Pages From Control of Induction Motors - 2

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

CHAPTER 4 / POWER ELECTRONIC CONVERTERS 71

RECTIFIER DC LINK INVERTER

FIGURE 4.20 Braking resistor arrangement.

switch, is inserted between the dc link and inverter, as illustrated in


Figure 4.20.

4.5 CONTROL OF VOLTAGE SOURCE INVERTERS

To explain the principles of control of inverters, it is convenient to intro-


duce the so-called switching variables, variously defined depending on
the type of inverter. For the most common, two-level voltage source
inverter depicted in Figure 4.13, three binary switching variables, a, ft,
and c, one per phase of the inverter, are sufficient. As already mentioned,
both switches in an inverter leg cannot be on simultaneously, because
they would short the dc supply source (i.e., the dc-link capacitor). The
situation when both switches are off is not dangerous, but the voltage at
the corresponding output terminal is undetermined. This is so because,
depending on the polarity of the load current, the terminal would be
connected, via one of the freewheeling diodes, to either the positive or
negative dc bus. Therefore, in practice, except for the very short blanking-
time intervals, one switch in each phase is on, and the other is off.
Consequently, each inverter leg can assume two states only, and the
number of states of the whole inverter is eight (2^).
Taking as an example phase A, the switching variable a is defined
to assume the value of 1 if switch SA is on and switch SA' is off. If,
conversely, SA is off and SA' is on, a assumes the value of 0. The other
two switching variables, b and c, are defined analogously. An inverter
state can be denoted as abc2^ For example, with a = 1,fo= 0, and c =
1, the inverter is said to be in State 5, because IOI2 = 5.
72 CONTROL OF INDUCTION MOTORS

It is easy to show that the line-to-line output voltages, V^B* Vgc and
VcA? of the voltage source inverter are given by

VAB 1 -1 0 a
VBC = v; 0 1 -1 b (4.3)
VrA -1 0 1 c

When the same control principle is applied to all three phases of an


inverter feeding a balanced wye-connected load, the individual line-to-
neutral voltages. VAN* ^BN^ ^^^ ^CN» ^ ^ balanced too; that is.

VAN + VBN + VcN = 0. (4.4)


In addition (unconditionally),
(4.5)
and

^BN ~ ^CN ~ ^BC- (4.6)

Solving Eqs. (4.4) to (4.6) for VAN» ^BN* ^^^ ^CN* yields

[VAN 1 1 0 -l"l [VAB]


VBN -1 1 0 VBC (4.7)
LvcN_ " ^ 0 -1 1 JL^CAJ
which, when combined with Eq. (4.3), gives

1 - 1 1 Va]
[VAN
VBN -A -1
3 1
-
• 2
-1
1 2 J\c\
b (4.8)
L^CN

The simplest control strategy for the inverter consists in imposing the
5-4-6-2-3-1 state sequence, resulting in the already-mentioned square-
wave, or six-step, mode of operation. Waveforms of the line-to-line and
line-to-neutral output voltages of the inverter in this mode are shown in
Figure 4.21. Each switch of the inverter is turned on and off once per
cycle only, and the peak value of the fundamental line-to-line output
voltage is 1.1 Vj. However, the load current is of poor quality, due to the
high content of low-order voltage harmonics. Also, the magnitude of
output voltage cannot be controlled within the inverter, which constitutes
another disadvantage of the square-wave mode. Therefore, in most practi-
cal inverters, transition to the square-wave operation occurs only when
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 73

STATE: 5 4

\ \y.\ 1 cjt
1 1 \ 1 1
y,
BC
1 ' 1 cot

'CA
i i i(jjt
^ ' '^—1 • : I cjt
\ \ ' i i 1
"BN
i * * 11 j
0)t
\ 1 i i '
j j i
CJt
1
I i 1
0 7T 27T

FIGURE 4.21 Output voltage waveforms in a voltage source inverter in the square-
wave mode.

