A Battery Equalisation System for

Electric Vehicle

A thesis submitted in partial fulfillment

of the requirements for the degree of
Master of Engineering
at the
University of Canterbury
Christchurch, New Zealand

Department of
Electrical and Computer Engineering

By
Ming-Kuang(Leo) HSIEH, BE (Hons)
09 July 2007

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
Abstract

In 1999, the Electrical and Computer Engineering Department at the University of Canterbury
started building their third electric vehicle (EV3) based on a TOYOTA MR2 with the goal of
building a higher performance vehicle to match present combustion engined vehicles. The car
is powered by 26 12volt sealed lead-acid batteries connected in series to achieve a nominal
312V DC source.

A battery voltage equaliser is a device that draws energy from a higher charged battery, then
discharges into a lower charged battery. The need for a voltage equaliser is principally due to
the differences in cell chemistry, temperature gradients along the battery string and the ages of
the batteries. During the charging or discharging process, some batteries reach their nominal
voltage or reach deep discharge states before the others. Then if the charger keeps charging
the batteries or the load keeps drawing energy from these batteries, it results in damage to the
batteries. Therefore maintaining the charge level on each battery becomes important. In
addition, it also improves the battery life and vehicle travelling range.

This thesis details the analysis of three different types of battery equaliser, which are based on
a 24W buck-boost converter, 192W buck-boost converter and 192W flyback converter. In this
design, all converters are designed to work under current mode control with average of 2A. To
make each converter install without significant effect on the performance and the cost, each
converter is also built with the goals of being small, lightweight, cost effective, flexible for
mounting, maintenance free and highly efficient.

At the end, the prototype battery equalisation converters were designed, constructed and
tested, and the efficiencies from each converter are measured around 90 ~ 92%. The
experimental results show two banks of series connected batteries can be successfully
equalised by the designed equaliser. This thesis covers the design, simulation and the
construction procedures of this battery equaliser system, and also details on some
considerations and possible future improvement that were found during the experimental test.

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

Acknowledgment

First and foremost, I would like to express my gratitude to my supervisors, Associate

Professor Richard Duke and Dr. Simon Round for their advice, guidance, encouragement and
being very patient throughout this project.

I would also like to thank all the members in the Power Electronics Research Group (William
Chen, Si-Kuok Ting, Irene Ting, John and Ari) for giving me good advice and being as a good
family in this research group. Special thanks to technician Ron Battersby, Ken Smart, Dudley
Berry, Scott Lloyd and Nick Smith in the laboratory and sharing their practical experiences.

To my family in Taiwan and in New Zealand, I like to express my appreciation to all my

family members, especially my parents and brother in Taiwan who supported me during my
entire study period. For my boarding family in New Zealand, Brian Baker and Veronica Baker,
who always look after my living and health over the years. Last but no least, I would also like
to say thank you to Pastor Albert Tang for being a great teacher, a brother and being a great
spiritual supporter to me.

ii

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Table of contents

Abstract…………………………………………………………………………………i
Acknowledgment……………………………………………………………………ii
Table of contents……………………………………………………………………iii
Table of Figures………………………………………………………………………v
Table of Tables………………………………………………………………………vii
Publication arising from this thesis……………………………………………viii

1 Introduction……………………………………………………………1
1.1. Project overview………………………………………………………………1
1.2. Thesis structure………………………………………………………………4

2 Battery Monitoring and Equaliser Topologies…………………………6

2.1. Battery monitoring……………………………………………………………7
2.2. Common bus topology………………………………………………………7
2.3. Common core topology……………………………………………………10
2.4. Ring topology…………………………………………………………………11
2.5. Selection of voltage equaliser configuration for EV3……………………13

3 Converter design………………………………………………………17
3.1. The 24W buck-boost converter……………………………………………17
3.1.1. The PWM controller……………………………………………………18
3.1.2. Average current mode control……………………………………………19
3.1.3. The gate drive……………………………………………………………20
3.1.4. The power converter of the 24W buck-boost converter…………………21
3.2. The 192W buck-boost converter……………………………………………31
3.3. The 192W flyback converter………………………………………………32
3.3.1. PWM controller for the 192W flyback converter………………………33
3.3.2. 192W flyback transformer………………………………………………34
3.3.3. Reducing the voltage stress on MOSFETs………………………………37
3.4. Summary……………………………………………………………………39
iii

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

4 Converters simulations…………………………………………………40
4.1. 24W buck-boost converter…………………………………………………41
4.2. 192W buck-boost converter…………………………………………………43
4.3. 192W flyback converter……………………………………………………45
4.4. Summary……………………………………………………………………48

5 The system construction………………………………………………49

5.1. The 24W Buck-boost converter……………………………………………49
5.1.1. Inductor design…………………………………………………………49
5.1.2. MOSFET selection………………………………………………………52
5.2. The 192W Buck-boost converter……………………………………………54
5.3. The 192W Flyback converter………………………………………………55
5.4. Summary………………………………………………………………………59

6 Performance records……………………………………………………57
6.1. Battery equalisation between non-isolated banks………………………59
6.2. Battery equalisation between isolated banks……………………………63
6.3 Summary………………………………………………………………………58

7 Conclusion……………………………………………………………70

Appendix …………………………………………………………………73
The schematics of 24W buck-boost converter…………………………………74
The schematics of 192W buck-boost converter………………………………75
The schematics of 192W flyback converter……………………………………76
The PCB of 24W buck-boost converter…………………………………………77
The PCB of 192W buck-boost converter………………………………………78
The PCB of 192W flyback converter……………………………………………79
The data sheet of E32 planar inductor……………………………………….....80
The data sheet of E43 planar inductor …………………………………………81
The data sheet of SUD35N05-26L MOSFET…………………………………82
The data sheet of IRF740 MOSFET……………………………………………84

iv

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Table of Figures

Figure 1.1. Battery location………………………………………….................2

Figure 2.1. The common bus equaliser…………………………………………8

Figure 2.2. The common core equaliser………………………………………………10

Figure 2.3. The ring equaliser…………………………………………………………11

Figure 2.4. The proposed voltage equaliser structure………………………………….13

Figure 2.5. The efficiency comparison of buck-boost and Cuk converters……………14

Figure 2.6. The proposed voltage equaliser structure…………………………………15

Figure 3.1. The Buck-boost converter…………………………………………17

Figure 3.2. The block diagram of SG3526……………………………………………18

Figure 3.3. The average current mode control circuit…………………………………19

Figure 3.4. The RM core power rating curve.…………………………………………23

Figure 3.5. 3C90 B-H curve……………………………………………………………25

Figure 3.6. The construction of the planar inductor……………………………………27

Figure 3.7. The track width versus temperature rise diagram…………………………28

Figure 3.8. The construction of the integrated inductor………………………………29

Figure 3.9. The flyback converter………………………………………………………32

Figure 3.10. The PWM controller for the 192W flyback converter.…………………33

Figure 3.11 Flyback converter with clamping circuit…………………………………38

Figure 4.1. 24W non-isolated buck-boost converter……………………………41

Figure 4.2. The switching waveforms of the 24W buck-boast converter.……………42

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

Figure 4.3. 24W buck-boost converter equalisation.…………………………………43

Figure 4.4. 192W buck-boost converter………………………………………………44

Figure 4.5. 192W buck-boost converter switching waveforms.………………………44

Figure 4.6. 192W buck-boost converter equalisation…………………………………45

Figure 4.7. The schematic of the 192W flyback converter……………………………46

Figure 4.8. 192W flyback converter switching waveforms……………………………46

Figure 4.9. 192W flyback converter equalisation.……………………………………47

Figure 5.1. RM, planar inductor and integrated inductors efficiency comparison……50

Figure 5.2. The inductor for the 24W buck-boost converter…………………………52

Figure 5.3. The photo of the 24W Buck-boost battery equaliser………………………53

Figure 5.4. 24W buck-boost converter action waveforms……………………………53

Figure 5.5. The 192W buck-boost converter…………………………………………55

Figure 5.6. The frequency response curve of the 192W flyback converter.……………56

Figure 5.7 The switching waveform with/without the clamping circuit………………57

Figure 5.8 The photo of the 192W flyback converter…………………………………57

Figure 5.9. The switching waveform of the 192W flyback converter…………………58

Figure 6.1. The equalisation setup……………………………………………60

Figure 6.2. Equalisation procedure……………………………………………………62

Figure 6.3. The setup diagram for non-isolated bank equalisation……………………63

Figure 6.4. Battery equalisation process for an eight series connected batteries………65

Figure 6.5. The setup diagram for isolated bank equalisation……………………66

Figure 6.6. Equalisation process for the isolated battery banks………………………68

vi

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Table of Tables

Table 2.1. Comparison of equaliser topologies………………..............................13

Table 4.1. The open-circuit voltages after 30A of 60 minutes discharge………………40

Table 6.1. The measured battery open-circuit voltages………………………………63

Table 6.2. The measured battery open-circuit voltages………………………………67

vii

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

Publication arising from this thesis
Ming-Kuang(Leo) HSIEH, Simon Round, Richard Duke, “An Investigation Of Battery
Voltage Equalization Topologies For An electric Vehicle”, accepted to be published at
the Australian Universities Power Engineering Conference (AUPEC 2004).
Christchurch, New Zealand, September 2004

viii

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 1: Introduction

1. Introduction

1.1. Project overview

The first electric car was invented by Thomas Davenport in 1834. At that time, the battery
used in the car was not rechargeable. In the next few decades, numbers of electric cars and
rechargeable battery technologies were developed, which brought more attention to
investigate electric vehicle (EV) development. In the late 1890s EVs outsold gasoline cars
ten to one. EVs dominated the roads and dealer showrooms. Some automobile companies,
like Oldsmobile and Studebaker actually started out as successful EV companies, only later
transitioning to gasoline-powered vehicles. In fact, the first car dealerships were exclusively
for EVs[1].

Early production of EVs, like all cars, was accomplished by hand assembly. In 1910,
volume production of gasoline powered cars was achieved with the motorised assembly
line. This breakthrough manufacturing process killed off all but the most well-financed car
builders. Independents, unable to buy components in volume died off. The infrastructure for
electricity was almost non-existent outside of city boundaries, limiting EVs to city-only
travel. Another contributing factor to the decline of EVs was the addition of an electric
motor (called the starter) to gasoline powered cars, finally removing the need for the
difficult and dangerous crank to start the engine. Due to these factors, by the end of World
War I, production of electric cars stopped and EVs became niche vehicles, serving as taxis,
trucks, delivery vans, and freight handlers[1].

In the late 1960s and early 1970s, there was a rebirth of EVs, which was prompted by
concerns about air pollution and the OPEC oil embargo. In the early 1990s, a few major
automakers resumed production of EVs, prompted by California’s landmark Zero Emission
Vehicle (ZEV) mandate. Those EVs were produced in very low volumes, essentially hand-
built like their early predecessors. However, as the ZEV mandate was weakened over the
years, the automakers stopped making EVs and Toyota was the last major manufacturer to
stop EV production in 2003[1].

Page 1

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 1: Introduction

In the past thirty years, the University of Canterbury has also been investigating and
developing its own EVs. The original EV was developed to test motor speed control ideas
for AC induction motors.[2] In 1982, the second EV was developed, and that was based on
a Austin Farina and powered by 20 series connected lead-acid batteries to form a nominal
240V dc bus. In 1999 the development of the third EV started with the goal of building a
higher performance vehicle to match present combustion engined vehicles.

The third electric vehicle (EV3) produced by the Electrical and Computer Engineering
Department at the University of Canterbury is based on a TOYOTA MR2. It is powered by
26 Hawker Genesis 12volt 26A-hour sealed lead-acid batteries connected in series to
achieve a nominal 312V DC source and these batteries are conveniently divided into four
banks by the constraints of their location in the car. The first bank consists of 8 batteries,
which are located under the front bonnet of the MR2. The second bank consists of 6
batteries and is placed in the engine bay at the back. The third and the fourth banks consist
of 6 batteries each and are located in the rear boot. These battery locations are shown in
Figure 1.1.

Figure 1.1. Battery location

Page 2

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 1: Introduction

Maintenance of cells at an equalised charge level is critical for enhancing battery life[1].
The need for a voltage equaliser is principally due to the differences in cell chemistry,
temperature gradients along the battery string and the ages of the batteries. During the
charging process, some batteries will consequently reach full charge before others and
before the overall battery terminal voltage reaches its nominal value[1]. Therefore, if the
charger continues to charge the remaining batteries, it would result in overheating the fully
charged batteries, thus reducing their useful life. The same principle can also be applied to
the discharging process. Any over-discharge would lead a battery into deep discharge,
which can also reduce the life of the battery and decrease the travelling range of the electric
vehicle.

