EE466/EE966 Part 2: Power Electronics

Topic 7: DC-AC power conversion

 DC-AC power conversion

1. Introduction
2. Power semiconductor devices
3. Thermal management
4. Passive components for power electronics
5. Non-isolated DC-DC power converters
6. Isolated-output DC-DC power converters
7. DC-AC power conversion
8. AC-DC power conversion
9. Control of DC machines
 DC-AC power conversion

As mentioned in the introduction to this part of the module, DC-AC conversion may be
required for:

• Machine control: most often three-phase (although occasionally single phase and
sometimes five or more phases)

• Grid-tie inverters: can be three-phase or single-phase

• Therefore, applications can be such as AC drive systems, AC sources, Uninterruptible

Power Supply (UPS), and alternative energy power sources.
 DC-AC power conversion

• To control an AC machine we require a variable frequency, variable voltage, AC

supply.
• This may be generated using a switched-mode DC-AC converter circuit: often referred
to as an inverter.
• The input to the switch mode inverter is assumed to be a DC voltage source (large DC
capacitor is used on the DC side). Such inverters are referred to as voltage source
inverters (VSIs)*

50Hz AC
AC Vd motor

Diode Filter Switch mode

rectifier capacitor inverter

*For information: as mentioned in Topic 5, there is another family of converter called the current
source converter (CSC). When this is used for inversion (DC-AC conversion) it is, of course, referred
to as the current source inverter (CSI).
 DC-AC power conversion

Shown here is a three-phase inverter:

VDC

TR1 TR3 TR5

D1 D3 D5
L1 iA

L2 iB

L3 iC

TR2 TR4 TR6

D2 D4 D6

0V

It is made up of three phase-leg, bridge-leg or half-bridge “sub-units”: all three terms are
commonly used. It may be controlled in one of several different ways, depending on the
fidelity of the output waveforms required, and the complexity of the control scheme.
 DC-AC power conversion

And a single-phase inverter is normally formed from two bridge-legs as shown:

VDC

TR1 TR3
D1 D3

L1 iout

L2 iout

TR2 TR4
D2 D4

0V
 DC-AC power conversion

• A simple AC waveform may be generated by switching the bridge legs at the

fundamental frequency.
• The converter gives square wave output at the desired power frequency.
• The maximum output van corresponds to a –VDC/2 to +VDC/2 square wave.
• In the square wave switching scheme, each switch of the inverter leg is on for half cycle
of the desired output frequency.

vAn
TA+ DA+ Vdc/2
Vdc/2
TA+ ON TA- ON
iA
Vdc n vAn A t

Vdc/2 TA- DA- vAN -Vdc/2

N 1/f1
 DC-AC power conversion

• One of the advantages of square wave switching is that each device only switches twice
per cycle, which results in lower switching power loss. This can be important at very
high power levels where the power loss is significant.
• With square wave switching, the fundamental component of the AC output voltage is
determined by the DC voltage and cannot be regulated by the inverter. Thus, VDC must
be adjusted in order to control VAn.
• High harmonic distortion exists in the output voltage waveform which can cause serious
problems to the AC motor connected.
• Within each fundamental period, the device is switched (on and off) once. Thus the
switching frequency is the same as the fundamental frequency and this is the reason
why it is also called fundamental frequency switching.
 DC-AC power conversion

 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.

INVERTER OUTPUT VOLTAGE

Vdc

 

-Vdc

FUNDAMENTAL COMPONENT
V1
4VDC


3RD HARMONIC
V1
3

5RD HARMONIC
V1
5
 DC-AC power conversion

But, this has the problem of a large low-frequency harmonic content.

