Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: A comparison between unbalanced distributed winding and balanced concentrated winding is presented
Received 8 October 2015 in this paper. The application is a torque motor where low speed and high torque are main requirements.
Received in revised form 7 July 2017 Concentrated winding slot/pole combination is selected so that winding factor is maximum and as well as
Accepted 20 November 2017
it is equal to the winding factor of the unbalanced winding motor. Stator and rotor Magneto-Motive Force
Available online 6 December 2017
(MMF) functions are obtained for both motor types to model torque ripple components in the frequency
spectrum. Finite element analysis (FEA) and experimental results are presented to compare back EMF
Keywords:
waveforms, magnetic loading, torque constant and thermal performance of two motors.
Concentrated winding
Direct drive motor
Two designs have identical stator outer diameter (OD) and active axial length. The detailed comparison
Magnetic saturation shows that two motors have similar torque constant. The behavior of two motors under heavy load differs
MMF harmonics and concentrated winding design performs better. Thermal performance of the designs is also compared
Permanent magnet synchronous motor and it is observed that concentrated winding design has also less winding temperature rise.
Torque motor © 2017 Elsevier B.V. All rights reserved.
Torque constant
Unbalanced winding
https://doi.org/10.1016/j.epsr.2017.11.013
0378-7796/© 2017 Elsevier B.V. All rights reserved.
E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74 65
Table 1
Some geometric and electrical data of 39/12 and 36/32 motors.
39/12 36/32
36/32 motor has almost 20% larger rotor diameter than 39/12 motor
has. This case implicates more than 40% higher torque capabil-
ity for the same electric and thermal loading this is well known
from the T = KD2 L sizing equation. However, 40% torque increase
would not be possible because larger rotor diameter reduces stator
Fig 1. An example of a stabilized turret platform: remote weapon station system —
®
MUHAFIZ (Photo Courtesy of ASELSAN Inc.).
winding slot area for the fixed stator OD. Hence the slot current
density of 36/32 motor would be higher than slot current den-
sity of 39/12 motor because of the smaller winding area. On the
problems based on the backlash, reduced efficiency, low rigidity other hand, due to the longer end winding, distributed wound
and mechanical complexity. Direct drive brushless servo motors 39/12 motor has higher phase resistance value than concentrated
eliminate those problems with the help of the direct torque and wound 36/32 motor as given in Table 1. Stator winding area versus
speed transfer to the mechanical system. So, direct drive servo current density versus phase resistance relation is investigated
motors come into prominence when high dynamic performance during thermal analysis for two motors. Analysis indicates that
and high position accuracy is required. In addition to that, direct concentrated-winding 36/32 motor has some advantages for ther-
drive brushless servo motors provide high torque/volume ratio, and mal loading of the machine.
they are integrated into the mechanical platform without mount-
ing elements. As a result making use of direct drive brushless
4. Comparing two winding structures
servo motors imply low weight and low volume turret system.
One example of turret system is shown in Fig. 1. In this paper,
Before starting performance comparison of two motors, it is
design considerations of a direct drive servo motor which is used
worth to give their characteristic features as cited in the literature.
in electro-optical director application.
The concentrated winding is well known for sub-harmonics in its
MMF distribution [4]. Although not commonly used, unbalanced
3. Basic design choices of two motors distributed winding also suffers from the low order magnetomo-
tive force (MMF) harmonic [5,6]. The unbalanced winding is a result
3.1. Concentrated winding motor with 36 slots/32 poles of improper slot/pole combination whose major advantage is lower
combination cogging torque. Sub-harmonics of MMF do not have any contribu-
tion to useful torque production. They may be responsible for many
Concentrated double layer winding is the design choice for the performance deteriorating features such as torque ripple, higher
motor. Shorter end turns and lower overall volume are two major radial force, and additional core loss. As far as MMF harmonics
driving motivations for the selection. 36 slots and 32 pole combi- are concerned, balanced concentrated and unbalanced distributed
nation is selected to maximize winding factor [1]. Pole number is windings resemble each other.
