Chapter 4 DC To AC Conversion (INVERTER)

Chapter 4
DC to AC Conversion
(INVERTER)
• General concept
• Single-phase inverter
• Harmonics
• Modulation
• Three-phase inverter
Power Electronics and 1

Drives (Version 3-2003):
Dr. Zainal Salam UTM-JB
DC to AC Converter (Inverter)
• DEFINITION: Converts DC to AC power by
switching the DC input voltage (or current) in a
pre-determined sequence so as to generate AC
voltage (or current) output.
• General block diagram
IDC Iac
VDC Vac
−
−
• TYPICAL APPLICATIONS:
– Un-interruptible power supply (UPS), Industrial
(induction motor) drives, Traction, HVDC

Simple square-wave inverter (1)
• To illustrate the concept of AC waveform

generation
S1 S3
S4 S2

AC Waveform Generation
S1,S2 ON; S3,S4 OFF for t1 < t < t2

vO
S1 S3 VDC
VDC
t
+ vO − t1 t2
S4 S2
S3,S4 ON ; S1,S2 OFF for t2 < t < t3

vO
S1 S3
VDC t2 t3
+ vO − t
S4 S2
-VDC

AC Waveforms
INVERTER OUTPUT VOLTAGE
Vdc
π 2π
-Vdc
FUNDAMENTAL COMPONENT
V1
4VDC
π
3RD HARMONIC
V1
3
5RD HARMONIC
V1
5

Harmonics Filtering
DC SUPPLY INVERTER (LOW PASS) FILTER LOAD
L
+ +
C
vO 1 vO 2
−
−
BEFORE FILTERING AFTER FILTERING

vO 1 vO 2
• Output of the inverter is “chopped AC voltage with

zero DC component”. It contain harmonics.
• An LC section low-pass filter is normally fitted at

the inverter output to reduce the high frequency
harmonics.
• In some applications such as UPS, “high purity” sine

wave output is required. Good filtering is a must.
• In some applications such as AC motor drive,
filtering is not required.
Variable Voltage Variable
Frequency Capability
Vdc2 Higher input voltage

Higher frequency
Vdc1 Lower input voltage

Lower frequency
• Output voltage frequency can be varied by “period”

of the square-wave pulse.
• Output voltage amplitude can be varied by varying

the “magnitude” of the DC input voltage.
• Very useful: e.g. variable speed induction motor

drive

Output voltage harmonics/
distortion
• Harmonics cause distortion on the output voltage.
• Lower order harmonics (3rd, 5th etc) are very

difficult to filter, due to the filter size and high filter
order. They can cause serious voltage distortion.
• Why need to consider harmonics?

– Sinusoidal waveform quality must match TNB
supply.
– “Power Quality” issue.
– Harmonics may cause degradation of
equipment. Equipment need to be “de-rated”.
• Total Harmonic Distortion (THD) is a measure to

determine the “quality” of a given waveform.

Total Harmonics Distortion (THD)
Voltage THD : If Vn is the nth harmonic voltage,
∞
(Vn, RMS )2
THDv = n= 2
V1, RMS
V2, RMS 2 + V3, RMS 2 + .... + V2, RMS 2

=
V1, RMS
If the rms voltage for the vaveform is known,
∞
(VRMS )2 − (V1, RMS )2
THDv = n= 2
V1, RMS
Current THD :
∞
(I n, RMS )2
THDi = n =2
I1, RMS
V
In = n
Zn
Z n is the impedance at harmonic frequency.

