Magnet Buyers Guide for Engineers
A Design-and-Specification Guide for Engineers Working With Magnets
This guide walks through the decisions an engineer actually makes when a magnet enters a design — from translating a force or field requirement into a spec, through material and tolerance trade-offs, assembly and thermal design, validation before production, and managing the design through to a shipped part. Where a topic needs real engineering depth — grade charts, tolerance limits, temperature derating, magnetization physics — this guide points to the dedicated reference rather than repeating it here.
How to use this guide
Read it in order if you're speccing a magnet into a new design for the first time — each part builds on the one before it. If you already know which stage you're at, use the sidebar or search to jump straight there. 18 chapters so far, with more parts added as the guide grows.
Part 1
Specifying the Magnet
Part 2
Designing the Assembly
Part 3
Validating Before Production
Part 4
Prototype to Production
Engineers Guide / Specifying the Magnet
Specifying the Magnet
From Design Requirement to Magnet Spec
7 min read
Chapter 1 of 4 — Specifying the Magnet
A design requirement usually arrives as a performance number — hold 3 lbs across a 2mm gap, sense a target at 15mm, transmit 2 N·m through a sealed wall. None of those is a magnet spec yet. The job of this chapter is the method for closing that gap: working backward from the physics to the handful of parameters a supplier actually needs.
Start from the working point, not the material
Every magnetic circuit has an operating point — the combination of flux density and field strength the magnet actually sees once it's installed in its air gap, steel circuit, or free-air mount. Picking a grade before establishing the operating point is backward: the same grade can be wildly overkill in a closed circuit and inadequate in an open one. The permeance coefficient of the installed geometry — set by the air gap, the steel return path (if any), and the magnet's own aspect ratio — is what actually determines where on the material's demagnetization curve the magnet operates.
The translation, step by step
- State the requirement as a field or force at a defined point. "Strong enough" isn't a spec; "holds 3 lbs across a 2mm air gap against a 1/8" steel plate" is.
- Establish the magnetic circuit. Open air, against a steel return path, or fully enclosed — this sets the permeance coefficient and therefore which part of the demagnetization curve is in play.
- Size the geometry against the requirement, using a pull-force or field-at-distance calculation rather than a guess — a size selector tool built around the actual physics gets you a real starting geometry, not a rule of thumb.
- Set the temperature ceiling before finalizing grade — this is often the constraint that actually picks the material family, more than raw field strength does (see the next chapter).
- Confirm the result against the original requirement with margin — real-world pull force runs lower than an idealized calculation due to surface finish, perpendicularity, and steel saturation; build in margin rather than specifying to the exact calculated minimum.
Why order matters here: engineers who pick a grade first and then check if it meets the requirement tend to over-specify — reaching for a higher grade than the circuit actually calls for, which drives cost and can even hurt performance if it worsens temperature margin for no functional gain. Starting from the operating point and sizing up from there almost always produces a cheaper, equally functional spec.
What to hand off once the math is done
The output of this process is the same seven-field spec used throughout magnet procurement — material/grade, dimensions and tolerances, magnetization direction, coating, max operating temperature, quantity, and need-by date — but arrived at through engineering derivation rather than assumption. The next three chapters in this part cover the decisions inside that spec that carry the most engineering weight: material family, tolerance, and magnetization pattern.
Engineers Guide / Specifying the Magnet
Specifying the Magnet
Choosing a Material Family: NdFeB vs. SmCo vs. Ceramic vs. Alnico
7 min read
Chapter 2 of 4 — Specifying the Magnet
Every material comparison chart shows NdFeB winning on raw field strength — which is true, and also the wrong basis for a lot of real design decisions. The actual selection question is which combination of temperature ceiling, coercivity, cost, and environmental resistance the application needs, and NdFeB isn't the answer to all four at once.
The four axes that actually matter
| Material | Where it wins | Where it loses |
|---|
| NdFeB (sintered) | Highest energy product for a given volume — the default when size and weight are constrained | Lowest standard temperature ceiling of the four (though high-temp grades push this up), needs a corrosion coating, most exposed to rare-earth supply volatility |
| SmCo | High coercivity at high temperature, inherently corrosion-resistant, stable to roughly 300–350°C in the right grade | More brittle, higher material cost, lower maximum energy product than top NdFeB grades |
| Ceramic / ferrite | Lowest cost by a wide margin, inherently corrosion-resistant, no coating needed | Lowest field strength of the four — needs more volume to hit the same force, and more brittle than it looks |
| Alnico | Highest temperature stability of the common families (usable well above 500°C), very stable output over temperature swings | Low coercivity — easily demagnetized by an opposing field, which rules it out for many sensor and motor applications despite the temperature advantage |
The question that actually picks the family
In practice, one axis usually dominates the decision faster than a full trade study: if the application exceeds roughly 150°C, NdFeB is likely out and the choice narrows to SmCo or Alnico depending on whether coercivity or absolute temperature ceiling matters more. If cost and size are unconstrained but the mounting will see repeated impacts or knocks, ceramic's brittleness may rule it out even at its lower price. If size is genuinely constrained and temperature stays moderate, NdFeB is very often the right and obvious answer — which is why it dominates consumer and most industrial applications; the exceptions above are exactly that, exceptions worth checking for deliberately.