selected as high as possible to increase torque density. Since the Before starting a comparison of two motors, it would be helpful
application requires low-speed operation of the motor, high pole to determine the sub-synchronous and super-synchronous com-
number does not constitute a harmful effect on the core loss and ponents in MMF. In the MMF harmonic content, a number of pole
efficiency. pair is the main space harmonic order. Harmonic orders smaller
than main space harmonic order are called sub-synchronous
3.2. Distributed winding motor with 39 slots and 12 poles components. Similarly, harmonic orders greater than main space
harmonic order are called super-synchronous components.
This motor is currently being used in the application. It was
designed with double layer distributed winding. 39 slots and 12 4.1. Stator MMF comparison
poles for three phase motor yields unbalanced winding.
36/32 motor is designed specifically for the application whereas Stator MMF distribution of the machine with 39/12 slot/pole
39/12 motor is an off the shelf product and adapted for the applica- combination is given in Fig. 2. The results are given for 14.5 Arms
tion. The design of 36/32 motor is driven by a geometric envelope phase current, and 16 turns per coil winding design. Fig. 3, on the
of 39/12 motor which is confined by stator outer diameter (OD), other hand, shows the MMF distribution of the machine with 36/32
inner rotor diameter (ID) and axial length. Other internal geomet- slot/pole combination. This machine has the same effective (RMS)
ric parameters are free variables, and they are optimized during the phase current. But its number of turn per coil is fixed at 18 dur-
design to meet torque output while obeying certain limits of slot ing the design. Fig. 4 shows the MMF harmonic contents of 39/12
current density and magnetic flux density. Because of the inter- and 36/32 machines. It should be noted that Fig. 2 is only valid for
mittent operation of the application, 8 A/mm2 of current density is symmetrical phase current. In this application, the current regula-
observed. Magnetic flux density limit is set to 1.7 T on average by tion algorithm of the motor driver assures the current symmetry.
considering saturation limit of selected magnetic material. Hence, balanced current is considered in the analysis.
Table 1 shows the summary of some key points of two motors. As seen from harmonic contents of two motors, sub and super
As mentioned outer envelopes of two motors are identical whereas harmonics occur in two motors. Fundamental components of 39/12
66 E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74
motor have quite substantial compared to other components. On frequency [9,10]. Furthermore, sub-synchronous space harmonics
the other hand, 36/32 motor has one sub-synchronous, and one have a relatively greater wavelength, and this causes deeper pene-
super-synchronous component and they are comparable in size tration of the flux to the core and further increase in losses [8]. On
with the main components as seen from Fig. 4. As studied in the the other hand, intended applications for the torque motors usu-
literature [7,8], sub-synchronous harmonics in space moves faster ally have very low operating speed. Hence, core and magnet losses
than a fundamental one which can be given mathematically, remain low, and the deteriorating effect of the sub-synchronous
w MMF components on the losses becomes a secondary issue.
wne = (1)
n
where w is stator phase current’s fundamental frequency, wne is 4.2. Rotor MMF comparison
the electrical rotation speed of the nth order space harmonic. For
a 16 pole pair machine, n = 16 is the main space harmonic and all 4.2.1. Calculating rotor MMF magnitude by magnetic equivalent
the harmonic orders with less than 16 are considered to be sub- circuit
synchronous. 1st order (fundamental) harmonic is 16 times faster This section explains the calculation of rotor MMF waveform
than the main component. By using this fact, we can easily conclude and its magnitude using basic magnetic circuit representation. The
that machine with sub-synchronous space harmonics would have method is very well described in the literature [11,12]. Fig. 5 shows
a higher rotor core and magnet losses because of relatively higher the model based on main flux path of one rotor pole pair. Leakage
E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74 67
Fig. 4. Stator MMF harmonic content of 36/32 (blue) and 39/12 (red) machines. (For interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article.)
Fig. 5. Rotor magnetic equivalent circuit for one pole pair in PMSM.
