What the material will not tolerate
Sintered magnet material behaves nothing like the metals most mechanical designers work with. Every retention decision follows from these five properties.
The interference fit trap
The instinctive way to retain a cylindrical component is a press fit into a bore. On a bare magnet this is one of the most reliable ways to destroy it: the hoop stress from an interference fit is tensile, in the direction the material is weakest, applied to a brittle solid with a plated surface that is being scraped during insertion.
Magnets can be captured in a housing, and they can be located by a light slip fit with adhesive. What they should not be is pressed into an interference fit as a bare component. Where a press fit is genuinely required, press a carrier — bond the magnet into a brass, aluminium or steel holder and press the holder.
The retention methods compared
| Method | How it retains | Strengths | Watch for |
|---|---|---|---|
| Adhesive bond only | Shear strength of the bond line | No stress concentration; accommodates tolerance; seals the joint | Temperature and chemical limits of the adhesive; surface prep on plated parts; cure time in the line |
| Mechanical capture (pocket + retaining feature) | Geometry — the magnet physically cannot leave | The most robust option; independent of adhesive performance and ageing | Needs room for a lip, cover or ring; magnet must not rattle |
| Capture plus bond | Geometry carries the load; adhesive prevents movement and seals | The default for demanding applications — belt and braces | Slightly more assembly cost; nothing else |
| Pot / cup assembly | Steel cup surrounds the magnet, bonded inside | Contains flux, concentrates working field, protects the magnet mechanically | Adds height; the cup is part of the magnetic circuit and changes the field |
| Over-moulding | Polymer moulded around the magnet | Excellent capture, environmental sealing, complex geometry in one operation | Moulding temperature and pressure; magnet must survive both; shrinkage stress on a brittle part |
| Carrier and press fit | Magnet bonded into a machined holder; the holder is pressed | Allows press-fit assembly without stressing the magnet; gives a datum for concentricity | Extra part and operation; carrier must be non-magnetic where field matters |
| Clamped between faces | Compressive preload through the magnet | Loads the material in its strong direction | Must be truly flat and parallel; preload changes with thermal cycling |
| Magnetic attraction alone | The magnet holds itself to a steel surface | Zero parts; fine for non-critical, non-vibrating, serviceable applications | Shear resistance is far lower than normal force; slides under vibration; not a design retention method |
The general rule
Let geometry carry the load and let adhesive prevent movement. A bond line that is the only thing between a magnet and a moving assembly is a single point of failure with a temperature limit, a chemical compatibility list and an ageing curve. A pocket with a retaining lip fails only if the housing fails. Where both are present, adhesive degradation becomes a maintenance observation instead of a released magnet.
Designing the pocket
Most retention problems are decided here, and most of them are tolerance problems rather than strength problems.
Clearance and the bond line
- Design a deliberate bond gap. Structural adhesives have an optimum bond line thickness — typically some tens of microns to a few tenths of a millimetre — and both a starved joint and an excessively thick one are weaker. A zero-clearance pocket squeezes the adhesive out during insertion and produces a joint with no adhesive in it.
- Provide adhesive escape. A blind pocket with no vent hydraulically locks: the magnet stops short, sits on trapped adhesive, and the assembly is dimensionally wrong. A small vent hole or a relief groove fixes it.
- Include the coating in the stack. Plating adds material per side, and the drawing may specify dimensions before or after coating. Confirm which — this single ambiguity causes more assembly interference than any other. See tolerances and acceptance criteria.
- Locate on the surfaces that matter. Datum the pocket from the functional face — usually the pole face at the working gap — and take the accumulated tolerance somewhere harmless.
Depth and capture
- Recess the magnet slightly below the housing face where the working gap allows. A magnet proud of its housing takes every impact directly on a brittle edge.
- Break the pocket edges. A sharp entry chamfers the magnet’s plating on the way in.
- Retaining features: a staked or rolled lip, a retaining ring, a cover plate, or a moulded snap. Each must apply load to a face rather than an edge.
- Never rely on a fastener passing through the magnet. If the design appears to need one, it needs a carrier instead.
Material choice for the housing
| Housing material | Magnetic effect | Use when |
|---|---|---|
| Aluminium | Non-magnetic; no circuit effect | Field must not be redirected; weight matters. Note the large expansion mismatch |
| Brass / bronze | Non-magnetic | Precision holders, sensor carriers, good machinability |
| Austenitic stainless (304 / 316) | Essentially non-magnetic | Corrosion resistance without disturbing the circuit |
| Ferritic stainless (430 / 410) | Magnetic — becomes part of the circuit | Deliberate flux return with corrosion resistance |
| Mild steel | Magnetic — acts as back iron | Concentrating field at the working face; containing stray flux |
| Engineering polymer | Non-magnetic | Over-moulded assemblies, cost-sensitive parts, electrical isolation |
The housing is part of the magnetic design
Choosing steel instead of aluminium for a bracket is not a neutral mechanical decision. Steel provides a flux return path, raises the magnet’s permeance coefficient, changes the field at the working gap and alters the stray field around the assembly. Sometimes that is exactly what you want — it is how pot magnets work. What it must not be is accidental. See flux containment.