The idea of a basic battery equaliser is to balance the charge level of two batteries by
drawing energy from the one with the higher charge and transferring it to the other. To
achieve this, a high frequency dc-dc converter is used. In power electronics, every
converter has its own energy storage, which can be an inductor, a capacitor, a transformer
or some combination of these. By controlling the switching signal, this energy storage
capacity can be charged from the source and then discharged to the load. In a battery
equaliser, the overcharged battery can be considered as the source and the undercharged
battery as the load.

The objective of this thesis is to investigate various battery equalisation topologies that
could be implemented into the EV3 to equalise the charge level of each individual battery
within the entire battery string. Since there is a total of 26 batteries that need to be equalised,
a number of equalisers have to be built for the entire battery string. Therefore, the design of
each equaliser must be small, lightweight, cost effective, easy to interface, flexible for
mounting, maintenance free and highly efficient.

An additional complication arises from the fact that these 26 batteries are not all located in
a single compartment of the vehicle. In addition to having an equaliser capable of
transferring energy between adjacent batteries, energy must also be able to be transferred
between the battery banks (Figure 1.1) and between the top and bottom of the entire string.

Page 3

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 1: Introduction

1.2. Thesis structure

In this thesis; various types of voltage equaliser topologies, which include the common core,
common bus and ring equalisers are investigated. In Chapter two the principles of each
equaliser topology are described. It demonstrates how the dc-dc converter can be used to
interface between two batteries or two banks of batteries and then transfer the energy from
one to the other. To design the most suitable equaliser for the EV3, the advantages and
disadvantages of each topology, based on this construction, cost and future upgradeability
are listed.

In Chapter three the design procedures of the dc-dc converters making up the equaliser are
detailed. To minimise the converter size and to increase the manufacturability, various type
of inductors, such as RM core, planar core and integrated inductor are also investigated. In
this chapter, all electric specifications including the power rating, current ripple and the
strategy of the control circuit of each converter are also defined.

In chapter four the simulations of all dc to dc converters making up the chosen equaliser
topology are described. To achieve realistic results of simulations, discharging test
measurements over eight series connected lead-acid batteries was first carried out. The
result of these tests gives an indication of the possible voltage variations for each battery.
The simulation is carried out by Pspice simulator and each rechargeable battery is modelled
as a 1F capacitor. The simulation results demonstrate that the selected converters can be
successfully used in the battery equalisation system.

In Chapter five the construction of each converter and their control circuits are detailed. In
this battery equalisation system, each converter is designed to operate at a fixed current rate
of 2A. Therefore current mode control is implemented into every converter. In the design of
the power converter, the performance comparison of the RM cored inductor, the integrated
inductor and the planar inductor are also investigated in order to determine the best solution
for each converter to meet the requirement of light weight and low profile format.

Page 4

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 1: Introduction

In Chapter six, two sets of equalisation results for the battery equalisation system are
obtained. These equalisation tests are based on equalising two banks of four series
connected batteries. Before each equalisation test, all batteries are equally charged to 13V.
In the first equalisation test, the two banks are connected in series, which forms an eight
series connected batteries string. Then the equalisation of individual batteries is done by
seven buck-boost converters after the entire string battery has been discharged for an hour
at a rate of 30A. After that, the 192W non-isolated buck-boost converter equalises the two
banks of batteries to ensure each bank has an equal charge level. In the second equalisation
test, the two banks of four series connected batteries are isolated from each other and the
equalisation between each bank is done by an isolated flyback converter.

In Chapter seven, the conclusion finalises the overall design process and summarises the
outcome. The difficulties and some considerations from the design of each converter are
also described. In addition, some possible future investigations are also pointed out.

References:
[1] Electric Vehicle History. May 2005. Electric Auto Association (EAA),
http://www.eaaev.org/Flyers/eaaflyer-evhistory.pdf
[2] Richard Duke, “Construction and Performance of an Electric Toyota MR2”,
Departmental seminar, 2005

Page 5

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

2. Battery Monitoring and Equaliser

Topologies

The idea of a battery equaliser is to balance the charge level of two or more batteries by
drawing energy from the one with the higher charge and then discharging to a lower level
battery. The most efficient means of achieving this transfer is by using a high frequency dc-
dc converter. In power electronics, every converter has an energy storage, which may be an
inductor, a capacitor, a transformer or some combination of these. By controlling the
converter’s switching signal, this energy storage capacity is normally charged from the
source then discharged to the load. In a battery equaliser the overcharged battery can be
considered as the source and the undercharged battery as the load.

To design the most suitable voltage equaliser for the EV3, there are a number of concerns
that have to be addressed. Due to the limited space in the car, the battery equaliser has to be
compact. Further, since there are 26 lead-acid batteries in the car, a large number of battery
equalisers are required in order to efficiently balance each individual battery. This makes
the cost, construction, manufacturability and efficiency of the battery equaliser important
aspects to consider.

The batteries used in the EV3 are rated at 26Ahr, and the equalisation rate is proportional to
the operational current of the equalisation system. Unfortunately higher operational current
brings larger power losses from the system, which results in increased switching losses. In
order to achieve a reasonably fast equalisation rate and to minimise the power loss from the
converter, the average operational current for this equalisation system is set to 2A.

In this Chapter, Section 2.1 describes two common battery monitoring techniques; the
coulometric and the open-circuit battery monitoring techniques, and how these techniques
could be implemented into the EV3. In Sections 2.2, 2.3 and 2.4; three battery equalisation
topologies common bus, common core and ring are detailed. Their construction, the
component counts, and the operating principles are also discussed. In Section 2.5 a
summary of these three topologies is given, and a decision is made to identify the best
solution for constructing a suitable battery equalisation network for the EV3.

Page 6

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

2.1. Battery monitoring

The maintenance of cells at an equalised charge level is critical for enhancing battery life[1].
There are numbers of ways to monitor the charge level of a lead-acid battery, and the most
common techniques are Coulometric measurement and open-circuit voltage measurement.
Coulometric measurement counts the ampere-hours either coming out of or going into the
battery bank. In its most basic form the battery capacity is assumed to be fixed, and then a
sensor has to be used for every battery in order to determine how much energy has been
drawn from or has flowed into the battery. In reality the total battery capacity varies with
the discharge current, the type of discharge, temperature and the age of the battery[2].
Open-circuit voltage can be used to determine the state of charge and is more suited to
battery monitoring in an electric vehicle, since the open-circuit voltage can be measured
directly from standard battery terminals. The open-circuit voltage of a sealed lead-acid
battery also relates directly to the battery’s state of charge[3]. The main drawback of the
open circuit voltage monitoring technique is that the open-circuit voltage must stabilise
before a reliable measurement can be made, and this can take from half an hour to several
hours depending on the type of battery[3].

2.2. Common bus topology

The common bus equaliser topology is shown in Figure 2.1. There are two different types
of common bus equaliser topologies. For both common bus equaliser topologies; energy
transformation for each battery is done using an isolated DC-DC converter, which can be a
flyback, push-pull, half-bridge or full-bridge converter.

Page 7

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

a b
Figure 2.1. The common bus equaliser

In the first common bus equaliser shown in Figure 2.1.a, the temporary energy storage is
the common bus, which can be made up from a capacitor bank or a separate rechargeable
battery. Energy transformation between any batteries in the series connected battery string
is done by using the isolated converter to draw the energy from the higher charged battery,
store the energy onto the common bus, and then using the other isolated converter to
discharge this energy into the lower charged battery[1].

In the second common bus equaliser shown in Figure 2.1.b, there is no intermediate energy
storage unit. When one battery is overcharged; the associated isolated converter will draw
the energy from this overcharged battery, and then recharge the entire battery string [1]. On
the other hand, if one battery is under-charged, the battery equalisation system would take
the energy from the whole battery string to recharge the under-charged battery via its
associated isolated converter.

Both common bus topologies would be good for a single long string of batteries, because
the converters are all individually attached to the common bus, and the batteries and
common bus are isolated from each other. Therefore only two steps are required to transfer
energy between any two batteries. Probably the most negative aspect of this topology is that
each and every converter requires a transformer to provide electrical isolation between the
battery and the common bus.

Page 8

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

Comparing these two common bus equalisers, the first common bus equaliser shown in
Figure 2.1.a has the advantage of flexibility of the transformer ratio and is better for future
expansion. Since the voltage rating on each side of the transformer is identical, the
transformer ratio of the primary and secondary windings can be made as 1:1, which means
the transformer ratio is not affected by the total number of the series string battery. So if
any battery is required to be added to or removed from the electric vehicle, this common
bus topology is very flexible for future expansion. However the major drawback of this
common bus equaliser topology is due to the extra set of capacitors or battery used on the
common bus side as the temporary energy storage, which brings an additional cost.

The second common bus equaliser does not require any additional capacitor or battery bank
as the temporary energy storage, therefore the component count is less than the first
common bus topology. Although this equaliser has the advantage of a low component count,
the transformer primary to secondary turns ratio for each converter has to be kept as 1:N,
where N is the total number of batteries in the whole series string. Therefore if the total
number of batteries is large, the weight of the transformer would be relatively heavier
compared with the first common bus topology, because the number of turns on one side of
the transformer has to be N times larger than the other side. This may also require the
converter to use larger transformer, since these extra turns require more winding space. The
other disadvantage of this common bus topology is because the turns ratio N has to be made
equal to the total number of batteries, therefore this restricts the possibility of future
changes such as adding some extra batteries to or withdrawing some battery from the
battery string. Further, since one side of the transformer is always connected to the high
voltage battery string and if batteries are spread in different compartments of the vehicle,
the high common bus voltage would be required to be routed throughout the vehicle, which
brings an additional safety concern.

Page 9

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

2.3. Common core topology

The common core topology is shown in Figure 2.2. Compared with the common bus
topology, the common core topology uses identical numbers of isolated dc-dc converters to
equalise every battery. The difference is that in the common core topology all windings
have to be coupled to the common core, which is the energy storage unit in this topology.
The principle of this topology is that once the overcharged battery is detected; the converter
will charge the common core by drawing the energy from the overcharged battery. Then
distributing the stored energy in the common core to every battery along the series
connected string battery via the diode, where the largest portion of the stored energy will be
directed to the lowest voltage battery without any additional control [1].

Figure 2.2. The common core equaliser

The common core topology is a good solution for a long string of batteries. However the
problem with this topology is that this scheme has a fairly high sensitivity to the leakage
inductance between secondary windings. Any slight mismatch of the secondary windings
would lead the charge on each battery to remain unbalanced after the equalisation process
[1]. To minimise this problem; the obvious requirement is that all batteries must be located
in the same compartment and the transformer winding all have to be wound on the same

Page 10

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

core in order that all the secondary windings have identical inductance. In the EV3; the 26
batteries are divided into four banks and placed in three different compartments, which
makes the common core topology difficult to implement. The other problem with this
topology is that when dealing with large numbers of batteries in the string, a larger
transformer core is required, which makes the equaliser difficult to mount, especially if
space is a concern [1].

2.4. Ring topology

The ring equaliser topology is shown in Figure 2.3. In this topology, the entire series string
of batteries is considered as a ring. Within the string of batteries; every two adjacent
batteries are linked by a non-isolated dc-dc converter and where the top and the bottom
batteries are linked by an isolated converter to overcome the potential difference [1].

Figure 2.3. The ring equaliser

The principal attraction of this topology over the other topologies is that only one converter
needs to be isolated. The non-isolated converters can be constructed in compact,
lightweight and low profile formats. However the disadvantage of this topology is because
each converter can only transfer the energy to or from adjacent batteries. Therefore for any
two batteries separated a few batteries away from each other, the equalisation system has to
draw the energy from the higher charged battery, and then transfer the energy through each
individual battery between them to charge the lower charged battery, which can make the
overall system inefficient.

Page 11

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

The non-isolated converters used in this topology could be either buck-boost or Cuk
converters. In terms of manufacturability, the buck-boost converter has the advantage of a
low component count, therefore the cost and the size of the converters can be kept to a
minimum. By comparing the converter efficiency, due to the lower current ripple, the Cuk
converter has higher efficiency than the buck-boost converter, which is important for high
power transformation. To choose the most suitable non-isolating converter for the ring
equaliser is dependent on how much power the converter needs to transfer. If the power
rating is small, the extra power loss from the buck-boost converter could be negligible, and
using fewer components could also reduce the manufacturing cost.

The isolated converter in this equaliser is used to provide electrical isolation and achieve
energy transformation between the top and the bottom batteries of the string. Unlike the
previous topologies, the ring topology only requires one isolated converter, and the turns
ratio of the transformer is 1:1. This makes the isolated converter relatively easy to design,
because this converter can work independently with no need to take the other converters
into account.

