3kw DCDC Converter For Electric Vehicle
3kw DCDC Converter For Electric Vehicle
3kw DCDC Converter For Electric Vehicle
Abstract
The Masters Thesis is devoted to quantify and verify the performance of
two full-bridge DC/DC converter topologies that are suitable for usage in a
hybrid electric vehicle, whereas possible improvements should be proposed
and implemented. The two topologies investigated are the full bridge with
full wave rectification and full bridge with current doubler synchronous rectification. The idea is that the converter topologies should be implemented
in Allegro AMS and a loss calculation script should be written in Matlab.
It should be possible to simulate and study the converters performance in
the ideal and real cases. In the project two different topologies were investigated and built in Allegro AMS. Simulations have been performed in Allegro
AMS and calculations carried out in Matlab. Zero-voltage switching (ZVS)
technique has been implemented and verified. A suitable transformer has
been designed and the whole procedure has been presented in the report.
Losses have been determined and the efficiency plots have been created and
presented.
The performed simulations have shown that it is possible to utilize both converters in a HEV and the normal performance has been verified. ZVS switching technique has been successfully implemented that has reduced switching
losses significantly, which has improved the efficiency of the converters. A
calculation script for loss determination has been successfully written and
losses have been determined, which are included in the results section. A
transformer has been designed and the size of it has been minimized with
the appropriate core being chosen. Efficiency has been determined and for
the full wave topology it reached a maximum value at 95.4 % at the load
level of 35% and for the current doubler it has reached 96.5 % at the load
level of 30 %. It has been shown that the current doubler is more efficient
than the full wave topology.
Full bridge DC/DC converter, modeling, zero-voltage switching, current doubler, synchronous rectification, switching loss, conduction loss, transformer
design, efficiency
Acknowledgments
We are grateful to Volvo Car Corporation, specially to Karl Klang for giving us
the opportunity to perform our project at their premises. We would like to express
special gratitude to all the guys who were helping and supporting us throughout
the project, especially to Mats Nilsson, Anders Magnusson and Leif Karlsson.
We would like to thank our examiner Prof. Torbjorn Thiringer for his immense
help and relevant advices. We would also like to thank our families and relatives
who were supporting and believing in us, without their exceptional support, successful completion of the project would not be feasible.
Contents
1 Introduction
1.1 Background . . .
1.2 Aim/Purpose . .
1.3 Simulation object
1.4 Method . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
2
2
3
3
2 Theory overview
5
2.1 Various topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Conventional full bridge DC-DC converter with full wave
synchronous rectification . . . . . . . . . . . . . . . . . . . . 6
2.1.2 Full bridge DC-DC converter with current doubler rectification 6
2.2 Zero voltage switching . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3 Set-up description
3.1 Case I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.1 Full-bridge DC-DC converter with synchronous full wave rectification . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Case II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Full-bridge converter with current doubler synchronous rectification . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Switching pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 Switching pattern of Case I . . . . . . . . . . . . . . . . . .
3.3.2 Switching pattern of Case II . . . . . . . . . . . . . . . . . .
3.4 Components selection . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.1 MOSFET . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
9
9
10
10
11
11
13
14
15
CONTENTS
4 Analysis
4.1 Waveforms verification . . . . . . . . . . . . . . . . . . . . . . . .
4.1.1 Full wave synchronous rectification . . . . . . . . . . . . .
4.1.2 Current doubler synchronous rectification . . . . . . . . . .
4.2 Switching techniques . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Full wave synchronous rectification - hard switching . . . .
4.2.2 Current doubler synchronous rectification - hard switching
4.2.3 Full wave synchronous rectification - soft switching . . . .
4.2.4 Current doubler synchronous rectification - soft switching .
4.3 Loss determination . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.1 Transformer design procedure . . . . . . . . . . . . . . . .
4.3.2 Semiconductors loss . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
17
17
17
22
24
25
28
30
32
38
38
42
5 Results
5.1 Transformer selection
5.2 Semiconductor losses
5.3 Efficiency . . . . . .
5.4 Cost estimation . . .
.
.
.
.
45
45
50
55
57
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6 Circuit improvements
63
6.1 Spike killer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7 Conclusions and Future work
65
7.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Bibliography
68
A Results - Tables
70
ii
1
Introduction
Growing concerns over environmental impact lead to the development of more
eco-friendly technologies in almost all industrial branches. Automotive industry
is highly concerned over governmental continuous legislation adjustments towards
the reduction of exhaust gas emissions. However, pollution is not the only vital
aspect that must be considered, another fact is that fossil fuels are depleting, hence
the road transportation sector has high dependency on it and convenient alternatives must be identified to maintain efficient and reliable performance.
All major car manufacturers have taken a new direction towards sustainable transportation, thus integrating electric powertrains in conventional ones or replacing it
totally with fully electrical. These changes infer to absolutely all new and different
concepts of perception for automotive industry sector. This all unique concept is
called hybrid electric vehicles, which already devised 100 years ago. The research
in this particular area has been accelerated in a rapid pace during recent years.
A lot of development strategies have been elaborated, relevant experiments have
been performed and practical models have been built.
However, new technologies are always supported by complex systems comprised of
power electronic devices, which enable more controllability whilst they contribute
towards robust, reliable performance of a particular object. Therefore, vehicles are
not exceptions, a bunch of power electronic devices are involved, where converters
play a tremendous role, allowing reliable workability of the hybrid electric vehicles.
Consequently, the following discussion and analysis will be devoted to investigate
and to quantify several DC/DC converter topologies used for auxiliary purposes
inside the hybrid electric vehicles.
1
1.1. BACKGROUND
1.1
CHAPTER 1. INTRODUCTION
Background
There are several different electrified vehicle topologies that are known today. All
of them come with rechargeable batteries that can be restored to almost full state
of charge by either connecting the vehicle plug to an external power source or by
an internal generator on board in the car. However, to achieve a smart control and
efficient charging, a good utilization of the power electronic converter is highly required. Therefore, all the following analysis will be oriented on deep investigation,
validation and quantification of DC/DC converters, needed for auxiliary purposes
in a hybrid electric vehicle. Formerly, it was the alternator that has been generating power to the internal power network inside the car, but now manufacturers
are heading towards utilization of more innovative decisions, those like DC/DC
converters. Those power electronic devices are much more efficient, allow better
controlability, safer and more powerful. They are fed from the main battery package of the car. There is a massively growing tendency towards replacing the power
supply for the auxilitary purposes with the DC/DC converters instead. Various
topologies are utilized for this particular purpose nowadays. Every manufacturer
has their own design and arrangement of such a DC/DC converter, which corresponds to the particular requirements every producer should strictly follow. In the
project, two different topologies that are quite common in electrified vehicle applications were investigated and their performance quantified. Those are full bridge
DC/DC converters with full wave and current doubler synchronous rectification.
1.2
Aim/Purpose
The aim of the thesis is to investigate, quantify and validate the performance of the
full-bridge DC/DC converter used in an electrified vehicle. The purpose, which is
implied in the project, is referred to quantification of losses and efficiency of both
converter topologies as well as verification of DC/DC converter functionalities for
a hybrid electric vehicle. The tasks to be performed in the project can be divided
into the following subtasks, where each will be further analyzed:
Simulation object
Waveforms attainment
Analysis & comparison
Validation & quantification
Cost estimation
1.3
CHAPTER 1. INTRODUCTION
Simulation object
The objects simulated are full bridge DC/DC converters with different types of
rectification and the location of it in the car is shown in Fig. 1.1. In this particular
project two different topologies of DC/DC converters will be implemented. One
of them is with full wave rectification and the another one is with current doubler
rectification. The converters are supposed to transform the voltage level, from 355
VDC to 14 VDC, which is to be used in a hybrid electric vehicle internal power
network needed to supply auxiliary services. The converter will be fed from a 355
VDC voltage source. It will have a transformer for galvanic insulation and fullbridge rectifier on the output with a LC filter to reduce output ripple as much as
possible.
Fig. 1.1: Part of the architecture of an electric vehicle, the investigated DC/DC
converter is marked red
1.4
Method
1.4. METHOD
CHAPTER 1. INTRODUCTION
Compare the ideal and real converters, verify the simulated performance
through studying relevant waveforms
Determine the appropriate switching technique, among hard- or soft-switching
Determine and quantify important parameters, ex. efficiency, losses etc.
Analyze and verify the obtained results, maybe suggest possible improvements, if necessary.
2
Theory overview
This chapter is devoted to give a short overview of the equipment being used and
implemented in the project.
2.1
Various topologies
There are several converter topologies that have been investigated and implemented in the project. In order to understand the behavior and the performance
of the converter, coherent explanations and fundamentals of them are included in
the following sections. In the meantime, DC/DC converters are the devices that
change one level of DC voltage to another (either higher or lower) level . They are
basically used in battery-powered applications and devices that require different
levels of voltage. A DC/DC converters allow such an equipment to be powered
from a battery with a single voltage level, avoiding the bulky and heavy battery
packs required for various individual purposes.
There are many topologies for DC/DC conversion, such as buck, boost, buck-boost,
forward, cuk, flyback, push-pull converters etc. Additionally, usually a distinction
is made between isolated and non-isolated converters. In many applications multiple outputs are required, safety standards need to be met and impedance matching
must be provided, therefore output isolation may be required to be implemented.
Since the thesis is dealing with an application in a hybrid electric vehicles, there
are several topologies that are utilized nowadays. However, this project will stick
to two of them, which are commonly exploited. They are the conventional fullbridge with full wave synchronous rectification and the current doubler topology
with synchronous rectification.
5
2.1.1
High power bi-directional DC/DC converters are required for applications such as
battery charges, hybrid electric vehicles, uninterruptible power systems [1]. The
bi-directional DC/DC converter is great for electric vehicle applications. The most
fundamental of the existing high power bi-directional DC/DC converter topologies
is depicted in Fig. 3.1, which can be operated alternately as a step-down or a
step-up converter. It is characterized by a traditional full bridge inverter, isolation
transformer and high current rectifier with synchronous rectification on the low
voltage side. The input to the converter could be either fixed or it could be
varying magnitude DC voltage. The output voltage can be controlled with either
duty cycle control or phase-shifting technique. Similarly, the magnitude and the
direction of the output current can be controlled [2].
