Mailadil T. Sebastian (Editor) - Microwave Materials and Applications, 2 Volume Set-Wiley (2017)

Islamic University of Gaza
Deans of Graduate Studies

Faculty of Engineering
Electrical Engineering Department
Coupled Resonator Diplexer for LTE

LTE-Advanced
Advanced
System
By
Hazem M. Abukaresh
Supervisor
Dr
Dr. Talal F. Skaik
A Thesis Submitted in Partial Fulfillment of the Requirements for the

Degree of Master in Electrical/Communication Engineering
‫ م‬2013 - ‫ھــ‬1434
Abstract
This thesis presents the design of a microstrip hairpin diplexer. The design is based on
coupled-resonator structure using U-Shaped resonators. It is designed to meet The Long
Term Evolution-Advanced (LTE-A) system Band 7, operation at uplink (UL): 2.50–
2.57 GHz and downlink (DL): 2.62–2.69 GHz for base transceiver station antenna. The
structure consists of three ports 10-coupled resonators with direct coupling to produce
diplexer with chebyshev filtering response. The diplexer does not involve any external
junctions for distribution of energy, so it can be miniaturized in comparison to
conventional diplexers.
The synthesis procedure of the proposed coupled-resonator diplexer is based on optimization of

coupling matrix. The optimization algorithm is based on minimizing a cost function that is
formulated with a minimum set of characteristics that can completely describe the response.
Both unconstrained and constrained local optimization techniques have been utilized to
synthesise the coupling matrix, and a numerical formula has been used to calculate the external
quality factors of the diplexer with symmetrical channels. A local gradient based optimization
technique has been employed here to produce coupling matrix for the coupled resonator
diplexer.
EM simulator (CST Microwave studio) has been used to extract the desired design
dimensions according to a prescribed diplexer design parameters. Global and local
optimization techniques are applied to whole structure to improve the initial response.
i
‫ملخص الرسالة‬
‫تھدف ھذه الرسالة الى تصميم جھاز المبدل التناوبي )‪ (Diplexer‬وھو يعمل على تناوب الھوائي ﻣا بين إرسال‬
‫واستقبال االشارة في الوقت ذاته ولكن على ترددين ﻣختلفين‪ ،‬أي استخدام قناتين ذات ترددين كل ﻣنھما في أتجاه و‬
‫بنفس الوقت‪ ،‬ﻣا يتطلب فصل ترددات الوصلة الصاعدة )‪ (Uplink‬وترددات الوصلة الھابطة )‪ (Downlink‬و‬
‫ھذا ﻣا يدعى بـ اإلرسال المزدوج بالتقسيم الترددي )‪ ،(FDD‬والذي بدونه سوف تتداخل ترددات الوصلة الصاعدة‬
‫وترددات الوصلة الھابطة ﻣع بعضھا البعض‪.‬‬
‫تقوم فكرة اإلرسال المزدوج على تقسيم نطاق الطيف الترددي )‪ (Spectrum Frequency‬الى نطاقين ‪ ،‬نطاق‬
‫لألرسال واالخر لالستقبال‪ ،‬كما يتم تقسيم كل نطاق ترددي الى عدد ﻣعين ﻣن القنوات ويتم تعيين قيمة ترددية لكل‬
‫كاف بين القنوات المتجاورة تفاديا ً للتداخل‪.‬‬
‫ٍ‬ ‫ﻣنھا ﻣع ﻣراعاة تحديد فاصل ترددي‬
‫صم َم ﺿمن نظام االتصاالت المتقدﻣة المتنقلة الدولية ‪ -‬شبكات التطور طويل األجل‬
‫يعمل جھاز المبدل التناوبي ال ُم َ‬
‫ﻣتقدم )‪ (LTE-A‬والتي تعرف بالجيل الرابع ﻣن الالسلكية الخلوية وھي خليفة لمعايير الجيل الثالث‪.‬‬
‫وقد تم تصميمه بحيث يخدم الطيف الترددي السابع للموجات الراديوية ﺿمن شبكات التطور طويل األجل ﻣتقدم‪،‬‬
‫والمستخدم في تراسل المعلوﻣات السلكيا والممتد في ﻣجال التردد ﻣن ‪ 2.50 GHz – 2.57 GHz‬للموجات‬
‫الصاعدة الى ‪ 2.62 GHz – 2.69 GHz‬للموجات الھابطة‪.‬‬
‫صمم جھاز المبدل التناوبي في ھذه الرسالة استنادا الى تقنية الرنين الذاتي والمزدوج المشكل لمصفوفة االقتران‬
‫)‪ (Coupling Matrix‬والتي يتم الحصول عليھا عن طريق التحسين )‪ (Optimization‬لمعادلة الكلفة ) ‪Cost‬‬
‫‪ (Function‬لھيكل المبدل التناوبي المكون ﻣن ثالثة ﻣنافذ تعمل كمدخالت وﻣخرجات بجانب ‪ 10‬رنانات بينھا‬
‫اقتران ﻣباشر )رنين ﻣزدوج ﻣباشر( للحصول على تصفية تشيبيشيف )‪ (Chebyshev Filtering‬في عملية‬
‫ترشيح االشارة‪.‬‬
‫استخدﻣت تقنية التحسين القائمة على التدرج )‪ (A gradient based optimization‬في انتاج ﻣصفوفة االقتران‬
‫والتي استعملت في ﻣا بعد للحصول على األبعاد المناسبة للرنانات والمسافات بينھما لتحقيق استجابة المبدل التناوبي‬
‫المطلوبة‪.‬‬
‫وظف برناﻣج للمحاكاة في تصميم جھاز المبدل التناوبي المؤلف ﻣن خطوط الشريط الدقيق )‪ ،(Microstrp‬ليصبح‬
‫التركيب النھائي عبارة عن ‪ 10‬أشرطة ناقله تم لويھا بشكل حدبة الحصان ﻣطبوعة فوق سطح لوحة عازلة ذات‬
‫سماحية نسبية ﻣرتفعة وسطحھا االّخر يمثل قاعدة أرﺿية ناقلة‪.‬‬
‫المبدل التناوبي ال يحتوي على أية رابطة خارجية لتوزيع الطاقة فأصبح باإلﻣكان الحصول على حجم أصغر إذا ﻣا‬
‫قورن بالمبدل التناوبي التقليدي‪.‬‬
‫‪ii‬‬
Acknowledgements
I have been indebted in the preparation of this thesis to my supervisor, Dr Talal Skaik,
who patience and kindness, as well as his academic experience, has been invaluable to
me.His patience and support helped me overcome many crisis situations and finish this
thesis. I hope that one day I would become as good an advisor to my students as Dr.
Talal has been to me.
I would like to thank my parents, Mousa and Sanaa, without their continuous support
and encouragement I never would have been able to achieve my goals. This one is for
you mom and dad!
I would like to express gratitude to my brothers and sister, Haythem, Hussam and Haya
whose constantly provide emotional support and took care of me in many aspects.
My wife Heba has been, always, my pillar, my joy and my guiding light, and I thank
her.
I also would like to thank to my daughter, Rahaf who joined us when I was writing my
thesis, for giving me unlimited happiness and pleasure.
Many friends have helped me stay sane through these difficult years. Their support and
care helped me overcome setbacks and stay focused on my graduate study. I greatly
value their friendship and I deeply appreciate their belief in me.
My acknowledgement also goes to the department of Electrical and Communication

Engineering at the Islamic University of Gaza to give me the chance to complete my
graduate study.
iii
Table of Contents
List of Figures............................................................................................................................ vi
List of Tables ........................................................................................................................... viii
Chapter 1: Introduction .......................................................................................................... 1
1.1 Technology Background ................................................................................................ 1
1.2 Evolution of wireless standards..................................................................................... 1
1.3 LTE-A Bands ................................................................................................................... 6
1.3.1 The 2.6 GHz Spectrum Band “Band 7”.............................................................. 6
1.4 Overview of Diplexers and their Application ........................................................... 10
1.5 Thesis Motivation.......................................................................................................... 11
1.6 Thesis Overview ............................................................................................................ 11
References............................................................................................................................. 13
Chapter 2: Overview of Diplexers ...................................................................................... 16

2.1 Introduction .................................................................................................................... 16
2.2 TDD Systems ................................................................................................................. 16
2.3 FDD Systems ................................................................................................................. 17
2.4 Conventional Diplexer.................................................................................................. 18
2.4.1 Literature Review ................................................................................................ 21
2.5 Coupled Resonator Diplexer........................................................................................ 25
2.5.1 Diplexers with star-junction ............................................................................... 25
2.5.2 Diplexers with a Common Resonator Junction ............................................... 26
2.5.3 General theory of couplings ............................................................................... 27
2.5.4 Three port networks ............................................................................................ 29
2.5.5 N-Port Coupled Resonator Circuits .................................................................. 30
2.5.5.1 Deriving Coupling Matrix of N-port Networks ................................. 31
2.5.5.1.1 Circuits with magnetically coupled resonators .................. 31
2.5.5.1.2 Circuits with electrically coupled resonators ..................... 35
2.5.6 General coupling matrix ..................................................................................... 39
2.6 Chebyshev Response .................................................................................................... 40
2.7 Summary ........................................................................................................................ 42
References............................................................................................................................. 44
Chapter 3: Synthesis of Coupled Resonator Diplexer .................................................. 48

3.1 Introduction .................................................................................................................... 48
3.2 Optimization Techniques ............................................................................................. 49
3.2.1 A gradient based local optimization ................................................................ 50
3.3 Cost function .................................................................................................................. 52
3.4 Optimization algorithm ................................................................................................ 55
3.5 Calculation of external quality factor ......................................................................... 57
3.6 Initial spacing of reflection zeros ................................................................................ 58
3.7 Diplexers with T-Topology ......................................................................................... 59
3.7.1 Diplexer with n=8, x=0.263157, and r=3 ......................................................... 60
iv
3.7.2 Diplexer with n=10, x=0.263157, and r=4 ....................................................... 64
3.7.3 Diplexer with n=12, x=0.263157, and r=5 ....................................................... 68
3.8 Frequency transformation ............................................................................................ 72
3.9 Summery ........................................................................................................................ 74
References............................................................................................................................. 75
Chapter 4: Diplexer Design .............................................................................................. 76

4.1 Introduction .................................................................................................................... 76
4.2 Microstrip ....................................................................................................................... 76
4.2.1 Microstrip Design ................................................................................................. 78
4.2.2 Resonators .............................................................................................................. 79
4.3 Hairpin Diplexers .......................................................................................................... 79
4.3.1 Hairpin model ...................................................................................................... 80
4.3.2 Hairpin Coupling Structures .............................................................................. 81
4.3.3 Unloaded quality factor ...................................................................................... 83
4.4 Coupling Coefficients and external coupling, Simulation and analysis ................ 84
4.4.1 Coupling in Physical Terms ............................................................................... 85
4.4.2 Extraction of coupling coefficient from physical structure ........................... 85
4.4.3 Extraction of external quality factor from physical structure ........................ 87
4.5 LTE-A 10-resonator diplexer ...................................................................................... 88
4.5.1 Diplexer Design ................................................................................................... 88
4.5.2 Diplexer physical structure and simulation...................................................... 92
4.6 Summary ........................................................................................................................ 95
References............................................................................................................................. 96
Chapter 5: Conclusions and Future Work..................................................................... 97

5.1 Conclusions .................................................................................................................... 97
5.2 Future Work ................................................................................................................... 98
Appendix A .......................................................................................................................... 99
ChebyshevLowpass Prototype Filters ............................................................................... 99
Appendix B ........................................................................................................................ 101

Gradient based Algorithm ................................................................................................ 101
v
List of Figures
Figure 1-1: Illustrated many European countries working in co-operation with each
other to develop GSM system ................................................................................................... 2
Figure 1-2: Evolution towards third generation mobile systems in terms of data rate
support .......................................................................................................................................... 4
Figure 1-3: Wireless evolution towards fourth generation mobile systems ....................... 5
Figure 1-4: ITU Options for the 2.6GHz Band....................................................................... 8
Figure 1-5: transceiver configuration example of multi-mode/multi-band ........................ 9
Figure 1-6: Conventional diplexer; Hardware structure (left) and frequency response
(right) ......................................................................................................................................... 10
Figure 2-1: The principle of operation in TDD systems ..................................................... 17
Figure 2-2: The principle of operation in FDD systems...................................................... 18
Figure 2-3: Conventional duplexer ........................................................................................ 19
Figure 2-4: Configuration of diplexer with a 1:2 divider multiplexing network ............. 20
Figure 2-5: Configuration of circulator-coupled diplexer ................................................... 20
Figure 2-6: Configuration of manifold-coupled diplexer .................................................... 21
Figure 2-7: Layout of the diplexer using the T-shaped resonator, R1, to combine two
second-order bandpass filters. Port 1 uses coupled feeding, ports 2 and 3 use tapped
feeding ........................................................................................................................................ 21
Figure 2-8: Fabricated microstrip diplexer............................................................................ 22
Figure 2-9: Fabricated microstrip hairpin diplexer .............................................................. 22
Figure 2-10: The Microstrip Diplexer.................................................................................... 23
Figure 2-11: The microstrip diplexer ..................................................................................... 23
Figure 2-12: Circuit layout of the diplexer............................................................................ 24
Figure 2-13: a compact diplexer using a square open loop................................................. 24
Figure 2-14: General schematic for star-junction multiplexers .......................................... 26
Figure 2-15: Diplexer with an extra resonator as common junction ................................. 27
Figure 2-16: General coupled microwave resonators. Resonators 1 and 2 can be
different in structure and have different resonant frequencies ........................................... 28
Figure 2-17: Inter-coupling between coupled resonators. (a) Coupled resonator circuit
with electric coupling. (b) Coupled resonator circuit with magnetic coupling. (c)
Coupled resonator circuit with mixed electric and magnetic coupling ............................. 28
Figure 2-18: three-port network.............................................................................................. 29
Figure 2-19: (a) Equivalent circuit of magnetically n-coupled resonators in N-port
network, (b) Equivalent circuit of electrically n-coupled resonators in N-port network 31
Figure 2-20: Network representation of 3-port circuit ........................................................ 33
Figure 2-21: Network representation of 3-port circuit ........................................................ 37
Figure 2-22: The frequency response of a sixh-order type I Chebyshev low-pass filter 41
Figure 2-23: The Inverse Chebyshev Response .................................................................. 42
Figure 3-1: n -resonator based diplexer ................................................................................. 48
Figure 3-2: Steepest ascent directions.................................................................................... 51
Figure 3-3: Flowchart of optimization algorithm ................................................................. 56
Figure 3-4: Lowpass prototype filters (a) passband edges of ±1 Hz, (b) passband edges
of x and 1 Hz ............................................................................................................................. 57
vi
Figure 3-5: (a) frequency spectrum allocation for band 7 in LTE-Advanced cellular
telephone, (b) normalized bandwidth for band 7 in LTE-Advanced cellular
telephone .................................................................................................................................... 58
Figure 3-6: Diplexer T-Topology ........................................................................................... 59
Figure 3-7: Topology of Diplexer with 8 resonators ........................................................... 60
Figure 3-8: Diplexer prototype response with 8 resonators from the first stage of
optimization process................................................................................................................. 63
Figure 3-9: Diplexer prototype response with 8 resonators from the second stage of
Figure 3-10: Topology of Diplexer with 10 resonators ....................................................... 65
Figure 3-11: Diplexer prototype response with 10 resonators from the first stage of
Figure 3-12: Diplexer prototype response with 10 resonators from the second stage of
Figure 3-13: Topology of Diplexer with 12 resonators ....................................................... 69
Figure 3-14: Diplexer prototype response with 12 resonators. (a) From the first stage of
optimization process. (b) From the second stage of optimization process ....................... 71
Figure 3-15: Lowpass to bandpass transformation .............................................................. 72
Figure 3-16: Diplexer bandpass response.............................................................................. 73
Figure 4-1: Microstrip structure.............................................................................................. 76
Figure 4-2: Hairpin resonator.................................................................................................. 80
Figure 4-3: Structural variations to miniaturize hairpin resonator. (a) Conventional
hairpin resonator. (b) Miniaturized hairpin resonator with loaded lumped capacitor.
(c) Miniaturized hairpin resonator with folded coupled lines............................................. 81
Figure 4-4: Hairpin Structures. (a) Tapped line input 8-pole Hairpin diplexer. (b)
Coupled line input 8-pole Hairpin diplexer .......................................................................... 82
Figure 4-5: Basic coupling structures of coupled microstrip hairpin resonators. (a)
Electric coupling structure. (b) Magnetic coupling structure. (c) Mixed coupling
structure...................................................................................................................................... 83
Figure 4-6: Two coupled hairpin resonators ......................................................................... 86
Figure 4-7: Amplitude response of S 21 for two coupled resonators .................................. 86
Figure 4-8: Externally coupled hairpin resonator................................................................. 88
Figure 4-9: Amplitude response of S 21 for externally coupled resonator ......................... 88
Figure 4-10: The layout of the LTE-A 10-resonator diplexer ............................................ 89
Figure 4-11: Diplexer topology .............................................................................................. 90
Figure 4-12: Diplexer prototype response ............................................................................. 91
Figure 4-13: The layout of the LTE-A 10-resonator diplexer design ................................ 94
Figure 4-14: The EM simulated performance of the diplexer ............................................ 95
vii
List of Tables
Table 1-1: Operation bands for LTE-Advanced ............................................................... 7

