ZERO VOLTAGE SWTCHING
FLYBACK ANI) FORWARD CONVERTER TOPOLOGIES
Youhao Xi
A Thesis
in
The Department
of
Electrical and Computer Engineering
Presented in Partial Fulfillment of the Requirements for the Degree of Master of Applied Science at
Concordia University Montréal, Québec, Canada
June 1997
O Youhao Xi, 1997
National Library 1*1 of Canada Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliographie Services services bibliographiques
395 Wellington Street 395. nie Wellington Ottawa ON K I A O N 4 Ottawa ON K1A ON4 Canada Canada
The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/film, de
reproduction sur papier ou sur format électronique.
The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts £iom it Ni la thèse ni des extraits subsbntieIs may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.
Zero voltage switching flyback and forward converter topologies
Youhao Xi
Advanced telecornmunication and cornputer systems demand high efficiency and
high power density, dihbuted power supplies. Both the flyback and fonvard converter
topologies are good candidates for these applications, owing to their simple circuitry.
These topologies, however, require practical performance improvements.
Improved zero voltage switching (ZVS) flyback and forward dddc converter
topologies are developed in this thesis. They employ an auxiliary circuit, with oniy a few
small components and switching devices, to achieve ZVS of the main switch. Steady state
analyses are performed and show that the proposed topologies have the following rnents:
lossless switching of the main switch independent of the line and load conditions. no
increase in conduction losses, simple power and control circuitry, and capability of
operating in either voltage or current mode control.
The srnall signal analysis and closed loop stability of the proposed topologies are
also derived. The resulting compensation required for stable operation is obtained and
shown to be straightfonvard.
The steady state and small signal analyses are verified on a 50W prototype
flyback converter switching at 200 kHz, and on a 100 W prototype fonvard converter
switching at 300 kHz, respectively. Experirnental results show that both the flyback and
forward converters have about 5 to 7% higher efficiencies than the conventional hard
switching converters for typicd efficiencies of 70% and 85%, respectively. These
topologies can therefore advantageously replace existing power supplies in advanced
telecommunications and cornputer applications.
1 would like to express my sincere gratitude to my supervisors, Dr. P. K. Jain and
Dr. G. Jobs, for their guidance, encouragement, fnendship and support, during the course
of this study.
Special th& to Norte1 Power Group (Ottawa) for their generous oEer of
components, devices and data sheds, for work related to this thesis. The financial support
provided through a BNR/NSEC Collaborative R&D Grant is highiy appreciated.
I am grateful for the International Tuition Remission through the agreement
between the Chinese Embassy and Québec Government.
I would like ro thank my colleagues at the Power Electronics Laboratory very
much for their kindness and help in every aspect, and for the nurnerous valuable
discussions we have shared.
Finally, 1 would like to express rny thanks to my wife, my siblings, and my
dearest parents, for their everlasting love, spiritual encouragement and strong support,
during these years.
................................................................................. List of Figures ... .............................. x
... .................................................................................................................. List of Tables x i i r
............................................................................................................ List of Acronyms xiv
List of Primary Symbols .......... .. ............................. .. ................................................ xv
CHAPTER 1 Introduction
1.1 Power Supplies for Telecommunication and Computer
1.2 Flyback and Fonvard Topologies ................................................................................
.................................................... 1.3 Problems with Standard Topolo~ies and Solutions 4
1.4 Review of Existing ZVS Flyback/Fonvard Topologies ............................................. 5
S ystems ............................... 1
9
Active Clamp Topology .................................................................................... 5
..................................................................... . Mirrored Structure' Topology -6
Passive Non-dissipative Snubber FlybacWFonvard Topologies ...................... 7
...................................................................... 1.5 Scope and Contributions of this Thesis 8
............................................................................................................. 1.6 Thesis Outline 9
CHAPTER 2 A ZVS Flyback Converter Topology
.............................................................................................................. 2.1 Introduction 1 1
. . .................................................................................................... 2.2 Circuit Descnption 12
............................................................................................... 2.3 Modes of Operation 1 3
................................................................................................ 2.4 Steady State Analysis 17
............................................................................................................. 2.5 Performance 26
............................................................. 2.5.1 Effects of the Non-Ideal Components 26
2.5.2 Determination of Duty Cycle for the Output Regdation ................................ 28
............................ 2.5.3 Advantages and Disadvantages of the Proposed Converter 29
......................................... 2.5 -4 RMS, Average, Peak Current and Voltage Values 30
.......................................................... 2.5.5 Losses Caused by the Auxiliary Circuit 32
......................................................... 2.5.6 Losses Removed by ihe ZVS Operation 33
............................................................................................... 2.6 Experimental Results 33
............................................................... 2.6.1 Experiment Setup ..................... ... -33
................................................................ 2.6.2 Key Waveforms ........................ ... 34
....................................................................................................... 2.6.3 Efficiency -38
.............................................................................................................. 2.7 Conclusions 39
CHAPTER 3 A ZVS Forward Converter Topology
.......................................................................... Introduction .... ................................. 41
. . . . Circuit Descnption .................................................................................................... 42
Modes of Operation ................................................................................................. 43
............................................................................................. Steady State Analyses -48
............................................................................................................. Performance 60
....... ..............................*.... Effects of the Non-Edeal Components/Devices ..... 60
................... Determination of Duty Cycle as to Regulate the Output Voltage 61
Advantages and Disadvantages of the Proposed Converter ............................ 62
................................................................... RMS or Average Current Values -63
vii
........................ .............................. 3 S.5 Losses Caused by the Auxiliary Circuit .... 65
......................................................... 3 S.6 Losses Removed due to ZVS Operation 65
............................................................................................. 3 -6 Experimental Results -66
............................................................................................ 3 .6.1 Experiment Setup 66
............................................................................................... 3 .6.2 Key Waveforms 67
.................................................................................................... 3 .6.3 Efficiency -73
.............................................................................................................. 3 -7 Conclusions 75
CHAPTER 4 Small Signal Models and Closed Loop Stability
......................................................................................... .......... Introduction .... -77
................................................................ Criteria of a Dynamicdly Stable System 78
.............................................. Small Signal Models of the ZVS Flyback Converter 79
............... 4.3.1 Small Signal Mode1 of the Current Loop of the Flyback Converter 81
.............. 4.3.2 Small Signal Mode1 of the Voltage Loop of the Flyback Converter 84
.............................................. Small Signal Models of the ZVS Fonvard Converter 87
...... 4.4.1 Small Signal Model of the Current Loop of the ZVS Forward Converter 88
................................................. 4.4.2 Voltage Loop of the ZVS Forward Converter 93
............................................................................................... Experimental Results -94
..................................................................................... ............. Conclusions .... -98
CRAPTER 5 Design Procedures
............................................................................................................... 5.1 Introduction 99
........................................................... 5.2 Design Procedure of the Flyback Converter 100
......................................................................... 5.2.1 Design of the Power Circuit 1 0 0
.................................................................. 5.2.2 Design of the Auxiliary Circuit 1 03
......................................... 5.2.3 A Design Example ........................................... 107
5.3 Design Procedure of the Forward Converter .......................................................... 108
............................................ 5 .3.1 Design of the Power Circuit ......................... .. 108
5.3.2 Design of the Awciliary Circuit .............................. .. ............................. 111
......................................................................................... 5.3.3 A Design Example 113
................... ........................ 5.4 Gating Generation for the Awliary Switch .... 115
5.4.1 Gating Pattern Generation by Additional Logic and Drive Circuit ............... 115
.............................................................................. 5 .4.2 The Design of the Splitter 117
........................................................................................ 5.4.3 A Design Example 1 8
CHAPTER 6 Conclusions
6.1 Summary ................................................................................................................. 120
............................................................................... 6.2 Conclusions and Contributions 121
................................................................................... 6.3 Suggestion for Future Work 1 2 2
............................................................................................................. REFERENCE -124
APPENDIX
1 The Control Chip UC3855 ..................................................................................... A-l
................................................................... II Schematic of the prototype converters A-2
Fig . 1.1
Fig . 1.2
Fig . 1.3
Fig . 1.4
Fig . 1.5
Fig . 2.1
Fig . 2.2
Fig . 2.3
Fig . 2.4
Fig . 2.5
Fig . 2.6
Fig . 2.7
Fig . 2.8
Fig . 2.9
Fig . 2.10
Fig . 2.1 1
Fig . 2.12
Fig . 3-1
............................................................................. A flyback converter topology 2
A fonvard converter topology ................................ .. ...................................... -3
An active clamp forward converter topology ..................................................... 6
The 'mirror-structured' ZVS flyback topology ................................................. -7
A fonvard converter with a non-dissipative snubber .......................................... 8
.............................................. The proposed ZVS flyback converter topology 12
. Key waveforms of the convener of Fig 2.1 ..................................................... 14
. ............................................... Modes of operation of the converter of Fig 2.1 15
The equivalent circuit of the discharging resonant loop in Interval 1 .............. 15
Theoretical results: the drain voltage and current waveforms of QI ................ 34
...... The experimental results: the drain voltage and curent waveforms of Q I 35
................ Theoretical results: the drain voltage and current waveforms of Q2 35
The expenmental results of the drain voltage and current waveforms of 0 2 - 3 6
The experimental results : the current and voltage waveforms of the main
3 switch in ZVS operation under light load conditions ...................................... 37
The current and voltage waveforms of the auxiliary switch under light load
............................................................................ conditions .................... l... 37
The overall effïciency of the converter vs . input AC voltage .................................. 38
The overall efficiency vs . load under given input voltage ................................ 38
............................................... The proposed ZVS forward converter topology 42
. *.................... Fig . 3.2 Key waveforms of the converter of Fig 3.2 .. .............. ... . ............................................... Fig . 3.3 Modes of operation of the converter of Fig 3.1 45
Fig . 3.4 The equivaient circuit of the discharging process in Interval 1 ........................ 46
Fig . 3.5 The equivalent circuit of the discharging process in Interval 5 ........................ 46
Fig . 3.6 The drain voltage and current waveforms of Ql in Fig . 3.1 (theoretical) ........ 67
Fig . 3.7 The experimental results of the drain voltage and current waveforms of QL.68
Fig . 3.8 The drairi voltage and current waveforms of Q2 in Fig . 3.1 (theoretical) ........ 69
Fig . 3.9 The experimental results of the drain voltage and current waveforms of Q1 ... 69
....... Fig . 3.10 Experimental results: the current and voltage of QI at high line full load 70
..... Fig . 3.1 1 Experimental results: the current and voltage of QI at high line Iight load 70
........ Fig . 3.12 Experimental results: the current and voltage of QI at low line full load 71
...... Fig . 3.13 Experimental results: the current and voltage of QI at low line light load 71
Fig . 3.14 Experimental results: the current and voltage of Q2 at medium line half load ....
......................................................................................................................... -72
..... Fig . 3.15 Experimental results: the current and voltage of Q2 at high line light load 72
....... Fig . 3.16 Experimental results: the current and voltage of Q2 at low line half load 73
Fig . 3.17 The efficiency vs . input voltage of the prototype converter ............................. 74
........................................... Fig . 3.18 The efficiency vs . load of the prototype converter 74
........................................................... Fig . 4.1 The Block diagram of the control loops 80
. ............................ Fig . 4.2 A typical implementation of the feedback loops of Fig 4.1 80
..... Fig . 4.3 The Bode plot of the open current loop of the prototype flyback converter 82
Fig . 4.4 The Bode plot of the open voltage loop of the prototype flyback converter ..... 86
Fig . 4.5 A typical Block diagram of the control loops of the fonvard converter ............ 87
Fig . 4.6
Fig . 4.7
Fig . 4.8
Fig . 4.9
Fig . 4.10
Fig . 4.1 1
Fig . 4.12
A typical implementation of the feedback loops o f Fig . 4.5 .......................... -87
The reconstruction of the sensed current in the proposed ZVS forward
............................................................................................................ converter 88
The Bode plot of the open current loop of the prototype forward converter ..... 91
The Bode plot of the open voltage loop of the prototype fonvard converter .... 93
Ccrnparison between the experimental and theoretical gain of the open voltage loop
power circuit ................................................................................................. ......95
Cornparison between the phase shift of the open voltage loop power circuit transfer
................................................................................... ............... function ...... -96
Experimental results of the fornard converter: transient response (Io step up) ....
97
Fig . 4.13 Experimental results of the fonvard converter: transient response (1, step
Fig . 4.14
Fig . 5.1
Fig . 5.2
Fig . 5.3
Fig . 5.4
................................................................................................................ doum) -97
Experimental results of the fonvard converter: transient response (K,, step
................................................................................................................ d o m ) -98
The secondary current and the output voltage ripples in the flyback converter ...
............................................................................................................. 101
A gating generation circuit, a splitter ............................................................. 1 16
. .................................................... Key waveforms of the splitter o f Fig 5.2 1 16
.............. Experimental results of the gating pattern generated by the splitter 1 18
xii
Table 1.1
Table 2.1
Table 2.2
Table 3.1
Table 3.2
Table 4.1
Table 5.1
Table 5.2
Table 5.3
Comparison between the two topologies based on the same powe :r leveI and
........................................ ................................... ... operating fiequenc y .. ... -3
Comparison between the theoretical and experimental results of the current
............. ...............*........*.......*...*................... and voltage of the main switch .. 35
Comparison between the theoretical and experimental results of the current
- . and voltage of the auxiliary switch.. .............................................................. 3 6
Cornparison between the theoretical and experimental results of the current
and voltage of the main switch .............. .. ..................................................... 67
Cornparison between the theoretical and experimental results of the current
................................................................. and voltage of the auxiliary switch 68
The theoretical and experimental resuIts of the power circuit transfer function
......................... .. .......................................................................................... 94
...................................... The parameters of die prototype flyback converter 107
................................... The parameters of the prototype forward converter 1 14
The parameters of the splitter in a design example ............................ . 1 1 9
AC
CA
DC
EMI
IC
MOSFET
PCB
PI
P W M
rrns
VA
ZCS
zvs
Alternative Current
enor Current Amplifier
Direct Current
Electromagnetic Interference
Integrated Circuit
Metal-Oxide-Semiconductor Field Effect Transistor
Printed Circuit Board
Proportional-Integral network
PuIse Width Modulation
Root Mean Square value
error Voltage AmpIifier
Zero Current Switching
Zero Voltage Switching
xiv
O, angdar frequency given by I/JK
a,, angdar fiequency given by I/ JG for the flyback topoiogy, and by
2 / (L C ) for the forward topology ,,/y or angdar frequency given by
C, input capacitor of a dddc converter
Co output capacitor
$Il* snubber capacitor
D, primary side blocking diode of the auxiliary circuit
D, secondary side blocking diode of the auxiliary circuit
Do, rectifier diode in the output stage
DO? freewheeling diode in the output stage
D nominal duty cycle of the main switch
Dm duty cycle of the auxiliary switch
Dm_ maximum duty cycle of the main switch
Dm, minimum duty cycle of the main switch
De effective duty cycle of the main switch in the forward topology
D, reduction in duty cycle of the main switch by L in the forward topology
d small signai deviation of D in Laplace transformation
2, small signal deviation of LIr in Laplace transformation
f , cross over frequency of an open Ioop system
4 a pole of a transfer function
f - a zero of a tramfer function
f , switching frequency
G,-+-(s) transfer fünction of the error current amplifier compensation network
Gi,,(s) closed loop transfer function of the current loop
GJs) transfer function of the voltage loop power circuit
Gu@) transfer function of the error voltage amplifier compensation network
Hl@) transfer function of current loop po wer circuit
H,(s) transfer function of the output stage of voltage loop
I~P primary peak current of the awciliary circuit
Iar secondary peak current of the auxiliary circuit
IO nominal output curent
lpeak peak switch current of the flyback topology
lm instantaneous magnetizing current in the fonvard topology
jS instantaneous secondary current of the power transformer
*
i, small signal deviation of the output inductor current in Laplace transformation
A
i, small signal deviation of the output current in Laplace transformation
K turns ratio of the current sensor transformer
k deviation factor of Vo from y,,
Lw primary of the coupled inductors of the auxiliary circuit
Lm secondary of the coupled inductors of the auxiliary circuit
xvi
magnetizing inductor of Tr
output inductor
additional secondary inductor in the forward topology
effective inductance seeing into primary winding of T in the forward topology
reflected inductance of Ls hto the primary side of Tr
number of tunis of the core resetting winding of Tr
nurnber of tums of the prirnary winding of Tr
number of t ums of the secondary winding of T'
reciprocal of the tums ratio of the power transformer, equal to NIN,
main switch power MOSFET
auxillary switch power MOSFET
clamp switch power MOSFET in the active clamp t~palogy
main switch power MOSFET in the active clamp topology
load resistor
on resistance of a power MOSFET
current sensor resistor
power switch
Laplace transformation variable
power transformer
switching cycle of the converter
instantaneous drain-to-source voltage of Q l
instantaneous secondary voltage of Tr
xvii
drain-to-source voltage of Q I at the beginning of each switching cycle
input voltage of the DClDC converter
nominal output voltage
reference voltage
peak value of the secondary voltage of T,
peak-to-peak voltage of the saw-tooth signai in the PWM control unit
small signal deviation of the output voltage of CA in Laplace transformation
small signal deviation of F(> in Laplace transformation
small signal deviation of the voltage resulted fiom the sensed current in Laplace
transformation
small signal deviation of the output voltage of VA in Laplace transformation
xviii
1.1 POWER SUPPLIES FOR TELECOMMUNICATION AND
COlMPUTER SYSTEMS
In recent years, telecornmunication and cornputer systems have experienced fast
growth. These advanced systems require distributed power supplies. The principal
features of the supplies are constant operating frequency, high efficiency and high power
density. The power levels are usually ranged fi-om hundreds of watts down to thirty watts
or lower.
