Where the boundary can actually sit
Make-or-buy on magnets is not one decision. There are four defensible boundary positions, and companies frequently sit at the wrong one because they inherited it from a prototype build rather than chose it.
| Boundary | You receive | You perform | Suits |
|---|---|---|---|
| Finished magnetized part | Magnet, coated, magnetized, inspected | Assembly only | Most programs; simple geometries; low to moderate volume |
| Finished magnetized assembly | Magnet bonded into a cup, rotor or bracket | Install the sub-assembly | Where the assembly step is specialised and volume is moderate |
| Unmagnetized blank | Magnet, coated, inert | Assemble, then magnetize and verify | Higher volume, multipole patterns, tight assembly tolerances |
| Components, full in-house build | Blanks, housings, adhesive | Bond, cure, magnetize, test | High volume with a genuine process advantage; rare |
The interesting boundary is the third one. Buying inert blanks and magnetizing after assembly is not an exotic choice — it is standard practice in motor and sensor manufacturing at volume — but it requires capital, fixtures and process control that a company buying finished parts does not have. The question is when it earns those.
Producing sintered NdFeB from alloy is a metallurgical operation involving strip casting, hydrogen decrepitation, jet milling in inert atmosphere, aligned pressing, vacuum sintering and multi-stage heat treatment. The production guide walks through it. Capital runs to tens of millions and the yield curve takes years to climb. No component manufacturer should consider it, and vertical integration proposals that reach this far are almost always solving a supply security problem that a second source and inventory would solve better.
The case for magnetizing after assembly
The argument for in-house magnetizing is rarely about cost per piece. Magnetizing is a fast operation and the charge for it inside a magnet price is modest. The argument is about everything that becomes easier once the parts arriving at your dock are inert.
Assembly gets dramatically easier
An inert blank can be placed by ordinary automation, held by conventional fixtures, and positioned to a tolerance the adhesive bond line actually requires. A magnetized part jumps, collects swarf, attracts its neighbours in the feeder, and applies force to the fixture that has to be designed against. On multipole rotors and close-pitch arrays the difference is not marginal — some assemblies are extremely difficult to build magnetized and straightforward to build inert.
Scrap and rework improve
Chipping from magnets snapping together during handling is a routine source of scrap, and sintered NdFeB is brittle enough that a collision is often terminal. Removing magnetization from the handling stages removes that mechanism. Rework also becomes possible: a misplaced inert magnet can be repositioned before cure, where a magnetized one frequently cannot.
Magnetic performance can improve
Magnetizing in the assembled state, in the final circuit, produces a more consistent result than magnetizing loose parts and then assembling them, because the steel present during magnetization is the steel present in service. On multipole rings the pattern is defined by the fixture rather than by placement accuracy, which tightens pole spacing and reduces variation.
A magnetizer that produces an adequate result on a bench sample may not fully saturate the same part inside a steel assembly, where flux paths and eddy currents differ. Under-saturated parts pass a casual gauss check and then behave badly — they lose output in service and show poor thermal stability. Establish a saturation curve during qualification by stepping the charge voltage until output stops rising, then run production with meaningful margin above the knee.
Freight and storage get simpler
Inert blanks ship without the field limits that constrain magnetized material by air, which removes a constraint from expedites and from air freight generally — see shipping magnetized material for the thresholds involved. Inert stock also stores without segregation, occupying ordinary racking rather than dedicated space.
What in-house magnetizing actually costs
The capital list is longer than the magnetizer, and the items after it are the ones that get underestimated in the business case.
Fixtures are the item that surprises people. They are part-specific, they are consumable — coils fatigue under repeated discharge and eventually fail — and every new part number needs one. A company with a wide, shallow catalogue of magnet parts can find fixture cost exceeding the magnetizer within two years. A company running three high-volume parts will not.
A magnetizing station stores substantial energy at high voltage and discharges it in milliseconds. The hazards are electrical, mechanical — fixtures can fail violently — and medical, since the field near an active coil is dangerous to anyone with an implanted device. This is not equipment to install without a documented safety programme, trained operators and access control. Follow the practices in the handling and safety guide and treat the exclusion zone as mandatory rather than advisory.
