Magnet Grade Selector

selector — geometry, circuit, temperature

why the grade table lies

Every magnet catalogue, including the reference tables in our own knowledge base, prints a maximum operating temperature against each grade suffix: N to 80 °C, H to 120 °C, SH to 150 °C. Those numbers are a property of the material, measured at a favourable reference geometry. They are not a property of your part.

What actually causes irreversible loss is the magnet's own demagnetizing field. A magnet generates a field that opposes its own magnetization, and the strength of that self-demagnetizing field is set entirely by shape. A short, wide magnet drives itself hard into the second quadrant; a long, narrow one barely does. Heat then pushes coercivity down until the working point falls past the knee of the curve — and whatever crosses that line does not come back on cooling.

The other half of the picture is the circuit. Putting the magnet against a steel plate roughly doubles its permeance coefficient, which can move the usable temperature by 30 °C or more without changing the part at all. A magnet that fails on the bench can be entirely safe once it is assembled.

method

what it can't see

• The nominal rating still binds. Load-line analysis covers demagnetization only. Sustained heat also drives oxidation and grain-boundary changes that no B-H curve predicts, so the tool never recommends a grade above its catalogue rating even when the load line would allow it. Where geometry is the tighter constraint, that is called out explicitly.
• Property values are typicals, not guarantees. Br and Hcj vary between mills and within a grade. For qualification, request the actual demagnetization curves at temperature from the supplier rather than relying on nominal figures — including ours.
• Uniform temperature is assumed. A thick magnet heated briefly can be far hotter at the surface than the core, producing uneven demagnetization the model treats as a single number.
• Pc is a first-order geometric estimate. The equivalent-radius treatment for blocks and rings is an approximation, and a real assembly with pole pieces, adjacent magnets or leakage paths can differ substantially. Opposing magnets in an array are a common and significant case this does not model.
• Squareness varies by mill. The knee is placed at 90% of Hcj. Lower-quality sintered material can knee earlier, which moves every temperature in the table down.
• Alnico is a special case. Its coercivity is so low that it demagnetizes from handling, not just heat, and it must be stored with keeper plates. Treat any Alnico verdict here as directional only.