Table 3-1: Coupling matrix of diplexer with 8 resonators from the first stage of
optimization process ........................................................................................................ 62
Table 3-2: Coupling matrix of diplexer with 8 resonators from the second stage of
Table 4-1: Normalized coupling matrix of 10-resonator diplexer................................... 90
Table 4-2: The actual coupling matrix of 10-resonator diplexer ..................................... 91
Table 4-3: Resonators and their resonant frequencies ..................................................... 92
Table 4-4: The LTE-A 10-resonator diplexer dimensions .............................................. 93
viii
Chapter 1
Introduction
1.1 Technology Background
Communication is everywhere, used by everybody, and at any time - Cellular

telecommunication system is considered as one of the fastest growing and most
confrontation telecommunication application ever. This will be clear from large and
continuously increasing percentage of all new telephone subscribers around the world.
Cellular telecommunication succeeded to have the ability to provide a wide range of
applications and may become the universal way of communication.
The objective of this Thiess is to analyze and design a coupled resonator

diplexer that meets the specification of the fourth generation (4G) of Cellular
telecommunication system Long-Term Evolution-Advanced (LTE-A) band 7,
specifically UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz for base transceiver station
antennas. LTE-A has been chosen for the current research because it is the future for the
cellular system, as most major mobile carriers in the United States and several
worldwide carriers have announced plans to convert their networks to LTE- A. Band 7
has been chosen since it’s a new band, and it has great potential to become a global
extension band for International Mobile Telecommunications-Advanced (IMT-
Advanced), and fixed broadband access infrastructure widespread.
1.2 Evolution of wireless standards
With the invention of microprocessors and the release of the cellular

communications concept in the period between 1970s and 1980s, the first generation
(1G) mobile communication systems were born [1]. The Major 1G standards are Nordic
Mobile Telephone (NMT) in Scandinavia, Advanced Mobile Phone Service (AMPS) in
USA, Total Access Communication System (TACS) and its extended version (ETACS)
in UK , C-450 in Portugal and Germany, and Nippon Telegraph and telephone (NTT)
in Japan[1,2].
The Second generation (2G) mobile communication systems were created to

address the need of the increased capacity over first generation analogue systems. 2G
standards utilizing a digital technology over switch circuit networks [3].
1
The main 2G standards are Interim standard 95 (IS-95) in USA, Personal Digital
Cellular (PDC) in Japan, IS-136 (Digital AMPS) in USA and Global System for Mobile
Communications (GSM) which has been specified and developed by many European
countries working in co-operation with each other as illustrated in Figure 1-1 [1].
Figure 1-1: illustrated many European countries working in co-operation with each other to
develop GSM system [3].
2
2.5G systems address the data capacity limitations correlating with the 2nd
generation systems. GSM introduced General Packet Radio Service (GPRS) which can
be used for services like Wireless Application Protocol (WAP) access, Multimedia
Messaging Service (MMS) and for internet services including email and World Wide
Web access [4]. IS-95 is upgrade to IS-95B which provides users with an efficient
mechanism for wireless internet browsing and email access [1]. NTT DoCoMo (Nippon
Telegraph and Telephone DoCoMo) developed their own proprietary packet based
technology called i-mode on its PDC network which supports games, color, graphics
and interactive web page [1]. IS-136 is developed to IS-136 HS.
During the transition from the second generation to the third generation a 2.75G
has appeared which is considered as a pre 3G that allows GSM operators to use existing
GSM radio bands to offer wireless multimedia IP-based services and applications [5].
In this generation GPRS network evolved to Enhanced Data Rates for GSM Evolution
(EDGE) networks which improved data transmission rates and services [5].
Third Generation (3G) Technology was developed in order to face up to the new
requirements of services of what were coming, as high-quality images and videos or to
provide access to the Web with higher data rates. The data-handling capabilities of 2G
systems are limited and are necessary for other mobile technology. 3G systems provide
a platform that is common for multiple wireless standards and technologies [6]. 3G
systems supported technologies is to create a single global standard that allows for
global roaming. The five 3G standards included in IMT-2000 terrestrial radio interfaces
are: The IMT-2000 defined highlights five distinct mobile/terrestrial radio interface
standards [7,8]:
1- IMT-MC: CDMA Multi-Carrier (known as cdma2000 or IS-2000)
2- IMT-DS: CDMA Direct Spread (known as Wideband Universal Mobile
Telecommunications System (UMTS) or WCDMA-FDD)
3- IMT-TC: CDMA TDD (WCDMA-TDD)
4- IMT-SC: TDMA Single Carrier (known as UWC-136 and EDGE)
5- IMT-FT: TDMA Multi-Carrier (well known as DECT, Digital Enhanced
Cordless Telecommunication)
3
Figure 1-2 show the evolution of mobile cellular standards until 3G which include
the above standards. CDMA 2000 3X Radio Transmission Technology (3xRTT) is
upgrade to Evolution for Data Optimized (EV-DO) [9,10].
Figure 1-2: Evolution towards third generation mobile systems in terms of data rate support [1,7].
3.5 G is a grouping of disparate mobile telephony and data technologies

designed to provide better performance than 3G systems. High-Speed Downlink Packet
Access (HSDPA) and CDMA2000 1x Evolution for Data Optimized Revolution A
(1xEV-DO REV. A) were among 3.5 systems [11].
3.75G provides mobile broadband access of several Mbit/s to smart phones and
mobile modems in laptop computers. The main standards of 3.75G are High-Speed
Uplink Packet Access (HSUPA), High-Speed Orthogonal Packet Access (HSOPA)
which known as HSPA Evolution (HSPA+) and CDMA2000 Evolution for Data
Optimized Revolution (1x-EV-DO Advanced) [12,13].
The Long-Term Evolution (LTE) is often called “3.9 G”, which the last major
addition of radio access features to 3G. The LTE project was start in 2004. The
4
motivation for LTE is the desire for a reduction in the cost per bit beside of lower cost
services with better user experience,
xperience, the flexible use of new and existing frequency
bands, a simplified and lower cost network with open interfaces, and a reduction in
terminal complexity with an allowance for reasonable power consumption [14].
The fourth generation (4G) is often ccalled LTE-Advanced (LTE-A)

A) which is
evolution to 2G and 3G wireless standards and provide a standard that ensures the
interoperability between satellite, terrestrial wireless mobile technologies and Wireless
Local Area Networks (W-LANs).
LANs). This evolution is illustrated in Figure 1-3,
1 which
shows that one standard, are being proposed for 4G than in previous generations [15].
Figure 1-3:
3: Wireless evolution towards fourth generation mobile systems [1,7,11,12,13,14].
5
1.3 LTE-A Bands
There is a growing number of LTE-A frequency bands that are being designated
as possibilities for use with this system. Many of the LTE-A frequency bands are
already defined for pervious standards, where other are new and being introduced as
other users are re-allocated spectrum elsewhere. Operating bands of LTE-A will involve
Evolved- (UMTS) Terrestrial Radio Access (E-UTRA) operating bands as well as
possible The International Mobile Union (IMT) bands identified by The International
Telecommunications Union-Radio communications sector (ITU-R). E-UTRA (LTE-A)
operating bands are shown in Table 1-1.
Band 7 has been chosen since it’s a new band, and it has great potential to
become a global extension band for IMT-Advanced, and fixed broadband access
infrastructure widespread [16,17].
1.3.1 The 2.6 GHz Spectrum Band “Band 7”
The 2.6 GHz band (2500-2690 MHz), sometimes also referred as the 2.5 GHz band
was allocated by the World Radio Communication Conference (WRC) in 2000 for
terrestrial mobile communications services. The band provides an opportunity to meet
rapidly rising demand for capacity to deliver mobile broadband services on a
widespread, common basis across the world. This possible outcome – a rare opportunity
in the frequency domain – would be beneficial to customers worldwide and support
national policy objectives to achieve (1) the direct economic benefits of economies of
scale (i.e. maximum affordability and coverage of broadband services) as well as (2)
ease of roaming and (3) interoperability of services on a global basis [18]. To date, the
2.6 GHz band is unique in that the band includes a substantial amount of spectrum (190
MHz) that has been allocated on a primary basis in all three ITU regions for terrestrial
mobile communications [18].
The ITU has defined (Recommendation ITU-R M.1036-3) three band plan options
for 2.6 GHz, as summarized below and in Figure 1-4 [18]:
• ITU Option 1 includes a mix of paired and unpaired spectrum in a standardized
configuration and has been formulated to avoid interference problems between
resulting FDD and TDD modes of operation.
6
Table 1-1: Operation bands for LTE-Advanced [16].
Operating Uplink (UL) operating band BS Downlink (DL) operating band UL-DL Band Duplex
Band receive UE transmit BS transmit UE receive Separation Mode
1 1920 MHz – 1980 MHz 2110 MHz – 2170 MHz 130 MHz FDD
5 824 MHz – 849 MHz 869 MHz – 894MHz 20 MHz FDD
6 830MHz – 840 MHz 875MHz – 885 MHz 35 MHz FDD
9 1749.9 MHz – 1784.9 MHz 1844.9 MHz – 1879.9 MHz 60 MHz FDD
15 Reserved Reserved – –
16 Reserved Reserved – –
18 815 MHz – 830 MHz 860 MHz – 875 MHz 30MHz FDD
19 830 MHz – 845 MHz 875 MHz – 890 MHz 30MHz FDD
20 832 MHz – 862 MHz 791 MHz – 821 MHz N/A FDD
33 1900 MHz – 1920 MHz 1900 MHz – 1920 MHz N/A TDD
41 3400 MHz – 3600MHz 3400 MHz – 3600MHz N/A TDD
7
• ITU Option 2 does not include unpaired spectrum and leaves the second member
of each pair undetermined. That is, the upper, center, and lower bands are paired
spectrum (downlink), but the location of the other pair member (uplink) is not
specified.
• ITU Option 3 allows freedom of choice about the respective amounts of

spectrum in the band that are attributed to paired (FDD operation) and unpaired
(TDD operation) blocks.
Figure 1-4: ITU Options for the 2.6GHz Band [24].

2.6 GHz band is already in use for LTE services in Scandinavia and is being
allocated in other European countries as well as in Americans countries [19] beside it is
considered roaming band in Japan and China [20].
LTE-Advanced transceiver configured for multi-mode and multi-band as an

example, Figure 1-5: It is a multi-mode since it supports GSM/GPRS, W-CDMA, LTE
and LTE Advanced systems. It is a multi-band since it uses these system bands [21,22].
8
Figure 1-5: transceiver configuration example of multi-mode/multi-band [22].
9
1.4 Overview of Diplexers and their Application
Diplexers are indispensable components in RF/microwave wireless

communication systems that were widely studied in the early 1960s by Matthaei et al.
[23] and Wenzel [24]. They are very important part of modern communication system,
and usually select and join channel. So they are generally used in microwave
communication, satellite communication and remote sensing system. With the rapid
development of microwave technology, the microwave communication frequency
resources are becoming precious and crowded. The high performance microwave filters
and diplexers have played more important roles in the microwave and wireless
communication system.
A diplexer is a three-port network which usually consists of two filters used to

isolate transmit and receive signal in a common frequency range that uses a common
antenna [25]. Diplexers are two channel versions of multiplexers.
They are multi channel filters that divide a frequency band into two or more
channels. Diplexers may consist of highpass and lowpass, bandpass and bandpass,
bandpass and bandstop, and other combinations [26]. They are usually designed using
two BPFs to separate each frequency pass-band connected to a common junction.
These two band-pass filters (BPFTx and BPFRx) as shown in Figure 1-6, are combined
to shape a diplexer.
Figure 1-6: Conventional diplexer; Hardware structure (left) and frequency response (right) [27].
The band-pass filter in the transmitter path (BPFTx) stops the transmitter noise
artificially increasing the receiver noise figure while the bandpass filter in the receiver
path (BPFRx) stops the transmitter signal overloading the receiver [28]. Diplexers are
used in a cellular base station; they split the available frequency band into partial bands,
10
to enable simultaneous sending and receiving. One partial band is assigned as uplink
from mobile to base station and the other partial band is assigned as downlink from base
station to mobile [1].
1.3 Thesis Motivation
The goal of this thesis is to design microstrip diplexer using 10 coupled

resonators. The diplexer will be designed for operation over the LTE-Advanced band,
specifically UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz for base transceiver station
antennas. This thesis studies the feasibility of replacing conventional diplexers with
coupled-resonator diplexers which enables miniaturization in comparison to the
traditional diplexers, and reduces the complexity in design.
Conventional diplexers consist of two channel filters connected to a junction for

energy distribution which increases the total size of diplexer. In order to decrease the
size of diplexer another design technique may be used. It is based on coupled resonator
structures that eliminate the need for external junctions [29]. These structures are
miniaturized since there are no external junctions and reduce complexity in design. The
design approach builds from three-port network theory for many resonators coupled
together with multiple outputs and the coupling matrix will be synthesized using
optimization by MATLAB. Microstrip “U” shape resonators will be used and
Stimulation software (CST Microwave studio) will be utilized for designing the
diplexer. The experimental results for the design prototype are shown and discussed in
the next chapters.
1.4 Thesis Overview
The objective of this research is to analyze and design microstrip diplexer. The
design is based on synthesis of coupling matrix of a 3-port coupled resonator circuit
using optimization. Diplexer meets the specification of LTE-Advanced system,
specifically band number 7 (i.e. UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz), and it is
to be used inside base transceiver station. Microstrp diplexer consists of 10 “U” shape
resonators that provide very small insertion loss and good isolation between the two
ports. The size of the proposed diplexer may be miniaturized in comparison to
conventional diplexers since a novel method presented in [29,30] will be used in the
synthesis.
11
Chapter 1 introduces cellular system evolution from 1980-2012 and beyond,
presents overview of LTE-Advanced system and its bands the licensing of the 2.6 GHz
spectrum band and assess the viability and implications of the three ITU band plan
Options. Finally the chapter provides a general overview about diplexers and their
applications.
Chapter 2 explains the conventional diplexers synthesis and the different types
of external junctions for distribution of energy. It discusses the theory of coupling, the
calculation of coupling coefficients between two resonators, and the calculation of
external coupling coefficients.
Chapter 3 conducts the synthesis of diplexer that specifies LTE-Advanced band

7 and shows how to synthesis coupling coefficients and external quality factors using
optimization cost function.
Chapter 4 explains the design procedure for the diplexer and the relationship
between the coupling coefficients and the physical structure of coupled resonators in
order to find the physical dimensions of the diplexer. Then, it shows the whole structure
of the diplexer and its response which resulting from CST simulation software.
The final chapter provides summary and conclusions drawn from this work.
12
References
[1] T. Rappaport, Wireless Communication Principle and Practice. 2ed edition,
Prentice Hall of India Private Limited, 2004.
[2] N. Boucher, Cellular Radio Handbook – A Reference for Cellular System

Operation, Quantum Publishing Inc., California, 1992.
[3] CP02 Introduction to Digital Cellular, ISSUE 5 REVISION 5, Motorola
[4] E. Seurre, P. Savelli and J. Pietri, GPRS for Mobile Internet, Artech House,
2003.
[5] E. Seurre, P. Savelli and J. Pietri, EDGE for Mobile Internet, Artech House,
2003.
[6] F. Muratore, UMTS Mobile Communication for the Future, John Wiley &
Sons,Inc, 2001.
[7] E. Dahlman, S. Parkvall, J. Sköld and P. Beming, 3G Evolution HSPA and

LTE for Mobile Broadband, Elsevier Ltd, 2007
[8] R. Sheriff and Y. Fun Hu, Mobile Satellite Communications Network, John
Wiely & Sons Ltd, 2001
[9] C. Cox, An Introduction to LTE: LTE, LTE-Advanced, SAE and 4G Mobile
Communications, John Wiely & Sons Ltd, 2012
[10] S. Yang, 3G CDMA2000 Wireless System Engineering, Artech House, Inc.
2004
[11] E.k Dahlman, S. Parkvall, J. Sköld and P. Beming, 3G Evolution HSPA and
LTE for Mobile Broadband, Elsevier Ltd, 2007
[12] P. Tapia, J. Liu, Y. Karimli and M. Feuerstein, HSPA Performance and
Evolution, A Practical Perspective, John Wiely & Sons Ltd, 2009
[13] CDG Organization [Online]. Available:
http://www.cdg.org/resources/files/fact_sheets/CDG_EVDO_Rev.B_FEB_2012.pdf
[14] S. Sesia, I. Toufik and M. Baker, LTE – The UMTS Long Term Evolution
From Theory to Practice, 2ed edition, John Wiely & Sons Ltd, 2011
[15] E.k Dahlman, S. Parkvall and J. Sköld, 4G LTE/LTE Advanced for Mobile
Broadband, Elsevier Ltd, 2011
[16] Moray Rumney, LTE and the Evolution to 4G Wireless Design and
Measurement Challenges,Reprinted for Agilent Technologies Publication by
John Wiely & Sons Ltd, 2009
13
[17] A. Taha, H. Hassanein and N. Abu Ali, LTE, LTE-Advanced and WiMAX
Towards IMT-Advanced Network, John Wiely & Sons Ltd, 2012
[18] The 2.6 GHz Spectrum Band, Scott Fox, M. Roetter, J. Walkenhorst, and P.
Rysavy, December 2009, [Online]. Available:
http://www.rysavy.com/Articles/2009_12_GSMA_2_6_GHz_Report.pdf
[19] Discussion Paper, Fixed or flexible? A survey of 2.6GHz spectrum awards,

