The Notice Arrives and the Clock Starts

A product change notification lands in a distributor email. A part you have been buying for three years is going end-of-life. There is a last-time-buy date, usually 6 to 12 months out, and a last-ship date, usually 12 to 18 months out. After that the part does not exist at any legitimate source.

The notice is not negotiable and the dates rarely move. What you control is the response, and the difference between a good one and a bad one is typically measured in months of lost shipments. Two things matter immediately: understand what is actually being discontinued, and calculate exposure before choosing a path.

Read the notice carefully first. A true EOL discontinues the part. A PCN may only change the package marking, die revision, or lead finish, which can break your process without breaking your design. A "not recommended for new designs" notice means the part is alive but its clock has started.

Size the Exposure Before You Decide

Do this in a day, not a month, because the last-time-buy window is the resource you are spending.

  • How many products use it. Search every active bill of materials, not only the one you are thinking of. The same regulator often appears in three products and a test fixture.
  • Units per year, and for how long. Multiply by parts per unit, add service and warranty demand, add scrap. A device with a 7-year service obligation carries far more demand than the sales forecast alone.
  • Where the part sits. A passive with fifty equivalents is an afternoon of work. A microcontroller whose firmware, pinout, and certification depend on it is a redesign.
  • What is downstream. Inside a certified subsystem, a change may trigger FCC or safety retesting, and in a medical device it engages design controls.
  • Current stock and pipeline. Your inventory, your contract manufacturer's, and what is already on order.

The output is one number: how many parts you need to cover the gap between now and a qualified alternative being in production. Every subsequent decision is a comparison against that number.

Option 1: Last-Time Buy

Buy enough parts to cover remaining demand. This is the default answer for parts that are hard to replace and cheap to store.

The advantages are obvious: no engineering work, no requalification, no certification risk. The costs are less obvious. Capital is tied up in inventory that generates nothing until it ships. Moisture-sensitive devices need dry packaging, controlled humidity, and attention to floor life. Solderability degrades, so parts stored more than about three years may need requalification. And if you forecast low, you are in the same position later with no options left.

Forecast generously: expected demand plus 20 to 50 percent, sized against confidence in your projections. Buying 30 percent extra of a $2.40 part is cheap insurance; the same margin on a $95 FPGA is a real capital decision. Treat the buy as a bridge and use the runway to execute one of the options below.

Option 2: A Drop-In Replacement

The best case: a part with the same package, the same pinout, and equivalent electrical behavior, either a second source or the manufacturer's own successor part.

"Drop-in" deserves suspicion until proven. Check the parametric differences datasheets bury: quiescent current, dropout voltage, ESR requirements on the output capacitor, gate charge, internal pull-up values, power-on reset behavior, and timing margins. A replacement regulator may need a different compensation network, and a sensor's replacement may return a different register map behind an identical footprint.

Then qualify it properly. Build a batch, run it across the full temperature and voltage range, and repeat the environmental and EMC checks the part could plausibly affect. A replacement regulator with a different switching frequency can move a radiated emissions peak into a limit line and cost a recertification. Budget 2 to 8 weeks.

Option 3: Redesign

When no equivalent exists, or the obsolete part is one of several aging components, a board revision is the honest answer.

Scope it deliberately. A minimum-change revision touches only the affected circuit, keeping the qualification burden small. A consolidation revision refreshes several at-risk parts at once and is worth doing on an older board, since it saves three more of these events.

Cost and schedule for a moderate board revision: 4 to 12 weeks and $15,000 to $60,000 including layout, prototypes, firmware changes, and qualification, plus certification retesting where the change reaches an emitter, a power supply, or a safety function. If the obsolete part is a microcontroller, add firmware porting time, which is usually the longest item, and revisit the selection criteria in how to choose a microcontroller so the replacement has a longevity program behind it.

Whichever route you take, run it through a controlled engineering change order. An undocumented part swap that reaches production without updating the BOM, the test procedure, and the service documentation creates a field population nobody can identify later.

The Gray Market Trap

When a part is gone, brokers appear offering exactly what you need at three times the price. Some are legitimate independent distributors with real testing capability. Many are selling parts recovered from scrapped boards, remarked to a different date code, or outright fake.

Counterfeit parts often pass a basic functional test and fail in the field or after a few months, which is the worst possible failure mode: they get through your line and into customers' hands. If you must buy on the open market, use a distributor with in-house testing, insist on decapsulation and X-ray sampling for high-value parts, and test every lot. The detection methods are in counterfeit electronic components.

Reducing the Odds Next Time

Obsolescence is not preventable, but exposure is manageable.

  • Check lifecycle status at design time. Distributor lifecycle data and manufacturer longevity programs, which commit to 10 or 15 years of supply for industrial lines, cost nothing to consult before you commit a part.
  • Prefer parts with real second sources. A jellybean regulator available from four manufacturers in the same package is worth a few cents over a single-sourced part with marginally better specs.
  • Subscribe to PCN notifications for every part on your BOM through your distributor. Most teams discover an EOL when an order fails, which wastes months of the window.
  • Design margin at the pinout level. A footprint that accepts either of two packages costs nothing at layout time.
  • Review the BOM annually for lifecycle risk, alongside the broader planning described in component shortage planning and the sourcing structure in supply chain management for a hardware startup.

Get an Obsolescence Response Planned

Projects House handles EOL events for shipping products: exposure analysis across the BOM, replacement evaluation, board revision when needed, and the qualification and change control that keeps certifications intact. Send your EOL notice and BOM through our contact form.