2.1.2
One of the most common alternative topologies for vehicle applications is current
doubler. The inverter side of this topology is the same as conventional full bridge
converter with four switches in two different legs. The difference is in the rectifier
side which has only two switches instead of four. Each switch sees the output current when it is switched on, but this output current is divided between transformer
and one of the output inductors. Therefore, only half of the output current flows
through the transformer and thats why it is called current doubler. Less current
in the transformer means less losses. This circuit has one additional filter inductor in the output side compared to the conventional topology, but each of those
inductors carries only half of the output current. Indeed, there is always current
in these two inductors and these currents will be integrated and flow to the load.
Comparing to other alternative topologies, which have center-tapped transformer,
this circuit has simpler structure of transformer which results in smaller size. The
topology is shown in Fig. 3.2 [7].
2.2
Conventional LC loop shaping circuit also called resonant circuit is often used in
shaping the switch voltage and current in order to yield zero-voltage and/or zerocurrent switching. These techniques have been developed using resonant switch
concept to tackle the switching losses in power devices as well as to suppress electromagnetic interference (EMI) produced, due to large di/dt and dv/dt caused
by switch-mode operation. The zero-voltage-switching (ZVS) can reduce the total switching losses by more than half, while zero-current-switching (ZCS) reduces
them by about 25 % when compared to hard switching. Hard switching can be
described by simple switching trajectory (loci) that is represented in Fig. 2.1.
The area under this trajectory represents the switching loss, hence it must be
reduced. This might be done in different ways. To relieve these switch stresses,
dissipative snubber circuits may be connected in series or parallel with the switches
in the switch-mode converters. However, these snubbers do not decrease the overall switching power loss, because it is simply shifted from the switch to the snubber
circuit. Consequently, to overcome these unwelcome stresses one may utilize the
technique of turning on and off each of the converter switches when either the
switch voltage or current is zero. Ideally, the switching transition must occur
when both voltage and current are zero, thus results in the following switching
trajectory depicted in Fig. 2.2, where the losses would be reduced.
ZVS topology is constituted, when the switch turns on and off at zero voltage.
The peak resonant voltage appears across the switch, but the peak switch current
remains the same as in its switched-mode counterpart. In contrast to ZCS, the
7
2.3. TRANSFORMER
Fig. 2.2: ZVS or ZCS switching loci that represent improved pattern
ZVS technique eliminates switching loss and di/dt noise due to the discharging of
MOSFETs junction capacitances and the reverse recovery diodes, and enables the
converters to operate at yet high switching frequencies.
2.3
Transformer
3
Set-up description
This chapter is devoted to the thorough description of the converter circuits being
investigated in the thesis work. The coherent analysis is conducted with simulated
results and an explicit explanation of the converters performance is carried out
through obtained waveforms.
3.1
3.1.1
Case I
Full-bridge DC-DC converter with synchronous full
wave rectification
In the circuit below the full-bridge DC-DC converter with full wave rectification
on the secondary side is depicted. It is commonly utilized in different applications,
mostly power supplies for various purposes. The proposed topology is investigated
since it is conventional and formerly was widely used; hence the comparison with
innovative ones would be important and perceptive. First, the topology shown in
Fig. 3.1 is discussed.
In this full-bridge converter, the input is a dc link with varying magnitude DC
voltage Vs and the input capacitor. The primary side consists of two legs, one
active and one passive; where in each leg there are two primary switches and their
diodes in anti-parallel to allow freewheeling, when there is no power transfer interval. The secondary side has the same configuration as the primary, but diodes
are connected in parallel to the switches, to allow current flow when some switch
is not conducting (current commutation interval). Both sides are coupled through
a transformer for safety purposes and to have galvanic insulation.
9
3.2. CASE II
On the output of the converter, an output filter is installed to prevent high current
and voltage ripple, consisting of an inductor and a capacitor. The performance of
the converter was investigated utilizing unipolar and bipolar switching technique,
whereas implementing hard and soft switching as well.
3.2
3.2.1
Case II
Full-bridge converter with current doubler synchronous
rectification
There are different possible topologies which can be studied and compared with
conventional full wave rectification converter. One of these choices is a full bridge
converter with current doubler rectification on the secondary side. By having a
current doubler structure, the secondary side of the transformer only sees half of
the output current with no center tapping need. Different topologies also exist
in order to achieve current doubling characteristics. They are made by changing
the location of switches and inductors on the secondary side and also by choosing
diodes or other semiconductors as switches. Using switches instead of diodes at
the rectifier side is preferred in order to achieve synchronous (active) rectification
for improving the efficiency.
On the other hand, the output voltage in this particular application is not high
and the diode voltage drops will be a high percentage of the output voltage in case
10
of using diodes which is not reasonable. A full bridge converter with a current
doubler synchronous rectification with ideal components is shown in Fig. 3.2.
Fig. 3.2: Full-bridge DC-DC converter with current doubler synchronous rectification
3.3
Switching pattern
There is a big variety of switching patterns utilized in modern switching converters. Depending on the purpose, the converter must be operated using appropriate
switching technique. Concerning the projects application purpose, one may say
that there is a vital need to deliver regulated low ripple voltage to the battery,
having a big freedom in controlling the output voltage. Therefore, the unipolar
switching technique based on enhanced phase-shifted controller (UCC2895 TEXAS
Instruments) has been implemented. However, for the analysis purposes, another
switching pattern has been implemented using almost 50% duty cycle which is
called bipolar switching.
3.3.1
different switching modes were implemented, considering hard and soft switching
techniques.
Voltage [V]
Vgate1
Vgate2
Vgate3
Vgate4
10
15
20
Time [us]
Hard switching implies that power switches must cut off the load current within
turn on and turn off time intervals, resulting in severe stress on the semiconductors.
The switching trajectories for different switching techniques may be illustrated according to Fig. 3.3 and Fig. 3.6. During the switching intervals, the power semiconductor device should withstand high voltage and high current at the same time,
resulting in substantial stress and switching losses. This causes many additional
problems, like heat dissipation, exaggerated components ratings etc. Therefore,
the concept is to move to resonant converter operation, incorporating zero voltage
(ZVS) or zero current (ZCS) switching with the help of a resonant tank, consisting
of inductances and capacitors. This creates oscillatory (usually sinusoidal) voltage and/or current that enables to accommodate ZVS or ZCS, hence reducing the
switching losses.
As it is seen from the circuit, the secondary side is implemented with controlled
switches; such a configuration is called synchronous rectification. It eliminates the
body diode conduction and reduces the rectifier conduction loss, hence improves
12
Voltage [V]
Vgate1
Vgate2
Vgate3
Vgate4
10
15
20
Time [us]
converter efficiency. However, there is a time instant when the parasitic body diode
unavoidably will carry the load current, which decreases the converters efficiency
because of the higher voltage drop over the body diode than over the switch itself.
Moreover, problems like increased switching losses and electromagnetic interference (EMI) occurring due to reverse recovery of body diodes arise. Consequently,
introducing a resonant LC-tank allows avoiding conduction of the body diode during the switching transitions. Therefore, with the help of the energy released from
the resonant tank, the switch junction capacitance is discharged, which allows the
power switch to turn on at ZVS.
3.3.2
On the full bridge side, the same pattern as mentioned in case I is implemented.
Bipolar switching, which means two pairs of diagonal switches work respectively
and unipolar switching, when diagonal switches work with a certain delay relative
to each other that is also called the phase shift. The difference between this case
and the previous one is the secondary side, where four switches are replaced by two.
In the bipolar pattern two secondary switches are working in the same manner as
13
the primary passive leg switches with the same duty cycle of slightly below 50%.
The gate voltages of the six switches are shown in Fig. 3.5.
VGate1
VGate2
VGate3
VGate4
VGate5
Voltage [V]
VGate6
10
12
14
16
Time [us]
In the phase shifted pattern, two secondary switches S5 and S6 start conducting
together with the switches S1 & S3 or S2 & S4. As it can be seen in Fig. 3.6,
the duty cycle for rectifier switches is more than 50% because they both need to
be on, during free-wheeling period, where the upper (S1 & S3) or the lower (S2
& S4) primary switches are working at the same time. On the other hand, S6
should not be turned off before S1 and similarly S5 can be turned off after S2
stops conducting.
3.4
Components selection
For choosing the appropriate rating for the semiconductor devices, a 200 % current
margin and 100 % voltage margin have been selected, due to the high power
application of 3 kW. To use the peak value for the voltage and the average value
for the current is sufficient due to the safety margin.
14
VGate1
VGate2
VGate3
VGate4
VGate5
Voltage [V]
VGate6
10
12
14
16
Time [us]
Fig. 3.6: Phase shifted switching pattern in the current doubler converter
3.4.1
MOSFET
The rated current is calculated using obtained data from the simulations in Allegro AMS with maximum power output and then imported to Matlab, where the
relevant calculations have been performed. Consequently, according to the carried
out calculation, the primary side rated current is Irated =8.9A for both investigated
topologies, full wave and current doubler converters, since they both have conventional full-bridge arrangement on the primary side. The transistors must conduct
at least twice the current regarding general safety instructions. However, in this
particular application, the DC/DC converter serves as a main source of power supply to the inner auxiliary network of the car, which means that if fault happens
auxiliary network will be out, which is of course not allowed.
Therefore, it is a bit unsafe to use just 18 A switch, hence a switch with a higher
current rating of 27 A was implemented. The voltage that the transistor must
withstand is 355 V, but some overvoltage is allowed for a short time. Then the
maximum voltage that the MOSFET has to withstand is 710 V. The switches that
were chosen are STB57N65M5 and STW69N65M5 from ST Microelectronics for
15
the full wave arrangement. For the current doubler, IPW65R080 from Infenion
with almost the same ratings was chosen.
The same procedure has been conducted for the secondary side, but the ratings
are different. The current measured and calculated at full load equals 210A and
the output voltage is 14V. The switches chosen were FDP8440 and FDP020N06B
from Fairchild for the full wave topology. Even though the secondary side of the
transformer in the current doubler does not experience the full load current, the
switches do. Therefore, the switches that were chosen are Fairchild FDP027N08B
and Vishay SUM110N06.
