Review of Sensorless Methods for Brushless DC
James P. Johnson M. Ehsani Yilcan Giizelgiinler
Member, E E E Fellow, IEEE
Power Electronics Laboratory
Texas A&M University
College Station, Texas
Abstract
This paper provides a review of the lireratiire addressing
sensorless operation of brushless Dc machines. Methods
explained ore stare-of-thenrr, including rhose I . using measured
currents, voltages.findomental machine equations. and algebraic
manipulations. 2. using observers, 3. using back-emf methods, 4.
sensorless starting rechniqices, and 5. with novel techniques not in
the previous categories.
1. Introduction
The purpose of this paper is to provide a review of
sensorless methods of BLDC. This paper will focus on the
analytical and qualitative basis of these methods.
2. Sensorless Review
Many reviews of sensorless operation of AC drives
have been seen in the literature [1,18,27,32,53]. Of
publications presently available in IEEEi and TEE conference
proceedings and transactions, past sensorless techniques have
been grouped into 5 categories:
1 . those using measured currents, voltages, fundamental
machine equations, and algebraic manipulations,
2. those using observers,
3 . those using back-emf methods,
4. sensorless starting techniques, and
5. those with novel techniques not falling into the previous
four categories.
2.1. Methods Using Measrirahles arid Math
The first type above consists of three sub-types, 1.
those using the voltages and currents to calculate flux
linkages, 2. those using the difference between a model’s
prediction of a voltage or current and the actual value, and 3.
those using the machine equations, measurables, known
machine parameters, and algebraic manipulations to calculate
position and speed. The first sub-type is seen in
0-7803-5589-W99/$10.00 0 1999 IEEE 143
Member, IEEE
[1,2,21,26,34,40,49,64]. The voltage equation of the
machine,
d
V=Rl+-v
dt
where,
V i s the voltage vector,
I is the current vector,
R is the resistance matrix, and
Y is the flux linkage vector.
This equation is then manipulated to obtain,
t
. v = I( V - R I ) dt.
0
Knowing the initial position, machine parameters,
and the relationship of flux linkage to rotor position, the rotor
position is estimated. Determining the rate of change of the
flux linkage from the integration results, the speed is
determined [34]. A variation includes using the previous
position data and polynomial fitting, extrapolating to obtain
the next step position prediction [21]. An advantage of the
flux-calculating method, as shown in [34] is that line-line
voltages may be used in the calculations and thus no motor
neutral is required. Others, have combined the flux-
calculating method with advanced control strategies such as a
state observer to compute the speed feedback signal [Z], an
adaptive control [26], and H2 and H? controls [40].
The second sub-type was published primarily by one
author, N. Matsui, in [17,20,39,45,70]. Using a d-q model of
the machine, the actual dq transformed currents and voltages,
and those on a hypothetical axis offset from the dq axis by a
small angle Ae, the output voltages of the model, on the
hypothetical axis, and those on the actual dq axis were
compared. The differences between the calculated voltages of
the hypothetical axis considering A9 = 0, and the actual dq
axis voltages on the hypothetical axis,
AV= Vdq - VhYpoh&al-dq = KEesin(A6).
yielded 88, the actual change in rotor position from the
previously known position. as with AI3 = 0, A8 = AV.
The third sub-type uses machine parameters and
equations, measurables, and algebra, reference frame theory
and transformations to calculate position and speed [25].
Initially the measured voltages and currents are transformed to
rotor and stator d-q reference frames. Relations between
stator and rotor reference frames, denoted with superscripts s
and r, respectively, are
as would be the actually measured variables as used in a
closed-loop system control. Specific rules apply for the
various types of observers. The interested reader is referred
to [73]. An example of a single output observer, from [73], is
as follows:
Consider the discrete-time state-space form of a single-output
system,

x[k + I]= Ax[k] + B u l k ]

y [kl = C x[ k] + D u[k]
ii = ii sine, - ii coser
which must be completely observable [73].
These relations allow the substitution of stator reference frame Next, a modal matrix must be determined for the system. A
quantities in the rotor reference frame voltage equations. good reference to obtain the knowledge of this procedure is
Once the substitutions are made and the equations relating
[74]. Using the modal matrix, P, the system is transformed
rotor and stator reference frames are totally in terms of stator into an observable form:
variables, in the stator frame, abc frame equivalents are
substituted in these equations, which are then manipulated to
yield rotor angle. Knowing that the machine is non-salient, x[k]= F'x'[k]= Q-lx'[k] ~

the equivalent rotor flux, and manipulating the same x'[k] = P-'x[k] = Qx[k]
equations, the rotor speed is also determined. The final
equations for position and speed are as follows:
x'[k + 11= A'x'[k] + B'u[k]

1
d . y[k]=ctTx'[k]+DTu[k]
v b - v y c - R ( i b -i,)-Ld .-(th - i C ) - & m ( L q - L d ) i ,
Br = ta,,i' dt
- Ria - Ld -dld- i., ) + a ( L q - I d ) ( & -ic)
where
A'=QAQ-'
B = QB
c' T = cT Q -1
Next, the full-order observer is designed using the designers
choice of eigenvalues. The eigenvalues are usually chosen to
as defined in [ZO]. Variations of this are found in
[20,23,3 1,42,43]. be slightly faster (of a larger negative value) than those of the
actual system so the state estimate error approaches zero as
time approaches infinity. If the observer eigenvalues are
2.2. Methods Using Ohsmws
chosen to be an order of magnitude larger, or more, than the
actual system's, then the estimated values will converge to the
The second type listed above determine rotor actual values within a sufficiently short time [24]. The
position and/or speed using observers. The Kalman filter is observer takes the form
used as a position estimator in [12,13,14,28,46,50]. One of
the earliest of these was by M. Schroedl in 1988. Utilizing c'[k + I] = F'<'[k]+g'y[k]+Hu[k]
various methods of measuring system voltages and currents
where
rough estimates of the angular rotor position were produced.
