- Published in Milestones.
Linking ECUs that controlled the engine, transmission and brakes so that realtime information could be transmitted was a big challenge for engineers in the early 1980s as they looked to boost vehicle performance and safety. The standard technology of the time, universal asynchronous receiver/transmitters, didn’t support the complex task of multidomain communications. So in 1983 Robert Bosch started a programme to develop a communications protocol that could link multiple domains in real time.
The process began with a three-month series of courses in information engineering led by university professors and attended by a chosen few engineers at the Tier One supplier. Among those engineers was Uwe Kiencke, who headed the department developing the system.
Kiencke’s team was small. There were seven other engineers on the project: Wolfgang Borst, Wolfgang Botzenhard, Siegfried Dias, Otto Karl, Martin Litschel, Helmut Schelling and Jan Unruh. Every six months they reported their results to their superiors, so decisions could be made quickly. In 1985 Intel was invited to take part in the project. In the eyes of Kiencke and his colleagues, that company was the inventor of the microprocessor, and therefore had to be involved.
Intel wanted assurances that the technology was suitable for the automotive sector and that demand for it would be strong enough. It asked Professor Lawrence from the Fachhochschule Wolfenbüttel engineering school in Germany to assess the system. Lawrence’s report was positive, and he even contributed the protocol’s name, Controller Area Network (CAN).
In 1985 Bosch and Intel began developing the required silicon chips. Two years later the first microprocessors for use in the CAN communication protocol became available. But challenges had to be overcome. Kiencke said later: “Engineers took the average message traffic, and introduced a so-called reserve factor, by which they multiplied message traffic. Calculating transfer rates from this can not guarantee message transmission within a given time under peak load. It is not the average latency time, but the maximum latency time for a subset of high-priority messages, which determines the transfer rate on the physical layer.” It was a steep learning curve.
Kiencke and his colleagues presented papers at various conferences. This was a key step in bringing European carmakers to accept the technology and in helping to open negotiations with US manufacturers.
Protracted discussions were held with the International Organization for Standardization, but no ISO standard was agreed upon. Kiencke said: “As competing communication protocols profited from the frank discussion with ISO, the penetration of CAN into applications was rather discouraged than enhanced. Classical standardisation seems to have lost its role as a driver of technological innovation.”
The setback didn’t stop the development of the system. In 1991 Kiencke’s work was fulfilled when the first CAN network was installed in the Mercedes-Benz S-Class. It linked five ECUs with a data transfer rate of 500Kbit/s.
Kiencke was proud of his work: “CAN has entered applications in distributed automation systems and automotive electronics,” he said. “CAN contains new approaches which optimise performance and support the control of safety-relevant systems. With more and more silicon implementations available, CAN will help to reconfigure the architecture of future automation systems,” he said.
Today CAN remains the backbone of most vehicles’ electric/electronic architectures. Without it, safety systems wouldn’t function.