To meet the requirements of these applications, a nurnber of pulse width
modulation (PWM) MOSFET topologies are cornmonly used to implement the
distributed power supplies. These topologies include the full bridge, half bridge, push-
pull, double switch forward, single switch fonvard and flyback topologies, which are
listed in the sequence fiom high power level down to low power level. Each topology has
unique properties which makes it best suited for a certain power level-
This thesis limits the discussion within applications with power level below 150
watts, where the single switch MOSFET forward and flyback topologies are best suited.
1.2 FLYBACK AND FORWARD TOPOLOGIES
Figs. 1.1 and 1.2 show the standard topologies of the single switch MOSFET
flyback and fornard converters, which are best suited for applications at power level
below 150 W and dc line voltages below 200 V. Their operating principles are briefed
below. Each topology has advantages and drawbacks, which are compared in Table 1.1.
The transformer of the flyback is operated in the indirect power transfer mode.
"men the switch is nimed on, a nsing current flows through the primary winding (an
inductor), thus an amount of energy is stored in the core of the transformer. When the
switch is turned off, the dotted end of the secondary winding (another inductor) reverses
the voltage polarity. Thus, the rectifier diode is forward biased and the stored energy is
transferred to the output via the secondary winding and the rectifier. Periodical switching
of the switch will keep the power continuously flowing h m the input to the output. By
modulating the pulse width of each switching signal, the output voltage is regulated.
R e c t i f i e r -..
T r a n s f o r m e r
8 Fig. 1.1 A flyback converter topology.
The transformer of the forward is operated in the direct power transfer mode.
When the switch is turned on, an amount of energy is transferred to the output via the
transformer and the rectifier. When the switch is -ed off, the core of the transformer is
reset by a tertiary winding. Penodical switching of the switch will keep the power
flowing from the input to the output. By moduiating the pulse width of each switchùig
signal, the output voltage is regdateci.
o a d
Table 1.1 Cornparison between the two topologies based on the same power level
Advantages
Disadvantages
Applications
The flyback topology The forward topology
1, Simpler circuitry: no output inductor, only one rectification diode in the output stage,
2. Simpler transformer structure, 3. Better regulating of slave outputs
against t h e load variations in multi-outputs appIications.
1. Higher RMS and peak currents, 2. Larger ripples in the output
voltage, 3. Bigger output capacitor needed.
and operating frequencv
-
1. Low output current, 2. Relatively lower power level, 3. Multiple outputs.
1. Less ms and peak currents,
2. Less nppk in the output voltage,
3. Smaller output capacitor needed.
1. Output inductor and two more diodes are needed,
2. One more set of winding of the power transformer is needed to reset the core,
3. Poor regulation of slave outputs against the Ioad variations.
1. High output current, 2. Relatively higher power level, 3. Better for single output.
1.3 PROBLEMS WITH THE STANDARD TOPOLOGIES AND
SOLUTIONS
Unfortunately, the standard flyback and forward topologies are operated with hard
switching, that is, there are significant overlaps between the switch current and voltage
during the tum-on and tuni-off transients. These overlaps causes the so cdled switching
Iosses. Thus, it is difficult to achieve high power density and high efficiency with the
standard topologies.
In order to achieve high power densities, the power supplies are operated at
increasingly higher fiequencies. However, when the switching fiequency increases, the
losses associated with the tum-on and nim-off of the power switching devices in the hard
switching MOSFET converters also increase. These losses are so significant that
operation of the converters above 50 H z is prohibitive, because of the low conversion
efficiency and high cooling requirements.
Over the years, the resistor-capacitor-diode snubber (RCD) has been employed to
reduce the turn-off voltage stress and switching losses in the switches of converters.
However, the power removed fiom the switching losses due to RCD dissipates
completely in RCD itself. Thus, RCD has no contributions to achieve high efficiency nor
to reduce the size of the cooling devices. Hence, RCD is not an solution to the
applications concemed herein.
Normally, the solution is to employ the soft switching techniques. There are two
types of soft switching techniques, namely zero voltage switchùig (ZVS) and zero current
switching (ZCS). Both can eliminate or greatly reduce the switching losses. But it has
been shown in [l] that ZVS is a better scheme than ZCS in the MOSFET converter
topologies. Owing to this reason, a lot of work has been done to develop ZVS
flybacWforward topologies.
1.4 REVIEW OF EXISTING ZVS FLYBACK/F'ORWARD
TOPOLOGIES
Several ZVS flyback /fornard converter topologies have been proposed, including
the active clamp converters [2-131, the "mirror-stnictured" converters [14], and the non-
dissipative snubber [15-171. Although ZVS is achieved in these converter topologies9
they have at least one of the following drawbacks: complicated control schemes,
increased conduction losses, loss of ZVS under the light load conditions, under utilization
of power transformer, and restrictions on use due to patents.
1.4.1 Active Clamp Topology
Fig. 1.3 shows an active clamp forward converter topology, where a clamp circuit
consisting of a switch and a capacitor is added to the standard forward circuit. With the
clamp circuit, ZVS is achieved in the main switch. The clamp circuit of Fig. 1.3 can also
be employed in the flyback topology, forming an active clamp flyback converter.
The drawbacks of these topologies include:
C apacitor M ain sw itch
Fig. 1.3 An active clamp forward converter topology.
(i) They need an isolated, variable d u t - cycle gate drive for the clamp switch,
(ii) They require a modified PWM control technique to properly program the associated
delays between gate drives of the main and clamp switches, so as to achieve ZVS,
(iii) ZVS is lost under light load conditions.
(iv) ZVS is achieved at the cost of increased conduction losses,
(v) The current mode control c m not be used,
(vi) The patent related legai issues make it difficult to use, because it is a patented
technique.
1.4.2 'Mirrored structure' Topology
Fig. 1.4 shows a ZVS flyback converter topology which has a mirrored structure
on both sides of the power transformer [14], where a MOSFET switch replaces the
rectifier diode of the standard flyback circuit. Other topologies like the fornard can be
converted into a ZVS operation in the same manner.
o a d Y'"
Fig. 1.4 The 'mirrored-structure' ZVS flyback converter.
I n p u t i O u t p u t - c a p a c i t o r
L &
The mirrored structure circuit employs fewer components than the active clamp
converter. As the output voltage is usually low (<24 V) for the telecornmunication and
cornputer systems applications, a low voltage rating MOSFET can be selected for the
ZVS switch. The drawbacks of this topology include:
(i) It needs an isolated gate drive for the ZVS switch,
(ii) The utilization of the power transformer is poorer,
(iii) The conduction losses are increased:
(iv) Larger ripples are resulted on the output capacitor.
1.4.3 Passive Non-dissipative Snubber Flyback/Forward Topologies
Fig. 1.5 shows a fonvard converter topology employing a non-dissipative snubber
[15-171. Two diodes and a capacitor and an inductor foms the snubber. No additional
switch is required. This snubber can also be applied to a flyback converter topology.
The drawbacks of the non-dissipative snubber include: (i) The switch has a hard
switching at tum-on, and (ii) The topology is not suitable for applications when the input
voltage varies in a wide range.
Fig. 1.5 A forward converter with a non-dissipative snubber.
1.5 SCOPE AND CONTRIBUTIONS OF THIS THESIS
The objective of this thesis is to propose and analyze ZVS flyba ck and f o ~
topologies that can overcome the following drawbacks of the existing topologies:
(i) complicated gate drive schernes.
(ii) loss of ZVS under light load,
(iii) no ability to employ current mode control,
(iv) increase in conduction losses,
and to compare performance of the proposed topologies with that of the standard
(conventional hard switching) fl yback and fonvard converters.
The scope of this thesis is lirnited to the steady state analyses, smdl signal
analyses, experimental verifications. and design of the proposed topologies for
applications at the power level below 150 W, switching fiequency between 200 kHz and
300 kHz, dc line voltage below 200V. Average current mode control is selected as the
control scheme.
The principle contributions of this thesis are as follows:
(i) ZVS flyback and forward converter topologies are proposed and analyzed. It is shown
that the proposed topologies have several advantages over the existing ZVS and
conventional flyback and forward topologies, including no increase in conduction losses,
simplicity in both power and control circuitry, lossless switching independent of lineAoad
conditions.
(ii) Prototype flyback and forward converters of the proposed topologies are built, which
are operated at 200 kHz and powering 50 W, and at 300 kHz and powering lOOW,
respectively. Expenmentd verifications of the theoretical analyses are performed with
these prototypes. Cornparisons of efficiencies between the proposed converters and the
conventional hard switching converters are made, which show that the proposed
converters have about 5% to 7% higher efficiency.
(iii) The design procedures for the proposed converter topologies are presented.
The contents of each chapter in this thesis are sumrnarized below.
In Chapter 2, a ZVS flyback converter topology is presented. The principles of
operation are illustrated and modes of operation are detailed. The steady state operation in
each mode is analyzed. Parameters evaluating the steady state performance of the
converter are denved. Experimental results of the prototype fiyback converter are
presented and are compared with the theoretical predictions.
In Chapter 3, a ZVS forward converter topology is presented. The principle of
operation is illustrated and modes of operation are detailed. The steady state operation in
each mode is analyzed. Some parameters evaluating the steady state performance of the
converter are derived. Experimental results of the prototype forward converter are
presented and are compared with the theoretical predictions.
In Chapter 4, small signal models are discussed, based on the average current
mode control. The flyback converter can utilize the same srnall signal model as a
conventional flyback converter. The small signal model of the fonvard converter is
derived. The compensation for both the current and voltage loops is discussed and the
small signal analysis of the fonvard converter is verified experimentally.
In Chapter 5, design procedures are presented. Design examples are given for both
the flyback and forward topologies. A simple implementation of generating of the gating
drive signals for the auxiliary switch is presented.
In Chapter 6, research and development in the thesis are summarized. Advantages
and disadvantages of the proposed topologies are discussed. Suggestions for future work
are made.
A ZVS FLYBACK CONVERTER TOPOLOGY
2.1 INTRODUCTION
Among the conventional dcldc converters. the flyback topology employs the
fewest power components and devices. It is best suited for multiple output applications at
total power levels below 150 W and dc line voltages above 100 V. However, as
mentioned previously, the standard flyback is unable to meet the requirements for high
power density and high efficiency. and the existing ZVS flyback converters have several
previously rnentioned drawbacks. Research for better solutions is hence required.
This chapter presents a ZVS flyback converter topology which is able to
overcome some of the drawbacks of the existing topologies. The topology employs an
auxiliary circuit that helps to achieve ZVS in the main switch, with addition of only a few
nurnber of small power rating components and devices.
The converter operates in the discontinuous mode. The continuous mode
operation is not recommended here since it is not easy to stabilize the converter. A steady
state analysis of the converter is presented in this chapter to understand the behavior and
to develop performance characteristics.
The circuit description is made in Section 2.2. The modes of operation are
detailed in Section 2.3. The steady state analysis is pe r fmed in Section 2.4.
Performance is discussed in Section 2.5 and experimental venfications are made in
Section 2.6.
2.2 CIRCUIT DESCRTPTION
Fig. 2.1 shows a ZVS flyback converter topology employing an auxiliary circuit
drawn inside the dashed line block. Outside the bIock is a standard flyback topology
shown in Fig. 1.1.
In Fig. 2.1, Citr is the input capacitor, T is the power transformer with windings Np
(pnmary) and N,. (secondary), L,,, is the rnagnetizing inductance of T , Do is the output
rectifier diode, Co is the output capacitor, R,, is the load and QI is the main switch.
The auxiliary circuit consists of (i) a snubber capacitor CStrb, whkh is connected in
parallel with the main switch QI, (ii) two coupled inductors L, (primary) and L,
(secondary), (iii) two blocking diodes D, and D2, and (iv) an auxiliary switch Q2.
2.3 MODES OF OPERATION
Fig. 2.2 shows key wavefoxms of the proposed converter. For each switching
cycle, T,, the converter operates in the following five intervals. In the operation of a
standard flyback, there are no Intervals 1 and 3. These two intervals are fulfilled by the
auxiliary circuit and are introduced here to achieve ZVS in the main switch, QI, during
both its turn-on and turn-off transients.
2.3.1 Interval 1 ( t , < t tJ
Fig. 2.3(a) shows the circuit operation during this interval. At the beginning of
this interval, 0 2 is turned ON. It has a zero current tum-on, because L, is in series with
it. Its drain-to-source voltage falls rapidly to zero. and a resonant loop consisting of C,,
La,, and L,, as s h o w in Fig. 2.4, is formed. C,,, starts to discharge through Lap. The
discharging current builds up a magnetic field in the core of the coupled inductors,
transfemng the stored energy in C,,,l, in the previous cycle to the core of L,,.
intervals
Main switch
Auxiliary switch
O
Main s w itch voltage and
curren
A uxiliar). sw itch voltage and
currefl
O
Transforrn e secodary curren
O
Coupled inducto sencondary O
c u m n
- . I
+ I I 0 I I I
, . 8 , 1 I I I
: t 0 , I I
4 , I I * , a I , I
4 ta r; ta l; ~ ' + T s t im e
Fig. 2.2 Key waveforms of the converter of Fig. 2.1. The converter operates in five modes per switching cycle. Variables are defined in Fig. 2.1
At the end of this interval, t = t,, - C,,,, is depleted. As a consequence, the drain of
Q l is pulled d o m to zero voltage: providing the ZVS condition for Ql at its tum-on.
During this interval, Co supplies the output current.
2.3.2 Interval2 (t, < t 5 4)
Fig. 2.3@) shows the circuit operation during this interval. At the beginning of
this interval, Ql is tumed ON under the zero voltage condition. Hence it has no tum-on
losses. At the same moment, Q2 is nimed OFF, stopping the current in Lw rapidly. The
* (a) Interval 1 (t,<lst2)
* (b) Interval 2 ( t p s t , )
t4 (d) Interval 4 (t,<tQ,)
O
(e) Interval 5 (t,<t<T,+i,) Fig. 2.3 Modes of operation of the converter of Fig. 2.1. The operation is divided
into five modes per switching cycle. The key waveforms are shown in Fig. 2.2.
Fig. 2.4 The equivalent circuit of the discharging resonant loop in Intervai 1. Variables are defined in Fig. 2.1.
rapidly falling current in L, reverses the voltage polarity on the dotted ends of the
coupled inductors. Thus D, becomes forward biased and is forced to conduct.
When D, is conducting, Lm sees a constant voltage y,,. Through coupling, L, sees
a reflected voltage. This voltage is the stress voltage on Q2. By increasing the ratio of L,
to L,, the stress will be lower. This in h m reduces the turnoff switching losses in Q2. On
the other hand, the current in Lar is decreasing linearly. In this way the stored energy in
the core of the coupled inductors is gradually fed back into the input line. At r = t,, the
process completes and D, becomes reverse biased again.
When Ql is ON, the magnetizing inductor of T sees a constant voltage y,,. Thus,
the current in Ql nses linearly. In this way the energy is stored in the core of T in the
sarne fashion as in a standard flyback converter. During this interval, Co supplies the
output current.
2.3.3 Interval 3 (t, < t r t,)
Fig. 2.3(c) shows the circuit operation during this interval. At the beginning of
this interval, Ql is turned OFF. C,, is in the process of charging and it slows down the
rise of the drain voltage of Q l . A sufficient value of capacitance of C,, will guarantee a
ZVS tunioff. The current in the primary side is decreasing in a resonant mode, with an
angular fiequency determined by L,,, and Cm*. The decreasing of the primary current
reverses the polarity on the dotted ends of T' But before the voltage of the secondary side
of T, reaches the output voltage y,' D,, maintains reverse biased. Co supplies the output
current during this interval.
2.3.4 Interval4 (t, < t s ta
Fig. 2.3(d) shows the circuit operation during this interval. At the beginnuig of
this interval, the voltage of the secondary side reaches V'- Then Do is forward biased and
begins to conduct. The secondary current refills Co, and also supplies the output current.
