Every year, major semiconductor manufacturers discontinue thousands of part numbers. The products built around those parts — industrial controllers, medical devices, avionics, automotive modules — do not retire on the same schedule. They carry certification cycles, installed bases and service obligations measured in decades. The result is a permanent gap between what franchise distributors list and what the world still needs. Filling that gap safely is what obsolete component sourcing is about, and it is a discipline with real rules, not a lottery.
This guide is written from the daily practice of a Hong Kong sourcing desk that handles end-of-life (EOL) inquiries every working day: where obsolete parts actually come from, how to decide between repair, last-time-buy and redesign, how to vet a supplier you have never used, and how to keep counterfeit risk near zero.
What "obsolete" actually means
A part number travels a predictable lifecycle: introduction, volume production, maturity, last-time-buy (LTB) notice, end-of-life, and finally obsolescence. The terminology matters because each stage changes your sourcing options:
- NRND (Not Recommended for New Designs) — production continues, but the manufacturer is steering new projects elsewhere. Supply is usually still normal; this is the moment to plan, not to panic.
- Last-time buy (LTB) — a final order window, typically 6–12 months, with delivery stretching beyond that. Miss the window and you enter the open market permanently.
- EOL / obsolete — production has stopped. Franchise distributors delist the part. Everything you buy from now on comes from residual factory stock, distributor inventory rotation, or the secondary market.
Two important nuances. First, obsolescence is per ordering code, not per family: a manufacturer may keep making the base silicon while dropping the industrial-temperature or leaded-package variant your BOM calls. Second, acquisitions scramble lineage — legacy Freescale, National Semiconductor, Burr-Brown, International Rectifier, Intersil and Microsemi parts all live on inside larger catalogs (NXP, TI, Infineon, Renesas, Microchip), sometimes under new codes, sometimes discontinued quietly.
Where obsolete parts actually come from
Understanding the supply channels tells you where the risks sit:
- Factory and franchised-distributor residual stock. The safest channel: sealed original packing, known date codes, traceable paper trail. It exists more often than buyers assume — manufacturers and franchises liquidate slow inventory continuously.
- OEM/EMS excess. Contract manufacturers regularly liquidate overbought or cancelled-project stock. Quality is typically good (parts came through franchise originally), but storage history varies — verify moisture sensitivity handling.
- Independent distributors with reserved EOL stock. Specialists who bought last-time-buy quantities years ago and hold them for exactly this demand. Pricing reflects scarcity; provenance should be documented.
- The open spot market. Shenzhen's electronics districts aggregate all of the above — plus, frankly, the counterfeits. This channel has the deepest availability and the highest variance. It rewards buyers with inspection discipline and punishes those without.
Decision framework: repair, last-time-buy, or redesign
When a critical part goes EOL, you have three real options. The right choice depends on remaining product life, volumes and certification constraints:
| Option | Best when | Cost profile | Main risk |
|---|---|---|---|
| Spot-buy EOL stock as needed | Low/uncertain volumes, service & repair demand | High unit price, zero NRE | Counterfeit exposure; availability gaps |
| Last-time-buy / stockpile | Stable production forecast, storage budget available | Moderate unit price, capital tied up | Over/under-estimating demand; storage degradation |
| Qualify an alternate / redesign | Long remaining product life, engineering capacity free | High one-time engineering cost | Re-certification delays; subtle parametric differences |
In practice, mature programs combine them: a stockpile sized to cover the certification window, spot purchases for service demand, and a redesign running in the background. What does not work is ignoring the LTB notice and hoping the market will provide forever — for commodity parts it usually will, but for FPGAs, precision analog and military-screened discretes, open-market supply thins out within 3–5 years of obsolescence.
The counterfeit problem, honestly stated
Obsolete sourcing carries the highest counterfeit exposure in the industry, because demand persists exactly where legitimate supply has ended. The typical failure is not sophisticated forgery — it is the recycled part: a used chip pulled from scrap boards, cleaned, re-surfaced ("blacktopped") and re-marked with a fresh date code. Such parts often work initially and fail in the field.
The defense is layered inspection, and it is non-optional: label and marking verification, solvent testing for re-surfaced packages, X-ray comparison of die and bond wires against known-good references, and electrical testing where the application justifies it. Any supplier who finds these requests annoying is telling you something important. Our counterfeit detection guide walks through the methods in detail, and our quality control page documents the pipeline we apply to every lot.
Supplier vetting checklist
Before placing a first obsolete-parts order with any supplier, work through this list:
- Physical address and warehouse — verifiable, in a known electronics trade hub. A mailbox address is a red flag.
- Inspection capability stated in writing — visual, X-ray, electrical: which stages are standard, which on request, and who performs lab work.
- Photos before payment — a serious supplier sends label, marking and date-code photos of your actual lot without being asked twice.
- Report-before-payment terms — inspection documentation should reach you before final funds release, especially on first orders.
- Warranty in months, not days — 365 days is the professional standard for independent distribution; 30 days is a hobbyist.
- Honest availability language — a supplier who quotes every impossible part instantly is quoting fiction. Real desks say "no" or "two weeks" regularly.
- Reference customers or repeat business — ask directly; legitimate suppliers will connect you.
- Traceability records — lot codes, packing records and country-of-origin documentation retained and shared.
How we source obsolete parts at RISEIC
Our Hong Kong operation sits one hour from the deepest component spot market on earth, and we use that position deliberately. We hold reserved EOL stock in the lines our customers need most — discontinued FPGA and CPLD families such as Xilinx Spartan-6 and Virtex legacy, Freescale-era microcontrollers, legacy Linear Technology and Burr-Brown precision analog, military-screened discretes from the Microsemi heritage line now under Microchip, and through-hole legacy memory in PDIP packages.
