Why does product lifecycle planning matter in electronics manufacturing?

I spoke with a Hardware Engineer at a mid-sized to large size electronics mfr (about 3 years ago). He had just found out that the sole mfr of a particular part that he had selected for a section of the upcoming design had EOL’d that part about 18 months prior. He was in the midst of a desperate attempt to ‘design around’ in order to meet his production launch date (about 4 months hence). I was given an estimate for the ‘redesign’ of about $200k. He would also stand to loose a major retailer that had been trying to get him to create this product for over a year.

I hear from a lot of companies of how that scenario plays out for them on a regular basis. The reality is, that components age.

The component obsolescence problem is worse than most people think

Life of an electronic component these days seems to be very short indeed, and the life of a semiconductor is probably less than that of most passive components. On average, components are taken out of life (end of life) about 2.5 years after introduction to the market, and in view of the fact that the component market is global, the life of components is subject to constant changes all over the world. Designers do not like to talk about obsolescence, and so most electronics are designed to last only a few years on the market. Many changes happen in supply chains for semiconductors, and so, for instance, a component can be introduced to a design only to be designated EOL before the design even goes through the first revision. Obsolescence notices from suppliers are not always publicized, and in many cases Product Change Notification (PCN) for such component can be found deep within supplier’s website. In many cases, nothing can be found at all.

As long as the component is still in production there is no concern regarding its obsolescence. However, once it is no longer in production it may be difficult to source, and designs may need to be updated to use a newer component. Most companies account for yield degradation of purchased parts, and longer lead times, but few account for component obsolescence. Obsolescence is typically tracked in a spreadsheet that is occasionally updated, but often not in a timely manner, and as a result is often discovered too late, after the problem has already occurred.

I want to emphasize that the problems described in the previous section are typically relevant to consumer-type products which typically have 18 month product life-cycles. Companies designing products for longer-lived markets, i.e., defense, medical devices, and industrial controls, must plan for component obsolescence to a far greater degree, as the components for their products may become obsolete while the product is still in production five-to-ten years later. There is little room for error in such cases, as components for a ventilator controller or a power grid sensor, for example, cannot be revised quickly and redeployed to the field.

What does good forecasting actually look like?

I’d define the component obsolescence forecasting or lifecycle forecasting (LFC) or component end of life (EOL) or component obsolescence (COB) planning as a process that can be easily set up by a procurement department or component purchasing department or even an engineering design organization. Once the process is set up the organization can then track all suppliers to get a heads up on any roadmaps that are being developed for any component. There are also many component lifecycle databases that track all active, discontinued, obsolete and end of components throughout their respective life cycles. Using these databases the organization can identify components that are already in the mature phase or have already entered the last time buy phase. The organization can then use this information to make an informed decision regarding the component.

Some of the signals worth tracking consistently:

• Single-source components with no pin-compatible alternatives on the market
• Parts that haven’t had a significant inventory replenishment in 12-plus months
• Suppliers who have shifted focus to newer process nodes (the older parts quietly starve)
• Components where distributor stock levels have been declining quarter over quarter

Forecastering component lifecycle does not require some sort of crystal ball to predict the future. In fact, there is nothing mystical about it at all. Component lifecycle forecasting can be reduced to a simple process of: 1) tracking supplier roadmaps, 2) tracking component’s lifecycle on various databases, and 3) updating a database (such as a spreadsheet) accordingly. Then simply apply some procurement discipline and throw the resulting data at design and procurement decisions as required.

Managing from launch to phase-out

Another point of view on product planning is that the launch of a product marks the beginning of another design effort and that the engineering team has to complete the design for the current product, then move on to the next design, and then also assist the next design team in bringing a new product to market. In other words, the engineering team is fully engaged in the design of new products while the current product is in full production in the factory. This style of program management is very typical in companies that are always on the look out for the next best thing in the market place. To treat the launch of a product as the end of a task can be very devastating to a newly launched product.

The key issue with many products is that they do not stop needing to be managed after the initial launch. Electronics components will continue to deteriorate over time, and any product that has a production life of 5 to 10 years will likely require changes along the way. As mentioned previously, products used in the Defense, Medical, and Industrial markets have particularly long production lives and will require monitoring and changes over extended time frames.

Managing a product through its full commercial life from initial design release through maturity to planned end of life (EOL) is the hallmark of companies able to deliver against long-term contracts. A strong Product Lifecycle Management (PLM) discipline is a major component in the ability to manage a product throughout its entire lifecycle. This is different from simply designing a product and then letting it run for years. And then, when problems arise, bringing in a PLM ‘expert’ to fix things. This is where the rubber meets the road for lifecycle management and where a solid PLM practice can provide a competitive edge.

A side-by-side comparison of the following will illustrate the problem.

Design continuity isn’t just about components

But also the knowledge of the design needs to be kept up-to-date. This is not only true for the design decisions that have been made by the engineers who originally designed the product. Often it is the knowledge of workarounds, that has been implemented in the design because of the quirks of certain components. A design is always made with the best components that are available at the time. If an engineer leaves the company a few years after the design has been released into production, he might take with him the knowledge of why a certain design decision was made. Often however, this knowledge is not written down anywhere. It is just in the head of the engineer, maybe even on a sticky note. When the time comes to redesign a part of the design, because it is no longer available, this knowledge is often lost. And the engineers who have to redo the design, have to reverse engineer their own product to understand why something was designed the way it was. This can be a real challenge.

Planning a product’s lifecycle is a great way to force design discipline on a product. Part of the designer’s job is to design a product with parts that fit within a product’s expected lifecycle. As a designer you must track your suppliers’ roadmaps for parts you have selected for your design. You must also track the component lifecycle databases for the parts you have chosen. The designer also monitors parts for obsolescence (as a part becomes ‘mature’ or goes into ‘last time buy’ status). The designer also thinks about potential redesigns for obsolete parts and who would make decisions for a revision versus a full redesign. The designer also thinks about the acceptable substitutions for obsolete parts. These are all very difficult questions that a designer does not want to have to answer. However the designer must.

(As an aside, I believe the costs to a company of not documenting enough far exceed the costs of component obsolescence over time. But, the cost of lack of documentation to materialize is slower, therefore harder to put on a P&L on a line by line basis).

The cost of not planning is deferred, not avoided

1. Component scarcity hits and panic buying inflates your BOM cost by 30-40%
2. A rushed redesign ships with insufficient validation and creates field reliability problems
3. A customer on a long-term supply agreement gets a delivery failure they weren’t warned about
4. The relationship takes years to repair, if it recovers at all

All of these problems can be solved if they are detected in time. I recently spoke with an engineer from a midsized manufacturer. Three weeks prior to start of production of a new product the engineer discovered that the planned to use microcontroller had been discontinued. The information about the part’s obsolescence had been available in a supplier’s database for a part of long. However, the supplier’s information was not linked to the engineer’s task list, therefore it never triggered any decision.

The lack of information in the industry is not the problem. It is the information / action gap and that is a process problem dressed up as a supply chain problem. In a market where component change is rapid and where the cost of being caught out is increasing rapidly, this type of problem is extremely expensive for companies.