In this way the stored energy in T is now transferred to the output.
The conduction of D, causes N,. to see a constant voltage Vo. This in turn clamps
the voltages on both primary and secondary windings, and consequently, it clamps the
voltage across the main switch QI.
2.3.5 Interval 5 (t, < t s t,+TJ
Fig. 2.3(e) shows the circuit operation during this interval. At the beguuiing of
this interval, the stored energy in T is completely transferred to the load. Then Do is
reversed biased again and C, supplies al1 the output current. The dif5erence between the
drain voltage of Q I and the dc line voltage y.,) causes a resonance in the network
consisting of L,,, and C,,.
At the end of Interval 5, a new cycle begins and Intervals 1 through 5 repeat.
2.4 STEADY STATE ANALYSIS
The operating principle has been described in last section. In this section the
steady state analysis is performed with the assumptions made below. In the andysis, the
time varying variables such as the cwent and voltage of the principal components and
devices are detemiined. Based on these variables, the performance of the converter is
illustrated, and the resultant quantities such as the rms, average or peak c w e n t and
voltage of the principal components and devices are obtained in Section 2.5. These
quantities are used in designing the converter as presented in Chapter 5.
In the analysis presented below, a closed form solution is obtained by solving a
set of differential equations in each interval and by matching the boundary conditions at
the boundary of each interval. The initial conditions are a fûnction of the operating
kequencyf;, the input dc line voltage y,, and the output power P,,. These initial conditions
can be obtained by the iterative process such as the Newton-Raphson method.
2.4.1 Assurnptions, Definitions and Initial Conditions
For convenience, following assumptions are made:
(i) The steady state conditions have be established. and the converter is running at the
nominal output voltage Vo and the static load condition: Po,
(ii) Each component and device has ideal properties, that is
(1) T,: the leakage inductance is ipnored, and the core does not saturate,
(2) L,, L,: the coupling factor is 1 .O, and the core does not saturate,
(3) Co, Cd: pure capacitors, and the capacitance of Co is infmitive,
(4) Dl, D2, Do: the fonvard voltage drop is O V, the recovery time is O s,
(5) Q I , 02 : the on resistances are O Cl, the inherent capacitances are O F.
The effects of the non-ided properties of the components and devices are investigated in
Section 2.5.1.
(iii) The magnetizing inductance of the transformer, Ln,, is much larger than L,.
Following parameters are defmed:
a,,--angular fiequency of the resonant tank of C,), and L,,
a,--angular frequency of the resonant tank of C,, and L, ,
--instantaneous current in Q2, I ~ P
Iap--peak current in Q2,
i,-4nstantaneous current in La-$,
1,--peak current in L,,
I,,l,,---peak magnetizing current in T,
i,,--instantaneous current in Q I ,
i.5--instantaneous secondary current,
1,--peak secondary current.
n--reciprocal of t u m s ratio of Troor NJV,,.
uQ,-4nstantaneous drain voltage of Ql? or voltage across C,,,,
um-4nstantaneous drain voltage of QI,
V,--initial drain voltage of Ql per switching cycle,
Vp-Q,--clamped voltage stress of Ql when it is OFF.
The analysis begins by stating the initial conditions of Interval 1 :
(i) Ql and Q2 are both OFF, and the drain voltage of the paralleled switches is V,,
u&,) = uO2(tJ = VO (2- 1
('1 ) = ) = 0 (2-2)
(ii) The magnetizing inductor Ln, and Cd undergoes a resonance, but the resonant cunent
in L, is negligible,
',,,(tJ = 0 (2-3)
(iii) Both D, and Do are reverse biased, and Co supplies the total output curent,
i&) = i&) = O (2-4)
2.4.2 Intewai 1 (t, -= t 5 t,) This interval starts at t = t,. QZ is nimed on and C.,, and LtIl foms a resonant tank
as s h o w in Fig. 2.4. Hence the following equations govem the resonant process in this
tank.
Combining Eqs. (2-5) through (2-7),
According to the assumption (iii), L,<<Ln,. Hence, Eq. (2-8) can be approximated as,
By the definition made above,
Giving the initial conditions expressed in Eqs. (2-1) through (2-4), Eqs. (2-9) and (2-6)
yield, respectively
u,, ( 0 = K, co$q,(t - t , ) ] (2-1 1)
iap(t) = ~ J w ' s i # ~ , , ( t - r l ) ] (2- 12)
Because D, and Do are reverse biased and Ql is OFF and Q2 is ON, the following
equations are fourid to govem the respective variables in this interval.
Lm = i.&) = iQ, (0 = (0 = 0 (2- 1 3)
= 0 (2- 1 4)
At the end of this interval, the resonance completes a quarter of its period, Le., o,(t, - 1,)
equals r/2. Thus- uQ, always reaches zero regardlesç of V, which is determined by f;. y,,
and Po,
and i, reaches the peak value given by
The final value of each variable in Interval 1 defines the initial conditions of Interval 2.
2.4.3 Intemai 2 (t, < t 4)
This interval starts at t = 1,. 0 2 is turned OFF and i, falls immediately to zero. A
current is forced to flow through L, and D, - and into the input dc line, with an initiai
current 1, determined by the energy conservative equation,
which yields
Substituting Eq.(2- 16) into (2- 1 8),
L, sees a constant voltage y,,. As Ql is ON, Lllt also sees a constant voltage y,,. Thus,
As Ql is ON and Q2 is OFF and Do is reverse biased. therefore
i., ( t ) = iU" ( t ) = O
At t = t,, i, reaches zero. Thus, by letting ia, be zero
Hence the voltage stress on Q-2 is govemed by
(2-23)
in Eqs. (ZZO), ta c m be found,
At the end of this interval, t = t,, iQ, reaches its peak value, as given by
where, f; is the switchuig frequency, and
D is the switching duty cycle of Q l , which is required to regulate the output
voltage at the static load conditions (discussed in Section 2.5.2) and is here equal
to (t3-t2)/T'.
The final value of each variable in Interval 2 defines the initial conditions of Interval 3.
2.4.4 Interval3 (t, < t 5 t,)
This interval starts at t = t,. Q2 is tumed OFF and C,.,l, is charged by in#. Refehg
to Fig. 2.4, the drain voltage of Ql in this interval satisfies the following equation
Giving the initial conditions determined by Eqs. (2-21) and (2-23): the solution of Eq. (2-
27) c m be obtained as
where, by the definition made previously.
As both Ql and Q2 are OFF and 4 is reverse biased, other variables in this
interval are found to be govemed by the following equations, respectively,
ici ( t ) = i&) = isz( t) = i J t ) = O (2-3 O)
Investigating the rising speed of uQ, by differentiating Eq. (2-28) and substituting
Eq. (2-29) into the result:
Eq. (2-31) reveals that the nsing speed of II,, is controlled by C,,. Giving the proper
value of C,, will suffikiently slow down the rise of uQ, and hence achieve ZVS in Q I at
the turn-O ff transient .
The magnetizing current is given by
At the end of this interval, t = t,. u,, reaches the clarnped voltage as given by
Thus, the duration of this interval cm be determined by substituthg Eq. (2-33) into (2-
28) and solving the resultant equation for (r,-t,). The final value of each variable in
Interval 3 defines the initial conditions of Intervd 4.
This interval starts at f = t,. 4, is fonvard biased and a current flows through N,.
and Do into the output end. As Nv has a inductance equal to n2Ln,, the initial value of the
secondary current, tP, can be obtained from the energy conservative equation,
which gives
Ns sees a constant voltage Va- therefore
Y, is (t ) = Isp - ( t - t4 ) (2-3 6 )
n- 4,
As Ql and Q2 are OFF and uQ, is clamped, thus, the following equations are obtained
ur,i ('1 = V p - y (2-3 7)
Because the converter is operated in the discontinuous conduction mode, the
secondary current i, will reach zero within one switching cycle. Thus, the relation
between the peak value of the secondary current and the output power Po is found to be
Hence, at the end of this interval. t = t,.
where, D'is the equivalent duty cycle of rectifier D,, and is equal to (t5-i4X.
The duration of this interval can be determined by combining Eqs. (2-39) and (2-40) and
solving the resultant equation for DI The final value of each variable in Interval 4
defines the initial conditions of Interval 5.
2.4.6 Intewal5 ( t5<t 5 Ts+ t , )
This interval starts at i = t,. 4 is reverse biased again. Ql and Q2 are OFF. C,,
and Lm undergo a resonance and this resonance obeys Eq. (2-27). Giving the initial
conditions as determined by last interval, and solving the equation for un,,
As Do and D, are reverse biased and Ql and QZ are OFF, thus,
is(t) = iQ1 (t) = iaq ( t ) = iap ( t ) = O
At the end of this interval, t = Tr+tl, uQ, reaches a voltage V,, which is the steady
state initial voltage of Interval 1 of each cycle. From Eq. (2-41), it is found that
1 Vfi = uQl(T + r l ) = y,, +-Y, cos[ao(T, + f l -QI (2-44)
n
Since LJtt is much larger than Lw, by comparing Eqs (2-42) to (2-12), in, is negligible and
hence ignored in the analysis of Interval 1 that is.
~J~ +f i ) * 0 (2-45)
The final value of each variable in Interval 5 defines the initial conditions of Interval 1 in
the next switching cycle' which just repeats the same process as analyzed above..
2.5 PERFORMANCE
2.5.1 Effects of the Non-Ideal Cornponents
The above analysis has been made mder the assunlptions of ideal components and
devices. In fact, each component and device of the circuit has some non-ideal properties.
For example, the leakage of the transformer, the ON resistance and the inherent
capacitance of switches, and so on. It is important to determine whether they make a
remarkable deviation.
It can be explained as follows. The ON resistance of switch causes the so called
conduction losses. The inherent capacitors indeed have some effects on the performance
of the switch. But the output capacitor of the switches c m be combined with Cd, and the
effects of input capacitor of the switches c m be made negligible by employing strong
gate drives. The blocking diodes in Fig. 2.1 can be chosen as neariy ideal ones, like the
Schottky diode or the ultra-fast diode. However. the effect of the Ieakage inductances of
the circuit are detectable on key wavefoms.
The effects of the leakage inductance of the transformer become apparent in
Interval 4. It is because in other intervals they c m be combined with magnetizing
inductance L,!,. However, in Interval 4, the flux linked inductance (L,) is clarnped at a
constant voltage ( V h ) , but the leakage inductance is not affected by t h i s clamp action.
The leakage, called Lkak in the following discussion, undergoes a resonance govemed by
where uQ, ,, is the voltage component of z iQI caused by the leakage, and
R, is the equivalent series resistance of the resonant loop.
The initial conditions of Eq. (2-46) are given by Eqs. (2-21) and (2-23). T'us, the
solution of Eq. (2-46) is
where,
Superimposing uQ, - ,(t) ont0 uQ, will give the total voltage of C,,.
Similarly, the leakage of L, and the inherent drain-to-source capacitor of Q2
undergo another resonance between the tumoff of Q2 and ta, as given by
where L is the leakage of L,,
Cm-, is the inherent capacitance of QZ,
a, ,, is given by
h is given by
where Ra , is the equivalent series resistance of the resonant tank.
Hence, superirnposing tiQ2-leak(t) ont0 uQ2 will give the total drain voltage of Q2.
2.5.2 Determination of Duty Cycle for the Output Regulation
Combining Eqs. (2-39) and (2-40),
Combining Eqs. (2-26), (2-33), (2-34). (2-40) and (2-53), the duty cycle required at the
input and Ioad conditions is found to be
When Po is so large that the second tenn in the bracket of right side of Eq. (2-54) is
negligible, Eq. (2-54) c m be approximated as
~ ~ ~ $ n x (2-5 5 ) K I
Then, substituting Eq. (2-55) into (2-54), and ignoring the second term, yield
n P:,, D = y, D' (2-56)
2.5.3 Advantages and Disadvantages of the Proposed Converter
The following major advantages of the proposed converter can be identified.
(i) No need of an isolated gate drive for the auxiliary switch, because Q2 and Ql
have common source connection as s h o w in Fig. 2.1.
(ii) Generating of the gate drive signals is simple, because the on time of Q2, unlike
in the case of the active clamp topology. is fixed at a quarter of the resonance penod of
the &,-Cs,, tank.
(iii) The ZVS conditions, provided by Intervals 1 and 3, are not affected by the line
and load conditions (refer to Eqs. (2-1 5) and (2-3 1)). Hence ZVS c m always be achieved
regardless of the linenoad conditions.
(iv) Less EMI is generated. Substituting Eq. (2-3 1) into (2-32) and differentiating the
equation with respect to time,
The fxst term in the bracket on the right hand of Eq. (2-57) principally govems the fdling
current in Ln,. It indicates clearly that the snubber capacitor C,, slows down the falling
speed of the current, and hence lowenng the EMI generated.
(v) Simple in circuitry, and al1 components of the auxiliary circuit require small
power ratings, since they only handle the stored energy in Cd,.
(vi) Conduction losses are not increased. Unlike in the active clamp topology, where
extra circulating power is required to flow in and out of the clamp capacitor, the proposed
topology does not involves the extra circulating power. Hence the conduction losses are
not increased.
The disadvantage of the proposed topology is that the auxiliary switch has a hard
switching turn-off. But the switching losses in Q2 can be reduced to a very low level by
selecting a MOSFET through trading off between the least ON resistance and the least
inherent capacitance, and also by increasing the ratio of L, to Lw, which lowers the
voltage stress on 0 2 at turn-off as given by Eq. (2-25).
2.5.4 RMS, Average, Peak Current and Voltage Values
In the design, the rms, average or peak current and voltage are required to
detemine the rating requirements in selecting the components. Some of these values have
been obtained in the analysis. Other values are calculated below.
(1) Rms current in Q I
The current in Ql has a triangle waveform. Hence the ms current is
Substiîuting Eqs. (2-26) and (2-54) into (2-58),
Correcting Eq. (2-59) for a conversion efficiency less than 1, Say ~l, the r r 113 - current
becomes
(2) Rms current in 0 2
The current in Q2 is one quarter of a sine wave, as given by Eq. (2-12). Hence. it
is found that
The average current is found to be
(3) Average current in D2
The current in D2 is ûiangular. From Eqs. (2-19) and (2-24), it is found that the
average current in D, is given by
1 t - t lm.-D2 = -In$ - =
2
(4) Average current in Df,
The current in Do is triangular. From Eqs. (2-39) and (2-40), the average current in
Do is given by
2.5.5 Losses Caused by the Auxiliary Circuit
(1) Switching losses in Q2: P.,. ,, It is estimated that
where C ( , s2 is the output capacitance of 02' and
[off -2 is the OFF transient time of 02.
(2) Conduction losses in @: P,.
The losses are found to be
P C - - ~ ~ = I;L~.Y-~)~ RD---
where RDQl is the ON resistance of Q2.
(3) Conduction losses in D, and D,: P,. .,,, The conduction losses in Dl and D, are found to be
Ir,, is the fonvard voltage drop of D,.
(4) Total losses by the auxiliary circuit
PI-0 = P r - x + P C . - ~ Z + Pc-DI&?
2.5.6 Losses Removed by ZVS Operation
(1) Switching losses in QI in a hard switching operation: Ps
The hard switching will cause switching losses in Ql as estimated by
where tOfl is the OFF transient time of QI, and
Cos, ,, is the output capacitance of QI.
(2) Net saving of power due to ZVS operation
The net saving of the power due to the ZVS operation of the proposed converter is
approximately
AP = P,-A - L e z (2-70)
2.6 EXPERIMENTAL RESULTS
2.6.1 Experiment Setup
A prototype ZVS flyback converter is built based on the design presented in
Chapter 5. The circuit operates under an input dc line varying fiom 90 V to 160 V. The dc
line voltage is obtained directly from a diode bridge rectiQing an ac line voltage in a
range of 75 V to 135 V. UC3855 is used as the PWM control chip. The schematic of the
circuit is shown in Fig. II. 1 in Appendix II.
The operating conditions of the circuit are surnmarized as follows: (i) the dc line
voltage range is 90 V to 160V (corresponding to a dc line voltage range 75 V to 135 V),
(ii) the full load is 50 W, with two outputs, of which the master output is 10 W at 5 V and
the slave output is 40 W at 20V, and (iii) the switching frequency is 200 kHz.