The Fourier series of single phase gives:

2
a n  V An sin n  d , 0      VAn  Vdc 2
 0

V   cos n  2Vdc
VAn,n  dc   V  n : odd number
  n  0 n
An,n

2Vdc V VAn,1
VAn,1   1.273 dc , VAn,n 
 2 n
vAn VAn/(Vdc/2)
1.2
Vdc/2 1.0
TA+ ON TA- ON 0.8
t 0.6
0.4
0.2
-Vdc/2
1/f1 0 1 3 5 7 9 11 13 15 n
 DC-AC power conversion

Single-phase inverter with square-wave switching

vAn
TA+ on TA- on Vd
t
TA+ DA+ TB+ DB+  180
Load vBn
io
Vdc n A B TB- on
TB+ on Vd
vAB t
TA- DA- TB- DB-
N 180
vAB 180-

Vd
• The switches in the two legs can each generate a 
square waveform. The output voltages from the t
-Vd
two legs has a controllable overlap angle α.
• During this overlap interval, the output voltage is 180-
zero as a consequence of either both top
switches or both bottom switches being on.
 DC-AC power conversion

• The fundamental and harmonic components of the output voltage can be

calculated using Fourier Transformation as
   / 2
0 v AB sinn d    / 2 Vd sin n d
2 2
v AB, n 

 n 
cosn    / 2  d cos
2Vd  /2 4V (n is an odd number)
 
n n  2 
1.2

1.0
Normalized amplitude

1st
• Normalised fundamental and 0.8

harmonic components with 0.6

variation of . 0.4
3rd
5th
0.2
7th
0.0 
0 30 60 90 120 150 180
 DC-AC power conversion

Id idc
vAN
TA+ TB+ TC+
Vdc/2 DA+ DB+ DC+
TA+ on TA- on
A Vdc
Vdc n B t
C 120 180
Vdc/2
TC- DC-
vBN
TA- DA- TB- DB-
N
TB+ on TB- on

t
• Three bridge-leg circuits can be vCN
180

modulated with a 120o phase shift. TC- on TC+ on

t
• Simple fundamental frequency 180
switching may be used vAB 120

Vdc
t
-Vdc
DC-AC power conversion

• Fourier analysis of output voltages gives

4 Vdc  1 1 
VAn  . sin ω1t  sin 3ω1t  sin 5ω1t  ...... 
π 2  3 5 

• and the line-to-line voltage

2 3  1 1 1 
VAB  Vd sin ω1t  sin 5ω1t- sin 7ω1t  sin11ω1t  ....... 
π  5 7 11 
• The fundamental output line-to-line RMS voltage

2 3 6
VAB ,1  Vdc   Vdc  0.78  Vdc
2 
 DC-AC power conversion

• Fortunately, there is the alternative of using PWM schemes for inverter control.
• For DC chopper circuits, a control signal vcontrol is compared with a repetitive switching-
frequency triangular waveform to generate the switching signals. Thus the output DC
voltage is controlled. (Recall control of the simple DC-DC converter discussed in Topic 1.)
• For a DC-AC inverter, PWM generation is more complicated as the inverter output is
preferred to be sinusoidal with controllable magnitude and frequency.
• In order to produce a sinusoidal output voltage waveform at a desirable frequency, a
sinusoidal control signal (m(t )  M sin1t , 0  M  1) is compared with a triangular
waveform having magnitude of ±1 (carrier signal).
• The frequency of the triangular waveform vtri (the carrier frequency) establishes the
inverter switching frequency and is generally kept constant.
• The control signal m(t) is used to modulate the switching duty factor and has a
frequency of f1, which is the desirable fundamental frequency of the inverter output (f1
is also called as modulating frequency and m(t) is call as modulation signal).
 DC-AC power conversion

m(t) vtri
1

-1
vAn 1/fs
Vdc/2

-Vdc/2

• It is recognised that the inverter output will not be a perfect sine wave, but will still
contain harmonics. Importantly, however, these are at a much higher frequency than
those in fundamental switching schemes.
 DC-AC power conversion

• M is called the modulation index (or modulation ratio)

• The frequency modulation ratio mf is defined as fs
mf 
f1
• In the inverter, the switches TA+ and TA- are controlled based on the comparison
of m(t) and vtri and the following output voltage results, independent of the
direction of iA:

mt   vtri , TA is on,

Vdc
v An 
2
Vdc
m(t )  vtri , TA is on, v An  
2 Vdc/2 TA+ DA+
iA
Vdc n vAn A

Vdc/2 TA- DA- vAN

N
 DC-AC power conversion

• Since the two switches are never off simultaneously (apart from short dead-
times), the output voltage vAn fluctuates between two values (Vdc/2 and –Vdc/2).
• Harmonic analysis of the inverter output voltage reveals that:
sin1t 
Vdc
‾ The fundamental component is given as v An ,1 ( t )  M
2
‾ Thus, by controlling M and ω1 of the modulation signal, the output fundamental
component is directly controlled.