Kalman filtering added refinement to the first estimates of F1=A'-g'cCT
position and speed. For additional information on Kalman H'= B-g'dT
filtering see [73], and those papers listed. Other observer-
based systems include those using non-linear [29.36,41]. full- with the observer output equation
order [2,33,38], disturbance [ 1,5,39], reduced-order [24,41].
and sliding-mode observers [3,16,47,48]. Observer theory w[k] = Pc[k] = Q-'Y,[k]
can be found in many texts and papers. The fundamental idea
of the basic full-order state observer was developed by David which allows the transformations
Luenberger of Stanford University in 1964 [73]. Observers
take as inputs the outputs and inputs of the actual system.
Typically, the state of the system is produced as the output.
The observer-produced state is then fed back into the system

144

which gives
<[k + 11 = FC[k]+gy[k]+ Hu[k]
where
g=Q-'d
F=Q- IFQ=Q- 1R Q - Q - ~ ~ ~ T Q = A - &
H=Q+= Q-'B&dT =B-gT
which is a full-order observer in x[k], meaning that <[k]
converges to x[k].
Disturbance observers implemented in the cited
papers, were reduced-order observers used with system states
which were assumed quasi-static (fast sampling and
calculating times). The slight variations of the state variables
from one sampling to the next is termed a disturbance, which
results in observations of small changes in rotor position
angles. Reduced-order observers depend on some of the
system outputs being linear transforhiations of the system
states. By coupling some of the system states to the output of
the observers through linear transformations (or actual
measurements if the states are readily available for
measurement), only the states of interest need be estimated by
the lower-order observer. This reduces the computational
burden of the observer [41]. In [1,5,24], reduced-order
observers estimate the back-emfwhich is used to obtain rotor
position and speed. In [41], the rotor position and speed are
estimated, as the outputs of a non-linear reduced-order
observer. Sliding-mode observers, offspring of sliding-mode
control, utilize a valying switching function in order to confme
the state estimation error to move toward zero on a phase-
plane sliding surface [3]. In [16] and [47], the sliding-mode
observers were based on the observation of d-q transformed
stator currents. From the resulting current estimations, the
back-emfs were estimated, producing speed and position
estimates. In [48], the full-order system model was utilized
and a reduced-order sliding-mode observer designed. In [3], a
full-order sliding-mode observer was described, and
modifications were made in order to overcome chattering
problems, inherent in sliding-mode systems. In general, the
sliding mode system starts with the state-space system
[3,16,47,48]
x = f(X,U),
Y=cx
where the usual definitions are made. In BLDC systems
(usually considering the PMSM or the fundamental of the
145
trapezoidal BLDC), X consists of the d-q transformed stator
currents, speed, and position, i.e. [id , 4 , 0,8 1, or just the
currents, as previously stated. The state observer takes the
same form with the addition of a term containing the sign of
the estimation errors:
X = f ( k , u ) - K .sgn(s),
where
the hat symbol denotes estimated values,
K is the variable gain matrix,
S is the vector of state estimation errors, i.e.
[ i ,-x1,?* - x 2 ; . . , i n -Xn]T.
n is the observer order, and
the superscript T denotes a transpose operation. In
order for the observer to operate in the sliding-mode, the
following condition must be met:
I d
--s T s<o
2 dt
Defining X - X = e, where e is the state estimation error, the
general error model and dynamics are described by
L = f'(e,U)+K.sgn(e)
where
f ' is the system representing the difference
f ( k , L J - f ( X , U ) , and disturbances due to parameter
variations have been neglected. Now, in order for
convergence, K must be continuously selected to insure that
e T e < O , a s -I- Sd T S = e T e
2 dt
2.3. Methods Usiiig Back-Emf Sensing
2.3.1. Terminal Voltage Sensing
In field-oriented operation of the BLDC, phase hack-
emf is aligned with phase current. Switching instants of the
converter can be obtained by knowing the zero-crossing of the
back-emf and a speed-dependent period of time delay [75].
The BLDC motor has a trapezoidal flux density distribution in
the air-gap, and thus the induced back-emf in the stator
windings is trapezoidal. Monitoring the phase hack-emf when
the particular phase current is zero (the silent phase), the zero
crossing is detected. Low pass filters are used to eliminate
higher harmonics in the terminal voltages. The zero-crossing
instants are decoded and appropriately time-shifted to produce
switching patterns.
Since back-emf is zero at rest and proportional to
speed, the terminal voltage sensing method is not possible at
low speeds. As the speed increases, the average terminal
voltage increases, and the frequency of excitation increases.
Capacitive reactance in the filters varies with the frequency of
excitation, introducing a speed-dependent delay in switching
instants, which disturbs current alignment with the back-emf
and field orientation thus, causes problems at higher speeds.
With this method, a reduced speed operating range is normally
used, typically around 1000-6000 rpm.