the maximum possible output voltage is needed. Other than that, the
inverter operates in the PWM mode.
Typical voltage waveforms in a PWM inverter are illustrated in Figure
4.22. In this example, the period of output voltage is divided into 12 so-
called switching intervals. One pulse of each switching variable appears
in each switching interval, and the adjustable pulse width varies from
zero to the interval width. The number, N, of switching intervals per cycle
of the output voltage, is given by

/s\
A^ (4.9)
/'
where/sw denotes the so-called switching frequency and/is the fundamen-
tal output frequency of inverter. The switching frequency is usually con-
stant; thus, A/^ depends on the output frequency only, and it is not necessarily
an integer. The voltage waveforms are pulsed, not sinusoidal, and clusters
of the most pronounced high harmonics coincide with multiples of the
switching frequency. Thus, with typical switching frequencies on the order
of several kHz, harmonic currents generated by those harmonics are weak,
thanks to the low-pass action of load (motor) inductances. As a result,
current waveforms in PWM inverters are close to ideal sinusoids, with
only a small ripple.
74 CONTROL OF INDUCTION MOTORS

AB Ji
I ly liiyuuiuu n
CJt

nu
'BC

rnnnn Q)t

u jy[ Q)t
iir
mMUMWfV%ta?knS#^<"'
"BN ^,^h;l^^^,^
Mlw¥^ ifV%i (j)t
Ml''¥'^t<yuJWfJt4'¥'':hmM Q)t

27Y

FIGURE 4.22 Output voltage waveforms in a voltage source inverter in the PWM
mode.

The most common class of PWM techniques for voltage source invert-
ers is based on the idea of voltage space vectors. Space vectors of electric
and magnetic variables are an important tool in the analysis, modeling,
and control of three-phase ac machines, and their physical interpretation
will be given in Chapter 6. Here, only the formal definition of voltage
space vectors is provided.
Considering three-phase voltages v^, v^, and v^, the corresponding
voltage space vector, v, is given by
Vd + 7Vq, (4.10)
where

fl 1 1 1
"2 2 fVal
[::]=;0 a^ :V32 J
L 2
Vb

|_^cj
(4.11)

Voltages Va, v^, and v^ can denote the line-to-neutral voltages, that is, v^
~ ^AN> ^b "= ^BN' ^c "^ ^CN» or the line-to-line voltages, that is, v^ = V^B*
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 75

^b ~ ^BC' ^c ^ ^CA- If ^a + ^b + ^c ~ 0, the abc-^dq transformation


defined by Eq. (4.11) can be inverted, to yield

2
0
3
1 1
3
1
V3
1
[::]• (4.12)

3 V3
The abc^dq transformation transforms three actual voltages into a
two-dimensional space vector in a complex plane. In particular, consider-
ing the classic two-level voltage source inverter, a voltage vector can be
assigned to each of the eight states.
EXAMPLE 4.1 To illustrate the idea of voltage space vectors, the
space vector, V5, of the line-to-neutral output voltage of a two-level
voltage source inverter in State 5 will be found. The inverter is assumed
to be supplied with an ideal dc input voltage, Vj.
In State 5, the switching variables are: a = 1, fo = 0, and c =
1. Thus, according to Eq. (4.8),

[VAN 2 -1 - 1 ] rn 11
VBN -1 2 -1 0 = Yi - 2
3 -1 -1 3
L^CN_ 2 J[1 1J
and, from Eq. (4.11),

[::]
2
V3
2 " 2
2
V3 -iM
that is.

^5

Space vectors of the line-to-neutral voltage associated with the eight


states of the voltage source inverters are shown in Figure 4.23 in the per-
unit format, with the dc input voltage, Vj, taken as the base voltage. There
are six nonzero vectors, v^ through v^, and two zero vectors, VQ and Vj,
76 CONTROL OF INDUCTION MOTORS

FIGURE 4.23 Space vectors of the line-to-neutral voltage (per-unit) in the two-level
voltage source inverter.