The hidden cost of magnetized inventory
Where the boundary sits determines whether magnetized parts exist in your building at all, and the carrying cost of magnetized stock is consistently underestimated because most of it does not appear on a purchase order.
| Cost | Magnetized stock | Inert blanks |
|---|---|---|
| Storage | Segregated racking, spacing between part families, non-ferrous shelving | Ordinary racking |
| Handling | Trained handling, controlled separation, two-person moves on larger parts | Conventional; automation-friendly |
| Air freight | Constrained by field limits at the package surface; screening and possible declaration | Unrestricted on this basis |
| Foreign object risk | Attracts swarf and debris; contamination of clean areas is a live concern | Minimal |
| Adjacent equipment | Distance from screens, instruments, cards and pacemaker-carrying staff | None |
| Injury | Pinch injuries between larger parts are the most common magnet injury in industry | Negligible |
| Scrap handling | Rejected magnetized parts require segregated disposal or demagnetization | Straightforward |
None of this argues against buying magnetized parts — the overwhelming majority of magnet users do, correctly, because at their volume the capital and process burden of magnetizing in house is not repaid. It argues for counting these costs when the volume is high enough that the comparison is close, because they systematically favour the inert side and they are systematically left out of the model.
Companies that magnetize in house retain air freight as a genuine expedite option on the inbound leg. Companies buying magnetized parts frequently discover, during their first real shortage, that the fast option they were counting on is constrained. If your supply plan assumes air expedites, check that assumption against the actual field limits before you rely on it.
Outsourced bonding, potting and over-mould
The other half of the boundary question is the assembly operation itself — bonding magnets into cups, rotors, brackets or housings. Here the calculus differs from magnetizing, because the capital is modest and what you are really buying is process qualification.
| Operation | Keep in house when | Outsource when |
|---|---|---|
| Adhesive bonding | You already control adhesive processes and have cure capacity | The bond is structural, qualified, and you have no surface-prep control |
| Potting and encapsulation | Volume is high and the geometry is stable | Tooling is specialised or the resin system needs vacuum degassing |
| Over-moulding | You have injection capability and the magnet survives the cycle | Nearly always — mould tooling and magnet thermal exposure are both specialist problems |
| Press-fit and retention | Assembly is straightforward and gauging is in place | Interference and hoop stress need analysis you would rather buy |
| Final magnetizing | Volume clears the breakeven and parts are few | Part variety is wide or volumes are modest |
Adhesive bonding is the operation most often brought in house on optimistic assumptions. The adhesive itself is cheap; the qualified process is not. Surface preparation, cure profile control, bond-line thickness, and validation across the temperature and humidity range the product will see are all real work, and a bond that fails in service on a rotating assembly fails destructively. The bonding and mounting guide and assembly and retention design set out what a defensible process involves — read them before deciding this is a simple operation.
On rotors, retention should be mechanical — a sleeve, a pocket, a retaining ring — with adhesive doing the positioning rather than carrying the load. Any make-or-buy analysis that assumes bonding is the whole retention strategy is understating both the risk and the cost of the in-house option.
The decision framework
Reduce the comparison to an annualised total for each boundary option, then find the volume at which they cross. The formula below is deliberately simple; its value is in forcing every term to be stated rather than assumed.
Annualmake = V × ( Pblank + cop + s ) + K/n + F + M
V = annual volume in pieces
P = piece price, magnetized or inert
h = per-piece handling premium on magnetized stock
c_op = magnetizing labour, energy and verification per piece
s = incremental scrap cost of the magnetizing operation
K/n = magnetizer capital amortised over n years
F = annual fixture cost, per part geometry
M = maintenance, calibration, training and compliance
Setting the two equal and solving for V gives the breakeven volume. Two things usually emerge. The per-piece saving is smaller than expected, because the magnetizing charge inside a magnet price is not large. And the fixed terms are larger than expected, because F recurs per part number and M does not go away.
Run the model at half and double your volume forecast before committing capital. A breakeven that only clears at the optimistic case is a decision to buy. Your should-cost model supplies the piece-price inputs; use the same figures in both so the comparison is internally consistent.
The answer most companies should land on
Stated as a general rule, with all the caveats that implies: buy finished magnetized parts unless you have concentrated volume on a small number of part numbers and a specific assembly problem that magnetization is causing.
The hybrid position is common and sensible: magnetize in house the two or three high-volume parts where assembly genuinely benefits, and buy everything else finished. It concentrates the fixed cost where volume repays it and avoids proliferating fixtures across a long tail.
One caution on the reasoning that most often drives these decisions. Vertical integration is frequently proposed as a supply security measure, and on magnets it does not deliver that. Magnetizing in house still leaves you buying sintered blanks from the same constrained supply base, with the same origin exposure and the same export-control considerations. It changes where the operation happens, not where the material comes from. If security is the objective, a qualified second source and a sized inventory position address it directly; moving the magnetizing step does not.
Boundary decisions get made at prototype stage, when volumes are tiny and buying finished parts is obviously right, and then never revisited as the programme scales. Put the make-or-buy model on the annual review alongside your planning parameters — the volume at which the answer flips will arrive without announcing itself.