Sie sind auf Seite 1von 8

Faculty of Engineering

Faculty of Engineering Papers


The University of Auckland Year

Magnetic Field Analysis of an Ironless


Brushless DC Machine
Udaya Mandawala John T. Boys

University of Auckland,
University of Auckland, New Zealand, j.boys@auckland.ac.nz
This paper is posted at ResearchSpace@Auckland.
http://researchspace.auckland.ac.nz/engpapers/5
2384 IEEE TRANSACTIONS ON MAGNETICS, VOL. 41, NO. 8, AUGUST 2005

Magnetic Field Analysis of an Ironless Brushless


DC Machine
Udaya K. Madawala, Member IEEE, and John T. Boys
Department of Electrical and Computer Engineering, The University of Auckland, Auckland, New Zealand
This paper presents an analytical solution for the magnetic field inside an unconventional ironless brushless (ILBL) dc motor design.
We discuss the unique characteristics of the design, which adopts an inside-out construction with an internal ironless stator. With
the intention of obtaining an analytical solution for the magnetic field, we propose a model based on a magnetic pole concept for rep-
resenting the magnetic circuit. The magnetic field inside the motor, which has no iron for guiding magnetic flux paths, is obtained by
solving Laplaces equation for magnetic scalar potential. We present both analytical and numerical solutions for the magnetic field and
compare them with measured results of a 20-pole prototype IL design to show the validity of the model. The IL design is simple, easy to
manufacture and, as indicated by the results, has an optimum number of magnets, for which its performance becomes maximum.
Index TermsBrushless machines, Laplace equation, magnetic analysis, permanent magnets.

I. INTRODUCTION

C ONVENTIONAL dc machines, excited by field windings,


are largely being replaced by low-cost permanent-magnet
dc (PMDC) machines [1]. This trend is mainly due to the
availability of low cost high-energy permanent magnets as
they can be used for replacing the bulky wound field excitation
and its associated losses in conventional dc machines. Today,
the continuation of this trend coupled with the advances in
semiconductor technology has transformed conventional dc
machines into efficient, compact, and high power/weight ratio
brushless dc (BLDC) machines with electronic commutators,
thereby eliminating the necessity for a troublesome mechanical
commutator. Moreover, BLDC machines are often equipped
with relatively inexpensive yet sophisticated drive systems to
cater for a wide range of load conditions. As a consequence,
they have today become the obvious and most popular choice Fig. 1. Rotor and stator of an ILBL motor.
for a variety of industrial applications. Nevertheless, even with
these very attractive characteristics, many attempts are still This paper describes a nontraditional design of a BLDC ma-
being made to further improve or optimize both the design chine that uses an ironless stator structure. Although the mag-
and operation of BLDC machines as opportunities are always netic field inside this ironless (IL) design can be analyzed by
created for improvement by the continual advances in the a finite-element software package, the authors have instead fo-
technological world. The improvements to the motor design, cused mainly on obtaining an analytical solution, which is more
in particular the magnetic circuit, are generally carried out by difficult to achieve for this design without a clearly defined mag-
making use of commercially available finite-element software netic circuit for guiding the magnetic flux, in comparison to con-
packages. This is mainly because magnetic field analysis is ventional designs. To facilitate an analytical solution, a model
complex and nonlinear in nature, and is best solved by numer- based on magnetic pole concept with magnetic scalar potential
ical techniques, as analytical solutions are often difficult to is proposed for representing the magnetic circuit. The magnetic
arrive at and may still require numerical integration. In addition, field inside the design is obtained by solving Laplaces equa-
there are many software packages currently available in the tion for the magnetic scalar potential using conformal transfor-
market, which are relatively inexpensive, accurate, easy to use, mation and SchwarzChristoffel mapping, as described in [7].
and suitable for such analysis. Naturally, this has led most au- The validity of the proposed model and the analytical solution
thors to focus largely on numerical techniques when obtaining is verified by experimental results and a numerical solution.
solutions for magnetic field problems [2][4]. Nevertheless,
attempts are still being made to obtain analytical solutions for II. ARCHITECTURE OF IRONLESS-BRUSHLESS (ILBL) MACHINE
magnetic field problems associated with electric machines in
order to facilitate efficient and optimum designs through the As the name implies, the unique feature of this proposed
change of motor parameters [5], [6]. ILBL machine is its ironless stator. The ILBL motor, shown in
Fig. 1, adopts an internal stator structure that can be viewed as an
inside-out construction of a conventional BLDC motor design.
The external rotor of ILBL dc machines is a thin cup-shaped
Digital Object Identifier 10.1109/TMAG.2005.852952 cylindrical steel drum, inside which a series of low cost Arnox
0018-9464/$20.00 2005 IEEE
MADAWALA AND BOYS: MAGNETIC FIELD ANALYSIS OF AN IRONLESS BRUSHLESS DC MACHINE 2385