Coercivity and remanence are not the same lever. A higher-energy-product NdFeB grade (N52 vs. N35) means more flux for the same volume, but typically lower coercivity — it demagnetizes more easily under an opposing field or elevated temperature. For an application with vibration, opposing fields, or moderate heat, a lower-energy grade with higher coercivity can genuinely outperform the "stronger" one over the product's life. Evaluate both together, not energy product alone.
Where to go next
Once a material family is chosen, the grade chart linked below gets specific about which grade within that family fits a given temperature ceiling. The temperature-and-design-rules guide covers the load-line effect in more depth than fits here — worth reading in full for anything running above roughly 60°C, since geometry and temperature interact in ways a headline temperature rating alone doesn't capture.
Engineers Guide / Specifying the Magnet
Specifying the Magnet
Tolerancing a Magnetic Design: Deciding What Actually Needs to Be Tight
6 min read
Chapter 3 of 4 — Specifying the Magnet
Tolerance is the cost lever engineers control most directly and use most carelessly. A dimension inherited from a CAD default, or tightened "to be safe," turns into a real cost and lead-time number the moment it hits a supplier's quote — and the fix is a design-stage decision, not a negotiation after the fact.
Two different kinds of tolerance, and they're not optional in the same way
Dimensional tolerance (length, diameter, flatness) and magnetic property tolerance (Br, Hcj spread across a grade) are governed by different physics and don't move together. A grade like N42 defines a Br range that's inherently several percent wide before any manufacturing variation is added — that spread is a property of the material specification itself, not something a tighter dimensional callout affects. A design that can't tolerate ordinary lot-to-lot field variation needs either a design that's insensitive to it, or a tighter magnetic-property acceptance criterion specified and tested for — which is a different (and more expensive) ask than a tight dimensional tolerance.
The design-stage question that actually matters
| Ask this | Because |
|---|
| Does this dimension control a press fit or interference fit? | If yes, the tolerance is functional and probably needs to be tight; if the magnet just needs to fit loosely into a pocket, it likely doesn't |
| Does this feature stack with other tolerances in the assembly? | A tolerance that's fine in isolation can become the tightest link in a stack-up — check the assembly, not just the part |
| Is this the CAD default, or a value someone actually calculated? | Defaults inherited from an unrelated part or software preset are the single most common source of unnecessarily tight callouts |
| What does the acceptance criterion actually get measured against? | A criterion with no defined measurement method or fixture invites disputes later — anchor it to real hardware, not a number derived from theory alone |
Anchor limits to real hardware, not theoretical ideals. The most defensible acceptance criteria are set by measuring approved first-article samples in the actual production test fixture and building limits around those measured values — not values derived purely from a drawing or a datasheet. Limits set from theory alone tend to produce fixture-to-fixture disputes down the line that limits anchored to retained reference hardware don't.
Where this saves real money
Relaxing a tolerance that isn't functionally load-bearing is very often a larger, faster cost reduction than negotiating price on an unchanged spec — because tight tolerance frequently means an added grinding or measurement step, not just a stricter inspection of the same process. The dedicated tolerances guide below goes deep into what's achievable at standard versus tight tolerance and what each actually costs; this chapter's job was only to flag which tolerances are worth that conversation in the first place.
Engineers Guide / Specifying the Magnet
Specifying the Magnet
Getting Magnetization Direction and Pole Pattern Right
6 min read
Chapter 4 of 4 — Specifying the Magnet
Magnetization direction isn't a property that gets added to a finished part — it's set during manufacturing, in a fixture, and changing it after the fact generally means remaking the part rather than reworking it. That makes it one of the few specs in a magnet design that's genuinely irreversible once tooling and fixturing are committed, which is exactly why it needs to be right before the RFQ goes out, not discovered wrong at first article.
The patterns, and what each is actually for
| Pattern | Typical use |
|---|
| Axial | Field runs through the thickness — the default for most disc, block, and simple ring geometries |
| Diametric | Field runs across the diameter of a cylinder or ring — common in rotary position sensing where a single sine/cosine field needs to be read as the part rotates |
| Radial (true radial) | Field runs outward (or inward) from the center at every point around a ring — used where a uniform radial field matters around the full circumference |
| Multipole | Multiple alternating poles around a ring or along a length — used for encoders and commutation, where pole count and pole-to-pole uniformity are themselves part of the spec |
Why encoder and commutation applications need extra care
For a multipole ring feeding a sensor, pole position relative to a mechanical reference — a keyway, a flat, a hub feature — is often the entire point of the design, not an incidental detail. Magnetizing a pre-formed, already-round ring can't hit that precision by hand-orienting it; the practical answer is usually bonding the ring unmagnetized and magnetizing the completed sub-assembly in a fixture that indexes directly to the mechanical datum. That's a process decision with assembly implications, and it has to be agreed before tooling exists — raising it at first article is too late.