Another method which similarly has a narrow speed
range is a method using phase-locked loop circuitry. The
phase-locked loop has a narrow speed range due to the
capabilities of the phase detector, and is sensitive to switching
noise. This method, provides gating signals much like hall-
effect sensor outputs by using a phase-locked loop circuit
which locks onto the back emf of the inactive phase during
every 60' period.
2.3.2. Third Harmonic Back-emf Sensing
The third harmonic of the back-emf can be used in
determination of the switching instants in the wye connected
120 degree current conduction operating mode of the BLDC
mo:or [37]. This method is not as sensitive to phase delay as
the zero-crossing method, as the frequency to be filtered is
three times as high. The reactance of the filter capacitor in
this case dominates the phase angle output of the filter more
so than with the lower frequency. On the other hand, greater
than 213 stator winding pole pitch is required. Provided that
this condition is satisfied, the third harmonic of the rotor flux
can link the stator winding and induces a third harmonic
voltage component which can be detected between the motor
neutral and the negative inverter bus terminal.
The terminal voltage equation of the BLDC motor
for phase a can be expressed as
Back-emf voltage, ea, is comprised ofmany voltage harmonic
components,
e, =E(cosoet+k~cos3w,t+kgcosSw,t+k-jcos7wet+ ...)
The third harmonic of the terminal voltages is acquired by the
summation of the terminal voltages.
Vas+Vbs+Vcs =3Ek3CoS3w,t+vhighfreq.
V 3 + Vhq$ freq.
146
The summed terminal voltages contains only the third and the
multiples of the third harmonic due to the fact that only zero
sequence current components can flow through the motor
neutral. This voltage is dominated by the third harmonic,
which does not necessitate much filtering. To obtain
switching instants, the filtered voltage signal which provides
the third harmonic voltage component is integrated to find the
third harmonic flux linkage.
h r 3 = jv3dt
The third harmonic flux linkage lags the third harmonic of the
phase hack-emf voltages by 30 degree. The zero crossings of
the third harmonic of the flux linkage corresponds to the
commutation instants of the BLDC motor. To acquire correct
commutation instants, sensing the positive or negative going
zero-crossing of the back-emf is essential
The third harmonic method provides a wider speed
range (100-6000 rpm) than the zero-crossing method, does
not introduce as much phase delay as the zero-crossing
method and requires less filtering.
2.3.3. Freewheeling Diode Conduction
This method uses indirect sensing of the zero-
crossing of the phase back-emf to obtain the switching
instants of the BLDC motor [6S]. In the 120 degree
conducting wye-connected BLDC motor, one of the phases is
always open-circuited. For a short period after opening the
phase, there remains phase current flowing, via a free-
wheeling diode. This open phase current becomes zero in the
middle of the commutation interval, which corresponds to the
point where back-emf of the open phase crosses zero. The
biggest downfall of this method is the requirement of six
additional isolated power supplies for the comparator circuitry
for each free-wheeling diode.
By this technique, 45-2300 rpm sensorless operation
has been achieved. This technique outperforms the previously
mentioned back-emf methods at low-speeds.
2.3.4. Back-emf Integration
In this method position information is extracted by
integrating the back-emf ofthe unexcited phase [30.52.66,69].
The integration is based on the absolute value of the open
phase's back-emf. Integration ofthe back-emf starts when the
open phase's back-emf crosses zero. A threshold is set to stop
the integration which corresponds to a commutation instant.
As the back-emf is assumed to vary linearly from positive to
negative (trapezoidal back-emf assumed), and this linear slope
is assumed speed-insensitive, the threshold voltage is kept
constant throughout the speed range. If desired. current
advance can be implemented by changing the threshold. Once
the integrated value reaches the threshold voltage, a reset
signal is asserted to zero the integrator output. To prevent
the integrator from starting to integrate again, the reset signal
is kept on long enough to insure that the integrator does not
start until the residual current in the open phase has passed a
zero-crossing. This type of control algorithm has been
implemented commercially. The integration approach is less
sensitive to switching noise, automatically adjusts to speed
changes, hut the low speed operation is poor. With this type
of sensorless operation scheme, up to 3600 rpm has been
reported [69].
2.1.Starling Techniques
To operate a BLDC without a rotor position sensor,
requires some method of starting the motor. Published
startup methods have belonged to one of five general types: 1
open-loop, 2. with known initial position, 3 using a machine
interrogation and sibqal processing, 4. using computationally
complex methods, and 5. those relying on a winding
inductance which varies with rotor position due to saliency.
2.4.1. Open-loop Starting
Open-loop [29,65,69] starting the motor is
accomplished by providing a rotating stator field which
increases gradually in magnitude and/or frequency. Once the
rotor field begins to become attracted to the stator field
enough to overcome friction and inertia, the rotor begins to
turn and the motor acts as a permanent magnet synchronous
machine. This type of system may be satisfactory in certain
applications, i.e. pump and fan drives. The disadvantage of
this method is that the initial rotor movement direction is not
predictable. When the stator field becomes just strong
enough, the rotor could move in either direction. If then the
speed of the stator field is slow enough, and the load torque
demanded does not exceed the pull-out torque, the motor will
operate synchronously in the desired direction. Initial
direction in certain applications such as disk drives strictly
require unidirectional motion, and thus this method is
inadequate. Another problem exists if the stator field is
rotating at too great a speed when the rotor field picks up.
This causes the rotor to oscillate, requiring the stator field to
decrease in frequency to allow starting.