Fourier Series
• Study of harmonics requires understanding of wave

shapes. Fourier Series is a tool to analyse wave
shapes.
Fourier Series
1 2π
ao = f (v )dθ (" DC" term)
π 0
2π
1
an = f (v) cos(nθ )dθ (" cos" term)
π 0
2π
1
bn = f (v) sin (nθ )dθ ("sin" term)
π 0
Inverse Fourier
1 ∞
f (v) = ao + (an cos nθ + bn sin nθ )
2 n =1
where θ = ωt

Harmonics of square-wave (1)
Vdc
θ ω
π 2π
-Vdc
1 π 2π
ao = Vdc dθ + − Vdc dθ = 0
π 0 π
Vdc π 2π
an = cos(nθ )dθ − cos(nθ )dθ = 0
π 0 π
Vdc π 2π
bn = sin (nθ )dθ − sin (nθ )dθ
π 0 π

Harmonics of square wave (2)
Solving,
V
[ π
bn = dc − cos(nθ ) 0 + cos(nθ ) π
nπ
2π
]
Vdc
= [(cos 0 − cos nπ ) + (cos 2nπ − cos nπ )]
nπ
Vdc
= [(1 − cos nπ ) + (1 − cos nπ )]
nπ
2V
= dc [(1 − cos nπ )]
nπ
When n is even, cos nπ = 1

bn = 0
(i.e. even harmonics do not exist)
When n is odd, cos nπ = −1

4Vdc
bn =
nπ
Spectra of square wave
Normalised
Fundamental
1st
3rd (0.33)
5th (0.2)
7th (0.14)
9th (0.11)
11th (0.09)
1 3 5 7 9 11
n
• Spectra (harmonics) characteristics:

– Harmonic decreases with a factor of (1/n).
– Even harmonics are absent
– Nearest harmonics is the 3rd. If fundamental is
50Hz, then nearest harmonic is 150Hz.
– Due to the small separation between the
fundamental an harmonics, output low-pass
filter design can be very difficult.
Quasi-square wave (QSW)
Vdc
α α α
π 2π
-Vdc
Note that an = 0. (due to half - wave symmetry)
bn = 2
1 π −α
π α
2V
nπ
π −α
Vdc sin (nθ )dθ = dc − cos nθ α [ ]
2Vdc
= [cos(nα ) − cos n(π − α )]
nπ
Expanding :
cos n(π − α ) = cos(nπ − nα )
= cos nπ cos nα + sin nπ sin nα = cos nπ cos nα
2Vdc
bn = [cos(nα ) − cos nπ cos nα ]
nπ
2Vdc
= cos(nα )[1 − cos nπ ]
nπ
Harmonics control
If n is even, bn = 0,
4Vdc
If n is odd, bn = cos(nα )
nπ
In particular, amplitude of the fundamental is :
4Vdc
b1 = cos(α )
π
Note :
The fundamental , b1 , is controlled by varying
Harmonics can also be controlled by adjusting α ,
Harmonics Elimination :
For example if α = 30 o , then b3 = 0, or the third
harmonic is eliminated from the waveform. In
general, harmonic n will be eliminated if :
90o
α=
n
Example
A full - bridge single phase inverter is fed by square wave
signals. The DC link voltage is 100V. The load is R = 10R
and L = 10mH in series. Calculate :
a) the THDv using the " exact" formula.
b) the THDv by using the first three non - zero harmonics
c) the THDi by using the first three non - zero harmonics
Repeat (b) and (c) for quasi - square wave case with α = 30
degrees

Half-bridge inverter (1)
S1 ON
Vdc S2 OFF
+
S1 2
VC1
-
− V +
Vdc o
G 0
t
RL
+
VC2 S2
- Vdc
−
2 S1 OFF
S2 ON
• Also known as the “inverter leg”.
• Basic building block for full bridge, three phase

and higher order inverters.
• G is the “centre point”.
• Both capacitors have the same value. Thus the DC

link is equally “spilt” into two.
• The top and bottom switch has to be

“complementary”, i.e. If the top switch is closed
(on), the bottom must be off, and vice-versa.
Shoot through fault and
“Dead-time”
• In practical, a dead time as shown below is required
to avoid “shoot-through” faults, i.e. short circuit
across the DC rail.
• Dead time creates “low frequency envelope”. Low

frequency harmonics emerged.
• This is the main source of distortion for high-quality

sine wave inverter.
S1
signal
+ S1 (gate)
Ishort
Vdc G
S2
RL
signal
− (gate)
S2
"Shoot through fault" . td td

