Analysis and simulation of the high-speed torque performance of brushless DC motor drives

S.K. Safi, PhD P.P. Acarnley, PhD, CEng, FlEE A.G. Jack, PhD, CEng, MlEE

Indexing terms: Direct-current drives, Inverters, Motors, Prediction, Pulse-width modulation

Abstract: A simulation model for the performance prediction of brushless DC drives is used to investigate the high-speed torque performance for a motor switched by a PWM inverter. The solution is given for both 120" and 180" conduction angles of inverter operation, with provision for selecting the advance commutation angle. The influence of the more important drive parameters on high-speed performance is described. The paper also presents two simplified analytical approaches for obtaining the optimum pbaseadvance angle in the high-speed ranges. The merit of these methods, compared with trial-and-error based on simulation, is described.
List of symbols

e
i

= phase

n R
v
w
U

v,,

back EMF, V current, A = phase self inductance, H = speed, rev/min = phase resistance, Q = phase voltage, V = DC bus voltage, V = angular velocity, rad/s = conduction angle, electrical degrees = phase-advance angle, electrical degrees
= phase

Subscripts a, b, c = stator phases

1
Introduction

the stator phase currents with respect to the corresponding back EMFs. A second method of improving the speed range of the motor is to use stator excitation applied over 180", rather than 120", of the electrical cycle. To aid an understanding of the basic principles and for the development of these high-speed torque-control techniques and to study their effect on the drive performance, modelling and simulation is important. In previous investigations by various authors, different models have been developed to analyse the performance of brushless DC motors. Most of these analyses are based on one of three types of approach: state-variable model, Fourier-series, and dq-axis model [4-91. These methods have proved useful in many cases. As the types of control strategy, supply condition and back-EMF waveform are fairly diverse in trapezoidal brushless DC drive schemes, the flexibility of the numerical time-step approach is advantageous. In this approach, the instantaneous voltages and currents are calculated at suitable time intervals allowing irregular supply and back-EMF waveforms to be readily incorporated. The paper deals with the application of this method to the analysis of brushless DC motors driven by a PWM inverter. The aim of the analysis is to simulate a brushless DC-drive system operating in closed-loop control modes using the high-speed control methods in conjunction with a PWM control technique. In addition, detailed investigations of the operating range employing phase-advance angle and the optimum phase-advance angle are presented. During these investigations, two simplified analytical approaches are presented for obtaining the optimum phase-advance angle in the whole speed range.
2
Methods of high-speed torque control

In general, for all brushless DC motor drives operated with contant DC supply voltage, the average torque falls as the rotor speed is increased toward the base speed and the motor which meets the specified average torque for a given application will frequently be found to have insufficient torque at higher speed [l-31. To maximise the available torque in the high-speed operation, two control methods are presented in this paper. The first is the fieldweakening control method which consists of time shifting

2.1 Field-weakening control method

In principle, it is not possible for the drive to operate 'above a base speed' in the situation where the DC supply voltage and the induced back EMF are equal. In a conventional brushed DC machine with a separate field winding, this problem is overcome by reducing the field The authors acknowledge financial support provided by Esprit project 2656: IDRIS, and technical support provided by Robert Bosch GmbH, particularly R. Kennel and E. Hoflin of that company.
191

0IEE, 1995 Paper 18088 (Pl), first received 4th August and in revised form 7th December 1994 The authors are with the Electric Drives and Machines Group, Department of Electrical and Electronic Engineering, University of Newcastle upon Tyne, NE1 7RU, United Kingdom
IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, May 1995

current and thereby the field flux in the machine; hence the use of the term 'field weakening'. This solution is not easily applicable in brushless DC motors where the source of field flux is a permanent magnet. However, operation into the field-weakening region of a brushless DC drive is possible by adjusting the timephase relationship between winding current and induced EMF, and it is this strategy which is described here.
2.1.1 Concept of phase-advance angle