2.4.2. Starting Methods Using a Known Initial Position
Methods which require a known position
[1,48,64,72] to start the motor suffer from an additional
constraint, e.g. a detent in which the rotor always freely stops.
or a stopping circuit (and initial positioning) which always
provides an initial position for the next movement. A simple
way of setting an initial position is to excite one phase to
cause the rotor to shift and lock into position.
147
2.4.3. Interrogation Techniques for Starting
Machine interrogation techniques [7,8,35,70] offer
an excellent solution for drives where absolute position
information is needed at startup. In these techniques, initial
rotational direction is controlled. Saliency of the interrogated
machine enhances the ability to detect rotor position. Usually
voltage pulses are imposed on the machine terminals while
monitoring the current peak or level after a time duration from
the start of the pulse, often a period approximately equa! to
the electrical time constant of the windings. The flux of the
permanent magnet, either aids or opposes the flux due to the
pulse and the current level indicates their interaction.
Possibilities include using the sign of the difference in the
current levels providing a sign vector which is compared to
values in a lookup table to obtain position information
[7,8,32]. Alternatively, interrogation of the phase
combinations with a salient machine using short (non-
saturating) and long (saturating) pulses has been reported
[35,70].
2.4.4. Control System Techniques for Starting
In [28], ti Kalman filter estimates rotor position from
measured voltages and currents. The machine equations of
non-sinusoidal waveforms are in a d-q stator reference frame.
At zero-speed the voltages and currents are zero and alone the
Kalman filter cannot provide rotor position information. By
injecting a d-axis current this condition is overcome and
position information at standstill is obtained.
2.4.5. Starting Using Machine Inductance Saliency
Saliency in permanent magnet machines is usually
present due to the use of interior permanent magnet (PM)
machine design. The interior PM allows a shorter air gap
which increases the flux intensity on the stator due to the
rotor. A thin bridge which usually separates the magnet from
the gap is often saturated during normal operation, increasing
the reluctance of the radial area where the magnets ( CI = 1 )
already exist, This allows a varying inductance profile for the
motor which depends on rotor position. Sensorless
operations utilizing the relationships between voltage pulses,
current amplitudes, rate of current rises, etc. have often been
used in switched reluctance and synchronous reluctance motor
drives [54-691 where saliency is inherent, and are applicable in
interior PM BLDC drives as well.
2.4.6. 1.C.s for Sensorless BLDC Control
In the past there have been several integrated circuits
which enabled sensorless operation of the BLDC. These
included Unitrode's UC3646, Microlinear's ML4425, and
Silicon Systems, Inc. 32M595. Each of these used back-emf
methods and open-loop starting.
2.5. Unique Sensore'* Techniques
The following sensorless methods are unique. These
range from artificial intelligence methods to variations in the
machine structure. There are few which fall into this class, and
therefore each is briefly discussed
A hackpropagation neural network (BPN) was
trained to provide a non-linear mapping between measured
voltages and currents, and rotor position [6]. This rotor
position was then used with the equations of section 2.1. to
calculate the flux linkage vector. This vector was compared
to the estimated flux linkage vector using the equations of
section 2.3.1. and the measured voltages and currents. The
difference was used as the error which was backpropagated to
modify the BPN weights. An initial flux linkage vector (initial
rotor position) was required to he known a priori.
A paper using fuzzy logic [ 8 5 ] , proposed two fuzzy
logic sub-systems. Using the conventional equations
consisting of phase voltages and currents, the rotor position
was calculated. With the knowledge of the relationships
between these measurahles and the rotor position, a fuzzy
Mamdami type system was developed to produce rotor
position estimates. It was noted that this could as easily been
accomplished with look-up tables, however, for the desired
resolution the size of the lookup tables becomes
unmanageahly large. The second fuzzy system used took as
input the estimated rotor position and produced reference
current values for two different drive strategies: unity power
factor, and maximum torque per amp.
In [15], a sensorless method was developed using the
calculation of flux linkage by integrating the back-emf The
motor was at the end of a 10 km cable. A difficulty in this
sensorless method was that the hack-emf had high-frequency
oscillations superimposed on the voltage waveforms at points
of current commutations. This was overcome by integrating
the back-emf over the whole 60' silent phase period which
produced pulses with amplitudes proportional to the speed
and at positions where the peak of the pulses was 30" prior to
a commutation instant.
Solutions to sensorless operation of BLDC by means
of alterations in machine design were shown in [68,71]. In
[68], an additional stator lamination with equally spaced slots
around the periphery was added to the machine. Each of the
equally spaced slots contained a small sensing coil. The local
magnetic circuit variations for each of the sensing coils is
effected by the permanent magnet rotor's position. A 20 kHz
signal was injected through the coils, and the signal distortions
were analyzed at the terminals of the sensing coils. The
second harmonic detected provided position information. h
artificial saliency was created in [71], by attaching small pieces
of aluminum to the surface of the permanent magnets allowing
148
eddy currents to flow (in the aluminum). The flow of eddy
currents in the aluminum acted to increase the reluctance of
the magnetic circuits creating saliency.
Two papers describe sensorless methods which rely
on irregularities in the measurahles [54,67]. These methods
utilize small deviations, or inherent properties of the shapes of
the current waveforms to obtain position information or
enable a control which tends to provide a phase current /
phase back-emf alignment. The method in [67] considers the
shape of the D.C. bus current. The authors show that the
shape of the D.C. bus current varies depending on the
alignment of the phase current and phase back-emf A clear
difference in the distortion patterns is seen between a leading,
aligned, and lagging operation. The method of control is to
count the number of samples on a consecutive rising and
falling half-cycle of the 6-pulse D.C. current ripple. When
these are equal, the machine is operating at unity internal
power factor. When unaligned, the difference in the rising and
falling counts is used as a multiplier in the D.C. link voltage
incremental update. The machine is operated a; a
synchronous machine with this control effort to maintain
phase current / back-emfalignment.
