How Different NdFeB Magnet Grades (N, M, H, SH, UH, EH, AH) Match Different Operating Conditions

NdFeB marks are not a ladder of “strength.” On a full designation such as N42SH, the leading N names the sintered neodymium-iron-boron family, the number is a nominal maximum energy product in MGOe, and the suffix is a minimum intrinsic-coercivity class. In the shorthand used in the title, N, M, H, SH, UH, EH, and AH are those coercivity classes: N means the base class with no temperature suffix, not a separate energy-product family. Most commercial sintered grades used in motors, actuators, and sensors have remanence roughly in the 1.15–1.45 T band. At a fixed energy-product number, a suffix step changes Hcj much more than it changes Br or BHmax. What separates the classes in hardware is resistance to irreversible flux loss under temperature and demagnetizing field. Selection is a temperature and demagnetization-margin problem, not a search for the highest room-temperature BHmax.

The engineering decision is which coercivity class survives the worst-case operating point, not which mark has the highest BHmax

Every commercial sintered grade mark has two independent axes. The number sets a nominal BHmax band. The suffix sets a minimum Hcj floor. Only the second axis decides whether the design holds at temperature. Published supplier charts commonly list indicative continuous-temperature windows of about 80 °C for the base N class, 100 °C for M, 120 °C for H, 150 °C for SH, 180 °C for UH, 200 °C for EH, and a higher figure for AH, often charted near 220 °C. Those figures are conventions, not guarantees. They assume a stated permeance coefficient and an irreversible-loss limit — a common catalogue convention is on the order of 5 % after the thermal excursion — and they move if either assumption changes. They are also far below the Curie point, which for NdFeB is typically above 300 °C and is not a usable operating limit.

A mark described as “150 °C capable” is 150 °C capable on a particular load line and against a particular loss threshold. A thin, radially magnetized ring on a shallow load line, in a poorly heat-sunk actuator, can lose margin well below the headline window. A well-loaded magnet in a closed steel circuit can sit above that headline and still clear its knee. The catalogue window is a starting screen, not the design limit.

Two procurement constraints follow from the same physics. Higher Hcj classes generally require more heavy rare earth, typically dysprosium and/or terbium, either in the alloy or introduced by grain-boundary diffusion. That raises cost and ties the price to a more volatile market than neodymium alone. Grain-boundary diffusion is often used because it raises Hcj with a smaller remanence penalty, and less total heavy rare earth, than bulk substitution; it does not remove Dy/Tb price exposure. Second, a higher coercivity class is not automatically a stronger magnet at room temperature. Relative to the highest-Br grade in the same family, the higher-Hcj mark usually gives up some Br and BHmax. The task is to find the lowest coercivity class that holds the margin, and stop there.

Compare grades on thermal margin, cost pressure, and the failure modes each class does and does not cover

The table is not a Br/Hcj/BHmax datasheet. It compares the axes that actually decide the class: thermal demagnetization margin, heavy-rare-earth cost pressure, and the exposures each class does and does not cover. Entries are qualitative. Certified curves for the exact grade and geometry override the table.

AH is not a synonym for EH. It is the higher Hcj class and is often given a higher indicative window; the certificate, not this table, sets the number. Cost pressure in the table is directional only. A current supplier quote is the only defensible bill-of-materials figure.

The comparison that a design review actually needs is a load-line map: permeance coefficient on one axis, temperature on the other, with each candidate knee overlaid. The knee must remain below the worst-case magnet temperature with margin left. The same map shows when a circuit change — a thicker magnet, a better return path, a smaller air gap — restores an N or M class, and when a move to UH would only hide a low permeance coefficient at a higher heavy-rare-earth cost.

The load line and the temperature-shifted knee set the class, not room-temperature BHmax

Irreversible loss begins when the load line intersects the demagnetization curve beyond the knee. The magnet then recoils on a lower line and does not recover the original flux. It is the load line that must clear the knee. The recoil line is the path after the damage, not the test of whether damage occurs.