H-8 ceramic permanent magnets (PMs) are firmly glued in place and
to provide the field excitation. It is mounted on bearings located (2)
inside the nonmagnetic stator, made out of plastic. PMs of flat
pole surfaces, instead of preferable arc-shaped, are used to lower where is the conduction current density in the medium. The
the manufacturing cost and they are assembled as close as prac- relationship between and is given by
tically possible to each other to make the rotor interior circular
and flux linkages maximum. As a result of the inside-out design, (3)
ILBL motors invariably have a high number of pole pairs and
high rotor inertia. Both these characteristics are advantageous where is the permeability of free space and magnetization
for lowering speed fluctuations as torque pulsations occur at a is due to the induced and residual magnetizations
high frequency and they are further damped out by the high rotor of the medium.
inertia. However, with a high number of poles, the commutator By substituting for in (1)
invariably has high switching losses and demands precision po-
sition sensors to meet the required high degree of switching ac- (4)
curacy for PM operation.
The plastic stator of the ILBL design contains a single-layer, If is zero, then is a curl free vector and, hence, may be
full-pitched three-phase concentrated winding. It is skein- derived from the gradient of a scalar given by
wound in a thin slotless configuration and glued on to the
(5)
stator core to keep the manufacturing cost low. This winding
arrangement results in back-electromotive force (back-EMF) where is the magnetic scalar potential of . Substituting (5)
voltages of 10%15% of third harmonic distortion in typical in (4) gives the Poissons equation for
ILBL motors, which are ideal for low-cost BLDC operation in
a unipolar or bipolar switching arrangement to keep the torque (6)
ripple low. With the adoption of a plastic core with no iron,
the ILBL design inevitably has a low operating flux density. where is the volume charge density of the medium. For
This is further compounded by the use of Arnox H8 magnets mediums where divergence of is zero, Poissons equation
as they appear as an additional air gap in the magnetic circuit reduces to Laplaces equation
of the design. Typical air gap flux density of an ILBL motor
is in the range of 0.140.18 T and, therefore, the motor does (7)
not appear to be capable of producing a comparable torque.
The drawback of this low operating flux density is, however, According to (6), is zero for a uniformly magnetized
largely offset by the inside-out construction of the design as magnet since all spatial derivatives inside the magnet are
it makes the split ratio of ILBL machines as high as 0.85. This essentially zero. Therefore, the magnetic charges are confined
is because the specific output of a machine is largely governed only to the pole surfaces of the magnet and can be represented
by its rotor diameter since an increase in rotor diameter causes by a surface charge density using a pole model [8]. This
a corresponding increase in both the magnetic flux and current implies that the alternate north and south pole surfaces of the
in the machine whereas an increase in axial length results in magnets in ILBL machines can be represented by equivalent
only an increase in the magnetic flux and not in the current. As potential surfaces in direct analogy to electrostatics. If the
a consequence, the air-gap radius of ILBL machines is larger magnets exhibit linear demagnetization characteristics with
than that of a similar sized conventional design and, hence, rigid magnetization [9], then the potential difference between
mostly compensates for the specific power reduced by the lower the surfaces can be determined by modeling the magnet as an
air-gap flux density. In comparison to similar sized motors, intrinsic magnetomotive force given by
ILBL motors with no iron are considerably lighter and quieter
in operation as there are no ferrostriction forces. The design (8)
is low in cost, robust, and suitable for applications such as
blowers, small pumps, conveyors, etc., where its characteristics in series with internal and external magnetic re-
are advantageous. luctances, respectively.