The second method, in [54], works on the principle
of machine saliency. In this paper, the crest of the current
waveform, which is controlled by hysteresis chopping, is the
medium of rotor position information gathering. The
elementary equations of the rise and fall of current in the
hysteresis waveform were examined. In these equations, it is
easily seen that there is a dependency on the instantaneous
inductance, as well as other machine parameters. Utilizing a
fast sampling and these equations, while considering the other
machine and drive parameters constant, and with measured
current and voltages, the instantaneous inductance, and thus
position is determined. This is possible only if the machine
has inductance saliency.
3. Conclusion
A review of prior work in rotor position sensorless
methods for brushless D.C. has been presented. The basis of
these techniques has been presented in an effort to provide a
reference for utilizing these methods. A new method of
sensorless operation which builds on those presented is
revealed in another paper in these proceedings,
Acknowledgments
The support of Texas Higher Education Coordinating
Board Advanced Technology Program, TRW Electronic Controlled
Steering, Bodine Electric Company, General Motors Research
Laboratories, and Texas Insments Digital Control Systems
Division, for this research is gratefully acknowledged.
 References
I. T. Senjyu and K. Uezato, “Adjustable Speed Control of B ~ s h l e s sDC
Motors without Position and Speed Sensors”, Inl’l. IEEWAS Con/
Proc. on Industrial Auranation and Conirol: Emerging Technologies,
pp. 160-164, 1995.
2. A. Consoli, S. Musumeci, A. Raciti, and A. Testa, “Sensorless Vector
and Speed Control of Brushless Motor Ixiws”, IEEE Trans. on
Indurhld Electronics, vol. 41, pp. 91-96, F e b w 1994.
3. Y.Kim, J. Ahn,W. You, and K. Cho, “A Speed Smsorless Vector
Control for Brushless Dc Motor using Binar). Observer‘’, Proc. o/the
1996 IEEE IECOV 2Tdbf‘I. Conf on lndushial Electronics. Control,
mdbsnumenrafion, vol. 3,pp. 1746-1751, 1996.
4. A. Lonsdale, N. Schofield, “An Integrated Low Cost Sensor for the
Direct Torque Control of B ~ s h l e s sDc Motors”, IEE Colloquium an
Machines for Automotive Applications (Digest No. 1996/166), pp. 611-
6n, 1996.
5. T. Senjyu, M. Tomita, S. Doki, and S. Okuma, “Sensorless Vector
Conml of Brushleis Dc Motors using Disturbance Observei’, PESC
‘9sRecord, 26” Annual IEEE Power Electronics Specialisis Conf, vol.
2, pp. 112-771, 1995.
6 F. H w g and D. Tien. “A Neural Network Approach to Position
Senrarless Control of Brushless I X Motors”, I’roc. of the 1996 IEEE
IECON 2rtdInt’l. CO,$ on Industrid Eleclro~ics, Control. ond
In.rtnmentution, vol. 2, pp. 1167-1 170, August 1996.
7. L. Cardoletti and A. Carsat, “Sensorless Closed-Lmp Control of a
Brushless I X Motor from Start-Up to Nominal Speed”, San Jose,
California: Microlinear Corporation, pp. 74-82, 1994.
8. Microlinear Corporation, “U.S. Patent #SO01405 : Position Detection
for a Bnrshless DC Motor”, San Jose, California Microlinear
Corporation, 1994.
9 M. Schr6dl. “New Rotor Position Detector for Permanent Magnet
Synchronous Machines llsing the “lnform”-Methad“, ETEP Vol. 1, No.
1, JanrwlylFebruar). 1991
IO. M. Schcdl, “Sensorless Control of Reluctance Machines at Arbitraly
Operating Conditions Including Standstill“, IEEE Transactions on
Powcr Electronics, Vol. 9, No. 2, March 1994.
I I. M. Schradl, “Sensorless Convol of Induction Motors at Low Speed and
Slimdstill“, Proceedings of Int’l Conference on Electrical Machines, Vol.
3, Septnnber 1992, Manchester University
12. M. Schrm,“Digital Implementntion of 8 Sensorless Control Algorithm
for P ~ m n e n Magnet
t S ~ ~ C ~ ~Motors“, O ~ O UPS~ o c h . i‘I. CO@ “SW
100”. ETH Zirich (.S’wirrprlandi,pp.430435, 199 I
13. M. Schrodl,“Operation ofthe Permanent Mamet Synchronous Machine
Without a Mechanical Sensor”, IEE /’roc. Inl’l. Conf on Power
electronics m i 1,brinble S p e r d M v r r , pp. 5 1-56, July 1990.
14. M Schodl, “Sensorless Control of Permonent Magnet Synchronous
Motors”, Elrchk Machines md Power ~Sysiemr,vol 22, pp. 173-185,
1993.