Ishort is very large
"Dead time' = td

Single-phase, full-bridge (1)
• Full bridge (single phase) is built from two half-
bridge leg.
• The switching in the second leg is “delayed by 180

degrees” from the first leg.
VRG
Vdc
2
LEG R LEG R' π 2π ωt
+
Vdc Vdc
S1 S3 −
2 VR 'G 2
+ - Vdc
+ Vo - 2
Vdc R R'
G π 2π ωt
-
+ −
Vdc
2
Vdc S4 S2 Vo
2
Vdc
-
π 2π ωt
Vo = V RG − VR 'G
G is " virtual groumd"
− Vdc

Three-phase inverter
• Each leg (Red, Yellow, Blue) is delayed by 120

degrees.
• A three-phase inverter with star connected load is

shown below
+Vdc
+
Vdc/2 S1 S3 S5
−
G R Y B
iR iY iB
+
S4 S6 S2
Vdc/2
−
ia ib
ZR ZY ZB

Three phase inverter waveforms
Inverter Phase
Voltage VDC/2
(or pole switching
waveform)
VRG -V /2
DC
1200
VDC/2
VYG
-VDC/2
2400
VDC/2
VBG
-VDC/2
VDC
lIne-to -ine
Voltage
VRY
-VDC
2VDC/3
Six-step
Waveform VDC/3
VRN
-VDC/3
-2VDC/3
Interval 1 2 3 4 5 6
Positive device(s) on 3 3,5 5 1,5 1 1,3
Negative device(s) on 2,4 4 4,6 6 2,6 2
Quasi-square wave operation voltage waveforms

Pulse Width Modulation (PWM)
Modulating Waveform Carrier waveform
+1
M1
−1
Vdc
2
0 t 0 t1 t2 t3 t4 t5
Vdc
−
2
• Triangulation method (Natural sampling)

– Amplitudes of the triangular wave (carrier) and
sine wave (modulating) are compared to obtain
PWM waveform. Simple analogue comparator
can be used.
– Basically an analogue method. Its digital

version, known as REGULAR sampling is
widely used in industry.

PWM types
• Natural (sinusoidal) sampling (as shown
on previous slide)
– Problems with analogue circuitry, e.g. Drift,
sensitivity etc.
• Regular sampling
– simplified version of natural sampling that
results in simple digital implementation
• Optimised PWM
– PWM waveform are constructed based on
certain performance criteria, e.g. THD.
• Harmonic elimination/minimisation PWM
– PWM waveforms are constructed to eliminate
some undesirable harmonics from the output
waveform spectra.
– Highly mathematical in nature
• Space-vector modulation (SVM)
– A simple technique based on volt-second that is
normally used with three-phase inverter motor-
drive
Modulation Index, Ratio
+1
M1
−1
Vdc
2
0 t0 t1 t 2 t 3 t 4 t5
Vdc
−
2
Modulation Index (Modulation Depth) = M I :

Amplitude of the modulating waveform
MI =
Amplitude of the carrier waveform
Modulation Ratio (Frequency Ratio) = M R (= p )

Frequency of the carrier waveform
MR = p =
Frequency of the modulating waveform
Modulation Index, Ratio
Modulation Index deterrmines the output

voltage fundamental component
If 0 < M I < 1,
V1 = M I Vin
where V1 , Vin are fundamental of the output
voltage and input (DC) voltage, respectively.
−−−−−−−−−−−−−−−−−−−−−−−−−−−−
Modulation ratio determines the incident (location)
of harmonics in the spectra.
The harmonics are normally located at :

f = kM R ( f m )
where f m is the frequency of the modulating signal
and k is an integer (1,2,3...)