After having performed simulations and a relevant verification, a quanitification is
carried out and the most suitable semiconductor devices are chosen according to
the highest efficiency reached in the converter, which is justified in the following
sections.
16
4
Analysis
4.1
Waveforms verification
4.1.1
The following analysis is performed based on the circuit shown in Fig. 3.1; different modes of operation are considered and explained by presenting waveforms
obtained from the simulations with the help of Fig. 4.1 to Fig. 4.3. This case
is considered to be closely related to the real one, since the transformer is not
perfectly coupled, which implies having some leakage inductance in the circuit.
The operation represents zero voltage switching on both sides of the transformer,
which was one of the aims in the project.
Mode I: initiation of normal active power exchange
Prior to t0 , diagonal switches on both sides of the converter are turned on, which
corresponds to the initiation of the active power transfer. The voltage Vin is immediately exerted on the transformer primary side, transferred to the secondary
side with transformers turns ratio. However, secondary voltage does not simultaneously follow the primary side, because of the negative current in the transformer
due to the leakage inductance. Therefore, it requires some time to reverse the
polarity of the current and to start the active power exchange.
17
CHAPTER 4. ANALYSIS
Fig. 4.1: Primary and secondary sides currents and voltages in the transformer as
well as output inductor current
18
CHAPTER 4. ANALYSIS
CHAPTER 4. ANALYSIS
has been initiated. However, there is no energy exchange between primary and
secondary sides of the transformer, which is indicated by zero voltage over it, but
there is a current that is flowing in the transformer, due to some leakage inductance
in the circuit. This corresponds to the output inductor discharge, hence leading
to inductor current decrease. On the secondary side, switch S5 and diode D8 conduct the same full current, sharing the current between switch S7 and diode D6
respectively. At t3 , the freewheeling interval finishes because of S1 and S5 turning
off, thus the next transition of the current is initiated.
REMARK
The important observation has to be noticed that there is a positive current flowing through S7, which might seem doubtful from the first sight. However, there is
a coherent explanation to this phenomenon. This could be explained by the fact
that the direction of the current through S7 is strongly dependant on the operating
point (loading) of the converter at that particular time instant and the size of the
leakage inductance of the transformer. The justification is brought down in Table
4.1. It is seen that the voltage potential over S7 is changing with respect to the
load change, seen from Fig. 4.4. At some load conditions, the direction of the
current flow through S7 during freewheeling interval may also change (indicated
by different voltage polarity), which might seem unreasonable, but it is actually
how the circuit operates.
20
CHAPTER 4. ANALYSIS
Case 2
Case 3
Vdc=14 V, L=nom
Vdc=13.5 V, L=nom
Vdc=14 V, L=1.5*Lnom
diLout/dt
-3.19 A/us
-3.09 A/us
-3.18 A/us
diLp/dt
-0.36 A/us
-0.5 A/us
-0.192 A/us
diLs/dt
-5.51 A/us
-7.7 A/us
-2.96 A/us
Ptransfer
67 A
96.17 A
46.7 A
Vs1
0.207 V
0.275 V
0.144 V
Vs3
-0.207 V
-0.275V
-0.144 V
Vp
1.09 V
1.5 V
0.877 V
0.057 V
Vs
0.071 V
0.098 V
VL1
-1.505 V
-2.055 V
-1.16 V
VL2
-0.099 V
-0.135 V
-0.077 V
Vs5
-0.176 V
-0.246 V
-0.13 V
Vs7
-0.0048 V
-0.013 V
0.0041 V
Vs6
-0.836 V
-0.813 V
-0.843 V
Vs8
-1.0081 V
-1.047 V
-0.978 V
VLout
-15.013 V
-14.56 V
-14.97 V
Fig. 4.4: Power transfer with different voltage level and transformer inductance
values
21
CHAPTER 4. ANALYSIS
4.1.2
This part describes different modes of operation for the full-bridge dc-dc converter
with current doubler synchronous rectification. As it will be explained later, the
best condition from soft switching viewpoint is using phase shifted switching pattern and incorporating leakage inductance in the transformer representation when
simulating. That is why unipolar switching pattern has been used for the analytical part and a coupling factor of 0.99 is chosen to represent the leakage inductance
inside the transformer. Different vertical magenta dashed-lines have been shown
in Fig. 4.5 and Fig. 4.6 which separate operational modes.
Mode I: (t0 -t1 )
This mode starts from t0 where S4 is turned off and the transformer voltages
are falling to zero. Before starting this mode, S1 and S4 were conducting and the
input voltage was applied to the transformer and energy was exchanging from the
primary to the secondary side. By turning off S4, the stored energy in L1 charges
the capacitance of C4 and discharges C3 which makes D3 to conduct. Conducting
D3 beside S1 prepare free-wheeling mode even before turning on S3 and results in
zero voltage switching as it was mentioned in the soft switching part. By turning
on S3 the current starts flowing through it and it should be considered that this
current is negative according to the switch positioning.
The transformer current starts decreasing, the same as inductors currents which
show their discharging behavior and results in supplying the output energy. There
is no energy exchange between the two sides of the transformer in this mode.
The output current starts to free wheel through the rectifier switches. The lower
switch (S6) is already on at this moment and can carry the free-wheeling current,
22
CHAPTER 4. ANALYSIS
but S5 is not turned on yet. Since the voltage over the transformer is zero and the
secondary lower switch is conducting, there is zero voltage over S5, and D5 does
not need to conduct for achieving ZVS. By turning on S5 at the same time as S2
the current starts to flow through it, but this current is much smaller than S6 current which was on before.This mode of operation ends up with turning off S1 at t1 .
CHAPTER 4. ANALYSIS
ity of the current in the transformer changes and this results in a current polarity
change in S2 and S3 as well. During the whole time of this mode, the rectifier
inductors currents are decreasing linearly until t2 which is the ending point.
Mode III: (t2 -t3 )
As it can be observed, this mode is starting at the time t2 where the current
in D6 reaches zero and the transformer secondary current magnitude is equal to
the lower inductor (L2 ) current, which is increasing linearly. The transformer secondary voltage is trying to reach to the final value. The primary voltage is equal
to the input voltage thanks to the S2 and S3 conduction. It can be said that at
this mode the power exchange period starts and energy is delivered to the load
from the source side until t3 where the next free-wheeling period will be initiated
and sequence of events will be repeated as it occurred in the mode I.
4.2
Switching techniques
This chapter is devoted to the description of the three cases which result in different
switching behaviors. The analysis starts from the most idealized case, utilizing
bipolar switching with no leakage inductance in the transformer (coupling factor
= 1). Next case will be unipolar switching pattern with the same coupling factor.
24
CHAPTER 4. ANALYSIS
The last case corresponds to the unipolar pattern with added external elements
such as resonant components. In this case the leakage inductance is represented
by changing the coupling factor of the transformer to 0.99.
4.2.1
Fig. 4.7: Bipolar hard switching for conventional DC/DC, primary side switches
As it is depicted in Fig. 4.7 and Fig. 4.8 turn on of the switches in the primary
side do not satisfy zero voltage condition, therefore switches are forced to commutate the full load current by the control device, resulting in high voltage stress
over the switch while turning on, hence increased switching losses. Furthermore,
turn-on high voltage levels induces a severe switching noise phenomenon known as
the Miller effect (increase in the equivalent input capacitance due to amplification
of the effect of capacitance between the input and output terminals) which is coupled into the drive circuit, leading to significant noise and instability.
Nevertheless, the same is happening on the secondary side switches, which is understandable because primary and secondary sides are coupled through a transformer.
Everything that is happening on the primary is reflected to the secondary sides,
but with a certain delay, due to the leakage inductance in the transformer, if it is
not perfectly coupled.
25
CHAPTER 4. ANALYSIS
Fig. 4.8: Bipolar hard switching for conventional DC/DC, secondary side switches
The overlap that is created during voltage/current transition causes energy loss
and it may be minimized by increasing di/dt and dv/dt. However, quick changing of di/dt and dv/dt will lead to EMI generation; consequently they must be
optimized to avoid EMI issues as much as possible. These rise/fall times may be
reduced by introducing dissipative passive snubber circuits, and they are usually
implemented in real power devices. Meanwhile, those snubbers transit switching
loss and stress into their circuit, but do not decrease the total losses. Therefore,
introducing only snubber circuits is not the best alternative; further analysis is to
be performed. Thereafter, the unipolar switching on the basis of UCC28950 from
TEXAS Instruments is implemented and the converters performance analyzed,
which is depicted in Fig. 3.6. This switching pattern is applied to justify if soft
switching is achievable and sufficient for this particular configuration.
As it is seen from Fig. 4.9 and 4.10, just implementing unipolar switching pattern
does not guarantee soft switching. The active leg switches turn on at maximum
voltage over it, conducting the full load current. It has to be avoided, if there is
a need of reducing the switching losses and stress over the semiconductor devices.
The secondary side switches behave almost the same as the primary, but in this
case, the active leg switches and the passive leg upper switch turn on in a hard
way as it is seen from the 4.10, when there is a voltage applied over them, but the
voltage is quite low and it is negative. To overcome this action the transformers
leakage or external inductances and MOSFETs parasitic capacitances are effec26
CHAPTER 4. ANALYSIS
Fig. 4.9: Phase-shifted hard switching for conventional DC/DC, primary side
switches
Fig. 4.10: Phase-shifted hard switching for conventional DC/DC, secondary side
switches
27
CHAPTER 4. ANALYSIS
Here it may be observed that during the transient, there is an undershoot of the
voltage over the switches in the passive leg, which is due to the reverse recovery
of the diodes, connected in parallel.
In order to achieve zero voltage switching over the switches, move on to adding
some external inductance, additionally changing the coupling factor of the transformer to introduce leakage inductance that is usually present in the real transformer. The conducted experiments and simulated waveforms are presented in the
next section.
4.2.2
As it can be seen in the simulation results in Fig. 4.11 and Fig. 4.12, all switches
are turned on when the voltage over them are not zero. So, there is no soft
switching in any switches during turn-on duration. The upper parts of the each
subplot show the gate pulse voltages in different switches and the lower parts show
the voltages over the switches.