15. E. C. Andresen, J. Xie, and C. Topfer, et al., “A Shearing-Wave
Generator for Operation in Deep Bore Ifales”, IEEE Trams. on Indusir),
.4pplicatinn.~,WI. 28, pp. 434440, ~ a r c h / ~ p 1992. nl
16. T. Puruhashi, S. Sangwongwanich. and S. O k m s , “A Position-and-
Velocity Sensorless Control for Brushless Dc Motors using an Adavtive
Slidmg~Mode O b m e r ” . IEEE Trms. on Indutrial Ele&mics,‘ vol.
39, pp. 89-95, A p d 1992.
17. 1’. TAcshita and N. Matsui. “Senrorlsrs Brushless DC Motor Urivc
with EMF Constant ldsntifiei’, IEEE btl’l. Conf an lndurrrid
i o ~1,~ pp.
Eleclronics, Conlrol, m ~ d l n s i ~ n ~ e , i i u tml. , 14-19, 1994.
18. P. W. Lee and C. Pollock, “Sensorless Commutntion Techniques for
Ilrushless I’ermancnl-Mqmr and Ilelucwn~.c Motor Drives’’, 1E.K
Colloyirmm OH 1 byiuble Speed D ~ i v e smd.Uorion Conlrol, pp. 8/1-8/6,
1992.
149
19. Kondo, “Armature Current Locus Based Estimation Methad of Rotor
Position of Permanent Magnet Synchronous Motor without Mechanical
Sensor”, Conference Record of the 1995 BEE I n d u e Applications
Sffiiety, IEEE, 1995.
20. H.Watanabe, H.Katsushims, and T. Fujii, “An Improved Measuring
System of Rotor Position Angles of the Sensorless D k t Drive
Servomotor”, IEEX Proc of the 1991 Int’ppl. Con/ on lndwrriol
Elechonics, Control, mdlnshumeniatpplon,pp. 165-170, 1991
21. P. Acarnley, “Sensorless Position Deteotion in Permanent Magnet
Drives”, IEE Colloquium on Pennonent Magnet Mochines ond Driver,
pp. 1011-1014, 1993.
22. A. H. Wijenayake, J. M. Bailey, and M. Naidu, “A DSP-Bssed Position
Sensor Elimination Method with On-line Parameter ldentlfication
Scheme for Permanent Magnet Sychronous Motor Dnves”, IEEE Conf
Recordofthe 1 3 * A ~ n u ~ I l A S ~ 4vol. ~ ~I,pp.
r i ~ 207-215,
~, 1995.
23. J. S.Kim and S. K. SUI, “New Approach for the Low Speed Operation
ofthe PMSM Drives without Rotational Position Sensors’’, IEEE Trms.
on Power Eleelronies, vol. I I, pp. 512-519, 1996.
24. J. Kim and S. SUI,“High Perfomnce PMSM Ixives without Rotational
Position Sensors using Reduced Order Observer”, Record o/ihe 199s
l E E E l n ~ ~ n y A p p l i ~ *Conf, ~ I, pp. 15-82, 1995.
t i ~ ” vol.
25. M. A Hoque and M. A. M m n , “Speed and Position Sensorless
Permanent Magnet Synchronous Motor Drives’’, IEEE Con$ Proc.,
1994 Cmrdim Cmf on Electrical ond Computer Engi,teeeering, vol. 2.
pp. 689.692, September 1994.
26. T. Liu and C. Cheng, “Adaptive Control for a Sensorless Permanent-
Magnet Synchronous Motor Drive”, IEEE Trans. on Aerospace ond
Electronic Sy~lems,vol. 30, pp. 900-909, July 1994.
27. J. Holtz. “Speed Estimation and Sensorless Contml of AC Drives”,
Proc. afthe IVhAnma1 Conf oJ’IEEE Indusnial Elecironicr, pp. 649-
654, 1993.
28. B. J. Brunsbach, G. Henneberger, and T. Klepseh, “Position Controlled
Permanent Magnet Excited Synchronous Motor without Mechanical
Sensors“, IEE Conf on Power Electronics wd Applications, vol. 6, pp.
38113, 1993.
29. K. R. Shouse and D. G. Taylor, “Sensorless Velocity Control of
Permanent-Magnet S)nchmnous Motors”, PNC. of ihe 3Yd Cmf OR
Decision mrdConho1, pp. 1844-1849, December 1994.
30. T. M. Jahns, R. C. B e m a , and M. Ehsani, %tegrated Current
Regulation for a Brushless ECM Drive”, IEEE Tran.~.on Power
Electronics, wl. 6, pp. 118.126, Janualy 1991.
31. P. Schrmdt: M. Gaspen, G. Ray, and A. H. Wi.ienaye,ke, “Inilial Rotor
Angle Detection of a Non-Salient Pole Permanent Magnet Synchronous
Machine”, Conf Proc. sf the IEEE lndmlv Appliculiom Society
Ann!ml.&ieeting, 1997.
32. K. Ohrushi, N. Matsui, and Y . lion; ‘Estimation, Identification, and
Sensorless Control m Motion Control System”, Proc. of rhe IEEE, wl.
82,pp. 1253-1265, August 1994.
33. R. B. Sepe, and J. H. Lmg, “Real-Time Obsener-Based (Adaptive)
Control of a Permanent-Magnet Synchronous Motor without Mechanical
Sensors”, IEEE Trans. on Industy ilpplicotions,vol. 28, pp. 3345-
1352, Novemberhember 1992.