R. Marsden, E. Sexton and A. Siong, DotEcon, June 2010, [Online].
Available:
http://dotecon.com/publications/dp1001.pdf
[20] Next Generation Mobile Networks Initial Terminal Device Definition,

NGMN, November 2010, [Online]. Available:
http://www.ngmn.org/uploads/media/NGMN_Initial_Terminal_Device_Definition.pdf
[21] N. Takaharu, "LTE and LTE-advanced: Radio technology aspects for mobile
communications,” General Assembly and Scientific Symposium, Istanbul,
13-20 Aug. 2011, pp. 1-4.
[22] Mobile Device Products, TriQuint Semiconductor "Connecting the Digital

World to the Global Network" , 2011, [Online]. Available:
http://www.triquint.com/prodserv/brochures/handset_brochure.pdf
[23] G. Matthaei, and G. Cristal, “Multiplexer channel-separating units using

interdigital and parallel-coupled filters,” IEEE Trans. Microw. Theory Tech.,
1965, pp. 328–334
[24] J. Wenzel, “Printed-circuit complementary filters for narrow bandwidth

multiplexers,” IEEE Trans. Microw. Theory Tech., 1968, 16, pp. 147–157.
[25] J. Reed, Software Radio: A Modern Approach to Radio Engineering. 1st

edition, Prentice Hall PTR, 2002
[26] C. Erich, and E .Eisenmann, Filter Design Tables and Graphs. John Wiley &
Sons,Inc, 1966.
[27] J. Eberspächer, H. Vögel, C. Bettstetter and C. Hartmann, GSM –
Architecture, Protocols and Services. 3rd edition, John Wiley & Sons,Inc,
2009.
[28] Jack Daniel Company, Duplexers – An Introductory Tutorial,
http://www.rfsolutions.com/duplex.htm
[29] T. Skaik and M. Lancaster, “Coupled Resonator Diplexer without External

Junctions,” Journal of Electromagnetic Analysis and Applications, 2011, pp.
238-241.
14
[30] T. Skaik, M. Lancaster and F. Huang, “Synthesis of Multiple Output Coupled
Resonator Microwave Circuits Using Coupling Matrix Optimization,” IET
Journal of Microwaves, Antennas and Propagation, Vol. 5, No. 9, 2011, pp.
1081-1088.
15
Chapter 2
Overview of Diplexers
2.1 Introduction
Full duplex systems allow simultaneous radio transmission and reception

between a subscriber and a base station, by providing two simultaneous but separate
channels (frequency division duplex or FDD) or adjacent time slots on a single radio
channel (time division duplex or TDD). These two popular ways of achieving full
duplex transmission and reception are explained in this chapter.
2.2 TDD Systems
The principle of operation of a TDD system is shown in Figure 2-1. Here the
radio communication system alternately transmits and receives a signal having the same
frequency. There is no simultaneous transmission in both directions at a given instant of
time. However, due to the fact that the data transmission rate is very much higher than
the user’s data rate, it is possible to provide the appearance of a full duplex operation to
the end user [1]. TDD is only possible with digital transmission formats and digital
modulation, and is very sensitive to timing. It is for this reason that TDD has only
recently been used, and only for indoor or small area wireless applications where the
physical coverage distances (and thus the radio propagation time delay) are much
smaller than the many kilometers used in conventional cellular telephone systems [2].
Compared with FDD, TDD is advantageous in that there is no problem in allocating a
frequency to each of the transmitting and receiving signals, and that the transmitting and
receiving efficiencies (propagation losses) can be made substantially equal [1].
TDD systems use a switch inside the subscriber unit to switch between
transmitter and receiver time slots, thus eliminating the need for a duplexer. This
reduces the cost associated with diplexer component. Further this system reuses the
filters, mixers, frequency sources and synthesizers, thereby eliminating cost and
complexity.
16
Figure 2-1: The principle of operation in TDD systems [2].
2.3 FDD Systems
The frequency duplex procedure has been used already in analog mobile radio
systems and is also used in digital systems. For communication between a mobile and a
base station, the available frequency band is split into two partial bands, to enable
simultaneous sending and receiving as illustrated in Figure 2-2. One partial band is
assigned for uplink (from mobile to base station) transmissions and the other partial
band is assigned for downlink (from base station to mobile) transmissions.
• Uplink band: transmission band of the mobile and receiving band of the base station.
• Downlink band: receiving band of the mobile and transmission band of the base
station.
To achieve good separation of both directions, the partial bands must be a sufficient
frequency distance apart, i.e. the frequency pairs of a connection assigned to uplink and
17
downlink must have this distance band between them. Usually, the same antenna is used
for sending and receiving [3].
Figure 2-2: The principle of operation in FDD systems [2].
2.4 Conventional Diplexer
The diplexer is a device that isolates the receiver from the transmitter while
permitting them to share a common antenna. It is often the key component that allows
two way radios to operate in a full duplex manner. An ideal diplexer provides perfect
isolation with no insertion loss, to and from the antenna. Conventional diplexers consist
of two channel filters connected to an energy distribution network. The channel filters
pass frequencies within a specified range, and reject frequencies outside the specified
boundaries, and the distribution network divides the signal going into the filters, or
combines the signals coming from the filters [4] (see Figure 2-3). FDD systems use two
singly terminated filters known as duplexing filters. The two duplexing filters are
18
connected at the terminating port to form a three terminal network. Diplexers are used in
cellular base stations to allow simultaneous transmission and reception using a single
antenna. Generally, the transmitter generates signals with relatively high power, and
hence the TX filter should have high power handling capability, and the receiver needs
to detect very weak signals. The RX filter is required to have high attenuation in the
transmit band in order to protect the low-noise amplifier in the receiver from the
transmitter high power signals. Similarly, the TX filter is required to have high level of
stopband attenuation in order to reject the out-of-band noise generated by the power
amplifier. Thus, the isolation between the receive and transmit channels is a crucial
parameter in the diplexer design [5].
Figure 2-3: Conventional duplexer [3].
The most commonly used distribution configurations are E- or H-plane n-

furcated power dividers [6,7], circulators [8], manifold structures [9-12], Y-junction
[13] and T-Junction [14].
Figure 2-4 shows the configuration of diplexer with a 1:2 divider diplexer
network. The power divider configurations can be designed for diplexer with wideband
channels or large channel separation [7]. Figure 2-5 show a circulator configuration,
where each channel consists of a bandpass filter and a channel-dropping circulator [4].
In [15], microwave diplexer uses a passive or active three-port clockwise circulator
device.
19
Figure 2-4: Configuration of diplexer with a 1:2 divider multiplexing network [7].
Figure 2-5: Configuration of circulator-coupled diplexer [4,15].
In manifold configurations, channel filters are connected by transmission lines:

microstrip, coaxial, waveguide, etc. and T-junctions, which illustrated in Figure 2-6 [5].
Manifold configurations provide low insertion loss and high power handling capability.
However, they have complex design.
20
Figure 2-6: Configuration of manifold-coupled diplexer [5].
The use of a circulator in place of a manifold structure in the diplexer means that
the filters can be tuned or even exchanged without the need to modify the manifold.
2.4.1 Literature Review
Many structures of diplexers have been proposed in literature. In [16], a novel

microstrip diplexer with a joint T-shaped resonator is presented. The diplexer using a
joint T-shaped resonator does not require combining circuits and matching networks. It
is designed to be used in the UMTS-WCDMA system and it has high isolation and wide
stopband in addition to satisfying the passband requirement. The diplexer structure is
shown in Figure 2-7.
Figure 2-7: Layout of the diplexer using the T-shaped resonator, R1, to combine two second-order
bandpass filters. Port 1 uses coupled feeding, ports 2 and 3 use tapped feeding [16].
21
In [13] a novel microstrip diplexer using step impedance resonator (SIR) with
folded hairpin structure is presented. The diplexer consists of two third-order
Chebyshev bandpass filters operating at 0.9GHz and 1.8GHz within GSM frequencies
band. A diplexer is constructed by connecting two filters to a Y-junction as illustrated in
Figure 2-8. There is no spurious responses reside within the two passband frequencies.
Figure 2-8: Fabricated microstrip diplexer [13].
Since conventional microstrip hairpin filter and diplexer are inherently formed
by coupled-line resonators, spurious response and poor isolation performance are
unavoidable.
In [17] a simple technique that is suitable for an inhomogeneous structure such
as microstrip to cure such poor performances. The technique is based on the stepped
impedance coupled-line as shown in Figure 2-9.
Figure 2-9: Fabricated microstrip hairpin diplexer [17].
22
In [18] is present novel microstrip diplexer design for UMTS and GSM ranges.
This device was implemented by using microstrip resonator with serial coupling. The
modified gap structure allows adjusting the operating frequency of the filters and
reduces the dimensions of the device as illustrated in Figure 2-10.
Figure 2-10: The Microstrip Diplexer [18].
In [19] a microstrip diplexer is proposed which has good performance on the

UMTS upload and download bands, The diplexer is based on a two-pole resonator
named H-type resonator as shown in Figure 2-11, and makes strong use of cross-
coupling to pass energy between ports.
Figure 2-11: The microstrip diplexer [19].
In [20] a compact diplexer based on folding microstrip lines is presented. The

folding microstrip lines have the characteristics of inductance and capacitance. They
have filter function and compact size as shown in Figure 2-12. The proposed diplexer
employs filters with folding microstrip lines for band selection and separation. A
23
prototype diplexer is designed at 900MHz and 1800MHz, which is proposed for
GSM/DCS dual band cellular system applications.
Figure 2-12: Circuit layout of the diplexer [20].
In [21], a compact diplexer using a square open loop with stepped impedance
microstrip resonators is proposed. The compact miniaturized two poles square open
loop resonators are used to design filters and a diplexer for IMT-2000 bands application
as shown in Figure 2-13.
Figure 2-13: a compact diplexer using a square open loop [21].
24
2.5 Coupled Resonator Diplexer
The synthesis procedure of the proposed diplexer in this thesis is based on

eliminates the need for an additional common junction this approach to diplexer design
is able to achieve reductions in the size and volume of the circuit. The coupled-resonator
based diplexers, without additional common junction is presented in [22-24]. The
method for synthesizing coupled resonator diplexers based on optimization of coupling
matrix of multiple coupled resonators representing a three-port network, and it is
performed in the normalized frequency domain. In [25], a method for synthesizing
coupled resonator diplexers composed of TX and RX filters (two types of junctions
connecting the TX and RX filters are considered). For the first junction type, the
common port is directly coupled to the first resonator of the TX and RX filters,
respectively. For the second junction type, common node is realized. Responses
comparisons for coupled resonator based diplexers and the relationships between the
responses and topologies of this type of diplexer structures are investigated in [26]. In
[27,28], an approach to the synthesis of diplexers that takes into account a three-port
junction in the initial synthesis of the two channel filters was presented. This approach
provides a very good starting point for the optimization of the whole structure, so the
convergence can be achieved with very little iteration. In [29,30], the synthesis of
multiport coupled resonator networks based on a procedure analogous to filters has been
reported.
2.5.1 Diplexers with star-junction
The star-junction, whose purpose in [31] is to connect the input port of the
diplexer (hereafter referred to as “port 1”) to the input ports of the two channel filters,
while the output filter ports represent the other ports of the diplexer. In the most simple
case, the star-junction is realized by the parallel (or series) connection of the channel
filters’ input ports. In [32], novel topologies of star-junction diplexer/multiplexers with
resonating junctions are proposed (Figure 2-14). The advantage of the proposed
topologies is that the number of connections to the resonating junction is reduced and
thus allowing diplexer/multiplexers with more channels to be implemented. An
optimization technique is used to synthesize the coupling matrix of the proposed
diplexer/multiplexers.
25
Figure 2-14: General schematic for star-junction multiplexers [32].
A general approach to the synthesis of microwave diplexer/multiplexers (RF

combiners) presenting a star junction topology (with a resonating junction) in [33] and
A planar microstrip diplexer/triplexer with compact size is proposed in [34]. The
diplexer/triplexer in [34] is composed of three pairs of bandpass filters and a star-
junction.
2.5.2 Diplexers with a Common Resonator Junction
Diplexers with a common resonator junction have a common port is coupled to

TX and RX filter by a common resonator (an extra resonator besides those of the TX
and RX filters) as illustrated in Figure 2-15 [35] a coupling matrix (see section 2.5.6)
synthesis procedure for diplexers with a common resonator employing TX and RX
filters with arbitrary topology is presented. It is based on the evaluation of the
characteristic polynomials of the TX and RX filters composing the diplexer with a
common resonator junction. An objective error function can be constructed using the
transmission and reflection zeros obtained from the polynomials of the TX and RX
filters. The coupling matrix of the TX and RX filters (MTX and MRX) are synthesized
independently by optimizing the objective error function. The coupling matrix of the
diplexers can be obtained from MTX, MRX and the equivalent capacitance of the
common resonator junction in the normalized frequency domain. In [36], miniaturised
microstrip diplexers designed using stepped impedance resonators by integrating two
bandpass filters with common resonator sections have been proposed.
26
Figure 2-15: Diplexer with an extra resonator as common junction [35].
2.5.3 General theory of couplings
In general, the coupling coefficient k of coupled microwave resonators which

can be different in structure and can have different self-resonant frequencies as referred
to Figure 2-16 may be defined on the basis of a ratio of coupled energy to stored energy,
i.e [37].
k=
∫ ∫ ∫ εE1 ⋅ E2 dv
+
∫ ∫ ∫ µH 1 ⋅ H 2 dv
(2.1)
∫ ∫ ∫ ε E1 dv × ∫ ∫ ∫ ε E2 dv ∫ ∫ ∫ µ H1 dv × ∫ ∫ ∫ µ H 2 dv
2 2 2 2
Where all fields are determined at resonance, and the volume integrals are over entire
effective regions with permittivity of ε and permeability of µ . The first term on the
right hand side represents the electric coupling while the second term the magnetic
coupling. It should be remarked that the interaction of the coupled resonators is
mathematically described by the dot operation of their space vector fields, which allows
the coupling to have either positive or negative sign. A positive sign would imply that
the coupling enhances the stored energy of uncoupled resonators, whereas a negative
sign would indicate a reduction. Therefore, the electric and magnetic couplings could
either have the same effect if they have the same sign, or have the opposite effect if their
signs are opposite. Obviously, the direct evaluation of coupling coefficient from
equation (2.1) requires the knowledge of the field distributions and needs to perform the
space integrals. This would never be an easy task unless analytical solutions of the
fields exist [37]. On the other hand, it would be much easier to find some characteristic
frequencies that are associated with the couplings. The coupling coefficient can then be
27
determined if the relationships between the coupling coefficient and the characteristic
frequencies are established. In what follows we derive the formulation of such
relationships.
Figure 2-16: General coupled microwave resonators. Resonators 1 and 2 can be different in
structure and have different resonant frequencies [37].
Before processing further, it might be worth pointing out that although the
following derivations are based on lumped element circuit models, the outcomes are
also valid for distributed element coupled structures on a narrow-band basis [37]. Figure
2-17 shows the different types of coupling which could be electric coupling, magnetic
coupling or mixed electric and magnetic coupling.
Figure 2-17: Inter-coupling between coupled resonators. (a) Coupled resonator circuit with electric
coupling. (b) Coupled resonator circuit with magnetic coupling. (c) Coupled resonator circuit with
mixed electric and magnetic coupling [37].
28
2.5.4 Three port networks
Linear three-port networks are characterized by a number of equivalent circuit

