On 11 March 2011 there was an earthquake about 40 miles east of Japan. The Tohuku earthquake, the world’s fourth most powerful recorded since 1900. It triggered a chain reaction that included a tsunami that killed nearly 20,000 people, as well as an explosion and partial meltdown at three nuclear power reactors at Fukushima Daiichi.
As a result of this disaster, many chip fabrication facilities located in Japan went offline. And even after the disaster was contained and life continued, many of those chip fabrication facilities did not continue to manufacture the older chipsets. Instead, they chose to focus on producing the next-generation chips rather than retool and continue making the old.
Very abruptly, this sequence of events spelled the end of KEB’s Combivert F4 range of drive controllers. The F4 had been in production for almost 20 years before it was replaced by the Combivert F5 model. This is good in the electronics world. However, outside of some limited parts stock, the product was no longer repairable because KEB did not have access to integrated circuits used on its control boards. KEB was not the only company affected in this way by the Fukushima disaster.
This leads to the topic of this article: how long will industrial automation components be manufactured and supported before they are made obsolete? This is an often overlooked aspect of vendor selection: namely, choosing suppliers that plan for long product lifecycles and support OEM customers for many years.
Obsolescence is a major hidden risk when selecting automation components.The engineering time that is required to design and validate systems is very expensive. So a premature ‘redesign’ shortly after a product launch carries both an opportunity expense and a capital investment expense. A ‘sticking plaster’ fix that is not thoroughly vetted can underperform and be unsafe.
Components that are mechanical in nature are relatively easy to repair or replace. Motors can be rewound. Bearings can be replaced. Gears can be re-cut and adapter plates machined. However, integrated circuits (ICs), memory and microprocessors found in inverters, PLCs and HMIs are much more difficult to replace. Sometimes used or overstocked parts can be found online from websites such as eBay or Alibaba. However, the quality and longevity of the electronic parts are dubious. The option to swap out ICs for different models or brands is often not even an option, particularly for small volumes. Chipset dimensions, voltage levels, heat dissipation and source code are a few obstacles that come to mind.
THE SHINY NEW MODEL
There exists a pocket in the market of automation companies whose product is somewhere between industrial and commercial grade. Industrial, because when the product is released it is a viable functioning product. The listed manufacturer performance specifications are great; maybe the best on the market. And they are sold at a very competitive price, which sounds like great value and a no-brainer.
In the author’s experience as a buyer of such consumer electronics, the product price, chip manufacturer and performance specs such as ‘clock speed’ are openly listed and very easy to compare. With the rapid technology advances in consumer electronics, it is fairly easy to quickly compare these specs and rank options.
However, later on customers have told KEB that they have had to ‘redesign’ the machine because the two-year old PLC or HMI they purchased from a supplier is no longer available, and there is no viable alternative. Occasionally, one year later, that same PLC or HMI model has been replaced by a newer model using even newer chipsets. This is crazy and incredibly frustrating.
In other words, the drawback to this approach becomes supporting the older components in the field. Almost always, the burden to redesign to accommodate the obsolescence falls onto the OEM or the end user.
Longevity and support are often more valuable than computing horsepower. Businesses do not (or should not) make decisions in the same way that consumers make decisions. End users make capital investment decisions to buy machinery which they expect to run for easily 10 years, 20 years or longer. Therefore, businesses should value machine designs and components that have long expected, supported lifetimes.
Companies do not always publish product lifecycle graphs, but hopefully their R&D and engineering teams make decisions to support a target product lifecycle. A product will go through various phases in its life (see diagram above). Most of its life will be in the ‘production’ phase. Eventually, it will be phased out in favour of a newer model. But even after production has ended, the product must continue to be supported and repaired – especially if the product is used in critical applications.
In general, KEB strives to design products for at least a 10-20 year lifecycle. The actual lifecycle of some products might end up being more. For example, the Combivert F5 drive controller is still in active production and the product is around 20 years old, although it has had a couple of facelifts throughout the years (the latest F6 drives are pictured below).
WHO DECIDES?
The actual lifetime varies by product and usually ends up being determined by the IC and component manufacturers. Companies such as Intel and ARM have many different chips, but they select a few models that they guarantee an extended lifetime and support. This allows manufacturers of industrial equipment to ensure that their product will be supported for a minimum number of years. Selecting components for longevity early in the design process has a large impact on lifetime. Other features such as using an SSD for memory can also work to extend the operating lifetime.
Obsolescence can be delayed, but never avoided altogether. For sure, customers should receive notice when a product has entered a ‘legacy’ or ‘mature’ stage. This is when the product is no longer being actively produced or developed but still supported for replacements and repairs. This is the time that the user should begin to proactively develop a plan for upgrading. If a product is abruptly made obsolete, a viable alternative can quickly be implemented. In addition, customers should be notified of a ‘last-time buy’ opportunity where they can strategically purchase parts to support their transition plan.
To summarise, engineering redesigns are time-consuming and costly. But component obsolescence will occur and cannot be avoided entirely. However, manufacturers of automation products can lessen the blow by designing for a long product lifecycle and giving adequate notice before end-of-life. When making a buying decision, it is not only about the performance and price. Ensure you ask your supplier how long the product will be supported and where it is in its product lifecycle.