Fig. 4.11: Hard switching in the primary side switches of the current doubler converter in case of bipolar switching pattern
28
CHAPTER 4. ANALYSIS
Fig. 4.12: Hard switching in the secondary side switches of the current doubler
converter in case of bipolar switching pattern
In order to solve the hard switching problem, unipolar switching pattern as the
second case has been performed and results have been compared with previous
case. The hard switching problem during turn-on is removed in four switches (36), but there are still some problems in switch 1 and 2 which are located in active
legs. Zero voltage switching is not achieved yet in these two switches even with
phase shifting pattern. All switching waveforms can be seen in Fig. 4.13 and Fig.
4.14.
Fig. 4.13: Hard switching in the primary side switches of the current doubler converter in case of phase shifted switching pattern
29
CHAPTER 4. ANALYSIS
Fig. 4.14: Hard switching in the secondary side switches of the current doubler
converter in case of phase shifted switching pattern
4.2.3
Fig. 4.15: ZVS switching with phase-shifted switching pattern for full wave DC/DC,
primary side switches
30
CHAPTER 4. ANALYSIS
Fig. 4.16: ZVS switching with phase-shifted switching pattern for full wave DC/DC,
secondary side switches
According to the performed simulations and obtained waveforms ZVS is successfully achieved on the primary as well as on the secondary side switches, which
can noticed in Fig. 4.15 and Fig. 4.16. This was achieved by modeling the power
semiconductor devices with parallel diodes, parasitic capacitances, and leakage
transformers inductance of 1.5% and external secondary side resonant inductance
of 120 nH.
All parasitic capacitances in the circuit including winding and heat sink capacitance have been lumped together as a switch capacitance. For example, when the
switch S4 turns off, the inductor current charges capacitance C4 and discharges
capacitance C3 until the voltage across C3 reaches zero, and the body diode D3
conducts current and after that the switch S3 turns on with zero voltage and the
inductor current freewheels through the switch instead.
All other switches operate according to the same principle described, but utilizing
different magnetic energy storages for different converters legs. To implement a
proper switching transition, the inductor energy stored at the start of the transition must be large enough to completely charge the switch capacitance from the
initial to the final voltage Vsw .
In the meantime, if the delay time determined is too high, the interval when
31
CHAPTER 4. ANALYSIS
the voltage across the switch is zero would be missed, which would lead to the capacitance being repeatedly recharged and the possibility of zero voltage switching
would be lost. Therefore, this has to be avoided by using proper duty cycle and
gate drive delay time determination. Another point corresponding to the proper
circuit operation would be body diode conduction prior to turn-on of the switches.
4.2.4
There are several ways to achieve zero voltage switching. One method which was
used above was changing the switching pattern that was mentioned as the second
case at the beginning of this chapter. Thanks to the phase shifting pattern the
hard switching problem was solved partially but not in all switches. Another solution was using transformer leakage inductance. The next step is changing the
coupling factor to a value less than one which represents the leakage inductance
in the transformer. Leakage inductance can be also be modelled by putting an
inductance in series with the primary winding of transformer.
It can be observed that leakage inductance will help to achieve zero voltage switching in all switches during turn on and off. The waveforms are depicted in Fig. 4.17
and Fig. 4.18. It should be mentioned again that the upper parts of the each
subplot show the gate pulse voltage in different switches and the lower parts show
the voltage over the switches.
A deeper study is needed on the zero voltage switching in order to undertand
how it is achieved in some conditions and not in the other. For this purpose two
transition points have been chosen and are depicted in Fig. 4.19 and 4.20 by
vertical dashed-lines in magenta color.
The first point is corresponding to the instant at which S4 turns off but S1 is
still conducting. Indeed, the power transfer is stopped here and the free-wheeling
period starts by turning on S3. According to the graphs, there is a delay between
turning off S4 and turning on S3. At this instant after turning off S4, the stored
energy in L1 charges the capacitance of C4 and discharges C3 which makes D3 to
conduct. S1 and D3 are on and the free-wheeling mode has started. Conduction
of D3 prepares a zero voltage over S3 and then by turning on S3, current will
flow through it under soft switching conditions. It should be mentioned that the
current flowing through the switch is considered negative according to the switch
positioning. The same story can happen while turning off S3 and turning on S4. In
this case D4 is conducting first and then S4 turns on under zero voltage condition.
32
CHAPTER 4. ANALYSIS
Fig. 4.17: Zero voltage switching in the primary side switches of the current doubler
converter in case of phase shifted switching pattern and coupling factor=0.99
Fig. 4.18: Zero voltage switching in the secondary side switches of the current
doubler converter in case of phase shifted switching pattern and coupling factor=0.99
33
CHAPTER 4. ANALYSIS
Fig. 4.19: Zero voltage switching behavior analysis in the current doubler converter
Fig. 4.20: Zero voltage switching behavior analysis in the current doubler converter
CHAPTER 4. ANALYSIS
exchange period between the primary and secondary sides of the transformer starts
at this point. After some instants the gate pulse voltage is applied to S2 and it will
be turned on and the current transferred from diode to the anti-parallel switch.
Again thanks to the anti-parallel diode there is a zero voltage switching over S2
since it will be turned on when its voltage is zero. The same sequence of events
will occur between turning off S2 and turning on S1.
As it can be seen and was mentioned before, S6 should not be turned off before S1
since it is needed to have both rectifier switches on during the free-wheeling period.
There is the same story for S5 and S2 during next free-wheeling time interval. It
was seen before that in case of having no leakage inductance there was not any soft
switching in the active leg switches during turn on. For the comparison purpose,
the waveforms are obtained and analyzed at the same instants as previous case
and are shown by magenta dashed-lines in Fig. 4.21 and 4.22. The first point is
showing the same behavior as before. When S4 is turned off, the stored energy
in secondary inductor charges and discharges C4 and C3 and D3 starts conducting. So the voltage over the transformer and S3 voltage fall to zero, and then
by turning on S3 zero voltage switching will be achieved. The story is different
for the next instant where the active leg switch S1 is turned off. Because of not
having any stored energy in the leakage inductance, C1 and C2 are not charged
and discharged, D2 does not conduct and the voltage over the other active leg
switch S2 stays constant at the input voltage value. Turning on S2 while it has the
maximum voltage is called hard switching which results in high switching losses.
This interval continues until turning off S6 which completes the free-wheeling
mode. The current flows through the secondary side of the transformer and primary side subsequently. Since S3 is still turned on, the primary current which is
negative goes through it. This negative current should close the circulation loop
and there is no way except passing through D1. With turning on S2 the current
moves from D1 to S2.
As it was explained before when there is switching in one of the active leg switches,
the energy in the leakage and magnetizing inductances charge and discharge the
parallel capacitors and forces the anti-parallel diode to conduct which makes zero
voltage switching to take place in the related switch. Since this stored energy is
not big, there is a limited time to achieve soft switching. There is a time between
turning off S1 and turning on S2 which is called dead time. This time can be
calculated according to the characteristics of the switch. In the last simulations,
this time is set to 45 ns. In Fig. 4.23 it is obvious that this time has been increased
in order to see how it affects the soft switching operation.
35
CHAPTER 4. ANALYSIS
Fig. 4.21: Hard switching behavior analysis in the current doubler converter
Fig. 4.22: Hard switching behavior analysis in the current doubler converter
When S1 is turned off, the energy charges and discharges the capacitors, D2 conducts and the voltage over S2 goes to zero. Since this energy is not that much big
36
CHAPTER 4. ANALYSIS
the current through D2 decrease to zero after a while. On the other hand, S3 is
on and its free-wheeling current closes the loop with D1 (The green curve which
is magnified 5 times higher). Therefore, the voltage over S2 is back to the input
voltage and the voltage over S1 reaches to zero. Under these circumstances if S2
turns on the switching will happen under the maximum voltage and this means
zero voltage switching opportunity is lost.
Fig. 4.23: Limited range for soft switching in case of having longer dead band in
the current doubler converter
In the case of switching in one of the passive leg switches this problem does not
occur since there is a bigger source of energy in the output filter which can help
to achieve soft switching even with high dead band between S3 and S4 turning
on and off. Indeed the anti-parallel diode can conduct until the related switch is
turned on.
37
4.3
CHAPTER 4. ANALYSIS
Loss determination
The following chapter is devoted to the determination of losses in the converter and
the choice of appropriate transformer for the particular application being described
in the project. The transformer design procedure and loss calculation script have
been conducted and implemented in Matlab.
4.3.1
V (t) = N
d
dt
(4.1)
dB
Am
dt
(4.2)
dB
B = BcostV
cos(wt)
1 (t) = N1 Am
dt
(4.3)
V1 (t) = N1 Am Bsin(wt)
(4.4)
38
CHAPTER 4. ANALYSIS
V1 = N1 Am Bcore
(4.5)
core
= Am B
(4.6)
where V1 (t) is the time-varying voltage on the primary side, Am is an effective area
of the core, d
is the time derivative of the magnetic flux in the core.
dt
Inserting (4.6) into (4.2), gives
Z DTs
Bcore
dB
V1 dt = N1 Am
(4.7)
Bcore
The voltage applied over the primary side, when there is an active power transfer
interval, is V1 , which is the input voltage. This voltage will create the peak flux
core in the transformer, which will be changing from negative to positive
density B
peak flux density, since this work deals with bidirectional operation of the fullbridge converter. This results in the following relation,
N1 =
Vd DTs
core
2Am B
(4.8)
According to the (4.7), the peak flux density also depends on the input voltage
that is applied over the transformer. Therefore, by changing the duty cycle DTs
core or N1 will also be changing, depending on
the peak flux density in the core B
the method through which it was calculated.
In order to define the appropriate core for the transformers, the maximum input voltage and the maximum duty cycle is considered.
N1 =
Vd,max Dmax Ts
core
2Am B
(4.9)
However, by changing the duty cycle, the maximum flux density will change, and
that will result in a smaller core loss respectively.
Two different ways to choose appropriate transformers size have been implemented:
39
CHAPTER 4. ANALYSIS
define the number of turns, calculate flux levels and then check if they occur
in the core or not;
define flux levels, calculate number of turns and then check if they fit in the
core or not.
