The four ways a magnet part disappears
Component obsolescence processes are usually built around semiconductors, where a manufacturer issues a product discontinuance notice with a defined last-order date. Magnets do not work that way. There is often no notice at all, and the four mechanisms below fail differently enough that a single response does not cover them.
| Mechanism | How it presents | Warning | Response |
|---|---|---|---|
| Grade or alloy withdrawn | Supplier offers an "equivalent"; the datasheet no longer lists your grade | Months, if you are reading datasheets | Requalify the substitute against your actual working point |
| Coating discontinued | A plating line closes or a chemistry is restricted; price jumps sharply | Weeks to months | Coating change is usually requalifiable; salt spray and adhesion testing required |
| Supplier exits or is restricted | Acquisition, plant closure, export licence denial, entity listing | Sometimes none | Second source, if one is already qualified; last-time buy if not |
| Tooling lost or worn | Dimensional drift, rising scrap, or a quotation for "new tooling" on a mature part | Visible in inspection data before it is announced | Fund replacement tooling, or take the opportunity to redesign |
The fourth is the one most specific to magnets and the least anticipated. Press dies and sintering fixtures wear, and a die that has run several million pieces produces parts drifting toward the edge of tolerance. The supplier will either absorb replacement quietly or present it as a new tooling charge on a part you have been buying for eight years — and which of those happens depends almost entirely on what your original agreement says about tooling ownership.
When a supplier offers an equivalent grade, they mean the nominal remanence and coercivity fall in the same band. They do not mean identical temperature behaviour, identical corrosion performance, or identical dimensional capability. A grade equivalence table is a starting point for requalification, not a substitute for it — see the grades chart for what the designations actually specify.
Early warning signals worth monitoring
Obsolescence is nearly always visible before it is announced, in signals that already pass through your own systems. The problem is that each one individually looks like something else.
Dimensional drift is the most reliable and the most overlooked, because it lives in inspection data that nobody trends. If your incoming inspection records a measurement rather than a pass, plot it by lot over two years. A monotonic march toward one limit is tooling wear, and it is a forecast rather than an observation.
Put one question in every annual supplier review: which of the parts we buy from you are you least committed to continuing, and what would change that? Suppliers answer this honestly far more often than buyers expect, because the alternative — a surprised customer six months later — is worse for them too. It costs nothing and it is the highest-yield obsolescence control available.
Sizing a last-time buy
A last-time buy converts a supply problem into an inventory problem and a cash problem. Sized too small it merely delays the crisis; too large it writes off working capital against a product that ends earlier than forecast. The structure below is straightforward — the difficulty is in the demand forecast, as always.
D_prod = annual production demand
Y_rem = remaining years of production, from the product roadmap
D_service = annual spares and warranty demand
Y_support = contractual support obligation after end of production
SS = safety stock over the whole window, not one lead time
Q_requal = pieces consumed by qualifying the replacement, if you intend to
Then adjust for the two things that make a raw quantity wrong:
Worked example
A latching magnet consumed at 40,000 pieces per year, with three years of production remaining and a seven-year service obligation running at 3,000 pieces per year. Assembly scrap is 2%. Requalification of an alternative will consume 4,000 pieces.
Including Qrequal means you are buying time to qualify a replacement, which is nearly always the better strategy than buying out the entire remaining life. Buying ten years of a part means storing it, financing it, and betting the roadmap is right. Buying three years and using them to qualify an alternative costs far less and leaves you with a supply chain rather than a warehouse.
How long magnets actually keep
The question every last-time buy raises: will these still be good in six years? For magnets the answer is generally yes, with two specific qualifications that determine how you store them.
The magnetism is not the problem
Sintered NdFeB held below its maximum working temperature and outside an opposing field loses a very small fraction of its magnetization over time. Long-term magnetic ageing is measured in fractions of a percent over years, and for practically every application it is irrelevant next to the temperature effects covered in magnets and temperature. Stored properly, a magnet does not go flat on a shelf.