When stock does not cover a requirement, the desk sources the Shenzhen and Hong Kong markets against audited channels, then runs the incoming lot through our standard inspection pipeline — marking verification, solvent screening, X-ray where indicated, and partner-lab electrical testing on request. The inspection report reaches your inbox before final payment. That sequence — source, inspect, document, then ask for payment — is the whole point of using a specialist instead of a marketplace listing.
Reading the EOL notice: PCN and PDN documents
Manufacturers announce discontinuation through Product Change Notifications (PCN) or Product Discontinuance Notices (PDN). These documents are worth reading carefully, because they contain your real deadlines: the last order date, the last ship date, and — often overlooked — the list of recommended replacement part numbers. Three habits make these documents useful instead of alarming:
- Track notices for your active BOM, not just your problems. Every manufacturer publishes PCN/PDN feeds; a quarterly review of notices against your BOM gives you 12–18 months of warning instead of a surprise at reorder time.
- Distinguish order deadline from ship deadline. The gap between them can be a year. An order placed on the last order date may not ship until well into the following year — plan cash flow and storage accordingly.
- Check whether the notice covers your exact code. Discontinuation is per ordering variant. A notice listing the commercial-temperature reel code may leave your industrial-temperature tray code in production — or vice versa.
Storage, shelf life and what date codes really mean
A date code is not an expiration date. Silicon does not spoil; packaging and solderability age. The real constraints on older stock are practical: solderability of leads degrades with oxidation over many years (recoverable through re-tinning), moisture-sensitive packages need verified dry storage, and electrolytic capacitors genuinely age. For hermetically sealed and properly stored ICs, parts from the 1990s routinely pass full functional test. What matters is documented storage conditions — which is why we state date codes on every quote and store stock in humidity-controlled space, and why you should ask any supplier how their older stock has been kept.
Three sourcing scenarios we see every week
The service-repair scramble. A failed industrial drive needs one obsolete driver IC; the machine's downtime costs more per day than the part costs in hundreds. Success factor: speed with verification — same-day market sourcing plus X-ray before dispatch beats a cheap untested part that fails on power-up.
The certification trap. A medical or aerospace product cannot substitute parts without re-certification, so the EOL part must be found, period. Success factor: provenance documentation. The inspection report is not bureaucracy — it is what keeps the certification file intact.
The quiet bridge buy. A redesign is approved but 18 months from production. Success factor: a sized stockpile — enough to cover the bridge window plus forecast error, bought once, stored correctly, cheaper than weekly spot purchases at rising scarcity prices.
The economics of EOL sourcing
Obsolete pricing follows scarcity, not cost curves. A part that sold for 2 dollars in production may trade at 20 two years after EOL — and at 6 briefly, if a large factory lot suddenly liquidates. Three implications follow. First, quote validity windows are real: spot prices move, so a professional quote states how long it holds, typically 24–72 hours on volatile lines. Second, batch economics favor commitment: buying a full available lot almost always prices better per unit than drip-feed purchases, and removes the availability risk entirely. Third, the cheapest offer is rarely cheap: below-market pricing on a scarce part is the single most reliable counterfeit indicator in this industry.
Negotiation is normal, but negotiate on volume and payment terms rather than squeezing unit price past the market floor — a supplier who accepts an impossible price has adjusted the product, not the margin.
Category notes: where obsolescence bites hardest
FPGAs and programmable logic are the classic EOL pain: high unit values, design-locked (a bitstream targets one exact device), and certification-locked in defense and medical. Discontinued families like Spartan-6 remain actively traded; inspection is non-negotiable because values justify sophisticated fraud. See our FPGA & CPLD category for current stock depth.
Microcontrollers sit in the middle: pin-compatible family siblings often survive the exact code's discontinuation, so alternates are frequently viable without a board spin.
Precision analog (legacy Burr-Brown, Linear Technology) is risky to substitute — offset, drift and noise specs differ between "equivalents" — but the parts themselves are counterfeited less than digital ICs because volumes are lower.
Legacy memory and through-hole parts persist surprisingly well: PDIP SRAM, EPROM and 5-volt logic move through surplus channels steadily. Storage condition is the variable to verify.
Military-screened discretes (JAN/JANTX series) anchor maintenance programs for decades-old equipment and are among the most remarked parts in the market; lot paperwork matters as much as the part.
Building your EOL response playbook
The organizations that handle obsolescence cheaply share one trait: they decided what to do before they had to. A one-page playbook is enough — who receives PCN/PDN notices, how fast a notice must reach the engineering team, who owns the bridge-buy decision, which supplier is first call for EOL sourcing, and what inspection standard applies to open-market lots. Write it once, review it yearly, and every future EOL notice becomes a routine execution instead of a fire drill. If you do not have an EOL supplier of record, make that the first line of the playbook — and put us on it.
Practical tips that save money
- Quote the full ordering code. Suffixes (package, temperature, reel) are where obsolete sourcing goes wrong. The more exact your inquiry, the more exact the match.
- Accept date-code ranges. Requiring a single date code shrinks the available pool dramatically; a two-year window often halves the price.
- Ask about family siblings. A pin-compatible variant with larger memory or a wider temperature range frequently remains available after the exact code vanishes — see our alternatives pages for worked examples.
- Consolidate with your regular BOM. EOL parts ship with the rest of your order from Hong Kong, saving freight and customs overhead — see our BOM sourcing guide.
- Plan the LTB before the deadline. If you see an NRND notice on a part you ship, start the conversation then — not after the last-time-buy window closes.
Have a hard-to-find part number right now? Send it to our desk — one part number is enough, and the answer comes back within 2 working hours.