B. A Model for the Magnetic Circuit


III. MODELING PMS USING POLE MODEL Because of the axisymmetric nature of the design, the mag-
netic field analysis is restricted only to a sector containing a
A. Pole Model
single magnet with the following assumptions.
From Maxwells equations under time invariant conditions, The effect of the armature reaction and the presence of stator
the magnetic flux density and the magnetic field intensity winding on the magnetic field distribution is negligibly
in a medium can be given by small.
Arnox-H8 magnets are uniformly and rigidly magnetized
perpendicular to the pole surface.
(1) The steel rim is infinitely permeable.
2386 IEEE TRANSACTIONS ON MAGNETICS, VOL. 41, NO. 8, AUGUST 2005

Fig. 2. A model representing a sector containing a magnet.

The end effects are negligible and, hence, a two-dimensional


solution is adequate.
With the absence of iron, the magnetic flux lines in ILBL ma- Fig. 3. A sector with a flat magnet in (a) Z -plane and (b) t-plane.
chines are in air. As a consequence, the magnets have an oper-
ating range on the curve instead of a static operating point,
Noting that any function that satisfies Laplaces equation is a
causing on the surface of the magnet to vary from a minimum,
harmonic function, is represented by a real part of a complex
where the magnets are virtually in contact, to a maximum at
function defined by
the center of the magnet where the external magnetic reluctance
is maximum. The potential distribution on the surfaces of the
magnet is, therefore, not deterministic for this particular design. (9)
However, on all interpolar axes is essentially zero because of
the alternate north and south pole configuration of the design. Since is an analytic function that satisfies the CauchyRie-
As a consequence, along the innermost steel rotor surface of mann equation, the magnetic flux paths in the machine can then
infinite permeability can also be considered as zero. With these easily be obtained by plotting the curves corresponding to con-
boundary conditions, a sector containing a magnet can be rep- stant values.
resented by a magnetic circuit model in Fig. 2. Fig. 3(a) shows a single sector, containing a flat surface
If there are no conduction currents, inside the machine may magnet in the complex -plane. The surface potential of the
be derived from using (5). The fact that is constant inside magnet is represented by a constant . If and are the polar
the magnet and zero elsewhere in the sector implies that inside coordinates of a point in -plane, then a point in any of the
the sector should satisfy Laplaces equation. Thus, the magnetic sectors may be represented by
field distribution inside an ILBL machine may be determined by
formulating the sector as a boundary value problem and solving
Laplaces equation for [10]. for (10)

where is an integer representing each sector with a magnet,


IV. ANALYTICAL SOLUTION FOR THE MAGNETIC is the number of magnets in the machine, and is the sector
FIELD DISTRIBUTION angle defined by

As described, the magnetic potential distribution over pole (11)


surfaces is indeterminate in this design. Therefore, in order to
obtain an analytical solution with an acceptable accuracy, a con- Since all sectors are identical, a solution can be obtained by
straint of the model was slightly changed as outlined below. considering a single sector in Fig. 3(a) with . The dif-
With the absence of iron, the radial air-gap magnetic field ficulty in obtaining an analytical solution with the two singu-
around the corners of the magnet is considerably small and larities at points A and B of this problem, can be overcome by
varying, whereas it is significant and reasonably constant over mapping the isosceles triangle onto a half-space in the -plane
the rest of the pole surface. This implies that the change in using the SchwarzChristoffel theorem with the transformation
surface potential is significant only around the corners of the
magnet, which is only a small part of the pole surface. In this
respect, it is reasonable to specify a constant surface potential (12)
distribution in the model by (8), as the error introduced at the
corners is insignificant and has only a little impact on the flux where is a constant. As shown in Fig. 3(b), the vertices A and
linkage, as justified by experimental results. B of the triangle are mapped onto 1 and 0 on the real axis of the
With the above change to the model, the problem now reduces -plane, respectively, while the remaining vertex C is chosen to
to solving Laplaces equation subject to only Dirichlet boundary be a point at infinity in the -plane. Substituting for in (12)
condition [11] and not to both Newmann and Dirichlet boundary
conditions. This can be achieved by formulating the problem in
the complex domain and using the method of conformal trans-
(13)
formation [12].
MADAWALA AND BOYS: MAGNETIC FIELD ANALYSIS OF AN IRONLESS BRUSHLESS DC MACHINE 2387