resulting from clamping all the output terminals to the negative dc bus
(State 0) or positive bus (State 7). The nonzero vectors divide the complex
plane into six sectors (sextants), numbered from 1 to 6. Comparing Figures
4.21 and 4.23, we see that the sequence of inverter states corresponding
to the consecutive nonzero voltage vectors results in the square-wave
mode of operation of the inverter.
It can easily be shown that if voltages v^, v^, and v^ form a balanced
set of three-phase voltages, the resultant voltage space vector, v, has the
magnitude, K 15 times greater than the peak value of those voltages.
The phase angle, a, of the voltage vector equals that of v^. Thus, if

Va(0 = Vjncos((o0, (4.13)


where V denotes a peak value, then

V = y^^« = l.SV^eJ^^K (4.14)


As the time progresses, vector v revolves in the complex plane with the
angular velocity o). Consequently, the goal of inverter control can be
formulated as follows: Make the space vector of the output voltage to
rotate with a desired speed, and adjust the magnitude of this vector to a
desired value.
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 77

As shown in Figure 4.23, the voltage source inverter can produce


only stationary voltage vectors. The popular Space Vector Pulse Width
Modulation (SVPWM) technique overcomes this limitation by means of
generation of the stationary vectors in such a manner that it is their time
average that follows the revolving reference vector, v*. Specifically, to
synthesize the desired voltage vector in a given sextant, the two nonzero
stationary vectors framing this sextant, plus a zero vector (or vectors),
are used. To explain this technique, the reference vector, at a certain
instant, is assumed to lie in Sextant 3, as illustrated in Figure 4.24. The
local (within the sextant) angular position of v*, p, is given by

-rr. / 3
a — —int\ —a (4.15)
3 VTT

and the vector of output voltage is assembled from framing vectors, Vx


and Vy (V2 and V3 in Sextant 3) and the zero vector (or vectors), v^ (VQ
or V7). Specifically, inverter states producing vectors Vx, Vy, and v^ are
imposed with durations ^x^sw ^Y^SW ^^^ ^z^sw respectively, where T^^
= 1/fsw denotes the switching period, that is, the length of the switching
interval. Coefficients dx, dy, and J^, which can be called state duty ratios,
express relative durations of individual states. Clearly,
dv + d\^ -\- dy = 1. (4.16)

'^2-yx

V^=Vy^

FIGURE 4.24 Illustration of the principle of the SVPWM technique.


78 CONTROL OF INDUCTION MOTORS

Duty ratios dx and (iy ^^^ be found from equation


V* = dxVx + dyVY (4.17)
decomposed into its real and imaginary parts to produce two real-
coefficient equations. The solution is

iH
dx = m s i n ( - - p ) (4.18)

and
dy = msin((5), (4.19)
where m denotes the so-called modulation index. The modulation index
can be defined in terms of the magnitude, V, of the voltage vector and
that, Vmax» of the maximum voltage vector possible to be generated using
pulse width modulation, as

m = --^. (4.20)
•^max

The duty ratio, d^, of the zero state is found from Eq. (4.16) as
dz= I - dx- dy. (4.21)
The maximum voltage vector is obtained when only the nonzero
vectors are utilized, that is, rf^ ~ 0- Then, the trajectory of the vector
forms the circle shown in Figure 4.24. Consequently (see also Figure
4.23), V^ax = V3^i/2. It means that the maximum available peak value
of the fundamental line-to-neutral output voltage of a PWM inverter is
[see Eq. (4.14)], and that of the fundamental line-to-line voltage
equals the supply dc voltage, V^. For comparison, as already mentioned,
the fundamental output voltage in the square-wave mode of operation is
1.1 V-,
Eqs. (4.18), (4.19), and (4.21) specify durations of individual states
of the inverter but not their sequence within the switching interval. Two
such sequences, tentatively called a high-performance sequence and a
high-efficiency sequence, are conmionly used. The high-performance state
sequence is | X - Y - Zj | Y - X - Z2 | ..., where zero states Z^ and Z2
are such that the transition from one state to another involves switching
in one phase only. It means that only one switching variable changes its
value from zero to one or vice versa. For example, in Sextant 2, where
X = 6 and Y = 2, the high-quality sequence i s | 6 - 2 - 0 | 2 - 6 - 7 |
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 79