Under the uniform-magnet approximation, the permeance coefficient is

where $B_m$ is the flux density at the operating point, $H_m$ is the demagnetizing field there, and ${\mu }_0$ is the permeability of free space. With $H_m$ entered as a negative second-quadrant value, the same quantity is $P_c=-B_m/({\mu }_0{H}_m)$. $P_c$ is dimensionless. A high $P_c$ is a thick, well-backed magnet in a closed circuit. A low $P_c$ is a thin or poorly returned magnet sitting closer to the knee. The expression is an average. Corners, thin radial sections, and local reverse fields can sit on a worse local load line than the average $P_c$.

The knee moves because Hcj falls much faster with temperature than Br does. For many commercial sintered grades the intrinsic-coercivity coefficient is roughly −0.4 to −0.7 %/K, and the remanence coefficient is roughly −0.10 to −0.13 %/K. Both figures are envelopes, not grade guarantees, and they shift with composition and with the route used to raise Hcj. A magnet that looks safe at 20 °C can be past its knee at 140 °C even though Br has fallen only a few percent. That statement is conditional on class and load line; it is not a universal crossing temperature.

The class is therefore set by where Hcj(T) and the knee sit relative to the load line at peak temperature, including fault field — not by the room-temperature BHmax column. The ratio $H_{cj}(T)/H_{cj}(20\mathrm{°}\mathrm{C})$ shows how fast the knee is approaching. It is not a sufficient specification. Br(T) and the actual permeance both enter the check.

A thin ring or a small bias magnet often has low $P_c$, so a headline temperature rating overstates the margin whenever temperature or reverse field is also high. Either price the class the low-$P_c$ curve requires, or change the circuit so $P_c$ rises. A tall magnet in a steel return path may hold the same functional margin on an H or even an M class, with less heavy rare earth. These coefficients describe broad sintered NdFeB behavior only. They do not replace the supplier’s measured curves for the exact grade. Bonded NdFeB, ferrite, and SmCo use different designation systems and different coefficients and should not be read from this suffix chart.

Match the coercivity class to the thermal window and the reverse-field exposure, and keep an explicit loss margin

The following windows are starting points for sintered NdFeB in actuators, motors, and sensors. The supplier curve at the real permeance coefficient overrides them.

• **Room temperature, well-loaded circuits ($P_c$ on the order of 2 or higher), no reverse field. N or M is usually enough. Coercivity that the load line never uses is only cost.
• Continuous operation toward roughly 100–120 °C, moderate load line. H is the usual screen. SH is the safer default if the thermal path is uncertain or duty-cycle peaks are aggressive.
• Continuous operation toward roughly 150 °C, or any reverse-field exposure. SH as the minimum. UH if the load line is shallow or the peak temperature has little margin.
• Continuous operation toward roughly 180 °C, including many high-current actuators and some traction-motor designs.** UH is the usual screen. Grain-boundary diffusion is a common way to limit the remanence penalty; it is not a requirement, and it does not fix a low $P_c$.
• Above roughly 180–200 °C. EH or AH, and a thermal-path redesign in parallel. At this point the suffix is only part of the solution. Confirm the AH window on the current certificate; do not borrow the EH figure.

Two review rules are worth writing down as practice, not as a coded limit. Size the class so the worst-case knee sits about 20–30 °C below the maximum expected magnet temperature, including transients and faults. And treat irreversible-loss allowance as a numeric specification — whatever the system actually tolerates — then verify it with a thermal soak on the real geometry and coating stack. A datasheet headline is not that test. If the soak data do not exist, that gap closes before release.

Grade selection fails when the circuit, the fault, or the environment is the limit, and no suffix fixes that

A higher suffix does not repair a bad magnetic circuit. The following limits are intrinsic to sintered NdFeB in real hardware.