Name the axis, not just the word "radial" or "axial." "Magnetized through the middle" or even "radially magnetized" can describe more than one geometry depending on the part. Specify the axis explicitly relative to a named feature — "through the thickness," "across the diameter, poles aligned to the flat" — so there's no ambiguity for a supplier building a fixture from the drawing alone.
What belongs on the drawing
- The magnetization pattern named unambiguously, referenced to a physical feature (flat, keyway, hub) where orientation matters
- Pole count and, for multipole parts, pole-to-pole uniformity tolerance if the sensor application is sensitive to it
- Whether the part ships magnetized or unmagnetized — and if unmagnetized, who owns the magnetizing fixture and process step
- The sensor or mating part number, if one exists — it's often the fastest way for a supplier to catch a pole-pitch mismatch before tooling is cut
Engineers Guide / Designing the Assembly
Designing the Assembly
Designing for Manufacturability: What Makes a Magnet Hard to Make
6 min read
Chapter 1 of 4 — Designing the Assembly
A magnet doesn't behave like the machined metal parts it usually sits next to in an assembly. Sintered material is hard and brittle rather than tough and ductile, it's typically magnetized before the part reaches its final process steps, and post-process modification that would be routine on a metal bracket — grinding a corner, drilling a hole — is often not possible at all. Designing around these constraints, rather than discovering them at first article, is what DFM means for a magnet.
Why "just modify it after" usually doesn't work
Cutting, grinding, or machining a finished sintered magnet risks cracking or shattering the brittle material, and — for a magnetized part — physically disrupting the material can locally scramble the domain alignment that produces its field, degrading performance in ways that aren't always visible by inspection. The practical takeaway: features that need machining have to be decided before sintering and often before magnetizing, not treated as a fixable afterthought.
Geometry choices that are easy versus hard to manufacture
| Easier to manufacture | Harder to manufacture |
|---|
| Simple discs, blocks, rings with generous wall thickness | Thin walls or sharp internal corners — brittle material concentrates stress at corners and can crack during sintering, cooling, or handling |
| Standard aspect ratios close to catalog geometry | Extreme aspect ratios (very thin and wide, or very long and thin) — these are both harder to press and more prone to handling damage |
| Features built into the pressing/sintering process | Features that require post-sinter machining (holes, chamfers, slots) — possible, but adds cost and a scrap-risk step |
| A single, unambiguous magnetization direction | Magnetization patterns that require a custom fixture — still very doable, but a real NRE and lead-time factor, not a checkbox |
Ask "how does this actually get made" before finalizing geometry. A quick conversation with a supplier's engineering team at the concept stage — before a drawing is locked — is the cheapest place to catch a feature that will be expensive or impossible to produce as drawn. This is a very different conversation than an RFQ clarification after the fact, and it costs nothing but a few minutes.
Where this connects to cost
Manufacturability and cost aren't separate conversations — a feature that's hard to make is, almost by definition, an expensive one, whether that shows up as a machining step, a lower yield from cracked parts, or a longer, more complex tooling design. The should-cost modeling guide linked below breaks down exactly which process steps scale with which design choices, for anyone wanting to go deeper than the qualitative guidance here.
Engineers Guide / Designing the Assembly
Designing the Assembly
Retention & Mounting: Choosing How the Magnet Stays Put
6 min read
Chapter 2 of 4 — Designing the Assembly
A magnet that performs perfectly in isolation still has to stay exactly where it's put — through vibration, thermal cycling, and the attractive forces it generates against its own steel circuit. Retention is a real design discipline, not an afterthought once the magnetic circuit works on paper.
Why press fits are usually the wrong default
Never press a sintered magnet directly onto a shaft or into a bore as the sole retention method. The hoop stress from an interference fit is tensile, and sintered rare-earth material is weak in tension — it cracks, sometimes immediately and sometimes only after it's already in service. This is one of the most common ring- and sleeve-magnet failure modes, and it's avoidable at the design stage.