At low speeds, the current waveform is close to the rectangular, with a short pulse of voltage applied at the beginning and end of each phase-conduction interval to force the necessary change in current level. During each interval, the average applied voltage slightly exceeds the EMF, by an amount equal to the voltage drop across the winding resistance. At higher speeds, the inductive reactance of the windings results in a significant time constant and, as a consequence, the time taken for the current to reach its rated value is a large proportion of the phase-conduction interval and the rated current level is attained only at the end of the interval. Hence the torque output of the drive decreases, since the current magnitude is low and it is out of phase with the EMF. This situation can be rectified to some extent by turning on each phase earlier. If this procedure is employed, the current rise is faster, since the applied voltage is counteracted by a lower EMF at turn on. Also, the phase difference between current and EMF waveform is reduced so that the torque output is improved.
2.2 Extended -conduction-angle control method

EMF are coincident in time. At speeds approaching the base speed, where the available direct supply voltage and the back EMF are comparable in magnitude, this argument becomes more difficult to apply, as it becomes increasingly difficult to force current into windings. For an explanation of operation at higher speeds it is necessary to consider energy flow in the drive. The power-switching circuit supplies electrical energy to the motor and this energy is either converted to mechanical energy which is the motor output or is stored as magnetic energy. When the switching transistor is turned off, the associated stored magnetic energy is either returned to the DC supply (through the freewheeling paths) or is converted to mechanical output energy, if the decaying winding current flows against back EMF. If the phase-advance procedure is employed, phase current is allowed to build up in the motor winding before the back EMF reaches any significant level. Thus, during the early part of the windingexcitation interval, electrical energy is being taken from the supply and stored as magnetic energy as the winding current attains its initial high value. When the back EMF reaches its maximum level, mechanical output energy is obtained by conversion of both the electrical energy input from the supply and the previously stored magnetic energy from the windings. In terms of the voltage balance in the windings, current is being forced to flow against the back EMF by the DC supply and the voltage caused by decaying current in the winding inductance.
4
Analysis and simulation approach

4.1 Analysis model and assumptions

The potential advantages of earler current turn on have been discussed in Section 2.1. One penalty to be paid for the use of the phase-advance method is that each winding is switched off before the corresponding back EMF has fallen to zero. Therefore it would appear to be advantageous to extend the interval for which the winding is excited. A particularly convenient method is to use windingexcitation intervals of 180" since the requirement for rotor-position signals is unchanged; excitation changes still occur at 60" intervals.
2.3 DC-chopper-control method This method is applicable only in systems with choppertype drive circuits, but is capable of very rapid and effective control of torque. A current chopper permits the use of a high direct supply voltage so that the actual currents flowing into the machine are maintained as closely as possible to the references during constant-torque operation. Nevertheless, the operating speed limit in a drive with current chopping is still dependent on the comparative magnitudes of direct supply voltage and back EMF. The action of the current chopper, in this context, may be viewed as facilitating the application of enhanced supply voltage by preventing excessive winding currents at low speed.
3
Combined effects of machine inductance and phase-advance angle

The analysis in this work is based on the model shown in Fig. 1. The analysis of such a system is complicated by the interactions between the machine itself and the power-electronic circuits used to provide the appropriate voltage and current waveforms. Consequently, it is common practice to make a number of simplifying assumptions in the analysis, namely (i) the motor is unsaturated (ii) eddy-current and hysteresis effects in the machine's magnetic materials have negligible influence on the winding current
I"

'0

'b

I"
LM

The discussion in Section 2 of the mechanism by which phase advance can benefit high-speed performance concentrated on the need to ensure that current and back
192

-.
It

L
"c

Fig. 1

System model and conuersion

IEE Proc.-Electr. Power Appl., Vol. 142, No.3, M a y I995

(iii) there is no saliency and therefore self and mutual inductances are constant and independent of rotor position (iv) the devices in the power-switching circuit are ideal.
4.2 Analysis and numerical solution

In this work, the analysis model explicitly represents the condition of current flowing in all three motor windings at commutation (Fig. 2b). At all other times only two phases are conducting (Fig. 2a). The representation of these conditions is accomplished by incorporating switching states into the model. The solution is achieved by using an appropriate system of differential equations to represent the system under the different operating conditions. These equations are presented in the following Section.