An important observation seen from (4.9) that has to be known is that increasing the number of turns on the primary side the peak flux density will decrease
respectively. Therefore, the core losses will decrease, however copper losses will
increase. Apparently, there is always a trade-off between core and copper losses.
Ideally, transformer loss should approximately split equally into core and copper
losses, but in practice it is hardly achievable, hence the balance between them has
to be maintained and the difference that results in the smallest deviation would
guarantee the most efficient transformer utilization. In applications, where the
transformer is under heavy load conditions, a design with larger core-losses and
smaller resistive losses could be favourable for example.
In order to calculate the cross-sectional area of the wiring, one must know the
RMS currents on the primary and the secondary sides of the transformer, which
are found with help of the fundamental equation described in [2]. Since the operating frequencies used in this project are quite high, in order to reduce the skin
and proximity effects losses in conductors Litz wire is utilized. It consists of many
thin wire strands, individually insulated and twisted or woven together, following
one of several carefully prescribed patterns often involving several levels.
Since we have relatively high currents in the transformer, a Litz wire with 0.94mm2
is used. In the project two different current densities have been chosen, 3 A/mm2
and 4 A/mm2 . This was done in order to see the difference in the copper loss and
identify the discrepancies. They were mounted in a bundle in order to withstand
those high currents and the number of wires in the bundle was calculated according
to.
The number of wires in the bundle for the primary side of the transformer
Xpri bundledwires =
Ipri,rms
JAlitz
(4.10)
(4.11)
CHAPTER 4. ANALYSIS
The number of wires in the bundle for the secondary side of the transformer is
Xsec bundledwires =
Isec,rms
JAlitz
(4.12)
Am 1,bundleN1 + A2,bundle N2
kcu
(4.14)
(4.15)
The core loss density must be scaled in terms of flux density and frequency, where
switching frequency f is given in kHz and B is the peak flux density in T , k
depends on the type of the material. The losses are closely proportional to the
switching frequency (a=1) and to the square of the flux density (b=2) for ferrite
materials.
Copper loss determination
In order to calculate the copper loss in the transformer, one must determine the
length of one Lturn turn that could be found in the core datasheet. Therefore,
having the number of turns of the windings, one can define the total length of the
conductors in the primary and secondary sides. Consequently,
Lpri,wind = Lturn N1 ; Lsec,wind = Lturn N2
41
(4.16)
CHAPTER 4. ANALYSIS
The next step is to calculate the resistance in the wires on each side of the transformer, which is given by
cu Lpri,wind
(4.17)
R1 =
A1,bundle
R2 =
cu Lsec,wind
A2,bundle
(4.18)
(4.19)
4.3.2
(4.20)
Semiconductors loss
The semiconductor losses can be divided into two different parts. Switching losses
which occur during turning on and off the switch, and the conduction losses which
occur when the switches are conducting.
Switching loss
By having the voltage and current waveforms over the switch during turn on and
off intervals, the dissipated power over the switch can be calculated, which is considered as switching losses. The left plot in Fig. 4.24 shows the voltage and current
waveforms over the switch during turn on. The voltage goes to zero from maximum value and the current increases from zero to the maximum value during the
turn on interval.
The right plot in Fig. 4.24 depicts the product of the voltage and current during
the same period. One triangle is formed which shows the instantaneous power
over the switch. By integration of this power versus time the dissipated energy
will be obtained. Finally, the average of dissipated power over the switch can be
calculated by dividing this energy by the switching period.
p(t) = v(t) i(t)
42
(4.21)
CHAPTER 4. ANALYSIS
Z
Wloss =
Ploss =
p(t)dt
Wloss
= Wloss fsw
Tperiod
(4.22)
(4.23)
Conduction loss
Since the MOSFET is behaving like a resistor during the conduction period it produces some conduction losses, which are dependent on the value of the MOSFET
on-state resistance and the current that flows through it.
2
Ponstate = Rdson Ion
(4.24)
In reality, usually there is a body diode inside the switch module, which is located
in anti-parallel with respect to the MOSFET and carries the current with reverse
polarity. It is the diode that conducts before turning on the MOSFET and prepares
zero voltage switching. Since there is a forward voltage drop across the diode when
it is on, some on-state losses can be produced and should be added to the MOSFET
on-state losses to form the total conduction losses for the switch module,
vonstate = Vsd + Rdson ion
43
(4.25)
Ponstate
CHAPTER 4. ANALYSIS
1
=
Tsw
Vsd
=
Tsw
Z
0
Tsw
Tsw
Rdson
ion dt +
Tsw
Tsw
i2on dt =
(4.26)
2
= Vsd Ion(avg) + Rdson Ion(rms)
The forward voltage drop and the on-state resistance of the switch module can be
obtained from the switch datasheet and the rms and average value of the current
can be calculated separately.
44
5
Results
The following chapter is devoted to the results that have been obtained in the
project. Relevant comparisons and justifications were performed and conclusions
were drawn as well as presenting the results.
5.1
Transformer selection
In this section the core for the transformer has been selected based on the flux that
is occuring in it and the volume of the winding that fits inside the core window.
The results are presented in Table 5.1.
Table 5.1: Peak flux density in different cores as a function of primary number of
turns at 100 kHz for the full wave converter
Core Type
Aw
B [T]
[cm3 ]
N1=10
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
ETD 29/16/10
210,56
1,056
0,528
0,352
0,264
0,211
0,176
0,151
ETD 34/17/11
304,56
0,826
0,413
0,275
0,207
0,165
0,138
0,118
ETD 39/20//13
398,56
0,642
0,321
0,214
0,160
0,128
0,107
0,092
ETD 44/22/15
492,56
0,464
0,232
0,155
0,116
0,093
0,077
0,066
ETD 49/25/16
586,56
0,380
0,190
0,127
0,095
0,076
0,063
0,054
ETD 54/28/19
680,56
0,287
0,143
0,096
0,072
0,057
0,048
0,041
ETD 59/31/22
774,56
0,218
0,109
0,073
0,055
0,044
0,036
0,031
Where Aw is the effective core area. Each cell represents different number of turns
in the secondary side and the magnetic flux magnitude formed in the core with
respect to the primary side. According to the datasheet of the magnetic material
45
CHAPTER 5. RESULTS
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
-13,77
103,06
219,89
336,72
453,55
570,38
687,21
ETD 34/17/11
-54,27
62,56
179,39
296,22
413,05
529,88
646,71
ETD 39/20/13
-117,47
-0,64
116,19
233,02
349,85
466,68
583,51
ETD 44/22/15
-161,77
-44,94
71,89
188,72
305,55
422,38
539,21
ETD 49/25/16
-226,55
-109,72
7,11
123,94
240,77
357,60
474,43
ETD 54/28/19
-294,97
-178,14
-61,31
55,52
172,35
289,18
406,01
ETD 59/31/22
-356,17
-239,34
-122,51
-5,68
111,15
227,98
344,81
ETD 29/16/10
Ve
Pcore [W]
[cm3 ]
N1=10
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
ETD 29/16/10
5.35
111,790
27,947
12,421
6,987
4,472
3,105
2,281
ETD 34/17/11
7.63
97,670
24,417
10,852
6,104
3,907
2,713
1,993
ETD 39/20/13
11.5
88,829
22,207
9,870
5,552
3,553
2,467
1,813
ETD 44/22/15
17.8
71,780
17,945
7,976
4,486
2,871
1,994
1,465
ETD 49/25/16
24.1
65,332
16,333
7,259
4,083
2,613
1,815
1,333
ETD 54/28/19
35.6
54,804
13,701
6,089
3,425
2,192
1,522
1,118
ETD 59/31/22
51.2
45,630
11,407
5,070
2,852
1,825
1,267
0,931
46
CHAPTER 5. RESULTS
lN
Pcu [W]
[mm]
N1=10
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
ETD 29/16/10
52.8
0,839
1,678
2,517
3,356
4,195
5,034
5,873
ETD 34/17/11
60.5
0,961
1,923
2,884
3,845
4,806
5,768
6,729
ETD 39/20/13
69
1,096
2,193
3,289
4,385
5,482
6,578
7,674
ETD 44/22/15
77.7
1,235
2,469
3,704
4,938
6,173
7,407
8,642
ETD 49/25/16
86
1,366
2,733
4,099
5,466
6,832
8,199
9,565
ETD 54/28/19
96
1,525
3,051
4,576
6,101
7,627
9,152
10,677
ETD 59/31/22
106.1
1,686
3,372
5,057
6,743
8,429
10,115
11,801
In Table 5.5 the total losses in the transformer are presented, since they are consisting of winding and core losses.
Table 5.5: Total transformer losses in the fullwave at 100 kHz
Core Type
Ptotal [W]
N1=10
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
ETD 29/16/10
112,629
29,625
14,938
10,343
8,666
8,139
8,154
ETD 34/17/11
98,631
26,340
13,736
9,949
8,713
8,481
8,722
ETD 39/20/13
89,925
24,400
13,159
9,937
9,035
9,045
9,487
ETD 44/22/15
73,014
20,414
11,679
9,425
9,044
9,401
10,107
ETD 49/25/16
66,699
19,066
11,358
9,549
9,446
10,013
10,898
ETD 54/28/19
56,329
16,752
10,665
9,527
9,819
10,674
11,796
ETD 59/31/22
47,316
14,779
10,127
9,595
10,254
11,382
12,732
It is known that in order to have ultimate design and selection of the transformer
the balance between the resistive transformer losses and the core losses must be
maintained. This means that the difference between them has to be minimized,
thus resulting in the most effective transformer design. In the ideal case, the difference must be 0, but it is hardly achievable in the real life. In Table 5.6 the sort
of balance between resistive and core losses is shown. It is preferable to choose the
core that has the lowest difference, which is marked with colors.
A good approximation for that balance to be maintained is 20% and as it is seen
from the table that all marked cells satisfy this condition, but not all of them
are suitable, since they may not fit in or may be over-saturated. According to all
aforementioned conditions in the project, the core ETD 59/31/22 with N1 =30
turns seems appropriate, therefore it has been selected. The size of the transformer
is 60x62x22 mm, since it consists of two pieces of ETD cores.