The three real retention strategies
| Method | When it fits |
|---|
| Bonding with a controlled glue line | The default for most cylindrical and planar mounting — a designed gap (commonly a few hundredths of a millimeter) both retains the part and accommodates thermal expansion mismatch between magnet and mating material |
| Bonding into a non-magnetic carrier, then mounting the carrier conventionally | Where real mechanical load exists — the carrier takes the load, and the magnet only has to be a magnet, not a structural member |
| Mechanical capture (pocket, lip, sleeve, band) | Rotating machinery, overhead mounting, or anywhere near people — mechanical capture means an adhesive failure degrades performance instead of releasing a projectile, which matters wherever safety is a factor |
Designing the bond joint itself
The adhesive only bonds as well as the surface it's applied to — for a coated magnet, that means the joint is only as strong as the coating's adhesion to the substrate beneath it, not the adhesive's rated strength on a clean metal surface. Attractive forces during assembly also load the joint before it's even cured, slamming parts together and sometimes shifting them mid-cure — a real assembly-process consideration, not just a bonding-strength number on a datasheet. For anything in rotating or cyclic service, add axial location (a shoulder, circlip, or end cap) so the part stays positioned even if the bond creeps at temperature over the product's life.
What to decide at the design stage, not the assembly stage
- Which of the three strategies fits the load case — don't default to bonding alone just because it's simplest to draw
- Whether the application is close to people or rotating equipment, which pushes toward mechanical capture regardless of what the adhesive alone could handle
- The glue line dimension, specified on the drawing rather than left to assembly discretion
- Whether thermal cycling in service will stress the bond over time, and whether axial/positional backup is warranted
Engineers Guide / Designing the Assembly
Designing the Assembly
Thermal Design: Keeping the Magnet Inside Its Operating Window
6 min read
Chapter 3 of 4 — Designing the Assembly
"This magnet is rated to 150°C" is a real number and also not the whole answer — because the same material, in a different geometry, can safely exceed that rating in one installation and fail well below it in another. The missing variable is the magnetic circuit the part is actually installed in.
The variable most engineers skip: permeance coefficient
A magnet's operating point on its demagnetization curve is set by the permeance coefficient of its installed geometry — determined by the air gap, whether there's a steel return path, and the magnet's own aspect ratio. A thin, wide magnet in open air has a low permeance coefficient: it sits near the "knee" of its demagnetization curve and can suffer irreversible loss well below its catalog temperature rating. The same material in a long geometry, or backed by a steel return path, has a high permeance coefficient and can safely run hotter than the headline number suggests. Evaluating temperature and geometry together — not temperature alone — is the actual design rule.
Reversible versus irreversible loss
| Type | What happens | Design implication |
|---|
| Reversible loss | Field drops as temperature rises, and recovers fully on cooling | Expected and normal — budget for it in the performance calculation at the hottest expected operating temperature, not at room temperature |
| Irreversible loss | Some of the field loss remains even after cooling back down | Occurs when the operating point drops below the knee of the curve — this is the failure mode that thermal design is actually trying to prevent |
The practical design workflow
- Establish the real thermal profile the part sees in service — not a nameplate ambient temperature, but the actual worst-case temperature at the magnet, which can be considerably hotter than ambient near a motor winding or under a hood.
- Model the magnetic circuit — air gap, steel backing or none, geometry — to establish the permeance coefficient.
- Request a load-line check from the supplier's engineering team using the actual drawing, circuit, and thermal profile — this is a standard calculation, not a special favor, and it's far cheaper than a field failure.
- If margin is thin, the fix is usually geometry or material family, not just a "hotter-rated" version of the same grade — see the material-family chapter earlier in this guide.
A 20mm × 1mm disc rated "80°C" is not the same part, thermally, as the same material at a 20mm × 10mm aspect ratio. The catalog rating describes the material; the permeance coefficient of your specific geometry decides whether that rating actually applies to your part in your application.
Engineers Guide / Designing the Assembly
Designing the Assembly
Designing for a Sensor or Motor Application
6 min read
Chapter 4 of 4 — Designing the Assembly
A holding or lifting magnet has one job: generate enough force at the working point. A sensor or motor magnet has a second, often harder job on top of that — produce a field shape that a sensor or winding can read accurately, consistently, across the part's whole operating life. That second job adds spec requirements that don't show up in a basic pull-force calculation at all.
What a sensor or motor application adds to the standard spec
| Additional requirement | Why it's there |
|---|
| Field at the working point, with tolerance | A Hall-effect or similar sensor reads absolute field strength at a defined distance — this needs its own tolerance and temperature-of-measurement callout, separate from the bulk material grade |
| Configuration and pole pattern, precisely named | Axial, diametric, true radial, or multipole with pole count and clocking — sensing applications are far less forgiving of ambiguity here than a simple holding magnet |
| Pole-to-pole uniformity (for multipole parts) | An encoder reading pole transitions needs consistent spacing and amplitude between poles — uniformity is a spec in its own right, not implied by pole count alone |
| Thermal stabilization for calibrated devices | Pre-baking a magnetized part above its worst-case service temperature moves the one-time thermal knock-down to before calibration instead of after — specify it on the drawing if the device is calibrated post-assembly |
Why this is a different conversation than "how strong does it need to be"
For a motor or actuator magnet, the relevant output is often torque or back-EMF waveform shape, not raw pull force — which depends on pole arc, magnetization direction (parallel vs. radial), and how the magnet interacts with the rest of the magnetic circuit, not on the magnet in isolation. For a sensor magnet, the acceptance test itself needs defining as part of the spec — fixtured field-at-point, Helmholtz moment measurement, or a pole scan — with limits anchored to first-article hardware, the same principle covered in the tolerancing chapter earlier in this guide, applied to a magnetic property instead of a dimension.