The solution of the differential equations which describe the switching states is calculated by applying a fourth-order RungeKutta algorithm for step-by-step integration. In this procedure, an artificial transient procedure has to be followed. Starting with arbitrary initial conditions, an excitation cycle (commutation and conduction) is studied; the conditions at the end of the cycle are taken as initial conditions for a second cycle and a third and so on until steady cycles are achieved. In this analysis, the system model includes the PWMswitching strategy. In this method, the current calculated at each interval is compared with the demand and an error current is generated. After proportional, integral and differential controlling actions, the scaled error current is compared with a triangular waveform and, as a result of this comparison, device-switching signals appropriate to the instantaneous rotor position are produced.
4.2.1 Circuit equations of system in operating conditions

.............................

L ,

As described above, the system model includes the PWM-switching strategy and thus it has switching states which operate on six modes for the commutation interval and six modes for the conduction interval over the 360" of an excitation cycle. For these modes it is not possible to calculate the current waveform for each phase in isolation when current is flowing in all three phases. In this case, the winding currents are calculated in terms of the line voltages. For example, by considering the case where the current flows in phases a and b during conduction (Fig. 2a) the line-voltage equation is given by
di,

+ e.)

-(Rib

di, +L dt + e.)

(1)

According to the relation of Kc = U, - ob, io = - i b , and eab = e, - e,, eqn. 1 is rewritten as

0

V,, = 2Ri,

+ 2L di, - + eab dt

.
c

This equation can be solved using numerical integration to find i, and hence the winding current i,. Conduction equations for other combinations of excited phases can be derived by transposition of winding and device numbering. During commutation (Fig. 2b), the independent currents must be calculated, so a second equation is generated by considering the series connection of windings a and c, to give the following equation in terms of V, = U, -U iL1 = -i, and e, = e , - e, i di, V& = 2Ria + 2L - + e, (3) dt
C )

Since the motor windings are star connected, the next relation can be obtained: i , + i, + i, = 0, Using this relationship and the definitions of iob (= i, - ib) and i , (= i, - i,), it is possible to express the actual winding currents in the terms of inb and ,i

(4)

Fig. 2 Conduction and commutation modes of the inverter ( 1 Conduction mode b Commutationmode

IEE Proc-Electr. Power Appl., Vol. 142, No. 3, May 1995

I93

Therefore, given the solution for i , and ,i , it is possible to calculate the three winding currents. The same procedure can be used to derive the circuit equations for other modes. Once these currents are obtained, the average torque, efficiency and torque ripple can be calculated.
4.3 Experimental verification

current is approximately three times the motor continuously rated current of 23 A, such current levels can occur as a result of rapid changes in speed demand. The main parameters of this drive are shown in Table 1.

To verify the model developed, the solution for the drive with 120" conduction angle obtained from simulation analysis was compared with the tested results of the experimental system. Figs. 3 and 4 show the comparison

L 3

-5-10 -151
,
I

0
0

50

100

150 200 250 angle, deg

300

350 LOO

Fig. 4 Measured phase currentfor 120" conduction angle Vertical axes 5 Aidivision, horizontal axes 5 msldivision B No phase advance b 15" electrical phase advance

Table 1 : Brushless-DC-motorparameters
751

Name

Symbol

Unit kW

Value

Rates power P , Number of rotor poles p I Continuous current Maximum current I,,, K, Voltage constant L Winding inductance R Winding resistance

20
6

A 23 A 158 VI1 000 rev/min 48.1 5 mH 3.1 n 0.26

5.1.1 Operating conditions

-s 1-

0
b

50

100

150 200 250 angle, deg

300

350

LOO

Predicted phase current for 120" conduction angle Fig. 3 No phase advance b IS" electrical phase advance
r?