parameters, such as impedance matrix, admittance matrix, and scattering matrix. Figure
2-18 shows a typical three-port network [38].
Figure 2-18: three-port network.
The impedance related to the three voltages V1,V2,V3 to the currents I1,I2,I3. The
admittance matrix is simply the inverse of the impedance matrix, Y = Z-1 [38].
V1   Z11 Z 12 Z13   I 1   Z 11 Z12 Z13 
V  =  Z Z 22 Z 23   I 2  , Z =  Z 21
   Z 22 Z 23  (Impedance matrix)
 2   21
V3   Z 31 Z 32 Z 33   I 3   Z 31 Z 32 Z 33 
(2.2)
Three-port network diplexer with one input and two outputs has a scattering
matrix with the following nine independent elements:
 b1   S11 S12 S13   a1   S11 S12 S13 
b  =  S S 22 S 23  a2  , S =  S 21
   S 22 S 23  (scattering matrix) (2.3)
 2   21
b2   S 31 S 32 S 33  a2   S 31 S 32 S 33 
The matrix elements S11, S12, S13, S21, S22, S23, S31, S32 and S33 are referred to as
the scattering parameters or the S-parameters. The parameters S11, S22 and S33 have the
meaning of reflection coefficients, by definition S11 is the reflection coefficient at port 1
with port 2 and port 3 terminated in a matched load. S21, S12, S31, S13 have the meaning
of transmission coefficients simply S21 mean the transmission coefficient from port 1
and port 3 to port 2 with port 2 terminated in a perfect match, it gives a measure of the
29
amount of signal that is transmitted to port 2 form other ports. The parameters S23, S32
referee to isolation between port 2 and port 3 [38].
The waves going towards( incident voltage wave ) the n-port are a = (a1, a2, a3),
the waves travelling away ( reflected voltage wave ) from the n-port are b = (b1, b2, b3).
By definition arrowed going into the n-port are counted positively and arrowed flowing
out of the n-port negatively [38].
If all lines entering the network have the same characteristic impedance, then
bi
Sij = (2.4)
aj
ak =0 k≠ j
For reciprocal networks, the S-matrix is symmetric.
S ij = S ji i≠ j (2.5)
2.5.5 N-Port Coupled Resonator Circuits
Coupled resonator circuits are the basis for the design of microwave filters. The
general coupling matrix of N-port circuit with n-coupled resonators, and a detailed
derivation of the general coupling matrix and its relation to the scattering parameters are
presented in the next sections. The derived coupling matrix is fundamental to the current
work, and it will be used as a basis for the synthesis of three-port coupled resonator
diplexer. In a coupled resonator circuit, energy may be coupled between adjacent
resonators by a magnetic field or an electric field or both as shown in Figure 2-19 [37].
The coupling matrix can be derived from the equivalent circuit by formulation of
impedance matrix for magnetically coupled resonators or admittance matrix for
electrically coupled resonators. This approach has been used to derive the coupling
matrix of an N-port n-coupled resonators circuit. Magnetic coupling and Electric
coupling will be considered separately and later a solution will be generalized for both
types of couplings.
30
(a)
(b)
Figure 2-19: (a) Equivalent circuit of magnetically n-coupled resonators in N-port network, (b)
Equivalent circuit of electrically n-coupled resonators in N-port network [37].
2.5.5.1 Deriving Coupling Matrix of N-port Networks
In a coupled resonator circuit, energy may be coupled between adjacent

resonators by a magnetic field or an electric field or both. The coupling matrix can be
derived from the equivalent circuit by formulation of impedance matrix for magnetically
coupled resonators or admittance matrix for electrically coupled resonators. Magnetic
coupling and Electric coupling will be considered separately and a solution will be
generalized for both types of couplings in section 2.5.3 In the case of magnetically
coupled resonators, using Kirchoff‘s voltage law, the loop equations are derived from
the equivalent circuit, and represented in impedance matrix form. Similarly, for
electrically coupled resonators, using Kirchoff‘s current law, node equations are derived
from the equivalent circuit, and represented in admittance matrix form [37].
2.5.5.1.1 Circuits with magnetically coupled resonators

Suppose only magnetic coupling between adjacent resonators, the equivalent
circuit of magnetically coupled n-resonators with multiple ports is shown in Figure 2-19
(a), where i represents loop current, L, C denote the inductance and capacitance, and R
denotes the resistance (represents a port). It is assumed that all the resonators are
31
connected to ports, and the signal source is connected to resonator 1. It is also assumed
that the coupling exists between all the resonators [37].
Using Kirchoff‘s voltage law, the loop equations are derived as follows,
 1 
 R1 + jwL1 + i1 − jwL12i2 L − jwL1nin = es
 jwC1 
 1 
− jwL12i1 +  R2 + jwL2 + i2 L − jwL2 nin = 0
 jwC 2  (2.6)
M
 1 
− jwLn1i1 − jwLn 2in + ... + jwL( n−1) 2i( n−1) +  Rn + jwLn + in = 0
 jwC n 
where Lab = Lba denotes the mutual inductance between resonators a and b. The matrix
form representation of these equations is as follows,
 1 
 R1 + jwL1 + jwC − jwL12 L − jwL1n 
 1
  i1  es 
 − jwL21 R2 + jwL2 +
1
L − jwL2 n  i2   0 
 jwC2   =   (2.6)
 M  M   M 
    
 1  in   0 
− jwLn1 − jwLn 2 L Rn + jwLn +
 jwC n 
or equivalently [ Z ].[i ] = [e] , where [Z ] is the impedance matrix. Assuming all
resonators are synchronized at the same resonant frequency ω0 = 1 / LC , where
L = L1 = L2 = L Ln and C = C1 = C2 = LCn , the impedance matrix [Z ] can be expressed
by [Z ] = ω0 L.FBW .[Z ] ,where FBW= is the fractional bandwidth, and [Z ] is the

normalized impedance matrix, given by [37],
 R1 jwL12 1 jwL1n 1 
 ω L( FBW ) + P − ω L FBW L − ω L FBW 
 0 0 0

− jwL 21 1 R 2 jwL 2n 1 
+P L −
[Z ] =  ω0 L FBW ω0 L( FBW ) ω0 L FBW  (2.7)
 M 
 jwL jwLn 2 1 Rn 
− 1
n1
− L + P
 ω0 L FBW ω0 L FBW ω0 L ( FBW ) 
j  ω ω0 
With P =  −  is the complex lowpass frequency variable.
FBW  ω0 ω 
32
Defining the external quality factor for resonator i as Qei = ω0 L / Ri , and the coupling
coefficient as M ij = L ij /L , and assuming ω/ω0 ≈ 1 for narrow band approximation, [Z ]
is simplified to,
1 
q + P − jm12 L − jm1n 
 e1 
 − jm 1
+ P L − jm2 n 
[Z ] =  21
qe 2  (2.8)
 M 
 1 
 − jmn1 − jmn 2 L + P
 qen 
where qei is the scaled external quality factor ( qei = Qei .FBW )and mij is the normalized
coupling coefficient ( mij = M ij .FBW ). The network representation for the circuit in
Figure 2-20, considering only three-ports, is shown in Figure 2-19 (a), a1 , b1 , a2 , b2 and
a3 ,b3 where are the wave variables, V1 , I1 , V2 , I 2 and V3 , I 3 are the voltage and current
variables and i is the loop current. It is assumed that port 1 is connected to resonator 1,
port 2 is connected to resonator x , and port 3 is connected to resonator y [22].
Figure 2-20: Network representation of 3-port circuit [22].

Three ports have been considered at this point since three-port devices such as
diplexers is the main focus in this thesis as shown in Figure 2-20.
The relationships between the voltage and current variables and the wave variables are
defined as follows [22],
33
1
VN = R (aN + bN ) and I N = (aN − bN ) (2.9)
R
Solving the equations (2.9) for aN and, bN the wave parameters are defined as follows,
1 V  1 V 
aN =  N + R I N  and bN =  N − R I N  (2.10)
2 R  2 R 
where N is the port number, and R corresponds to R1 for port 1, Rx for port 2, and R y
for port 3. It is noticed in the circuit in Figure 2.2 that I1 = i1 , I 2 = −ix , I 3 = −i y , and
V1 = es − i1 R1 . Accordingly, the wave variables may be rewritten as follows,
es es − 2i1 R1
a1 = b1 =
2 R1 2 R1
a2 = 0 b2 = ix Rx (2.11)
a2 = 0 b3 = i y R y
The S-parameters are found from the wave variables as follows,

b1 2R1i1
S11 = a2 =a3 =0 = 1−
a1 es
b2 2 R1 Rx i x
S 21 = a2 = a3 =0 = (2.12)
a1 es
b3 2 R1 R y i y
S 31 = a2 = a3 =0 =
a1 es
Solving (2.6) for i1 , i x and i y
es −1
i1 = [Z ]11
ω0 L.FBW
es −1
ix = [Z ]x1 (2.13)
ω0 L.FBW
es −1
iy = [ Z ] y1
ω0 L.FBW
and by substitution of equations (2.13) into equations (2.12), we have,
34
2 R1 −1
S11 = 1 − [Z ]11
ω0 L.FBW
2 R1 Rx −1
S 21 = [ Z ] x1 (2.14)
ω0 L.FBW
2 R1 R y −1
S 31 = [ Z ] y1
ω0 L.FBW
In terms of external quality factors qei = ω0 L.FBW / Ri , the S-parameters become,
2 −1
S11 = 1 − [Z ]11
qe1
2 −1
S21 = [Z ]x1 (2.15)
qe1qex
2 −1
S31 = [ Z ] y1
qe1qey
where qe1 , qex and qey are the normalised external quality factors at resonators 1, x , and
y , respectively. In case of asynchronously tuned coupled-resonator circuit, resonators
may have different resonant frequencies, and extra entries mii are added to the diagonal
entries in [Z ] to account for asynchronous tuning as follows,
1 
 q + P − jm11 − jm12 L − jm1n 
 e1 
 1 
− jm21 + P − jm21 L − jm2 n
[z] =  qe 2  (2.16)
 M 
 1 
 − jmn1 − jmn 2 L + P − jmnn 
 qen 
2.5.5.1.2 Circuits with electrically coupled resonators
The coupling coefficients introduced in the above section are all based on
mutual inductance, while this section presents the derivation of coupling coefficients for
electrically coupled resonators in an N-port circuit, where the electric coupling is
represented by capacitors. The normalized admittance matrix [Y ] will be derived here
in an analogous way to the derivation of the [Z ] matrix in the previous section. Figure
2-19 (b) show the equivalent circuit of electrically coupled n-resonators in an N-port
35
network, is where represents the source current, vi denotes the node voltage, and G
represents port conductance. It is assumed here that all resonators are connected to
ports, and the current source is connected to resonator 1. Also, it is assumed that all
resonators are coupled to each other. The solution of this network is found by using
Kirchhoff‘s current law, which states that the algebraic sum of the currents leaving a
node is zero. Using this law, the node voltage equations are formulated as follows [37],
 1 
 G1 + jwC1 + v1 − jwC12v2 L − jwC1n vn = is
 jwL1 
 1 
− jwC12i1 +  G2 + jwC2 + v2 L − jwC2 n vn = 0
 jwL2  (2.17)
M
 1 
− jwCn1v1 − jwCn 2 vn + ... + jwC( n−1) 2v( n−1) +  Gn + jwCn + vn = 0
 jwLn 
where Cab = Cba denotes the mutual capacitance between resonators a and b. The matrix
form representation of these equations is as follows,
 1 
G1 + jwC1 + jwL − jwC12 L − jwC1n 
 1
  v1  is 
 − jwC21 G2 + jwC2 +
1
L − jwC2 n   v2   0 
 jwL2   =   (2.18)
 M  M   M 
    
 1  vn   0 
− jwC n1 − jwC n 2 L Gn + jwC n +
 jwLn 
or equivalently [Y ].[v ] = [i ] , where [Y ] is the admittance matrix.
Assuming all resonators are synchronized at the same resonant frequency ω0 = 1 / LC ,
Where L = L1 = L2 = L Ln and C = C1 = C2 = LCn , the admittance matrix [Y ] can be
expressed by [Y ] = ω0C.FBW .[Y ] , where FBW is the fractional bandwidth, and [Y ] is

the normalized admittance matrix, given by [37],
36
 G1 jwC12 1 jwC1n 1 
 ω C ( FBW ) + P − ω C FBW L − ω C FBW 
 0 0 0

 − jwl 21 1 G 2 jwC 2n 1 
[Y ] =  ω0 L FBW ω0 L( FBW ) + P L − ω0C FBW  (2.19)
 M 
 jwC jwCn 2 1 Gn 
− 1
n1
− L + P
 ω0C FBW ω0C FBW ω0C ( FBW ) 
where P is the complex lowpass frequency variable.

By defining the coupling coefficient as M ij = Cij /C , and the external quality factor for
resonator i as Qei = ω0C / Gi , and assuming ω/ω0 ≈ 1 for narrow band approximation, the
normalized admittance matrix [Y ] may be simplified to,
1 
q + P − jm12 L − jm1n 
 e1 
 − jm 1
+ P L − jm2 n 
[Y ] =  21
qe 2  (2.20)
 M 
 1 
 − jmn1 − jmn 2 L + P
 qen 
where qei = Qei .FBW is the scaled external quality factor, and mij = M ij .FBW is the
normalized coupling coefficient.

A 3-port network with n-coupled resonators is considered here, with port 1
connected to resonator 1, port 2 connected to resonator x, and port 3 connected to
resonator y. The network representation is shown in Figure 2-21, where all wave and
voltage and current variables at the network ports are the same as those in Figure 2-20
[22].
Figure 2-21: Network representation of 3-port circuit [22].
37
By comparing the variables at the ports in the circuit in Figure 2-19 (b) and the network
representation in Figure 2-21, it is identified that V1 = v1 , V2 = −vx , V3 = −v y , and
I1 = is − v1G1 , where vx and v y are node voltages at resonators x and y , respectively.
Accordingly, the wave parameters may be expressed as follows [22],

is is − 2i1G1
a1 = b1 =
2 G1 2 G1
a2 = 0 b2 = v x Gx (2.21)
a2 = 0 b3 = v y G y
The S-parameters are found from the wave variables as follows,

b1 2G1v1
S11 = a2 =a3 =0 =1−
a1 is
b2 2 G1G x v x
S 21 = a2 = a3 =0 = (2.22)
a1 is
b3 2 G1G y v y
S31 = a2 = a3 =0 =
a1 is
The node voltage variables v1 , vx and v y are found from (2.18) as follows,
is −1
v1 = [Y ]11
ω0C.FBW
is −1
vx = [Y ]x1 (2.23)
ω0C.FBW
is −1
vy = [Y ] y1
ω0C.FBW
and by substitution of equations (2.23) into equations (2.22), we have,
2G1 −1
S11 = 1 − [Y ]11
ω0C.FBW
2 G1G x −1
S 21 = [Y ] x1 (2.24)
ω 0 C.FBW
2 G1G y −1
S 31 = [Y ] y1
ω0C.FBW
38
The S-parameters can now be expressed in terms of the normalized external quality
factors, qei = ω0C.FBW / Gi as follows,
2 −1
S11 = 1 − [Y ]11
qe1
2 −1
S 21 = [Y ]x1 (2.24)
qe1qex
2 −1
S31 = [Y ] y1
qe1qey
To account for asynchronous tuning, the normalized admittance matrix will have extra
terms mii in the principal diagonal as follows,
1 
 q + P − jm11 − jm12 L − jm1n 
 e1 
 1 
− jm21 + P − jm21 L − jm2 n
[Y ] =  qe 2  (2.25)
 M 
 1 
 − jmn1 − jmn 2 L + P − jmnn 
 qen 
2.5.6 General coupling matrix
From previous sections formulations, the most notable is that the formulation of
normalized impedance matrix [Z ] is identical to that of normalized admittance matrix
[Y ] . Accordingly, a unified solution may be formulated regardless of whether the

couplings are magnetic or electric or even the combination of both.
So the S parameters of a three-port coupled resonator circuit may be generalized as,
2 −1
S11 = 1 − [ A]11
q e1
2 −1
S 21 = [ A] x1 (2.26)
qe1 qex
2 −1
S 31 = [ A] y1
q e1 q ey
Where it is assumed that port 1 is connected to resonator 1, ports 2 and 3 are connected
to resonators x and y respectively [22]. A general normalized coupling matrix [A] in
39
terms of coupling coefficients and external quality factors has been derived as shown in
equation
[A] = [q]+ p[U]− j[m] (2.27)
 1 
q L 0 L 0 
 e1  1 L 0 0  m11 L m1( n −1) m1n 
 M O M M M   M
1 M O M M  O M M 
[ A] =  0 L L 0 + p − j (2.28)
 q ex  0 L 1 0 m( n −1)1 L m( n −1)( n −1) m( n −1) n 
 M M M O M     
 L 0 1  mn1 L mn ( n −1) mnn 
1  0
 0 L 0 L 
 q ey 
where qୣ୧ is the scaled external quality factor (qୣ୧ = Qୣ୧ . FBW) of resonator I, FBW is
ఠమష ఠభ
the fractional bandwidth given by ‫= ܹܤܨ‬ ఠ೚
, [U ] is the [n × n] identity matrix, n is
the number of the resonators, p is the complex lowpass frequency variable, [m ] is the
coupling matrix and entry mij is the normalized coupling coefficient between resonators
i and j, (m୧୨ = M୧୨/୊୆୛ ), and the diagonal entries m୧୧ account for asynchronous tuning,
so that resonators can have different self-resonant frequencies [23].
After determining the required coupling matrix for the desired filtering
characteristic, a next important step for the filter design is to establish the relationship
between the value of every required coupling coefficient and the physical structure of
coupled resonators in order to find the physical dimensions of the filter.
2.6 Chebyshev Response
The Chebyshev response is a mathematical strategy for achieving a fast roll-off