The sensor or motor part number, if one exists, is often the single fastest way for a supplier to catch a mismatch. Sending the mating sensor or winding spec alongside the magnet drawing lets a supplier check pole pitch and field amplitude against what the sensor actually needs to read — catching a problem before tooling exists rather than during a first-article test that fails for a reason that traces back to an assumption on the magnet side.
Engineers Guide / Validating Before Production
Validating Before Production
What to Simulate vs. What to Test
5 min read
Chapter 1 of 4 — Validating Before Production
A pull-force calculator or FEA model gives a fast, cheap answer early in a design — and it's an idealized answer, built on assumptions about perpendicularity, surface finish, saturation, and perfectly uniform material that real parts don't perfectly satisfy. Knowing which gap matters for a given design is what separates useful simulation from a number that quietly misleads.
What simulation and calculation are good for
- Comparing design options quickly. Should this be a disc or a ring, what happens if the air gap grows by a millimeter — relative comparisons between options are usually reliable even when the absolute numbers carry error.
- Establishing an order-of-magnitude starting point. Sizing a magnet roughly before committing to a first physical sample.
- Modeling temperature and permeance interactions that are genuinely hard to measure directly, like the load-line position at an elevated temperature.
Where physical testing is not optional
| Situation | Why simulation alone isn't enough |
|---|
| Final pull-force validation for a safety-relevant holding application | Real-world pull is consistently lower than idealized calculation — paint, shear loading, and dynamic loads all subtract from the theoretical number in ways a calculator doesn't fully capture |
| Sensor field-at-point acceptance | The actual measurement method and fixture define what "passing" means — a simulated field profile isn't the acceptance criterion, the measured one is |
| Thermal performance near the material's rated ceiling | Real material variation and manufacturing tolerance mean a part right at the edge of its calculated margin needs a physical measurement, not just a model that assumes nominal material properties |
| Anything with a manufacturing tolerance that could plausibly fail | Simulation typically models a single nominal geometry — first article inspection is what confirms the actual distribution of real parts meets the requirement |
A reasonable rule of thumb: use calculation and simulation to narrow a design space quickly and cheaply, and reserve physical testing for the design that's actually going to ship — especially anywhere close to the calculated margin. Treating a simulated number as final for a safety-relevant or tightly-margined design is one of the more common and avoidable sources of field failures.
Engineers Guide / Validating Before Production
Validating Before Production
Working With a Supplier's Engineering Team
5 min read
Chapter 2 of 4 — Validating Before Production
A magnet supplier's engineering team sees a wider range of designs, failure modes, and manufacturing constraints than any single design engineer does in the course of one program. Bringing them in during design — not just at RFQ, and not only after a first article fails — is one of the highest-leverage things an engineer can do on a magnet-containing design.
The three points where it's worth reaching out
| Stage | What to ask |
|---|
| Concept, before geometry is locked | "Is this feature manufacturable, and roughly what would drive cost here?" — cheap to change now, expensive later |
| Before finalizing tolerance and acceptance criteria | "What's actually achievable at standard vs. tight tolerance for this geometry, and what does the difference cost?" — an engineering team can usually answer this faster and more concretely than a generic reference guide |
| When a physical test result doesn't match the calculated one | "Here's what I measured and here's what I calculated — where's the gap likely coming from?" — often faster to resolve with someone who's seen the specific failure mode before than by re-deriving from first principles alone |
What makes that conversation actually productive
- Bring the application, not just the spec. "This needs to hold 3 lbs" is less useful than "this holds an access panel closed near an engine bay at roughly 250°F" — the second version lets an engineer catch problems the first doesn't reveal.
- Share the CAD model or drawing, not just a description. A rough sketch resolves ambiguity that a paragraph of text can't — the CAD download center linked below covers what file formats are useful to exchange.
- Say what's fixed and what's still flexible. If the geometry is locked but coating is open, or vice versa, saying so up front focuses the conversation on the decisions that are actually still live.
This isn't the same relationship as procurement's supplier vetting. A buyer evaluates a supplier's quality system and commercial terms; an engineer is evaluating whether their engineering team can actually engage with a technical problem. Both matter, and they're evaluated differently — a supplier can have excellent quality documentation and a thin engineering bench, or the reverse.