..., that is, in the binary, abc, notation, | 110 - 010 - 000 | 010 - 110 -
111 |.... This state sequence, which for a given number, A^, of switching
intervals per cycle of the output voltage results in the best quality of
output currents, yields N/2 switching pulses per switch and per cycle.
The number of switchings can further be reduced, at the expense of
slightly increased distortion of current waveforms, when the high-
efficiency state sequence, | X - Y - Z | Z - Y - X | ..., is employed. Here,
Z = 0 in even sextants, and Z = 7 in odd sextants. If Sextant 2 is again
used as an example, the high-efficiency sequence i s | 6 - 2 - 0 | 0 - 2 -
6 I ..., or I 110 - 010 - 000 I 000 - 010 - 110 |.... Note that switching
variable c is zero throughout the whole sextant in question. With this
state sequence, the number of switching pulses per switch and per cycle
is N/3 + 1, that is, almost one-third lower than that with the high-quality
sequence, which results in a proportional reduction of switching losses
in the inverter.
EXAMPLE 4.2 A switching interval within which the space vector
of output voltage is to follow the reference vector v* = 240 Z170°
V is considered. The inverter is supplied from a 430-V dc voltage
source, and the switching frequency is 2 kHz. Find the duration and
high-quality sequence of inverter states in the interval in question.
The reference voltage vector is in Sextant 3 and, according to
Eq. (4.18), the local angular position of v* is 50°. The maximum
available magnitude of the space vector of output voltage is Vj^ax ~
V s X 430/2 = 372.4 V. Thus, the modulation index is m = 240/
372.4 = 0.644. Consequently, the duty ratios of individual states are
dx = 0.644 sin(60° - 50°) = 0.112, Jy = 0-644 sin(50°) = 0.493,
and dz= 1 — 0.112 — 0.493 = 0.395. Because the switching interval,
T ; ^ is 0.5 ms long, the State X should last 0.112 X 0.5 = 0.056 ms.
State Y should last 0.493 X 0.5 = 0.247 ms, and State Z should last
0.395 X 0.5 = 0.197 ms.
Sextant 3 is framed by vectors V2 and V3. Hence, assuming the
X - Y - Zj state sequence in the considered switching interval, the
sequence of states is 2 (0 to 0.056 ms), followed by 3 (0.056 ms to
0.303 ms), and 7 (0.303 ms to 0.5 ms). Note that in the next switching
interval the state sequence should be Y - X - Z2, that is, 3 - 2 - 0
(assuming that the reference vector will still be in the same sex-
tant). •
Typical waveforms of the output current, i^, in the voltage source
inverter in the square-wave and PWM operation modes are shown in
Figures 4.25(a) and 4.25(b), respectively, with an induction motor as a
80 CONTROL OF INDUCTION MOTORS

FIGURE 4.25 Waveforms of the output current in a voltage source inverter feeding
an induction motor: (a) square-wave operation, (b) PWM operation.

load. Note the similarity of the output current in a PWM inverter to that
of the input current in a PWM rectifier (see Figure 4.10).
In the three-level inverter, three states are employed for each leg, so
the switching variables are of the ternary format. Specifically,
0 if SI S2 are OFF and S3, S4 are ON 1
1 if SI S4 are OFF and S2, S3 are ON (4.22)
2 if SI S2 are ON and S3, S4 are OFF
The output line-to-line voltages are given by

[VAB
A 1 -1 0 "1fal
VBC 0 1 - 1 \b (4.23)
2 -1 0
L^CAJ 1 J\c\
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 81

