The demand for intelligence in machines is permanently on the rise. Recent IoT and Industry 4.0 trends accelerate this demand. Development in this area is cost-intensive and time-consuming. Engineers are under pressure to optimise their bill of material. The deployment of industrial-grade platforms based on commercial IT standards, such as x86 and Mini-ITX, paves the way for engineers to profit from highest re-use efficiency. So it comes as no coincidence that congatec’s first platform of this type is equipped with 2nd generation AMD G-Series SOCs.
Machine builders and automation vendors constantly have to improve the quality and efficiency of their applications in order to stay competitive in a globalised market. Machines are being equipped with more and more intelligence and additional functions as well as improved user experience thanks to innovative 3D graphics and intuitive multi-touch operation.
On top of all this, there is a prevailing demand for improved machine flexibility and production agility which calls for new ways of monitoring, management and control. For nearly all automation and machine builder applications, the quest is to provide extended connectivity and embedded intelligence for the IoT and Industry 4.0 integration. The aim is to offer new user values generated for both machine operators and machine vendors by the analytics of all the big data created by the machines and applications.
We are already familiar with this type of intelligence boost. We experienced it when smartphones took over from cellphones. Instead of only being capable of phone calls, smartphones provide a host of additional functions, making it a multi-functional all in one device that includes additional services. This development is now also taking place in industrial applications. Machines are evolving into smart machines. And this is the future of Industry 4.0 applications: fully network-connected machines creating added value thanks to an unlimited information flow and increased functionality. However, the main task of the machines is still to provide the same core functionality as before. But there are more competitors vying to supply this core task as now additional IoT players have come onto the market. Therefore engineers need a way to ideally balance cost with all these new features which are to be added. Commercially available building blocks help to cut down costs and to increase functionality in the most efficient way.
Gaining more efficiency through maximised re-use
The re-use of common technology standards guarantees a broad range of offerings as well as generic long-term availability. In the case of intelligent control systems and human machine interfaces it is more or less common practice to re-use standards (de-facto or real) of the commercial IT world, as this market has the largest, most homogenous user group. So, re-using x86 architectures is recommendable, as they enable OEMs to most efficiently re-deploy existing software, development tools and OS. OEMs, however, have to be aware of the fact that some requirements in industrial installations differ significantly from traditional IT. First and foremost is the question of longevity. Ruggedness and industrial interfaces are further differentiating factors. Therefore, processors and boards, which are designed for use in an office computer, are not suitable for use in industrial applications. Platforms with dedicated designs for the embedded sector are required.
Re-using processor technology
The AMD Embedded G-Series SoC platform that comes in an industrial-grade system-on-chip design fits this bill. AMD has developed the Embedded G-Series SoCs specifically for power-, and cost-sensitive SFF designs with high graphics demands. They are available as dual- and quad-core versions and 8000-series AMD Radeon graphics.
For HMIs and Panel-PCs which today need to deliver impressive user experiences, the discrete-class graphics, which are integrated into the AMD Embedded G-Series SoC, support applications that previously required a separate graphics processor. Thanks to the DirectX 11 and OpenGL support, OEMs can re-use the same tool-chains that engineers use to develop commercial high-performance graphics solutions, also offering all the features required for very impressive machine GUIs. Their capability to control up to two independent displays broadens the range of options of HMIs and starkly enhances user experience. Furthermore, with OpenCL – originally designed for high performance embedded computing – compute-intensive tasks, like image processing in vision systems, can be reassigned to the graphics processor with high parallelism. For this type of application, the integrated GPU provides a computing performance of up to 256 GFLOPs.
So it becomes apparent that standard IT technology delivers highly attractive features for automation engineers and these come ‘out of the box’ on a single chip. AMD is one of the leaders in this x86 SoC area and also boasts an excellent reputation in the graphics area. The company is also pushing the heterogeneous system architecture, which paves the way for flexible code handling on the CPU or GPU. Even more advantageous though is the ability to re-use code everywhere for highest investment security – within a dedicated platform, a product family or the entire x86 ecosystem. This re-use is most important for machine builders and automation companies because their code requires a long-term stable generic ecosystem for re-use that also offers seamless migration paths.
Re-using standard form factors
In order to implement this processor technology into an application, machine builders and automation vendors are faced with choosing amongst a huge range of form factors. All of them have been developed to offer dedicated benefits. ATX-based motherboards enjoy the highest deployment rates – which can also be seen as re-use – as they originate from the standard IT sector with its massive field deployments. Like all form factors, motherboards offer a defined set of interfaces and, first and foremost, standardised mechanics. This includes the PCB footprint, the interface plate as well as the mounting holes which also enables the re-use of all the different components built around this standard.