Where Rm : Magnet reluctance, Rg : Airgap reluctance, Ørem1 : North pole remenance, flux, Ørem2 : South pole remenance flux, Ørem1 = Ørem2 = Br .A, Br : Magnet remenance flux
density, A: Magnet radial flux surface area.
Fig. 6. Rotor MMF waveform. Fig. 7. Rotor MMF waveforms of 39/12 and 36/32motors in one electric cycle.
flux is ignored in the model and permeability of the core is assumed By using Eq. (2), r1 and r2 values can be calculated for each motor.
to be infinite. As seen in the equation, magnet thickness (lm ), air gap length (lg ),
By solving magnetic circuit; magnet relative permeability (r ) and magnet remanent flux den-
sity (Br ) effect the result. Table 2 summarizes the values of these
Rm Rg lg /o parameters for 36/32 and 39/12 motors.
r1 = −r2 = Br .A. = Br (2)
Rm + Rg 1+
lg
r39/12
o lm
= 0.8 (3)
r36/32
where
r1 and r2 are ampere turns provided by magnets to the magnetic Two motors have the same magnet embrace ratio of 0.81 in one
circuit, pole pitch. This information and the peak magnet MMF values r1
is magnetic permeability of the magnet and given by = r o . and r2 given in Table 2 are used to plot approximate MMF wave-
Hence the waveform of the one rotor pole pair MMF will be as forms shown in Fig. 7. This study about rotor MMF indicates that
shown in Fig. 6. 36/32 motor has the ability to create higher rotor MMF as given by
68 E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74
Table 2
Comparison of rotor design parameters and rotor MMF magnitude for two motor types.
Motor type Air gap length (lg ) Magnet thickness Magnet relative Magnet remanent flux r1 (A − t) r2 (A − t) Hm = −r/lm (kA/m) Bm (T)
(lm ) permeability (r ) density (Br )
net flux density can be interpreted as higher air gap flux density. If
two motors had the same stator slot and yoke dimensions, some-
one should have expected 5% more flux density in the core. This
may raise the concerns about saturation.
Measurements made for 39/12 motor’s stator yoke thickness
and stator tooth width show that they are smaller than the associ-
ated dimensions of the 36/32 motor. Having given this geometric
comparison, and by using the operating point summary given in
Table 2 and Fig. 8, someone could easily reach the conclusion that
the core of the 39/12 motor has the tendency to saturate even before
the armature reaction starts. With deep armature reaction under
heavy load conditions, saturation would become more severe. This
is validated by finite element analysis in the upcoming sections.
Eq. (3). For the geometric dimensions of two motors given in Table 2,
a simple 1-D calculation is performed to figure out the operating 4.3. Torque output comparison
point of the magnets. Calculated operating points of two motors
are shown in Fig. 8 where M1 (Hm1 , Bm1 ) and M2 (Hm2 , Bm2 ) are Stator and rotor MMF harmonic content can be used to calculate
the magnets operating points of balanced, concentrated winding two motors’ torque output by applying Lorentz Force Law as in Ref.
and unbalanced distributed winding motor, respectively. Although [13].
36/32 motor has higher magnet MMF magnitude effective in its
magnetic circuit, 39/12 motor has almost 5% higher magnet flux
density effective in its magnetic circuit. This is due to shorter air P o
T1 = kstk kskew rg lstk fs,1 fr,1 sin (d ) (4)
gap length and longer magnet length in 39/12 motor. Higher mag- 2 g
Fig. 9. Rotor MMF harmonic content of 36/32 (blue) and 39/12 (red) machines. (For interpretation of the references to colour in this figure legend, the reader is referred to
the web version of this article.
E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74 69
Fig. 10. Back EMF waveforms and harmonic contents of 39/12 (upper) and 36/32 (lower) motors.