by allowing ripple in the frequency response, either passband ripple (type I) or stopband
ripple (type II) [39].
Type I Chebyshev filters are the most common Chebyshev filters. The gain (or
amplitude) response as a function of angular frequency ω of the nth-order low-pass
filter is [39]:
1
G n (ω ) = H n ( jω ) = (2.28)
ω 
1 + ε 2Tn2  
 ω0 
40
ω
where ε is the ripple factor, ω 0 is the cutoff frequency and Tn ( ) is first kind
ω0
Chebyshev polynomial of the nth order, which defined by [38]:
 −1 ω ω
 cos( n cos ) ≤1
ω  ω0 ω0
Tn ( ) =  (2.29)
ω0 cosh( n cosh −1 ω ) ω
≥1
 ω0 ω0
The passband exhibits equiripple behavior, with the ripple determined by the
ripple factor ε .
L Ar
ε = 10 10
−1 (2.30)
where L Ar is the passband ripple in dB. In the passband, the Chebyshev polynomial
alternates between 0 and 1 so the filter gain will alternate between maxima at G = 1 and
minima at G = 1 / 1 + ε 2 . This behavior is shown in Figure 2-22.
Figure 2-22: The frequency response of a sixh-order type I Chebyshev low-pass filter [40].
Type II Chebyshev filters is known as inverse Chebyshev, this type is less

common because it does not roll off as fast as type I, and requires more components. It
has no ripple in the passband, but does have equiripple in the stopband [39]. The gain is
41
1 (2.31)
G n (ω ) =
1
1+
ω 
ε 2Tn2  0 
ω 
In the stopband, the Chebyshev polynomial will oscillate between 0 and 1 so that
the gain will oscillate between zero and G = 1 / 1 + 1 / ε 2 . The parameter ε is decibels
by following equation:
1
ε= L Ar
(2.32)
10 10
−1
Figure 2-23 shows Type II Chebyshev filters behavior
Figure 2-23: The Inverse Chebyshev Response [40].
2.7 Summary
The derivation of the coupling matrix of multiple coupled resonators with

multiple outputs has been presented. Magnetic coupling and Electric coupling have been
considered separately and a solution has been generalized for both types of couplings.
The relationships between the scattering parameters and the coupling matrix of a 3-port
coupled resonator circuit have been formulated.
42
The derived equations in this chapter will be used as a basis in the next chapters.
The design procedure will base on optimization of the coupling matrix of a 3-port
circuit with multiple coupled resonators.
A cost function will be used to be minimized in the optimization algorithm, a
gradient based local optimization iteration method will be employed to synthesize the
coupled resonator diplexer and examples of diplexers with different topologies will be
illustrated.
43
References
[1] Y. Urabe, H. Takai and H. Yamasaki, “Time Division Duplex Transceiver,” U.
S. Patent 5 446 770, 29 August 1995.
[2] T. Rappaport, Wireless Communication Principle and Practice. 2ed edition,

Prentice Hall of India Private Limited, 2004.
[3] J. Eberspacher and H. Vȍgel, GSM Switching, Services and Protocols. 2ed
edition, John Wiley & Sons,Inc, 2001.
[4] I. Carpintero, M. Padilla-Cruz, A. Garcia-Lamperez, and M. Salazar-
Palma,”Generalized multiplexing network,” U.S. Patent 0114082 A1, Jun. 1,
2006.
[5] Jack Daniel Company, Duplexers – An Introductory Tutorial,

http://www.rfsolutions.com/duplex.htm
[6] J. Ruiz-Cruz, J. Montejo-Garai, J. M. Rebollar, and S. Sobrino, "Compact full

ku-band triplexer with improved E-plane power divider," Progress In
Electromagnetics Research, Vol. 86, 2008, pp. 39-51.
[7] J. Dittloff, J. Bornemann, and F. Arndt, "Computer aided design of optimum E-
or Hplane N-furcated waveguide power dividers," in Proc. European Microwave
Conference, Sept. 1987, pp. 181-186.
[8] R. Mansour, et al., “Design considerations of superconductive input
multiplexers for satellite applications,” IEEE Transactions on Microwave Theory
and Techniques, Vol. 44, no. 7, pt. 2, 1996, pp. 1213-1229.
[9] J. Rhodes and R. Levy, “Design of general manifold multiplexers,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 27, no. 2, 1979, pp.
111-123.
[10] D. Rosowsky and D. Wolk, “A 450-W Output Multiplexer for Direct
Broadcasting Satellites,” IEEE Transactions on Microwave Theory and
Techniques, Vol. 30, no. 9, Sept. 1982, pp. 1317-1323.
[11] M. Uhm, J. Lee, J. Park, and J. Kim,“An Efficient Optimization Design of a
Manifold Multiplexer Using an Accurate Equivalent Circuit Model of Coupling
Irises of Channel Filters,” in proc. IEEE MTT-S Int. Microwave Symp., Long
Beach, CA, 2005, pp. 1263-1266.
[12] R. Cameron and M. Yu, “Design of manifold coupled multiplexers,” IEEE
Microwave Mag., Vol. 8, no. 5, Oct. 2007, pp. 46–59.
[13] T. King, A. Ying Ying, S. Tiong, “A Microstrip Diplexer Using Folded
Hairpins,” Proceeding of the IEEE International RF and Microwave Conference
(RFM),Seremban, Malaysia, 12th-14th Dec. 2011, pp. 226-229.
44
[14] H. Zhang and G. James, “A Broadband T-junction Di-plexer with Integrated Iris
Filters,” Microwave and Opti-cal Technology letters, Vol. 17, No. 1, 1998, pp.
69-72.
[15] C. Saavedra, “Diplexer Using a Circulator and Inter-changeable Filters,”
Proceeding of the 7th International Caribbean Conference on Devices, Circuits
and Systems, Mexico, 28-30 April 2008, pp. 1-5.
[16] M. Chuang, , and M. Wu, “Microstrip Diplexer Design Using Common T-
Shaped Resonator,” Microwave and Wireless Components Letters, IEEE, Vol.
21 , Issue: 11, Nov. 2011, pp. 583 – 585
[17] S. Srisathit, S. Patisang, R. Phromloungsri, S. Bunnjaweht, S. Kosulvit, and M.
Chongcheawchamnan, “High Isolation and Compact Size Microstrip Hairpin
Diplexer,” Proceeding of the IEEE Micrwave and Wireless Components Letters,
Vol. 15, No. 2, Feb. 2005, pp. 101-103.
[18] S. Bezerra, and M. Melo, “Microstrip Diplexer for GSM and UMTS Integration
Using Ended Stub Resonators,” Proceeding of the IEEE MTT-S International
Microwave and Optoelectronics Conference, Oct.-Nov.2007, pp. 954-958.
[19] H. Cabral, S. Bezerra and M. de Melo, “A Diplexer for UMTS Applications,”

Proceeding of the IEEE MTT-S International Microwave and Optoelectronics
Conference (IMOC), Nov. 2009, pp. 215 - 217.
[20] Q. Zhang, X. Liu, H. Guo, Y. Wang, “A Novel Compact Diplexer Based on

Folding Microstrip Lines,” Proceeding of the Microwave, Antenna, Propagation,
and EMC Technologies for Wireless Communications (MAPE), Nov. 2011, pp.
785 - 788.
[21] J. Konpang, “A Compact Diplexer Using Square Open Loop with Stepped
Impedance Resonators,” Microwave Conference, 2008. APMC 2008. Asia-
Pacific, 16-20 Dec. 2008, pp. 1 -4.
[22] T. Skaik, “A Synthesis of Coupled Resonator Circuits with Multiple Outputs
using Coupling Matrix Optimisation”, PhD Thesis, March 2011, School of
Electronic, Electrical and Computer Engineering, the University of Birmingham.
[23] T. Skaik, M. Lancaster, and F. Huang, “Synthesis of Multiple Output Coupled
Resonator Microwave Circuits Using Coupling Matrix Optimization,” IET
Journal of Microwaves, Antenna & Propagation, Vol.5, no.9, June 2011, pp.
1081- 1088.
[24] T. Skaik, Michael Lancaster, “Coupled Resonator Diplexer without External
Junctions” Journal of Electromagnetic Analysis and Applications, Vol. 3, 2011,
238-241.
45
[25] R. Wang, J. Xu, M. Wang, and Y. Dong, “Synthesis of microwave resonator
diplexers using linear frequency transformation and optimization,” Progress In
Electromagnetics Research, Vol. 124, Feb. 2012, pp 441-455.
[26] W. Xia, X. Shang, M. Lancaster, “Responses Comparisons for Coupled-
Resonator Based Diplexers,” Passive RF and Microwave Components, 2012, pp.
67- 75.
[27] G. Macchiarella and S. Tamiazzo, “Novel Approach to the Synthesis of
Microwave Diplexers,” IEEE Transactions on Microwave Theory and
Techniques, Vol. 54, No. 12, 2006, pp. 4281-4290.
[28] G. Macchiarella and S. Tamiazzo, “Synthesis of Diplexers Based on the

Evaluation of Suitable Characteristic Polynomials,” IEEE MTT-S International
Microwave Symposium, San Francisco, 11-16 June 2006, pp. 111-114.
[29] A. Garcia-Lamperez, M. Salazar-Palma, T. Sarkar, “Analytical synthesis of

microwave multiport networks,” IEEE MTT-S Int. Microwave Symp. Digest,
USA, Vol. 2, June 2004, pp. 455–458.
[30] A. Garcia-Lamperez, M. Salazar-Palma, T. Sarkar, “Compact multiplexer

formed by coupled resonators with distributed coupling,” IEEE Antennas and
Propagation Society Int. Symp., USA, Vol. 1A, 2005, pp. 89–92.
[31] G. Macchiarella, “ynthesis of Star-Junction Multiplexers,” Microwave

Magazine, IEEE, Vol.12 , Issue: 6, 2011, pp. 101 – 109.
[32] T. Skaik, “Novel Star-Junction Coupled –Resonator Multiplexer Structures,”

Progress in Electromagnetics Research Letters, Vol. 31, April 2012, pp. 113-
120.
[33] G. Macchiarella, S. Tamiazzo, “Synthesis of Star-Junction Multiplexers,”

Microwave Theory and Techniques, IEEE Transactions on Vol. 58, Issue: 12,
2010, pp 3732 – 3741.
[34] C.W. Tang and M.G. Chen, “Packaged microstrip triplexer with star-junction
topology,” Electronics Letters Vol. 48 , Issue: 12, 2012, pp. 699 – 701.
[35] R. Wang, and J. Xu, “Synthesis and Design of Microwave Diplexers with a
Common Resonator Junction,” Microwave and Millimeter Wave Technology
(ICMMT), Vol. 2, May 2012, pp.1-4
[36] C. Chi-Feng, H. Ting-Yi, C. Chi-Ping, W. Ruey-Beei, “Microstrip diplexers

design with common resonator sections for compact size, but high isolation,”,
IEEE Trans. Microw. Theory Tech., Vol. 54, Issue: 5, 2006, pp. 1945–1952.
[37] J. Hong, Microstrip filters for RF/Microwave applications. New York, NY: John
Wiley, 2011.
46
[38] D. Pozar, Microwave Engineering, 4th edition, John Wiley & Sons, Inc., 2012.
[39] A. Williams and F.Taylors, Electronic Filter Design Handbook, 4th edition,
McGraw Hill’s, 2006.
[40] R. Schaumann, M. Valkenburg, Design of Analog Filters, 2nd edition, Oxford
University Press, 2000.
47
Chapter 3
Synthesis of Coupled Resonator Diplexer
3.1 Introduction
There are many possible topologies for n -coupled resonators that can achieve a
Chebyshev response. One example is illustrated in Figure 3-1; it is a schematic of a
diplexer with n resonators. Each circle represents a resonator, and the lines between
resonators are internal couplings. The arrowed lines between resonators and ports
represent external couplings. Resonators 1 and 2 we call the stem [1]; two branches are
coupled to the second resonator of the stem and every one consists of r resonator where
r = (n - 2)/2 .
Figure 3-1: n -resonator based diplexer.

The stem of the coupled-resonator based diplexer works as a dual-band bandpass
filter. It attenuates signals outside of two passbands but not have effects on splitting
signals on these two bands. The two branches of the diplexer work as Chebyshev
bandpass filters with different centre frequencies. Each branch considers one of the two
passbands of the responses diplexer and eliminates signals outside of the passband of
the branch. Signals passing through the stem will be oriented to one of these two
branches and reflected by the other. As a result, signals are split to different ports [1-3].
In the next sections, diplexers with different topologies will be proposed. The
design procedure is based on optimization of coupling matrix for multiple coupled
48
resonators in a three port network. These diplexers will not have any energy distribution
networks in their structures, unlike the conventional diplexers. This enables
miniaturizing the diplexer structure. A cost function will be used to be minimized in the
optimization algorithm, and diplexers with different number of resonators will be
presented.
3.2 Optimization Techniques

This section is an introduction to optimization technique to solve problems
arising in diplexer design. In the most general terms, optimization technique is body of
mathematical results and numerical methods for finding and identifying the best
nominee from a collection of alternatives without having to evaluate all possible
alternatives [4]. The process of optimization lies at the root of engineering, since the
classical function of the engineer is to design new, better, more efficient and less
expensive systems as well as to devise plans and procedures for the improved operation
of existing systems [4].
The power of optimization methods to determine the best case without actually
testing all possible cases comes through performing iterative numerical calculations
using clearly defined logical procedures or algorithms implemented on computing
design.
To apply optimization theory to specified problems, it is necessary to define the

system boundaries, and then selects a criterion on the basis of which the performance or
design of the system can be evaluated so that the best design or set of operating
conditions can be identified. After that make selection of the independent variables that
are suitable to characterize the possible choice designs. Finally you can build the system
model.
There are many optimization techniques that can be used to synthesis coupled
resonator diplexer. These techniques can be based either on optimization of the
dimensions of the physical structure of the diplexer using EM simulations (CST
Microwave studio) [5], or on optimization of the coupling coefficients in a coupling
matrix. The advantage of use coupling matrix optimization approach that it is required
less computational time and iterations than full scale EM simulations (CST Microwave
studio) to complete the synthesis process.
49
The main key element in formulating a problem for optimization is the selection
of the control variables that are adequate to characterize the possible candidate designs
or operating conditions of the system. In case of optimization of the coupling matrix of
a coupled resonator circuit, coupling coefficients and external quality factors represent
the control variables. The control variables may be either unconstrained, so that the
search space is unbounded, or constrained by lower and upper limits to prevent the
optimization algorithm from giving an unfeasible solution.
In microwave coupled resonator optimization problems, the cost function of

many variables will have multiple local optimal and the global minimum can be found
only by locating all local optimal and selecting the best one.
During optimization process multiple local minimum can be possible to find, one of
them is the global minimum. However, finding the global minimum is more challenging
and need complex technique to search them.
Local optimization algorithms strongly depend on the initial values of the

control parameters. The initial guess used as an input to the algorithm that will seek a
local minimum within the local neighborhood of the initial value. However, this local
minimum is not ensured to be the global minimum. In comparison to local methods,
global optimization methods are much slower and may take hours or even days to find
the optimal solution for problems with tens of variables. A gradient based local
optimization method has been used in the current work to produce coupling matrices of
coupled resonator diplexer. The method has been successful and efficient for all of the
designs illustrated in this chapter, and the convergence of the algorithm was very fast.
3.2.1 A gradient based local optimization
Gradient method is an algorithm to solve problems of the form minn f(x) with the
x∈ℜ
search directions defined by the gradient of the function at the current point. The
gradient of a function is an n-component vector given by [6]:
 ∂f / ∂x1 
∂f / ∂x 
 2
∇f =
M
 (3.1)
n x1  
∂f / ∂xn 
50
The gradient has a very important property. If we move along the gradient
direction from any point in n-dimensional space, the function value increases at the
fastest rate. The gradient direction is a local property and not a global one. The gradient
direction is called the direction of steepest ascent [6].
Figure 3-2 shows where the gradient vectors ∇f evaluated at points 1, 2, 3, and
4 lie along the directions 11', 22', 33', and 44', respectively. The function value increases
at the fastest rate in the direction 11' at point 1, but not at point 2. Similarly, the function
value increases at the fastest rate in direction 22' (33') at point 2(3), but not at point 3
(4). In other words, the direction of steepest ascent generally varies from point to point,
and if we make infinitely small moves along the direction of steepest ascent, the path
will be a curved line like the curve 1-2-3-4. Since the gradient vector represents the
direction of steepest ascent, the negative of the gradient vector denotes the direction of
steepest descent. Thus any method that makes use of the gradient vector can be expected
to give the minimum point faster than one that does not make use of the gradient vector
[6].
Figure 3-2: Steepest ascent directions [6].
A gradient based local optimization is a general iteration algorithm including