of the simulation results and the experimental current waveforms for 0" and 15" phase-advance angles, respectively. One can note a good agreement between the prediction and the practical measurement.
5
Results and discussion

5.1 Performance prediction The drive system, including its controller, is simulated using the analysis technique described in Section 4. The performance characteristics, such as torque/speed, RMS current, percentage torque ripple and percentage eficiency are examined for these control schemes and the effect of variation of parameters such as the machine inductance on the high-speed-torque performance is also examined. The results presented in this work were obtained for a 20 kW industrial machine-tool axis drive with rated static torque of 21 Nm. Performance predictions are made for a current demand of 60A. Although this
194

The test results presented in this Section are for different operating conditions. These operating conditions can be divided into three ranges according to the speed: (a) Low speed: Low speed is defined here as the range of speed over which the current is maintained constant by the current controller and thus full constant torque is available. (b) Base speed: Base speed is the highest speed at which maximum torque can be obtained. (c) High speed: When considering the characteristics of a drive at high speeds, two main operating-speed ranges are distinguished: (i) the high-speed range below the base speed, where current changes are forced by the voltage difference of the supply voltage available and the back EMF. Current and torque control can be achieved by both a current controller and phase advance angle (ii) the high-speed range above the base speed, where the available torque decreases with increasing speed. Current and torque can only be controlled by the commutation angle.
5.2 Low-speed performance for 120" and 180" conduction angles 5.2.1 Conduction angle of 120" Fig. 5 is used as a basis for comparing aspects (e.g.

voltage, current, average torque and torque ripple) of the

IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, M a y 1995

machine performance at low speed (100 revlmin). As expected, the winding-current waveform is near rectangular in shape and easily reaches the current demand of

current waveform is near quasisquare in shape and easily reaches the demanded current of 60A. Also note that torque ripple increases as the windings are excited at rotor positions where the induced EMF is not constant.
5.3 High -speed/torqueperformance for 120" and 180" conduction angles

5.3.1 Conduction angle of 120" To examine the drive performance at high speed, a speed of 3000 rev/min and a current demand of 60 A are considered. The overall shape of the characteristics and details of the results for this case is shown in Fig. 7. It is

-60

0
0

r
-8
50

1 6oo

100

150 200

250

300

350

LOO

angle, deg

601-

401 0
b

50

100

150 200 250 a n g l e , deg

300

350

LOO

Current, EMF and torque for I20" conduction angle at = 3.1 mH and phase advance ofzero current ~-~~ back EMF

Fig. 5

I W revlmin, L

50

100

150 200 250 angle, deg

300

-600 350 400

60 A, since there is a large margin in available voltage. There is no reduction in torque at low speed and under this condition the drive requires zero phase-advance angle. Also note that the opposite-polarity spikes at the centre of the current waveform are caused by the stator currents commutating from one phase to another.
5.2.2Extended conduction angle of 180"

angle, deg

To demonstrate the effect of the extended conduction angle method at low speeds Fig. 6 shows that winding8o

Current, EMF and torque for 120" conduction angle at 3000 rev/min, L = 3.1 mH and phase advance ofzero current _ _ _ _ back EMF

Fig. 7

1 6o

- 80 0 6 -/E z

clear from this Figure that a reduction in output torque is observed, owing to the increasing size of the switch-ontime constant relative to an electrical cycle. At higher speeds, the machine runs out of voltage and/or time to force the phase current up to the level demanded, and the reducing difference between direct supply voltage and back EMF (& - e) now dictate the output. At this speed the motor phase current only just reaches the demandmaximum value of the current and so only starts to chop at the tail of the period. In addition, note from Fig. 7 that, in the high-speed region, the current waveform becomes near sinusoidal in shape. As will be seen in Section 5.4, it is very helpful if the phase-advance technique is used to keep the current in phase with respect to the EMF.
5.3.2Extended conduction angle of 180"