P o
Tripple = −kstk kskew rg lstk hfs,h fr,h sin ((h ± 1) ωe t ± d ) (5) ripple in torque although their spread in the spectrum is signif-
2 g
icantly different. On the other hand, average torque comparison
shows that 39/12 motor has approximately 6% higher output than
h = 6m ± 1
36/32 motor has.
m = 1, 2, 3, . . .
where
P: number of poles 5. Comparison of basic performance metrics
g: air gap length
rg : air gap radius 5.1. Back EMF harmonics
lstk : stack length
fs,h and fr,h : hth order stator and rotor MMF harmonics, respec- Back EMF data is an important metric for the performance. High
tively. h = 1 for the main component back EMF constant as V/(rad/s) is an indicator for high torque con-
kstk : stacking factor of laminations stant. Hence this may be interpreted as high torque density motor.
kskew : coefficient to account for the effect of skewing Measurements show that 39/12 motor has slightly higher back EMF
d : spatial angle between main stator and rotor harmonic. It is constant than 36/32 motor. Fig. 10 shows back emf waveforms and
taken as 90 electric degrees by assuming no flux weakening. harmonic contents for two designs at around 100 Hz electrical fre-
Based on the harmonic contents calculated in the preceding sub- quency. 36/32 design is skewed by 0.5 slot pitch whereas 39/12
sections and also by using Eqs. (4) and (5), Table 3 is prepared to design has no skewing. The effect of skewing on 36/32 design has
show the torque output of two motors. been quantified by comparing back emf waveforms of skewed and
Tripple is defined for the first four harmonic torque components non-skewed motors. The ratio of back emf peak values has been
as found to be 0.88. This number has been used as kskew in the torque
calculations.
Tripple = T6 2 + T12 2 + T18 2 + T24 2 (6)
As far as back-EMF harmonics are concerned, concentrated
Table 3 shows two motors have different ripple components in winding has lower harmonic content. Fig. 10 shows back EMF har-
their torque spectrums. 39/12 motor’s overall torque ripple defined monic content of two motors. It should be noted here, there is a DC
by Eq. (6) is approximately 37% higher than 36/32 motor’s overall component (meaning zero order harmonic) in the back-EMF wave-
torque ripple. As a result, two motors have comparable cumulative form. The reason is measuring the waveform with non-calibrated
70 E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74
Table 3
Average torque and torque ripple comparison of two motors for 14.5 RMS phase current.
Where T1 is the average torque and T6 , T12 , T18 , T24 are the harmonic torque components.
Fig. 11. (a) Cogging torque variations of 36/32 (PMSM-1) and 39/12 (PMSM-2) motors at 400 rpm. (b) instantaneous torque variations of 36/32 (upper) and 39/12 (lower)
motors at 32 rpm.
digital oscilloscope. In the reality, generating DC component in be used in the application. 36/32 motor slot pole combination also
PMSM is not possible. offers significantly low cogging torque. Fig. 11a shows the exper-
imental cogging torque comparison. Similar argument is valid for
the torque ripple comparison of two motors. Lower harmonic con-
5.2. Cogging torque and torque ripple
tent in back EMF means lower torque ripple as cited by many
researchers [13–15]. From this point of view, we can conclude
39/12 slot/pole combination is wound by unbalanced winding
concentrated winding design has lower torque ripple as well. This
which can be considered as a disadvantage. On the other hand, it
statement has also been validated in the preceding subsection by
offers very low cogging torque hence it is the only motivation to
E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74 71
Table 4
Comparison of torque constant in concentrated and distributed winding motor.
Two motors are compared for torque constant values. During the
test, the rotor position is aligned to d axis. After position alignment,
the shaft is locked then the motor is tested. By giving controlled
current to the q axis, torque is observed with torque transducer.