47
CHAPTER 5. RESULTS
Table 5.6: Difference between transformer core and copper losses in the fullwave at
100 kHz
Core Type
N1=20
N1=30
N1=40
N1=50
N1=60
N1=70
ETD 29/16/10
110,951
26,270
9,904
3,631
0,277
-1,928
-3,591
ETD 34/17/11
96,709
22,495
7,968
2,259
-0,900
-3,055
-4,736
ETD 39/20/13
87,732
20,014
6,581
1,166
-1,929
-4,111
-5,862
ETD 44/22/15
70,545
15,476
4,272
-0,452
-3,302
-5,414
-7,177
ETD 49/25/16
63,966
13,600
3,160
-1,383
-4,219
-6,384
-8,232
ETD 54/28/19
53,278
10,650
1,513
-2,676
-5,435
-7,630
-9,559
ETD 59/31/22
43,944
8,036
0,013
-3,891
-6,604
-8,847
-10,869
As it was mentioned earlier, for the current doubler case another approach has
been chosen for determining the proper number of turns in the transformer. By
having the maximum allowable flux density according to the material datasheet,
different steps from zero to this maximum value have been chosen and based on
these steps, different number of turns for primary side are determined. These values are shown in Table 5.7 and 5.8 for different flux densities, which are given in
[T] and various core types. As it can be seen, the EPCOS PQ cores have been
chosen for this topology, which is different from the full wave rectification case. All
parameters have been calculated at the maximum load power which is obtained
by setting the duty cycle and shifting the delay to the maximum possible value.
Table 5.7: Primary number of turns as a function of different flux density in the
current doubler at 100 kHz
Core types
Ae
N1
[mm2 ]
B=0,05
B=0,1
B=0,15
B=0,2
B=0,25
B=0,3
B=0,35
PQ16/11.6
42,1
421,6
210,8
140,5
105,4
84,3
70,3
60,2
PQ20/20
62,9
282,2
141,1
94,1
70,5
56,4
47,0
40,3
PQ26/25
122
145,5
72,7
48,5
36,4
29,1
24,2
20,8
PQ32/30
153,8
115,4
57,7
38,5
28,9
23,1
19,2
16,5
PQ35/35
169,7
104,6
52,3
34,9
26,1
20,9
17,4
14,9
PQ40/40
188,3
94,3
47,1
31,4
23,6
18,9
15,7
13,5
PQ50/50
330
53,8
26,9
17,9
13,4
10,8
9,0
7,7
Table 5.9 and 5.10 depict the values for the core and copper losses for different cores
in different flux densities. In order to keep the aforementioned balance between core
and copper losses inside the transformer the green cell in Table 5.11, PQ32/30,
has been chosen to be the proper core for this case. The difference between copper
and core losses for this operating point is in the desired range but maybe not the
48
CHAPTER 5. RESULTS
Table 5.8: Secondary number of turns in the current doubler at 100 kHz
Core types
Ae
N2
[mm2 ]
B=0,05
B=0,1
B=0,15
B=0,2
B=0,25
B=0,3
B=0,35
PQ16/11.6
42,1
47,2
23,6
15,7
11,8
9,4
7,9
6,7
PQ20/20
62,9
31,6
15,8
10,5
7,9
6,3
5,3
4,5
PQ26/25
122
16,3
8,1
5,4
4,1
3,3
2,7
2,3
PQ32/30
153,8
12,9
6,5
4,3
3,2
2,6
2,2
1,8
PQ35/35
169,7
11,7
5,9
3,9
2,9
2,3
1,7
PQ40/40
188,3
10,6
5,3
3,5
2,6
2,1
1,8
1,5
PQ50/50
330
1,5
1,2
0,9
minimum possible. Indeed, there might be some other cores which have better
balance between the two loss components but from the fitting viewpoint this core
is the best choice for this current doubler converter, where the core and copper
losses are 4.4 W and 3.83 W respectively. The primary and secondary number of
turns corresponding to this choice are about 38 and 4 turns.
Table 5.9: Transformer core losses in the curerent doubler at 100 kHz
Core types
Ve
Pcore [W ]
[cm3 ]
B=0,05
B=0,1
B=0,15
B=0,2
B=0,25
B=0,3
B=0,35
PQ16/11.6
1,14
0,053
0,213
0,479
0,852
1,331
1,917
2,609
PQ20/20
2,84
0,133
0,533
1,199
2,132
3,332
4,798
6,530
PQ26/25
6,53
0,306
1,224
2,755
4,898
7,652
11,019
14,999
PQ32/30
10,44
0,489
1,958
4,404
7,830
12,234
17,618
23,979
PQ35/35
13,44
0,630
2,520
5,670
10,080
15,750
22,680
30,870
PQ40/40
17,43
0,817
3,268
7,353
13,073
20,426
29,413
40,035
PQ50/50
37,27
1,747
6,988
15,723
27,953
43,676
62,893
85,605
Table 5.10: Transformer copper losses in the curerent doubler at 100 kHz
Core types
lN
Pcu [W ]
[mm]
B=0,05
B=0,1
B=0,15
B=0,2
B=0,25
B=0,3
B=0,35
PQ16/11.6
20
13,548
6,774
4,516
3,387
2,710
2,258
1,935
PQ20/20
44
19,950
9,975
6,650
4,987
3,990
3,325
2,850
PQ26/25
56
13,091
6,545
4,364
3,273
2,618
2,182
1,870
PQ32/30
62
11,497
5,748
3,832
2,874
2,299
1,916
1,642
PQ35/35
76
12,772
6,386
4,257
3,193
2,554
2,129
1,825
PQ40/40
87
13,177
6,588
4,392
3,294
2,635
2,196
1,882
PQ50/50
100,5
8,685
4,343
2,895
2,171
1,737
1,448
1,241
49
CHAPTER 5. RESULTS
Table 5.11: Transformer total losses in the curerent doubler at 100 kHz
Core types
Ptotal [W ]
B=0,05
5.2
B=0,1
B=0,15
B=0,2
B=0,25
B=0,3
B=0,35
PQ16/11.6
13,60157
6,987
4,995
4,239
4,040913
4,175
4,545
PQ20/20
20,08311
10,508
7,849
7,119
7,321609
8,123
9,379
PQ26/25
13,39688
7,769
7,118
8,171
10,2705
13,201
16,869
PQ32/30
11,98606
7,705
8,237
10,704
14,534
19,534
25,623
PQ35/35
13,4023
8,906151
9,927
13,27308
18,305
24,809
32,695
PQ40/40
13,99372
9,85647
11,745
16,36667
23,061
31,609
41,917
PQ50/50
10,43243
11,33082
18,618
30,12385
45,413
64,341
86,845
Semiconductor losses
There are always losses in semiconductor devices and they are presented in the following section. Each part of these losses have been calculated separately, switching
and conduction losses, according to the equations that were introduced before. Although the switch module Allegro AMS model has been used in the simulation and
just one current waveform for each switch module can be observed, it is important
to split the current between the body diode and MOSFET in order to calculate
the right value of conduction losses for each switch. This can be done by looking
at the gate pulse voltage of each switch. The results for the converter with full
wave rectifier are depicted in Appendix B and in Fig. 5.3.
According to the results, it is clear that the main dominant part of the total
losses is the conduction losses, especially at the secondary side because of higher
current rating. Therefore, a big improvement is achieved in reducing the losses and
consequently increasing the efficiency, if some methods are used to decrease the
conduction losses. One of these methods is paralleling the switches. If we consider
just one switch in Fig. 5.4 and write the equations for conduction losses for both
cases.
50
CHAPTER 5. RESULTS
250
250
IPW65 & SUM110
IPW65 & FDP027
IPW65 & Double SUM110
IPW65 & Double FDP027
200
200
150
150
Losses [W]
Losses [W]
100
50
0
0
100
50
20
40
60
80
100
0
0
20
40
60
80
100
Fig. 5.1: Total conduction converter losses for full wave and current doubler topologies
2
Pcon1 = Rdson1 .IM
OSrms1 + Vf 1 .IDavg1
(5.1)
IM OSrms2 =
IDavg2 =
IM OSrms1
2
(5.3)
IDavg1
2
(5.4)
Rdson2 = Rdson1
51
(5.2)
(5.5)
CHAPTER 5. RESULTS
70
60
60
50
Losses [W]
Losses [W]
50
40
30
30
20
20
10
10
0
0
40
20
40
60
80
0
0
100
20
40
60
80
100
Fig. 5.2: Total switching converter losses for full wave and current doubler topologies
Vf 2 = Vf 1
Pcon2 = 2[Rdson1 .(
(5.6)
IM OSrms1 2
IDavg1
2
) +Vf 1 .
] = 0.5Rdson1 .IM
OSrms1 +Vf 1 .IDavg1
2
2
(5.7)
So it means that by paralleling two switches and making a new module we have
managed to halve the on state resistance of the MOSFET and therefore reduce
the total conduction losses to a high extent. By this method some new results
were obtained, which imply lower conductions losses in the semiconductors and
consequently lower total losses in the device.
52
CHAPTER 5. RESULTS
250
250
200
150
Losses [W]
Losses [W]
200
100
50
0
0
150
100
50
20
40
60
80
100
0
0
20
40
60
80
Fig. 5.3: Total converter losses for full wave and current doubler topologies
53
100
CHAPTER 5. RESULTS
Comparing the waveforms in Fig. 5.1, 5.2 and 5.3, it must be noticed that the
loss difference is not that dramatic, but there is a small deviation in total losses
in both cases. This is justified that the switches used have different performance
characteristics. From Table A.2 and Fig 5.2 it is seen that switching losses for
STW69N65M5 has increased significantly after reducing the load to 30%. This
can be explained by the fact that the ZVS on the primary side is lost. After implementing the switch paralleling method the conduction losses in the secondary side
has been reduced by around 50 W for maximum load that is presented in Appendix
B and Fig. 5.3 with the dashed line, which is a quite significant fraction from the
initial result. Tables A.7, A.8 and Fig. 5.1 also show the changes in conduction
losses by paralleling two switches in the current doubler converter. Consequently,
the overall efficiency has been improved, but the present-worth analysis should be
carried out in order to check if the measure is cost-effective.