50

100

150 200 250 a n g l e , deg

300

350

LOO

60 r

I 0

. 50

, 100

, 150 200 250 angle, deg

300

350 400

Current, E M F and torque for 180" conduction angle at 100 revlmin, L = 3.1 mH and phase advance of zero __ current _ _ _ ~ back EMF IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, May 1995

Fig. 6

At high speeds, an extended conduction angle has a different effect on the torque, resulting from windinginductance and high-speed limitations. The results of these constraints may be explained by referring to the current waveforms which are shown in Fig. 8, which shows the current and back EMF at various speeds with zero phase advance. It can be seen that an increase of the rotor speed causes the current to change shape considerably. Also, note that, as speed increases, the distortion of current effectively delays the waveform and increases the shift between the current and back EMF.
195

It is important to note that, with 180" conduction, when positive excitation is started, there is a large negative current in the phase winding, which must decay to

at high speeds, as it becomes increasingly difficult to establish the rated phase currents in the motor winding. As can be seen in Fig. 9, the torque improvement in the high-speed region is increased as more phase advance is

B " 20
U ,

- 80

0
a

50

100

150 200 250 angle, deg

_____ -' , 300 350

-200 400

ii
10

0 0
a

1000

2000 3000 4000 speed, revtrnin

5000

4 6000

"
3

60r 50 I

0 b

50

100

150 200 250 angle, deg

300

350

400

LO -

Fig. 8
~

Current and EMF waveformfor 180" conduction angle phase advance zero current _ _ _ ~back EMF a Speed = 2MxI rev/min b Speed = 3MxI rev/min

30 -

-..

zero before positive current can be established. At high speeds, the decay time of the initial negative occupies a considerable part of the excitation interval and the negative torque produced by the negative current reduces the average torque. As the conduction angle increases, the initial negative current becomes larger and requires a longer time to decay to zero. Improvements in high-speed-torque performance with high inductance can be achieved with the extendedconduction-angle control method only if the phaseadvance method is used, as will be shown in Section 5.4.
5.4 Effect of high-speed-torque control methods on the machine-performancecharacteristics 5.4.1 Conduction angle of 120"

e F *

g20-

10-

-10 O0
b

1000

2000

3000

4000

5000

6000

speed, revlmin

Mnchine-performe curuesfor 1.20" conduction angle advance angle = 0 A advance angle = 15" advance angle = 30" 0 advance angle = 45" x advance angle = 60" (1 RMS currentlsped performance b Torquelspad performance

Fig. 9

The effect of phase advance on the torque/speed characteristics is demonstrated in Fig. 9 which shows torque/ speed characteristic of the machine as a function of speed with phase-advance angle of a = 0",15", 30", 45"and 60". The torque/speed curves of Fig. 9 provide a useful summary of the motor's capacity to operate under closed-loop control over the whole range of load conditions. The curves of this Figure show that, as the speed increases, the torque produced at any phase-advance angle will reach a maximum before falling to zero, and with large phase-advance angles a positive torque is not produced at low speeds. The envelope of the torque/ speed curves defines the maximum torque which the motor can produce at each speed. This torque decreases
196

introduced until an optimum is reached for that particular speed. Fig. 10 illustrates how this improvement in performance is obtained by employing the phase-advance procedure at high speed (3000rev/min). The waveforms of Fig. 10 also show that, with the correct value of phaseadvance angle for a particular operating speed, the current is nearly in phase with the back EMF and thus there is a large improvement in the values of torque obtained at a speed of 3000 rev/min. Although torque ripple is of less concern at high speeds, owing to the lowpass-filtering effect of the combined motor and load inertias, it is worth comparing the torque waveforms in Figs. 7 and 10. These show that phase advance leads not only to higher levels of average torque, but also to lower levels of torque ripple.
IEE Proc.-Electr. Power Appl., Vol. 142, No.3, May 1995

5.4.2 Extended conduction angle of 180" Similar trends are observed for the extended-conductionangle method with varying phase-advance angle. Fig. 11

that this is a near-optimal value if there is no phase advance. However, if the total phase inductance is doubled, to 6.2 mH, torque production is improved from
50 -

I
\'.