2.6.2 Key Waveforms
Fig. 2.5 shows the theoretical wavefoms of II,, and in,, which are drawn in
MathCAD based on the equations which are found to govem ilp, and i,, in the analysis
presented in the previous sections. Fig. 2.6 shows the experimental results of the drain
voltage and current of Q I in the prototype converter. Cornparison between the theoretical
and experimental results are made in Table 2.1. It is seen that they are in good agreement.
l L I voltage ;\. ,9., - l t
i \ : : ,?., ,n 1 ! \, ; 1. : ;,J L j i i ;
\ current ; I ldf I
Fig. 2.5 Theoretical results: the drain voltage and current waveforms of Ql in Fig. 2.1, which are predicted by the steady sate analyses made in Section 2.4 and 2.5. The negative
drain current is due to the latching of the body diode of Q1 before it is turned on. Operating conditions: V,,,=13OV, Pn=50 W.f;=2001cHz
Fig. 2.6 The experimental results: the drain voltage and current wavefoms of Ql in the prototype ZVS converter of Fig. 2.1. The theoretical waveforms are shown in Fig. 2.5.
Scales: vertical--5OV/div., 1 Mdiv.; horizontal-0.5ps/div. Operating conditions: yl,=l 3 OV, P0=50W, f;=200kHz
Table 2.1 Cornparison between the theoretical and experimental results of the current, voltage and duty cycle of the main switch.
I 1 I I I I I 1 I
-
current - scales: -
-
Fig. 2.7 The drain voltage and current waveforms of Q2 in Fig. 2.1, which are predicted by the steady sate analyses in Sections 2.4 and 2.5.
Operating conditions: V,= 13 OV, <=50WYf;=200kHz
(V,,,=l30V, P,=5O W,1;=200kHz)
vg8 1 &d
210V 3.16A 220V 1 3.22A
Vpmk
273V 280V
Theoretical Experimental
D 0.23 0.22
Fig. 2.7 shows the theoretical wavefoms of u, and i,, which are drawn in
MathCAD based on the equations which are fomd to govem up and i , in the analysis
presented in the previous sections.
Fig. 2.8 shows the experimental results of the drain voltage and current of Q2 in
the prototype converter. Comparison between the theoreticai and experimental results are
made in Table 2.2. It is seen that they are in good agreement.
Fig. 2.8 The experimental results of the drain voltage and current wavefoms of Q2 in the prototype ZVS converter of Fig. 2.1. The theoretical waveforms are shown in Fig. 2.7.
Scales: verticaMOV/div. 1 Ndiv.; horizontal-0.5ps/div. Operating conditions: y,1=130V, <,=5O W,j;=200kHz
Table 2.2 Comparison between the theoretical and experimental results of the current, voltage and duty cycle of the auxiliary switch.
(ytI=1 30V, Pl,=50 W,f;=200kHz)
Theoretical Vo!
47.3V D
0.23 I,,k
2.69A ta-t2
1 . 6 9 ~
Figs. 2.9 and 2.10 show the experirnental key waveforms of the main switch Ql
and the auxiliary switch Q2, respectively, under light load condition. It is seen that ZVS
is not lost at light load.
Fig. 2.9 Experimental results: the current and voltage waveforms of the main switch QI in ZVS operation under light load condition:
V II 1 =120V, P,,=l OW' f;=ZOOkHz. Scales: vertical-50Vidiv. 1 Ndiv.; horizontal-OSps/div.
Fig. 2.10 The current and voltage waveforms of the auxiliary switch Q2 under light load condition: y,,= 120V, Po=I O W, f;=200kHz.
Scales: verticaL50V/div. 1 A/div.; horizontd-O.Sps/div.
2.6.3 Effïciency
Experimental resdts of the efficiency from the prototype converter are obtained.
The input dc voltage is obtained fiom a diode bridge rectifier which rectifies the ac line
voltage varying between 75 V to 135 V (ac). For convenience, the experimental results of
efficiency inciude the effect of this input rectifier stage.
I Li - - -'C -
ZVS Flyback convener efficiency
8 . - O 66- conventiona1 Flyback converter
58 ' input ac voltage (V)
56 75 85 95 105 115 125 135
Fig. 2.11 The overall eficiency of the converter vs. input AC voltage (the range of AC voltage corresponds to a dc input voltage range of 90 to 160V).
The eff~ciency of hard switching operation of the converter is made in cornparison here. The output power is P0=50W, and f;=ZOOkHz.
A 75 t ZVS flyback converter
6 conventional flyback converter
I Output Power (W) 50 - b
10 20 30 40 50
Fig. 2.12 The overall efficiency vs. load under given input voltage for both the proposed and conventional circuits. The efficiency measured includes the efficiency of the input ac-to-dc bridge
rectifier stage. YI,= 1 20V, 1;=200kHz.
Fig. 2.1 L shows the efficiency of the converter vs. input voltage under the full
load condition (50W). The comparison behveen the efficiencies of the ZVS and the hard
switching flyback converters is also made in Fig. 2.1 1. The comparison shows that ZVS
indeed makes a remarkable increase in efficiency: an increase of about 7% is obtained.
Fig. 2.12 shows the overall efficiency of the converter vs. load under given input
voltage for both the proposed and conventional circuits. The proposed circuit always has
higher efficiency than the conventional hard switching flyback converter.
2.7 CONCLUSIONS
In this chapter a ZVS flyback converter topology has been presented and
analyzed. The steady state operation of the circuit can be divided into five distinct
intervals. The steady state analysis of the circuit is performed and verified
experimentally .
Analysis and experiments show that the performance of the proposed converter is
enhanced with the auxiliary circuit. 7% higher eficiency has been achieved on the
prototype circuit as compared with the conventional hard switching converter. Other
major merits of the proposed ZVS flyback topology include:
(i) Lossless switching of the main switch, independent of the line and load conditions,
(ii) Simple power and control circuitry,
(iii) No increase in conduction iosses.
(iv) Ability to operate in either voltage or current mode control.
Therefore, the proposed converter topology provides an attractive alternative to
the existing ZVS flyback topologies and it is prornising in wide applications in
telecommunication and cornputer systems in the f h r e .
A ZVS FORWARD CONVERTER TOPOLOGY
3.1 INTRODUCTION
The single switch fonvard converter topology is extremely popular in
telecommunication and distributed dc bus applications. This topology offers a cost
effective solution to fil1 the void created between the low power flyback topology and the
more complex high power bridge types. Operating in continuous inductor current mode,
the fonvard converter utilizes a much lower peak current than its flyback counterpart.
This is advantageous with low voltage inputs.
As reviewed in Chapter 1, the conventional forward is not able to achieve high
power density and high efficiency. and the existing ZVS forward converters have several
previously mentioned drawbacks. Better solutions should be searched for.
This chapter presents a ZVS fonvard converter topology employs a similar
auxiliary circuit to that in chapter 2, with a little modification. The auxiliary circuit helps
to achieve ZVS in the main switch of the converter.
The converter operates in continuous mode in order to reduce the peak current in
the switch. A steady state analysis of the converter is presented in this chapter to
understand the behavior and to develop performance charactenstics.
The organization of this chapter is as follows. In section 2.2, the circuit
description is presented. In Section 2.3, the modes of operation are detailed. In Section
2.4, the steady state analysis is performed. in Section 2.5 performance is discussed and in
Section 2.6 experimental verifications are made.
3.2 CIRCUIT DESCRIPTION
Fig. 3.1 shows a ZVS fonvard converter ropology employing the sirnilar auxiliary
circuit to that in Fig. 2.1, drawn in the dashed Iine block. A modification on the auxiliary
circuit is that there is an additional small inductor L , in series with the secondary winding.
The reason of this modification is revealed by the analysis in Section 3.4.2.
-
Fig. 3.1 The proposed ZVS fonvard converter topology.
In Fig. 3.1, the basic foward circuit is comprised of (i) the input capacitor Ci,,,
(ii) the power transform T, which has three windings, the primary Np the secondary N,,
and the tertiary N, that helps to reset the core, (iii) the main switch QI, (iv) the rectifier
Do, and the fieewheeling diode D,,?, (v) the output inductor Lo, (vi) the output capacitor
Co, (vii) the blocking diode Dr, and (viii) the load Ru.
n i e auxiliary circuit consists of (i) a snubber capacitor Cd, which is in parallel
with 01: - (ii) two coupled inductors L, and La, (iii) two blocking diodes D, and 4, (iv)
an auxiliary switch Q2, and (v) a srnall inductor L-,, which is inserted into the secondary
side of the power circuit.
3.3 MODES OF OPERATION
Fig. 3.2 shows key waveforms of the proposed converter of Fig. 3.1. The
operation of the circuit can be divided into the following seven intervals. In a standard
fonvard converter, there are no Intervals 1, 2 and 4. These intemals are fulfilled by the
auxiliary circuit in order to achieve ZVS in Q I .
3.3.1 Interval 1 ( tl < t t2 )
Fig. 3.3(a). At the beginning of this intervai, 0 2 is switched ON. Thus Cm,, LoP
and L, form a resonant tank, as show in Fig. 3.4. C,,, begins to discharge through Lw
and Q2. The discharging current rises resonantly and builds up a magnetic field in the
core of Lw. In this way the discharged energy fiom C,, is transferred to the core. The
voltage of Cd drops to zero within this interval.
3.3.2 Intewal2 ( t, < t t3 )
Fig. 3.3@). At the beginning of this interval, Ql is switched ON under the zero
voltage condition. At the sarne moment, QZ is switched OFF, which stops the current in
L, rapidly. The rapidly falling current reverses the voltage polanty at the dotted ends of
the coupled inductors. Thus, & becomes foward biased and is forced to conduct,
releasing the stored energy in the core of L.v.
Intervals: 7 1 3 3 4 5 6 7 1 3 -
Gating for main sw itch O N
O tj W T s w t7 l3 l 4 r6 t7 t
G ating for auxi- A I
O N
+ + 1
A L I Voltage and - - - -!
" t d Q , /
current o f 'Y I
/ m a h sw itc h 0 1 f
I I O b t
v O ltag current o f I
'Y m a h sw itc h 0 1 f I I
O b t 4 1
V oltage and current o f aux i - \
liary sw itc h - U @ l 1 / f
b f
Secondary 4 current
O b t Currentin
2nd coupied O . ta I +
inductor t 1 h t3 t4 t5 t 6 f7 f i + & l
tim e
Fig. 3.2 Key waveforms of the proposed converter of Fig. 3.1. The converter operates in seven modes each switching cycle. Variables are defined in Fig. 3.1
* (a) interval 1 (t,< t 5 t,)
( d ) Interval 4 (rd< r l r,)
O (e) Interval 5 (t,< t 5 1,)
v (f) Interval 6 (r,< t 5 t,)
(g) Interval 7 (t,< t 2 T,+tI)
Fig. 3.3 Modes of operation of the converter of Fig. 3.1. The operation is divided into seven modes per switching cycle. The key waveforms are shown in Fig. 3.2.
The curent in Ql reflects the secondary current, which nses slowly owing to the
existence of L.$. Both Dot and D,,, are conducting to give the total current in L,, with the
current in D,, nsing and that in 4, falling linearly.
Fig. 3.4 The equivalent circuit of the discharging process in Interval 1. Variables are defined in Fig. 3.1.
Fig. 3.5 The equivalent circuit of the charging process in Interval 5. The variables are defined in Fig. 3.1.
3.3.3 Intewal3 ( t 3 < t 5 t , )
Fig. 3.3(c). At the beginning of this interval. cment in D,, reaches the value of
the current in Lu, and current in D,, falls to zero. Power is delivered to the load in the
same rnanner as in the standard fonvard converter.
3.3.4 Intewal4 ( t4 < t 5 t5 )
Fig. 3.3(d). At the beginning of this interval, QI is switched OFF. The pnmary
cwent charges C,,,. With a sufficient value of capacitance, CIlb can satisfactonly slow
down the speed of nse of the drain voltage of Q l so as to facilitate a ZVS tumoff in Q I .
Starting at the beginning of this interval, the current in Do, decreases and Do, is
forced to conduct by L,. L, and C,, f o m s a resonant tank shown in Fig. 3.5. The current
in pl and that flowing into C.$), fa11 resonantly, and the drain voltage of Ql rises
resonantly. The current in Do, rises accordingly.
33.5 Interval 5 ( S < t < t6 )
Fig. 3.3(e) shows the circuit operation during this interval. At the beginning of
this interval? the drain of Ql reaches a voltage of 2V,,). 9 is forced to conduct. Thus, the
voltage of the primary winding of T is clamps to a voltage of - V_, and the drain voltage
of Ql is clamped to 2y,,. In this way the core of T is reset. During this interval, Ls sees a
clamped voltage and the current in D,,, falls linearly.
3.3.6 Intewai 6 ( t6 < t I, )
Fig. U(f ) shows the circuit operation during this interval. At the beginning of this
interval, the current in L,% exhaust, D,, is reverse biased and Do2 freewheels the total
inductor current in L,. The voltage clamp action lasts until the current in Dr fa11 to zero at
the end of this interval.
3.3.7 Interval 7 ( fi < t TS+tl )
Fig. 3.3(g) shows the circuit operation during this interval. At the beginning of
this interval, the clamp action stops. C,, and the magnetizing inductance Lm of T form a
resonant loop. The drain voltage of Ql starts to oscillate about the input dc line voltage
y,:,,, starting fiom 2y1:,,.
Following Interval 7, the process repeats fiom Intervals 1 through 7.
3.4 STEADY STATE ANALYSIS
The operating principle has been described in last section. In this section, the
steady state analysis is performed with the assumptions made below. Based on the
analysis, the tirne v q i n g variables such as the current and voltage of the principal
components and devices are detemined, the performance of the converter is illustrated,
and the resultant quantities such as the rms? average, or peak current and voltage of the
components/devices are obtained. These quantities are used in designing the converter as
presented in Chapter 5.
The analysis below uses a similar approach to that in analyzing the flyback
converter in Section 2.4. Similady. the initial conditions can be obtained by the
previously mentioned iterative process.
3.4.1 Assumptions, Definitions and Initial Conditions
For convenience, the following assumptions are made:
(i) The steady state conditions have been established, and the converter is ninning at the
nominal output voltage Va and the static load conditions: Po-
(ii) Each component and device has ideal properties, that is,
(1) T,: the leakage inductance is ignored. and the core does not saturate,
(2) Lw, L,: the coupling factor is 1.0' and the core does not saturate,
(3) Co, pure capacitors, and the capacitance of Cc, is infinitive,
(4) Lu: the inductance of L,, is infinitive,
( 5 ) Dz? Du,, D; the fonvard voltage drop is O V, the recovery time is O s,
(6) Q I , Q2: the on resistances are O R, the inherent capacitances are O F.
The effects of the non-ideal properties of the components and devices are investigated in
Section 2.5.1.
(iii) The magnetizing inductance of the transformer, Ln,, is much larger îhan L,.
The following parameters are defined:
on--angular frequency of the resonant tank of $,,,. L, and L,(Interval l),
a,--angular fiequency of the resonant tank of C.v,l, and L,,l(Inte~al 7),
a;-angular fiequency of the resonant tank of C,, and L,(Interval4),
' --instantaneous current in 02.
IV--peak current in QZ,
i_--instantaneous current in L,, . I_--peak current in Lm,
i_--instantaneous input current.
(,IP4ak--peak magnetizing current in TF'
Iboutput current,
i,,--instantaneous current in Q I ,
is---intantaneou secondary current,
I,peak secondary current.
k--factor by which V, deviates fiom y,,, LI--reflected value of L, seen from the primary side of T ,
L,-magnetizing inductance of Tr7
Li-effective inductance of paralleling LI and L,,,,
n--reciprocai of tums ratio of T , equal to NJN,,
V--output voltage,
V --clamped voltage stress of QI when it is OFF, P-Q 1
V P m --clamped OFF voltage of QZ when it is OFF,
u,,--instantaneous drain voltage of QI, or voltage across C,$,,,,
Vo--initial drain voltage of QI per switching cycle.
According to the definitions and basic circuit theory,
J 1 L,+L, a,, = -*
C,,, La,, LI-
The analysis begins by stating the initiai conditions of Interval 1 :
(i) Both QI and 0 2 are OFF, and the drain voltage of the paralleled switches Ql and Q2
is Y,,
where k is the factor by which V, deviates from Y i , and
$J~J = $&tJ = 0 (3 -7)
(ii) The magnetihg inductor L,,, and C,), undergoes a resonance in the end of last
switching cycle, but the resonant current in Lll1 is negiigible?
illI(tl) = 0 (3 -8)
(iii) D, , Do, and Dr are reverse biased. and D, , fieewheels the total output current,
i&J = i,(tl) = O (3-9)
3.4.2 Interval 1 ( t , < t 5 t2 )
This intenta1 starts at t = t , . 0 2 is turned on and a resonant tank shown in Fig. 3.4
is fomed. The resonance process obeys the foilowing equations:
4. L,. - + rrc,, = C.:il dt
Combining Eqs. (3- 10) through (3- 12):
Giving the initial conditions determined by Eqs. (3-6) through (3-8)' the solution of Eq.
(3-1 3) is
Eq. (3-14) indicates that, if L were absent fiom the circuit of Fig. 3.1, narnely L, = 0, uQ,
could never reach zero. Under this condition, no ZVS would be achieved, To soive this
problem, L, is hence introduced.