Regular sampling
t1 t2 Sinusoidal modulating Carrier, vc(t)

waveform, vm(t)
2π
t
π
Regular sampling waveform, vs (t )

t'1 t'2
vpwm t
Regular sampling PWM

Asymmetric and symmetric
regular sampling
T
+1 M1 sin ω mt
sample
point
t
T 3T 5T π
4 4 4 4
−1
Vdc
2 asymmetric
sampling
t
t0 t1 t2 t3
symmetric
sampling
V
− dc
2
Generating of PWM waveform regular sampling

Bipolar Switching

+1
M1
−1
Vdc
2
0 t0 t1 t2 t 3 t 4 t5
Vdc
−
2

Unipolar switching
A Carrier waveform B
(a)
S1
(b)
S3
(c)
V pwm
(d)
Unipolar switching scheme

Bipolar PWM switching: Pulse-
width characterization
∆ modulating
∆ δ= carrier
4 waveform waveform
π 2π
kth
pulse
π 2π
δ 1k
δ 2k
αk

The kth Pulse
∆
δ0 δ0 δ0 δ0
+ Vdc
2
δ 2k
δ1k
+ Vdc
2
αk
The kth PWM pulse

Determination of switching angles
for kth PWM pulse (1)
AS2
v Vmsin( θ )
AS1
+ Vdc
2
Ap1 Ap2
V
− dc
2
Equating the volt - second,

As1 = Ap1
As 2 = Ap 2

PWM Switching angles (2)
The Volt - second during the first half cycle
of the PWM pulse is given as :
Vdc Vdc
A p1 = (δ1k ) − (2δ o − δ1k )
2 2
= (Vdc )(δ1k − δ o )
Similarly for the second half,
Vdc
Ap2 = (δ 2k ) − Vdc (2δ o − δ 2k )
2 2
= (Vdc )(δ 2k − δ o )
The volt - second supplied by the sinusoid,
αk
As1 = Vm sin θdθ = Vm [cos(α k − 2δ o ) − cos α k ]
α k − 2δ o
= 2Vm sin δ o sin(α k − δ o )
Similarly,
As 2 = 2δ oVm sin(α k + δ o )

Switching angles (3)
For small angle δ o
sin δ o → δ o ,
As1 = 2δ oVm sin(α k − δ o )
As 2 = 2δ oVm sin(α k − δ o )
To derive the modulation strategy,
A p1 = As1;
A p 2 = As 2
Hence, for the the first half cycle of PWM pulse,
(Vdc )(δ1k − δ o ) = 2δ oVm sin(α k − δ o )
2Vm
(δ1k − δ o ) = (δ o sin(α k − δ o )
Vdc
By definition, the Modulation Ratio,
Vm
MI = is known as modulation
(Vdc 2 )
Thus, the pulse width for the first half cycle
of the PWM waveform is given by :
δ1k = δ o [1 + M I sin(α k − δ o )]

PWM switching angles (4)
Thus the leading edge switching angle of
the kth pulse is :
α k − δ1k
Using similar method, pulse width of the

second half cycle of PWM waveform :
δ 2k = δ o [1 + M I sin(α k + δ o )]
And the trailing edge angle :
α k + δ 2k
The above equation is valid for Asymmetric

Modulation, i.eδ1k and δ 2k are different.
For Symmetric Modulation,

δ1k = δ 2k = δ k
Hence
δ k = δ o [1 + M I sin α k ]
Example
• For the PWM shown below, calculate the switching
angles pulses no. 2.
carrier
2V
waveform
1.5V
π 2π
modulating
waveform
1 2 3 4 5 6 7 8 9
π t13 t15 t17

t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12 t14 t16 t18 2π
α1

Harmonics of bipolar PWM
∆
δ0 δ0 δ0 δ0
Assuming the PWM + Vdc
waveform is half 2
δ 2k
δ1k
wave symmetry,
harmonic
content of each
(kth) PWM pulse + Vdc
2
can be computed as : αk
1T
bnk = 2 f (v) sin nθdθ
π 0
α k −δ1k
2 V
= − dc sin nθdθ
π α −2δ 2
k o
α k +δ 2 k
2 Vdc
+ sin nθdθ
π α −δ 2
k 1k
α k + 2δ o
2 V
+ − dc sin nθdθ
π α +δ 2
k 2k

Harmonics of Bipolar PWM
Which can be reduced to :
Vdc
bnk = − {cos n(α k − 2δ o ) − cos n(α k − δ1k )
nπ
+ cos n(α k + δ 2 k ) − cos n(α k − δ1k )
+ cos n(α k + δ 2 k ) − cos n(α k + 2δ o )}
Yeilding,
2Vdc
bnk = [cos n(α k − δ1k ) − cos n(α k − 21k )
nπ
+ 2 cos nα k cos n 2δ o ]
This equation cannot be simplified
productively.The Fourier coefficent for the
PWM waveform isthe sum of bnk for the p
pulses over one period, i.e. :
p
bn = bnk
k =1
Next slide shows the computation of this equation.