‾ The harmonics in the inverter output voltage waveform appear as sidebands,

centred around the switching frequency and its multiples, that is, around fs, 2fs, 3fs,
and so on.
VAn,n/(Vdc/2)
1.0
M=0.8
0.8
0.6
fs-2f1 fs+2f1 2fs-f1 2fs+f1
0.4
2fs-3f1 2fs+3f1 3fs-2f1 3fs+2f1
0.2
0
f1 fs 2fs 3fs
frequency
 DC-AC power conversion

Bipolar voltage PWM switching

• With bipolar switching, opposite switches (TA+, TB-) and (TA-, TB+) from the two legs are
treated as two switch pairs.
• Switches in each pair are turned on and off simultaneously. One of the two switch
pairs is always on.
• The waveform has same shape as the output from the single-leg converter with
amplitude of ±Vdc, and the fundamental voltage is

v AB ,1 (t )  M  Vdc sin1t 
vAB
Vdc
TA+ DA+ TB+ DB+
Load
io
Vdc n A B
vAB
TA- DA- TB- DB- t
N
-Vdc
 DC-AC power conversion

Unipolar voltage PWM switching

• With unipolar voltage PWM switching, the switches in the two legs of the full-bridge
inverter are not switched simultaneously as in the bipolar scheme.
• Two separate control (modulation) signals are used for the two legs:
ma (t )  M sin1t  mb (t )  M sin1t 
• The two control signals are compared to the same triangular carrier waveforms to
generate the switching patterns for each leg.
• In this type of PWM scheme, when a switch occurs, the output voltage changes
between zero and +Vdc or between zero and –Vdc voltage levels. For this reason, this
type of PWM is called unipolar voltage switching.
• The fundamental component of the output voltage is the same as bipolar voltage
switching which is given by

v AB ,1 (t )  M  Vdc sin1t 
 DC-AC power conversion

Unipolar voltage PWM switching ma(t) vtri 1/fs

mb(t)

t

v AN
Vdc

t
v BN

Vdc

t
vAB=vAn-vBn
Vdc

t

-Vdc
 DC-AC power conversion

• Similar to the single-phase inverter, the objective in PWM controlled three-

phase inverter is to shape and control the three-phase output voltages in
magnitude and frequency.
• To obtain balanced three-phase output voltages in a three-phase PWM
inverter, the same triangular voltage waveform is compared with three
sinusoidal control voltages that are 120 degrees out of phase.

ma (t )
ma (t )  M  Sin (1t ) Van,1 (t )   Vdc
2
m (t )
mb (t )  M  Sin (1t  1200 ) Vbn,1 (t )  b  Vdc
2
m (t )
mc (t )  M  Sin (1t  2400 ) Vcn ,1 (t )  b  Vdc
2
0  M 1
3M
Vab,1  Van ,1 (t )  Vbn,1 (t )   sin(t  30)
2
 DC-AC power conversion
vtri ma(t) mb(t) mc(t)

Three-phase inverter - PWM

t

vAN

Vdc

t

vBN

Vdc

t
vAB

Vdc

t

Vdc
 DC-AC power conversion
Three-phase inverter - PWM

• In the three-phase inverter, normally only the harmonics in the line-to-line

voltage are of concern.
• The harmonics in any of the phase voltages are identical to the harmonics
shown earlier.
• However, the harmonic spectrum of the line-to-line voltage shown below
indicates some harmonic components, which exist in the output voltage of
one phase, have been cancelled in the line-to-line voltage.

VAB,n/Vdc
0.8 M=0.8
0.6

0.4 fs-2 fs+2 2fs-1 2fs+1

3fs-2 3fs+2
0.2

0
f1 fs 2fs 3fs
frequency
 DC-AC power conversion

To get rid form unwanted harmonics and extract the fundamental components, the
power filter is required to achieve this task.

 L-Filter - First order

L L

 LC–Filter - Second order

VI Vg VI C Vg
 LCL-Filter – Third order

L1 L2

VI C Vg
 DC-AC power conversion

An important practical issue: The inverter circuits covered so far are formed from a generic
sub-circuit called a voltage source bridge-leg or half-bridge. Essentially, the half-bridge is
formed by two power devices in series. As mentioned, both devices must never be turned
on at the same time.