Cauchy’s Method, Newton’s Method, Modified Newton’s Method, Marquardt’s
Method, Conjugate Gradient Methods, Quasi-Newton Methods and Quasi-Newton
Methods. Such techniques are usually quite efficient in converging to a local minimum
51
in the objective function. However, there is no guarantee that this will be the global
minimum. In general, the objective function is likely to have many minima. Only those
minima that give a sufficiently good match to the data are of interest [4,6]. A gradient
based optimization algorithm is presented in Appendix B [6].
3.3 Cost function
The cost function used in current work [7,8] has been selected since it depends
on minimum set of variables that specified the desired response, which makes the
algorithm requires less iterations and converges faster.
A cost function that is used in the optimization of the coupling matrix of coupled
resonator diplexer is formulated here. For a coupled resonator diplexer, the reflection
and transmission functions may be defined in terms of polynomials as follows [7,8],
F (s) P1 ( s) / ε P2 ( s ) / ε
S11 ( s ) = , S 21 ( s ) = , S 31 ( s ) = (3.2)
E (s) E ( s) E (s)
where the roots of F (s) correspond to the reflection zeros, the roots of P1 (s) , and
P2 (s) correspond to the transmission zeros of the diplexer frequency response at ports
2, and 3 respectively, ε is a ripple constant, and E (s) roots correspond to the pole
positions of the filtering function. The initial cost function is written in terms of the
polynomials F , P and E and it is evaluated at the frequency locations of transmission
and reflection zeros as follows [6]:
2
T1 2 T2 2 2 LR
R R−2 F ( s pv )
Ω = ∑ P1 ( sti ) + ∑ P2 ( stk ) + ∑ F ( s rj ) + ∑ − 10 20
(3.3)
i =1 k =1 j =1 v =1 E ( s pv )
where S ti , S tk are the frequency locations of transmission zeros of S 21 , S 31
respectively, T1 , T2 are the numbers of the transmission zeros of S 21 , S 31 respectively,
R is the total number of resonators in the diplexer, LR is the specified return loss in dB
( LR < 0 ), S rj and S pv are the frequency locations of the reflection zeros and the peaks
frequency values of S11 in the passband. The last term in the cost function is used to set
52
the peaks of S11 = F / E to the required return loss level. It is assumed here that both
channels of the diplexer have the same return loss level.
From equation (2.3), recall that for a 3-port network of multiple coupled
resonators where port 1 is connected to resonator 1, port 2 and port 3 are connected to
resonators a and b respectively:
2 −1
S11 = 1 − [ A]11
q e1
2 −1
S 21 = [ A]a1 (3.4)
qe1 qea
2 −1
S 31 = [ A]b1
qe1qeb
The inverse of the matrix [A] can be described in terms of the adjugate and determinant
by employing Cramer‘s rule for the inverse of a matrix,
−1 adj ([ A])
[ A] = , ∆A ≠ 0 (3.5)
∆A
where adj([A]) is the adjugate of the matrix [A], and ∆ A is its determinant. Noting that
the adjugate is the transpose of the matrix cofactors, the (x,1) element of the inverse of
matrix [A] is:
−1 cof 1 x ([ A])
[ A] = (3.6)
∆A
where cof1x ([ A]) is the (1,x) element of the cofactor matrix of [A]. By substitution of
equation (3.6) into equation (3.4), the following equations are obtained,
2 cof 11 ([ A])
S11 = 1 −
qe1 ∆A
2 cof1a ([ A])
S 21 = (3.7)
qe1qea ∆A
53
2 cof1b ([ A])
S31 =
qe1qeb ∆A
where it is assumed that the common port is connected to resonator 1, port 2 is

connected to resonator a, and port 3 is connected to resonator b.
by equating the S-parameters in equation (3.7) to their equivalent in equation (3.2), the
polynomials F(s), P(s) and E(s) are expressed in terms of the general matrix [A] as
follows,
2.cof11 ([ A( s )])
F (s) = ∆ A (s) − ,
q e1
P1 ( s ) 2.cof1a ([ A( s )])
= ,
ε q e1 qea (3.8)
P2 ( s ) 2.cof1b ([ A( s )])
= ,
ε q e1 qeb
E (s) = ∆ A (s)
By substitution of the polynomials in equation (3.8) into equation (3.3), the cost
function is now expressed in terms of determinants and cofactors of the matrix [A] and
the external quality factors as follows [7],
2
T1
2.cof1a ([ A( sti )])
Ω=∑
i =1 q e1q ea
2
T2
2.cof1b ([ A( stk )])
+∑
k =1 qe1q eb
(3.9)
2
R 2.cof11 ([ A( s rj )])
+ ∑ ∆ A ( s rj ) − )
j =1 qe1
2
LR
R−2 2.cof11 ([ A( s pv )])
+ ∑ 1− − 10 20
v =1 q e1 .∆ A ( s pv )
where qe1 , q ea , and q eb are the external quality factors at ports 1, 2 and 3 respectively,
∆ A (s = x) is the determinant of the matrix [A] evaluated at the frequency variable x,
and cof mn ([ A( s = y)]) is the cofactor of matrix [A] evaluated by removing the mth row
54
and the nth column of [A] and finding the determinant of the resulting matrix at the
frequency variable s=y.
The first two terms in the cost function are used if the diplexer characteristics contain
transmission zeros. However, for a Chebyshev response, these terms may be used to
minimize the transmission of each channel at the passband of the other channel, thus
increasing the isolation between channel ports. So, the frequency locations sti are
chosen to be the band edges of the channel at port 3, and similarly the frequency
locations stk are chosen to be the band edges of the channel at port 2.
The optimization of the coupling matrix [m] is based on minimization of cost

function that is evaluated at frequency locations of the reflection and transmission zeros.
The third term in the cost function is used to minimize the cost function at the frequency
locations of the reflection zeros.
The cost function will be used in a gradient based optimization algorithm to

synthesize coupling matrices of coupled resonator diplexer. The initial values of control
variables, coupling coefficients, were set to 0.5 for direct coupling and the values of the
external quality factors are numerically calculated, as will be shown in the next sections.
The lowpass frequency positions of the reflection zeros of the diplexer are
initially set to be equally spaced in the optimization algorithm which will modified their
positions until became equiripple level at specified insertion loss is achieved. Or, the
initial values of the frequency locations of reflection zeros may be scaled from standard
lowpass prototype Chebyshev filters, but any way these locations still need to be
modified during optimization process until equiripple is achieved. The frequency
locations of the peaks of |S11| are found using numerical differentiation at each iteration
in the optimization algorithm. The variables that will be optimized in the optimization
algorithm are the coupling coefficients and the frequency locations of the reflection
zeros.
3.4 Optimization algorithm
A gradient based optimization technique has been utilized to synthesize the

coupling coefficients, and the cost function in equation (3.9) has been used. To avoid
convergence to a local minimum, the optimization has been preceded in two stages. In
55
the first stage, the cost function in equation (3.9) has been used without the last term,
and with equally spaced reflection zeros and unconstrained gradient-based local
optimization method. In the second stage, the entire cost function in equation (3.9) has
been used, and the output coupling coefficients from the first stage are used as initial
values, and the frequency locations of the return zeros are allowed to move until peaks
in the passbands are equal to –20 dB. The flowchart of the algorithm is given in Figure
3-3.
Figure 3-3: Flowchart of optimization algorithm.
56
3.5 Calculation of external quality factor
The responses of two lowpass prototype filters have the same order and filtering
function but with different bandwidth are shown in Figure 3-4. The first is with
frequency edges of ±1 Hz and a bandwidth of BW±1 , and the second is with frequency
edges of x and 1 Hz and a bandwidth of BW x1 [7].
Figure 3-4: Lowpass prototype filters (a) passband edges of ±1 Hz, (b) passband edges of x and 1
Hz.
The normalized external quality factors of these filters are related by [7],
BW ±1 2
q ex1 = q e ±1 = q e ±1 (3.10)
BW x1 1− x
where qex1 is the normalized external quality factor of the filter with edges of x and +1,
and qe±1 is the normalized external quality factor of the filter with edges of ±1, that can
be calculated from the g-values as shown in Appendix A .
Because of a symmetrical diplexer with channel edges of (-1,-x) and (x,1), the
normalized external quality factors at ports 2 and 3 are calculated from equation (3.10),
and the normalized external quality factor at the common port is equal to qex1 / 2 .
Figure 3.5 (a) illustrates the spectrum allocation for band 7 in LTE-Advanced
cellular telephone uses which specifically UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz
for base transceiver station antennas.
The normalized bandwidth in Figure 3-5(b) can be calculated from the following
scaling equation:
57
2 190
= (3.11)
2 x 50
From equation (3.11), x=0.263157.
Figure 3-5: (a) frequency spectrum allocation for band 7 in LTE-Advanced cellular telephone, (b)
normalized bandwidth for band 7 in LTE-Advanced cellular telephone.
3.6 Initial spacing of reflection zeros
The initial guess of the locations of reflection zeros within a diplexer channel
with edges of ( x, f c ) Hz, the leftmost reflection zero is located at ( x + 0.02)i Hz, and
the rightmost reflection zero is located at ( f c − 0.02)i Hz. The other reflection zeros are
equally spaced between ( x + 0.02)i and ( f c − 0.02)i with frequency spacing as follows
[7],
( f c − x − 0.04)
y −1
Where y is the total number of reflection zeros within a diplexer channel.
58
3.7 Diplexers with T-Topology
Figure 3-6 shows the general T-Topology for a symmetrical diplexer where n is
the total number of resonators, ±x define the inner edges of the two channels, and r is
the number of resonators located between either output port, and the junction resonator
[7]. Resonators in each arm should have different self-resonant frequencies to separate
the diplexer channels from each other.
In this T-topology, the resonators are directly coupled with Chebyshev response.
The resonators in the vertical branch have different self-resonant frequencies; this is to
achieve two frequency bands at the ports 2 and 3. Consequently, for the high frequency
channel to be at port 2, the resonators above the junction resonator should have positive
frequency offsets M ii > 0 , and for the low frequency channel to be at port 3, the
resonators below the junction resonator should have negative frequency offsets M ii < 0 .
A gradient based constrained optimization technique has been utilized to

synthesize the coupling coefficients, and the following conditions for coupling
coefficients have been applied to simplify the optimization problem [7],
mn − 2 r ,n − 2 r +1 = mn − 2 r ,n − r +1 K mn − r −1, n − r = m n −1, n
m n − r +1, n − r +1 = − m n − 2 r +1,n − 2 r +1 K m n −1, n −1 = − m n − r −1,n − r −1 ,K m nn = − m n − r ,n − r
Figure 3-6: Diplexer T-Topology [7].
59
3.7.1 Diplexer with n=8, x=0.263157, and r=3
An LTE-Advanced band 7 diplexer, specified at UL: 2.50–2.57 GHz and DL:

2.62–2.69 GHz has been synthesized with Chebyshev response. The diplexer topology
is illustrated in Figure 3-7, which has a total number of resonators of n=8, a return loss
of 20 dB for both channels, and inner edges for both channels of ±0.263157Hz. The
value of r is taken as 3 in this example, which means 3 resonators should be located
between the junction resonator and either of the output ports.
The coupling coefficients between any adjacent resonators are initially set to 0.5
in the initial coupling matrix, and the same for self coupling coefficients m33 , m44 , m55 ,
− m66 , − m77 and − m88 . The coefficients m11 , m22 and the couplings that do not exist
between resonators are set to zero. Since the desired response of the diplexer has
symmetrical frequency bands, the following conditions are set for the coupling
coefficients m23 = m26 , m34 = m67 , m45 = m57 , m33 = −m66 , m44 = −m77 and m55 = −m88 .
Figure 3-7: Topology of Diplexer with 8 resonators.
60
A gradient base optimization technique has been utilized to synthesize the
coupling coefficients, and the cost function in equation (3.9) has been used. The
optimization has been carried out in two stages. In the first stage, the cost function in
equation (3.9) has been used without the last term, and with equally spaced reflection
zeros and unconstrained gradient-based local optimization method. In the second stage,
the whole cost function in equation (3.9) has been used, and the output coupling
coefficients from the first stage are used as initial values, and the frequency locations of
the return zeros are allowed to move until peaks in the passbands are equal to –20
dB. Moreover, the third term in the cost function was assigned more weight than the
other terms. Although a Chebyshev response is required, the first two terms in the cost
function have been used to increase the isolation between diplexer channels by
minimizing the transmission of each channel at the passband of the other channel. In the
second stage, some constraints have been applied on the control variables to prevent the
optimization algorithm from giving an unfeasible solution for the coupling coefficients,
and also to prevent the locations of reflection zeros from moving outside the passband.
The variables in the optimization algorithm are:
{m12 , m23 , m26 , m33 , m34 , m44 , m45 , m55 , m55 , m67 , m78 , sr1 sr 2 , s r 3 , sr 4 }
The last four variables represent the reflection zeros of the upper band, and the
reflection zeros of the lower band are equal to − s rj . The lower and upper bounds of the
variables are as follows,
0,0,0,0,0,0,0,0,0,0,0.2633,0.2633,0.2633,0.2633 lower bounds


1,1,1,1,1,1,1,1,1,1,0.9999,0.9999,0.9999,0.9999 upper bounds
The cost function is evaluated at the return zeros locations s rj , s11 peaks
frequency locations s pv . The initial values of return zeros locations s rj have been set
with equal spacing as explained in section (3.5), with lower channel edges of -1 and -
0.263157Hz, and upper channel edges of 0.263157 and 1 Hz, as follows:
±0.2832i, ±0.5154i, ±0.7477i, ±0.9800i
61
These values are then allowed to be moved in the optimization algorithm until
the peaks of s11 in the passband are equal to the specified return loss. The final
locations of return zeros are:
±0.2633i, ±0.4846i, ±0.7974i, ±0.9999i.
The locations of transmission coefficients at passband edges sti = [-0.263157i -i],
and stk = [0.263157i, i]. The normalized external quality factors are numerically
calculated from equation (3.9) and found to be qe1 = 1.2640 and q e 5 = q e 8 = 2.5281.
The optimized normalized coupling matrix that result from the first stage of
optimization process is shown in Table 3-1 and the prototype response of the diplexer is
Table 3-1: Coupling matrix of diplexer with 8 resonators from the first stage of optimization
process.
1 2 3 4 5 6 7 8
1 0 0.7981 0 0 0 0 0 0
2 0.7981 0 0.3866 0 0 0.3866 0 0
3 0 0.3866 0.5455 0.2683 0 0 0 0
4 0 0 0.2683 0.6065 0.3345 0 0 0
5 0 0 0 0.3345 0.6074 0 0 0
6 0 0.3866 0 0 0 -0.5455 0.2683 0
7 0 0 0 0 0 0.2683 -0.6065 0.3345
8 0 0 0 0 0 0 0.3345 -0.6074
62
Figure 3-8:
8: Diplexer prototype response with 8 resonators from the first stage of optimization
process.
The final optimized normalized coupling coefficients from the second stage of
optimization are shown in Table 3-2,
3 2, and the prototype response of the diplexer is
Table 3-2:
2: Coupling matrix of diplexer with 8 resonators from the second stage of optimization
process.
1 2 3 4 5 6 7 8
1 0 0.8101 0 0 0 0 0 0
2 0.8101 0 0.4009 0 0 0.4009 0 0
3 0 0.4009 0.5496 0.2793 0 0 0 0
4 0 0 0.2793 0.599 0.3594 0 0 0
5 0 0 0 0.3594 0.5992 0 0 0
6 0 0.4009 0 0 0 -0.5496 0.2793 0
7 0 0 0 0 0 0.2793 -0.599 0.3594
8 0 0 0 0 0 0 0.3594 -0.5992
63
Figure 3-9:
9: Diplexer prototype response with 8 resonators from the second stage of optimization
process.
An LTE-Advanced
Advanced band 7 diplexer has been syntheised with symmetrical bands
and a Chebyshev response, with r=4, a total number of resonators of n=10, inner edges
for both channels of ±0.263157Hz, and a return loss of 20 dB for both channels. The
diplexer topology is shown in Figure 33-10. The values of normalized external quality
factors are calculated using equation (3.9) as q e 6 = q e10 =2.636 and q e1 =1.318. Since
the diplexer has symmetrical bands the coupling coefficients are set as follows:
m23 = m27 , m34 = m78 , m45 = m89 , m 56 = m 9 ,10 , m33 = −m77 , m44 = −m88 , m55 = −m99
and m 55 = − m10 ,10 .
64
The initial values of the coupling coefficients between adjacent resonators are
set to 0.5, and the same for self coupling coefficients m33 , m44 , m55 , m66 . The cost
function in equation (3.9) has been used and the optimization has been done in two
stages. Frequency locations s ti and s tk in the first two terms in the cost function have
been taken as sti = [-0.263157i -i], and stk = [0.263157i, i]. The values of return zeros
locations s rj have been initially equally spaced between lower channel edges of -1 and -
0.263157 Hz, and upper channel edges of 0.263157 and 1 Hz, as follows, (from section
3.5),
±0.2832i ± 0.4574i ± 0.6316i ±0.8058i ±0.9800i
In the first stage of optimization process, the cost function in equation (3.9) has
been used without the last term, and with equally spaced reflection zeros and
unconstrained gradient-based local optimization method. The optimized normalized
65
coupling matrix resulting from this stage is shown in Table 33-3,
3, and the response of the
diplexer is shown in Figure 3--11.
Table 3-3:
3: Coupling matrix of diplexer with 10 resonators from the first stage of optimization
process.
1 2 3 4 5 6 7 8 9 10
1 0 0.7866 0 0 0 0 0 0 0 0
2 0.7866 0.3623 0 0 0 0.3623 0 0 0
3 0 0.3623 0.5539 0.2359 0 0 0 0 0 0
4 0 0 0.2359 0.6172 0.2304 0 0 0 0 0
5 0 0 0 0.2304 0.621 0.3062 0 0 0 0
6 0 0 0 0 0.3062 0.6176 0 0 0 0
7 0 0.3623 0 0 0 0 -0.5539 0.2359 0 0
8 0 0 0 0 0 0 0.2359 -0.6172 0.2304 0
9 0 0 0 0 0 0 0 0.2304 -0.621 0.3062
10 0 0 0 0 0 0 0 0 0.3062 -0.6176
Figure 3-11:
11: Diplexer prototype response with 10 resonators from the first stage of optimization
process.
66
In the second stage, the cost function in equation (3.9) has been used with the
last term and the coupling coefficients, locations of reflection zeros have been
constrained to prevent the optimization algorithm from giving an unfeasible solution
and prevent the locations of reflection zeros from moving outside the passband.
The lower and upper bounds of the variables are as follows,
0,0,0,0,0,0,0,0,0,0,0,0,0,0.2633,0.2633,0.2633,0.2633,0.2633 lower bounds