Engineers Guide / Validating Before Production
Validating Before Production
First Article and Design Validation Testing
5 min read
Chapter 3 of 4 — Validating Before Production
A first article inspection is the point where a design stops being a drawing and becomes a measured, physical thing — and how the acceptance criteria are set at this stage determines whether disputes six months from now have a clean answer or turn into a debate about what was actually agreed.
What an FAI actually verifies
A proper first article inspection checks the first parts off a new or changed tool against every dimension and magnetic property on the drawing — not a sample-based spot check. This matters because a new tool or process can introduce a variable that a prototype (made by a different, more manual process) never exercised — meaning a design that passed prototype testing can still fail first article, and that's the process working correctly, not a false alarm.
Setting acceptance criteria that actually hold up
Anchor limits to measured reference hardware, not to a number derived purely from theory. The defensible method: measure the approved first-article samples in the actual production test fixture, and set acceptance limits around those measured values — then retain the samples as physical reference hardware. Limits set from a datasheet or calculation alone tend to produce fixture-to-fixture disputes later that limits anchored to retained hardware simply don't.
A short list of what belongs in the criteria
- The measurement method and fixture, named specifically — "acceptable field strength" without a defined measurement setup isn't a criterion, it's a hope.
- Visual standards with reference images, since "minor chip" and "significant chip" mean different things to different inspectors without a picture to anchor the line.
- What happens on a failure — a documented containment and root-cause process, not just "remake it," so a real process problem doesn't repeat identically on the next lot.
Feeding disputes back into the design
Most tolerance and acceptance disputes in year one of a program trace back to a criterion that was vague rather than a part that was actually wrong. Every resolved dispute is an opportunity to tighten the specific line on the drawing that caused the ambiguity — that feedback loop, run consistently, is what makes a program's second year quieter than its first.
Engineers Guide / Validating Before Production
Validating Before Production
Common Engineering Mistakes That Cause Magnet Problems in the Field
6 min read
Chapter 4 of 4 — Validating Before Production
Most magnet field failures trace back to one of a small number of design-stage decisions, not to a bad batch of material. None of the items below are exotic — each is a known failure mode with a known, usually cheap, design-stage fix.
The recurring list
| Mistake | What it causes | The fix |
|---|
| Specifying temperature ceiling from the material datasheet alone, ignoring geometry | Irreversible demagnetization in service, well below the "rated" temperature | Evaluate temperature and permeance coefficient together — see the thermal design chapter |
| Press-fitting a sintered magnet as the sole retention method | Cracking, sometimes immediately and sometimes only after time in service | Bond with a controlled glue line, or use mechanical capture — never rely on interference fit alone |
| Leaving coating unspecified or under-specifying it for the environment | Corrosion that degrades both the mechanical surface and, eventually, the magnetic material beneath it | Specify coating and thickness deliberately based on actual service environment, not a default assumption |
| Assuming a magnetized part can be trimmed or drilled after the fact | Cracked parts, scrapped assemblies, or degraded field from disrupted domain alignment | Lock features that require machining before magnetizing — see the manufacturability chapter |
| Ambiguous magnetization direction on the drawing | A part built to the wrong pattern, discovered at first article rather than before tooling | Name the axis explicitly, referenced to a physical feature — see the magnetization chapter |
Nearly all of these share a pattern: they're decisions that felt safe to defer or assume during design, and turned out to be irreversible or expensive once the part existed. The five minutes it takes to check each one against this list before a drawing is released is cheap relative to a field failure, a redesign, or a scrapped production lot.
Where to go if something has already failed
If a part is failing in the field and the cause isn't obvious, the most useful first step is usually a direct conversation with the supplier's engineering team with the actual failed part and its service history in hand — not a guess based on this list alone. This list is meant to catch problems before they happen; diagnosing one that already has often needs the specific part in front of someone who can measure it.
Engineers Guide / Prototype to Production
Prototype to Production
Design Changes and Revision Control on a Magnet Drawing
5 min read
Chapter 1 of 4 — Prototype to Production
A dimension change that would be a trivial CNC program edit on a machined part can mean an entirely new tool for a magnet — because the tool that produces the geometry and the fixture that produces the magnetization pattern are both built around the specific dimensions on the drawing at the time they were made.
Changes that look small but aren't
| Change | What it actually affects |
|---|
| A dimension change, even a small one | May require new tooling entirely, not a tool adjustment — worth confirming before assuming it's minor |
| A tolerance tightened after first article | The existing FAI and retained reference hardware may no longer apply — a new first article may be needed against the new criteria |
| A coating change | Can change the part's dimensions (coating adds thickness) and its bonding characteristics — a retention design validated on the old coating isn't automatically valid on the new one |
| A magnetization pattern change | Almost always means a new or modified fixture — treat this with the same weight as a geometry change |
What a disciplined process looks like
- Route every drawing change through the supplier before assuming it's minor. "Is this a tool change?" is a fast question with a consequential answer.