34. R. Wu and G. R. Slemon, “A Permanent Magnet Motor Drive without a
Shaft Sensor”, IEEE Trans. on Indusny Apppplicationr, vol. 21, pp. 1005-
10I I, SeptemberlCaober 199 I
3s. N. Matsui and T ‘Takeshita, “A Novel Starling Method of Sensorless
Salient-Pole B N S ~ ~Motor”,~ S S Conf Record o/ rhe 1991 IEEE /AS
.AnnuoIMeeling,vol. 1, pp. 386-392, October 1994
36. J. Hu, D. M. Dawson, and K. Anderson, “Position Control of a
B N S ~ ~DC ~ S Motor
S without Velffiily Measurements”, IEE Pmc. on
Elechonie PowerApplimtions, vol. 142, pp. 113-119, March 1995.
37. J. Moreira, “Indirect Sensing for Rotor Flux Position of Permanent
Magnet AC Motors W a t i n g in a Wide Speed Range’‘, IEEE T r a s . on
Industy Applicaiiom Society, vol. 32, pp. 401407,
Novemberficember 1996.
38. Ludovic, '-Rmshless DC Motor Control without Mechanical Sensors",
Power Electronics Specialist Conference, IEEl
39. N. Matsui and M. Shigyo, "Rmshless D(
Position and Speed Scnnors", IEEE Trm.7.
vol. 28,pp. 120-127, JanuaryIFebruary 1992.
40. T. Liu and C. Cheng, "Controller Design for a Sensorless Permanent
Magnet Synchronous Drive System", IEE Proc.-B, vol. 140, pp. 369-
378, Novembzr 1993.
41. J Solsma, M. I. Valla, and C. Muravchik, "A Nonlinear Reduced Order
Obs'ruer for Permanen1Magnet Synchronous Motors", IEEE Trans. on
Inchrm~ialEl~ct~~rics, vol. 43, pp. 38-43, August 1996.
42. J. C y m a , T. Ab,T.Higuohi, E. Yamad, and K. Shihahard, "Sansor-
loss Contml of a tlalf-Wave Rectified RNshless Sylchronous Motor",
Coqf Record qflhe 19951EEE industry ilpplicottons ConJ, vol. 1, pp.
69-74, 1995.
43. J. S Kim and S.K. Sul, .'New Approach for High Performance PMSM
Drives without Rotational Position Sensors", IEEE ConJ Pmc. 1775.
Applied P o ~ , r Eleecrronicr
r Con/ and Erposilion, vol. I , pp. 381-386,
1995.
44. Ostlund, "Initial Rotor Position Detcdion for an Intepted PM
Synchronous Motor Dnve", Conference Record of the 1995 IEEE
Industry Applications Society, IEEE, 1995.
45. N. Malsui, "Sensorless Operation of HNshless Ix' Motor Dnves'',
/'roc. of the IEEE hi'/. Con/. on Industrial Elemonics, Conlrol, and
Imlrunientalion,vol. 2 , pp. 739-744, November 1993.
46. R. Dhaouadi: N.Mohan, and I,. Nurum, --Designand Implcmentltion of
an Exladed Kaiman Filler for the Statc Estimation of 8 Permanent
Magnet Synchnous Motor", IEEE T m m on Power Electronics, vol.
6, pp. 49 1-497, July 1991.
17. 2. lbinotn and P. Seinas, "Application of Sliding Mode Observer for
Lndnced e.m.f., Position and Sped Estimation of P e m n e n t Magnet
Motors': IEEE Con/ Proc. o/lEl"ON, pp. 599-604, 1995.
48 J. Hu, D. Zhu, Y.Li, and J. Gao, "Application of Sliding Observer to
Sensdess Permanent Magnet Synchronous Motor Drive System". IEEE
I'owv E1e:leCmnicrSpecialist Cor$, vol. I, pp. 532.536 1994.
49. N. Erlugml, P P. Acarnley, and C. D French, "Real-time Estimation of
Rotor Position in PM Motors I)unng Transient Operation", 1EE FlJV? 68 D. E. Hesmondhalgh, D. Tippbg, and M. Amrani, "Performance and
European CO,$ on Power Electronics und .Applicdi.olions,pp. 3 11-3 16,
1993.
50 A. Rado, S.Rolognani, and M. ?.tgliotto, "Effective Estimation of Speed
and Rotor Position of a PM Synchronous Motor Ilrive by a Kalman
Filtering Technique", PESC '72 Record, 23" .leiual IEEE Power
lileclr"nicsSppeciuli31 CO,,/, MI. 2,pp. 951-957, 1992.
51. Shinhawa, "Wide Speed Operation of a Smsoiless B r u s h l ~Dc Motor
ltdring An Intenor Permanent Magnet Rotor", Power Conversion
Conference, EEE: 1993
52. R. C, h e m , T. M. Ja hsani, "Four-Quadmnt Sensorless
Rrurhicss ECM IXve'',
Iiypoporiliun,pp. 202-209,
53 J. Holtz, "State of the Art of Controlled AC Dnvc without Speed
Sensor'', IEEI? I'mc. o/tlre 1795 Itit 'I. C.on/ on Power Elctmnics ond
I h w S'sten,s. pp. 1-6. 19Y5.
54 A. Kulkami and M Ehsani, "A Now1 Position Sensor Elimination
'Technique for the Intenor Pennancnt-Magnot Syichronous Motor
Dnvc": IEEE T~airr.mi Indt~si~ifll .Applicol!o,zs, vol. 28, pp. 144-1j0.