1 600

400

40 -

<
30

!?

" 3
v)

20-

LL

10
0
0

50

100

150 200 250 angle, deg

300

-600 3 5 0 400

0 0 a
60 r 1000

2000

3000

4000

5000

6000

speed, revlmin

"b
b
~

i0

IO0

150 2dO 2 i O angle, deg

3b0

3kO

460

Fig. 10 Current, EMF and torque for 120" conduction angle at 3000 revlmin, L = 3.1 mH and phose uduance o f 15" current

backEMF _ _

summarises the drive performance with extended conduction angle. Improvements in performance are obtained by advancing the winding excitation relative to the induced EMF as speed increases. Fig. 12 illustrates how these improvements in performance are obtained by using the phase-advance procedure at a speed of 3000 rev/min. As an example, the results of Fig. 12 show that, when the phase-advance angle is 30", the average torque at a speed of 3000 rev/min is typically 1.5 times the maximum torque when the phase-advance angle is zero.
5.4.3Performance comparison between 120" and 180"angles It is of interest to compare the difference in machine performance when operating with the two conduction angles. Fig. 13 is used as a basis for comparing the torque/speed characteristics of a brushless DC motor drive with 120" and 180" conduction angles. For both conduction angles, the phase-advance method is effective in improving torque at high speeds, in the range 40006000 rev/min. As can be seen from the results of these Figures, the benefits of an extended conduction angle with phase advance are most pronounced at very high speed. For this particular case, the greatest torque improvement appears in the range 4000-6000 rev/min.
5.5 Combined effect of winding inductance and phase-advance angle To examine further the combined effects of both phase-

-1 O 00
b

1000

2000

3000

4000

5000

6000

speed, revlrnin

Fig. 11 Machine-performance curves Jor 180" conduction angle 0 advance angle = 0 A advance angle = I S advance angle = 30" 0 advance angle = 45" x advance angle = 60" a RMS current/speed performance b Torque/speed performance

24Nm to 46Nm at 3000rev/min, provided that the phase advance is adjusted appropriately. The details of the results for this case are shown in Fig. 15. To realise these higher levels of torque, it is essential to apply the phase-advance method, with more phase advance being required for higher phase inductances.
6
Analytical investigation of the optimum phase-advance angle

6.1 General considerations

advance angle and winding inductance on machine performance, Fig. 14 shows how torque and RMS current vary with winding inductance for different values of phase advance at a typical speed of 2000rev/min. The phase inductance of the actual motor is 3.1 mH and it can be seen from the torque/inductance characteristic
IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, May I995

The critical importance of phase-advance angle in the successful high-speed operation of the brushless-DCmachine drive makes an analytical approach especially valuable, since time spent on trial-and-error setting up in the laboratory is eliminated. Although it would be possible to implement trial and error using simulation, this approach, which is presented in Section 4, is involved and requires a substantial
197

amount of computation time to establish best steadystate operating points. It is therefore desirable to establish simplified analytical representations which yield a
100

more direct means of finding the optimum phase-advance angle when the system is operating in the high-speed ranges.

1600

20

0
a

6 8 inductance, rnH

I 0 1 2 1 4

0
0

50

100

150

200

250

300

350 400

-600

angle, deg

"k
b
~

i0

I00

1 5 0 200

2;O

300

350 LdO

anqle. deg

Fig. 1 2 Current, EMF and torque for 180" conduction angle at 3000 revlmin, L = 3.1 mH and phase advance o f 30" current ~~-~ back EMF

101
0
b

6 8 inductance, rnH

1 0 1 2 1 4

Fig. 14
A

Machine-performance curves for 120" conduction angle

Speed = 2oM) rev/min

40
4

advance angle = 0" advance angle = 15" advance angle = 30" 0 advance angle = 45" x advance angle = 60" a RMS current against inductance characteristics b Torque against inductance characteristics
#

"
in

30-

16oo

20.