It can be extended fiom Eq. (3- 14) that, if there exists a relation
un, will always reach zero when the fxst
impossible to satisfy Eq. (3-15) over al1
half of the resonant process finishes. But it is
the operating range, because k is not a fixed
parameter and it is detemined by the line and load conditions. For simplicity in design,
let
L , = Lap (3-1 6)
and set the duration of this interval to be a half of the resonant period. Thus, if the
condition k > O is always valid. ir,,, - will always reach zero within this duration and is
automatically clamped at zero by latching in of the body diode of Ql in the rest of this
interval. It will be shown in Section 3.4.8 that this condition c m be reached by selecting a
large 4,-
Referring to Eqs. (3-1) and ( M ) , Eq. (3-16) can be satisfied by selecting the
value of Ls as given by
Thus, Eq. (3-14) c m be rewritten as
From Eqs. (3- I l ) , (3-IO), (3-18) and the initial conditions (3-7) and (3-8). the
following variables can be determined as
Y I 1 1 i ( ) ( t - t , ) + ( T + k ) - sin[co ,, ( t - t, )] (3-19) 2L, OJ. 4,,7
Refer to Fig. 3.4, the magnetizing and the secondary currents of the transformer c m be
detennined, respectively, by
illm = =' ~ll lw L I + Li
Since D, is reverse biased and Ql is OFF.
As explained previously, at the end of this interval.
Thus, the ZVS condition for turning on of Ql is established.
The final value of each variable in Interval 1 defines the initial conditions of
Interval 2. For convenience in the following analysis, let
1 LI,, ISI = i , ( tz) =-- C I 1 Ytl -(t2 --il) " -.-(f2 - t , ) L I + LOP 2Lap 2Lap
3-43 Interval2 (t, < t 2 t,)
This interval starts at t = t2. Sirnilady to Section 2.4.3, it can be determined that
From E q . (3-30), the duration when id., flows
t" - r2 =
is determined
During (ta-t,), uQ2 will be clamped at a voltage the same as given by Eq. (2-25).
As L-< sees a constant voltage n y,, the secondar). current rises linearly
As Ln[ sees a constant voltage YI[, thus,
Hence the primary current is
As Q2 is off and QI is on,
At the end of this interval, the secondary cment reaches the value of the output
current Io
i \ U 3 ) = Ir, (3 -3 7)
Hence, fkom Eqs. (3-32) and (2-37), the duration of this interval can be detennined as
4 f3 - t , = ( I o - I J - (3 -3 8)
Y I ,
If Io < I,,, which is a case under the light load conditions, Interval 2 will not exist. It is
because the secondary current has reached 4, in Interval 1. Thus, Do, is reverse biased and
the total output current flows through Q,,. Under such conditions, the operation directly
enters Interval 3 fiom Intenirai 1.
The final value of each variable in Interval 2 defmes the initial conditions of
Interval 3.
This interval starts at r = t,. In this interval, the operation of the circuit is the same
as the standard fonvard converter: LI,, sees a constant voltage y.,, i5 is cornmanded by L, to
be constant at Io, and the total input current flows through QI. Therefore, the following
equations govern the respective variables during this interval.
The fmal value of each variable in Interval 3 defines the initial conditions of
Interval 4. For convenience in the following analysis, let
I l I l 2 = ilIl ( t4 )
3.4.5 Intewal 4 (t4 < t -< t5)
This interval starts at r = r,. Q l is tumed OFF at the beginning of this interval. C.vllb
is charged and the equivalent circuit is shown in Fig. 3.5. Thus. uQ, is govemed by the
following differential equation,
Giving the initial conditions determined by Eqs. (3-41) and (3-43). it is found that
As L , sees a voltage (y,; uQ,) and L,. sees a voltage n(yjTus,), thus,
And the input current is
As both QI and Q2 are OFF.
iQ,(t) = i&) = O
At the end of this interval, uQ, reaches 2 yll, i.e.,
uc,(4) = 2 YI, (3-5 1)
Then L,,, sees a negative voltage -y,, and Dr becomes fonvard biased and starts to
conduct, clamping u,, at 2Tl1.
The final value of each variable in Interval 4 defines the initial conditions of
Interval 5 . For convenience in the following analysis. let
I,? = i&) (3 -52)
' n t 3 = iw ('5 ) (3 -53)
This interval starts at t = t,. The core of T is reset as Dr conducts. Ltll sees a
constant voltage - ij1, and L,, sees a constant voltage -n ytl. Thus,
As Dr conducts, the tertiary current which consists of and the r e f h e d secondary
current flows into the dc line,
~ / l t ~ o = - ~ / / / ( ~ l -ni.&)
As both QI and Q2 are OFF and uu, is clarnped, therefore
ici ( 1 ) = i&) = O
q,IW =
At the end of this interval, < reaches zero
qf,) = O (3 -59)
Thus, Do,becomes reverse biased again and D , freewheeis the total output current 1,. 0,
From Eqs. (3-55) and (3-59). the duration of this interval is determined by
The final value of each variable in Interval 5 defines the initial conditions of
Interval 6.
3.4.7 Interval 6 (r, -= t 2 t,)
This interval starts at r = r,. The resetting of the core continues. If the resetting
finishes in Interval 5, this interval will be skipped and the operation enters directly into
Interval 7 fiom Interva1 5.
As Ln, sees a constant voltage -yl, and u,, is still clarnped,
The tertiary current consisting of i flows into the dc line,
i,,, (0 = -ilIl (0
As Do, is reverse biased and Ql and QZ are both OFF,
i , J f ) = iQ , ( f ) = i&) = O
At the end of this interval. ilIl fails to zero, finishing the resetting
inlO,) = 0 (3-65)
Dr becomes reverse biased again. From Eqs. (3-61) and (3-65), the duration of this
interval is detennined by
Ill13 LI11 42 =.T t , - t o = ( t , - ~ ~ ) - ( t ~ - t ~ ) = - - - (3-66) Y I 1 f i Y,,
The final value of each variable in Interval 6 defines the initial conditions of
3.4.8 Interval 7 (t, < r < t8)
This interval starts at r = t,. C , , , and L,,, undergo a resonance owing to the
difference between uQ, and the input dc line voitage Similarly to Section 2.4.6,
uc>,(t) = cl {l +cosbO(t - r d ) (3 -67)
Substituting Eq. (3-5) into (3-68),
7
Owing to the large value of L,,,, i,,, c m be ignored and regarded as O. This is consistence
with assurnption (ii) in Section 3.4.1 (Eq. (3-8)). As 4, is reverse biased and both Ql and
Q2 are OFF, thus
ia,,(t) = iQ, ( t ) = iJt) = O (3-70)
iJt) = i,,,(t) = O (3-71)
At the end of this interval, one switching cycle finishes. As in the steady state
operation, uQ, reaches Vo. Referring to Eqs. (3-6) and (3-67), the factor by which the
steady state initial drain voltage of QI deviates from y,, is detennined by
k =cos[o,(T, +t , -t,)] (3 -72)
Selecting a large LI## will result in a small a, (for exarnple. let o,T < rr) and hence k>O so
as to achieve ZVS over al1 the line/load range (refer to Section 3.4.2).
The final value of each variable in Interval 7 defines the initial conditions of
Interval 1 in the next switching cycle. which repeats the same process as in this cycle.
3.5.1 Effects of the Non-Ideal Components/Devices
Each realistic component and devices has some non-ideal properties as rnentioned
in Section 2-51. For the sarne reasons as explained in Section 2.5.1, the non-ideal
properties will not make a significant deviation fiom the operation of the circuit discussed
above.
A similar process to that in Section 2.5.1 can be followed to detennine the effects
of the Ieakage inductances, and it will not be repeated here.
3.5.2 Determination of Duty Cycle as to Regulate the Output Voltage
Refer to Fig. 3.3. In one switching cycle, L, sees a constant voltage n t - V o for a
duration of (D-lI,)T$, where D is the duty cycle of the main switch Q l and D, is the
reduction of the effectua1 duty cycle owing to L,, which is given by
Q = U? -Qf, (3-73)
and Ln sees another constant voltage -y , for the rest of the cycle, thus, the volt-second
balance of L , gives the following equation:
(D-D,) (ny,,- Q - (1 -D-D.JV0=O
which yields
( D - D, ) 17 ï<,, = y,
Assuming the output power is P,,. thus
Substituting Eq. (3-38) into (3-37), and substituting (3-26) and (3-76) into the result
Let D,, represent the duty cycle of the auxiliary switch Q2,
Dm= (12 + , I f ;
Getting (1, - . -t,) fiom Eq. (3-78) and substituthg the result and Eq. (3-17) into (3-77), it
yieIds
Eq. (3-75) indicates that, in order to regulate the output voltage at certain Line/load
conditions, the duty cycle of the main suitch should be
3.5.3 Advantages and Disadvantages of the Proposed Converter
The following major advantages of the proposed converter can be identified fiom
the above analysis.
(i) No need for an isolated gate drive for the auxiliary switch, because Q2 and Ql
have a common source connection (refer to Fig. 3.1 ).
(ii) Generating the gate drive signal is simple. because the on time of Q2, unlike in
the case of the active clamp topology. is fised ar a half of the resonance penod of the L,
L6Cd tank (refer to Section 3.4.2)
(iii) The ZVS turn-on transient of Q I , facilitated by Intervals 1, is not afZected by the
line and load conditions. T h e turn-off transient of Q I is affected by the load current Io as
shown in Eq. (3-47). With a suficient value of capacitance, C, will guarantee a ZVS
tunioff in Q I under the full load condition. Under Iight load, the rise of uQ, will be
slower. Hence, ZVS will always be achieved regardless of the line/load conditions.
(iv) Less EMI is generated owing to that C,, slows down speed of change of the
inductor current and capacitor voltage, as indicated by Eqs. (3-46) and (347).
(v) It is simple in circuitry, and al1 components of the auxiliary circuit require smdl
power ratings.
(vi) Conduction losses are not increased. Because C,, is always there to slow down
the voltage rise speed and to lower QI's voltage stress. Unlike in the existing techniques
where extra circulating power is need by the clamp circuit, the auxiliary circuit does not
draw extra power.
The disadvantage of the proposed topology is that the auxiliary switch has a hard
switching turnoff. But the switching losses in QZ can be minimized by selecting a
MOSFET through trading off between the least ON resistance and the least inherent
capacitance, and also by increasing the ratio of L,, to Lup, which lowers the voltage stress
on 0 2 at turnoff.
3.5.4 RMS or Average Current Values
In the design, the rms, average or peak current and voltage are required to
determine the rating requirements in selecting the components. Some of these values have
been obtained in the analysis. Other values are calculated below.
(1) RMS current in Ql
The current in Ql can be approximated with a square waveform (Fig. 3.2), thus,
For a conversion eficiency less than 1. Say 7, the RMS current becomes
(2) RMS current in Q2
The curent in QZ consists of two components, one is in a triangle wavefonn, the
other is a half penod of a sinusoidal waveform, as given by Eq. (3-19). Thus,
The average current is found to be
(3) -Average current in D,
The current in D, - is triangular. From Eqs. (3-22). (3-26) and (3-28). it is found
that the average current in D, is given by
(4) Average currents in Do, and D,,?
The current in D,, is of a complex waveform. The average current in Do can be
calculated as
and the average current in Do? is determined by
3.5.5 Losses Caused by the Auxiliary Circuit
(1) Switching losses in Q2: P, ,, It is found that
1 1 P , ~ ~ = - c,,-Q~ V: f , + 1, vil, JZJZLF-Q~ f,
2 3
where Co-xT uz is the output capacitance of Q2. and
t~ al is the OFF transient tirne of Q2.
(2) Conduction losses in QZ: P,. oz
The Iosses is found to be
P < - - ~ ? = I Xif-s-CJZ R I > - ~ Z
where RD-@ is the ON resistance of m. (3) Conduction losses in Dl and D,: P,. ,,,, ,
The conduction fosses in D, and D, are found to be
Pc-DIU = L'I..-I)I I c t ~ . - c ~ z + V l - . _ ~ loi-^^
where V' ., is the fonvard voltage drop of Dl. and
V' , is the forward voltage drop of D,.
(4) Total losses by the auxiliary circuit
3.5.6 Losses Removed due to ZVS Operation
(1) Switching Iosses in Ql in a hard switching operation: P,
The hard switching will cause a switching losses in Q l as given by
where tofl ,, is the OFF transient time of Ql? and
C,,, is the output capacitance of QI.
(2) Net saving of power due to the ZVS operation
The net saving of the power due to the ZVS operation of the proposed converter is
approximately
3.6 EXPERIMENTAL RESULTS
3.6.1 Experiment Setup
A prototype ZVS fonvard converter is built based on the design presented in
Chapter 5. Since the circuit outputs a power of 100 W at an output voltage of 5 V,
synchronous rectifiers are chosen in the output rectification stage of the circuit so as to
reduce the losses in the stage. An improved self-driven gate drive technique for
synchronous rectifier in the forward converter [18] is used to optimize the performance of
the synchronous rectifiers. The schematic of the prototype converter is shown in Fig. 11.2
in Appendix 11.
The operating conditions of the circuit are summarized as follows: (i) the input dc
line voltage varies h m 40 V to 60 V, (ii) the full load is 100 W at a output voltage of 5
V, and (iii) the switching fiequency is 300 kHz.
3.6.2 Key Waveforms
Fig. 3.6 shows the theoretical waveforms. which are drawn in MathCAD based on
the equations of uQl and iQl obtained in the steady state analysis in Section 3.5. Fig. 3.7
shows experimental results of the waveforms of drain voltage and current of Ql in the
prototype converter. These two figures are under the same operating conditions: V,,,=5 OV,
Po=lOOW. The cornparison between the theoretical and experimental results is made in
Table 3.1. It is seen that they are in good agreement.
Fig. 3.6 The drain voltage and current waveforms of Q1 in Fig. 3.1, which predicted by the steady state analysis made above. The negative drain current
to the latching of the body diode of Ql before it is turned on Verticai: 2A/div., 20V/div., Horizontal: O.Sps/div.
Table 3.1 Cornparison between the theoretical and experimental resdts
are is due
of the current, voltage and duty cycle of the main switch (~,,=5OV, P,,=1 OOW,f;=3OOkHz)
D VO 71 .OV 77.0V
Vpeak Ipeak 6.67A 6.80A
Theoreticai Experimental
0.29 0.3 1
1 15V 118V
Fig. 3.7 The experimental results of the drain voltage and current of Ql in a prototype ZVS converter of Fig. 3.1. The theoretical waveform is shown in Fig. 3.6
Vertical: 2Aldiv.. 2OV/div.. Horizontal: O.Sps/div.
Fig. 3.8 shows the theoretical waveforms of u,,, and io, based on the steady state -
anaiysis in Section 3.4 and 3.5. Fig. 3.8 is drawn with the software MathCAD. Fig. 3.9
shows experimental results of the waveforms of drain voltage and current of Q2 in the
prototype ZVS converter of Fig. 3.2. These two figures are under the same operating
conditions: r , 4 O V . P,l=lOOW. The cornparison between the theoretical and
expenmental results is made in Table 3.2. tt is seen that they are in good agreement.
Table 3.2 Cornparison between the theoretical and experimental results of the current. voltage and duty cycle of the auxiliary switch
(vjJ=50V, C1=l 00W,f;=300kHz) heak
3 .56A v ~ !
9.12V Theoretical vp,al, 1 15V
t I 1 I l I t I I I I Fig. 3.8 The drain voltage and current waveforms of Qt in Fig. 3.1,
which are predicted by the steady state analysis made above. Vertical: 1 A/div.. 20V/div.. Horizontal: O.Sps/div.
Fig. 3.9 The Experimental results of the drain voltage and current wavefoms of Q2 in Fig. 3.1. The theoretical waveform is shown in Fig. 3.8.
Vertical: lA/div., 2OV/div., Horizontal: 0Spsldiv.
Figs. 3.10 through 3.16 show the current and voltage wavefoms of the main and
auxiliary switches at different Iinelload conditions, veriQing the analysis in that ZVS is
always achieved in the main switch.
Fig. 3.10 Experimental results: the current and voltage of Ql at high line full load. Po= 1 00 W, V,=6OV. f.=300kHz. Scales: vertical--20V/div., 2Ndiv.; horizontal--
O.jps/div.
Fig. 3.11
A .spc
Experimental results: the current and voltage of Ql at hi& line light load. P,=20W, ~.11=60V,f;=300kHz. Scales: vertical--20V/div., ZA/div.; horizontal--O.Sps/div.
Fig. 3.12 Experimental results: the current and voltage of Ql at low line full load. Pt,= 100 W. yll=40v'f;=300kHz.
Scales: vertical--20Vldiv. 2Mdiv.: horizontal--OSps/div.