The advantage of the ring topology is that all the converters work independently and the
non-isolated converters can be made in a compact lightweight and low profile format. Since
any battery and converter can be added on or taken off the battery string without affecting
the others, this topology becomes easy for future expansion and mounting.

Page 12

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

2.5. Selection of voltage equaliser configuration for EV3

A summarised comparison table of equaliser topologies is shown in Table 2.1.

Common bus Common core Ring

• Good for long • Good for long battery string • Can be made in small size

For battery string • Simple control and light weight

• Easy for future • Easy for future expansion
expansion
• Converter requires • Converter requires transformer
Inefficient to transfer the

Against transformer • Inflexible for future expansion energy to the remote

• Inflexible for mounting battery

Table 2.1. Comparison of equaliser topologies

For the EV3 application, the ring structure is the most appropriate because most of the
converters can satisfy the requirement of being small, non-isolated and easily
manufacturable. In fact only one converter connected between the top and bottom of the
battery string needs transformer isolation, and the proposed configuration is shown in
Figure 2.4.

Figure 2.4. The proposed voltage equaliser structure

In Figure 2.4, the overall battery string is divided into four banks. Inside each bank, a
number of 24W non-isolated converters are used to equalise the batteries inside the bank.
The batteries inside each bank would be equalised first by the 24W non-isolated converters,
then the bank equalisation would be done by the 144W or 192W converters.

Page 13

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

The key points for selecting the types of converters are the cost, manufacturability and
efficiency. Unfortunately there is no single solution to satisfy all the requirements. For the
non-isolated converters, the buck-boost configuration has the advantage of a low
component count, therefore the buck-boost would have the advantages of high
manufacturability and low cost. The Cuk converter has theoretically the highest efficiency
when the converter is used for transferring energy between any adjacent batteries, but the
drawback of the Cuk converter is if the converter is desired to have bi-directional energy
transformation capability, the component count would be double that of the buck-boost
converter. Therefore in order to decide which converter is the best choice, the converter
efficiency comparison of the buck-boost and the Cuk converters was calculated through
simulation. In this performance test both converters have a fixed input voltage of 12V, and
by changing the input current, the efficiency of each converter over a range of power rating
from 10W to 50W was measured. The results of these simulations are shown in Figure 2.5,
which indicate there is no significant efficiency improvement when using the Cuk converter.
Therefore the buck-boost converter was chosen as the non-isolated converter of the
equaliser.

Buck-boost and Cuk converter efficiency

comparison
100

98
Efficiency (%)

96
Buck-
94 boost

92 Cuk

90

88
10W 20W 30W 40W 50W
Power rating

Figure 2.5. The efficiency comparison of buck-boost and Cuk converters

The isolated converter provides the electrical isolation between the top and bottom bank.
Various isolated converters can be selected, such as flyback, push-pull, forward, half-bridge
and full-bridge converters. By comparing both electrical and mechanical characteristics, the
decision for the isolated converter was made. Due to the significantly low component
counts over the others, the single switch flyback converter was chosen because only one
MOSFET is required on each side of the transformer to achieve bi-directional energy

Page 14

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

transformation. However the price to pay for the single switch flyback converter is that the
single switch flyback converter must dissipate the energy stored in the leakage inductance
of the transformer in either the switch or a snubber associated with the switch. Other
isolated converters with more than one switch such as the two switch flyback converter or
half bridge forward converter perform better because the energy stored in the leakage
inductance does not need to be dissipated.

The disadvantages of the isolated converter are the size, cost and the manufacturability of
the transformer. For any converter that uses transformer as its energy transfer unit, if the
operational voltage is low, the required minimum number of turns would be low, and the
ratio of the primary leakage inductance to the primary inductance becomes too high,
reducing the converter efficiency. Increasing the turns of the winding can solve this
problem, but it would increase the size of the transformer and winding resistance.

The proposed ring voltage equaliser structure is shown in Figure 2.6. Within this structure,
three 192/144W non-isolated buck-boost are used to balance the banks from the top to the
bottom, and a isolated flyback converter is used to balance the energy between the top and
the bottom bank. Inside each bank, the adjacent batteries are equalised by 24W non-isolated
buck-boost converters. In this design, all the converters are designed for bi-directional
energy transfer, with average currents selected as 2A, so that approximately 10% of total
charge can be balanced in an hour.

Figure 2.6. The proposed voltage equaliser structure

Page 15

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 2: Battery Monitoring and Equaliser topologies

In the next chapter, the design procedures of each converter are detailed. This includes the
control strategy, and the design of each controller and converter. The calculations of each
required component values are listed and several inductor core construction technologies
are also introduced to achieve a more compact design.

References:
[1] PowerDesigner, “Dynamic Equalization Techniques For Series Battery Stacks”
http://www.powerdesigners.com/InfoWeb/design_center/articles/NDCD/ndcd.shtm
[2] Linden, D., Handbook of Batteries, 2nd ed. 1995, USA: McGraw-Hill.
[3] Sinclair P., Duke R.M. and Round S.D. “An Adaptive Battery Monitoring System for an
Electrical Vehicle”. Proc. Intl. Conf. on Power Electronics, Drives and Energy Systems,
Perth, Australia, 1998

Page 16

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 3: Converter design

3. Converter design

In Chapter 2, the ring equaliser topology was chosen as the proposed equaliser structure for
the EV3, and the primary reason for choosing this topology is that fewer isolated converters
are required to make up a low profile, low cost and easily expandable voltage equalisation
system. This battery equalisation system requires three different types of converter to be
built, the 24W buck-boost converter, the 192W buck-boost converter and the 192W flyback
converter.

In this chapter, the detailed design procedures of each converter are demonstrated, and the
key factors of each converter are also discussed. The operational current for all converters
are limited at 2A, and the target efficiency for all converters are 90%.

3.1. The 24W buck-boost converter

In the EV3, all 26 batteries are connected in series and divided into four banks. Within any
bank, a 24W buck-boost converter is used to interconnect any two adjacent batteries. The
design of this converter can be split into two parts: the controller and the power converter.
The controller includes the PWM generator and the gate drive, which is used to drive the
buck-boost power converter. The schematic diagram of the buck-boost converter for the
24W battery equaliser is shown in Figure 3.1.

Figure 3.1. The Buck-boost converter

Page 17
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

3.1.1. The PWM controller

The PWM controller is the heart of the converter. In this design, the two main functions
that the controller provides are:

Generating the PWM switching signals to drive the power converter.

Sensing the pulse by pulse current signal from the converter, and converting this
pulsed current signal into an average current signal, which is used to adjust the duty-
cycle of the PWM signal to switch the MOSFETs.

The selected PWM control IC is SG3526 and its block diagram is shown in Figure 3.2. The
SG3526 is a high performance monolithic pulse width modulator circuit designed for fixed-
frequency switching regulators and other power control applications. Included in the 18-pin
dual-in-line package are a temperature compensated voltage reference, sawtooth oscillator,
error amplifier, pulse width modulator, and two low impedance power drivers. Also
included are protective features such as soft-start and undervoltage lockout, digital current
limiting, double pulse inhibit, a data latch for single pulse metering, adjustable deadtime,
and provision for symmetry correction inputs. For ease of interface, all digital control ports
are TTL and B-series CMOS compatible. Active LOW logic design allows wired-OR
connections for maximum flexibility. This versatile device can be used to implement
single-ended or push-pull switching regulators of either polarity, both transformerless and
transformer coupled. The SG3526 is characterized for operation from 0°C to +125°C, and it
is capable of generating PWM signals up to 350kHz with a duty cycle from 10 to 95%.[1]

Figure 3.2. The block diagram of SG3526

Page 18
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

As mentioned in the previous chapter, the operational current of this battery equalisation
system is set to 2A. To achieve this, all the PWM controllers will be designed to operate
under average current mode control.

3.1.2. Average current mode control

The control strategy for the operation of any converter can be either voltage mode or
current mode. In voltage mode, the output voltage of the converter is regulated. Therefore if
the output impedance is high, the output current would be low, but if the load impedance is
low, the output current would be high. Alternatively, with current mode control the output
current from the converter is regulated, and the output voltage varies in response to load
impedance changes. In this project, every converter was designed to transfer the energy at a
constant rate of 2A; therefore, current mode control was employed.

A small shunt resistor or a current transformer are the common choices for current sensing.
The shunt resistor is easier and cheaper to apply, but when a large current is involved, the
shunt resistor generates larger conduction loss than the current transformer. In this design,
since the regulated average current is only 2A, the shunt resistor has been chosen to sense
the current signal.

To ensure current mode control functions properly, choosing a suitable value for the current
sensing resistor is also important. If the resistance is too small, the current signal may be
too small and easily affected by noise, thus reducing the accuracy of the measurement.
Alternatively if the resistance value is too high, the power loss from the current sensing
resistor affects the converter overall efficiency. In this battery equaliser design, since some
converters are placed close to the motor and other high power electronic circuit, it is
important to ensure the measured current signal is accurate enough for the rest of control
circuit. According to the experimental results, a 40mΩ shunt resistor can provide a good
quality current signal and keep the efficiency degradation at an acceptable level.

Page 19
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

As shown in Figure 3.1, the current sensing resistor RCS is placed in series with the
inductor with neither end connected to ground. Therefore, a differential amplifier with a
common mode range greater than 12V is required to accurately measure the current signal
across RCS, and in this design the CA3140E op-amp was chosen, and the average current
mode control circuit is shown in Figure 3.3. This current mode control circuit can only
measure one way of current flow. Therefore for this bi-directional battery equaliser, another
set of this current mode control circuit is required.

Figure 3.3. The average current mode control circuit

This current mode control circuit can be divided into two parts. The first part is the
differential amplifier, which is the current sensing part, and this is made up by CA3140E
op-amp. The second part is the current limiting part, which is made up by the internal error
amplifier of the SG3526 PWM controller. The operating principle of this current mode
control circuit is first of all, the differential amplifier measures the signal from current
sensing resistor. Then the 10nF capacitor will smooth the current signal, which makes it
more like a dc voltage for the following current limiting circuit. In Figure 3.3, the 10kΩ
potentiometer set the average operational current level, and in this application the average
current is limited to 2A.

3.1.3. The gate drive

To efficiently drive the MOSFET in a power converter, minimising the turn on/turn off
time of the MOSFET is important. To reduce the switching time, sufficient driving current
must be supplied to charge the gate capacitor in the MOSFET. During the turn off time, the
MOSFET also requires a fast path to discharge the energy in the gate capacitor. Therefore,
a gate drive is used in order to reduce the switching time, minimising the switching loss.

Page 20
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

The IR2112(S) is a high voltage, high speed, power MOSFET and IGBT driver with
independent high and low side referenced output channels. Logic inputs are compatible
with standard CMOS or LSTTL outputs, down to 3.3V logic. The output drivers feature a
high pulse current buffer stage designed for minimum driver cross-conduction. Propagation
delays are matched to simplify use in high frequency applications. The floating channel can
be used to drive an N-channel power MOSFET or IGBT in the high side configuration,
which can operate at up to 600 volts.[2]

3.1.4. The power converter of the 24W buck-boost converter

The design of the 24W buck-boost converter is characterised by two key components, the
inductor and the MOSFET. The inductor is the energy storage unit in the buck-boost
converter, and the MOSFET is the switching device of this converter. In the following
section, the detailed inductor design procedure is listed, and to improve the
manufacturability, the technology of the planar inductor and the integrated inductor are also
investigated.

A - Inductor design on RM core

This inductor design procedure is carried out by the standard RM core, and the design
consideration includes the core size, core material, required inductance and airgap.
However, this design procedure can also be applied to other types of inductor cores.

Step 1: Select core material and core size

Different materials have different magnetic characteristics. An inductor's magnetic core is

made of specially formed material with “soft” magnetic properties. Although physically
hard, a magnetic core is said to be “soft” when it does not retain significant magnetism.
Silicon steel, iron powder and ferrite are common core materials for inductor design.

Page 21
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Silicon steel is relatively inexpensive and easy to form. In addition, silicon steel is a metal
with low resistivity. Low-core resistivity means silicon steel readily conducts electrical
current. The result is that undesirable eddy currents can flow in the core material. Eddy
currents contribute to heating and core loss. In addition, a silicon steel core tends to reach
the point of saturation rather easily. When saturated, a core is unable to store additional
magnetic energy. Rapid saturation results in reduced operating range[4].

Iron powder has higher resistivity than silicon steel. By special processing, iron particles
are insulated from each other. The particles are mixed with a binder. The cores are then
pressed into their final shape. Next, a baking process is used to cure the cores. After curing,
many tiny air gaps combine to provide a distributed air gap effect, which also serves to
insulate the particles from one another thereby reducing eddy current flow in the core. This
extends the useful frequency range, but there is a trade-off. The binding material adds a
distributed air gap to the core. The distributed air gap reduces the permeability of the core
[4].