and the line-to-neutral ones by

[VAN 2 -1 -11 Vol


VBN -1 2 -1 b (4.24)
6 -1 -1 \c\
L^CN_ 2J
The abc^ notation can be used for the 27 states of the three-level inverter.
Corresponding per-unit space vectors of the line-to-neutral voltage are
shown in Figure 4.26. States 0, 13, and 26 produce zero vectors.
The greater number of states allows for higher quality currents than
those generated in two-level inverters. Even the square-wave mode of
operation, illustrated in Figure 4.27, results in line-to-neutral voltage
waveforms well approximating sinusoids. In the PWM mode, to produce
currents of comparable quality, a three-level inverter can operate with a
switching frequency much lower than that of a two-level inverter.
As shown in Figure 4.28, in the RDCL inverters, individual "pulses"
of the output voltage are actually trains of resonant, usually clipped, pulses
of this voltage, an integer number of pulses in each train. This somewhat
limits the flexibiUty of control of these inverters. In contrast, as akeady
mentioned, the ARCP inverters are capable of true pulse width modulation,
that is, the output voltage pulses are similar to those in hard-switching
inverters.

t^n=fi3=t'26=0

1
V, V2,
^2)

Vi
e // / t^21

VA = ^'l8

^^ '^K r . ^ \ — d
1

^^19
Vs

^^Vn X
V2 ^20

FIGURE 4.26 Space vectors of the line-to-neutral voltage (per unit) in the three-
level voltage source inverter.
82 CONTROL OF INDUCTION MOTORS

STATE: 18 21 24 15 6 7 8 5 2 11 20 19

FIGURE 4.27 Output voltage waveforms in a three-level inverter in the square-wave


operation mode.

"AB

BC

'CA

0 • a)t

FIGURE 4.28 Output voltage waveforms in a resonant dc link inverter.


CHAPTER 4 / POWER ELECTRONIC CONVERTERS 83

In many ASDs, the open-loop voltage control is insufficient to ensure


the required quality of dynamic performance of the drive. In these drives,
the motor current is closed-loop controlled, the voltage control being
subordinated to the current control. Two basic approaches to current
control, each with a number of variants, have emerged over the years,
the bang-bang control and linear control
The simplest version of the bang-bang control, based on the so-called
hysteresis controllers, is illustrated in Figure 4.29. In each phase, the
actual current is compared with the reference current, and the difference
(current error) is applied to a hysteresis controller, whose output signal
constitutes the switching variable for this phase. Taking phase A as an
example, the characteristic of the hysteresis controller is given by

0 if A/A < " 2


(4.25)
h '
1 if M^>

INVERTER

MOTOR

CO

§
o
(/)
UJ
Q:
IxJ
(/)
>-
X

FIGURE 4.29 Block diagram of the bang-bang current control scheme.


84 CONTROL OF I N D U C T I O N MOTORS

where A/^ denotes the current error and h is the width of the tolerance
band. With the current error within the tolerance band, the value of a
remains unchanged. The bang-bang current control is characterized by a
fast response to rapid changes of the reference current. Many modifications
of the basic scheme have been proposed to stabilize the switching fre-
quency and reduce the interaction between phases.
In a linear current control system, linear controllers are used to generate
reference signgils for the inverter's pulse width modulator. Such a system
is shown in Figure 4.30. Output currents /^ and IQ are measured and
converted into the i^ and i^ components of the current space vector i.
Specifically,
i = k + Jiq^ (4.26)
where
• - 3. (4.27)

and

^q = (4.28)

WVERTER

MOTOR

UNEAR
CONTRaLERS
ta |6 i
i . 1 \ L

PULSE WDTH MODULATOR

^A

FIGURE 4.30 Block diagram of the linear current control scheme.


CHAPTER 4 / POWER ELECTRONIC CONVERTERS 85

Signals i^ and i^ are compared with their reference counterparts, / | and


/*, and the respective current errors, A/^ and A/q, are applied to linear
controllers, typically of the proportional-integral (PI) type. These produce
components v$ and v* of the reference voltage vector, v*, for the pulse
width modulator, which generates the switching variables, a, b, and c,
for the inverter. Usually, the SVPWM technique is used in the modulator.