Solving for constant from set boundary conditions a ring magnet. This is justifiable because the rotor interior of
70.75 mm radius with flat surface magnets, is very much sim-
ilar (with only a 0.85 mm deviation at the center) to that with
(14) curved pole surfaces.
Fig. 4(a) shows a single sector of rotor radius , containing
a magnet with a curved pole surface in the complex -plane.
This problem has two singularities at points B and D where
Since these two integrals are functions [13], (14) can be rep- changes from 0 to on the interpolar axis when the mag-
resented by nets are virtually in contact. A solution for this problem can be
obtained by mapping the two singularities in the -plane onto
(15) a half-space in an intermediate -plane, as in Fig. 3(b) with a
single singularity, by the transformation from Kober [14]
where and represent
the incomplete and complete functions, which can be evalu-
(21)
ated by a series expansion. From Abramowitz and Stegun [14],
an incomplete function can be described by a hypergeometric
Substituting for and solving for gives
function as

(22)

(16) Therefore, the solution for Laplaces equation for in


which may be expanded to an absolutely convergent function -plane [15] is given by

(23)

Substituting for yields

(17) (24)
or
Therefore, the transformation, which maps the isosceles triangle
in the -plane onto a half-space in -plane, is (25)

This is the solution for Laplaces equation for complex po-


tential . By substituting for from (9) and considering the
(18)
real part, at any point inside the sector can be found by
Thus, the solution for Laplaces equation for complex potential
in -plane is given by

(19)
(26)
By substituting for and in equation
Since is derived from the gradient of , the corresponding
radial and tangential , components of are given by

(27)
and
(20)

As evident from (19), even an analytical solution for mag-


nets with flat pole surface is achievable, it is not still possible
to obtain an explicit expression for in terms of because of
the complexity of the solution. A workable analytical solution (28)
can, however, be obtained if the magnets of flat pole surfaces
used in the prototype design, are considered to be as curved Equations (27) and (28) can thus directly be used for deter-
pole surface magnets in the model, resembling to a motor with mining the magnetic field at any point inside a sector.
2388 IEEE TRANSACTIONS ON MAGNETICS, VOL. 41, NO. 8, AUGUST 2005

Fig. 4. A sector with a curved magnet in (a) Z -plane and (b) t-plane.

Fig. 5. Computed flux patterns due to magnets. (a) Analytical. (b) Numerical.

V. RESULTS AND DISCUSSION


All theoretical calculations and experimental measurements Fig. 6. Computed radial flux density distribution. (a) Analytical. (b)
discussed in this section have been based on a 20-pole prototype Numerical.
ILBL machine and its details are given in the Appendix. Both
the validity of the proposed magnetic model and the accuracy of
conclusion that can be arrived at from these patterns is that the
the analytical solution were verified by solving Laplaces equa-
erroneous flux paths at the corners of the magnets of the ana-
tion numerically by computer simulation with no modification
lytical solution have only a little contribution to the flux linkage
to the model, and comparing the results with those obtained ex-
and, hence, the analytical solution can be considered to be of
perimentally on the prototype.
acceptable accuracy when determining the performance.
Fig. 5 is a comparison of the flux patterns computed from the
analytical and numerical solutions. The flux patterns are similar The radial component of the air-gap magnetic field is a very
but all flux paths from the analytical solution emanate at right good indication of the shape of the back-EMF voltages gener-
angles to the pole surface. This was expected from the analyt- ated in the motor terminals. Therefore, as shown in Fig. 6, it
ical solution because of the imposed constant surface potential was computed analytically at different radial distances and com-
(calculated from (8) with an estimated average reluctance) and pared with that obtained numerically to investigate the shape of
curved pole surface. In this situation, the crowded flux paths, back-EMF voltages and also to justify the modifications made
where the magnets are virtually in contact, indicate a strong to the model in obtaining the analytical solution.
magnetic field with a significantly large radial component, es- As evident from Fig. 6, the magnetic flux paths are more ra-
pecially in close proximity to the pole surface. However, at the dial and stronger closer to the surface, than further away from
center of the magnet, the flux lines are sparse and, hence, the the magnet where flux paths are weak. The results are in good
magnetic field is weak. But they tend to follow the natural cir- agreement except those around the corners of the magnets at
cular path, resulting in a large radial component as evident from large radii, where the analytically computed radial component
the results. In contrast, the numerically computed flux paths of is noticeably large. This in fact was identified when formulating
a real machine neither emanate from nor enter into the magnet the analytical solution, and is also obvious from the flux pat-
at right angles as a consequence of estimating the actual surface terns of Fig. 5(a) as they emanate at right angles to the pole sur-
potential iteratively using Dirlichlet boundary condition. They face. However, the air-gap field distributions are very much sim-
tend to bend more toward the adjacent magnet at the corners ilar in both cases, a little further away from the magnet surface
and are more radial in the middle of the magnet. This suggests where a practical winding with typical dimensions would be lo-
that in both cases the flux paths with a larger radial penetration cated. This can be attributed to rapidly decaying radial compo-
around the center have a larger contribution to the flux linkage nent of the magnetic field around the corners of the magnets,
in a practical winding than those around the corners, as indi- where magnetic flux paths tend to bend naturally more toward
cated by the winding location with dotted lines. An important the adjacent magnets. These results, therefore, suggest that the
MADAWALA AND BOYS: MAGNETIC FIELD ANALYSIS OF AN IRONLESS BRUSHLESS DC MACHINE 2389