- Version the drawing and reference which revision any retained FAI hardware and acceptance criteria apply to. A dispute over "does this part conform" is much easier to resolve when it's clear which drawing revision the reference hardware was measured against.
- Treat a coating or material substitution with the same rigor as a dimensional change, even though it can feel like a smaller decision — the downstream effects on retention and tolerance are real.
The cost asymmetry is the reason this matters: a five-minute check with a supplier before releasing a revision is nearly free. Discovering after the fact that a "minor" change actually required new tooling — after parts have already been ordered against the assumption that it didn't — is not.
Engineers Guide / Prototype to Production
Prototype to Production
When to Freeze the Design (and Why Magnets Punish Late Changes)
4 min read
Chapter 2 of 4 — Prototype to Production
Every program has a moment where the design is "done enough" to commit to tooling — and for a magnet-containing design, that moment carries more weight than it does for many other components, because the cost of being wrong after that point is higher.
Why magnets are less forgiving of late changes
A late change to a machined part is often a program edit and a re-cut; a late change to a magnet's geometry or magnetization pattern is frequently a new tool or fixture, on top of the manufacturing lead time that follows it. That compounding effect — tooling lead time added on top of, not instead of, production lead time — is the single biggest scheduling risk of freezing a magnet design too early, before the requirement that actually drives its geometry is settled.
Signals it's genuinely time to freeze
- The magnetic circuit (air gap, steel return path, mounting geometry) is set by the surrounding mechanical design, not still moving
- A prototype or sample has validated the calculated performance against the real requirement, not just the idealized one
- Tolerance and acceptance criteria have been discussed with the supplier's engineering team, not just assumed
- The production volume estimate is stable enough that tooling amortization makes sense at that volume
Signals it's premature
If the surrounding assembly is still changing, the magnet almost certainly still is too — even if the magnet's own drawing hasn't been touched recently. A magnet's spec is derived from its installed circuit; if the air gap, mounting, or mating part is still in flux, the magnet spec built around today's version of those may not survive to the version that actually ships. Prototyping in something close-enough-to-final, rather than committing to production tooling, is usually the right call while the surrounding design is still moving.
Freezing early feels like progress and can genuinely be the wrong call if it locks in assumptions the rest of the design hasn't settled yet. The should-cost modeling guide below is useful context here too — understanding where tooling cost actually comes from makes the freeze decision a more informed trade-off than a default "we're on schedule, let's commit."
Engineers Guide / Prototype to Production
Prototype to Production
Handling Design-for-Cost Trade-offs Without Breaking Function
5 min read
Chapter 3 of 4 — Prototype to Production
Cost pressure on a program almost always eventually reaches the magnet, and the instinct is often to ask for a price concession on the existing spec. The more productive question is usually a design one: is there a spec change that removes cost without touching function at all.
Levers that are usually safe
| Lever | Why it's often free savings |
|---|
| Relaxing a non-functional tolerance | If the tolerance was a CAD default rather than a calculated requirement (see the tolerancing chapter), loosening it removes a manufacturing step without touching performance at all |
| Converting a custom size to a stock-adjacent one | A small geometry adjustment on the mating design can sometimes let a stock magnet work — see the custom-vs-stock discussion in the Buyer's Guide for the commercial side of this same decision |
| Simplifying a coating spec where the environment allows it | A lighter or more standard coating can be a real saving where corrosion exposure was over-specified relative to actual service conditions |
Levers that look like savings and usually aren't
Downgrading grade to hit a price target without re-verifying performance is not a safe lever. Material cost tracks a real market index — it's not padding — so a lower grade genuinely does less work, and the margin that was there for a reason (thermal headroom, coercivity against an opposing field) may not survive the substitution. Any grade change needs to go back through the same validation the original spec went through, not be treated as a drop-in swap.
The order that actually finds savings
- Separate tolerances and coating decisions that were calculated from ones that were defaulted — only the second group is free to revisit.
- Check whether the current geometry is genuinely custom or could be a stock-adjacent modification instead.
- Only after those are exhausted, consider whether a lower grade could work — and if so, treat it as a new spec requiring its own validation, not a substitution.
This is the engineering side of a conversation procurement is often having from the commercial side at the same time — the cost-driver chapters in the Buyer's Guide cover that half; this chapter is about which of those levers are safe to pull without a redesign and which aren't.
Engineers Guide / Prototype to Production
Prototype to Production
Documentation Engineers Should Demand From a Supplier
4 min read
Chapter 4 of 4 — Prototype to Production
Procurement has its own list of documents to request — CoCs, material certs, compliance statements, covered in the Buyer's Guide. This chapter is the engineering-side list: the records that actually matter for validating and maintaining a design over its life, some of which overlap with procurement's list and some of which don't.
