03-08-2021 | | By Robin Mitchell
While the consumer industry does everything it can to replace old tech, some applications continue to use old tech, and with good reason. Why can old tech be hard to replace, should Ethernet replace technologies such as CAN, and how can newer technologies replace older technologies in critical applications?
One of the biggest challenges faced by consumers in the electronics industry is the need to constantly replace ageing technology. The need to replace old technology can come from a multitude of reasons including newer standards that don’t work with older technology and failing hardware which could include batteries no longer charging or faulty parts on a PCB. This need to replace older hardware is also having a knock-on effect on the environment with large amounts of e-waste being generated.
If you think having to replace technology is difficult, imagine how it must be for an engineer who has to do the exact opposite and try to keep old systems functioning reliably. There are many examples of systems that are extremely dated which still use technologies such as floppy disks to function. In such scenarios, trying to repair equipment can be a nightmare as such hardware will no longer be in production. Thus, it is down to the engineer to either find spare parts on the market or find ways to breathe new life into dysfunctional equipment.
In such applications, surly new technology could easily replace old hardware? As it turns out new hardware could very easily replace old hardware, but doing so is not always advisable. The reason for this comes down to reliability and dependability. Imagine a nuclear power plant that has been operating safely since the early 90s. The computers used to control the plant will undoubtedly use a command-line disk operating system, and a single desktop PC using Windows 10 would probably have more processing power than the entire power plant.
However, upgrading the plant’s computers with new systems could lead to a seriously dangerous situation for one simple reason; the old computers are proven and the new computer is not. Basically, the computer system controlling the plant has proven to operate well for decades, but a new PC from a local hardware store has had a grand total of 0 proven years of continuous operation. If a new computer is used in an application whereby reliability is critical, there is no telling if the system will function correctly for extended periods of time. For a desktop PC application, this is an acceptable risk, but for a nuclear power plant, no risk is acceptable.
This inability to use newer technology also explains why NASA craft have typically used very old computing systems, why the HMS Queen Elizabeth was found to be using Windows XP, and many businesses use decades-old mainframes. To see evidence of what happens to systems when trying to implement new technology without decades of reliability evidence, simply look towards Tesla. From car crashes to dashboard errors, Tesla vehicles are based on newer technology which does not have the same degree of reliability as other vehicles using an older technology.
This may seem like a very specific question to answer, but it is being asked after reading an online article that discusses the advantages of Ethernet over CAN and why automotive engineers should switch over. To sum up, what was being said, engineers should look towards a “clean-sheet-architecture” meaning that everything they know about standards and hardware preferences should be forgotten, new technology should be explored, and then a new design standard can be formed around the latest technologies. Ethernet technologies provide many software advantages over CAN, provide simpler wiring with the use of standard RJ45 cables, operate at higher speeds, and can meet the high demands from software-defined systems.
If automotive vehicles were to be designed from scratch today, Ethernet would be a good candidate for communication between different devices in vehicles. As it was rightly stated in the original article, Ethernet is a high-speed communication system that works well with modern technologies. Furthermore, features such as Power-over-Ethernet could help to reduce the amount of wiring in vehicles, and the use of standardized connectors would make car electrical systems far easier to route and maintain.
However, vehicles are not being designed today, and replacing CAN with Ethernet is something that simply won’t happen for at least a decade. One of the most important concepts around CAN is that it is designed with reliability in mind. Unlike consumer electronics, automotive products are under strict requirements to be safe and dependable due to the fact that they are controlling 3-ton death machines driving at high speed.
CAN also have the added advantage that it uses termination resistors with all devices connected to the same bus. Any wiring mistakes or failed devices can be determined rapidly by any other device connected to the network. Furthermore, CAN uses a priority system to ensure that devices whose messages are most important are transmitted on the network. Ethernet however uses messages that are sent to a hub which then routes the message to a controller. The lack of message priority means that it can take time for a controller to respond to a major fault or error message.
However, the most fundamental point that will not see CAN replaced anytime soon is that it has evidence of decades of reliable operation. The CAN standard has been upgraded over time, but it is internationally recognized, and the many testing procedures formed around CAN show that it is a reliable bus for automotive systems. While Ethernet also has decades of use, it has not been used in the automotive industry, and as such cannot be reliably put into vehicles.
While vigorous testing can be done on systems to prove their safety and reliability, the best method for replacing older technologies is to put newer technologies into applications that are not safety-critical but treated as such. From there, reliable operation for long periods of time will prove system reliability and thus make it appropriate for such a system to replace ageing systems.
However, this means that whatever technology is used to replace an older system will always be outdated. This should not disused engineers from trying to replace older systems with the latest technology, but it should be appreciated that if an old system works well, replacing it for the sake of replacing it can be unnecessary.
There is a good reason why CAN has prevailed in automotive systems, and trying to replace it with unproven technology could end very badly. But, in anticipation of the future, engineers should start to experiment with current technology systems to see how they could be used to transform technology.