Microprocessors and microcontrollers are the brains of Industry 4.0. In fulfilling that role, they are not only having to become ever smaller and more powerful, but also process and relay signals in real time.
What machine is going to be used to make a product? What is the workflow sequence? And what happens if one of the machining stations fails? Finding the answers to those questions is currently still the job of production planning – based on high levels of employee know-how. In Industry 4.0, such decisions will be automated – that is to say, made by microprocessors. They will be built-in to tools, machines, plant components and end-products too, communicating with other components and areas of the Smart Factory, and using the information they gather to make autonomous judgements in fractions of a second to handle rapidly changing scenarios.
Ever smaller structures
Complex high-performance processors are thus the brains behind the intelligent elements that make up Industry 4.0. The huge volumes of data and complex tasks, such as the application of algorithms in decision-making, demand enormous computing power. At the same time, power consumption has to be minimised, because most processors are mobile units, powered by batteries or using energy harvesting.
Multi-core chips, featuring multiple main processors, are currently standard. In addition to data processing, they provide a wide range of control and program functions embedded in the firmware or in programmable software. The increased performance demanded by Industry 4.0 applications can best be achieved by configuring more and more transistors per chip, while at the same time making the structures smaller. But ever smaller structures entail a number of problems: The amount of leakage current as a proportion of the total power consumption of the integrated circuits increases. As a result, semiconductor manufacturers are employing alternative materials, such as High-k Metal Gate technology, to reduce leakage currents by several orders of magnitude while maintaining a consistent switching speed.
Cutting power consumption
Low power consumption was also a development goal for the new “Tomahawk 2” microchip from the Technical University of Dresden’s Center for Advancing Electronics Dresden (cfaed). As opposed to current multi-core chips, it is able to adjust the power supply to each of its 20 cores individually up or down within nanoseconds as a means of saving power. The chip is already working reliably at the lab in Dresden, and is achieving speeds six times faster than similar micro-electronic components used in mobile communications based on the same power consumption. It is able to easily interconnect a wide variety of different devices, making it a key element of Industry 4.0 for connected factories and development departments. With the new prototype, the researchers at the Technical University of Dresden’s Micro-electronics Excellence Cluster are preparing for the so-called “Tactile Internet”. This will enable very large volumes of data to be transferred in real time. “It is the next stage of the digital revolution,” says Cluster Co-ordinator Prof. Dr Gerhard Fettweis.
Chips for real-time applications
Real-time capability is a key challenge which chips used in Industry 4.0 applications such as servo drives, actuator-based motion control and inverter control have to meet. Such applications demand not only fast processing and response sensitivity, but also very good real-time behaviour. Conventional solutions such as high-frequency microprocessors with cache memory are not suitable for real-time applications owing to the difficulty in predicting cache behaviour. So microcontrollers with on-chip flash memory are used for real-time controls. But the rising demands in terms of processor power have meant that those chips, too, are reaching their performance limits. One solution is offered by chips in which the processor core accesses the high-speed on-chip memory directly instead of via a cache. This means no disturbances due to cache latency can occur in real-time applications.
Challenges for chip production
However, miniaturisation is also posing new challenges for chip production: as Dr Randhir Thakur, Executive Vice President and General Manager of the Silicon Systems Group at Applied Materials, explains: “The reliability and performance of the wiring that connects the billions of transistors in a chip is critical to achieve high yields for device manufacturers.” The company supplies systems for the production of semiconductors. “As wire dimensions shrink to keep pace with Moore’s Law, interconnects are more prone to killer voids and electromigration failures.” Consequently, Applied Materials has developed a new process whereby the copper connections are encapsulated in a thin film of cobalt. This means reliable chips can be produced even as their sizes get progressively smaller. State-of-the-art manufacturing techniques are enabling ever smaller processors. This year, the first chips in 14 nm technology will be launched onto the market – and manufacturers are already talking about 7 nm chips.
So a further step along the road to ever smaller “super-brains” for Industry 4.0 has been achieved, enabling the integration of ever smaller tools and equipment, and so permitting ever increasing penetration of a value chain with intelligence.