PWM Spectra
M I = 0.2
Amplitude
M I = 0.4
1. 0
0 .8
M I = 0.6
0.6
Modulation
Index
0.4 M I = 0 .8
0.2
0 M I = 1 .0
p 2p 3p 4p
Fundamental
NORMALISED HARMONIC AMPLITUDES FOR
SINUSOIDAL PULSE-WITDH MODULATION

PWM spectra observations
• The harmonics appear in “clusters” at multiple of
the carrier frequencies .
• Main harmonics located at :
f = kp (fm); k=1,2,3....
where fm is the frequency of the modulation (sine)

waveform.
• There also exist “side-bands” around the main

harmonic frequencies.
• Amplitude of the fundamental is proportional to the

modulation index.
The relation ship is given as:
V1= MIVin
• The amplitude of the harmonic changes with MI.

Its incidence (location on spectra) is not.
• When p>10, or so, the harmonics can be

normalised. For lower values of p, the side-bands
clusters overlap-normalised results no longer apply.

Tabulated Bipolar PWM Harmonics
0.2 0.4 0.6 0.8 1.0

n
MI
1 0.2 0.4 0.6 0.8 1.0
MR 1.242 1.15 1.006 0.818 0.601
MR +2 0.016 0.061 0.131 0.220 0.318
MR +4 0.018
2MR +1 0.190 0.326 0.370 0.314 0.181

2MR +3 0.024 0.071 0.139 0.212
2MR +5 0.013 0.033
3MR 0.335 0.123 0.083 0.171 0.113

3MR +2 0.044 0.139 0.203 0.716 0.062
3MR +4 0.012 0.047 0.104 0.157
3MR +6 0.016 0.044
4MR +1 0.163 0.157 0.008 0.105 0.068

4MR +3 0.012 0.070 0.132 0.115 0.009
4MR+5 0.034 0.084 0.119

4MR +7 0.017 0.050

Three-phase harmonics
• For three-phase inverters, there is significant

advantage if MR is chosen to be:
– Odd: All even harmonic will be eliminated

from the pole-switching waveform.
– triplens (multiple of three (e.g. 3,9,15,21, 27..):
All triplens harmonics will be eliminated from
the line-to-line output voltage.
• By observing the waveform, it can be seen that with

odd MR, the line-to-line voltage shape looks more
“sinusoidal”.
• As can be noted from the spectra, the phase voltage

amplitude is 0.8 (normalised). This is because the
modulation index is 0.8. The line voltage amplitude
is square root three of phase voltage due to the
three-phase relationship

Effect of odd and “triplens”
π 2π
Vdc
2 V RG
Vdc
−
2
Vdc
2 VYG
Vdc
−
2
Vdc
V RY
− Vdc
p = 8, M = 0.6
Vdc
2 V RG
Vdc
−
2
Vdc
2 VYG
Vdc
−
2
Vdc
V RY
− Vdc
p = 9, M = 0.6
ILLUSTRATION OF BENEFITS OF USING A FREQUENCY RATIO
THAT IS A MULTIPLE OF THREE IN A THREE PHASE INVERTER

Spectra: effect of “triplens”
Amplitude
0.8 3 (Line to line voltage)

1.8
1.6
1.4
1.2
1.0
0.8
0.6
B
0.4
41 43 61 65 83 85
19 23 37 47 59 67 79 89
0.2
0 A
21 63
83 85
19 23 41 43 61 65 81 87
39 45 59 67 79 89
37 47 57 69 77 91 Harmonic Order
Fundamental
COMPARISON OF INVERTER PHASE VOLTAGE (A) & INVERTER LINE VOLTAGE
(B) HARMONIC (P=21, M=0.8)