Ferrite is a crystalline magnetic material made of iron oxide and other elements. The
mixture is processed at a high temperature and formed into a crystalline molecular structure.
Unlike the other material, ferrites are ceramic materials with magnetic properties. Ferrites
have high magnetic permeability and high electrical resistivity. Consequently, undesirable
eddy currents are greatly reduced by ferrite cores. With their high resistivity, ferrites are
ideal for use as inductors. For example, ferrite beads are frequently used to reduce parasitic
oscillations and for general filtering at the component lead level. This type of broadband
component requires a broadband low-Q in order to provide high impedance over a wide
frequency range[4].

By comparing the characteristics of each core material, ferrite becomes the better choice for
inductor design. However there are a numbers of different ferrite cores that can be chosen.
Considering the cost, availability, the frequency and the flux density, the 3C90 material is
selected.

There are many possible inductor core geometries. A core's geometry depends on various
factors, including the application, the available mounting area and volume, the allowable
radiation, the limitations on windings, the operating temperature, and how the inductor will

Page 22
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

be mounted. Consequently, a core's geometrical shape can take the form of a cylinder,
bobbin, toroid or several other complex shapes[4]. In the beginning of the inductor design
the standard RM core is used to carry out the design process. Then the planar inductor and
the integrated inductor technology will be demonstrated later on.

The selection of the core size depends on the power that the converter is designed to deliver.
If the core is too small, the inductor would be saturated during the operation. However there
is harm to the converter efficiency when using a large inductor core, but choosing the right
core size can reduce the manufacturing cost, size and the weight of the converter. The
power rating to the core size curve is shown in Figure 3.4 [2], for this 24W buck-boost
converter, the RM10 core is chose.

Figure 3.4. The RM core power rating curve.

Step 2, Calculating the minimum inductance.

The minimum inductance is required to maintain the continuous conduction of the

converter when the maximum input voltage is applied. However it is also important not to
make the inductance too large, because a larger inductance requires more turns, which leads

Page 23
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

to higher conduction loss during operation. The standard terminal voltage of the battery is
12V, but when the battery is fully charged, the battery voltage can go up to 13V. Assuming
the duty-cycle of the switching signal is 50%, if the converter operates at average current of
2A, the allowable peak current would be 8A in order to maintain the converter operates
under continuous conduction.

However this assumption can only be made when the converter has 100% efficiency, but in
reality it is not possible to achieve this. To make the design more practical, the maximum
current ripple is set to 6A and the minimum inductance can be calculated by Equation 3.1.

V ⋅ DT
L= Equation 3.1
∆I

L = minimum inductance
V = maximum input voltage
D = duty-cycle
T = switching period
ΔI = allowable ripple current flow through the inductor

Assuming the maximum input voltage is 13V, switching frequency is 100kHz, 50% duty-
cycle and 6A of ripple current. The minimum inductance is calculated as 11.9µH.

Step 3, Selecting maximum “on” period

Since the charge level of the battery is proportional to the terminal voltage of the battery,
the input voltage of the converter will not stay at 13V at all times, as mentioned in the
previous step. Therefore, in order to maintain the constant power flow when low input
voltage occurs, the duty-cycle of the switching signal has to be increased, and the
maximum “on” period is defined when the lowest input voltage occurs.

The minimum inductance was calculated from Step 2. If the input voltage drops to 11.5V,
and the switching frequency and the current ripple remain unchanged, the maximum duty-
cycle can be calculated as 62% from Equation 3.1.

Page 24
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Step 4: Calculating the air gap

For every dc to dc converter that uses an inductor, a sufficient air gap has to be carefully
calculated. The air gap is used for storing energy from the dc source. An insufficient air gap
would lead the inductor to saturate, which decreases the converter efficiency. Alternatively,
if the air gap is too large, the inductance will be small, which leads the converter to operate
in discontinuous conduction, which can also affect the converter efficiency. The air gap size
can be calculated by Equation 3.2.

µ r ⋅ N 2 ⋅ Ae
α= Equation 3.2
2⋅L
µr = 4π×10-7
N = number of turns
Ae = area of the core, mm2
L = inductance, mH

From Step 1 and 2, the RM10 core was selected as the initial core size and the minimum
inductance 11.9µH was also calculated. Assuming the number of turns N is equal to 10, the
required air gap can be calculated as 0.507mm.

Step 5: Check core flux density and saturation margin

After calculating the required inductance, number of turns and the air gap size, it is
necessary to check the maximum flux density in the core to ensure an adequate margin
between the maximum working value. Figure 3.4 shows the B-H curve of the 3C90 core,
and the flux density must not reach the saturation region.

Page 25
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Figure 3.5. 3C90 B-H curve

The flux density consists of the ac flux density Bac and dc flux density Bdc. The sum of the
ac and dc flux densities is the maximum flux density of the core, where the ac and dc flux
density can be calculated by Equation 3.3 and 3.4.
V ⋅t
Bac = Equation 3.3
N ⋅ Ae
µ 0 ⋅ N ⋅ I DC
Bdc = µ ⋅ H = Equation 3.4
2α × 10 −3
V = Maximum supply voltage t = on time, µs
N = Number of turns Ae = Area of core, mm2
µ0 = 4π × 10-7 H/m IDC = Average DC current
α = air gap, mm

The B-H curve is temperature dependent. Figure 3.4 shows the B-H curves of the inductor
core at 25°C and 100°C. However both temperatures are unlikely to be the working
temperature of this battery equalising system. The working temperature of the converter
depends on the allocation of the converter, and along the EV3 the warmest area will be the
engine bay. During the operation, the temperature of the engine bay would be around 40°C.
By allowing 20°C temperature rises on the inductor, the core temperature would be 60°C,
which means the saturation level of the core would be around 375mT.

Page 26
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

At the initial stage, if the maximum supply voltage V is set as 13Volts, the on time t = 5µs,
number of turns N = 10, and the Ae is equal to 96 mm2. The Bac can be calculated as 67.7mT.

To calculate the Bdc, the number of turns N and the average current IDC were assumed as 10
and 2A. The air gap size α was calculated from Step 4, which is 0.5mm. Therefore the dc
flux density is calculated as 25.1mT. The total flux density is equal to 92.8mT, which is
well under the maximum allowable flux density of 350mT.

B - The planar inductor

Since 25 24W buck-boost converters need to be built and ideally located in a recess in the
top of each battery, the manufacturing process for a low profile design has to be as simple
as possible. The disadvantage of the conventional inductor is that it has a relatively large
physical size. Therefore if 25 converters have to be installed into the EV3, they would
occupy too much space and create a mounting problem. To overcome this mounting issue,
the planar inductor and the integrated inductor technologies were investigated.

A planar inductor is constructed using either one E and one I core or two E cores, and the
actual winding is made up by the loops of printed circuit board (pcb) tracks, where higher
inductance can be achieved by multiple loops of a multilayer pcb. The biggest advantage of
the use of the planar inductor is that since the windings are made up by the pcb tracks, the
inductor can be made in a low profile, high precision and easily manufactured format. The
constructions of the planar inductor are shown in Figure 3.5.

(a) (b) (c) (d)

Figure 3.6a. E core and I core
3.6.b. Fitting two multilayer pcb into the E core
3.6.c. Interconnecting the two pcb together
3.6.d. Putting the I core on then completing the player inductor

Page 27
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

A major problem of the planar inductor is that if the pcb contains more than two layers, any
intermediate windings would not have as much heat dissipating capacity as the top and
bottom layers. Therefore, as the current flows through the winding, the pcb temperature
will build up, then winding resistance and conducting losses will increase. To overcome
this problem, a wider pcb track can be used to reduce the winding resistance and hence
increase the current capacity and reduce the conduction losses from the inductor. The trade
off by using the wider pcb track is the allowable numbers of turns per layer would be
decreased. To emphasise the idea, Figure 3.6 shows how the pcb temperature can be varied
by changing the track width.

Figure 3.7. The track width versus temperature rise diagram

Page 28
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

C – The Integrated Inductor

The integrated inductor is an inductor that has the form of an integrated circuit. Unlike the
plastic package from the standard integrated circuit, the whole package of the integrated
inductor is the inductor core, which has a built in air-gap. The leads of the package are the
partial wirings of the inductor, and the inductor is completed by the pcb wiring. The
construction of the integrated inductor is shown in Figure 3.7.

Figure 3.8. The construction of the integrated inductor

Compared with the standard inductor and the planar inductor, the integrated inductor has
the advantages of the smallest physical size, and because there is no multi-layer pcb
required, the pcb manufacture also becomes relatively simpler compared with the others.
However, apart from these advantages, there are two disadvantages of this integrated
inductor. The first drawback is due to this compact core size; the maximum inductance
from a single integrated inductor can only be made around 5µH, which means the converter
requires 250kHz or higher switching frequency. The second drawback is the cross section
area of the pin out. The cross section area of the lead is 0.18mm2, which is eight times
smaller than the litz wire that can be used on the RM core. The smaller cross section area
brings higher dc resistance, which means the conduction loss from the inductor would be
eight times higher.

D - MOSFET selection
To select an appropriate MOSFET for the 24W bock-boost converter, there are four main
considerations - voltage rating, current rating, on resistance Rds-on and switching loss.

Page 29
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

The voltage rating tells how much voltage MOSFET can handle between Drain and Source.
For the buck-boost converters, the voltage between drain and source Vds would be equal to
twice the supply voltage; therefore, the voltage rating of the MOSFET must be at least two
times the supply voltage plus 20 percent of margin. Ideally the current rating is defined by
the MOSFET peak operational current. However, in the converter design the pulse current
also needs to be considered. Therefore, the continuous current and the pulse current
determine that MOSFET rating.

The Rds-on is the on resistance when the MOSFET is in the on state. The low Rds-on
MOSFET has low conducting loss and in most cases is proportional to the voltage rating of
the MOSFET. Therefore, it is not wise to select a high voltage rating MOSFET and operate
it under low voltage conditions.

Ideally, higher switching frequency requires less inductance to maintain continuous

conduction. Unfortunately, the high switching frequency means larger switching loss. The
switching losses are created as a result of the simultaneous exposure of a MOSFET to high
voltage and current during the transition between the open and closed states [3]. The fast
transient requirement pushes us towards possibly high switching frequencies, which in turn
increases the relative importance of the switching losses[3].

In reality it is difficult and uneconomical to manufacture and purchase the MOSFET, which
has both low Rds-on and low switching loss. The low Rds-on MOSFET is normally made up of
a larger silicon area to reduce the on resistance, and the switching speed of MOSFETs is
limited by the gate capacitance on the silicon, which is called the gate charge. The gate
charge is the gate capacitance that needs to be charged or discharged during the on or off
switching transition. The drawback of this larger silicon area is the creation of a larger gate
capacitance. That makes the MOSFETs need more energy to be turned on, hence increasing
the switching time and the switching loss. Therefore the compromise between the
conducting loss and the switching loss has to be considered.

In this equalisation system, the overall battery string is divided into four banks. The first
bank contains 8 batteries. The second, the third and the fourth banks each contains 6
batteries. So the required number of 24W buck-boost converters for each bank is 7 for the
first bank, and 5 for the second, the third and the fourth bank. Therefore the total number

Page 30
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

for the required 24W buck-boost converters is 22. In each 24W buck-boost converter, each
converter requires two MOSFETs, therefore the cost becomes a significant concern. In
general, the price of the MOSFET is dominated by two factors; the voltage rating and Rds-on.
By allowing 20 percent allowance on both voltage and current signal, the voltage and
current rating could be set to 29V and 2.4A. Then any MOSFET, which has a voltage rating
higher than 29V and has a reasonably low Rds-on and price can be considered.

The gate charge defines whether the MOSFET has low switching loss or not. It determines
the gate capacitance of the MOSFET, and it is mainly related to the Gate to Source voltage
and is slightly affected by the Drain to Source voltage. In practice, the logic MOSFETs
normally have lower gate charge because they only need 5V to be switched on instead of
the 9 to 10V of the standard MOSFETs. Therefore by considering the factors of cost,
conduction loss and switching loss, the logic MOSFET SUD35N05-26L is chosen for the
24W buck-boost converter and more details on this MOSFET will be discussed in chapter 5.