The test system is shown in Fig. 12. The torque–current relationship
for the magnetically linear case can be expressed with Eq. (8). For
magnetically nonlinear (saturated) case, current torque relation is Fig. 16. Flux density map of 36/32 motor for Iq = 10 A current.
expressed as in Eq. (9).
situation may occur in torque motor applications. The underlying
T = Kt irms (8)
reason for the performance difference could be armature reaction
100 − x and its effect on the magnetic saturation [16]. Finite element anal-
T = Kt ( )irms (9)
100 yses have been performed to compare magnetic loading of two
Here, x is the percentage torque reduction due to magnetic sat- motors. Figs. 15 and 16 show flux density map for normal load-
uration. Kt represents the torque constant which is measured by ing conditions where both motors are excited with Iq = 10 A q axis
injecting small current through motor windings so that magnetic current.
saturation does not occur. Comparison of two motors’ Kt values is As seen from flux density maps, 39/12 motor subjects heavy
given in Table 4. saturation in the stator back iron (yoke) even for normal loading
For normal loading conditions, torque constant of 39/12 motor conditions. Heavy loading conditions aggravate the saturation. This
is a bit higher than 36/32 motor’s torque constant as shown in is shown in Figs. 17 and 18 for 40 A load where flux density maps of
Figs. 13 and 14. two motors are given for 39/12 and 36/32 motors; respectively we
From Figs. 13 and 14, 36/32 design has higher torque constant can conclude that 39/12 motor has already been designed with high
than 39/12 motor under heavy load conditions.This is an impor- magnetic loading, and its torque constant gives away its linearity
tant aspect in comparing two motors hence frequent heavy loading as electric loading increases with rising mechanical load.
72 E. Mese et al. / Electric Power Systems Research 156 (2018) 64–74
Fig. 19. Pictures of motors under investigation. (a) 39/12 distributed winding motor
(b) 36/32 concentrated winding motor.
Fig. 20. Steady-state thermal analysis results of 39/12 and 36/32 motor.
Fig. 21. Steady-state thermal results of (a) 39/12 and (b) 36/32 motors.
Acknowledgements
References
[7] N. Bianchi, E. Fornasiero, Impact of MMF space harmonic on rotor losses in [12] J. Stepina, Matrix analysis of space harmonics of asymmetrical stator
fractional-slot permanent-magnet machines, IEEE Trans. Energy Convers. 24 windings, IEE Electric Power Appl. 134 (Pt. B, July (4)) (1987).
(June (2)) (2009) 323–328. [13] D.C. Hanselman, Minimum torque ripple, maximum efficiency excitation of
[8] N. Schofield, K. Ng, Z.Q. Zhu, D. Howe, Parasitic rotor losses in a brushless brushless permanent magnet motors, IEEE Trans Ind. Electron. 41 (3) (1994)
permanent magnet traction machine, in: Eighth International Conference on 292–300.
Electrical Machines and Drives (Conf. Publ. No. 444), 1–3 Sep 1997, 1997, pp. [14] J.Y. Hung, Z. Ding, Design of currents to reduce torque ripple in brushless
200–204. permanent magnet motors, IEE Proc. B 140 (4) (1993) 260–266.
[9] F. Magnussen, H. Lendenmann, Parasitic effects in PM machines with [15] M.S. Islam, S. Mir, T. Sebastian, S. Underwood, Design considerations of
concentrated windings, IEEE Trans. Ind. Appl. 43 (September–October (5)) sinusoidally excited permanent magnet machines for low torque-ripple
(2007) 1223–1232. applications, IEEE Trans. Ind. Appl. 41 (4) (2005) 955–962.
[10] J.R. Hendershot, T.J.E. Miller, Design of Brushless Permanent Magnet Motors, [16] R.P. Deodhar, D.A. Staton, T.J.E. Miller, Variation of torque constant with
Magna Physics Publishing and Clarendon Press, Oxford, 1994. armature current in brushless PM motors, in: Seventh International
[11] H. Seok-Hee, T.M. Jahns, W.L. Soong, M.K. Guven, Illindala, MS, Torque ripple Conference on Electrical Machines and Drives (Conf. Publ. No. 412), 11–13 Sep
reduction in interior permanent magnet synchronous machines using stators 1995, 1995, pp. 405–409.
with odd number of slots per pole pair, IEEE Trans. Energy Convers. 25 (March
(1)) (2010) 118–127.