4.6 CONTROL OF CURRENT SOURCE INVERTERS

Current source inverters are less commonly used in induction motor ASDs
than voltage source inverters, mostly due to the poorer quality of output
currents. Still, they have certain advantages, such as imperviousness to
overcurrents, even with a short circuit in the inverter or the load. The
absence of freewheeling diodes further increases the reliability. Also,
current source inverters are characterized by inherently excellent dynamics
of the phase angle control of the output current. Such control is realized
by changing the state of inverter and the resultant redirecting of the source
current. However, the magnitude adjustments of output currents can only
be performed in the supplying rectifier. The rectifier allows bidirectional
flow of power, and, because the input current is always positive, the input
voltage becomes negative when the power flows from the load to the
supply power system. Therefore, semiconductor power switches used in
a current source inverter must have the reverse blocking capability.
In contrast with voltage source inverters, the simultaneous on-state
of both switches in the same inverter leg is safe and recommended for a
short period of time initiating a state change of the inverter. This is to
avoid the danger of interrupting the current, which would result in an
overvoltage. Consequently, switching variables are defined differently
than those in the voltage source inverter. In the subsequent considerations,
variables a, b, and c are assigned to switches SA, SB, and SC (e.g., a =
1 means that SA is on), and variables a', b\ and c' to switches SA', SB',
and S C (see Figure 4.14). Then, the output line currents, ip^, /g, and /(-,
of the current source inverter can be expressed as

/A = (a - a')/i,
(4.29)
/B = {b- b%,
and

ic = (c - c%.
86 CONTROL OF INDUCTION MOTORS

where I^ denotes the constant input current. If the motor has a delta-
connected stator, then the currents, /AB» ^*BC' and /CA» ^^ the individual
phase windings are given by

'AB 1 1 -1 01 'A
'BC 0 1 -1 'B (4.30)
"3
'CA -1 0 1 ic
The peak value of the fundamental line currents is approximately 1.1 7^.
Switching variables in the square-wave operation mode of the current
source inverter are shown in Figure 4.31 and the resultant output current
waveforms in Figure 4.32. It can be seen that at any time only two switches

(j)t

\(J)t

-\(A)t

27T

FIGURE 4.31 Switching variables in the current source inverter in the square-wave
operation mode.

Hri i ''ut
1 1 1 1 i
i i i i \f'\f
1 i 1 1
1 1 i i i ' ut
1 1 1 ^
"AB j i 1 1 .
! 1 { IWI
: i 1 I
"BC i i
- 1 = \ ! 1 —. \(j)t
! \ r ' 1
"CA 1 i 1 i i \
i ] U 1 i ! 1 Cut

0 TV 27T

FIGURE 4.32 Output current waveforms in a current source inverter in the square-
wave operation mode.
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 87

conduct the output currents. This conduction pattern is typical for the
voltage source rectifier which, indeed, is an inverse of the current source
inverter. Space vectors of the line currents associated with individual
states of the inverter are depicted in Figure 4.33 in the per-unit format,
with the dc input current, /j, taken as the base current. The states are
designated by letters denoting conducting switches of the inverter. For
example. State AB represents the situation in which the conducting
switches are SA and SB'. Simultaneous conduction of both switches in
the same leg results in a zero vector (States AA, BB, and CC).
Similarly to the voltage source inverter, the square-wave mode of
operation requires each switch to be turned on and off once per cycle
only. This is one of the reasons that current source inverters have typically
been used in high-power drives, with large and slow semiconductor power
switches employed in the inverter.
PWM current source inverters, equipped with output capacitors (see
Figure 4.14), are characterized by significantly higher quality of output
currents than that in square-wave inverters. Several PWM techniques have
been developed, one of them, based on a trapezoidal modulating function,
illustrated in Figure 4.34. The modulating function signal, jc, is compared
with a triangular carrier signal, y, and the intersection instants determine

FIGURE 4.33 Space vectors of the line currents (per unit) in the current source
inverter.
88 CONTROL OF INDUCTION MOTORS

FIGURE 4.34 Illustration of a PWM technique for a current source inverter.

the value of switching variable a. With respect to a, switching variables


b and c are delayed by one-third and two-thirds of the period of output
current. Variables a\b\ and c' are shifted with respect to a, b, and c by
a half period. The best attenuation of low-order harmonics in the output
currents is achieved with the peak value of the modulating function equal
0.82 of the peak of the carrier signal. PWM current source inverters are
not feasible for high-power drives because of the excessive size of the
required capacitors.

