IEEE Electrification Magazine - March 2017 - 25

Figure 5(b) is the large sample-to-sample variation. It is
impossible to discern between property variations due to
the diffusion process, and variations due to sample preparation. Some of the factors contributing to this include the fact that the
original magnets had rounded corners and edges, not exactly parallel
opposite walls in the prepared cubes,
and that these cubes may not have
been mounted exactly parallel to the
poles of the VSM.
The fundamental shortcoming of
this approach is that only the two
extreme BH curves were measured
(within the allowable sample size),
and these do not represent the many
regions with different boundary concentrations of Tb. In fact, depending
on the conditions of the diffusion,
even the BH curve of the corner cube might represent a
gradient of concentrations of Tb. For the design of EV and
HEV motors, it would be preferred to have BH curves that
correspond to each of the regions with different magnetic properties.

regardless of final part shape. However, for diffused magnets the magnetic properties are not homogenous
across the sample; accurate characterization requires the
measurement of the specific part and
not a standardized sample. This measurement of the nonhomogeneous
magnet is useful for sample-to-sample
comparison and for general specification purposes.
For some electromagnetic rotor
designs, however, it is important to use
BH curves corresponding to the regions
containing different concentrations of
the HRE element. Ideally, these BH
curves will be accurately simulated for
the same regions as determined by
EPMA. At this time, however, such simulated BH curves are not readily available from all magnet manufacturers.
Thus, the often-preferred method is to cut the nonhomogeneous magnet into small regions and then stack regions
from the same location obtained from different magnets
and measure the BH properties with a hysteresisgraph. The
stacking is necessary because the regions are usually so
thin that several must be stacked to exceed the height of
the hysteresisgraph detection coil. In Figure 6, the magnet
is sliced in three layers, and each layer in turn is cut into
three regions. Region 1 from one magnet (A) can then be
stacked with region 1 from another magnet (B) and the BH
curves are measured on stacked magnet AB1.
Usually several assumptions are made when using this
method. It is assumed that the regions into which the
magnet is divided are somewhat homogeneous, which,
based on the EPMA data we already discussed, is a simplification that may not appropriately reflect the reality of
the HRE element distribution.

For the design of
EV and HEV motors,
it would be preferred
to have BH curves
that correspond to
each of the regions
with different
magnetic properties.

magnetic Properties Using a hysteresisgraph
As of late it appears that most magnet manufacturers prefer to determine the BH curves of diffused magnets via
closed-circuit magnetic measurements using a hysteresisgraph. The closed-circuit measurements discussed in this
article were performed on a PERMAGRAPH L with iron-
cobalt (FeCo) heating poles and a J-compensated surrounding coil by Magnet-Physik. The typical hysteresisgraph
measurements of homogeneous magnets often use standardized samples cut from sintered magnet blocks to allow
for the characterization of bulk magnetic properties

AB1

AB2

AB3

AB4

AB5

AB6

AB7

AB8

AB9

Figure 6. Slicing and stacking of the various regions in diffused magnets. The BH curves are measured on regions coming from the same location
of different magnets.

IEEE Electrific ation Magazine / march 2 0 1 7

Table of Contents for the Digital Edition of IEEE Electrification Magazine - March 2017