Corrosion is the problem
NdFeB corrodes readily, and the coating is the only thing preventing it. Over long storage the failure mode is coating degradation — nickel plating with a pinhole, an epoxy layer with an edge defect — progressing to substrate corrosion that propagates along grain boundaries and eventually causes the material to crumble. Humidity accelerates this substantially. The coatings comparison covers which systems tolerate long storage best; nickel-copper-nickel and epoxy over plating both perform well when uncompromised.
SmCo is far more corrosion-resistant and stores essentially indefinitely, though it is brittle and chips easily. Ferrite is chemically stable and stores without difficulty. If your last-time buy is on a samarium cobalt or ceramic part, storage risk is much lower and you can size the buy on demand and cash considerations alone.
Practically: a properly coated, properly packaged NdFeB magnet stored in a dry, stable environment will be fine for the five to seven years a typical last-time buy needs to cover. The failures come from inadequate packaging and uncontrolled humidity, not from the material. Budget for the packaging — it is a small fraction of the inventory value and it is what the buy depends on.
Tooling ownership, and what it is actually worth
Tooling is where obsolescence planning meets contract law, and where most companies discover that a clause they never negotiated determines their options.
| Tool | Purpose | Typically owned by | Transferable in practice |
|---|---|---|---|
| Press die | Forms the aligned green compact | Buyer, if paid for; otherwise supplier | Sometimes — but sized to that supplier's press and shrinkage |
| Sintering fixtures | Support parts through the furnace cycle | Supplier, almost always | Rarely; furnace-specific |
| Machining fixtures | Hold parts for grinding and slicing | Supplier | Rarely worth moving |
| Magnetizing fixture | Defines the magnetization pattern | Buyer, if paid for separately | Yes, if the receiving magnetizer matches — check energy and voltage |
| Inspection gauges | Verify dimensions and output | Buyer if specified | Yes, and worth retaining |
The uncomfortable reality about transfer
A tooling clause giving you title to the press die is worth having, and worth considerably less than it appears. Sintered magnet tooling is dimensioned around a specific producer's powder characteristics, press behaviour and sintering shrinkage — shrinkage running in the region of fifteen to twenty percent, and compensated for in the tool. A die taken to a different producer will frequently need modification or outright replacement to hit the same finished dimensions.
What genuinely transfers is the information: the drawing, the specification, the qualified process parameters, the inspection method and the historical capability data. Magnetizing fixtures and gauges often transfer usefully. The press die mostly does not.
The clause worth fighting for is not ownership of the die. It is the right to receive, on request or on termination, the complete process record — drawings, grade and coating specification, magnetizing parameters, inspection methods and capability history — in a form usable by another producer. That is what shortens a requalification, and it costs the supplier nothing to grant at the point of contracting.
Also fix, at the outset, who pays for tooling replacement at end of life. A die that wears out after six years is a foreseeable event, and the agreement should say whether replacement is at supplier cost, buyer cost, or amortised into the piece price. Leaving it silent means it becomes a negotiation at the worst moment, with a worn tool and no alternative source.
The clauses that prevent the surprise
Every mechanism in this guide is cheaper to handle with notice than without. Notice is contractual, and these are the provisions that produce it. None is exotic and most suppliers will agree to them at contracting, when they cost nothing.
The change-control clause and the named production site do most of the work. An unannounced move of your part to a different plant — within the same supplier, under the same part number — is a common and under-recognised source of trouble, because the new site has different equipment, different process parameters and different capability, while the paperwork suggests nothing changed. Naming the site makes that a contractual event rather than a discovery at incoming inspection.
Discontinuance notices arrive by email to a purchasing address and are frequently read as routine correspondence. Name a recipient in the contract, and route these to whoever owns the obsolescence process rather than to a general mailbox. A twelve-month notice discovered at month eleven is a one-month notice.
The most effective control remains structural rather than contractual. A part with a qualified second source is far less exposed to any of the four mechanisms above, and the qualification has to be funded while the incumbent is still healthy — the second-source qualification guide sets out the realistic timeline and cost. Obsolescence planning and dual sourcing are the same programme approached from different ends, and the parts that justify one generally justify the other.