4.7 SIDE EFFECTS OF CONVERTER OPERATION IN


ADJUSTABLE SPEED DRIVES

Although effective and efficient, power electronic converters used in ASDs


give birth to certain undesirable side effects. Their extent and severity
have spawned the popular phrase about "secondary issues, but primary
concerns."
The problem of low-order current harmonics generated by diode recti-
fiers in the supply system has already been mentioned in Section 4.2. To
avoid bulky passive input filters, much smaller active filters or PWM
rectifiers can be used instead. However, their mode of operation results
in clusters of high-order harmonics in the input currents. These high-
frequency currents cause, in turn, high-frequency components of voltage
in the utility grid, called voltage noise or, more generally, conducted
electromagnetic interference (EMI), Also, certain amount of voltage noise
leaks to the supply system from PWM inverters via the dc link and input
rectifier. In addition, the very process of switching fast semiconductor
power switches produces electromagnetic noise in the megahertz range
CHAPTER 4 / POWER ELECTRONIC CONVERTERS 89

of frequency. Part of the energy carried by high-frequency noise is radiated


by overhead transmission Unes and other wiring acting as antennas. Be-
cause both the conducted and radiated EMI tend to disturb the operation
of communication systems and other sensitive electronic equipment, a
special low-pass filter, called the EMI or RF (radio frequency) filter, must
be installed at the input to the rectifier-inverter cascade.
High values of the rate of voltage change, dv/dt, resulting from the
short rise times of voltage pulses generated by the inverter, cause insulation
degradation in stator windings due to unequal voltage distribution among
individual coils. Another voltage hazard is associated with the cable
connecting the motor to the inverter. For voltage pulses, the cable, if
sufficiently long, constitutes a transmission line. The ringing and reflected-
wave phenomena result in overvoltages reaching twice the normal ampli-
tude of the pulses if the cable exceeds certain critical length (the rule of
thumb for the critical length is 60 meters per each microsecond of the
rise time). Clearly, the long-cable overvoltages are hazardous for the stator
winding, and they must be reduced by resistive-capacitive (RC) filters
placed at either end of the cable. The filters, a necessary evil, increase
the bulk and cost of the drive system.
The induction motor is a three-wire load, with individual phases
coupled to ground via stray capacitances of the motor. The stray capaci-
tances, although small, provide low-impedance paths to ground for tran-
sient currents generated by high values of dv/dt of the inverter output
voltage. One of these capacitances is that of the motor bearings, in which
the thin film of lubricant constitutes a dielectric. Electric charge accumu-
lates on the rotor assembly until the resultant shaft-to-ground voltage
exceeds the dielectric capability of the bearing lubricant. The flashover
currents significantly accelerate wear of the bearings due to so-called
electrical discharge machining (EDM). Transverse grooves, pits, or frost-
ing may appear in the bearing race after just a few months of operation
of the drive. Direct preventive measures proposed include the outer-race
insulation, conductive lubricant, dielectric-metaillic Faraday shield in the
motor airgap, or shaft grounding system. To reduce the common-mode
(neutral to ground), dv/dt induced leakage current, a common-mode choke
or transformer can be added at the motor terminals.
The common-mode voltage, responsible for the shaft voltage and
bearing and leakage currents, can be reduced by software and hardware
means. Denoting by v^, v^, and v^ voltages of stator terminals with respect
to ground, the common-mode voltage, v^^ is given by

Vcm = | ( V A + VB + Vc). (4.31)