A - 5 p s
C
- 0
Fig. 3.13 Expenmental results: the current and voltage of Ql at low line light load. P,,=20 W. V,,=qOV.f;=3 00kHz.
Scales: vertical--20Wdiv.. 2Ndiv.: horizontal--0Spddiv.
curent 1
Fig. 3.14 Experimental results: the current and voltage of Q2 at medium line half load. P,=50W. C;,,=jOV.~=300kHz.
Scales: vertical--2OV:div.. 1 illdiv.: horizontal--0.5jddiv.
HAG current ,,,
Fig. 3.15 Experirnental results: the current and voltage of Q2 at high Iine light load. P,,=20 W. V,,=60V?f;=3001cHz.
Scales: vertical--2OV/div., 1 N d i v . horizontal--0.5ps/div.
Fig. 3.16 Experimental results: the curent and voltage of Q2 at low line light load. P =20W. Vt,=4OV.f,=30OkHz.
Scales: vertical--2OV/div.. 1 Ndiv. ; horizontal--0Sjddiv.
3.6.3 Effciency
Expenmental results of the efficiency from the prototype converter are obtained.
The input dc voltage varies from 40 to 60 V. The full load power is 100 W. Synchronous
rectification is used in the circuit ourput rectifying stage.
Fig. 3.17 shows the experimental results the efficiency vs. input voltage under full
the load condition (100W). The c o m p ~ s o n between the efficiencies of the ZVS and the
hard switching operation of the fonvard converter is made in Fig. 3.17. The hard
switching forward also employs synchronous rectifiers. The c o m p ~ s o n shows that ZVS
indeed makes a significant increase (about 5%) in efficiency. ZVS operation at fiequency
of 200 kHz is also tested. where the snubber capacitor and the prirnary of the coupled
inductor are adjusted to Suit for this frequency.
The efficiency is about 1% higher than that at 3001cHz The gain in efficiency is
due to the reduced switching losses in the auxiliary switch, and the gate drives of the
synchronous rectifiers.
40 45 50 55 60
Input voltage (V)
Fig. 3.17 The efficiencies vs. input voltage of the prototype converter in ZVS operation and non-ZVS operation of the forward converter
(conditions: P,,= 100 W. y,=5V).
+ ZVS operation Hard switching
0 - . -. -- - . - -- . - -- b
O 20 40 60 80 100
Output power ( W) - Fig. 3.18 The efficiencies vs. load of the prototype converter in ZVS operation and non-ZVS operation. Conditions: V,,,=50 V,f;=300 kHz.
Fig. 3.18 shows the eff~ciency vs. load under given input voltage for both the
proposed and conventional circuits. The proposed circuit always has better efficiency
than the conventional hard switching fonvard converter.
The efficiency is expected to be higher if optimized magnetics are used, and if a
PCB circuit is built. That means, using the proposed forward topology, a 90% forward
converter outputting 20A at SV is possible.
3.7 CONCLUSIONS
In th is chaprer a ZVS forward converter topology has been presented and
analyzed. The steady state operation of the circuit can be divided into seven intervals. The
steady state analysis of the circuit is perfom~ed and verified experimentally.
Analysis and experimental results silo\\- tliat the performance of the proposed
converter is enhanced with the awiliary circuit. -4bout 5% higher efficiency has been
achieved on the prototype circuit as compared with the conventional hard switching
converter. Higher efficiency is espected on PCB circuit with optimal rnagnetics. Other
major merits of the proposed ZVS fonirard topology include
(i) Lossless switching of the main switch. independent of the line and load conditions,
(ii) Simple power and control circuit,
(iii) No increase in conduction losses,
(iv) Ability to operate in either voltage or current mode control.
Therefore, the proposed converter topology wi l l liopefully be widely used in indusaial
application in the fùture.
SMALL SIGNAL MODELS AND CLOSED
Loop STABILITY
4.1 INTRODUCTION
The steady state analyses of the proposed converters have been performed in the
previous chapters, which have shown the advantages of the converters over the
conventional hard switching and existing ZVS circuits. However, only the steady state
analysis is not enough. The circuits should rolerate the line and Ioad variations, and they
should regulate the output against these variations by means of closed loop control. In
order to study the behavior of converters under these conditions, the small signai analyses
for both the converters are presented in this chapter. These analyses are also used to
optirnize the closed loop response.
This chapter presents the small signal models and the closed loop stability of the
converters. The small signal mode1 of a circuit differs with the loop structure, or the
control technique used. Normally, there are three types of control technique developed for
the dddc converters, narnely, the voltage mode control, the peak current mode control,
and the average current mode control. It has been s h o w in [1920] that the average
current mode control brings in the best dynamic performance, hence, it is discussed in
this chapter. As the flyback converter can utilize a model previously developed in [20],
only the srnall signal mode1 of the forward converter is derived below. During the
discussion, the equivalent senes resistance (ESR) of the capacitors is ignored because the
effects of ESR can be reduced with high quality capacitors and these effects is negligible
in the fiequency range of our concem (<IO0 W).
The structure of this chapter is as follows. Section 4.2 reviews the basic cnteria of
a stable closed Ioop system. Section 4.3 discusses the stabilization of the flyback
converter. Section 4.4 presents the denvation of the small signal model of the forward
converter, and discusses the stabilization issues. Section 4.5 presents the experimental
verifications of the model derived.
4.2 CRITERIA OF A DYNAMTCALLY STABLE SYSTEM
According to the Nyquist stability criteria, the closed loop stability of a switch
mode converter is optimal, if (i) the open loop gain has a slope of -20 dB/dec. in the
vicinity of the crossover frequencyf,, and (ii) the open loop transfer function has a phase
margin of about 45" to 63" [19-211. Proper compensation should be made around the loop
to meet these two criteria. Hence the loop should be tailored such that the above two
conditions are satisfred.
4.3 SMALL SIGNAL MODELS OF TIiE ZVS FLYBACK
CONVERTER
To simplify the discussion, it is assumed that the auxiliary circuit works for a very
short t h e compared to the main circuit, which is the practical case when UC3855 (shown
in Appendix 1) is employed as the PWM control chip. The duty cycle of the auxiliary
circuit Dm is about 0.07 to 0.1. Also, it is assumed that the converter operates in the
discontinuous conduction mode. Under such conditions. as analyzed in Section 2.4.2. the
effect of the auxiliary circuit on the transformer current is negligible, and hence it is
undetectable on the secondary side of the power transformer. In other words, it has litde
dynamic effect on the operation of the converter. Therefore, from the control point of
view, the auxiliary circuit cm be excluded from the closed loop. and importantly, the
analysis is that of a basic flyback converter.
Fig. 4.1 shows the block diagram of a conventional feedback control loops of the
standard flyback in the average current mode control. Fig. 4.2 shows a typical
implementation of the feedback loops of Fig. 4.1 [20].. The current loop is formed by the
following elements: (i) a current sensor resistor R,,? (ii) an error current amplifier CA, (iii)
a PWM comparator and the gate drive. and (iv) the power switch S. Current in the switch
is sensed. Because the control chip has a current synthesizer inside UC3855, the sensed
current is transformed to reconstnict the inductor current in L,,,.
Fig. 4.1 The Block diagram of the control loops.
Fig. 4.2 A typical implementation of the feedback loops of Fig. 4.1. The elernents enclosed by the dot-dash line are inside the UC3855 chip.
Through the current synthesizer, R,, reflects the actual current signal in Lm.
The voltage loop is compnsed of (i) the closed current loop, (ii) the voltage sensor
(R4 and R5), (iii) an error voltage amplifier VA, and (iv) the power circuit. The
compensation elements are selected below.
4.3.1 Small Signal Mode1 of the Current Loop of the Flyback
Converter
(1) Transfer function of the power circuit
From the previous work [20], the small-signal control-to-input gain of the flyback
current loop power circuit of Fig. 4.1 (fiom v,, at the CA output, to v,, the voltage across
RJ is given by
where, IL is the averaged inductor current in L,,,.
Incorporating this mode1 into the isolated flyback topology shown in Fig. 1.1-
where the tums ratio of the power transformer is Un, and that of the current sensor
transformer is K. the gain becomes
where F,(s) is the current sensor transfer function given by
Eq. (4-2) has a pole at O Hz and a zero at
This zero moves with IL, which corresponds to certain line and load conditions. In order
to maintain the stability under al1 line/load conditions, the crossover fiequency of the
current loop gain should be selected lower than the lowest zero given by Eq. (4.4).
Fig. 4.3 shows the bode plot of the current loop of the prototype flyback converter
(refer to the design example in Chapter j), where K=1/100, n=1/18. V0=5 V, Vw=5.5 V,
Rs=30 Q, and L,=50 pH, a n d m 0 0 kHz. In th is case it is found that the lowest value of
L, which corresponds to the full load and lowest input voltage, is about 190 kHz.
f (Hz) ----,
(a) Gain vs. frequency
(b) Phase shift vs. fiequency
Fig. 4.3 The Bode Plot of the opened current loop of the prototype fly back converter.
(2) Compensation for the closed loop stabilization
Concepts of the compensation for the closed loop stabilization c m be illustrated
with the example of the prototype converter.
The crossover fiequencyL should not go higher than the lowest zero fiequency of
the power circuit transfer function. Also, it should be seIected below one sixth of the
switching frequency [20]. In the example, iff;=200 kHz andJ=l90 kHz, then is set at
The resdt of the compensation is shown in Fig. 4.3(a). and has two poles and one
zero to give a proper loop gain. The implernentation is shown in Fig. 4.2. Thus the
transfer function of CA is determined by
where the poles and zero are found to be
In order to suppress the high fiequency noise caused by switching. it requires
additional pole around f,z by adding the network Rp-Cp, which gives
The value of each component is determined as 1201
In the example of the prototype converter, by selecting R,=3.9 kQ it is found that
C,=100 pF and R, 4 . 7 kQ. By putting a zero at J,=6 kHz' there are C, =4700 pF and
42=1 00 kHz. By Ietting Cp=l 000 pF, it yields Rp= 1.5 kn.
The compensation is shown in Fig. 4 3 a ) . Fig. 4 3 b ) shows the Bode plot of the
total open loop phase shifi. The open loop phase rnargin atf; is found to be 61 .jO.
The equivalent transfer function of the closed current loop is
G,,,(s), is an element of the voltage loop.
4.3.2 Small Signal Mode1 of the Voltage Loop of the Flyback Converter
The stability criteria for the voltage loop is the same as that for the current loop.
The compensation is needed to tailor the total voitage loop gain to fit the stability criteria.
(1) Transfer function of the power circuit
For the flyback converter output stage shown in Fig. 4.1, the small signal mode1 is
By the power balance equation,
(1, + Mu)( t + A C ) = (',rj +hl,rr)('itj +AYlj) (4- 1 6 )
where hl,, A Vo, Mllj, and A y,# are the smdl signal perturbations in the related signals.
Ignoring the second order terms and assurning a constant V _ and applying the Laplace
transformation, Eq. (4-1 6) yields the smdl signal mode1 for the output stage:
I#,<(, +y,<) = ylj[,,
where
If,= PyRU
From Eqs.(l- 14) and (3- 16)- the transfer function of the output stage is
Then the power circuit transfer function. from the output of VA to the output end of the
converter. is
Fig. 3.4 shows the Bode plot of the voltage loop gain of the prototype flyback.
(2) Compensation for the closed loop stabilization
Refemng to Fig. 4.4(a). the compensation for the voItage loop c m be achieved by
a PI controller. The implementation is s h o w in Fig. 4.2. Thus the transfer function of the
VA compensation network and the voltage sensor is
A
R3 1 1 _+ G,, (s)& ( s ) = Y-- = ( V" R.: SR&
)
In the example of the prototype flyback converter, it is found that, by selecting the
crossover frequency at 10 ldIt and choosing R,=3 kR and R,=4.5 kR, the needed
compensation requires R,=4.9 WL and C3=7.8 nF. The phase margin is thus found to be
63 -2".
J!Hd -----+
(b) Phase shift vs. fiequency
Fig. 4.4 The Bode plot of the gain of open voltage Ioop of the prototype flyback converter.
4.4 SMALL SIGNAL MODEL OF THE ZVS FORWARD
CONVERTER
Fig. 4.5 shows the block diagram of the typical control loops of a forward
topology. where the inductor current i, is controlled.
Fig. 4.5 A typical block diagram of the control loops of forward converter.
Fig. 4.6 A typical implementation of the feedback loops of Fig. 4.5. The elements enclosed by the dot-dash line are inside the UC3855 chip.
Through the current synthesizer, R,v reflects the actual current signal in La.
Fis. 4.6 shows a typical implernentation of the feedback loops of Fig. 4.5 in the
case of the fonvard topology. The current Ioop is foxmed by the folIowing elements: (i) a
current sensor resistor R,, (ii) an error current amplifier CA, (iii) a P WM comparator and
the drive, and (iv) the power switch S.
Small Signal Modeis of the Current Loop of the Forward
Converter
Transfer function of the power circuit
Current in the switch is sensed. But, owing to the current reconstruction element,
is called the 'current synthesizer' in the PWVI chip UC3855, the current in Lt,.
narnely i,. is virtually sensed and hence controlled.
~ a t i n ~ f of main:
Sensed, current: - T,
by R.,I 4
Reconstructed ,* -
synthesizeq : I
Fig. 4.7 The reconstruction of the sensed cturent in the proposed ZVS fonvard converter. The sensed switch current shown in Fig. 3.2, is reconstructed by the
synthesizer inside UC3855 to reflects the waveshape of the output inductor current. D is the main duty cycle, De is the effective duty cycle, and D, is the difference between them.
Fig. 4.7 shows the current reconsiructing function in the proposed ZVS fonvard
converter. The sensed switch current shown in Fig. 3.2, is reconsmcted by the
synthesizer inside UC3855 to reflect the waveshape of the output inductor current. D is
the main duty cycle, De is the effective duty cycle, and D, is the difference between them.
Dr is due to the effects of Ls in Fig, 3.1 and is determined by Eq. (3-79). From the loop
point of view, such a reconstruction fünction simply hides the effects of the auxiliary
circuit, making it look like a conventional forward converter. The only exception here is
that the effectua1 duty cycle is no longer D, but is
0, = ( 0 - 0,) (4-22)
Therefore, the state space average equation of the converter is,
where, n = AT, !!V,,.
Then the DC mode1 of the converter is
which is confimed by Eq. (3-75). With assumption of constant y,l and y)_ the small
signal AC mode1 is represented as
According to Eqs.(3-25), (3-26)' (XZ), (3-34), (3-37) and (3-78), it can be obtained
Substituting Eq. (3- 17) into (4-26). and substituting L+dl,,, y,,+A yll and D,+AQ for their
respective variables, and ignoring the second order tems, gives
Hence,
u Dx + 0.5% AD, =-AIo - IZ C l Y t I
A Y I 1
By the assurnption AIo = and AP:tt = 0,
In the Laplace transformation, Eq. (1-29) can be written as:
Substituting it into Eq.(4-25), and rearranging the terms!
Then the sensed voltage on the sensor resistor, on the secondary of the sensor
transformer. is
CR, = nKR,iLO (4-32)
where K is the t u s ratio of the current sensor transformer? R, the sensor resistor.
Let Vcw represent the peak to peak voltage of the saw tooth waveform, then
where v,, is the output of the current amplifier of the current loop.
Therefore, f?om Eqs. (4-31) through (4-33), the control-to-output gain of the forward
current loop power circuit (fiom v,, at CA output, to v,, the voltage across the sensor
Fig. 4.8 shows the Bode plot of the open loop gain of the current loop of the
prototype forward converter (refer to the design example in Chapter 5 and Appendix II),
where ~ 1 / 3 , L0=6 pH, L,=O.3 pH, f;=300 kHz, Rs=12.5 R, K=1/100, yn=50 V and
V,=5.5 V. Then the pole of the power circuit is found to be at 2.6 kHz.
(2) Compensation for the closed loop stabilization
The needed compensation network is similar to that in the flyback circuit, i.e.. two
pole and one zero compensation. The implementation of the compensation is shown in
Fig. 4.6. Similar process to that in the case of flyback topology is followed.
Eqs. (4-5) through (4-9) define the transfer function of the compensator and the
poles and zero. Eqs. (4-1 1) and (4-12) determine the values of R, - and C,. C, is determined
fiom Eq. (4-7) by setting the second pole. 4,. In the example of the prototype fonvard
converter. seIectin_gf,=30 kHz R,=3.3 KS2.f.=8.0 kJ3z andf,,=l00 kHz. it is found that,
R,=33 - kQ, C, =56 pF and C,=560 pF.
Fig. 4.8 (b) shows the Bode plot of the open loop phase shifi of the current loop of
the prototype forward converter. The phase rnargin is about 61 -5'.