Thermal mismatch and dynamic loads
Differential expansion
Sintered NdFeB has anisotropic thermal expansion — different along and across the magnetization axis, with one coefficient often very small or negative. Neither matches aluminium, and neither matches steel. Over a wide service temperature range, a rigidly constrained magnet in a metal housing will see stress it did not see at assembly.
- Aluminium housings are the demanding case. Aluminium expands considerably more than the magnet, so a pocket that is a close fit when hot may clamp when cold, or a joint that was sound cold may go slack hot.
- Let the adhesive absorb it. A resilient bond line accommodates differential movement; a rigid, thin, highly filled joint transmits it into the magnet. This is one of the strongest arguments for a designed bond gap rather than a minimal one.
- Model the extremes, not the nominal. Check the fit and the joint stress at both temperature limits of the service profile, including any process excursion such as reflow, paint bake or wash.
- Beware assembly at one temperature and service at another, particularly where a heated or chilled fit is used to ease insertion.
Vibration and shock
- Any clearance becomes an impact. A magnet loose in a pocket under vibration hammers its edges against the housing and chips. Retention must eliminate free movement, not merely prevent escape.
- Adhesive fatigue in shear is the usual long-term failure path. Mechanical capture removes the dependency.
- Mechanical shock rarely demagnetizes modern sintered NdFeB meaningfully — but it readily cracks it. The structural concern is the real one. Alnico is the exception, being genuinely shock-sensitive magnetically.
Rotating assemblies
The demanding case, and the one where retention failure is most consequential.
- Centrifugal load scales with the square of speed. Doubling rotational speed quadruples the retention requirement, so a design validated at one speed is not validated at a higher one.
- Surface-mounted magnets need positive retention — a containment sleeve of carbon fibre or non-magnetic alloy, or a banding. Adhesive alone is not a rotor retention strategy at speed.
- Interior permanent magnet designs place the magnet in a slot in the lamination stack so the steel bridges carry the load — which is one reason the topology dominates high-speed traction motors.
- A containment sleeve is a magnetic air gap. Its thickness comes directly out of the working field, so the mechanical and magnetic designs must converge.
- Balance matters. Magnet mass and position tolerance feed into rotor balance; specify concentricity and positional tolerance against the rotational axis explicitly.
The consequence of getting this wrong
A released magnet in a rotating assembly is not a quality escape; it is a projectile inside a machine, followed by secondary damage as fragments are drawn into every ferrous surface nearby. For rotating designs, retention should be validated by test at overspeed and at the temperature extremes, not by calculation alone.
The assembly process
Design decisions that only reveal themselves on the line. Involving manufacturing engineering before the housing is released saves considerably more than it costs.
Handling magnetized parts
- Feeding is genuinely difficult. Magnetized parts bridge in bowl feeders, cling to chutes and stick to steel tooling. Plan for non-magnetic tooling, controlled presentation, and often manual or semi-automatic placement.
- Insertion force is not what the fit suggests. A magnet approaching a steel housing is pulled in by attraction and can snap into place hard enough to chip. Damped or guided insertion protects the part.
- Polarity is invisible. If a reversed magnet assembles without obstruction, add a poka-yoke — an asymmetric feature, a fixture that only accepts one orientation, or 100% polarity verification at the station. See incoming inspection.
- Keep magnetized parts away from instruments, measurement equipment and any magnetic media at the station.
- Handling injury is a real hazard at larger sizes. Gloves, eye protection and a controlled separation technique belong in the work instruction.
The alternative: magnetize after assembly
Sintered magnets can be assembled unmagnetized and magnetized in a fixture once the assembly is complete. Where it fits, it removes most of the problems above at once.
| Gains | Costs |
|---|---|
| Ordinary handling, feeding and automation | Capital cost of a magnetizer and a fixture per part family |
| No attraction-driven insertion damage | The surrounding assembly must tolerate the magnetizing pulse |
| No polarity errors — orientation is set by the fixture | Fixture design must reach saturation through the assembly |
| Simpler packaging and no air-freight field limits inbound | Verification that full saturation was achieved becomes your responsibility |
| Lower handling injury risk on the line | Adds a process step and a control point |
Saturation of sintered NdFeB requires a pulsed field several times the material’s intrinsic coercivity, which is why this is an industrial capacitor-discharge operation rather than something improvised. It must be agreed at RFQ stage, because the magnet ships in a different state and the fixture becomes part of your process. Related shipping consequences are in shipping magnetized material.
Design review checklist
- Is the magnet loaded in compression, with no tension, bending or point loads?
- Does geometry retain the part if the adhesive fails completely?
- Is there a designed bond gap, and can adhesive and air escape during insertion?
- Does the tolerance stack include coating thickness, on the correct basis?
- Has the fit been checked at both temperature extremes and after any process excursion?
- Is the housing material’s effect on the magnetic circuit intentional?
- Can the part be assembled the wrong way round, and if so what prevents it?
- For rotating assemblies: has retention been validated by test at overspeed and temperature?
- Can the assembly be serviced, or is the magnet captive for life — and is that intended?