Recall Topic 2 where we emphasised the importance of configuring the gate drive
connections to address the effects of parasitic inductances.

+V
Another factor that must be TR1
considered when designing
gate driver circuitry is dv/dt- D1

induced conduction. Any

power device is potentially Vout
affected by this, but we’ll
look at the IGBT here. TR2

D2

0V
 DC-AC power conversion

From their datasheets it can be seen that most

IGBTs are virtually fully off with if the gate-emitter
voltage, vGE, is simply held at 0V. +V
TR1
So, if for example, vGE of TR2 in the circuit shown
D1
here is held at 0V, then this should mean that TR1
can be safely turned on with no risk of causing a
shoot-through current. Vout

TR2
Not necessarily: a vGE = 0V may be inadequate.
To protect against shoot-through, or dV/dt- D2
induced conduction, it may be necessary to apply
a negative off state voltage. 0V

What is dV/dt-induced conduction, and why does

it occur?
 DC-AC power conversion

Consider what happens when TR1 turns on. This

action imposes a very rapid voltage change across
TR2. V(+)

This inevitably causes a current istray to flow through

TR1
the collector-gate (Miller, or reverse transfer)
capacitance CCG of the complementary device (TR2
here).
istray
vE1
If this current is sufficiently large it may temporarily
turn on TR2 by lifting its gate-emitter voltage above
its turn-on threshold value, with consequent losses
Rg Rint
and possible device failure. TR2
vg
Even if Rg is set at zero to hard-clamp vGE(TR2) to 0V vGE
through a low-impedance source, shoot-through 0V 0V
can still occur. This is because of the internal device
“gate spreading resistance”, shown here as Rint.
 DC-AC power conversion

Use of negative off-state gate

bias voltage to prevent vE1 V(+)
dv/dt-induced conduction:

By negatively pre-biasing vGE2 t

when it is off, the peak
istray
transient voltage
superimposed onto it by TR1
turning on can be submerged vGS vTH
vGS rises above threshold voltage
below the threshold voltage
value.
vGS vTH Biasing gate with a negative off-
state voltage prevents vGS from
voff rising above above threshold
voltage

Because of dv/dt-induced conduction, most commercially-available gate-driver modules

output a negative gate-emitter voltage when in the off-state. A typical IGBT driver will
apply vGE = 15V in the on-state and vGE = -8V in the off-state.
 DC-AC power conversion

Because of dv/dt-induced conduction, most commercially-available gate-driver modules

output a negative gate-emitter voltage when in the off-state.

A typical IGBT driver will apply vGE = 15V in the on-state and vGE = -8V in the off-state.

A further consideration is the need to provide dead- +V

TR1
times or underlapping to avoid unwanted conduction
through both devices in a phase-leg. D1

Ideally it should be possible to simply drive TR1 and Vout

TR2 with complementary gate-drive signals as
described in the bipolar switching scheme. TR2

D2

0V
 DC-AC power conversion

However it’s necessary to account for worst-case propagation delays through the
driver circuitry and the devices themselves.

If, for example, TR1 turns on faster than TR2 turns off then both devices will be
on simultaneously for a short period. This can be disastrous in an inverter where
a virtual short-circuit may be placed across a voltage rail

For this reason, the safety delays referred to as dead- +V

TR1
times or underlap intervals are introduced.
D1
Tdead Tdead
Drive Vout
signal to
TR1
TR2

Drive D2
signal to
TR2 0V
 DC-AC power conversion

Further reading
N. Mohan, T. Undeland and W. P. Robbins, “Power Electronics: Converters, Applications
and Design”, John Wiley and Sons Inc., 2nd Edn., 1995
B. W. Williams, “Power Electronics: Devices, Drivers, Applications and Passive
Components”, 2nd Edn. Macmillan, London, 1992
D. A. Grant and J. Gowar, “Power MOSFETs: Theory and Applications”, John Wiley and Sons
Inc., 1989

Commercial websites
Semikron, www.semikron.com, manufacturer of power modules and gate driver circuitry
used in inverters
Vincotech, www.vincotech.com, again, manufacturer of inverter hardware