The equivalent transfer fùnction of the closed current loop has the same definition
as in Eq. (4-14).
4.4.2 Voltage Loop of the Forward Converter
(1) The characteristic of the power circuit gain
The closed inner Ioop is an element of the voltage loop. Refening to Fig. 4.6, the
output stage of the circuit has a transfer function as given by
Then the power
tram fer h c t i o n
f (Hz) - (a) Gain vs. frequency
f (Hz)-
(b) Phase shifi vs. fiequency
converter? has a
Fig. 4.9 The Bode Plot of the open voltage loop gain and phase-shifi of the prototype forward converter.
Fig. 4.9 shows the Bode plot of the open voltage loop of the prototype forward
converter.
(2) Compensation for the closed loop stabilization
With reference to Fig. 4.9(a), the compensation can be performed by a PI
controller. The implementation is shown in Fig. 4.6. The transfer function of the VA
compensation network and the voltage sensor is defmed by Eq. (4-21). In the case of the
prototype forward converter. if selecting the crossover fiequency around 10 WIz R,=3
ki2 and R,=4.5 kS2. the components for the PI are found to be R 9 . 3 kQ and Cj=6.8 nF.
The open voltage loop gain is show in Fig. 4.1 1. The phase rnargin is 64.8'.
4.5 Experimental Results
Experimental verification of the srnall signal models discussed above is made on
the prototype forward converter. For convenience. the transfer function of the voltage
loop power circuit. G,,s(s)t is tested.
Figs. 4.10(a) and 4.1 1 (a) show the experimental results of the Bode plots of
G,,.(s). Figs. 4.10(b) and 4.1 1 (b) show the theoretical predictions of G,,,,.(s) in the same
scale as Fig. 4.10(a).
Table 4.1 The theoretical and experimental results of the power circuit transfer function.
results Experimental
results
Theoretical
phase at 100 kHz beyond
phase at 10 kHz
-89.6 deg.
phase in low freq. range
O deg.
gain in low freq. range
6.0 dB
5 .O dB
gain at 100 kHz -37.7 dB
gain at 10 kHz 3.2 dB
O deg 3.0 dB -98.0 deg 1 -180deg
-43.0 dB beyond -180 deg
>.: I/R<dB) 0: 8 O MKR 58 084.366 Hz A MAX 63.00 dB GAIbl -34.8354 dE
OdB
A 3 I N -60.00 dB ÇTART 10.000 Hz w n r i f 50.00 dea STOP 100 OOO. CG0 Hz - - -
STOP= 1000t30.QXl q Z (a) Experimental results. Vertical: 12 dB/div.
frequency - (b) Theoretical results. Vertical: 20 dB/div.
Fig. 4.10 Comparison between the experimental and theoretical gain of the open voltage loop power circuit.
Comparison between theoretical and experimental results are shown in Table 4.1,
which shows that they are in a good agreement.
-O deg.
h ? A I N -60. 00 . d 3 START 1 0 - 000 HZ B r ~ ; I ~ K - 1 eo. o d+g STOP 100 ooo, occ Hz .=S€P>LE= L 1 1 HGP\BC
(a) Experimental result. The theoretical prediction is s h o w in @). Vertical: 36 degdiv.
(b) Theoretical prediction.
Fig. 4.11 Cornparison between the phase shift of the open voltage loop power circuit transfer fiction.
Figs. 4.12 through 4.14 shows the experimental results of transient responses of
the fonvard circuit under step force. The circuit is stable under these step changes.
Experiment of the step up of the input voltage is not able to cary out for the difficulty in
getting a f a t step change in the dc line with available instruments.
Fig. 4.12 Experimental results of the fonvard converter: transient response. Io steppes up fiom 8A to 20A. Conditions: V,=50V,f;=300kHz.
Vertical--SA/div., 1 Vldiv.. Horizontal--50pldiv.
Fig. 4.13 Experirnental results of the forward converter: transient response. Io steppes down fiom 20A to 8A. Conditions: y.,r=50V,f;=300kHz.
Vertical-SNdiv., 1 V/div., Horizontal--5Ops/div.
Fig. 4.14 Experimental results of the forward converter: transient response. y,, steppes down fiom 60V to 40A. Conditions: PO=l 00Wof;=300kHz.
Vertical--5Ndiv.. O.SV/div.. Horizontal-- jOOps/div.
4.6 CONCLUSIONS
Smail signal models of the proposed circuits in the average current mode conuol
and the closed loop stability compensation have been discussed in this chapter. The
proposed flyback converter can utilized the previously established model for the standard
flyback and the compensation c m follow the conventional scheme and process.
As there is no suitable model for the proposed forward converter, the model
thus derived in this chapter. The additional inductor L, of the auxiliary circuit bnngs
pole at the fiequency in proportional to the switching frequency
the are simple and straightforward.
and Ls. Compensation of
This chapter presents the design procedures of the proposed converters. The
design is based on the required specifications, such as the input voltage range. output
voltage, output power. switching frequency, permitted ripple in the output voltage. etc..
and uses the analysis presented in the previous chapters.
Design for generating gating pattern for the auxiliary switch in the proposed
converter topologies is also presented in the case where a single output PWM control chip
other than UC3855 is employed.
The structure of this chapter is as foIlows. In Section 5.2, the design of the flyback
converter is presented and an example is given. In Section 5.3, the design of the forward
converter and an example are presented. In Section 5.4, an implementation of the gating
pattern generating for the auxiliary switch in the proposed converters is illustrate and
verified expenmentally.
5.2 DESIGN PROCEDURE OF THE FLYBACK CONVERTER
Assume the following specifications and parameters are known.
Dm,--Maximum duty cycle
f;--Switching f'kequency, Hz
P;-Full load output power, W
ytt ,--Maximum input voltage, V
y,, ,in--Minimum input voltage, V
Va--Output voltage. V
ATl,--Perrnitted peak-peak npple in y,,' V
Ay;-Perrnitted peak-peak ripple in V,,. V
5.2.1 Design of the Power Circuit
The following design procedures are referred to the circuit s h o w in Fig. 2.1. A11
parameters are in SI units.
(1) Turns ratio of the transformer, NdlV,
According to Eq. (2-56), and considering the forward voltage drop V,.. of the
rectifier diode,
where, D'(cl -Dm,) is the equivalent duty cycle of the rectifier current, and
VF is the fonvard voltage drop of the rectifier.
Eq. (5- 1) gives
and
(2) Magnetizing inductance, Lm
In the discontinuous conduction mode. fiom Eq. (2-40). there should be
Combining Eqs. (5-4) and (2-39) gives
(3) Output capacitor, Co
Fig. 5.1 shows the relationship between the output rectifier current i and load
current 5 in the case of discontinuous conduction mode operation. Assurning al1 the ripple
current component of i, flows through $, and the average current flows through R,,. It can
be calculated that
Fig. 5.1, The secondary current and the output voltage ripple in the converter of Fig. 2.1. The operation mode is assumed to be discontinuous conduction.
As IV is deterrnined by Eq. (2- 39), and
and
AV, 2 AQ/C,
The output capacitor can be found fkom Eqs. (2-39) and (5-6) tbrough (5-8):
(4) Selection of the main switch, Ql
By Eq. (2-33). the voltage rating of Q l is determined by
where: V,, is the drain to source breakdown voltage rating of QI.
The rrns current is determined by Eq. (2-60). and the average current rating is detemined
b Y
where, the constant 1.5 accounts for the worst case efficiency of 70%, and
If, ,,, is the continuous drain current rating.
From Eqs. (2-26) and (2-55), the peak pulse drain current rating is
where. r,, Q, is the pulse drain current rating.
102
Besides these ratings, a MOSFET which has the lowen ON resistance on the
product catalogue should be selected.
(5) Selection of the output rectifier, Do
The rectifier should be fast rectifier diode or Schottky diode. Its voltage rating is
determined by
v~-L!ri > v~r-rnax + (5-13)
where. VR , is the reverse voltage rating.
Its current rating is determined by
where. Io , is the average rectified current rating.
(6) Selection of the input capacitor, Ci,
The input capacitor is determined by [Z]:
where, A y l l is the allowable peak-to-peak ripple voltage in the input line. and
AI.,_ is the time for which Cl,, must supply the current in case the input line fails.
5.2.2 Design of the Auxiliary Circuit
(1) Auxiliary duty cycle,
Dm should be limited by
Dm (1 - 2 DmJ (5- 16)
In this way. the auxiliary circuit does not affect the resetting of the core of T,
Dm may also be detennined by the control chip like UC3855, where it is fixed. In
this case the design should observe the fixed factor.
(2) Snubber capacitor, C,,
The value of Cm, determines the rise time of QI's voltage at its tumoff. as
described in Eq. (2-28). For a very short duration, we c m approximate
Limit the nse of u,,, to be below yj1 within the demanded time tr. Thus, fiom Eqs. (2-29)
and (5-1 7),
On the other hand. the rise time should not exceed the gap left by 2DlIlm and DaUr. which
eives a Iimit C
From Eqs. (2-28) and (2-33). there is an approximation
Combining Eqs. (2-28). (5-19) and (5-20),
The voltage rating of the capacitor should not be less than that of Q I .
(3) Primary of the coupled inductor, Lw
As analyzed in Section 2.4.1, Dom is fixed and corresponds to a quater of the
resonant penod of the network consisting of L,, and Cd. Referring to Fig. 2.2, there exists
a relationship:
Dm = ( t z - t i ) / ~ ,
According to Eqs. (2-10) and (5-22),
which gives
The core of the coupled inducton shouId be selected so that it can handle a power
as given by
1 P. = 7 LI i&f, -
where. iop is defined by Eq. (2-1 6).
(4) Selection of the auxiliary switch, Q2
The on-resistance of Q2 should be as srna11 as possible. Its voltage rating is the
same as that of Ql. It should have a current rating capable of handling the peak
discharging current I,, and an RMS current I,,, Q2 as given by Eq. (2-6 1 ).
(5) Secondary of the coupled inductor, L,
According to Section 2.4.3, L, should be as big as possible, so as to rninimize the
voltage stress(refer to Eq. (2-25)) and hence the -off losses in PZ. But a bigger La,, will
also extend the energy releasing time.
From Eqs. (2-18) and (2-24), and with reference to Fig. 2.2, the energy releasing
time is determined by
In order to release the stored energy in the core completely within switching cycle. there
should be a limitation,
&a - &2 < (1 - DOA TS
n u s , fiom Eqs. (5-26) and (5-27), it is determined that
( 6 ) Selection of the blocking diodes, D, and D,
Both the diodes are fast recovery diodes with low voltage drop. D, is optional. and
is only applicable to cases where the releasing current of L, may reach zero before Ql is
turned OFF. or
r , - r 2 = - J E > D ~ ~ ~ T . , K t ,
Without LI,, the charged output capacitor of Q2 will discharge into Ql after the tum-off
of Q2, increasing the losses. Therefore DI is only dispensable in the design where Eq. (5-
29) is valid.
DI should be selected so that it is able to handle a peak current of IOP. As it is in
parallel with Q2 when Ql is ON and QZ is OFF, thus, fiom Eq. (2-25), its voltage stress
i s
Similarly, D, should be able to handle a peak cuITent of 1, given in Eq. (2-19). Its
voltage stress is
5.2.3 A Design Example
A prototype flyback converter works under the following conditions:
(i) Input voltage: ytl ,,=16O V. and i:,) ,in=90V - -
(ii) Outputs: two outputs, (a) Vo1=5 V, P,,=10 W, and (b) V O- ,=20V. <,=20W. Total: 5OW
(iii) Permitted ripples in the outputs: A Vo,=50 rnV (Le., 1 %) , A VO2=200 mV (Le., 1 %)
(iv) Maximum duty cycle: Dm_=0.4
(v) Switching fiequency : f;=200kHz
(vi) Auxiiiary duty cycle: DOlLr=0.075 (fixed by UC3855)
(vii) Input capacitor voltage sustaining tirne: Af,T_=lOO/j;
Table 5.1 The parameters of the prototype flyback converter
parameter
I
L n , Co, Cd
c,!, Do, DO?
selection 1/18 1 /5
50 p H 220 pF 100 pF 200 pF MBR735 MBR8 10
ratings
35V, 7SA 1 OOV, 8A
parameter
Qf
C.Wh
Lm Lm 0 2
Dp D2 Controller
selection IRFP450 3.3 nF IS p H 90 p H IRF730
HFAOSTB UC3855
ratings 500V, 14A,0.4R
400V, 5.5A. 1 .OC2 600V, 8A
Table 5.1 shows the design and selection of each component and device of the
prototype converter, which is made by following the design procedure presented above.
5.3 DESIGN PROCEDURE OF THE FORWARD CONVERTER
The design of the forward converter is also based on the specifications of the
application. Assume the following specifications and parameters are known.
D, ,,--Effective maximum d u 6 cycle
f;--Switching frequency. Hz
P,J--Full load output power? W
A y,--permitted peak-peak ripples in cl, V
y,, max--Mvia><imurn input voltage. V
P;,! ,i,,--Minimum input voltage, V
PL--Output voltage. V
Ay;-permitted peak-peak ripples in V, V
5.3.1 Design of the Power Circuit
The following design procedures are referred to the circuit shown in Fig. 2.1. Al1
parameters are in SI units.
(1) Turns ratios of the transformer, N p s , N p i
From Eq. (3-75)
where, De ,, is the effective duty cycle of the main switch, Le., De ,, =D-D,.
Taking into account the forward drop of the rectifier, it is modified as
When De ,,<OS, usually
P= * 1 NP
If Dem_>0.5, to guarantee successfûl core resetung, the turns ratio shouid be
(2) Output inductor, L,
Assume the converter is at the boundary of the continuous conduction mode
operation when its output power is a twentieth of Pf1. Hence there is [23]
which gives
(3) Output capacitor, Co
The ripple voltage can be found as given by [23]
17
Hence, Co should be chosen as
(4) Magnetizing inductance, Lm
Assume the magnetizing current accounts for Iess than one twentieth of the peak
input current. that is,
Because,
Hence,
(5) Selection of the main switch, QI
The switch should be selected so that it has the lowest ON resistance among the
product catalogue. The voltage rating is determined by Eq. (3-5 1 ) as
['D.S.S_~I L'IN
where. VDSs (,, is the drain to source breakdown voltage rating of Q I .
The rms current is determined by Eq. (3-85): and the average and peali current ratings are
determined by
Po I , i ~ - ~ i > 15
Vin-min Dmax
where, the constant 1.5 accounts for the worst case efficiency of 70%,
IFei is the continuous drain current rating, and
{,, ,,, is the pulse drain current rating.
(6) Selection of the output rectifier, Do, and Do,
The rectifiers should be fast rectifier diodes, Schottky diodes or synchronous
rectifier MOSFETs. Their voltage rating is detennined by
F R - ~ o > n Vrrr-ma.
where V,-, is the reverse voltage rating .
Its current rating is determined by
where, I, ,], and I,, ,,,, are the average current rating of Do, and Do,- respectively.
(7) Selection of the input capacitor, $,
The selection criterion for C,,, is the same as Eq. (5-1 5).
5.3.2 Design of the Auxiliary Circuit
(1) Auxilia- du@ cycle, Du,',
The maximum Da, should be limited by Eq. (5-16) so as not to affect the resetting
of the core of T,. It should be noted that, in the case of the Forward converter topology?
Dm, corresponds to half the resonant period of the network consisting of Lp. C,,,Ii and T,
(2) Snubber capacitor, C,,
The value of C-,,,, determines the rise time of the drain voltage of Ql at its turnoff.
as described in Eq. (3-46). For a very short duration, it is approximated to
Limit the nse of u,, below P;,) within the demanded
On the other hand. the rise time should not exceed
gives a limit
time tr. Thus, fiom Eq. (Mg),
(5-50)
the gap lefi by 2D and Dar, which
t s - t ~ 5 (1-2Dmm - Daw) T.v
Combining Eqs. (5-49) and (5-51).
11 I o Cwh 5 -(l - 2 Dmzu - Dam) Ts
2 v ,l,
(3) Primary of the coupled inductor, L,
From Eqs. (3-3) and (3-16).
Because D, corresponds to a half of the resonant period, Le..
Tt gives
The core of the coupled inductors should be selected so that it is able to handle the power
given by Eq. (5-25), where Iap is determined by Eq. (3-25).
(4) Additional inductor, L-v
From Eq. (3-1 7), it c m be determined that
7
L, = n- L,
Ls can be integrated into the secondary windings of the power transformer.
