Mistakes

Common Mistakes in Rare Earth Magnet Manufacturing and How to Fix Them

The nine most expensive errors in rare earth magnet production, each with the symptom on the floor, the root cause, and a fix you can verify with a number.

Most scrap and margin loss in sintered NdFeB magnet production traces to the same handful of repeatable errors. A plant running 90 percent through yield against an achievable 96 percent loses roughly 600 kg of alloy per month on a 10 ton per month line, and at 60 to 80 dollars per kg for NdPr bearing alloy that is 36,000 to 48,000 dollars of metal before any labor or furnace time. Each mistake below gets three things: the symptom you see on the floor or in the ledger, the root cause, and a fix with a number attached so you can verify the correction actually worked.

Mistake one is applying a single linear shrinkage factor to an anisotropic material. Die pressed NdFeB shrinks about 17 to 20 percent parallel to the alignment direction and only 13 to 15 percent transverse, so one averaged factor puts you out of tolerance in one axis while the other measures fine. Symptom: sintered blocks consistently long or short in the same direction, batch after batch. A 2 point shrinkage error on a 50 mm green dimension is a 1 mm miss, which either scraps the block or forces extra grind stock. Fix: measure shrinkage per axis on every alloy lot and press orientation, then set die compensation with the Sintering Shrinkage calculator instead of a plant wide average.

Mistake two is planning material as if powder yield were 100 percent. Between hydrogen decrepitation, jet milling, cyclone holdup, and rejection of oxidized fines below about 1 micron, realistic yield from strip cast alloy to pressed green compact is 93 to 97 percent. Symptom: alloy purchases run 5 to 10 percent above the bill of materials every quarter and nobody can say where the metal went. Root cause: losses booked as overhead instead of tracked per batch. Fix: log every milling campaign in the Powder Yield calculator, set a control limit at 94 percent, and investigate any batch below it the same day, since a 1 point yield drop on 10 tons is 100 kg of alloy.

Mistake three is quoting finished parts from sintered weight without modeling kerf and grind stock. Slicing 2 mm thick segments with a 0.3 mm blade throws away 13 percent of the block in kerf alone, and total machining loss on thin small magnets commonly reaches 25 to 40 percent of sintered mass. Symptom: finished kg shipped per kg sintered sits near 0.6 when the quote assumed 0.85. Fix: lay out block nesting and blade selection in the Grinding Loss calculator before quoting. Switching from a 0.3 mm blade to 0.15 mm multi wire slicing on 2 mm parts cuts kerf loss roughly in half, worth about 6 points of material yield.

Mistake four is treating grinding sludge and solid scrap as waste. Swarf from NdFeB machining carries 28 to 32 weight percent rare earth content, and with NdPr oxide trading between 50 and 80 dollars per kg over recent years, dewatered sludge is often worth 8 to 15 dollars per kg to a recycler, not the 1 to 3 dollars per kg many plants accept on a flat rate hauling contract. Solid scrap, cracked blocks and out of tolerance parts, fetches 30 to 50 percent of virgin alloy value. Fix: price every stream with the Scrap Recovery Value calculator and index scrap contracts to a published NdPr oxide price rather than a fixed number.

Mistake five is undersizing the magnetizer for high coercivity grades. Full saturation needs a field of roughly 2 to 2.5 times intrinsic coercivity, so an SH grade with Hcj of 20 kOe wants about a 4 T pulse, and a fixture sized for N42 at 12 kOe will leave it partially magnetized. Symptom: Helmholtz coil flux reads 2 to 4 percent low, but only on SH, UH, and EH grades, and only on that fixture. Fix: check every grade and fixture pairing in the Magnetization Capacity calculator, then verify saturation by stepping charge voltage upward until measured flux plateaus within 0.5 percent between steps.

Mistake six is accepting chronic plating rejects as normal. NiCuNi coating rejects of 5 to 10 percent for edge coverage and blistering usually trace to sharp edges below a 0.2 mm radius and overloaded barrels, not to the plater. Adding a 0.2 to 0.4 mm chamfer at grinding typically pulls coating rejects under 2 percent, and the Coating Yield calculator shows whether rework or scrap is the cheaper disposition. A related quoting error is pricing coated parts off a rare earth price that is months old; run the Rare Earth Cost Sensitivity calculator whenever NdPr moves more than 10 percent so the mistake surfaces before the purchase order does.

Mistake seven is letting final inspection set the line rate. If grinding delivers 600 pieces per hour and 100 percent flux and dimensional checking runs 400 per hour, WIP grows by 200 pieces every hour and lead time quietly stretches by days. Size gauge count with the Inspection Bottleneck calculator; two gauges at 400 per hour clear the line with 33 percent headroom. Mistake eight is running vacuum sintering furnaces half loaded. A 12 to 20 hour cycle costs nearly the same at 60 percent load as at 95 percent, so energy per kg roughly doubles when the load halves. Batch scheduling against the Furnace Utilization calculator recovers most of that loss.

Mistake nine is single sourcing heavy rare earths and calling it a savings. Around 90 percent of separated dysprosium and terbium supply comes from one country, and the 2025 export licensing episode turned routine purchase orders into 45 to 90 day waits for plants with no buffer. Symptom: a force majeure or allocation letter arrives and your Dy bearing grades stop overnight. Fix: qualify a second alloy or magnet source before you need it, hold 8 to 12 weeks of Dy and Tb exposure in inventory, and score each vendor quarterly in the Supplier Risk calculator so concentration shows up as a number, not a surprise.

Published 2026-07-02.