Professional Documents
Culture Documents
Power
Abstract
The theory of a permanent-magnet d.c. machine that has an axial field and a disc-shaped armature is
discussed. Consideration is given to the effect that the distribution of magnetic field in the airgap will
have on the generation of e.m.f. The machine parameters in the angular and radial directions are related
to this distribution, so that a pole shape may be chosen that maximises the output power.
List of principal symbols The axial-field motor described here is one that has been
developed for a few years, without having any special theory
e
ij = e.m.f..in conductor element derived for it. The ..basis of the theory not only aids the fur-
ej = e.m.f. in conductor at angular position/ ther development of the machine, but reminds us of the par-
ij = subscripts describing radial and angular positions ticular nature of conventional theory. The advantages that
/ = length of active conductor this motor offers as an automobile-radiator cooling-fan motor
p = number of poles and as an electric-vehicle traction motor have recently been
rt = radius corresponding to i described.1'2
v = linear velocity of conductor at its average radius
Ac = specific electric loading 2 Topology of the axial-field motor
Bm = specific magnet loading
The machine described in this paper has become
Bij = flux density at a point i, j known as the d.c. disc-armature motor. It was chosen to be
Bj = flux density at angular position/ a permanent-magnet machine because this field system, un-
E — total armature e.m.f. like that of electromagnets, requires no power input once the
Ec = average conductor e.m.f. poles have been fully magnetised. This leads to a greater over-
H = magnetising force all machine efficiency for any given power rating. This ad-
Hx = magnetising force in axial direction vantage is increased further by eliminating any iron from the
/ = armature current armature, thus making the associated eddy-current and
Ic = conductor current hysteresis losses insignificant. A stationary flux-return circuit
M — intrinsic magnetisation on the remote side of the armature is therefore required if
Ms = pole strength on magnet face this gain is to be realised. This is very difficult to achieve, in
P = moment of flux density all but the smallest motors, unless an axial-field.arrangement
Pj = moment of flux density at angular position / is adopted.
/?i/?2= inner and outer radii of active length of conductors The machine now has a very large airgap, comprising the
Z = total number of armature conductors armature thickness and running clearance on either side of
a = pole arc it. The armature coils are therefore located towards the mid-
4> = total flux per pole dle of this airgap, and each is aircored. This means that
<hj = flux at point ij armature-reaction effects are negligible in such a machine.
i// = magnetic potential In an axial-field system, the active conductors run radially
CJ = conductor angular velocity from the axis of the machine. The coils are connected as a
1 Introduction double-layer winding (Fig. 1) and the complete armature is
Faraday's disc was an axial-field electrical machine, therefore disc shaped. This assembly is made possible by
and yet multipolar machines of this type are not the most bending the ends of the coils up, so that they no longer lie
commonly used today. Although they are still comparatively in the plane of the disc. One side of each coil is longer than
rare, there are a number of recent advances, particularly in the other, by an amount that makes it possible for the end
the development of permanent-magnet materials, that make windings to nest closely together. Finally, the winding is
them preferable to conventional machines in a number of connected to the commutator, and the complete armature
applications. is encapsulated in epoxy resin, giving it mechanical strength.
The airgap length is proportional to both the magnetising
Paper 7313 P, first received 28th January and in final form 2nd force and the length of the magnets. One of these quantities
August 1974 must therefore increase greatly to accommodate such a large
Mr. Campbell is with Cambridge University Engineering Department, airgap. So that the machine weight did not become too large,
Trumpington Street, Cambridge CB2 1P2, England permanent magnets with high coercivities were chosen to
PROC. IEE, Vol. 121, No. 12, DECEMBER 1974 1489
provide the field. The arrangement of the active parts of neighbouring pole edges required to produce optimum per-
this machine is shown in Fig. 2. formance will be approximately constant over most of the
radial active distance. This will become less true towards the
extremities of this distance, where the effects of the poles
are diminished. In these regions, also, the original assumption
of a constant flux density around the pole-face edges will be
least valid. However, for most of the active length of a con-
ductor, an optimum performance would be achieved if the
adjacent pole edges were parallel. There would be a loss of
material, though, if the magnets were cut from large rings,
and so, in the prototype disc-armature machines, the magnet
edges lie on radial lines.
Fig. 2
Exploded view of disc-armature motor
X=H
4
in good agreement with experimental results in Section 5. a—B
rs cos0 — x2 — s1
dd (15)
H [x2 + s2sin2d)'(r2 + x2 + s2 -2rs cos 0) 1/2
r=R,
I_ B The significance of eqn. 15 is that the field in the airgap can
be calculated without reference to a potential distribution
G
on the pole face. The intrinsic magnetisation M exists only
area 1 area 2
on the pole face asMs, and is not dependent upon radial or
F angular position. The drawback to eqn. 15 is that the re-
maining integration can only be performed by numerical
H A
means, and it will only be possible to find a discrete field
distribution.
