent products combine smaller form factors with the latest processor architectures and features. The Open Standard Modules™ specification is among the latest and smallest. It adds robustness with its soldered-down ball grid array (BGA) outline.
Standards accelerating embedded development
The COM approach puts common computing resources into a standard module. A standardized COM specification, such as COM Express introduced by the PCI Industrial Computer Manufacturers Group (PICMG) in 2005, establishes a common form factor and interfaces that facilitate integration with carrier boards to allow flexibility and scalability as the project evolves.
The COM Express form factors are:
- Mini (84 mm x 55 mm).
- Compact (95 mm x 95 mm).
- Basic (125 mm x 95 mm).
There are various pinouts:
- Type 6 interfaces emphasize graphics applications.
- Type 7 was conceived for server-grade platforms.
- Type 10, in the 84 mm x 55 mm mini form factor, supports compact embedded designs.
Single-board computers, as an alternative, are ready to use out of the box and typically easier to set up. They combine computing components with I/O ports, connectors and other features in a single unit.
The 90 mm x 96 mm PC/104 form factor, introduced in 1987, is an example. Oriented around the ISA bus, PC/104 makes PC technology accessible for embedded development with features such as a self-stacking bus and by providing pin headers instead of edge-card connectors.
As PC technologies change, additional specifications arrive. An example is PCI/104 Express, offered in the same 90 mm x 96 mm form factor. This format makes the high-speed PCI Express® bus available for modular embedded applications.
Other platforms have emerged, taking advantage of technologies such as Arm Cortex-A application processors. Examples include the Raspberry Pi Compute Module and projects popular in the maker community, such as BeagleBone and Arduino. These offer SBC-like convenience in a compact form factor and are often used for prototyping, small-scale embedded projects, and industrial and IoT applications.
Compute module form factors are shrinking
The embedded development trend is toward easier, faster development, using a smaller form factor with higher performance and lower power. This trend is leveraging the latest processor, memory and interface architectures.
Avnet is a founding member of the Standardization Group for Embedded Technologies (SGET), a not-for-profit association. SGET specifications include the SMARC® (“Smart Mobility ARChitecture”) computer module, Qseven® and Open Standard Modules™ (OSM).
SMARC comprises standardized 82 mm x 50 mm and 82 mm x 80 mm modules sizes for portable and stationary embedded systems typically under 6 watts. The Qseven module is good for mobile and IoT applications. Its main features are:
- A 70 mm x 70 mm or 40 mm x 70 mm form factor
- A slim profile
- Graphics capability
- Audio processing ability
- Mass storage
- Network connections
- Multiple USB ports
SGET’s Open Standard Modules (OSM) specification makes the most of ultra-low-power application processors. About the size of a postage stamp, OSM can replace predecessors the size of a credit card. OSM represents a significant miniaturization of modular COM/carrier designs.
OSM is aimed at embedded and edge IoT systems running open-source operating systems. The modules are based on various processor architectures including MCU32, Arm, and x86. The modules are directly solderable, BGA, surface-mount packages. This makes them rugged and suitable for harsh industrial environments. In addition, the BGA design can reduce assembly costs in production.
Detailing the Open Standard Module specification
The OSM specification sets out four different sizes:
- Size-0 "Zero" measuring 30 mm x 15 mm with 188 contacts,
- Size-S “Small” measuring 30 mm x 30 mm, with 332 contacts,
- Size-M “Medium” measuring 30 mm x 45 mm with 476 contacts,
- Size-L, “Large” at 45 mm x 45 mm with 662 contacts.
As a comparison, the 45 mm x 45 mm Size -L module is 28% smaller than the 40 mm x 70 mm µQseven and 51% smaller than SMARC (82 mm x 50 mm).
Using these modules, applications become processor agnostic, as well as offering scalability. OSM is future-proofing applications by simplifying and accelerating processor design-in. Both the hardware and software specifications of OSM modules are open source, encouraging collaboration and innovation within the embedded and IoT system development community.
The specification for each module includes basic interfaces that are routinely needed in embedded systems, including Ethernet, I2C, SPI, UART, and USB with the option of USB 3.0 on modules above Size-0. Size-S and larger modules have an RGB display interface, DSI and CSI display and camera serial interfaces, and PCIe x1.
Sizes M and L have embedded DisplayPort, and Size L additionally includes an LVDS display interface and PCIe x4. The specifications also provide scope for vendor-defined contacts, ranging from up to three contacts on Size-0 modules to 19 on Size-L.
The specification supports wireless communication by defining antenna connection positions. In addition, a standardized evaluation carrier board provides access to the features of all module sizes. Using a carrier board can accelerate development by delaying surface mount assembly until later in the project.
OSM supports different processors
Open standard module vendors have the flexibility to create modules with a wide variety of different processor types and interfaces. Some examples currently available (from Avnet Embedded) include Size-M modules built with NXP i.MX 8M Mini application processors that contain up to four Arm Cortex-A53 application processors, combined with a graphics processing unit (GPU) and Cortex-M4 microcontroller core with DSP extensions that provide real-time capabilities.
Other examples include Size-S modules based on the i.MX 91 Cortex-A55 processor and Arm Neon media processing engine, the i.MX 93 processor with Cortex-A55 cores and the Arm Ethos-U65 microNPU that efficiently handles machine-learning applications. As i.MX 9 processors, these modules also benefit from NXP’s EdgeLock on-chip security designed to protect IoT devices against network-based attacks.
The four OSM sizes
Modules designed for small IoT applications
OSM modules provide a compact, robust and power-saving nerve center for mobile robots such as industrial autonomous guided vehicles, lawnmowers or drones. They can capture and process data from multiple sensing channels including inertial and ranging sensors for guidance and obstacle avoidance. The modules can also leverage on-device artificial intelligence (AI) to accelerate machine-vision response, all with a minimal power budget.
Applications may include smart city devices, such as traffic flow and parking monitors, and surveillance for congestion and emissions charging. The modules can power down into microwatt idle modes, enabling operation from harvested solar energy stored in a small battery. This power profile can help save installation costs and accelerate rollout. With the CSI interface permitting high-resolution image data, and edge inferencing for fast number-plate recognition, developers can also take advantage of the native security features to protect against online threats including secure over-the-air (OTA) firmware upgrades.
Conclusion
Standardized computing platforms have accelerated the pace of embedded systems development. Their high integration and standard format are opening the field of deeply embedded design to more organizations. OEMs that may not have hardware design skills are no longer at a disadvantage.
The OSM specification takes miniaturization to the next level, adding application processors and the latest low-power technologies, heterogeneous processing including AI, and hardware security. Moreover, the solder down BGA connection enhances robustness and production readiness.