54
5.3. EFFICIENCY
5.3
CHAPTER 5. RESULTS
Efficiency
The following section represents the final results obtained from the simulations
with the real components in Allegro AMS. For the comparison reasons, both full
wave and current doubler topologies are plotted together in order to have explicit
understanding in the efficiency discrepancies.
96
96
94
94
92
92
Efficiency [%]
Efficiency [%]
All calculated efficiencies for all arrangements are depicted in Fig. 5.5 that have
been simulated in the project. It is noticeable that having two parallel switches
instead of one increases efficiency by around 1 %, which is significant improvement.
It is also seen that efficiency in the current doubler is greater than that in the full
wave set-up by around 1.5 %. This is a great achievement and it coincides with
the theory behind it.
90
88
86
84
82
0
90
88
86
STB57FDP8440
STW69FDP020
STB572*FDP8440
STW692*FDP020
50
Load level [%]
84
100
82
0
IPW65SUM110
IPW65FDP027
2*IPW652*SUM110
2*IPW652*FDP027
50
Load level [%]
Fig. 5.5: Efficiency plots for both full wave and current doubler arrangements
55
100
5.3. EFFICIENCY
CHAPTER 5. RESULTS
96.5
96.5
96
96
95.5
95.5
95
95
Efficiency [%]
Efficiency [%]
In Fig. 5.6 the two best of achieved efficiency curve are depicted for single switches
on the primary, single and double switches on the secondary sides. It is quite
clear that even paralleling only on the secondary side increases the total efficiency
substainally.
94.5
94.5
94
94
93.5
93.5
93
93
FWSingle FDP8440
CDSingle FDP027
92.5
0
50
Load level [%]
FWDouble FDP8440
CDDouble FDP027
100
92.5
0
50
Load level [%]
100
Fig. 5.6: Best efficiency plots for both full wave and current doubler arrangements
with single and double switches on the secondary side
In Fig. 5.7 it is seen that the maximum efficiency is achieved with placing two
switches in parallel on both sides of the converter. Apparently, this is logical since
paralleling decreases the on-state resistance of the MOSFET, hence the conduction
loss of the switch, which comprise a significant portion of the total converter losses.
56
CHAPTER 5. RESULTS
97
96.5
Efficiency [%]
96
95.5
95
94.5
94
93.5
FWDouble STB57 & Double FDP8440
CDDouble IPW65 & Double FDP027
93
0
10
20
30
40
50
60
Load level [%]
70
80
90
100
Fig. 5.7: Efficiency plots for both full wave and current doubler arrangements with
double switches on both sides
5.4
Cost estimation
Since it is not the most efficient way to utilize just a single switch arrangement, the
decision towards paralleling of the switches has been implemented. However, it has
to be checked if this measure is cost-effective. Therefore, a present worth analysis was carried out and the calculated results are presented in the following section.
Calculations were performed taking into consideration the load cycle of the converter, using current distribution throughout almost one hour that is depicted in
Fig. 5.8 with three different modes being driven, which are city, highway and
heavy cycles.
Fig. 5.9 shows the distribution plot which is obtained by the data from Fig. 5.8.
By having the distribution histogram the probability for different currents can be
57
CHAPTER 5. RESULTS
Velocity [km/h]
150
100
50
0
0
500
1000
1500
Time [s]
2000
2500
3000
500
1000
1500
Time [s]
2000
2500
3000
Current [A]
200
150
100
50
0
0
Fig. 5.8: Converter loading cycle, taken from the real test
obtained, which is shown in the lower plot of Fig. 5.9 that must also be considered
in order for the analysis to be more accurate.
After replacing the single switches with double ones on the secondary side, the
conduction losses have been decreased. The differences between the total losses
in two different cases, single and double switches at the secondary part of the
converter, have been plotted and shown in Fig. 5.10 and 5.11 for two different
topologies. It can be seen that the difference is around 15 W at 50% of the maximum load which can be considered as the converter most common operating point.
Concerning the loss differences and probability plots for different load currents
the normalized power loss differences can be obtained and can easily be converted
to the energy, which is shown in Fig. 5.12 and 5.13. This energy can be saved
throughout the operating time if the paralleling method is used instead of single
switches in the rectifier side of the converter. In this calculation it was assumed
58
CHAPTER 5. RESULTS
8000
6000
4000
2000
0
0
20
40
60
80
100
120
Current [A}
140
160
180
200
20
40
60
80
100
120
Current [A}
140
160
180
200
Probability
0.06
0.04
0.02
0
0
59
CHAPTER 5. RESULTS
50
40
30
20
10
0
10
0
50
100
150
200
Fig. 5.10: Total power loss difference between single and double switches at the
secondary side of the full wave converter
cost-effective. The switches prices are given in Table. 5.12 according to the suppliers websites. The future worth price analysis is based on Compound Interest
Factor (CIF) calculation that is shown in 5.8.
CIF = (1 + i)n
(5.8)
60
CHAPTER 5. RESULTS
50
40
30
20
10
10
0
20
40
60
80
100
120
Current [A]
140
160
180
200
Fig. 5.11: Total power loss difference between single and double switches at the
secondary side of the current doubler converter
500
450
400
350
300
250
200
150
100
50
0
0
20
40
60
80
100
120
Current [A}
140
160
180
200
Fig. 5.12: Energy saved throughout a year in the full wave converter
61
CHAPTER 5. RESULTS
350
300
250
200
150
100
50
0
50
0
20
40
60
80
100
120
Current [A]
140
160
180
200
Fig. 5.13: Energy saved throughout a year in the current doubler converter
Cost
Cost
Total cost
Saved energy
[$]
[SEK]
[SEK]
[SEK]
[SEK]
4,12
27,14
108,56
225,6
230,4
3,3
21,74
43,48
90,36
153
62
6
Circuit improvements
One of the aims of the following project was to investigate the performance of the
converter and suggest some improvements, if issues were identified. Therefore, the
following chapter is devoted to the description of the procedure that was applied
to get the presented results.
6.1
Spike killer
The whole analysis is carried out on both investagated converter topologies with
real components. It is clearly observed in Fig. 6.1 that during turn-off interval
voltage over the switch on the secondary side overshoots its reference value and
starts to oscillate. We think that this occurs because of the diode reverse recovery
current that is conducting prior to the switch. Diode snaps off, resulting in an
overshoot and causes a ringing effect inside the module, which brings parasitic
influence. This ringing may generate significant noise and it may be sustained for
many cycles after the spike has happened, which may damage the switch, consequently it has to be suppresed. Since the amount of time to suppress its oscillations
is not sufficient, steady-state cannot be reached, therefore those oscillations must
be damped. One way to damp them is to implement the RC snubber in parallel
to every switch module on the secondary side.
63
Fig. 6.2: Voltage over the switch after snubber was implemented, ringing effect
eliminated
64
7
Conclusions and Future work
7.1
Conclusion
The set aim of the project has been successfully fulfilled, models of two different
DC/DC converters utilized in an electrified vehicle were built and their performance investigated. Relevant analysis and comparison have been carried out and
presented in the report. Phase shifted switching pattern and leakage inductance
contributed significantly in achieving zero voltage switching, so that switching
losses were minimized. Losses in the converters have also been determined with
help of Matlab, where the calculation script was fully written. By using parallel
main switching components, less conduction losses were achieved, which was the
dominant part of total losses. The transformer design was carefully performed and
the core was appropriately selected, considering core and copper losses and the
condition of reducing the size and the volume of it.
After loss calculation, the efficiencies of both converters have been quantified and
they are in agreement with the theoretical justification behind it that the current
doubler topology is more efficient than the full wave one. Present worth analysis
has been conducted and proved that the paralleling of the switches is cost-effective
in both cases. Since some oscillations were observed in the simulation, a snubber
circuit was implemented to supress the spikes and smooth the voltage waveforms.
7.2
Future work
Concerning the future work that could be done as continuation of the project, we
think that the following tasks could be carried out:
65
66
Bibliography
BIBLIOGRAPHY
[10] H.-J. Chiu, L.-W. Lin, A phase-shifted zero voltage transition full-bridge converter with current doubler synchronous rectification, Annual conference in
Sapporo 6.
[11] D. Gautam, F. Musavi, M. Edington, W. Eberle, W. Dunford, A zero voltage
switching full-bridge dc/dc converter with capacitive output filter for a plugin-hybrid electric vehicle battery charger, Delta-Q Technologies 6 (2012) 1381
1386.
[12] T. Mishima, Y. Ooue, Y. Fukumoto, M. Nakaoka, A secondary-side phaseshifting zvs-pwm dual full bridge dc/dc front-end converter for plasma rf power
generator, IEEE Power Electronics 6.
[13] S. Waffler, J. W. Kolar, A novel low-loss modulation strategy for high-power
bi-directional buck-boost converter, IEEE Power Electronics 6.
[14] H.-J. Chiu, L.-W. Lin, A bidirectional dc/dc converter for fuel cell electric
vehicle driving system, IEEE Transactions on power electronics 9 (2006) 950
958.
[15] N. H. Kutkut, A full bridge soft switched telecom power supply with a current
doubler rectifier, IEEE 8 (1997) 344354.
[16] S. Mappus, Control driven synchronous rectifiers in phase shifted full bridge
converters, Tech. rep., Texas Instruments (march 2003).
[17] E.-H. Kim, B.-H. Kwon, Zero-voltage- and zero-current switching full bridge
converter with secondary resonance, IEEE Transactions on power electronics
9 (2010) 10171025.
[18] H. Mao, O. A. Rahman, I. Batarseh, Zero-voltage switching dc/dc converters
with synchronous rectifiers, IEEE Transactions on power electronics 10 (2008)
369378.
[19] C. Henze, H. Martin, D. Parsley, Zero-voltage switching in high frequency
power converters using pulse width modulation, IEEE 8 (1988) 3339.
0
68
BIBLIOGRAPHY
Appendices
69
A
Results - Tables
In this chapter the explicit results from the simulations with real components are
presented. All calculations were performed by the devised script within the project.