1,1,1,1,1,1,1,1,1,1,1,1,1,0.9999,0.9999,0.9999,0.9999,0.9999 upper bounds
Where the last 5 terms represent lower and upper bands of the reflection zeros of the
upper band. The realised locations of return zeros are
±0.2700i ±0.4500i ±0.6300i ±0.8100i ±0.9900i
The optimized normalized coupling matrix is shown in Table 3-4, and the response of
the diplexer is shown in Figure 3-12.
Table 3-4: Coupling matrix of diplexer with 10 resonators from the second stage of optimization
process.
1 2 3 4 5 6 7 8 9 10
1 0 0.7943 0 0 0 0 0 0 0 0
2 0.7943 0.3795 0 0 0 0.3795 0 0 0
3 0 0.3795 0.5435 0.2582 0 0 0 0 0 0
4 0 0 0.2582 0.6066 0.2429 0 0 0 0 0
5 0 0 0 0.2429 0.6267 0.3312 0 0 0 0
6 0 0 0 0 0.3312 0.6137 0 0 0 0
7 0 0.3795 0 0 0 0 -0.5435 0.2582 0 0
8 0 0 0 0 0 0 0.2582 -0.6066 0.2429 0
9 0 0 0 0 0 0 0 0.2429 -0.6267 0.3312
10 0 0 0 0 0 0 0 0 0.3312 -0.6137
67
Figure 3-12:
12: Diplexer prototype response with 10 resonators from the second stage of optimization
process.
An LTE-Advanced
Advanced band 7 diplexer has been synthesized with symmetrical
bands with x=0.263157Hz, r=5, and a return loss of 20 dB at both channels. The
response of the diplexer is Chebyshev and it consists of 12 resonators directly coupled
together. The diplexer topology is shown in Figure 3-13.
13. The normalized external
quality factors are calculated using equation (3.9) as q e 7 = qe12 =2.6980 and q e1 =1.3490.
To obtain symmetrical bands as specified, the following conditions for coupling
fficients have been set in the optimization algorithm: m23 = m28 , m34 = m89 ,
coefficients
m 45 = m 9 ,10 , m 56 = m 9 ,10 ,11 , m 67 = m11,12 , m33 = −m88 , m44 = −m99 , m 55 = − m10 ,10 ,
m 66 = − m11,11 , m 77 = − m12 ,12 . The initial values of the coupling coefficients for adjacent
resonators and the coefficients m ii (i=3,…7) are set to 0.5.
The cost function in equation (3.9) has been used and the optimizati
optimization
on has been
done in two stages as explained earlier with an unconstrained local optimization
68
technique in first stage and constrained in the second stage. Frequency locations s ti
and stk in the first two terms in the cost function have been taken as s ti = [-0.263157i
-i], and stk = [0.263157i, i]. The initial values of return zeros locations s rj are set with
equal spacing as follows,
±0.2832i ±0.4225i ±0.5619i ±0.7013i ±0.8406i ±0.9800i
The final locations of return zeros are
±0.2700i ±0.4140i ±0.5580i ±0.7020i ±0.8460i ±0.9900i
69
The optimized normalized coupling matrix from stage 1 and stage 2 are shown
in Table 4-5 and Table 4-6 respectively, and the response of the diplexer illustrated in
Figure 3-14.
Table 3-5: Coupling matrix of diplexer with 12 resonators from the first stage of optimization
process.
1 2 3 4 5 6 7 8 9 10 11 12
1 0 0.7779 0 0 0 0 0 0 0 0 0 0
2 0.7779 0 0.3494 0 0 0 0 0.3494 0 0 0 0
3 0 0.3494 0.5532 0.2225 0 0 0 0 0 0 0 0
4 0 0 0.2225 0.617 0.2111 0 0 0 0 0 0 0
5 0 0 0 0.2111 0.6239 0.2149 0 0 0 0 0 0
6 0 0 0 0 0.2149 0.6238 0.2912 0 0 0 0 0
7 0 0 0 0 0 0.2912 0.6215 0 0 0 0 0
8 0 0.3494 0 0 0 0 0 -0.5532 0.2225 0 0 0
9 0 0 0 0 0 0 0 0.2225 -0.617 0.2111 0 0
10 0 0 0 0 0 0 0 0 0.2111 -0.6239 0.2149 0
11 0 0 0 0 0 0 0 0 0 0.2149 -0.6238 0.2911
12 0 0 0 0 0 0 0 0 0 0 0.2911 -0.6215
Table 3-6: Coupling matrix of diplexer with 12 resonators from the second stage of optimization
process.
1 2 3 4 5 6 7 8 9 10 11 12
1 0 0.7933 0 0 0 0 0 0 0 0 0 0
2 0.7933 0 0.3623 0 0 0 0 0.3623 0 0 0 0
3 0 0.3623 0.5462 0.2311 0 0 0 0 0 0 0 0
4 0 0 0.2311 0.6208 0.2103 0 0 0 0 0 0 0
5 0 0 0 0.2103 0.617 0.2215 0 0 0 0 0 0
6 0 0 0 0 0.2215 0.6333 0.3045 0 0 0 0 0
7 0 0 0 0 0 0.3045 0.6275 0 0 0 0 0
8 0 0.3623 0 0 0 0 0 -0.5462 0.2311 0 0 0
9 0 0 0 0 0 0 0 0.2311 -0.6208 0.2103 0 0
10 0 0 0 0 0 0 0 0 0.2103 -0.617 0.2215 0
11 0 0 0 0 0 0 0 0 0 0.2215 -0.6333 0.3045
12 0 0 0 0 0 0 0 0 0 0 0.3045 -0.6275
70
(a)
(b)
Figure 3-14:
14: Diplexer prototype response with 12 resonators. (a) From the first stage of
optimization process. (b) From the second stage of optimization process.
71
3.8 Frequency transformation
The design of a diplexer is given in the bandpass frequency domain, in which

the real diplexer operates. From the previous sections, the design of the proposed
diplexer takes place in the normalized frequency domain as a lowpass prototype.
Therefore, a frequency transformation from normalized frequency domain to bandpass
frequency domain is needed. This section presents frequency transformation formulas of
a lowpass prototype diplexer with given specification to bandpass [7]. An illustration of
the frequency mapping is shown in Figure 3-15. The frequency edges of the lowpass
prototype of the diplexer are (-Ωc, x1 ) and ( x2 , Ω c ) . These frequencies are mapped
into bandpass frequencies using the following transformation formula [9],
Figure 3-15: Lowpass to bandpass transformation.
 ω ω0 
Ω = α  −  (3.12)
 ω0 ω 
To map the band edges Ω = −Ω c to ω = ω1 , and Ω = Ω c to ω = ω2 ,
ω ω 
− Ω c = α  1 − 0  (3.13)
 ω0 ω1 
ω ω 
Ω c = α  2 − 0  (3.14)
 ω0 ω 2 
Solving equations (3.13) and (3.14) yields,
ω 0 = ω1ω 2 (3.15)
72
Ωc ω 2 − ω1
α= , where FBW = (3.16)
FBW ω0
That’s yelled
 ω ω0 
x = α  −  (3.17)
ω
 0 ω 
Solve equation (3.17)
xω 0
ω2 − ω − ω 02 = 0
α
2
 xω 0  (3.18)
−  + 4ω 02
xω 0  α 
ω= +
α 2
Where the second solution is ignored
Figure 3.16 shows response of the LTE

LTE-Advanced
Advanced band 7 diplexer in the
bandpass frequency domain after applying equation (3.18). It is specified at UL: 2.50–
2.57 GHz and DL: 2.62–2.69
2.69 GHz.
Figure 33-16: Diplexer bandpass response.
73
3.9 Summery
In this chapter, coupled resonator diplexer topology has been proposed. The
design procedure (topology and transmission characteristics of the diplexer) is based on
optimization of coupling matrix for multiple coupled resonators in a three-port network.
This diplexer can be designed using any type of resonators, and it does not involve
energy distribution networks in its structure, unlike the conventional diplexers. This
enables miniaturizing the diplexer structure.
A cost function has been formulated to be minimized in the optimization

algorithm, and diplexers with different topologies have been successfully synthesized. A
comparison between diplexers with the same specification and different numbers of
resonators has been presented. It has been found that higher isolation occurs with
increasing number of resonators between output ports but the level of complexity of
synthesis increases for coupled resonator structure.
According to the diplexer electrical specifications of different numbers of

resonators, the topology with 10 resonators has acceptable isolation and perfect
transmission and reflection response, moreover low computational complexity.
The next chapter shows the design and the physical structure of LTE-A-band 7
10-resonator diplexer with bandpass Chebyshev filtering response. The diplexer is
designed using microstrip resonators.
74
References
[1] W. Xia, X. Shang, M. Lancaster, “Responses Comparisons for Coupled-

Resonator Based Diplexers,” Passive RF and Microwave Components, 2012,
pp. 67- 75.
[2] T. Skaik, M. Lancaster, and F. Huang, “Synthesis of Multiple Output

Coupled Resonator Microwave Circuits Using Coupling Matrix
Optimization,” IET Journal of Microwaves, Antenna & Propagation, vol.5,
no.9, June 2011, pp. 1081- 1088.
[3] T. Skaik, M. Lancaster, “Coupled Resonator Diplexer without External

Junctions” Journal of Electromagnetic Analysis and Applications, 2011, pp.
238-241.
[4] A. Ravindran, K. Ragsdell and G. Reklaitis, Engineering Optimization:

Methods and Applications. 2ed Edition, John Wiley & Sons, Inc. 2006
[5] J. Bandler, R. Biernacki, S. Chen, D.G. Swanson, and S. Ye, “Microstrip

filter design using direct EM Field simulation,” IEEE Transactions on
Microwave Theory and Techniques, vol. 42, July 1994, pp. 1353-1359.
[6] S. Rao, Engineering Optimization Theory and Practice. 3rd Edition, John
Wiley & Sons, Inc. 2009
[7] T. Skaik, “A Synthesis of Coupled Resonator Circuits with Multiple Outputs

using Coupling Matrix Optimisation”, PhD Thesis, March 2011, School of
Electronic, Electrical and Computer Engineering, the University of
Birmingham.
[8] A. Jayyousi and M. Lancaster, “A gradient-based optimization technique

employing determinants for the synthesis of microwave coupled filters,”
IEEE MTT-S International Microwave Symposium, USA, 2004, pp. 1369-
1372.
[9] J. Hong, Microstrip filters for RF/Microwave applications. 2ed Edition, A

John Wiley & Sons, Inc. 2011.
75
Chapter 4
Diplexer Design
4.1 Introduction
In the previous Chapters, a design methodology of coupled resonator diplexers

has been presented with numerical examples. This Chapter exhibits design of diplexer
for band 7 in LTE-Advanced cellular telephone for base transceiver stations. It is
specified at UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz. The design of this device has
been done using hairpin resonators. The theory of hairpin relevant to the design process
of microstrip components is first discussed in this chapter. The extraction of coupling
coefficients and external quality factors from physical structure will be discussed.
4.2 Microstrip
The microstrip structure consists of a conducting strip with thickness t and width
w located on top of a dielectric material (substrate) with dielectric constant ε r and
height h as shown in Figure 4-1. The bottom of structure is ground plane. The structure
of microstrip is inhomogeneous because the field exists in both air above the strip line
and the dielectric material and hence the field is quasi TEM [1].
Figure 4-1: Microstrip structure [1].
76
Due to inhomogeneous structure of microstrips, the effective dielectric constant
ε re is defined as the ratio between the capacitance per unit length with the presence of
the dielectric ( C d ) and the capacitance per unit length with air replacing the dielectric
substrate ( C a ) as follows [1]:
Cd
ε re = (4.1)
Ca
The characteristic impedance is defined as follows:
1
Zc = (4.2)
c Ca C d
Where c is electromagnetic wave velocity in free space (300,000 km/s).
Equations for effective dielectric constant and characteristic impedance within

1% accuracy for very thin conductors are provided [1]:
For w/h ≤1
ε r + 1 ε r − 1   w  
−0.5 2
h
ε re = + 1 + 12  + 0.041 −   (4.3)
2 2  w  h  
η  8h w
Zc = ln + 0.25  (4.4)
2π ε re w h
Where η is the wave impedance in free space and is equal to 120π.
For w/h≥1:
−0.5
εr +1 εr −1 h
ε re = + 1 + 12  (4.5)
2 2  w
−1
η w w 
Zc =  + 1.393 + 0.677 ln + 1.444  (4.6)
ε re  h h 
77
4.2.1 Microstrip Design
The design equations of microstrip line are given in terms of Z c and ε r as

follows [1]: For w/h ≤ 2
w 8 exp( A)
= (4.7)
h exp(2 A) − 2
With
0.5
Z ε  ε r −1  0.11
A = c  r +1  + 0.23 +  (4.8)
60  2  ε r +1  εr 
And for w/h ≥2
w 2 ε −1
= (B − 1) − ln(2 B − 1) + r [ln(B − 1)] + 0.39 + 0.61 (4.9)
h π 2ε r εr 
With
60π 2
B= (4.10)
Zc ε r
The guided wavelength for a microstrip line is given in millimetres in terms of

the operation frequency in GHz as follows:
300
λg = mm (4.11)
f (GHz ) ε re
and the propagation constant β is given by:
2π
β= (4.12)
λg
If the physical length of a microstrip line is l, then the electrical length θ = βl ,

and if the microstrip is quarter-wavelength ( l = λ g / 4 ) or half-wavelength ( l = λ g / 2 ),
then the electrical length will be θ = π / 2 , θ = π respectively [1]. Quarter wavelength

and half-wavelength microstrip lines are used in the design of microstrip diplexers [1].
78
4.2.2 Resonators
A resonator is a structure that contains one oscillating electromagnetic field at

least. There are various types of microstrip resonators. Quarter wavelength resonators (
λ g / 4 long) and half wavelength resonators ( λ g / 2 long) are used in microwave
diplexers, where λ g is the guided wavelength at the resonant frequency. λ g / 4
resonators have higher resonant frequencies at f ≈ (2n − 1) f 0 , for n = 2,3,..., whereas
λ g / 2 resonators have higher resonant frequencies at f ≈ nf 0 for n = 2,3,... . The

distributed line resonators can be shaped into various configurations such as hairpin,
spirals and open loop resonators [1].
4.3 Hairpin Diplexers
In microwave communication systems, high performance and small size

bandpass filters are essentially required to enhance the system performance and to
reduce the fabrication cost.
Microstrip diplexers are already small in size compared with other diplexers,
such as waveguide diplexers. However, for some applications where the size reduction
is of primary importance, smaller microstrip diplexers are desirable, even though
reducing the size of a diplexer generally leads to an increase in dissipation losses in a
given material and, consequently, reduced performance. Miniaturization of microstrip
diplexers may be achieved by using high dielectric constant substrates or lumped
elements. However, very often, for specified substrates, a change in the geometry of
diplexers is required and, therefore, many new diplexer configurations become possible
[1]. A conventional Parallel coupled diplexer with half-wavelength resonator is too
large to be used in the modern communication system such as LTE-Advanced system,
their large size is incompatible with these systems where size is an important
consideration [2]. The length of parallel coupled diplexer increases with the order of
diplexer. To solve this problem, hairpin line diplexer using folded λ/2 resonator
structures were developed [3,4]. The traditional design of the hairpin topology has the
advantage of compact structure, but it has the limitation of wider bandwidth [5]. In
addition to small size, high selectivity and narrow bandwidth, good Return Loss (RL)
and low cost are desirable features of narrowband bandpass microstrip diplexers.
79
4.3.1 Hairpin model
Hairpin-line bandpass diplexers are compact structures. They may conceptually

