Object lessons in designing with neodymium magnets

Edited by the Engineeringtalk editorial team Mar 23, 2004

Three case histories show that problems in applying neodymium-iron-boron magnets are often attributable to insufficient knowledge of the material's capabilities.

Since neodymium-iron-boron magnets were first introduced more than 20 years ago, their performance has improved considerably and prices continue to fall.

Yet many design engineers still struggle to use them in sensors and electric motors.

Three case histories from Magnequench show that problems are often attributable to insufficient knowledge of the material's capabilities.

The first is the conversion of a two-pole motor for a fuel pump from ferrite to a bonded neodymium, with an increase in energy product from 4 to 10MGOe.

The change was necessary because the ferrite magnets were unable to meet a performance specification at low temperatures.

(Uniquely among permanent magnet materials, the intrinsic coercivity of ferrite decreases as the temperature falls).

Initially, the solution appeared to be simply to replace the ferrite with an identically sized bonded neodymium magnet, keeping the original two-pole design.

There was a slight increase in flux and the performance specification was met - although marginally - but finite element analysis (FEA) of the magnetic circuit showed that the return path is saturated; a clear sign that there is too much magnet in the circuit.

Performance can be improved by making the return magnet thicker so that it carries all the available flux - a straightforward approach, but not the most effective.

Instead, better use of the magnet can be made by reducing the included angle, or by changing to a four-, six- or eight-pole design.

From a manufacturing perspective, a single ring is preferable to arc segments - and the geometry is well suited for multipole configurations, provided the assembly is properly magnetised (normally by using a special magnetising fixture).

The solution recommended by Magnequench was a multipole ring with a thicker return path.

The second example was a case where the performance of a motor using isotropic neodymium was less than expected, giving rise to excess cogging.

FEA revealed magnetising as the problem - a common occurrence with isotropic materials, because it is the magnetising rather than the material itself that determines the flux pattern.

The field created by the magnetiser was not perpendicular to the surface of the magnet, as was assumed at the design stage, which meant that the flux was off slightly from the intended angle.

When the FEA model was altered to reflect the true pattern of the flux, the excess cogging was obvious.

Altering the direction of the field delivered by the magnetising fixture was not possible in this case, so the only option was to magnetise before assembly.

(In this connection, it is worth pointing out that the tables of magnetic properties published by manufacturers are from samples of their products that have been magnetised to saturation, and that measurements are made parallel to the direction of magnetisation.

If one or both of these conditions are not met, the published properties are invalidated, which can result in under-performance of the kind seen in this case).

The final example concerns a diametrically magnetised ring of isotropic bonded neodymium intended for a sensing application, for which a relatively uniform magnetic field inside the ring was required.

In practice, most of the flux travelled within the wall of the ring rather than where it was needed.

Magnequench's solution was to create the field using a dipole Halbach ring - a configuration that allows the field at the centre to be calculated before manufacture.

A further advantage was that the ring could be manufactured and magnetised as a single piece, eliminating the complexities of assembling many segments, each of which has a specific direction of magnetization.

The resulting flux pattern made possible by the special fixturing achieved an increase in density from 5 to 65mT at the centre of the ring, with very little wasted outside it.

These three examples - each of which shows how the performance of devices can be improved by a better understanding of the magnetic materials employed - are by no means untypical of the technical enquiries that Magnequench deals with every day.

They demonstrate that, whatever the design, it is always prudent to take the advice of a magnet applications engineer, even before a first prototype is built.