Table A.1: Converter losses in [W] for fullwave at 100 kHz with STB57N65M5 as
primary switches and FDP8440 as secondary switches
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
13
7,350
0,511
0,672
15,724
0,529
3,127
27,914
406,981
18
4,434
0,617
0,896
20,808
0,689
3,216
30,660
531,062
23
4,906
0,767
1,280
27,213
0,884
3,342
38,392
684,943
28
5,042
0,954
1,780
34,968
1,082
3,397
47,223
831,664
35
4,920
1,175
2,418
43,730
1,317
3,583
57,143
1023,005
40
5,082
1,390
3,224
54,293
1,548
3,678
69,215
1196,276
50
5,081
1,736
4,781
73,721
1,926
3,813
91,058
1470,640
60
5,142
2,100
6,739
96,411
2,331
4,030
116,753
1774,646
69
5,505
2,377
8,351
113,700
2,644
4,274
136,852
2030,427
79
5,428
2,729
10,480
136,475
3,000
4,525
162,637
2314,128
89
5,537
3,101
12,890
161,807
3,374
4,804
191,513
2617,437
100
5,720
3,532
14,962
188,129
3,793
5,070
221,206
2937,920
Where Psw,tot,P is the total switching losses in the primary side, Psw,tot,S is the
total switching losses in the secondary side, Pcond,tot,P is the total conduction losses
in the primary side, Pcond,tot,S is the total conduction losses in the secondary side,
Pcopper is the copper losses in the transformer, Pcore is the core losses in the transformer, Pout is the output power and the load level is expressed in percentage of
70
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
13
59,680
0,513
0,633
16,110
0,528
3,127
80,591
404,583
18
46,291
0,627
0,760
21,509
0,686
3,216
73,090
527,484
23
30,234
0,769
0,965
28,333
0,881
3,342
64,524
681,033
28
17,859
0,906
1,249
36,121
1,080
3,397
60,613
829,401
35
6,551
1,083
1,747
46,864
1,336
3,546
61,127
1030,770
40
7,209
1,242
2,181
56,081
1,549
3,659
71,922
1198,245
50
7,500
1,562
3,154
74,946
1,939
3,813
92,915
1490,679
60
7,386
1,903
4,273
93,654
2,280
4,030
113,525
1761,436
69
7,791
2,241
5,462
113,318
2,624
4,274
135,710
2041,507
79
7,953
2,542
6,840
135,435
2,975
4,525
160,271
2329,564
89
7,964
2,847
8,338
157,738
3,310
4,804
185,002
2619,003
100
7,738
3,230
10,127
185,810
3,695
5,070
215,670
2932,378
Table A.3: Current Doubler converter losses in [W] at 100 kHz with IPW65R080CFD
as primary and SUM110N06 as secondary switches
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
15,635
7,592
1,571
1,571
0,319
1,381
28,069
246,384
14
10,395
8,167
2,408
4,007
0,542
1,482
27,000
418,222
20
13,118
7,418
3,501
7,501
0,764
1,586
33,888
589,487
28
16,281
7,807
5,563
14,330
1,083
1,748
46,811
835,710
36
18,039
8,656
7,964
22,760
1,388
1,919
60,726
1073,818
42
18,480
9,504
9,869
29,616
1,596
2,037
71,102
1235,314
48
18,554
11,553
12,134
37,881
1,819
2,158
84,099
1403,755
56
18,414
12,842
15,545
50,594
2,123
2,347
101,865
1639,326
64
18,191
15,612
19,368
65,350
2,434
2,544
123,499
1876,868
68
18,214
18,844
21,695
74,331
2,609
2,680
138,373
2021,544
74
18,173
22,135
24,715
86,313
2,830
2,819
156,984
2183,773
84
18,008
27,789
30,532
109,929
3,230
3,108
192,594
2491,040
100
18,764
36,908
39,808
148,764
3,825
3,568
251,635
2947,893
71
Table A.4: Current Doubler converter losses in [W] at 100 kHz with IPW65R080CFD
as primary and FDP027N08B as secondary switches
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
15,514
0,697
1,577
1,675
0,355
1,381
21,199
274,596
15
10,419
0,558
2,383
3,920
0,568
1,482
19,329
439,574
21
12,947
0,536
3,480
7,118
0,786
1,586
26,452
608,230
29
16,192
0,346
5,564
13,517
1,105
1,748
38,472
855,004
37
18,066
0,257
8,086
21,650
1,419
1,919
51,396
1097,964
43
18,521
0,312
10,082
28,383
1,634
2,037
60,970
1262,956
48
18,168
0,345
12,063
35,019
1,823
2,158
69,577
1410,262
56
18,463
0,362
15,476
47,024
2,129
2,347
85,801
1648,889
64
18,282
0,397
19,324
60,723
2,440
2,544
103,711
1886,005
69
18,195
0,342
21,846
70,055
2,633
2,680
115,752
2037,728
74
18,081
0,404
24,654
80,532
2,836
2,819
129,326
2193,634
84
18,017
0,474
30,489
102,896
3,236
3,108
158,220
2501,381
92
18,127
0,558
35,106
121,169
3,538
3,334
181,833
2731,863
100
18,615
0,646
39,769
140,031
3,832
3,568
206,462
2959,688
Table A.5: Converter losses in [W] for full wave at 100 kHz with STB57N65M5 as
primary switches and eight switches of FDP8440 in the secondary side
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
13
8,505
0,315
0,630
14,260
0,517
3,181
27,408
403,635
18
4,911
0,373
0,901
19,379
0,694
3,216
29,474
539,161
23
5,146
0,431
1,261
24,889
0,882
3,270
35,879
682,717
28
5,087
0,534
1,786
31,766
1,101
3,306
43,580
843,219
35
4,990
0,639
2,340
37,585
1,317
3,490
50,361
1023,943
40
5,260
0,729
3,124
45,604
1,543
3,546
59,806
1188,270
50
5,102
0,902
4,755
60,695
1,938
3,678
77,070
1477,785
60
5,260
1,065
6,537
74,158
2,314
3,931
93,266
1782,532
69
5,026
1,202
8,393
89,105
2,663
4,070
110,460
2037,376
79
5,358
1,346
10,725
105,243
3,032
4,274
129,978
2322,839
89
5,306
1,525
13,203
122,951
3,420
4,482
150,888
2618,364
100
5,584
1,698
15,849
138,753
3,783
4,804
170,472
2944,207
72
Table A.6: Converter losses in [W] for fullwave at 100 kHz with STW69N65M5 as
primary switches and eight switches of FDP020N6B in the secondary side
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
13
61,055
0,311
0,583
14,208
0,501
3,145
79,804
387,333
18
44,404
0,424
0,739
19,853
0,690
3,216
69,324
532,719
23
30,841
0,451
0,966
25,890
0,876
3,270
62,293
673,702
28
16,753
0,562
1,282
33,199
1,096
3,306
56,198
835,476
35
6,298
0,656
1,734
40,827
1,343
3,490
54,348
1037,149
40
7,315
0,691
2,180
47,544
1,545
3,565
62,840
1191,917
50
7,640
0,855
3,080
59,731
1,900
3,755
76,962
1474,447
60
7,471
1,030
4,335
76,000
2,306
3,872
95,014
1770,192
69
7,554
1,169
5,607
90,585
2,665
4,030
111,609
2041,052
79
7,626
1,294
6,989
105,024
3,009
4,274
128,216
2327,839
89
7,738
1,464
8,601
121,024
3,374
4,482
146,682
2617,065
100
7,793
1,598
10,279
135,946
3,731
4,804
164,152
2948,298
Table A.7: Current Doubler converter losses in [W] at 100 kHz with double switches,
IPW65R080CFD as primary and SUM110N06 as secondary switches
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
15,635
7,592
0,786
0,914
0,319
1,381
26,627
246,384
14
10,395
8,167
1,205
2,387
0,542
1,482
24,177
418,222
20
13,118
7,418
1,753
4,505
0,764
1,586
29,143
589,487
28
16,281
7,807
2,784
8,641
1,083
1,748
38,345
835,710
36
18,039
8,656
3,988
13,760
1,388
1,919
47,749
1073,818
42
18,480
9,504
4,941
17,949
1,596
2,037
54,507
1235,314
48
18,554
11,553
6,077
23,004
1,819
2,158
63,165
1403,755
56
18,414
12,842
7,785
30,845
2,123
2,347
74,356
1639,326
64
18,191
15,612
9,699
40,023
2,434
2,544
88,504
1876,868
68
18,214
18,844
10,867
45,589
2,609
2,680
98,803
2021,544
74
18,173
22,135
12,383
53,104
2,830
2,819
111,443
2183,773
84
18,008
27,789
15,294
68,036
3,230
3,108
135,464
2491,040
100
18,764
36,908
19,946
92,888
3,825
3,568
175,898
2947,893
73
Table A.8: Current Doubler converter losses in [W] at 100 kHz with double switches,
IPW65R080CFD as primary and FDP027N08B as secondary switches
load [%]
Psw,tot,P
Psw,tot,S
Pcond,tot,P
Pcond,tot,S
Pcopper
Pcore
Ptot,loss
Pout
15,514
0,697
0,789
1,082
0,355
1,381
19,817
274,596
15
10,419
0,558
1,192
2,535
0,568
1,482
16,754
439,574
21
12,947
0,536
1,742
4,604
0,786
1,586
22,200
608,230
29
16,192
0,346
2,785
8,761
1,105
1,748
30,938
855,004
37
18,066
0,257
4,050
14,061
1,419
1,919
39,771
1097,964
43
18,521
0,312
5,050
18,479
1,634
2,037
46,033
1262,956
48
18,168
0,345
6,038
22,874
1,823
2,158
51,407
1410,262
56
18,463
0,362
7,751
30,831
2,129
2,347
61,883
1648,889
64
18,282
0,397
9,682
39,977
2,440
2,544
73,322
1886,005
69
18,195
0,342
10,942
46,259
2,633
2,680
81,052
2037,728
74
18,081
0,404
12,350
53,367
2,836
2,819
89,858
2193,634
84
18,017
0,474
15,276
68,625
3,236
3,108
108,736
2501,381
92
18,127
0,558
17,588
81,209
3,538
3,334
124,354
2731,863
100
18,615
0,646
19,926
94,309
3,832
3,568
140,896
2959,688
74