be obtained by folding the resonators of parallel-coupled, half-wavelength resonator,
into a “U” shape. Consequently, the same design equations for the parallel-coupled,
half-wavelength resonator may be used [1]. U-shaped resonators make progress in
circuit size reduction from the parallel coupled line structure. However, to fold the
resonators, it is necessary to take into account the reduction of the coupled-line lengths,
which reduces the coupling between resonators [1]. If the two arms of each hairpin
resonator are also closely spaced, they function as a pair of coupled lines, which can
also have an effect on the coupling. An element of hairpin bandpass resonator are shown
in Figure 4-2. The advantage of hairpin diplexer over end coupled and parallel coupled
Microstrip realizations, is the optimal space utilization. This space utilization is
achieved by folding of the half wavelength long resonators.
Figure 4-2: Tapped Hairpin resonator [1].
A conventional hairpin resonator in Figure 4-3 (a) may be miniaturized by

loading a lumped-element capacitor between the both ends of the resonator, as indicated
in Figure 4-3 (b), or alternatively with a pair of coupled lines folded inside the resonator
as shown in Figure 4-3 (c) [1]. In [1], it has been demonstrated that the size of a three-
pole miniaturized hairpin resonator diplexer is reduced to one-half that of the
conventional one.
80
Figure 4-3: Structural variations to miniaturize hairpin resonator. (a) Conventional
hairpin resonator. (b) Miniaturized hairpin resonator with loaded lumped capacitor.
(c) Miniaturized hairpin resonator with folded coupled lines [1].
Tapped line input and coupled line input are the two types of hairpin structures
that are commonly used in diplexer realization and are shown in Figure 4-4(a) and (b)
respectively. Tapped line input has a space saving advantage over coupled line input,
while designing the coupling line is required for the input and output high external
quality factor.
4.3.2 Hairpin Coupling Structures
Shown in Figure 4-5 are the three basic coupling structures. The coupled
structures result from different orientations of a pair of identical hairpin resonators. It is
clear that any coupling in those coupling structures is that of the proximity coupling,
which is, basically, through fringe fields [6]. The nature and the extent of the fringe
fields determine the nature and the strength of the coupling. It can be shown that at
resonance, each of the hairpin resonators has the maximum electric field intensity at the
side with an open side, and the maximum magnetic field intensity at the opposite side.
Because the fringe field exhibits an exponentially decaying character outside the
region, the electric fringe field is stronger near the side having the maximum electric
field distribution, while the magnetic fringe field is stronger near the side having the
maximum magnetic field distribution [6].
81
(a)
(b)
Figure 4-4: Hairpin Structures. (a) Tapped line input 8-pole Hairpin diplexer. (b) Coupled line
input 8-pole Hairpin diplexer.
It follows that the electric coupling can be obtained if the open sides of two
coupled resonators are proximately placed as Figure 4-5 (a) shows, while the magnetic
coupling can be obtained if the sides with the maximum magnetic field of two coupled
resonators are proximately placed as Figure 4-5 (b) shows. For the coupling structure in
Figure 4-5 (c), the electric and magnetic fringe fields at the coupled sides may have
comparative distributions so that both the electric and the magnetic couplings occur [1].
In this case the coupling may be referred to as the mixed coupling.
82
Figure 4-5: Basic coupling structures of coupled microstrip hairpin resonators. (a) Electric coupling
structure. (b) Magnetic coupling structure. (c) Mixed coupling structure.
4.3.3 Unloaded quality factor
The unloaded quality factor Qu is a figure of merit for a resonator. It describes

the quality of the resonator in terms of losses and energy storage. For example, a high Q
resonator implies low energy loss and good energy storage, whereas a low Q implies
higher losses. A general definition for the Qu that applies to any type of resonator is [1],
Time - average energy stored in the resonator

Qu = ω (4.13)
Average power lost in the resonator
The losses in a resonator can generally be associated with the conductor,

dielectric material, and radiation. The total Qu may be defined by adding these losses
together as follows [1],
1 1 1 1
= + + (4.14)
Qu Qc Qd Qr
where Qc , Qd and Qr are the quality factors associated with losses from conductor and
dielectric making up the resonator and radiation from the cavity. The loaded quality
factor QL may be defined in terms of the unloaded quality factor Qu and the external
quality factor Qe as follows [7],
1 1 1
= + (4.15)
QL Qu Qe
where Qe is the quality factor associated with effective losses through the external
coupling circuit, and it is defined as the ratio of the energy stored in the resonator to the
energy coupled to the external circuit. The extraction of the external quality factor from
the physical structure will be described in the next section. The conductor quality factor
83
is adversely proportional to the surface resistance of the conductor sheets. At low
frequencies, the total dc resistance of hairpin is defined as [8]:
l
Rdc = (4.16)
Wtσ
where l is the total length of the conductor in the coil, W is the track width of the
hairpin, t is the thickness, and σ is the conductivity of the conductor. At high
frequencies, the resistance is defined as [8]:
l
Rrf = (4.17)
 1

 δ 
Wσδ 1 − e 
 
where δ is the skin depth and is defined by [10]:
2
δ= (4.18)
ωµσ
For thick conductors, t / δ < 1, and hence Rrf ≅ Rdc .
The dielectric quality factor Qd is adversely proportional to the loss tangent of

the dielectric substrate. For a substrate with very high resistivity, the loss tangent is very
small, and the ohmic losses in the dielectric material are very small, whereas for a
substrate with very low resistivity, the electric field penetration inside the substrate is
limited and the ohmic losses take place in this case [8].
The radiation quality factor Qr is generally defined as [1]:
time average energy stored in resonator

Qr = ω (4.19)
average power radiated
Normally, the filters are shielded in housing walls, so the power radiated will be lost in
the imperfect conducting walls [1].
4.4 Coupling Coefficients and external coupling, Simulation and analysis
This section discusses the experiments done such as adjusting the hairpin
resonator to a specific resonant frequency, finding and analyzing the coupling
coefficients for different structures, and finding and controlling the external coupling.
84
4.4.1 Coupling in Physical Terms
After determining the normalized coupling matrix [m] for a coupled resonator
topology, the actual coupling matrix [M] of a coupled resonator device with given
specification can be calculated by prototype demoralization of the matrix [m] at a
desired bandwidth and a centre frequency, as follows [1],
M ij = mij .FBW (4.20)
where FBW is the fractional bandwidth. The actual external quality factor Qe is related
to the normalized quality factor qe by [1],
qe
Qe = (4.21)
FBW
The next step is to construct a structure of coupled resonators and implement the
required coupling coefficients of the matrix [M] physically. The extraction of the
coupling coefficient M ij of two coupled resonators and the external quality factor Qe
from the physical structure is presented in the next subsections.
4.4.2 Extraction of coupling coefficient from physical structure
In coupled resonator circuits, the coupling coefficient for a selected resonator

pair can be obtained from the physical structure using EM simulation. To extract the
coupling coefficient of two asynchronously coupled resonators, a general formula that
applies to any type of resonators is used [9],
2 2
1 ω ω   ω 22 − ω12   ω02
2 2
− ω01 
M = ±  02 + 01  ×  2 
2 
− 
 2 2 
 (4.22)
2  ω01 ω02   ω 2 + ω1   ω02 + ω01 
where ω01 and ω02 are the resonant frequencies of the two coupled resonators, ω1 and
ω 2 are the lower and higher frequencies in the magnitude of S 21 response of the two
coupled resonator structure with the ports are very weakly coupled to the resonators.
The characteristic parameters ω01 , ω02 , ω1 and ω 2 can be determined using full-wave
EM simulations. The coupling between two resonators of the diplexer configuration can
be extracted from EM simulation (CST Microwave studio) using an arrangement of
85
Figure 4-6. The two resonators have different size and different resonant frequency. The
coupling is mainly controlled by the spacing between them. For the EM simulation
(CST Microwave studio), the coupled resonators are very weakly excited by the two
ports as arranged. Two resonant peaks, which result from the mode split because of the
coupling between the two resonators, can clearly be observed from the EM-simulated
frequency responses as shown in Figure 4-7. The coupling coefficient can then be
extracted using equation (4.22).
Figure 4-6: Two coupled hairpin resonators.
Figure 4-7: Amplitude response of S 21 for two coupled resonators [9].
The formula in equation (4.22) is applicable for synchronously coupled

resonators, and in this case it is simplified to [9],
ω 22 − ω12
M =± (4.23)
ω 22 + ω12
86
The coupling coefficient usually corresponds to a magnetic coupling or an
electric coupling. These two types of coupling exhibit opposite signs for the coupling
coefficient.
The self-coupling coefficient M ii in the principal diagonal in the coupling matrix [M]
that accounts for asynchronous tuning is related to the self-resonant frequency of
resonator i by [1],
 ω 2 − ω02 
M ii = 2 ×  02i 
2 
(4.24)
 ω 0i + ω 0 
where ω 0 is the desired centre frequency of the coupled-resonator device and ω0i is the
resonant frequency of resonator i .
4.4.3 Extraction of external quality factor from physical structure
The external quality factor of a single resonator can be found by simulating S 21
response with one port weakly coupled. Shown in Figure 4-8 is an arrangement to
extract the external quality factor of the I/O resonator. The resonator, which is assumed
to be lossless in the simulation, is excited at port 1 through a 50-Ω coupled line. Port 2
is very weakly coupled to the resonator in order to find a 3-dB bandwidth of the
magnitude response of S 21 for extracting the single-loaded external quality factor Qe .
The external quality factor Qe can then be calculated from the simulated S 21 response
using the following formula [1],
ω0
Qe = (4.25)
∆ω ±3dB
where ω 0 is the resonant frequency of the loaded resonator and ∆ω ±3dB is the 3 dB
bandwidth, as shown in Figure 4-9.
87
Figure 4-8: Externally coupled hairpin resonator.
Figure 4-9: Amplitude response of S 21 for externally coupled resonator [1].
4.5 LTE-A 10-resonator diplexer
4.5.1 Diplexer Design
When a microstrip diplexer design involves full-wave EM simulations, it is

computationally superior to decompose the diplexer into different parts that are
simulated individually by the EM simulator (CST Microwave studio) to extract the
desired design parameters according to a prescribed general coupling matrix, as
discussed in previous sections. They are then combined to obtain the response of the
overall diplexer [1], then to improve the design; optimization techniques are applied to
whole structure. This CAD approach, which is particularly effective for narrowband
diplexer designs, is demonstrated with a ten-pole direct-coupled diplexer, which
88
consists of ten folded half-wavelength resonators resulting in a compact diplexer
topology on a dielectric substrate with a thickness denoted by h . Resonators 1,6 and10
are the input and outputs (I/O) resonators, respectively, and there exists a direct
coupling between them as shown in
Figure 4-10.
Figure 4-10: The layout of the LTE-A 10-resonator diplexer.
LTE-Advanced 10-resonator diplexer has been designed using hairpin microstrip

resonators. The diplexer Used for LTE-Advanced system-band 7, specifically UL: 2.50–
2.57 GHz and DL: 2.62–2.69 GHz. The diplexer topology is shown in Figure 4-11, and
the ten-pole microstrip direct-coupled diplexer is designed based on the optimized
normalized coupling matrix which shown in Table 4-1 with the scaled external quality
factor qe1 =1.318046133, qe 6 = qe10 =2.636092266. The prototype response of the
diplexer is shown in Figure 4-12.
89
Figure 4-11: Diplexer topology.
Table 4-1: Normalized coupling matrix of 10-resonator diplexer
1 2 3 4 5 6 7 8 9 10
1 0 0.7943 0 0 0 0 0 0 0 0
2 0.7943 0 0.3795 0 0 0 0.3795 0 0 0
3 0 0.3795 0.5435 0.2582 0 0 0 0 0 0
4 0 0 0.2582 0.6066 0.2429 0 0 0 0 0
5 0 0 0 0.2429 0.6267 0.3312 0 0 0 0
6 0 0 0 0 0.3312 0.6137 0 0 0 0
7 0 0.3795 0 0 0 0 -0.5435 0.2582 0 0
8 0 0 0 0 0 0 0.2582 -0.6066 0.2429 0
9 0 0 0 0 0 0 0 0.2429 -0.6267 0.3312
10 0 0 0 0 0 0 0 0 0.3312 -0.6137
90
Figure 4-12:
4 Diplexer prototype response.
The fractional bandwidth for the diplexer is FBW = 0.073267 and the center
frequency of channel 1 is 2.535 GHz and that of channel 2 is 2.655 GHz. Thus, the
desired design parameters (external quality factor and the actual cou
coupling
pling matrix) can be
found from Equations (4.20) and (4.21) as follows and as listed in table 4-2.
4
Qe1 = 17.98966833, Qe6 = Qe10 = 35.97933665.
Table 4-2:
2: The actual coupling matrix of 10
10-resonator diplexer.
1 2 3 4 5 6 7 8 9 10
1 0 0.057632 0 0 0 0 0 0 0 0
2 0.057632 0 0.026545 0 0 0 0.026545 0 0 0
3 0 0.026545 0.040583 0.017284 0 0 0 0 0 0
4 0 0 0.017284 0.04522 0.016881 0 0 0 0 0
5 0 0 0 0.016881 0.045499 0.022434 0 0 0 0
6 0 0 0 0 0.022434 0.04525 0 0 0 0
7 0 0.026545 0 0 0 0 -0.04058 0.017284 0 0
8 0 0 0 0 0 0 0.017284 -0.04522
0.04522 0.016881 0
9 0 0 0 0 0 0 0 0.016881 -0.0455 0.022434
10 0 0 0 0 0 0 0 0 0.022434 -0.04525
91
The resonators with non-zero self coupling coefficients (Mii≠0) have frequencies
other than the centre frequency of the whole diplexer (2.59326 GHz). The
corresponding resonant frequencies of each resonator are calculated using equation
(4.24) and they are listed in table 4-3.
Table 4-3: Resonators and their resonant frequencies.
Resonator Resonant
Number Frequency (GHz)
1 2.59326
2 2.59326
3 2.6464
4 2.6526
5 2.6529
6 2.6526
7 2.5412
8 2.5353
9 2.5349
10 2.5352
4.5.2 Diplexer physical structure and simulation
Full-wave EM simulations are carried out to extract the desired external quality
factors Q and coupling coefficients M against the physical dimensions, using the
methods described in section 4.
For this CAD design, all EM simulations are performed using commercially
software, CST which is a specialist tool for the 3D EM simulation of high frequency
components. CST uses finite integration method (FIM) as a numerical analysis
technique [10]. The hairpin length is about λ g / 2 long with λ g in equation (4.11) the
guided wavelength at the resonator self-frequency. The hairpin resonators have a line
width that depends on the characteristic impedance of the hairpin line Z c = 68.3Ω on
the substrate used. The diplexer is designed to have coupled lines input and outputs. The
coupled line is chosen to have characteristic impedance that matched a terminating
impedance Z o = 50Ω [1]. A commercial substrate (RT/D 6006) with a relative
dielectric constant of 6.15, a thickness h =1.27 mm has used for microstrip
representation. The layout of the final diplexer design and all determined dimensions are
92
illustrated in Figure 4-13 and Table 4-4 respectively. The diplexer has an overall size of
80 by 53 mm. The input and output resonators are slightly shortened to compensate for
the effect of the coupled line and the adjacent coupled resonator. The EM simulated
performance of the diplexer is shown in Figure 4-14.
Table 4-4: The LTE-A 10-resonator diplexer dimensions.

Parameter Final Value (mm) Parameter Final Value (mm)
L1 11.935 L6 13.365
W1 4 W6 2
D1 1 D6 1
LP1 10.590 S3 0.778
WP1 1.867 LP3 13
S1 0.308 WP3 1.867
S12 0.674 T3 12.539
F1 8.793 L7 14.076
F2 8.957 W7 2
L2 12.321 D7 1
W2 4 S 78 3.154
D2 1 L8 14.155
S 23 1.630 W8 2
S 27 1.394 D8 1
L3 13.387 S89 3.286
W3 2 L9 14.152
D3 1 W9 2
S 34 3.341 D9 1
L4 13.342 S 910 2.369
W4 2 L10 13.965
D4 1 W10 2
S 45 3.301 D10 1
L5 13.402 S2 0.483
W5 2 LP2 13
D5 1 WP 2 1.867
S 56 3.004 T2 11.02218
93
Figure 4-13: The layout of the LTE-A 10-resonator diplexer desig
94
Figure 4-14: The EM simulated performance of the diplexer
The performance of the entire diplexer is illustrated in Figure 4-14. It can be shown
from the simulation results that the return loss is better than 12 dB in the transmit and
receive band, the insertion loss is only about 0.3 dB in transmit and receive band and
isolation greater than 60 dB in the uplink channel and greater than 35 dB in the
downlink channel.
4.6 Summary
The design of coupled-resonator diplexer using hairpin microstrip resonators has been
presented. The diplexer design of LTE-A Band 7 has been presented for the use as
component in the base station transceiver of a cellular communications system. 10
resonators were needed in the LTE-A Band 7 diplexer to achieve the diplexer electrical
requirements.
95
References
[1] J. Hong , Microstrip Filters for RF/Microwave Applications,2nd edition, John

Wiley & Sons, Inc.,NY, 2011.
[2] R. Collin, Foundation for Microwave Engineering, 2nd edition, IEEE Press,
Wiley, New York, 1992.
[3] G. