Shielding redirects, it does not block
Magnetic flux forms closed loops. It cannot be absorbed or terminated — every line leaving a north pole returns to a south pole. A shield works by providing a low-reluctance route that carries the flux around the region you want protected.
The relevant material property is relative permeability (µr): how much more readily a material carries flux than air does. Air has a µr of 1. Ordinary steel is in the hundreds to low thousands. Mu-metal, annealed, can reach the tens of thousands at low field levels. Flux takes the easy path in almost exactly the way current takes the low-resistance path in an electrical circuit — and the same intuition applies, including the failure modes.
The consequence engineers most often miss
Because a shield carries flux rather than destroying it, adding shielding changes the magnet’s operating point. Enclosing a magnet in steel lowers the reluctance of its circuit, raising the permeance coefficient and moving the working point up the demagnetization curve — which is beneficial for demagnetization resistance but also alters the field at every point in the design. A shield added late to fix a stray field problem will change the field your sensor sees. Shield and magnet have to be designed as one circuit.
Materials and their limits
Two properties govern selection, and they trade against each other: permeability determines how well the material attracts flux, and saturation flux density determines how much it can carry before it stops behaving like a shield at all.
| Material | Relative permeability | Saturation (approx.) | Best used for |
|---|---|---|---|
| Mu-metal / high-nickel permalloy | 20,000–100,000+ | ~0.7–0.8 T | Weak fields, precision instruments, the inner layer of a nested shield |
| Nickel-iron, moderate nickel | 5,000–30,000 | ~1.0–1.5 T | General instrument shielding with more headroom than mu-metal |
| Silicon steel | 2,000–7,000 | ~1.8–2.0 T | Motor and transformer circuits; laminated for AC |
| Low-carbon steel / soft iron | 200–2,000 | ~2.0–2.1 T | Strong fields, back iron, yokes, pot magnets, shipping shields |
| Cobalt-iron | 2,000–10,000 | ~2.3 T | Where maximum flux-carrying capacity in minimum section is worth the cost |
| Stainless 304 / 316 | ~1 (non-magnetic) | — | Structure, not shielding — austenitic stainless is not a shield |
| Aluminium, copper, brass | ~1 | — | No static shielding at all; useful for eddy-current damping of changing fields |
Saturation is the failure mode that surprises people
Mu-metal has spectacular permeability and a low saturation point. Placed close to a strong NdFeB magnet, it saturates — and a saturated shield has an effective permeability approaching that of air. It stops shielding, completely, and the failure is not gradual.
This is why the intuitive move of wrapping a strong magnet in the highest-permeability material available frequently performs worse than plain mild steel. For strong fields, choose on saturation first and permeability second. Mu-metal belongs where fields are weak — instrument enclosures, magnetometer shields, the inner layer behind something else.
Two properties that matter in practice and are usually forgotten
- Permeability is destroyed by cold work. Bending, stamping, shearing and even dropping a mu-metal shield can reduce its permeability dramatically. High-permeability alloys must be annealed after forming, in a controlled atmosphere, and handled carefully thereafter. A shield fabricated and installed without post-forming anneal may deliver a fraction of its datasheet performance.
- Permeability is not constant. It varies with the field level, so a datasheet figure quoted at a low field says little about behaviour near a strong magnet. Design against the curve, not the headline number.
The four design approaches
In rough order of what to try first.
1 — Distance
The cheapest shield is space. Field from a compact magnet falls very steeply with distance — roughly with the cube of distance in the far field for a dipole. Doubling the separation between a magnet and a sensitive component often achieves more than any practical shield, at zero material cost and zero weight. Exhaust this before designing hardware.
2 — Close the magnet’s own circuit
Give the magnet a deliberate low-reluctance return path so its flux never enters the surrounding space in the first place. This is what a pot magnet does, and what back iron does in a motor or a holding assembly:
- Back iron or yoke behind the magnet returns flux and concentrates it at the working face.
- Cup or pot construction surrounds the magnet, containing the field except at the working gap — which is why a pot magnet holds harder than the bare magnet inside it and leaks far less field.
- Keeper plates across the poles close the circuit completely for storage and transport.
- Halbach arrangements use the magnets themselves to cancel field on one side and reinforce it on the other, with no ferrous material at all.
This is nearly always the better engineering answer than adding a shield around a leaky assembly: it improves the useful field and reduces the stray field simultaneously. The mechanical side of building these structures — pocket geometry, retention and the housing material decision — is in magnet assembly and retention design.
3 — Shield the victim, not the source
Where the sensitive item is small and the field source is large or unavoidable, enclosing the victim is more material-efficient. This is the standard approach for magnetometers, photomultipliers, CRTs and precision instruments. Two rules govern it:
- Enclosure completeness dominates. Flux enters through openings. A shield with a large aperture performs far worse than its material would suggest, and seams, joints and fastener holes all leak. Overlapping joints outperform butt joints.
- Aspect ratio matters. A deep enclosure with a small opening shields well; a shallow tray shields poorly.
4 — Nested shields
For demanding attenuation, use multiple layers separated by air or a non-magnetic spacer, with material chosen by position:
- Outer layer: high saturation — mild steel or silicon steel. It takes the brunt of the field and drops it to a level the next layer can handle without saturating.
- Inner layer: high permeability — mu-metal. Now operating in the weak-field region where it excels.
- The gap between layers is functional, not incidental. Layers in direct contact behave much more like a single thicker shield and lose most of the benefit.
Nesting works because attenuation compounds across layers, and because each layer places the next in a field regime suited to its material.
Practical design guidance
| Design question | Guidance |
|---|---|
| How thick should the shield be? | Thick enough that the flux it must carry stays below saturation. Estimate the total flux from the magnet, divide by the shield’s cross-sectional area, compare against the material’s saturation flux density, and add margin. Thickness is a saturation calculation, not a rule of thumb. |
| Where should it be placed? | Close to the source contains flux most efficiently but changes the magnet’s operating point most. Close to the victim leaves the source circuit undisturbed. Decide which constraint dominates before choosing. |
| Does it need to be continuous? | Yes, as far as practical. Joints, seams and holes are the dominant leakage paths. Overlap joints; keep apertures small and deep rather than large and shallow. |
| Will stainless steel work? | Austenitic grades such as 304 and 316 are essentially non-magnetic and provide no shielding. Ferritic and martensitic grades such as 430 and 410 are magnetic and will shield. Confirm the grade, not the word “steel”. |
| Will aluminium work? | Not for static fields. It provides eddy-current damping of changing fields, which is a different requirement and worth distinguishing clearly in the specification. |
| What about temperature? | Permeability varies with temperature and collapses above the material’s Curie point. Rarely an issue at ordinary service temperatures; check for hot applications. |
| Does the shield need protection? | Mild steel and soft iron corrode. Plate or coat them, and account for the coating in the magnetic gap — a non-magnetic coating between magnet and back iron is an air gap. |
Common mistakes
- Mu-metal against a strong magnet. It saturates and stops working. Mild steel almost always outperforms it in that position.
- Ignoring the effect on the magnet. Adding steel raises the permeance coefficient and changes the working field everywhere in the design.
- Assuming stainless is a shield. It usually is not.
- Forming mu-metal and not annealing it. The performance loss can be an order of magnitude.
- Perforating the shield for fasteners or cable entries and not accounting for the leakage.
- Shielding to a calculated figure and never measuring. Shielding is very sensitive to geometry and construction detail; predicted attenuation and achieved attenuation frequently differ.
Specifying and verifying the result
Shielding requirements written as adjectives cannot be tested. Write them as fields at points.
Where a requirement usually comes from
- A sensitive component’s datasheet limit — magnetometers, Hall devices, imaging sensors, some displays.
- Medical device proximity requirements, particularly minimum distances for implanted cardiac devices.
- Workplace exposure limits where personnel work near strong fields.
- Air freight limits — the IATA magnetized material thresholds are, in effect, a shielding specification applied to a package. The measurement method and the limits are in shipping magnetized material.
- Adjacent-assembly interference in dense equipment, where two magnetic subsystems affect each other.
Where simulation earns its cost
Analytical estimates handle simple closed geometries reasonably. They handle apertures, seams, saturation and nested layers poorly — which is to say they handle real shields poorly. Finite element analysis is genuinely worthwhile for shielding work because the non-linearity that matters most, saturation, is exactly what hand calculation cannot capture. Model the assembly as built, including joints and holes, then verify on hardware. Our engineering team runs FEA-backed analysis for customer assemblies where the requirement warrants it.