3.2. The 192W buck-boost converter

The 192W buck-boost converter is for transferring energy between the adjacent banks. The
design procedure of the 192W buck-boost converter is similar to the previous 24W version,
but there are three exceptional concerns that need to be considered. First, due to the input
voltage is higher than the 24W buck-boost converter, so the selected MOSFETs should
have higher voltage ratings. Second, since the power rating of this buck-boost converter is
higher than the previous 24W version, a larger inductor core is required for storing larger
energy. Third, assuming 90 percent efficiency is achieved, the total power loss from the
converter would be 19.2W. Due to the majority of power loss being caused by the
MOSFETs, heatsinks for each MOSFET are required in order to prevent overheating the
MOSFET.

The input voltage of this 192W buck-boost converter is eight times higher than the previous
24W version. Minimising the switching loss becomes important. From the design of the
24W buck-boost converter, it mentions that minimising the switching loss can be done by
either using the low gate charge MOSFETs or reducing the switching frequency. However,

Page 31
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

the problem brought from choosing low gate charge MOSFETs and how to compromise
between the low Rds-on or the low gate charge was also detailed. Then the result tells that
reducing the switching frequency is the better option to minimising the switching loss from
the 192W buck-boost converter. In order to meet the electrical specification and minimising
the cost, IRF740 is chosen for the 192W buck-boost converter.

The inductor design procedure is identical to the inductor design of the 24W buck-boost
converter. However because the power rating of the converter has been increased, a larger
inductor core is used. According to the RM core power rating curve from Figure 3.4, the
minimum inductor core size for the 192W buck-boost converter would be RM14.

Alternatively the planar inductor core can also be used in this 192W buck-boost converter.
According to the cross section area and the material, the E42 planar core would also be
suitable for this application. However, due to the insufficient core size and the wiring space,
the integrated inductor can not be used on this converter.

Reducing the switching frequency requires larger inductance to maintain the continuous
conduction. By repeating the inductor design procedure in the design of the 24W buck-
boost converter, if the switching frequency is reduced from 100kHz to 50kHz, and the duty-
cycle and the allowable current ripple remain unchanged, the minimum inductance can be
calculated as 192µH for the 192W buck-boost converter.

3.3. The 192W flyback converter

The flyback converter can provide an electrical isolation between two battery banks.
Therefore, it is used for transferring energy between the top and the bottom banks. The
schematic of the design flyback converter is shown in Figure 3.8.

Page 32
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Figure 3.9. The flyback converter

This flyback converter is also designed for bi-directional energy transformation, and 2A of
average operational current. In Figure 3.12, each side of the flyback transformer has its
own PWM controller. Bank 1 represents the top bank and Bank 2 represents the bottom
bank of the battery string; therefore, the reference voltage Vref_1 would be 216V higher
than the Vref_2 because there are 18 batteries between these two reference points. The two
Rcs are used for current sensing purposes, and the sensed current signal would be fed into
Vcs, which is the current sensing input of the PWM controller. Then the PWM controller
would adjust the duty-cycle to switch the MOSFETs.

3.3.1. PWM controller for the 192W flyback converter

The PWM controller for this 192W flyback converter is similar to the buck-boost PWM
controller. The control strategies of both controllers are by sensing the pulse by pulse
current signal then convert it into a dc voltage, and then the SG3526 PWM controller to
adjust the duty-cycle of the PWM signal. However, there are two differences between these
two controllers. The first difference is that each controller only needs to drive one
MOSFET, which makes the switching control strategy simpler than the buck-boost
controller. The second difference is since the current sensing resistor is connected to the
reference point of the battery bank, the differential amplifier used in the buck-boost
controller is not required. Instead of the differential amplifier, a negative feedback amplifier

Page 33
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

is used to sense the current signal and then the amplified current signal would be sent to the
SG3526. The schematic of the flyback PWM controller is shown in Figure 3.9.

Figure 3.10. The PWM controller for the 192W flyback converter.

3.3.2. 192W flyback transformer

The key design of the flyback converter is the flyback transformer. The flyback transformer
is used for storing and transferring energy from one side of the transformer to the other.
Therefore, to design an appropriate flyback transformer becomes the most important factor
of the flyback converter design. The construction of the flyback transformer can be seen as
the combination of two inductors. Therefore, the inductor design procedure from the buck-
boost converter is again used; moreover, the primary and secondary inductances can not be
treated individually. Then some reiterative calculation would be required in order to make
both primary and the secondary turns meet the electrical requirement, and the detailed
transformer design procedure is shown below.

Step 1, Core selection

The core size depends on how much power the converter needs to deal with. From the
design of the 192W buck-boost converter, the RM14 core was selected. According to the
identical power rating, the RM14 core can also be used in this flyback transformer design.

Step 2, Calculating minimum primary inductance

The minimum inductance is required to maintain the continuous conduction of the
converter when the maximum input voltage is applied. Assuming the switching frequency
and the duty-cycle are 50kHz and 50 percent, the average battery voltage inside the bank

Page 34
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

and the allowable current ripple are 13V and 6A. The minimum primary inductance can be
calculated again from Equation 3.1, and the value of the minimum primary inductance is
174µH.

Step 3, Selecting maximum “on” period

In order to maintain the constant power flow, the maximum “on” period will occur at
minimum input voltage and maximum load.[3] Assuming the input voltage is dropped to
11.5V, and the rest of the parameters from Step 2 remain unchanged, the maximum on
period can be calculated as 57% from Equation 3.1.

Step 4, Select ac flux density swing

From Figure 3.4, the maximum allowable flux density is 350mT. The flux density swing
also consists of both ac and dc flux densities, and from Figure 3.4, the ac flux density
swing is initially selected as 250mT.

Step 5, calculate minimum primary turns

After the ac flux density is selected. By rearranging Equation 3.3, the minimum number of
primary turns is shown in Equation 3.5, and according to the manufacture’s datasheet [4] of
the RM14 core, the effective area of the centre leg is 198mm2.
V ⋅t
N min = Equation 3.5
∆Bac ⋅ Ae
Nmin = minimum primary turns
V = applied dc voltage = 12 × 8 = 96V
t = “on” times, µs = 10µs
∆Bac = maximum ac flux density, T. Assuming ∆Bac =250mT
Ae = minimum cross-section area of core, mm2. Ae = 198 mm2.

Therefore the minimum primary turns Nmin can be calculated as 20.

Step 6, Calculate secondary turns

The output voltage of the transformer has the nominal voltage of 72V (6 × 12V batteries).
Therefore, the secondary turns can be calculated by Equation 3.6, which converts the
primary turns into volts per turn, then calculates the required secondary turns.

Page 35
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Primary volts per turn V/N = Vp/primary turns = 96/20 = 4.8 V/N
VS ⋅ N P
And then the secondary turn Ns = Equation 3.6
VP

72 ⋅ 20
= = 15
96

Where Vs = secondary voltage

Ns = secondary turns

The secondary inductance can be calculated by Equation 3.7.

L = Al ⋅ N 2 Equation 3.7
L is the value of the inductance
Al is the equivalent quantity of the inductance
N is the numbers of turn

Al is the equivalent quantity of the inductance in mH with a winding of 1000turns, which

can be calculated from the primary turn and primary inductance, and the calculated Al is
3×10-7. Then the secondary was calculated as 67.5µH.

Step 7, Calculating the air gap size

The air gap calculation is very important in the flyback transformer design. An insufficient
air gap would saturate the core, hence reduce the converter efficiency. Alternatively, if the
air gap is too large, the inductance would become too small, which may lead to the inductor
working under discontinuous conduction, and thus decrease the converter efficiency. The
air gap can be calculated as 0.24mm from Equation 3.2.

Step 8, Check core flux density and saturation margin

The overall flux density of this flyback converter consists of the ac flux and the dc flux, and
they can also be calculated by Equation 3.3 and 3.4. At this stage, the supply voltage V is
made up by eight 12Volts batteries, which makes up a total voltage of 96 Volts. Assuming
50% duty-cycle, the on time t = 10µs, and from Step 4, the primary number of turns N is 20;
and the effective area Ae of the E43 core is 198 mm2. Therefore, the ac flux density is
calculated as 240mT.

Page 36
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

To calculate the dc flux density, the number of turns N and the average current Idc were
assumed as 20 and 2A. The air gap size α was also calculated as 0.23mm. Therefore the dc
flux density is calculated as 109mT. The total flux density is equal to 349mT, which is
under the maximum allowable flux density of 350mT.

In the buck-boost converter design, the planar core was introduced to improve the
manufacturability. However, in the flyback converter design, due to the manufacturability,
it is very difficult to make a pcb that has greater than four layers. Therefore, the planar core
was not considered.

3.3.3. Reducing the voltage stress on MOSFETs

The voltage stress is the Drain voltage to overshoot during the turn-off edge, caused by the
transformer leakage inductance.[3] Since the current flowing through the inductor can not
be stopped instantaneously at the time that the MOSFET is switched off, the small amount
of leakage inductance and this residual current would result in an overshoot across the drain
and source of the MOSFET.

To overcome this effect, the voltage overshoot problem is best dealt with by ensuring that
the leakage inductance is as small as possible, then clamping the tendency to overshoot by
dissipative or energy recovery methods.[3] Minimising the leakage inductance can be made
by shortening the wiring distance from the transformer to the other components. To achieve
this, it is better to place the other components close to the transformer.

The flyback converter with the voltage clamping circuit is shown in Figure 3.10. The idea
of this clamping circuit is when MOSFET Q1 is turned on, the reverse polarity of the
clamping diode Dc1 makes the clamping circuit act like an open circuit. However, during
the instant when the MOSFET is turned off, due to the current flow through the inductor it
can not be instantaneously turned off. Therefore, the current from leakage inductance from
the primary inductance would try to keep flowing through Q1, and then cause a voltage
spike. The Clamping circuit 1 provides an alternative path for the primary inductance to
dissipate remaining energy from the primary leakage inductance.

Page 37
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

Figure 3.11 Flyback converter with clamping circuit

In Figure 3.15, the flyback converter with the clamping circuit is displayed, and it shows
each clamping circuit consists of three components, a diode Dc, a capacitor Cc and a
resistor Rc. The diode is used to provide a current flow path for the energy stored in the
leakage inductance. Therefore, when the MOSFET is turned off, the energy from the
leakage inductance can flow into the clamping circuit. Then the Rc and Cc inside the
clamping circuit would dissipate the energy from the leakage inductance.

There is no rule for selecting the value of the Rc and Cc. The values of Rc and Cc depend
on how fast the energy from the leakage inductance needs to be dissipated. If the RC time
constant is large, less voltage stress appears on the MOSFET. However, the drawback of
the clamping circuit is that during the off state the voltage across the MOSFET is equal to
the sum of the input voltage and the voltage across the flyback transformer. Therefore, if
large clamping resistance and capacitance are used, the clamping circuit would require
more time to charge the clamping capacitor Cc, which increases the MOSFET turn off time,
hence increases the switching loss.

Increasing the value of Rc and Cc can significantly reduce the voltage stress, but it is
undesireable to make the clamp voltage too low, as the flyback overshoot has a useful
function. It provides additional forcing volts to drive current into the secondary leakage
inductance during the flyback action. This results in a more rapid increase in flyback
current in the transformer secondary, improving the transfer efficiency and reducing the
losses incurred in the clamp resistor Rc[3].

Page 38
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 3: Converter design

3.4. Summary

In this chapter, the design procedures of three different converters - 24W buck-boost, 192W
buck-boost and 192W flyback converters were detailed. The PWM controllers for all three
types of converters are very similar. However there are two exceptional design
considerations. First, as there is only one switching device on the 192W flyback converter
and the MOSFET is connected to the reference point via the current sensing resistor.
Therefore, there is no high-side drive required for this converter. Second, unlike the buck-
boost converter, the current sensing resistor of the flyback converter is not sitting on the
voltage source like the buck-boost converter does. So the additional op-amp for providing
wide common mode range is not needed.

In the next chapter, the simulations of each converter will be made by Pspice simulator. The
component values calculated in this chapter are also applied into these simulation tests. The
simulation results include the detailed equalisation process and the switching waveforms on
each key component, and finally it shows the energy between the two batteries can be
successfully balanced by the designed converter.

References:
[1] Data sheet of SG3526
[2] Philips inductor core material
[3] Keith Billings, “Switchmode power supply handbook second edition”, 1999
[4] Andy Chow, Chief Design Engineer, J. W. Miller Magnetics, Gardena, Calif. “Inductor
Core Material: The Heart of an Inductor”

Page 39
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 4: The simulations

4. Converters Simulations

In this Chapter the results from the simulations of each converter making up the chosen
voltage equaliser topology are discussed. The simulation models include one 24W non-
isolated buck-boost converter, one 192W non-isolated buck-boost converter and one 192W
isolated flyback converter. The simulations were carried out by Pspice simulator and in
these simulations two rechargeable batteries or banks of batteries were modelled as two 1F
capacitors, and the battery internal resistance and the wiring resistances were modelled as
two 100mΩ resistors.

Before the simulation, a series connected battery discharging test at a rate of 30A for 60
minutes through a resistive load was done in order to measure what the nominal voltages
from each battery would be after the batteries have been discharged. In the EV3 there are
26 batteries connected in series. Drawing energy from each battery at a rate of 30A is equal
to drawing 10kW of energy, which is roughly equal to the power consumption of standard
city driving. In this battery discharging test, eight sealed lead-acid batteries were first
parallel connected and equally charged, and then connected in series, and the discharging
result is shown in Table 4.1.

Average
Batt 1 Batt 2 Batt 3 Batt 4 Batt 5 Batt 6 Batt 7 Batt 8
voltage
Starting
12.62 12.78 12.74 12.76 12.65 12.72 12.71 12.76 12.72
voltage
End voltage 11.34 11.64 11.5 11.5 10.87 10.76 10.88 11.72 11.28
Difference
from the
+0.06 +0.36 +0.22 +0.22 -0.41 -0.52 -0.40 +0.44
average end
voltage
Table 4.1. The open-circuit voltages of each battery after 30A of 60 minutes discharge

In Table 4.1, the result shows that after an hour of discharge, the open-circuit voltages from
these batteries end up between 10.76 and 11.72 volts, which means after an hour of driving,
the voltage difference from each battery could be as high as 1V. Therefore, for the

Page 40

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

following simulations, the initial voltage of the source and the target batteries will have 1
volt difference.

4.1. 24W buck-boost converter

The voltage equaliser based on the 24W non-isolated buck-boost converter is shown in
Figure 4.1. In this configuration, C1 represents the higher voltage battery and C2 is the
battery required to be recharged, and the initial voltage these batteries are set to 12.5V and
11.5V respectively. In this simulation, the switching frequency was set to 100kHz, the
inductor wiring resistance R3 is 15mΩ and the inductance value is set to 18µH, which is
the minimum inductance value to maintain continuous conduction.

Figure 4.1. 24W non-isolated buck-boost converter

The aim of this simulation is to transfer the energy from C1 to C2, and the detailed
switching waveforms of the MOSFET and the inductor are shown in Figure 4.2. In these
simulation results, the PWM switching signal is represented by the switching signal V1.
Vds indicates the drain to source voltage of Q1. IC1 and IC2 show the current flow from the
source and target batteries C1 and C2. Finally, IL represents the current flow through the
inductor L1.

Page 41

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

When the PWM switching signal V1 of this equaliser is high, Q1 turns on, then inductor L
is charged by C1 via R1 and Q1. Later when V1 is switched low, Q1 turns off, the current
stops flowing from C1, the inductor current keeps circulating and the stored energy in the
inductor is discharged into C2 through R2 and the internal diode of Q2 without any
additional control.

Figure 4.2. The switching waveforms of the 24W buck-boast converter.

The charge level of the lead acid battery is proportional to its open-circuit voltage. By
measuring the open-circuit voltages across both C1 and C2, the charge level of each battery
can be monitored. The voltage equalisation result of this 24W buck-boost converter is
shown in Figure 4.3, which demonstrates that in principle the battery voltages can be
balanced. However the test result that is shown in Figure 4.3 is the accelerated version of
the voltage equalisation process. In the real world application, the voltage equalisation
speed would be much slower than this.

Page 42

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

Figure 4.3. 24W buck-boost converter equalisation.

4.2. 192W buck-boost converter

The setup configuration of this 192W Buck-boost converter simulation is similar to the
24W version. However, there are four changes required. First, in this simulation the source
and the target banks are made up of 8 and 6 batteries connected in series. Therefore, the
initial voltage for the source and the target banks are set as 100V (12.5V×8) and 69V
(11.5V×6). Second, because the operational voltage has been increased, the switching
frequency is reduced to minimise the switching loss, and in this simulation the switching
frequency is set to 50kHz. Third, in Chapter 3.2, the minimum required inductance to
maintain the continuous conduction for 50kHz switching frequency was calculated as
192µH. Moreover a larger inductance requires more turns, therefore, the dc resistance of
the inductor will also be increased, and in this simulation the inductor dc resistance is set to
70mΩ. Fourth, the MOSFETs have to be changed to handle larger the input voltage, and so
in this converter the IRF740 was chosen. The schematic of the 192W buck-boost converter
is shown in Figure 4.4.

Page 43

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

Figure 4.4. 192W buck-boost converter

In the simulation results, the PWM switching signal is represented by the switching signal
V1. Vds indicates the drain to source voltage of Q1. IC1 and IC2 show the current flow from
the source and target batteries C1 and C2. Finally, IL represents the current flow through the
inductor L. The operation principle of the 192W buck-boost converter is identical to the
24W version. This assumed, C1 has a higher voltage than C2, and the detailed operational
waveforms are shown in Figure 4.5.

Figure 4.5. 192W buck-boost converter switching waveforms.

Page 44

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

From Figure 4.5, the shape of the switching waveform of the 192W buck-boost converter is
similar to the 24W version. The only difference is due to the input voltage being eight times
higher than the 24W buck-boost converter, the required MOSFET drain to source voltage
Vds also becomes eight times higher. In Figure 4.5, the current waveform from the source
battery has a large current overshoot during the transition from off to on period. This is
caused by the reverse recovery of the body diode in Q2 which has been conducting
immediately prior to Q1, and then creates the current overshoot. The voltage equalisation
result of the 192W buck-boost converter is shown in Figure 4.6, which demonstrates in
principle that battery bank voltage can also be balanced.

Figure 4.6. 192W buck-boost converter equalisation.

4.3. 192W flyback converter

The use of the flyback converter is to provide both electrical insolation between the top and
the bottom battery banks and to balance the voltage of those bank. The schematic of the
single switch flyback converter is shown in Figure 4.7. In this simulation the initial voltage
of each bank is set to 100V (12.5V×8) and 69V (11.5V×6), and the switching frequency is
set to 50kHz. In Chapter 3.3, the primary and secondary inductances were calculated as
174µH and 98µH respectively. To simulate the clamping circuit, the primary and the
secondary leakage inductances were estimated as 2µH and 1µH respectively. The

Page 45

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

transformer primary and secondary wiring resistances in this simulation were estimated as
50mΩ and 25mΩ respectively.

Figure 4.7. The schematic of the 192W flyback converter

In the simulation, C1 was assigned has higher charge level than C2. To transfer the energy
from C1 to C2, the switching signal diagrams of this flyback converter are shown in Figure
4.8. In the simulation results, the PWM switching signal is represented by the switching
signal VSW1. Vds indicates the drain to source voltage of Q1. IC1 and IC2 show the current
flow from the source and target batteries C1 and C2. Id specifies the current flow through
the MOSFET Q1 and finally Id_clamp symbolises the current flow through the clamping
diode D_clamp1.

Figure 4.8. 192W flyback converter switching waveforms.

Page 46

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

As Figure 4.8 indicates for the flyback converter, when the switch VSW1 is high, Q1 turns
on. The primary inductance is charged by C1. During this period, because the clamping
diode D_clamp1 polarity is reversed, the clamping circuit is inactive. When V1 turns low,
the leakage inductance energy flows into the clamping circuit and the power dissipating
component C_clamp1 and R_clamp1 dissipates this energy. Then the stored energy from
the primary inductance is transferred to the secondary inductance. During this turn off state,
the current flow at the secondary side goes through the internal diode of Q2 to C2 to
complete one equalisation cycle without any additional control. During this turn off period,
clamping circuit 2 does not do anything because it only functions when energy is taken
from C2 to C1. The result of this simulation is shown in Figure 4.9, which demonstrates in
principle that the battery bank voltages at either end of the battery string can be balanced.

Figure 4.9. 192W flyback converter equalisation.

Page 47

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 4: The simulations

4.4. Summary

In this Chapter, three different power converters 24W buck-boost, 192W buck-boost and
192W flyback converters were simulated by PSpice simulator. The calculated component
values for each converter were obtained in Chapter 3. The detailed switching waveforms
for each converter were also described, and the simulation results show that the energy
equalisation between each individual battery and between banks of batteries can be
achieved. In the next chapter, the construction of each converter are detailed.

Page 48

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

5. The system construction

In Chapters 3 and 4, the design procedures and the simulations of this voltage equaliser
were detailed. In this chapter, the construction procedures of the 24W buck-boost converter,
the 192W buck-boost converter, and the 192W flyback converter are described. The
concept of the system’s construction is to make each converter compact, light weight, easy
to manufacture and highly efficient, with the target efficiency for each converter at 90%.

For the 24W buck-boost converter, the main design concentrates on the performance
investigation of the different inductor cores: the RM core, the planar core and the integrated
inductor. Then the investigation results are applied to the design of the 192W buck-boost
converter. In the design of the 192W buck-boost converter, due to the converter structure
being identical to the 24W version, the design procedure would be focused on minimsing
the power losses of the converter. In the design of the 192W flyback converter,
investigating the frequency response of the flyback converter will be first discovered. Then
the clamping circuit will be made to reduce the voltage stress on the MOSFET from the
leakage inductance of the chosen flyback transformer.

To turn each equaliser on and off easily, two 4N35 optical couplers are employed on each
PWM controller to enable and disable PWM control. This makes the PWM controllers
easier to interface with the micro-controller. However, in order to minimise the component
count and accelerate the construction process, the current mode control circuits are built on
a separate board, which would be connected to the equaliser when the converter is required
to be turned on.

Page 49

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

5.1. The 24W Buck-boost converter

The overall voltage equalisation system requires twenty-two 24W buck-boost converters.
To minimise the weight and size, a heatsinkless design is essential. To achieve this, the
right MOSFET and right switching frequency have to be carefully selected in order to
improve the converter efficiency.

5.1.1. Inductor design

In Chapter 3, the three different types of inductors, RM, planar and integrated inductors
were introduced, and their advantages and disadvantages were also discussed. In this
section, the performance analysis of these inductors is measured.

For this performance test, three buck-boost converters were built based on these three types
of inductor. In order to accurately compare the performance of each inductor, each
converter used an identical MOSFET, and had identical pcb layouts. For this test, the
switching frequency was set from 100kHz to 200kHz, and the maximum current swing was
set to 6A. The required minimum inductance of 12µH was calculated from Equation 3.1.

The first inductor was based on a RM core, and it was wound with 1.2mm diameter litz
wire. The number of turns and the air gap were 8 and 0.8mm respectively. The inductance
and the dc resistance of this inductor were recorded as 16µH and 33mΩ. The second
inductor was based on an E32 planar core. This planar inductor was made up of two double
layer pcbs, and the width of the pcb track was 95mil. The inductance and the dc resistance
of this inductor were recorded as 16µH and 59mΩ respectively. The third inductor was the
integrated inductor. Due to the core size of the integrated inductor, each integrated inductor
can only provide 5µH of inductance. Therefore in this performance test, three integrated
inductors were connected in series to make up a 15µH inductor, and the efficiency
comparison of the three inductors is shown in Figure 5.1.

Page 50

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

Figure 5.1. RM, planar inductor and integrated inductors efficiency comparison

This efficiency measurement is made by applying a constant input voltage 12 Volts and
constant input current 2A into the converter. Then varying the switching frequency from
100kHz to 200kHz, and then measure the output voltage and output current from the
resistive load and calculate the output power to determine the efficiency.

The experimental results in Figure 5.1 show the efficiency of the converter that uses the
planar inductor is typically around 1.5 ~ 2% lower than the inductor wound on the RM core.
This extra loss from the planar inductor is caused by the higher winding resistance; whereas
the litz wire that was used in the RM cored inductor has a much larger cross section area
compared to the pcb track. The advantage gained by using a planar inductor to produce a
low profile pcb outweighs the small loss in efficiency. However, by looking at the
efficiency of the integrated inductor, its performance is unsatisfactory. Due to the track
width of the integrated inductor being much narrower than the others; it ends up with the
highest dc resistance and the worst efficiency. Therefore, from considering the performance,
ease of construction and cost, the planar inductor was chosen.

The chosen planar inductor core for the 24W Buck-boost converter was Philips 3C90 E32
core. Due to the manufacturability, it is difficult to make more than two layers per pcb.
Therefore, the inductor was made by connecting two double layer pcb to achieve a total
four layer board.

Page 51

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

In the power converter design, minimising the temperature rise on the pcb is important for
reducing the conducting loss from the pcb tracks. The pcb track material used in this project
is 1oz copper. According to the temperature rise versus track width diagram shown on
Figure 3.6, for the single side pcb and the average current of 2A, the temperature rise can
be maintained under 20°C if the track width is wider than 50 mils. The same theory can
also be applied to the double side pcb; by having an identical operational current and 20°C
limitation on temperature rise, the minimum pcb width would be 100 mils.

In the E32 planar inductor, the winding space allows two turns of 110 mils wide pcb tracks
per layer. Therefore, for two double layer pcbs, eight turns in total were achieved in this
E32 planar inductor, and by adding the 0.8mm air gap, the inductance was measured as
35µH. The photo of the inductor is shown in Figure 5.2.

Figure 5.2. The inductor for the 24W buck-boost converter

5.1.2. MOSFET selection

In this battery equalisation system, 22 24W buck-boost converters need to be built, and in
Chapter 3 the requirements of the MOSFET were detailed. Ideally, to achieve a heatsinkless
design, the MOSFETs have to have low on-resistance to minimise the conduction loss, and
low gate charge to reduce the switching loss. However, due to the concerns of the cost and
the availability, it is not practical to select a MOSFET that has both low on-resistance and
gate charge.

Page 52

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

To minimise the power losses for this 24W buck-boost converter, the logic MOSFET was
chosen. The logic MOSFET is designed for low power converter use, and it normally has
the slightly lower gate charge than the standard MOSFET. Therefore, it can be switched as
low as 5V compared with 10V from the standard MOSFET. However, applying higher
switching voltage to the logic MOSFET can reduce the charging time on the gate
capacitance, which results a quick switching transition, and then reduces the switching loss.

The MOSFET used in this 24W buck-boost converter was SUD35N05-26L. The
SUD35N05-26L is a logic type MOSFET, which has the voltage rating of 55V and the
average current rating of 25A. When switching at 12V gate signal, the Rds-on is 20mΩ and
the gate charge is 22nC. It is also a surface mount component; therefore, the converter can
be made in low profile format. The final design of the 24W Buck-boost converter is shown
in Figure 5.3, and according to the experimental results, the maximum efficiency was
recorded as 92% when switching at 70kHz. Since the power loss from this converter is only
2W, the heatsink is not required in this converter.

Figure 5.3. The photo of the 24W Buck-boost battery equaliser

The measured switching signal of the design’s 24W buck-boost converter is shown in
Figure 5.4. It includes the waveforms of the PWM signal from the PWM controller, the
MOSFET drain to source voltage and the inductor current waveforms. The PWM controller
generates a 70kHz switching signal. When the switching signal is high, MOSFET turns on
and then the inductor is charged by the source battery; therefore, the inductor current ramps

Page 53

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

up. Then when the switching signal is low, MOSFET turns off, and then the stored energy
in the inductor discharges the stored energy into the lower charged battery.

Figure 5.4. 24W buck-boost converter action waveforms

5.2. The 192W Buck-boost converter

The design procedure of the 192W buck-boost converter is identical to the 24W buck-boost
converter. In the design of the 192W buck-boost converter, the Philips E43 planar inductor
was selected. In Chapter 3.2, the required inductance for this 192W buck-boost converter
was calculated as 166µH. Comparing with the previous inductor from the 24W buck-boost
converter, the larger inductance requires more turns to be made up, thus it requires
increasing the wiring resistance of the inductor. In this inductor, the wiring pcb was also
constructed by 2 double layer pcbs, but each layer has 3 turns, and the track width is 135
mil to increase the inductance and reduce dc resistance. The required air gap was calculated
as 0.4mm, and the inductance was measured as 191µH. Then, assuming input voltage
equals 96V, the ripple current is 6A and 50% duty-cycle, the minimum switching frequency
is calculated as 50.4kHz. by Equation 3.1.

Page 54

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

The chosen MOSFET for this 192W buck-boost converter was IRF740. The voltage and the
current ratings of this MOSFET are 400V and 10A respectively. The on-resistance Rds-on
and the gate charge of this MOSFET are 0.55Ω and 36nC, therefore the conducting loss
from the higher on resistance and the switching loss from the higher gate capacitance and
higher input voltage would result a larger power loss than the 24W buck-boost converter.

In order to dissipate the power losses from the MOSFETs, the heatsinks were required in
this 192W buck-boost converter. According to the experimental results, the best efficiency
was recorded as 90% when the switching frequency was 60kHz. A photo of the 192W
buck-boost converter is shown in Figure 5.5.

Figure 5.5. The 192W buck-boost converter

5.3. The 192W Flyback converter

In this 192W flyback converter, the required voltage and current rating are identical to the
192W buck-boost converter. Therefore, IRF740 was selected in this converter. The use of
the flyback converter is to provide an electrical isolation between the source and the load.
Therefore, the design of this flyback converter would be focused on the flyback transformer
design and reducing the voltage stress of the MOSFET.

Page 55

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

In Section 5.1 and 5.2, the inductors for both 24W and 192W buck-boost converters are
constructed based on the planar inductors, which result in an overall design with high
productivity, low profile and low cost. However due to the manufacturability and the
concern for the converter efficiency, the flyback transformer was wound on the standard
RM core, and in order to improve the transformer efficiency and experimental frequency
capability, litz wire was used.

The design procedure of the flyback transformer is described in Chapter 3.3. By making
different flyback transformers for different switching frequencies, the frequency response
curve was made to find the maximum efficiency point and the efficiency curve is shown in
Figure 5.6.

192W flyback converter frequency response

100
95
90
Efficiency (%)

85
80
75
70
65
60
50 60 70 80 90 100 110 120 130
Switching frequency (kHz)

Figure 5.6. The frequency response curve of the 192W flyback converter.

In this efficiency test, the converter reaches the highest efficiency at 60kHz. At this
frequency, the flyback transformer was designed to have 20 turns on the primary winding
and 16 turns on the secondary winding.

In this performance test, the converters were made without the clamping circuit. As
mentioned in Chapter 3.3, the clamping circuit is designed to reduce the voltage stress on
the MOSFET, hence improving the safety of the converter. However, due to the clamping
circuit slows down the MOSFET turn off time, the increased turn off time will decrease the
converter efficiency.

Page 56

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

To design the clamping circuit for this flyback transformer, the leakage inductances of the
primary and secondary windings were measured as 1.9µH and 1.2µH. To select the value of
power dissipating components Rc and Cc in the clamping circuit, the clamping voltage has
to be defined. In general practice, the clamping voltage can be set up to 30% higher than 2
times the input voltage without affecting the converter efficiency too much. Therefore, in
this design, a 1kΩ resistor and a 3.3nF capacitor were chosen. In Figure5.7 the switching
waveforms show the switching signal of the MOSFET.

(a) (b)
Figure 5.7.a. Switching and drain to source signals without clamping circuit.
5.7.b Switching and drain to source signals with clamping circuit.

Finally, this 192W flyback converter was made, and the photo of the converter is shown in
Figure 5.8. The maximum efficiency was recorded as 89% at the switching frequency of
60kHz

Figure 5.8 The photo of the 192W flyback converter

Page 57

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

The switching waveform of this 192W flyback converter is shown in Figure 5.9. It shows
when the primary switching signal is high, the primary MOSFET turns on and then the
primary inductor is charged by the source battery. Then when the primary switching signal
is low, the primary MOSFET turns off. The flyback action would transfer the stored energy
to the secondary side and then flow through the internal diode of the secondary side
MOSFET to recharge the load battery.

Figure 5.9. The switching waveform of the 192W flyback converter

In this Chapter, the construction procedures of each converter were described. In the design
of the 24W buck-boost converter, the investigation of several of inductor cores was made,
and the result shows the planar inductor becomes the better solution for this application in
terms of the construction and the efficiency. Further, the logic level surface mount type
MOSFET was used to reduce the switching loss and then making the construction of the
converter into a low profile, heatsinkless format. In the design of the 192W buck-boost
converter, the planar core was also chosen. However, because higher operating power and

Page 58

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 5: The system construction

larger power loss was involved, the surface mount type MOSFET was not suitable and the
heatsink was required in this 192W buck-boost converter. In the design of the 192W
flyback converter, due to the difficulty on making the multi-layer pcb and the concern for
the energy efficiency, the RM core was chosen to make the flyback transformer.

5.4. Summary
In this chapter, the efficiencies from each converter were also measured. In both 24W and
192W buck-boost converters, the efficiency were recorded as 92 and 90% respectively. In
the design of the 192W flyback converter, due to the voltage clamping circuit was
implemented to improve the safety of the converter, the increased switching time also
increases the switching loss. Then the efficiency was recorded as 89%.

In the next chapter, experimental battery equalisation tests using these converters are
detailed. The equalisation process will be done by equalising series connected batteries
after the batteries have been discharged at rate of 30A for an hour, and the required time for
equalising the entire battery string will be measured.

Page 59

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Chapter 6:Performance Records

6. Performance Records

In Chapters 3, 4 and 5, the designs, the simulations and the construction of each converter
making up the proposed battery equaliser topology are detailed and operational tests are
described. To confirm the correct operation of the overall equaliser topology, experimental
measurements of the battery equalisation on two banks of 4 × 12V batteries can provide a
reasonably clear indication, and with this configuration, the designed 192W converters,
which have already been tested to their rated capacities in Chapter 5 will be tested under
96W of load. These equalisation test results are given in this chapter.

The overall test was divided into two parts, the equalisation using the 192W non-isolated
buck-boost converter and equalisation using the 192W isolated flyback converter, which
are shown in Figure 6.1. In the first equalisation test, using the 192W non-isolated buck-
boost converter, two banks of batteries were connected in series. Within each bank, the four
batteries were equalised using three 24W non-isolated buck-boost converters. In the second
equalisation test, the two banks were isolated from each other, and the bank equalisation
was done using the 192W flyback converter. However, within each bank, the four series
connected batteries were still equalised by three 24W non-isolated buck-boost converters.

Figure 6.1. The equalisation setup

For these equalisation tests, each battery is rated 76Ahr, which is three times higher than
the batteries used in the EV3. Because of this, the equalisation speed will also be three
times slower. Before each equalisation test, all eight batteries were connected in parallel
and equally charged to around 13.5V. After that, these batteries were disconnected, settled
for three hours, and then reconnected into two banks of four series connected batteries and

Page 60
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

their open-circuit voltages are recorded. At this stage, four older batteries were intentionally
grouped into one bank to deliberately make the bank perform like a weak battery bank, and
the other four batteries were formed into a stronger battery bank. Since old batteries always
draw energy faster than newer ones, separating the old and the new batteries can make a
clear voltage difference between the two banks. Next, these two banks of batteries were
connected in series and then discharged by a carbon resistor at a rate of 30A for an hour,
which is approximately the average power consumption that is equal to the EV3 being
driven in the city area in an hour. After that, the whole battery string was left to settle for
three hours, and then the open-circuit voltage from each battery was recorded again.

During equalisation, the decision to turn on or off on each converter is not straightforward.
Because, for the open circuit battery monitoring measurement, the rate of change of voltage
on each battery is different. Besides, the rate of change of voltage on each battery when it
approaches full charge or nears the under charge condition are also different. Therefore a
comprehensive control system is required to accurately and efficiently turn on or off each
converter. Such control would be necessary when the equalisation system was installed in
the EV3. However the design of the control system for the equaliser was not part of the
brief for this project, because extensive testing of the system would be required first to
ascertain an appropriate control strategy. For the purposes of these results, each converter
was controlled manually according to a set of rules, which are described by Figure 6.2.

The rules for this equalisation process are simply when the open-circuit voltage between
any two adjacent batteries is greater than 0.2V, the 24W buck-boost converter will be
activated to balance these two batteries, or if the bank voltage difference is greater than
0.6V, the bank equalisation will start. Because the measured open-circuit voltage during the
equalisation process is much higher than its settled open-circuit voltage, the target battery
would be charged until its open-circuit voltage becomes 0.3 to 0.5V higher than the source
battery.

Page 61
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

Start

Measure the
open circuit voltage
of each bank

Measure the
open circuit voltage
of each battery inside
the banks

Battery Batteries
Yes
equalisation in equalisation
side each bank inside the bank ?

No

Bank Yes Bank

equalisation ? equalisation

No

End

Figure 6.2. Equalisation procedure

6.1. Battery equalisation between non-isolated banks

The first battery equalisation test was set up to balance two series-connected battery banks,
which were formed by four series connected batteries in each bank. Within each bank, three
24W buck-boost converters were used to equalise two adjacent batteries, and the set up
diagram is shown in Figure 6.3. After the entire battery string has been discharged at rate of
30A, the open-circuit voltages from each battery are recorded in Table 6.1.

Page 62
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

Figure 6.3. The setup diagram for non-isolated bank equalisation

Batt1 Batt2 Batt3 Batt4 Batt5 Batt6 Batt7 Batt8

Voltage
before 13.52 13.67 13.54 13.57 13.58 13.57 13.55 13.62
discharge
Voltage
after 11.05 11.08 11.09 11.2 11.45 11.57 11.59 11.42
discharge
Table 6.1. The measured battery open-circuit voltages

Page 63
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

In Table 6.1, Batt 1 to 4 was formed by four older batteries to make up the first battery bank.
Batt 5 to 8 was formed by four newer batteries to make up the second battery bank. The
following steps detail the battery equalisation process, which was based on the 192W and
24W buck-boost converter, and the equalisation results are shown in Figure 6.4.

Equalisation starts – Bank 1 voltage = 44.42V, Bank 2 voltage = 46.03V.

No battery equalisation required inside each bank.
Bank 2 starts transferring to Bank 1 by 192W buck-boost converter
Interval 1 – Bank 2 transfers energy to Bank 1 by 192W buck-boost converter
Point A – Bank 2 stops transferring energy to Bank 1 (Bank 1 idles)
Batt7 starts transferring energy to Batt8 by 24W buck-boost converter
Interval 2 – Batt7 transfers energy to Batt8 by 24W buck-boost converter
Point B – Batt7 stops transferring energy to Batt8
Bank 2 starts transferring energy to Bank 1 by 192W buck-boost converter
Interval 3 – Bank 2 transfers energy to Bank 1 by 192W buck-boost converter
Point C – Bank 2 stops transferring energy to Bank 1 (Bank 2 idles)
Batt2 starts transferring energy to Batt3 by 24W buck-boost converter
Interval 4 – Batt2 transfers energy to Batt3 by 24W buck-boost converter
Point D – Batt2 stops transferring energy to Batt3
Bank 2 starts transferring energy to Bank 1 by 192W buck-boost converter
Interval 5 – Bank 2 transfers energy to Bank 1 by 192W buck-boost converter
Point E – Bank 2 stops transferring energy to Bank 1 (Bank 2 idles)
Batt4 starts transferring energy to Batt3 by 24W buck-boost converter
Interval 6 – Batt4 transfers energy to Batt3 by 24W buck-boost converter
Point F – Batt7 starts transferring energy to Batt8 by 24W buck-boost converter
Interval 7 – Batt4 transfers energy to Batt3 by 24W buck-boost converter
Batt7 transfers energy to Batt8 by 24W buck-boost converter
Point G – Batt4 stops transferring energy to Batt3
Batt1 starts transferring energy to Batt2 by 24W buck-boost converter
Interval 8 – Batt7 transfers energy to Batt8 by 24W buck-boost converter
Batt1 transfers energy to Batt2 by 24W buck-boost converter
Point H – Batt7 stops transferring energy to Batt8 (Bank 2 idles)
Interval 9 – Batt1 transfers energy to Batt2 by 24W buck-boost converter
Point I – Batt1 stops transferring energy to Batt2. Bank 1, Bank 2 idle. Equalisation Ends

Page 64
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

Figure 6.4. Battery equalisation process for an eight series connected batteries

In this equalisation test, two banks of battery were connected in series, and this equalisation
process shows these two banks can be successfully equalised by the combination of the
192W and 24W buck-boost converters.

Page 65
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

6.2. Battery equalisation between isolated banks

The second battery equalisation test was set up to balance two isolated battery banks. Inside
each bank there were four batteries connected in series. In this equalisation test, the bank
equalisation was done by the 192W flyback converter and the individual battery inside each
bank was balanced by the 24W buck-boost converter. The set up diagram is shown in
Figure 6.5.

Figure 6.5. The setup diagram for isolated bank equalisation

Page 66
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

Again, a battery discharging process has done before the equalisation process, and the
measured open-circuit voltage are recorded in Table 6.2.

Batt1 Batt2 Batt3 Batt4 Batt5 Batt6 Batt7 Batt8

Voltage
before 13.76 13.77 13.74 13.68 13.75 13.73 13.7 13.79
discharge
Voltage
after 11.76 11.74 11.92 11.71 12.01 11.88 11.79 11.9
discharge
Table 6.2. The measured battery open-circuit voltages

The following steps detail the battery equalisation process, which was based on the 192W
flyback converter and 24W buck-boost converters, and the equalisation results in Figure
6.6 shows two isolated battery bank can be successfully balanced.

Equalisation starts – Batt3 starts transferring energy to Batt4 by 24W buck-boost converter
Batt5 starts transferring energy to Batt6 by 24W buck-boost converter
Batt8 starts transferring energy to Batt7 by 24W buck-boost converter
Interval 1 – Batt3 transfers energy to Batt4 by 24W buck-boost converter
Batt5 transfers energy to Batt6 by 24W buck-boost converter
Batt8 transfers energy to Batt7 by 24W buck-boost converter
Point A – Batt8 stops transferring energy to Batt7
Interval 2 – Batt3 transfers energy to Batt4 by 24W buck-boost converter
Batt5 transfers energy to Batt6 by 24W buck-boost converter
Point B – Batt5 stops transferring energy to Batt6
Interval 3 – Batt3 transfers energy to Batt4 by 24W buck-boost converter
Point C – Batt3 stops transferring energy to Batt4
Bank 2 starts transferring energy to Bank 1 by 192W flyback converter
Interval 4 – Bank 2 transfers energy to Bank 1 by 192W flyback converter
Point D – Bank 2 stops transferring energy to Bank 1
Interval 5 – System idle
Point E – Batt3 starts transferring energy to Batt2 by 24W buck-boost converter
Interval 6 – Batt3 transfers energy to Batt2 by 24W buck-boost converter
Point F – Batt6 starts transferring energy to Batt7 by 24W buck-boost converter
Interval 7 – Batt3 transfers energy to Batt2 by 24W buck-boost converter
Batt6 transfers energy to Batt7 by 24W buck-boost converter
Point G – Batt3 stops transferring energy to Batt2

Page 67
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

Batt6 stops transferring energy to Batt7

Equalisation ends

Figure 6.6. Equalisation process for the isolated battery banks

Page 68
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 6:Performance Records

6.3 Summary

In this chapter, two battery equalisation tests are detailed and the results indicate that each
individual battery can be successfully equalised to adjacent batteries. However, there was
one disadvantage found during the test. That is, when the 192W converter is turned on, the
24W converters have to be off. Otherwise the 192W and the 24W converters would affect
each other, and making the overall system unstable. When two banks are equalised by the
192W converter, the rest of 24W converters have to wait until the 192W converter finishes
the bank equalisation. To solve this problem, reducing the number of batteries per bank can
increase the chance of multiple banks working at the same time. However by doing this, the
system would require more converters to do the bank equalisation, which increases the cost
and increases the size and weight of the system. Within each bank, multiple 24W
converters can be turned on at the same time without any problem. Therefore how to
configure the number of converters required to equalise the bank, the batteries inside the
bank and the future expansion become an important future consideration.

Additionally, in this chapter, each converter was turned on or off manually, but an
automatic control scheme is required in the real world application. A possible solution
could be a microprocessor-based control system with an intelligent control algorithm to
monitor the condition of each battery and then give commands to each converter. However,
this controller needs to accurately measure how much charge each battery has and how long
the charging process needs to be done.

Page 69
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 7: Conclusion

7. Conclusion

A prototype battery equalisation system has been designed, constructed and tested for the
EV3, which is powered by 26 series-connected lead acid batteries divided into four banks.
This prototype battery equalisation system is based on the ring equaliser topology. The
overall system requires three different types of converters, each designed to operate under
average current mode control at a rate of 2A. The three types are the 24W buck-boost
converter, the 192W buck-boost converter and 192W flyback converter. The 24W and the
192W buck-boost converters are non-isolated converters and are designed for transferring
energy between any two adjacent batteries or banks. The 192W flyback converter is an
isolated converter that links the top and bottom battery banks to complete the ring structure.

Simulations of the designed equaliser were performed using a PSpice simulator. In the
simulation section, two rechargeable batteries or banks of batteries were modelled as two
1F capacitors, and the battery internal resistance and the wiring resistances were modelled
as two 100mΩ resistors. To determine the suitable initial nominal voltage for each battery
and the voltage variation from each battery, a 30A, one hour battery discharge test over
eight equally charged series connected batteries was done. Drawing energy from each
battery at a rate of 30A is equivalent to drawing 10kW of energy from the EV3, which is
approximately equal to the power consumption of standard city driving. In the simulation
test, each converter was also designed to operate at a rate of 2A, and the simulation results
showed that two rechargeable batteries can be successfully equalised by the proposed
converter.

The aim of the design and construction is for each converter to have an efficiency of at least
90% and to be physically compact, light-weight and of a low profile format. To achieve a
high efficiency at a reasonably low cost, the switching frequency and the MOSFET for each
converter were carefully selected. In the design of the 24W and 192W buck-boost
converters, the planar inductor was introduced to improve the manufacturability and to
enable the converter to be made in a low profile format without significant efficiency
reduction.

Page 70
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 7: Conclusion

However, planar technology is not the universal solution for all inductor design. In the
design of the flyback converter, the required numbers of turns for both primary and
secondary winding are much higher than the number of turns used in the buck-boost
converter. Therefore, in order to make the pcb track wide enough to minimise the dc
resistance and at the same time fit a greater numbers of turns into a limited area, the multi-
layer pcb is essential. However, due to the difficulties of making a multi-layer pcb, the
planar technology was abandoned in the flyback converter design, and instead, a RM core
was used and the primary and secondary windings were wound with litz wire to improve
frequency capability and reduce the dc resistance.

Correctly sensing the current is the most important part in the buck-boost converter design.
The current sensing resistor is not connected to the voltage reference point like the flyback
converter. Therefore, to make the current sensing function work, a differential amplifier had
to be used to sense the voltage difference across the current sensing resistor. A differential
amplifier, which has a wide common mode range, is required. Most modern op-amps can
be used in the 24W buck-boost converter, but the op-amp selection for the 192W buck-
boost is very limited. The reason for this is the controller is powered by 12V, but the
current sensing resistor is sitting on top of and connected to the 72V battery bank.
Therefore, if the common mode range of the op-amp is not high enough, the current sensing
will not work properly or will become less sensitive to the current signal.

In the 192W flyback converter design, the biggest challenge was the design of the voltage
clamping circuit. The voltage clamping circuit consists of a diode, a resistor and a capacitor,
and the use of this was to reduce the voltage stress on the MOSFET during the transition of
the turn off time. To ensure the voltage can be clamped under the expected value, the
expected turn off speed, the converter efficiency, the amounts of leakage inductance and
leakage power need to be considered. The component value used in this flyback converter
may not be the most suitable value, but it significantly reduced the MOSFET voltage stress
with only 2% efficiency reduction.

Page 71
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Chapter 7: Conclusion

Overall, a prototype of a battery equalisation system for the EV3 has been successfully
designed, built and tested. The equalisation results show series-connected batteries can be
equalised by either the combination of 24W buck-boost converters and a 192W buck-boost
converter, or the combination of 24W buck-boost converters and a 192W flyback converter.
However, there are two disadvantages found during these equalisation tests. First, because
there is not any control system to accurately measure the open-circuit voltage from each
battery and calculate how much time does the equalisation is required. Manual control the
action of each converter is impractical. Second, since the 192W and the 24W converters
can not be turned on at the same time, or any two 24W converters can not operate from the
same battery at the same time. The time consumed on waiting for the other converter to
complete their part of the equalisation process may be significant, which contributes to
overall system inefficiency.

To improve the system capability, a microprocessor-based control system and a high

accuracy battery monitoring system need to be implemented, and hence provide a more
flexible control strategy and bring about a more efficient equalisation processes. Overall,
this project gives a clear indication on how the designed battery equalisation system can be
used on a series-connected battery string and what are the design considerations required.

Page 72
PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com
 Appendix

Appendix

The schematics of 24W buck-boost converter

192W buck-boost converter
192W flyback converter

The PCB of 24W buck-boost converter

192W buck-boost converter
192W flyback converter

The data sheet of E32 planar inductor

E43 planar inductor
SUD35N05-26L MOSFET
IRF740 MOSFET

Page 73

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The schematics of 24W buck-boost converter

Page 74

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The schematics 192W buck-boost converter

Page 75

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The schematics 192W flyback converter

Page 76

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The PCB of 24W buck-boost converter

Page 77

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The PCB of 192W buck-boost converter

Page 78

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The PCB of 192W flyback converter

Page 79

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The data sheet of E32 planar inductor

Page 80

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The data sheet of E43 planar inductor

Page 81

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The data sheet of SUD35N05-26L MOSFET

Page 82

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

Page 83

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

The data sheet of IRF740 MOSFET

Page 84

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com

 Appendix

Page 85

PDF Created with deskPDF PDF Writer - Trial :: http://www.docudesk.com