(5) Selection of the auxiliary switch, QZ
The on-resistance and inherent capacitances of Q2 shouid be low in the product
catalogue. Its voltage rating is the same as of Q I . The current rating should be able to
handle the peak discharging current Iap given by Eq. (3-17), and an RMS current I,,, Q,
given by Eq. (3-87): and an average current given by Eq. (3-88). C
(6) Secondary of the coupled inductor, L,
From Eq. (2-37). Lm should be large in value so as to reduce the turnoff voltage
stress on 03. However: Eq. (5-27) should aiso be satisfisd so as to release the energy
stored in the core of the coupled inductor within one switching cycle. From Eqs. (3-22):
(3-25), (3-3 1 ) and (5-27)-
(7) Selection of the blocking diodes, D, & D,
D, and D, should be able to handle a peak current of l a p and L, repectively.
Their voltage ratings should be higher than their voltage stress given by Eqs. (2-23) and
( H Z ) , respectively.
5.3.3 A Design Example
The specifications of the prototype forward converter is as follows.
(i) Input voltage: - ,,75 V, and y?, _,,,=35V -
(ii) Output: Vo=5 V, Po=lOO W
(iii) Permitted ripple in the output: AV0=50 mV (i.e., 1%)
(iv) Maximum duty cycle (effective): De - ,,,=0.4
(v) Switching fiequency :f;=3OOkHz
(vi) Auxiliary duty cycle: Dmr=0.075 (fixed by UC3 855)
(vii) Input capacitor voltage sustainkg tirne: A<ux=lOO/f;
Table 5.2 shows the design and selection of each component and device of the
prototype converter. which is made by following the design procedure presented above.
The design of the synchronous rectification stage is based on the design procedure
presented in [ 1 81.
Table 5.2 The parameters of the prototype fonvard converter
parameter selection n 1 /3
* Two in parailel in order t c
ratings
50V,75A, 9.5mQ
200V, 18A, o. 18Q
reduce the conduction losses in the switch.
parameter
c.v,rtj
Ln,7
Lm 4 0-3
Dl Br
Controller
selection IO nF
3 pH 90 p H 330 nH RF640
HFA08TB
UC3 8%
ratings
200V. 18-4: 0.18Q
600V. 8-4
5.4 GATING GENERATION FOR T E E AUXILIARY SWITCH
UC3855 can be directly employed as the PWM control chip of the proposed
converters. It produces two gating signais to drive the main and auxiliary switches. The
auxiliary gating signal has a fixed duty cycle of about 0.075. The major features of
UC3855 include (i) average current mode control, allowing better stability and easier
contml loop design? and (ii) internai current synthesizer, simpli&ing the current sensing
method (see Appendis 1).
A problem in using UC3855 is that the auxiliary duty cycle is not adjustabie. As a
result. the auxiliary duty cycle may not be the optimal one' which will result in the least
losses in the auxiliary circuit. Another problem is that UC3855 is expensive as compared
to most of the single output PWM chips.
#en a PWM chip of single output is selected as the control chip. gating for the
auxiliary switch can be generated with an additional sub-circuit as presented below.
5.4.1 Gating pattern generation by additional logic and drive circuit
Fig. 5.2 shows an example of the simple implementation of the auxiliary gating
pattern. It consists of a Schmitt trigger. a Op-Amp, inverters and AND gates. Other logic
combinations exist but the function it fulfills is the sarne: to split the original PWM pulse
from the single output PWM chip into two pulses: one for the auxiliary switch and the
other for the main switch. The additional logic and drive circuits form a splitter.
Fig. 5.3 shows key waveforms at the indicated points in Fig. 5.2. The auxiliary
duty cycle is adjusted by varying the time constant of the RB-CB network of Fig. 5.2.
Fig. 5.2, The pating generation circuit, a splitter, which generates the auxiIiary gating on basis of the original PWM signal
time
Fig. 5.3, The operationa1 waveforms in the splitter: voltage waveform at the corresponding point in Fig. 5.2
In Fig. 5.3, V, and V, appear to have the sarne wave shape. The AND gate
between point E and seemingly unnecessary, is of practical importance. This is
because d l the gates have delays in their response. In order to synchronize the two
gatings to an accurate timing sequence, a symrnetrical structure in the splitter is required,
which results in the least difference in delay for the two output gating signals.
5.4.2 The design of the splitter
(1) Selection of R, C,j and D,
The duty cycle of the auxiliary switch, or the pulse width of the auxiliary gating
signal. c m be set by the proper value of R, and C, in the spiitter in Fig. 5.2. The
following equation should be satisfied
where c,,,, is the high voltage level from the PWM chip output.
Eq. (5-58) yields
The diode D, is introduced to follow the fast &op of PWM signals. It should be a
fast diode.
(2) The voltage divider, RA, and R,,
Because V,,, is usually about 10- 1 5V, while the high output voltage V,, (logical
level) is usually 6V. a direct connection of IT,,l,, to the logic circuit may destroy it.
Therefore, the linear voltage amplifier is inserted between the PWM output and the logic
circuit. The magnitude of the amplification is less than one, and should obey the relation:
An additional capacitor CA is introduced to balance the stray capacitors of RA, and
R A , making the amplifier a purely linear device. Its value is decided experimentdly.
(3) The seleetion of the logic gates and the driver
All the logic gates should be capable of high speed response. CMOS logic circuits
are suitable for the splitter. High speed dnvers should be selected.
5.4.3 A Design Example
Table 5.3 shows the parameters of the spliner of Fig. 5.2 in a design example.
UC2844 is used as the PU'M control chip. It has only one gate drive. The splitter is thus
ernployed to generate the gate drive for the auxiliary switch. Fie. 5.4 shows the
experimental results of the gating signalsl which verïQ the design.
I C H I . SV CH2 . SV
Fig. 5.4 The expenmental results of the gating patterns generated by the splitter of Fig. 5.2.
channel 1 : original PWM signal fiom UC2844 channel2: gating for auxiliary switch Q.2 charme13 gating for main switch Q I
Table 5.3 The parameters of the splitter in a design example
' * n i e Schmik trigger and the Op-amp are subçtituted with 1
parameter
RA L RE CA
RB CR
inverters in the example.
180 kC2 parameter 1 DB IN4 148
TC74HC04P MC74HCO8A
IR21 10
91 ki2 1 Inverters* 0.22 pF 1 ~ ~ ~ g a t e 2.2 ki2 500 nF
Drivers
This thesis has discussed zero voltage switching (ZVS) flyback and forward
converter topologies. Such a topic is of special importance. because both the topologies
are very popular in industrial applications. especially in the distributed power systems of
the advanced communication and computer ne tworks.
This thesis can be summarized as follows. In chapter 1. existing ZVS flyback and
fomrard converter topologies are bnefly reviewed and their drawbacks are sumrnarized.
These topologies lose ZVS at light load. require complicated gate drive schemes. achieve
ZVS at the cost of increasing the conduction losses, have restrictions on use due to
patents.
In Chapters 2 and 3, ZVS flyback and forward converter topologies which employ
an auxiliary circuit have been presented. The modes of operation of each circuit is
identified. During each switching cycle, the flyback converter operates in five distinct
intervals and the forward converter operates in seven intervals. By selecthg the values of
the snubber capacitor and the coupled inductors, ZVS cari be always achieved. Steady
state anaIysis is perfomed. and the factors influence the performance of the circuit is
investigated. Experiments on the prototype converters are carried out and the analysis and
design are verified.
In Chapter 4, the smdl signal models and closed loop stability are discussed.
Transients responses are obtained experirnentally. Finally, in Chapter 5. design
procedures of both converters and of the gating pattern generation for the auxiliary switch
are presented and design exarnples are given.
6.2 CONCLUSIONS AND CONTRIBUTIONS
The follouing conclusions can be made.
(i) The auxi1ia.q circuit is simple, and o d y a few lo~v power rating components and
devices are required.
(ii) The gate drive scheme for the auxiliary suitch is simple. No isolated. variable
duty cycle gate drive is required by the auxiliary switch.
(iii) ZVS can be always achieved whatever line/load conditions, and no conduction
Zosses are increased.
(iv) The prototype flyback converter has about 7% higher efficiency than its hard
switching counterpart. The efficiency is rneasured under the following conditions:
switching fiequency is 200 kHz, input dc line voltage range is fiom 90 V to 160 V. two
outputs, one of which is 10 W at 5 V and the other is 40 W at 20V.
(v) The prototype forward converter has about 5% higher efficiency than its hard
switching counterpart. The efficiency is measured under the following conditions:
switching fiequency is 300 1<Hz, input dc line voltage range is fiom 40V to 60VI the load
power is 100 W at 5 V, synchronous rectifiers are employed.
(vi) The small s ipa l analysis show that the flyback converter can utilize the srnaIl
signal model previously developed for the conventional flyback. The forward converter
has a pole proportional to the switching fiequency and the additional inductance inserted
into the secondary circuit. The closed loop design based on the denved model has a good
dynarnic response.
(vii) The disadvantage of the auxiliary circuit is that it has hard switching at tum-off.
though the losses associated with it c m be reduced by increasing the ratio of the two
coupled inductors.
The major contributions of this thesis include:
(i) Systematically analyses of the proposed converters are performed,
(ii) Design procedures are developed.
(iii) Prototype converters are built and experiments are carrïed out and the analysis and
design are verified experimentally,
(vi) Higher efficiencies have been obtained.
6.3 SUGGESTIONS FOR FUTURE WORK
The following suggestions c m be made for future work.
(i) The resuits obtained in this thesis have shown the advantage of the proposed
converters. The prototype converters on printed circuit board (PCB) should be built and
use optimal magnetics to M e r optimize the performance.
(ii) To further kprove the efficiency, new synchronous rectification techniques with
resonant gate drives rnust be investigated in the proposed circuit topology.
Mohan, N., Undeland, T.M.. Ro bbins, W.P., Power Electronics: Converter,
Applicationsy and Desi@, John WiIey & Sons? 1989, pp. 1 85- 1 86.
Carsten, B., "Design techniques for transformer active reset circuits at high
fiequencies and power levels," High Frequency Power Conversion Proceedings.
1990, pp. 235-246.
Harada. K., Sakamoto, H., "Switched snubber for high frequency switching," ZEEE
PESC '90 Record. Vol.1, pp. 18 1-1 88.
Jitam, 1.' "Zero voltage PWM. double ended converter, " Higi~ Frequency Power
Conversion Proceedings. 1997, pp. 394-404.
Tsai. F., Ng? W., '' A low cost. low-loss active voltage-clamp circuit for interleaved
single-ended Fonvard P WM converter," IEEE APEC '93 Record, pp. 729-73 3.
Watson? R., Lee, F.C., Hua, G.C., "Utilization of an active-clamp circuit to achieve
sofi switching in Flyback converters', IEEE PESC ' 91 Record, pp. 909-916.
Harada K., Sakamoto, H., "On the saturable inductor commutation for zero-voltage
switching," pp. 1 89- 1 96
Kim, H. J., Leu, C. S., Fanington, R., Lee, F. C., "Clamp mode zero-voltage-
switched multi-resonant converters," IEEE PESC '92 Record, pp. 78-84.
Tang, W, Tabisz, W., Lofii, A.? Lee, F. C., Vorpenan, V., '?DC analysis and design
of Forward zero-voltage-switched multi-resonant converter,'? IEEE PESC '90
Record, pp. 333-340.
[l O] Andreycak, B., "Active clamp and reset technique enhances Fonvard converter
performance," Unitrode Design Seminar '91, Section 3, pp. 1-1 8.
[ I l ] Zaity T., Ninomiya, T., Shoyarna, M., Tanaka, H., "P WM-controlled current-mode
resonant converter using an active-clamp technique,'' IEEE PESC '96 Record, pp.
89-93.
[12] Joug., G. B., "New sofi switched PWM converter," IEEE PESC '96 Rem- pp.
63-68.
[13] Hamada. S.? Mii- T., Hiraki, E., Nakaoka. M., 3anirable reactor & lossless
capacitor-assisted sofi-switching asymmetrical PWM DC-DC converter with
Forward-Flyback transformer link." IEEE PESC '96 Record, pp. 100- 105.
[14] Steigenvald, R. L., "A review of sofi-switching techniques in high performance DC
power supplies." IEEE IECOAr '95 Proceedings. pp. 1-7.
[1 51 Pressman_ A. 1.. Switching Poiver Supply Design. McGraw-Hill. 199 1, pp. 42 1-427.
[16] Domb, M., Redl, R., Sokal, N., "Non-dissipative tum off snubber alleviates
swtiching power dissipation, secon breakdown stress, and Vm overshoot: analysis.
design procedure and experimental verification," IEEE PESC '82 Recod pp. 445-
454.
[17] Ninomiya, T., Tanaka, T., Harada, K., "Optimum design of non-dissipative snubber
by evaluation of transistor's switching loss, surge voltage, and surge current," IEEE
PESC '85, pp. 283-290.
El81 Xi, Y., Jain, P. K., Joos, G., "An irnproved gating technique for the synchronous
rectifier MOSFETs in the fomard converter topology?" presented on IEEE
CCECE797.
[ 193 Dixon, L. H., "Control loop cookbook," Unitrode Seminar SEM4 100, 1 994.
[20] Dizon, L. H., "Average current mode controI swtiching power supplies," Uniîrode
Seminar SEM-700, 1990.
[2 11 Kassakian, I. G., Schlecht, M. F., Verghese, G. C., Principles of Power
Electronics, Addison-Wesley, 1 99 1. pp. 368-379.
[22] Chryssis. G., High Frequency Switching P o w r Supplies: theoiy and design. 2nd
ed., McGraw-Hill, 1989. pp. 8-9.
[23] Mohan. N.' Undeland. T.M.. Robbins. W.P.. Power Electronics: Converter,
Applications, and Design. John Wiley & Sons. 1989. pp. 69-74.
THE CONTROL CHlP UC3855
- - INI.ORAT100
k , CIRCUIT.
UNITRODE
High Performance Power Factor Preregulator FEATURES
Coritrds Boost PWM ta Near Unity Power Factor
Fixed ~ r e ~ & n q Average Cunent Mode Control Minimizes Lin8 Current Distorù'on
Buitt-in Actiw3 Snubber (ZVT) âllows Operaiion to SOOkHz. improved EMI and Etiiciency
lnductor Current Synthesizer allows Single Cunent Transformer Current Sense for Impmved Efficiency and Noise m g i n
Acatrate Analog Multiplier with Lin9 Cornpensaor allows for Univerd Input Vdtage Operation
High Bandwidth (~MWZ), Low Offset Current Amplifier
Overvdtage and Overcunent protecu'on
Two M O Threshold Options
150pA Startup Suppiy Current Typical
Precision t % 7.5V Reference
PREUMINARY
DESCRIPTION The UCt 8 5 W provides al1 the control features neçeççary for high power. high frequency PFC bost converters. The average cwrent mode canuol method aflows for stable. low distortion AC Ilne cunerit programrning without the need for slope comperrsation. In addiion. the UC1855 utilizes an active snubbing or M (Zero Vottage Tanri- tion technique) to drarnaticaliy reduce diode reaovery and MOÇFET tum-on losses, resulting in lower €MI emissions and h i g M effidency. Boost converter mvitching frequencies up to 500Wz am m w realiz- able. requiring only an additional SrnaIl MOSFEfI diode, and induclor to resonantly soft switch tbe b o s t diode and switcti. Amrage anent sensing can be employed using a sirnpie resistive shunt œ a curent sense bar&xm8f. Using the cumnt sense transfomer rnelhod, the ifi temal current synthesizer circuit buffers the indudor currenî during the switch on-tirne. and reconstructç the inducbDr arment during the switch off-time. lrnproved signal to noise ratio and negligibie airrent serrsing lbsses make this an attradRR solution for higher power applications.
The UC1855CVB aiso features a single quadrant multiplier, squarer, and divider circuit which provides the programrning signai for the cur- rent lmp. The interna1 multiplier current limit reduces output power during low line conditions. An overvoltage protedion arcuit disables both controller outputs in ttte event of a boost output OV condition.
Low startup supply current. U M O with hysteresis, a 1% 7.5V refer- ence. voltage amplifier with softstart, input supply valtage cîamp, en- able comparator. and overcurrent comparator m p l e t e the Iiçt of features. Àvaitabte packages include: 20 pin N, DW, a, JI and L
BLOCK DlAGRAM
LIcenso Parent tom P l ~ n i a r Mdgnetlca Pin n u m k s reibr 10 DIL-20 J or N packages.
APPENDIX II
SCHEMATICS OF THE PROTOTYPE CONVERTERS
VCC
ION
VREF
VSENSE GND
Vir
20 CA t
33k t- 11560pF I CAO
zvs GTDRIVE
II 56pF
Y
in
3 3k VAOüT 15 4.3k 7
Vsn
3 I ZVDRIVE 17
CT - F --Cr --
1 1 0 0 0 ~ ~ - IN4148 - 1 N393 4 13 - ni in
300pF 1 1 13
ZV5 - 10 i
-
l b II 1 17 SS VSENSE
0
1 lu
- GND uF
1" - Fig. II.2 Prototype forward converter
~ ~ J A P F
16