• Time spent above the class window. A short excursion and a long soak are different. Held above its window, the magnet takes irreversible magnetic loss as the operating point passes the knee. A higher class buys headroom. It does not buy immunity, and it does not turn that loss into a reversible Br shift.
• Reverse field and short-circuit faults. An N or M magnet in a machine that sees a short-circuit current can demagnetize locally and stay demagnetized. This is the case where UH or EH is required by the fault field, not merely comfortable on temperature.
• Shallow load lines and thin sections. A thin radial ring or a small sensor magnet can pass the knee well below the headline temperature. Geometry and return path have to change; another suffix only moves the same problem.
• Air-gap and concentricity errors. An effective air gap larger than nominal lowers $P_c$ and walks the operating point toward the knee. Tolerance stack-up is a demagnetization risk, not only a torque or flux-density risk.
• Incomplete magnetization. Higher-Hcj grades need a substantially higher magnetizing field to reach the catalogue polarization. An under-magnetized part misses both the Br and the knee that the margin calculation assumed. Magnetizing-fixture access is part of the grade decision.
• Eddy-current heating. Slot harmonics, high electrical frequency, and rotating fields can heat the magnet from the inside, or heat a conductive sleeve around it. Suffix selection does not remove that heat. Segmentation, insulation paths, and the thermal design do.
• Coating and environment. The suffix says nothing about salt spray, humidity, or coolant. Corrosion loss is a coating and plating problem. It is not the same mechanism as thermal demagnetization, and a coercivity discussion does not close it.
• Over-specification. UH and EH carry real heavy-rare-earth cost and supply exposure. If the circuit can be changed to raise $P_c$, that is usually the cheaper and more stable fix.

If more than one of the circuit, fault-field, and environment limits is active, the next step is a circuit and thermal redesign, not a suffix upgrade. Long-term loss above the knee is magnetic and irreversible. Oxidation, corrosion, and coating breakdown are separate mechanisms and are not cured by Hcj.

AIC Engineering sizes the class from the circuit and the worst-case temperature, not from a catalogue lookup

AIC Engineering treats this decision as a magnetic-circuit problem. The capabilities that bear on it are:

• Magnetic circuit and magnetic-application structure design. The load line and the demagnetization knee are evaluated for the specific geometry and worst-case temperature, so the class follows from the operating point rather than from the headline window alone.
• Special motor permanent-magnet assemblies. Multi-pole rings, radially oriented rings, Halbach arrays, and linear-motor magnets — the geometries in which shallow load lines and thermal risk tend to concentrate.
• Rapid prototyping in 3–7 days. Samples in that interval, so the geometry exists before a production bill of materials and a heavy-rare-earth content are frozen. A thermal soak is a separate test plan; the 3–7 day interval is the prototyping lead time, not a claim that the soak is finished in that window.
• Permanent-magnet product quality inspection. Delivered magnetic properties can be checked against the Br and Hcj assumed in the margin calculation, rather than left at nominal catalogue values.

For a high-temperature or reverse-field-sensitive program, the sequence is to start from the worst-case temperature and the fault scenario and work back to the minimum Hcj class, rather than to start from a preferred grade and discover the knee in test.

Record the worst-case operating point before the grade and the heavy-rare-earth content are frozen

1. Write down the maximum magnet temperature, including transient peaks and fault conditions, and the peak reverse field if there is one.
2. Estimate $P_c$ for the as-built geometry, including tolerance-driven air gaps, not the nominal gap alone.
3. Overlay candidate knees on that operating point and keep the review margin described above.
4. Set a numeric irreversible-loss specification and plan a thermal soak on the actual geometry and coating stack.

Magnetic-circuit design and rapid prototyping in 3–7 days are how that sequence is closed on hardware rather than on a headline window. Contact the AIC Engineering team at https://www.aicmagnetics.com.

Grade conventions and temperature behavior should be read from standards and supplier curves, not from a single headline

1. GB/T 13560, Sintered neodymium iron boron permanent magnets — commercial grade designation, including the energy-product number and the coercivity suffix. Use the current revision; the top of the suffix series has been extended over time.
2. IEC 60404-5, methods of measurement of the magnetic properties of permanent magnets. This standard governs how properties are measured. It does not replace a supplier demagnetization curve, and it is not the source of the indicative temperature windows.
3. Supplier technical catalogues and application notes from Arnold Magnetic Technologies, Shin-Etsu Chemical, and Proterial (formerly Hitachi Metals, NEOMAX) — Hcj classes, indicative maximum operating temperatures, and temperature coefficients for specific grades.
4. Peter Campbell, Permanent Magnet Materials and their Application — permeance coefficient and load-line methods under the uniform-magnet approximation.
5. U.S. Department of Energy, Critical Materials Assessment — dysprosium and terbium supply context for high-coercivity grades. It is context for cost volatility, not a price list.