Fig. 9. Flux linkage for different magnet thicknesses.

Fig. 7. Measured versus computed radial flux density distribution.

Fig. 10. Flux linkage for different rotor radii.

used, the thickness of the winding and the magnets , and the
dimensions of the machine. Thus, for a given winding thickness
Fig. 8. Comparison of indirect versus direct flux density. and machine dimensions, there should be an optimum number
of magnets for which the flux linkages become maximum. This
proposed approximate analytical solution can be used for an- effect is illustrated in Fig. 9 by numerically calculating the flux
alyzing the magnetic field inside which a practical winding is linkages for different winding thicknesses (represented as a per-
located, with a reasonable degree of accuracy. centage of rotor radius ) for a rotor radius of 70.75 mm. As
The validity of the developed magnetic model was verified di- evident, the thinner the magnet the higher the optimum number
rectly by measuring the flux density inside the machine by a flux of magnets that gives the maximum flux linkages as weaker
meter and indirectly by locating search coils inside the machine. magnetic flux produced by thin magnets cannot penetrate deep
The comparison of the measured radial flux density distribution into the stator core.
of a typical ILBLDC machine by search coils and that computed The effect of the rotor radius on the optimum pole number is
from the numerical solution, is shown in Fig. 7. As apparent, the shown in Fig. 10 for magnets of 9.5 mm thickness. In this situ-
measured results are in good agreement with those computed ation, the optimum pole number increases with the rotor radius
numerically at different radial distances and the small discrep- since larger rotors results in larger effective air gaps and winding
ancy can be attributed to the assumed infinitely permeable back thicknesses.
iron and the variation of during magnetization.
The comparison of both indirectly and directly measured flux
VI. CONCLUSION
density and the numerically computed flux density is shown in
Fig. 8. Since it is not practicable to use the flux meter to mea- An unconventional ironless BLDC machine design has been
sure the flux density at different radial distances over the pole described. The magnetic field inside the design that has no iron
surface, flux densities along the pole axis instead were measured for guiding flux paths was obtained by using a model based on
for comparison. Results show that directly measured flux densi- the magnetic pole concept. To facilitate an analytical solution,
ties, which can be considered as the closest to the actual values, the magnetic scalar potential instead the magnetic vector poten-
are in very good agreement with the values computed numeri- tial, has been used. An approximate analytical solution has been
cally, verifying the validity of the proposed magnetic model. presented and the results are in good agreement except those at
Flux linkages in a typical ILBL machine largely depend on the corners of the magnets which have only a little contribution
several design parametersparticularly, the number of magnets to the flux linkages, as verified by both a numerical solution and
2390 IEEE TRANSACTIONS ON MAGNETICS, VOL. 41, NO. 8, AUGUST 2005