10

1000

2000

3000

4000

5000

6000

speed, rev I rn I n

50r

0
0

50

100

150

200

250

300

350

400

-800

angle, deg

d
0

$50
401
"
" "

120'
10

50

100

150 200

250

xx)

350 LOO

"

0
b

1000

2000

3000

4000

5000

6000 Fig. 15

speed, revlmin

angle, deg

Comparison ofperformance at 120" and 180" RMS current/speed performance b Torqueispeed performance
a

Fig. 13

~~~~

Current, EMF and torque for 120" conduction angle at 3 O O O rev/min, L = 6.2 mH and phase advance of45" current backEMF
~

198

IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, M a y 1995

When all the available phase-current waveforms at different operating speeds from the simulation test presented in Section 5 had been analysed and studied, it was concluded that the drive system, in which the stator current is regulated by a voltage-source PWM inverter, has two main control regions (a) region 1: the PWM action regulates the stator current to a limit value and the drive is said to be operating in the constant-current region (b) region 2: the current limiting using the PWM action becomes ineffective and is practically nonexistent, so the drive is said to be operating in the constantvoltage region. In this second region, the electrical time constant of the winding is relatively large compared with the excitation interval of each phase. Therefore the higher-frequency components of excitation are heavily filtered and the major current component is at the fundamental frequency. As a consequence of this, the current waveform becomes sinusoidal in shape. Hence, under this operating condition it is appropriate to derive an analytical expression for the optimum phase-advance angle based using the sinusoidal analysis. This Section describes two analytical approaches to obtaining an optimum phase-advance angle. The first method is based on the controlled current, where the phase voltage must rise before back EMF in order that the current has attained the desired value at the start of the on period, and the second is based on the sinusoidal analysis, where the stator current is effectively sinusoidal at high-speed operation.
6.2 Analytical methods to determine optimum phase-advance angle 6.2.1 Method for constant-currentregion

This result shows that the angle a, is a function of the time constant and operating speed. Thus, at all speeds, the voltage applied to the winding must advance in phase to compensate for the increasing phase lag between current and applied voltage.
6.2.3 Comparison of optimum phase-advance angle calculated by analytical and simulation methods

To investigate the validity of both analytical methods set forth in the preceding sections, their results are verified by comparison with results obtained by the simulation approach. Fig. 16 shows the phase-advance angle plotted against speed using contant-current, constant-voltage and simulation methods. It is seen that the results of the constantcurrent formula are in close agreement with those of the
120E

speed, revlmin

To ensure a reasonable square current waveform and to operate effectively over a range where a drive is operating with current control, the switches must be closed in advance of the rising-back-EMF region so as to allow time for the current to build up. The applied phase voltage must switch on a degrees before the phase back EMF. During this switch-on angle, the current is considered to rise linearly up to its value of I and the angle a is derived, (assuming e < V ) ,as
a. = -

Fig. 16 Comparison of optimum phase-advance angles calculated by analytical ahd simulation methods ~1 Simulation b Constant voltage c Constant current

6pI Ln

simulation method at low speeds and differ widely from the contant-voltage formula which gives satisfactory results only in the high-speed range. This should be expected, as the method of the constant-current region has been developed for a drive operating with PWM control whereas the method for constant voltage was developed for a drive operating in the constant-voltage region.
6.2.4 Transition-control algorithm

'

v,,

(5)
To combine the control-region methods, which are derived in Section 6, in the whole speed range a transition factor K ( o ) is introduced which specifies a relation between the transition speed wT (i.e. base speed) and speed regions o. The general effect on the phase-advanceangle/speed curve, taking account of control regions, is shown in Fig. 16. The analysis for the combination of control regions due to the transition factor of control regions is given by the expression