For embedded designs, the Mini-ITX form-factor is especially recommendable. At only 170 mm x 170 mm it is the smallest ATX compatible form factor – allowing for space-saving designs. These small dimensions make the Mini-ITX, for example, a perfect fit for small stand-alone Box-PCs, book-sized PCs for industrial cabinets as well as HMIs and Panel PCs with screen sizes starting from 12”. For system design and housing, there are hundreds of different dedicated casings or modular casing families available. So OEMs can usually find a perfectly suited housing for Box- or Panel-PCs off-the shelf. Plus, a high re-use quote has a positive impact on the pricing. Just think about small box towers for Mini-ITX boards, which are available for less than €50 in the commercial sector, and consider all the mechanical components inside a box like this. No other standard can deliver so many different mechanical components with such a price tag. And many of them can be re-used in industrial applications. By implementing riser cards for PCIe expansions, Mini-ITX can even be used in the most space-restricted applications so it’s advisable to first check embedded Mini-ITX boards before considering any other board level embedded form factor. And this recommendation is seconded by the huge I/O eco system that is also available for this form factor.
Re-using the I/O eco-system
Mini-ITX motherboards provide many long-term available I/Os which also constantly profit from throughput improvements thanks to the success of different interface standards like USB, Ethernet and PCIe. The same arguments are valid for the re-use of commercially available peripherals. OEMs can re-use, for example, a broad range of industrial cameras for industrial imaging based on USB or PCI Express based frame grabbers. Further to this, today, a vast number of sensors and individual extensions are based on USB and can therefore easily be connected to any standard motherboard. Plus, a broad range of PCI Express and mini-PCIe expansions is available to integrate field buses, digital and analog I/O cards as well as wireless connectivity such as WLAN and/or UMTS/LTE. By re-using these proven and tested off-the-shelf components, OEMs can easily build their application-specific embedded control and visualisation core with only minimal efforts and time. Plus, thanks to the high economies of scale, these off-the-shelf components also offer an excellent price performance ratio. So even extending a system is quite an efficient process when using commercially available components. One can even say that – compared to any other standard – ATX offers the richest ecosystem which in turn provides highest level flexibility. This makes it a perfect fit for many different Box-PCs and Panel-PCs configurations which definitely differentiates this standard from any backplane-orientated form factor standard.
Re-using only dedicated embedded versions
Machine builders and automation vendors will find a broad range of commercial Mini-ITX motherboards, which technically may fit their requirements. These boards, however, follow the fast developments of the consumer market with short lifecycles, which are insufficient for embedded engineers. OEMs cannot manage the integration of a new board every 6 to 12 months. Instead they require a lifecycle of several years, since changing a component usually makes time-consuming requalification necessary – not forgetting the amount of work involved in spare parts management and support handling. A long-term available board is therefore required to minimise work. Product availability, however, is not the only reason to opt for embedded motherboards.
Another argument for embedded motherboards lies in the fact that consumer boards are often not prepared for the harsh environmental conditions in the industrial sector. Therefore, to fulfill the demand for reliable, 24/7 operation, a rugged board design with industrial-grade components is required which can work in industrial temperatures ranges such as 0°C to +60 °C and which also offers electromagnetic compatibility (EMC) and electromagnetic susceptibility (EMS). Additionally, passive cooling solutions differentiate industrial designs from commercial ones ensuring high-level shock and vibration resistance.
Furthermore, the industrial segment still requires legacy interfaces. These include LVDS, to connect attractively priced displays, and serial ports for common industrial peripherals as well as legacy supervision interfaces. All these interfaces are not offered on commercial IT boards, but can frequently be found on embedded motherboards. Often, embedded boards also offer additional interfaces; for example for intrusion detection or General Purpose I/Os which enable more added value in industrial system designs without the need to design-in dedicated solutions for these features. So it definitely makes sense to go for industrial-grade quality.
Benefitting from single source re-use of the embedded vendor’s expertise
Small BIOS adoptions like boot logo, BIOS map, or display adoptions, are often demanded by OEMs. These small, yet necessary modifications call for a supplier that is service-orientated as well as experienced in this type of customization. Vendors headquartered in the Asian timezone often are not the right partners for such purposes. But even within the European time zone, differences in the quality of services are recognizable. Some vendors only support large scale applications and they tend to forget that the new small customers might one day become the biggest. Other vendors plan to shift their focus from boards to systems so that the interest in customisation on board level is shrinking. Instead they forward these questions to third party service or distribution partners, who usually do not have such deep insight into the board design. This approach is always destined to create information delays as well as information gaps or even mistakes in the communication chain. Unnecessary delays in the development and increased costs are the result.