55. and A t1. Kulkami, 'Elimination of Discrete
merit Sensor in Switchcd Keloctnncr Motor
I>iws.'~IEEE Tram on hrdmlv .Ipplicaliorrr, vol. 28, pp 128-135.
Janui~ffifehmsy1992.
56. M Ehrani. I. Husain. S.Mahajan, and K. K. Kamani, ..Nee. Modulation
Ilncding 'Tcechniques for Indirect Rotor Position Sensing in Snitched
Reluctance Motors". I H X Trms. or8 lndmlr) .-lpplrcolio,n. vol. 30, pp.
85-91. Januayil'ebrwy 1994
150
57. M. Ehsani, K. R. Ramani, "Direct Contml Strategies Rased on Sensing
Inductance in Switched Reluctance Motors", IEEE Tranr. on Power
Eleelrmics, vol. 11, pp. 74-82, January 1996.
58. S. S&, Elelectromagnetic Gearing Applicotiom in Hybrid Electric
Vehicles,M.S. Thesis, Texas A&M Ilniversity, 1994.
59. M. Ehsani, and I. Husain, "Simplification of Switched Reluctance
Motor lxlves far Commercial Applications", Pwc. oJIhe 1773 IEEE
PowerEIecltonics Congress, pp. 3043, 1993.
60. M. S. Arefeen, M. Ehsani, and T. A. L i p , "Indirect Startup Rotor
Position Sensor for Synchronous Reluctance Motor'', Proc. of the 1974
IEEE .4pplied Power Eleetronic.~Col$ and Erposition, vol. I,pp. 78-
82, 1994.
61. M. S . Arefcen, M. Ehssni, and T. A. L i p , .'An Analysis of the
A C C W ~of C Indirect
~ Shaft Sensor for Synchronous Reluctance Motor".
IEEE Trans. an Inhustry ilpplications, wl. 30, pp 1202-1209;
SeptcmkrIOctober 1994.
62. M. Ehsani, and I Huussain, '%mr Position Semshg in Switched
Reluctance Motor Drives by Meas-g Mutually Induccd Voltages".
IEEE Trms. on Industry Applrcorionr, vol. 30, pp. 422-429, MayIJune
1994.
63. I.fiussnin and M. Ehsani, "Error Analysis in Indirect Rotor Position
Sensing of Switched Reluotance Motors", IEEE Trrnis. on lndumlal
Electronics, vol. 41, pp. 301-307, June 1994.
64. N. Emgrul and P. Acsmley, "A New Algoithm for Smsorleas
+ration of Permanent Magnet Motors", IEEE Trms. on Induslv
ilpplieations, vol. 30, pp. 126-133, Januaryffebnwry 1994.
65. S. ogarawara and H. Akagi, "An Approach to Position Sensorless
Drive for Blushless DC Motors", IEEE Trans. on lndustp Applicalions;
vol. 27, pp 928-933, SeptcmberIOctobi 1991
66. D. R e e r , C. Oudet, and D. h d h a m , "Starling Brushless Dc Motors
Utilizing Volocity Sensors", Proc. O/ Ilw 14" Annual Syniposium 011
Incremental Motion Coniml Sysrenrs and Devices, Champaign, Illinois:
Incremental Motion Control Systems Society: pp. 99.107, June 1985.
67. P. S. Fredeiksen, J. Bkk, and F. Blaab-ilerg, "Comparison of Two
Enera Optimizing Techniques for PM-Machines", IEEE Cot$ oil
Industruil Electromes. Control, mrdlnslmmorlalion, pp. 32-37, 1994.
Desi@ of an Electromagnetic Sensor for Hrushlcss DC Motors", I&?
Proc., vol. 137, pp. 174-IX3,May 199U.
69. D. Peters and 1. Harth, Y C . s Provide Control for Sensorless DC
Motors", ED&;,pp. 85-94, April 1993.
70. N.Matsui, "Smsorless PM Brushless K Motor Drives",
on lnduslrirrl Electnmics, vol. 43, pp 300.308, April 1996.
71. M. Tom!ta, M Satoh, A. Yamguchi; S. Doki, and S Okwna;
"Sensorless Hstimation of Rotor Position of Cylindrical Brushless L X
Motors using Eddy Clurcnt", IEEE Inl'l. ll'o?krhop on .idvanced
,Morion Conrrol, pp. 24-28. 1996.
72. M. Park and H Lee, .3morless Vector Contml of Permanent Magnet
Synchronous Motor Using Adaptive Identification", IEEE ( h r $ on
I n d u s h i l E l ~ ~ t ~ Soociely. ~ I , pp. 209-2 14, 1989.
o ~ ~ i =vol.
73. M. S. Smtini, A. R Stubbzmd, and G. H. Hostetier, Di@lal C m t m l
Syslem Deri@i, New York Saunders College Fublishing, 1994.
74 R. Rronson, Afatri.~ .&lethods- A,, hrlroducfion,New York Academic
Prcrs, Inc., 1991.
75 K. Ikuka, el al.. ,' Micnmomputcr Contml for Sensmlrss Bmshlesr
Motor.., E R E 'Transactions on Industy Applicntiona, Vol 1A-27,
pp.595-601, MayJune 1985.
76. M. Hamdi and M. Ghribi, "A Sensorless Control Scheme Uased on
Fur& 1.ugic for AC Servo Dnws using a Fernanant-Magnet
Synchronous Motor'.. IEEE Cmodmn Con$ on Elrcrricnl m d
Co!qmti??gE n p i n e ~ m gpp. , 300-309, 1995.