aT = K(o)ai + { 1 - K ( o ) } a , (7) where ai is phase-advance angle for the constant-current region and is given by eqn. 5, a, is phase-advance angle for the constant voltage and is given by eqn. 6, and K ( w ) = 1/{ 1 + (o/oT)}. It is clear from this analysis that the transition between control regions are governed by a transition factor which determines when transition takes place. Eqn. 7 can be used effectively in the whole speed
199

The current I is maintained constant by the PWM control and then turned off before the corresponding back EMF has fallen to zero. Eqn. 5 has some similarities to the expression adopted by Bose et al. [lo]. However, their expression is derived for the phase-advance-angle control of a reluctance machine.
6.2.2 Method for constant-voltage region

To predict the phase-advance angle throughout the region of constant-voltage control, it is appropriate to use sinusoidal analysis. The approach developed here is based on two-axis machine theory where machine performance is derived from the phasor diagram of the machine. Using this procedure, the optimum phaseadvance. angle is found as follows
a, = arctan 2n

(-)( -)
60

nL R

IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, M a y 1995

range and will also easily determine the base speed at which the current controller runs out.
7 Conclusions

References

1 ACARNLEY, P.P.: Stepping motors: a guide to modem theory and practice, IEE Control Engineering Series 19 (Peter Peregrinus,
1982) 2 ACARNLEY, P.P., JACK, A.G., and JOWETT, P.T.: Speed stahil-

The performance of brushless DC motors switched by a PWM inverter is investigated, with a view to improving the high-speed-torque performance. During the course of this work a practical computation technique for predicting the steady-state performance of a brushlesspermanent-magnet-motor drive is presented. This method provides a convenient and economical means of evaluating the effects of system parameters and modes of operation on the performance of the drive. The results of analysis have clearly shown that the high-speed torque characteristics of the brushless DC motor can be improved considerably simply by adjusting the chosen setting of the phase-advance angle and using appropriate conduction angles. It has been shown that the torque available from a commercial drive at high speeds (i.e. 3000rev/min) can be improved by 12.5% using phase advance. The results have also shown that high-speed torque performance can be improved by phase advance, combined with modest increases in winding inductance. Two simplified analytical approaches to determine the optimum phase-advance angle in the low-speed and highspeed ranges have been presented. The merit of these methods, compared with trial-and-error based on simulation, is described.

ity in permanent-magnet synchronous and brushless DC drives (ICEM, Pis% 1988) 3 JAHNS, T.M.: Torque production in permanent magnet synchronous motor drives with rectangular current excitation, IEEE Trans., 1984, IA-20, (4), pp. 803-813 4 LIPO, T.A., and TURNBULL, F.G.: Analysis and comparison of two types of squarewave inverter drives, IEEE Trans., 1975, IA-11, (2), pp. 137-147

5 CHALMERS, B.J., PACEY, K., and GIBSON, J.P.: Brushless DC traction drive, Proc. IEE, 1975, 122, (2), pp. 733-738 6 WEH., H., and BOULES, N.: Field analysis for a high-speed permanent magnet synchronous of the disc construction type, Electr. Mach. Electromech., 1980.5, (11, pp. 25-37 7 ACARNLEY, P.P., and HUGHES, A.: Predicting the pullout torqueispeed curve of variable-reluctance stepping motors, Proc. IEE, 1981,128, (2). pp. 109-113 8 BINNS, K.J., BARNARD, W.R., and JABBAR, M.A.: Hybrid permanent-magnet synchronous motors, Proc. IEE, 1978, 125, (3). pp. 203-208 9 DEMERDASH, N.A., and NEHL, T.W.: Dynamic modelling of brushless DC motor for aerospace actuation, IEEE Trans., 1980, AES16, (6). pp. 811-821 10 BOSE, B.K., MILLER, T.J.E., SZCZESNY, P.M., and BICKNELL, W.H.: Microcomputer control of switched reluctana motor. IEEE Trans., 1986,1A-22, (4), pp. 708-715

200

IEE Proc.-Electr. Power Appl., Vol. 142, No. 3, M a y 1995