A product that has shipped for six years suddenly cannot be built, because one part on the board has reached end-of-life and the last reels are gone. It is one of the most common reasons a working design has to be touched again — and how much it costs depends almost entirely on decisions made long before the part disappeared.
The status is written on the distributor's page. The action is yours, and it is different in each row.
| What the distributor says | What it actually means | What you can still do |
|---|---|---|
| Active | In production, no restriction. | Nothing to do. Keep watching it. |
| NRND (not recommended for new designs) | Still made, still fully available, but the manufacturer is steering new designs away from it. No end date is given. | This is the cheapest moment you will ever get. Qualify an alternative now, while the current part is still available to build with. |
| Last-time buy (LTB) | The manufacturer has announced the final date on which an order can be placed. | Place your final order, sized to the number of years you actually need. This is the only moment at which it can be done. |
| End-of-life (EOL) | The last-time-buy date has passed. Distributor stock is whatever is left in the channel. | Buy only what is genuinely available from authorised channels, and start the replacement process immediately. |
| On allocation | Still in production, but supply is deliberately restricted and quantities are assigned by the manufacturer. | Do not redesign for this. Allocation is temporary; obsolescence is not. Reserve what you can and buffer your schedule. |
| Obsolete / discontinued | No longer manufactured by anyone. What exists is what is in the world, plus whatever has been imitated. | Three options only: a drop-in equivalent, a functional equivalent with a design change, or a redesign. See below. |
The distinction matters because the response is different in each case.
Two of those statuses are commonly called obsolescence and are not. Allocation is a supply problem — the part is still being made and will be again; redesigning around it is an expensive overreaction to a temporary shortage. NRND is a warning rather than a failure: nothing has been discontinued, and treating it as an emergency wastes the one advantage it gives you, which is time. A part is not obsolete when it disappears. It is obsolete the day the manufacturer stops recommending it for new designs — and that is the day the clock starts.
When an LTB window opens, the instinct is to buy for the lifetime of the product. That is sometimes right and often not, because a component in storage is not the same thing as a component in stock. Most surface-mount parts have a moisture sensitivity level, and their floor life — the time they can spend exposed to ambient air before reflow — is measured in days. Parts that sit for years need baking before use, and their solderability degrades rather than holding steady.
So the LTB order should be sized to a defined horizon, stored under the manufacturer's conditions, sealed with desiccant, and recorded with date codes as it is consumed. And it is a bridge, not a strategy: if the part carries firmware, an LTB stock that outlasts the ability to recompile for its replacement has bought time you cannot spend.
It solves the schedule problem and creates a worse one.
The open market for electronic components is made up largely of excess inventory: parts a large manufacturer over-ordered, or stock left over when another customer cancelled a build. Nothing about that is inherently dishonest. What is missing is the manufacturer's warranty and the traceability chain — so the part cannot be verified as what it claims to be, and if it fails there is no lot to trace back to.
That matters because counterfeits are not only non-working parts. The more dangerous ones are refurbished units sold as new, re-marked components of an inferior grade, parts from a failing production lot, or a faster-rated part sold at a slower part's price. Incoming inspection can catch a percentage of these — X-ray, decapsulation and electrical test each catch something — and none of them catches everything, and each of them costs money and time on every lot.
The administrative cost is just as real. An untraceable part breaks the traceability chain your quality system depends on, which is a problem rather than a preference once you build for medical or automotive customers: under ISO 13485 and IATF 16949, traceability is a requirement, and a brokered part can make an otherwise-compliant build non-compliant. The rule that keeps this simple is to ask for the manufacturer's certificate of conformance. Authorised surplus stock has paperwork. If the answer is that there is none, the discount is being paid for with risk you are carrying.
In ascending order of cost, and of how much of the original design survives.
Same footprint, pinout, voltage range and timing — usually a successor part from the same manufacturer, or a second source that genuinely matches. This is the cheapest outcome and the one to look for first. The trap is the phrase “pin-compatible”, which describes the footprint and not the behaviour. Electrical, thermal and timing differences that do not show up in the pinout have a habit of showing up on the board, so a drop-in candidate is verified against the datasheet and then verified again on an assembled unit before the whole order is built.
The part does the same job but not identically: a different register map, a different peripheral configuration, sometimes a different package on the same die family. The board may not change at all while the firmware does. This is the expensive middle path, and it is easy to underestimate: the physical build is unchanged, so it looks like a small piece of work, but it carries a full engineering change order, a new validation cycle and a new set of test limits.
New silicon family, new layout, and for medical devices a regulatory submission attached. It is almost always the most expensive option in the first year and frequently the cheapest over five, because it ends the cycle instead of deferring it — which is worth remembering when the second part on the same board reaches end-of-life eighteen months later.
Less than a distributor, more than nothing — and the difference is whether it happens before you need it.
Why traceability is a requirement rather than a preference under ISO 13485 and IATF 16949 — and what that means for where parts are bought.
Component minimums and long lead times are two of the largest hidden costs at low volume — and both are lifecycle problems in disguise.
MOQ, lead times by order quantity, component sourcing, certifications, board size limits and testing.
Gerber files and a BOM — with manufacturer part numbers if you have them. We will come back within 24 hours with a price, a lead time, and a note on any part whose lifecycle status we would not want to be relying on in two years.
