Component obsolescence – what to do next?
Sometimes it is ICs, sometimes connectors or passives. Every component has a finite longevity and even the cheapest, smallest, seemingly insignificant part can jeopardise production of a whole product line.
While the majority of mainline semiconductor manufacturers are very good at labelling where their parts are on this list, here’s a roundup of the typical product cycle of a component with some of the different terms used by different suppliers: Marketing -> Preproduction / Beta / Sampling -> General Availability (GA) / Active-> Not recommended for New designs (NRD) -> End of Life (EoL) -> Obsolete
Marketing
Despite the marketing brief being posted, for some reason you just can’t get hold of it, or the data sheet. Perhaps you can get on an alpha program, but do you need this pain as you will be helping to qualify the part? Dates for GA are written in sand at this stage, so beware as if they slip, that will likely adversely affect your own schedule.
Preproduction / Beta / Sampling
That hot new part which your engineers want to use is one that you can only get samples of, if you’re lucky, and you might even need to be on a beta program to get hold of those. Again, you will probably find some TBDs in the data sheet, or it’s watermarked with “Draft”. GA dates should be more solid now and the part is “safe” to use. Occasionally however, parts don’t go from Beta to GA, such as when a technical snag is discovered.
General Availability (GA) / Active
The part is in full production. The question now is just how far through GA it is. Is it relatively new or near to NRD? Looking at the dates on the data sheet will help. At this point, errata and apps notes may be available, which can be very useful.
Not recommended for new designs (NRD)
As the label says, don’t use it unless you have a very specific need or a small production run. Time to look at last time buy options. Beware of the grey market (parts of indeterminate providence).
End of Life (EoL)
May be rolled into the NRD stage, so now is last time buy time.
Obsolete
You missed the last time buy and specialist suppliers are finding pockets of them at the back of dusty warehouses around the world. They may even be genuine.
Why components go obsolete
It is worth understanding the rationale for obsolescence, as it can affect the initial choice of parts and the route taken to replace those parts. There are a number of reasons why components go obsolete, including:
1. Declining sales making continued production uneconomical for the manufacturer.
2. Acquisition or merger of companies resulting in rationalising of product lines.
3. Technology changes (e.g. Fab geometry) making the means of production no longer available.
4. Regulatory changes e.g. RoHS.
5. Sub-components going obsolete (e.g. telephony modems have their chip-set obsoleted).
In most cases, you, the designer, will have no influence over any of these factors unless you are a large user. However, when selecting components, it is worth considering the likelihood of any of these events taking place. A good example is with LCD panels, where they are often made with a specific market in mind and perhaps a single customer. When that customer obsoletes their product and no longer buys the product, factor 1 above applies and all the other minority customers will see the panel suddenly go EoL.
When selecting parts, it has to be said that not all suppliers are equal with regards to longevity. Some like NXP and ST publish longevity plans on their core processors, allowing you to clearly see how long their parts will be available for. But beware! Just because the processor has a 15-year longevity doesn’t mean the memory chips in the system have longevity to match, in fact this is highly unlikely. So, in this case, it is merely that the processor can be confirmed as unlikely to be the part that calls time on your product. Other semiconductor manufacturers can be much more focussed on fast moving consumer goods and frequently obsolete parts.
With LCDs, there is an element that you get what you pay for. The lowest-cost LCDs typically are tied to other consumer products and will have a short shelf life. Not a problem if your own product longevity is a couple of years, but if it is longer, some defensive design around the LCD interface may be required (e.g. the ability to add conversion boards to other panels).
For connectors and passives, you can’t beat designing in generic parts or multi-sourced options. Where this isn’t possible, again some companies are more “agile” than others. It’s worth asking the suppliers the question on key parts as you may get an honest answer.
What can design engineers do when a component becomes obsolete?
Normally a supplier will offer a last time buy on components that they are planning to make obsolete for a part. You can sometimes sign up for email updates on a part to alert you of status changes. This is well worth doing as it is free and could save you any headache. Likewise, some of the catalogue distributors (e.g. Farnell) also send out product alerts when the status of parts change.
Let’s assume though that you have run into trouble as your part is completely obsolete. You know that’s you’re going to have to find a resolution to sort the problem out, so here are the only options that you now have:
I. Find a drop-in alternative
It may be that there is an alternative that is pin compatible. Perhaps one that didn’t exist when the design was done. If there are enough other people in the same boat, that latent demand can help alternatives to appear. Be inventive in how you look for an alternative as parts that are the same are sometimes categorised in different ways by various manufacturers.
2. Scrap your product
If your own product is heading towards the exit door, you might have an alternative that can do most of what the terminally ill product can do. This is, however, a business decision that’s outside of the scope of this particular article.
3. Modify your circuit board
If the product is selling well, a re-design may be a no-brainer, but you will likely have to face these costs for this option:
a) Redesign effort of the schematics, BOM and PCB
b) Prototype run to test non-trivial changes
c) Software changes (depending on what component is changing)
d) Compliance re-testing where appropriate (e.g. probably not for a connector change but for a new microcontroller).
This is the time to do a complete sweep of the whole bill of materials to look for any of those NRD labels indicating that other parts whilst not EOL are nearing their own obsolescence. Like the road being dug up by the gas men, just after the same stretch has been newly laid by the water works, you don’t want to launch your revised product just to find another problem.
There can be in-between solutions, such as making daughter boards that replace a single component on a circuit board, or sit between one PCB and an LCD, and convert the signals for a new panel. Thus, the development cost is much lower than for a full re-design. If you do a full-redesign, this gives you an opportunity to perhaps get some cost out of the product or to add new features. Seize the moment and it will make the procedure less depressing! Perhaps it could turn a good product into one that is great.
About the author:
Dunstan Power is Director of ByteSnap Design – www.bytesnap.com
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
