Microsecond-Level Real-Time Ethernet Deterministic Bus (REDBUS): Architecture and Motor Control Experiments

For decades, real-time digital control systems have enabled countless applications across industry, transportation, defense, and science. A more recent trend is enhancing their interconnections using hardware and protocols, delivering timing performance comparable with that of the controllers themselves, thus enabling considerably more complex forms of real-time coordination over multiple devices, such as robot axes. Several industry standards have emerged, primarily leveraging existing IEEE802.3 Ethernet hardware and differing levels of the associated stack. However, these are generally proprietary and involve complex software libraries, which are desirable in terms of abstraction and multi-vendor interoperability but are not well-suited for situations where low-cost, small-size, and high-reliability remotization of individual sensors and actuators are required. In this paper, we introduce a complementary approach known as the Real-time Ethernet Deterministic Bus (REDBUS), which is characterized by an atypical usage of the IEEE802.3 physical layer, rewired according to a daisy-chain topology, yielding a sort of token-passing network. The cascaded devices replace control values with sensor readings on the fly at predetermined frame locations under a minimalist framework devoid of any software layers. The concepts and implementation are described in detail, an example code is provided, and representative results on the real-time control of high-speed stepper and brushless motors are presented. Due to the lack of abstraction, security, scalability, diagnostics, and multi-protocol integration features, REDBUS cannot replace the existing industrial Ethernet standards on the factory floor. Rather, it aims to complement them within predetermined designs, such as individual pieces of equipment that represent closed ecosystems, where complexity, size, and timing requirements are more pressing and where the entire data flow allocation and timings can be fixed ab initio. Diverse engineering fields can directly benefit from this simple and unconstrained architecture for meeting demanding experiment control and data acquisition requirements.