experimental results. Results indicate that for a given set of di- [6] J. Charpentier and G. Lemarquand, Optimal design of cylindrical
mensions the IL design has an optimum number of poles for air-gap synchronous PM couplings, IEEE Trans. Magn., vol. 35, no.
4, pp. 22672273, Jul. 1998.
which its flux linkages are maximum. [7] M. Markovic, M. Jufer, and Y. Perriard, Analyzing an electromechan-
ical actuator by Schwarz-Christoffel mapping, IEEE Trans. Magn., vol.
APPENDIX 40, no. 4, pp. 18581863, Jul. 2004.
[8] B. Krstajc, Z. Andelic, S. Milojkovic, S. Babic, and S. Salon, Non-
DETAILS OF 200W PROTOTYPE ILBLDC MACHINE linear 3D magnetostatic field calculation by the integral equation method
with surface and volume magnetic charges, IEEE Tran. Magn., vol. 28,
no. 2, pp. 10881091, Mar. 1992.
[9] N. Bouls, Prediction of no-load flux density distribution in permanent
Rotor (pressed cup) magnet machines, IEEE Trans. Ind. Appl., vol. IA-21, no. 4, pp.
633643, Jul./Aug. 1985.
Outer radius mm [10] Z. Q. Zhu, D. Howe, E. Bolte, and B. Ackermann, Instantaneous mag-
Inner radius mm netic field distribution in brushless permanent magnet dc motors, Part 1:
Open-circuit field, IEEE Tran. Magn., vol. 29, no. 1, pp. 124135, Jan.
Length mm 1993.
[11] C. W. Steel, Numerical Computation of Electric and Magnetic
Stator (plastic) Fields. New York: Van Nostrand, 1987, pp. 3334.
Radius mm [12] E. Kreyszig, Advanced Engineering Mathematics. New York: Wiley,
1988, pp. 908917.
Length mm [13] G. F. Carrier, M. Krook, and C. E. Pearson, Functions of a Complex
Variable. New York: McGraw-Hill, 1966, pp. 136150.
Winding resistance Ohms [14] M. Abramowitz and I. A. Stegun, Eds., Handbook of Mathematical
Winding inductance H Functions. New York: Dover, 1972, pp. 555567.
[15] H. Kober, Dictionary of Conformal Representations. New York:
Back EMF constant V rpm Dover, 1957, pp. 5152.
Arnox-H8 Permanent Magnets [16] B. Davis, Integral Transforms and Their Applications. Berlin, Ger-
many: Springer-Verlag, 1978, pp. 110112.
Thickness mm
Width mm Manuscript received March 22, 2005; revised May 17, 2005.
Length mm
Remanent flux density T
Coercivity kA m Udaya K. Madawala (M96) was born in Sri Lanka in 1962. He received
Recoil permeability the B.Sc. degree with honors in electrical engineering from the University of
Moratuwa, Sri Lanka, in 1987 and the Ph.D. degree in power electronics from
Number of magnets the University of Auckland, Auckland, New Zealand, in 1992.
He was employed by Fisher & Paykel Ltd., New Zealand, in 1992, where
he worked as a research and development engineer in the area of motor design
and control. In 1997, he joined the department of Electrical and Computer En-
REFERENCES gineering at the University of Auckland as a Research Fellow and became a
[1] P. Campbell, Permanent Magnet Materials and Their Applica- Lecturer in 2002. His research interests are in the fields of BLDC motor design
tions. Cambridge, U.K.: Cambridge Univ. Press, 1994, pp. 189200. and control, resonant converters, supercapacitor applications, and unbalanced
[2] F. Fouad, T. Nehl, and N. Demerdash, Permanent magnet modeling operation of three-phase induction machines.
for use in vector potential finite element field analysis in electrical ma-
chinery, IEEE Trans. Magn., vol. MAG-17, no. 6, pp. 30023004, Nov.
1981.
[3] R. E. Neubayer, T. Nehl, and N. Demerdash, Continuous field compo-
nents at FE-interfaces due to complex magnetic scalar potential, IEEE John T. Boys graduated from the University of Auckland, Auckland, New
Trans. Magn., vol. 32, no. 3, pp. 639642, May 1996. Zealand, in 1962.
[4] M. A. Alhamadi, Coupled vector-scalar potential method for 3D mag- After receiving the Ph.D. degree, he worked for SPS Technologies (USA) for
netostatic field computations using hexahedral finite elements, IEEE five years before returning to academia where he is currently Professor of Elec-
Trans. Magn., vol. 32, no. 5, pp. 43474349, Sep. 1996. tronics at The University of Auckland. His fields of interests are motor control
[5] P. Elies and G. Lemarquand, Analytical optimization of the torque of a and inductive power transfer. He has a particular interest in the transfer of tech-
PM coaxial synchronous coupling, IEEE Trans. Magn., vol. 34, no. 2, nology from the research laboratory to industry, and is the holder of more than
pp. 10371046, Mar. 1999. 20 patents.

Das könnte Ihnen auch gefallen