90 CONTROL OF INDUCTION MOTORS

If the dc-link capacitance is composed of two capacitors with a grounded


common point, v^ = V^(a — 0.5), v^ = V^(b — 0.5), and v^ = V^ic —
0.5). Consequently, if only inverter voltage vectors Vj, V3, and V5 or V2,
V4, and V6 are used, the conmion-mode voltage remains constant, that is,
it becomes a dc voltage, at the level of -V/6 or V^/6, respectively. Such
PWM strategy is impractical, but it indicates control possibilities for
alleviating the problems associated with the common-mode voltage. Sev-
eral PWM techniques for common-mode voltage reduction have been
developed.
Hardware solutions include extended inverter configurations, such as
multilevel, double-bridge, and four-leg topologies. Another option is to
connect, in parallel with the standard voltage source inverter, an active
(switched) circuit for cancellation of the common-mode voltage. Such a
circuit, based on an emitter follower, is shown in Figure 4.35. It imposes,
via a transformer coupling, a compensating voltage at the inverter output.
Voltage pulses of fixed frequency result in annoying tonal noise in
motors fed from PWM inverters. To quiet a drive system, the switching
frequency should be raised to a supersonic (greater than 20 kHz) level.
This is not always feasible, and random PWM can be used instead. It
consists in random changes of the switching frequency, so that the har-
monic power (watts) in the voltage spectrum is transferred to continuous
power density (watts per hertz). As a result, the tonal noise is converted
into "static," easily blending with the background noise. The random

RECTFER MVERTER
DC LIK
MOTOR

CONM)N-MODE VOLTAGE
CANCELLER

FIGURE 4.35 Common-mode voltage canceler for a voltage source inverter.


CHAPTER 4 / POWER ELECTRONIC CONVERTERS 9 I

PWM also reduces the peak and quasi-peak EMI conducted from the
drive system to the utility grid.

4.8 SUMMARY

Induction motors in ASDs are supplied from inverters, which are dc to


ac power electronic converters. The dc supply voltage for inverters is
provided by rectifiers. These usually have the three-phase bridge topology
and are based on power diodes or, if control of the input current or inverted
power flow is required, on SCRs. Poor quality of currents drawn from
the power system is the major disadvantage of such rectifiers, and expen-
sive input filters must be installed to alleviate this problem. The size of
input filters can greatly be reduced when PWM rectifiers are employed.
They can be of the voltage source or current source type, the latter
converters being capable of boosting the output voltage above the peak
value of the ac input (line-to-line) voltage.
Two-level voltage source inverters are most common in practice, but
alternative topologies, such as soft-switching and three-level inverters,
are gaining ground. Current source inverters are also used, usually in
high-power drives. These inverters are more robust than voltage source
inverters, and the dynamics of current phase control is excellent. The
current quality in the square-wave mode of operation is poor, though. In
that respect, PWM current source inverters, with output capacitors, are a
better solution, but the required size of the capacitors limits the power
range of these converters.
A cascade of the rectifier, dc link, and inverter constitutes a frequency
changer. PWM frequency changers, characterized by high-performance
operation and capable of the reversed power flow, can be of the voltage-
source or current-source type. The voltage source PWM frequency changer
is based on a voltage source PWM rectifier and a current source inverter.
Conversely, in the current source PWM frequency changer, a current
source rectifier and a voltage source inverter are employed.
The most popular PWM technique for voltage source inverters is
based on the concept of voltage space vectors. A voltage space vector is
obtained by transformation of values of actual voltages in a three-phase
system into a complex number. Three-phase currents in a three-wire system
can similarly be transformed.
In contrast to voltage source inverters which, in the PWM mode,
allow control of both the frequency and magnitude of the output voltages,
current source inverters are incapable of the magnitude control of output
92 CONTROL OF INDUCTION MOTORS

currents. This can only be done in the current source (controlled rectifier)
supplying the inverter.
Power electronic converters used in ASDs with induction motors
cause serious side effects. These include harmonic pollution of the supply-
ing power system, conducted and radiated electromagnetic interference,
insulation degradation in stator windings, overvoltages in the cable con-
necting the inverter and motor, common-mode voltage resulting in acceler-
ated bearing deterioration and leakage currents, and annoying acoustic
noise. Most of these side effects can be remedied using appropriate filters.
Other measures, such as multilevel, double-bridge, and four-leg inverters,
active circuits for the common-mode voltage cancellation, and modified
PWM strategies are also employed in practical drive systems.

You might also like