Comments on PWM scheme
• It is desirable to have MR as large as possible.
• This will push the harmonic at higher frequencies

on the spectrum. Thus filtering requirement is
reduced.
• Although the voltage THD improvement is not

significant, but the current THD will improve
greatly because the load normally has some current
filtering effect.
• However, higher MR has side effects:

– Higher switching frequency: More losses.
– Pulse width may be too small to be constructed.
“Pulse dropping” may be required.

Example
The amplitudes of the pole switching waveform harmonics of the red
phase of a three-phase inverter is shown in Table below. The inverter
uses a symmetric regular sampling PWM scheme. The carrier frequency
is 1050Hz and the modulating frequency is 50Hz. The modulation
index is 0.8. Calculate the harmonic amplitudes of the line-to-voltage
(i.e. red to blue phase) and complete the table.
Harmonic Amplitude (pole switching Amplitude (line-to

number waveform) line voltage)
1 1
19 0.3
21 0.8
23 0.3
37 0.1
39 0.2
41 0.25
43 0.25
45 0.2
47 0.1
57 0.05
59 0.1
61 0.15
63 0.2
65 0.15
67 0.1
69 0.05


Chapter 4 DC To AC Conversion (INVERTER)

Cargado por

Información del documento

Descripción original:

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Chapter 4 DC To AC Conversion (INVERTER)

Cargado por

Copyright:

Formatos disponibles

Chapter 4

Power Electronics and 1

• General block diagram

Power Electronics and 2

• To illustrate the concept of AC waveform

Power Electronics and 3

S1,S2 ON; S3,S4 OFF for t1 < t < t2

S3,S4 ON ; S1,S2 OFF for t2 < t < t3

Power Electronics and 4

Power Electronics and 5

BEFORE FILTERING AFTER FILTERING

• Output of the inverter is “chopped AC voltage with

• An LC section low-pass filter is normally fitted at

• In some applications such as UPS, “high purity” sine

Vdc2 Higher input voltage

Vdc1 Lower input voltage

• Output voltage frequency can be varied by “period”

• Output voltage amplitude can be varied by varying

• Very useful: e.g. variable speed induction motor

Power Electronics and 7

• Harmonics cause distortion on the output voltage.

• Lower order harmonics (3rd, 5th etc) are very

• Why need to consider harmonics?

• Total Harmonic Distortion (THD) is a measure to

Power Electronics and 8

V2, RMS 2 + V3, RMS 2 + .... + V2, RMS 2

Power Electronics and 9

• Study of harmonics requires understanding of wave

Power Electronics and 10

Power Electronics and 11

When n is even, cos nπ = 1

When n is odd, cos nπ = −1

• Spectra (harmonics) characteristics:

Note that an = 0. (due to half - wave symmetry)

Power Electronics and 16

• Also known as the “inverter leg”.

• Basic building block for full bridge, three phase

• G is the “centre point”.

• Both capacitors have the same value. Thus the DC

• The top and bottom switch has to be

• Dead time creates “low frequency envelope”. Low

• This is the main source of distortion for high-quality

"Shoot through fault" . td td

Power Electronics and 18

• The switching in the second leg is “delayed by 180

Power Electronics and 19

• Each leg (Red, Yellow, Blue) is delayed by 120

• A three-phase inverter with star connected load is

Power Electronics and 20

Quasi-square wave operation voltage waveforms

Power Electronics and 21

• Triangulation method (Natural sampling)

– Basically an analogue method. Its digital

Power Electronics and 22

Modulation Index (Modulation Depth) = M I :

Modulation Ratio (Frequency Ratio) = M R (= p )

Modulation Index deterrmines the output

The harmonics are normally located at :

Power Electronics and 25

t1 t2 Sinusoidal modulating Carrier, vc(t)

Regular sampling waveform, vs (t )

Regular sampling PWM