Fig. 5
2-dimensional representation with simplified magnet
O 24 -
O 2O
O2O -
>; O I 6
OI6 -
x O 12
5 OI2 -
OO8
O O8 -
OO4
O O4 -
O 4 6 IO 16 2O 24
distance from magnet <£. deg
8 IO 16 2O 24
distance from magnet C^, deg
Fig. 9
Field distribution at first layer of conductors for various
pole-arcI pole-pitch ratios (radius = 47 mm)
Fig. 8
Field entering flux-return ring normally (radius = 47 mm)
O28 r
a Calculated curve
b Experimental values
O24
The accuracy of the calculation is demonstrated by
evaluating the field entering the return ring normally (Fig. 8)
at a particular radius. Under these conditions the distances * O2O
JC and s (Fig. 6) are constant in eqn. 15. Fig. 8 shows the E
field at the point Q over half a pole pitch, and this variation
is achieved by altering the position of the r axis (normal to - O 16
Q) by means of the angle 0. The computed values (solid line)
are in good agreement with the experimental values, found
by inserting a Hall-effect element on the surface of the re- x' O I 2
turn ring.
008
5.1 Pole-arc/pole-pitch ratio
It is more useful to calculate the field distribution
normal to each of the conductor layers in the armature. For 004
the layer nearest to the magnets, the effect of different pole-
arc/pole-pitch ratios on the field distribution is shown in
Fig. 9. The useful flux is seen to increase as the ratio in- 34 38 42 46 5O 54 58
creases. However, this gain becomes less as higher ratios are radius, mm
approached, owing to the increase in leakage flux between
adjacent magnets. Since this leads to a poorer utilisation of Fig. 10
the magnet material, the ratio of 0-8 used in this machine Radial field distribution along the magnet centre line
may have been a wise choice. Pole-arc/pole-pitch ratio = 0-8
5.2 Radial-field distribution, and a relationship The most common criterion for relating R^ to R2 is to
between Ri and R2- I maximise the armature power. This power El must first be
The importance of treating the machine as a 3- obtained in terms of specific magnetic and electric loadings.
dimensional problem can be seen from a plot, at the same The former is usually identical to the average useful flux
conductor layer, of the variation of field with radius at the density in the airgap, but it has already been shown that
midpoint of a pole (Fig. 10). This shows that the field at such a quantity is of little use in an axial-field machine. If
Ri and R2 is approximately half its maximum value. That Bij and r{ really are insperable, then a specific magnetic
maximum occurs at a radius slightly greater than the average loading, that is independent of machine dimensions, can
radius, as might be expected from the earlier discussion. As never be defined. For this purpose, the assumption must
with the angular direction, there is clearly a need to optimise therefore be made that Bj, defined by eqn. 3, is independent
PROC IEE, Vol. 121, No. 12, DECEMBER 1974 1493
ofr,-. It is then possible to obtain the average conductor the two variables:
e.m.f. Ec from eqns. 3, 7 and 8:
2-n/p \p(x,d) = X(x)e(d) (20)
Ec = {u(Rl-Rl)pl4n}f Bj
o By first considering the 0 direction, the solution must be
The specific magnetic loading Bav can then be defined as the periodic, and will have the general form
average of Bj over one pole pitch. Hence
0(0) = (21)
J, K, and X are determined from the boundary conditions
The specific electric loading Ac makes it possible to find, I//(JC,O) = 0 and \p(x,C) = 0. Hence, eqn. 21 becomes
from the machine dimensions, the total electric loading in
the armature. Conductor current Ic is approximately inde- ^ (22)
pendent of machine dimensions, but the number of con- where n= 1, 2, 3 etc. This is true for both areas 1 and 2 in
ductors Z will depend on R{. Ac is therefore defined as Fig. 5.
The solution in the JC direction will be slightly different
(17) for each area. Area 1 will be considered in detail. X is
common between the solutions in ea6h direction, and so,
After rewriting eqn. 16 as the total armature e.m.f., the in the x direction, corresponding to eqn. 22, will be exponential.
armature power becomes It is given by
Xn (x) = Ln sinh — (x - x0) (23)
El = ftu3BntAc{R\-R\)Ri (18)
The maximum armature power will therefore occur when where JC0 is a constant and is zero, owing to the boundary
R2 = \/3/?i (19) condition 0(0,0) = 0.
The total solution is found by combining eqns. 20, 22
For the motor shown in Fig. 7, Ri is 34 mm and R2 is 59 mm. and 23, and summing for all values of n:
v-, , nnd , rvnx
6 Conclusions = 2- kn sin — sinh — (24)
n=l C <-
This paper has developed the basic theoretical .
principles of an axial-field permanent-magnet machine. It The final boundary condition is the potential distribution on
is now possible to investigate the advantages of this mach- the magnet face, defined as 0(0) along the line HI. In eqn. 24
ine in a number of applications, with a greater accuracy.
Depending upon the design adopted, these advantages can be
high efficiency, high power/weight ratio, low inertia, high
speed, or merely the physical shape of the machine.
from which the sine Fourier coefficients of i//(0) are
The ability to study the variation of conductor e.m.f.
with angular position, by the use of eqn. 6, is particularly nnH 2 c . nnd
useful in a machine of this type. Since armature reaction — = - / 0(0)sin— (25)
is negligible, it is possible, by a suitable addition of the indi-
vidual conductor e.m.f.s, to calculate the voltage between The complete solution in area 1 is therefore found by sub-
two adjacent commutator segments for a rotating armature. stituting eqn. 25 in eqn. 24, from which eqn. 10 is obtained.
Using this method, work is in progress to study the quality The solution in area 2, given by eqn. 11, is found in a
of commutation in this machine, with a view to determin- similar way, allowing for the slight difference in the JC direc-
ing the maximum operating speed of any particular design. tion.