The engineering-relevant documents
| Document | Why an engineer specifically needs it |
|---|
| First Article Inspection (FAI) report | The actual measured dimensional and magnetic properties of the samples validation was based on — not just a pass/fail statement |
| Retained reference hardware | Physical samples the acceptance criteria were set against — invaluable for resolving any future dispute about whether a part conforms |
| Demagnetization curve / material certification | Proves the grade is the grade — relevant for anyone validating a thermal or performance margin, not just a procurement traceability requirement |
| Process flow and control plan (PPAP-style, where warranted) | For a long-life or safety-relevant program, understanding the manufacturing process itself — not just its output — matters for assessing what could change it later |
When to ask for the full package versus the basics
Not every purchase needs a full PPAP-style submission — that level of documentation is warranted for a long-life, safety-relevant, or regulated program, not for a low-risk stock purchase. Matching the documentation request to the actual consequence of a problem mirrors the same principle covered in the Buyer's Guide's vetting chapters, applied here to engineering records rather than commercial ones.
Ask for these at RFQ or design-validation stage, not after a problem occurs. An FAI report and retained reference hardware are natural byproducts of a first article inspection that already happened — requesting them costs the supplier almost nothing if asked for at the right time, and can be genuinely difficult to reconstruct after the fact if asked for later.
Engineers Guide / Reference
Reference
Engineering Glossary: Quick Reference
3 min read
Chapter 1 of 2 — Reference
The terms used throughout this guide's chapters, defined quickly for reference. The full Magnet Terminology Glossary covers grades, materials, and additional technical vocabulary beyond this shortlist.
| Term | Definition |
|---|
| Br (remanence) | The magnetic flux density remaining in a material after an external field is removed — the baseline measure of a grade's field strength |
| Hcj (intrinsic coercivity) | The reverse field required to demagnetize the material — the property that determines resistance to demagnetization from opposing fields or heat |
| BHmax (energy product) | The maximum energy density a magnet can deliver — the headline number in a grade name (e.g., N42) but not the whole performance picture |
| Permeance coefficient | The operating point set by a magnet's installed geometry (air gap, steel circuit, aspect ratio) — determines where on the demagnetization curve the magnet actually operates |
| Load line | The line representing a magnetic circuit's operating conditions, plotted against the material's demagnetization curve — where they intersect is the operating point |
| Knee (of the demagnetization curve) | The point past which small additional field loss causes disproportionate, often irreversible, demagnetization |
| Reversible / irreversible loss | Temperature-driven field loss that fully recovers on cooling (reversible) versus loss that remains after cooling (irreversible) |
| DFM | Design for Manufacturability — designing a part so it can actually be produced efficiently, given the real constraints of the manufacturing process |
| FAI | First Article Inspection — full dimensional and magnetic-property verification of the first parts off a new or changed tool |
| PPAP | Production Part Approval Process — a structured documentation package proving a production process can consistently meet spec |
Engineers Guide / Reference
Reference
Engineer's Design Checklist: Spec to Production
4 min read
Chapter 2 of 2 — Reference
This is the guide in checklist form — one line per decision, in the order they typically come up in a design. Each links back to the chapter covering it in full. Print this page for a working reference; the print stylesheet strips the sidebar and navigation.
Specifying the magnet
- Requirement stated as a field or force at a defined point, with the magnetic circuit established (air gap, steel return path or none)
- Material family chosen against temperature ceiling, coercivity need, and cost — not defaulted to NdFeB without checking
- Tolerances separated into functionally necessary versus inherited-default — only the first group locked in as tight
- Magnetization direction and pole pattern named unambiguously, referenced to a physical feature
Designing the assembly
- Geometry checked against manufacturability constraints — no features assuming post-magnetization machining
- Retention strategy chosen deliberately (bonding, carrier, or mechanical capture) — not press-fit as a default
- Thermal margin evaluated using permeance coefficient and real service temperature, not the catalog rating alone
- Sensor or motor-specific spec additions included where applicable (field-at-point, pole uniformity, thermal stabilization)
Validating before production
- Simulation used to narrow options; physical testing reserved for the design that's actually shipping, especially near calculated margin
- Supplier engineering team engaged before geometry was locked, not only at RFQ stage
- Acceptance criteria anchored to measured, retained reference hardware — not theoretical limits alone
- Common-mistakes list checked against the current drawing before release
Prototype to production
- Every drawing revision routed through the supplier before assuming it's a minor change
- Design freeze timed to when the surrounding mechanical design has actually stabilized, not just the calendar
- Cost-reduction levers checked in order: non-functional tolerance, stock-adjacent geometry, then (with full re-validation) grade
- FAI report and reference hardware requested at validation stage, matched to the program's actual risk level
Using this checklist: a low-volume, low-consequence design doesn't need every item at full rigor — use this as the ceiling of diligence and scale down deliberately. If you're the one placing the purchase order rather than owning the design, the
B2B Magnet Buyers Guide covers the procurement side of the same journey.
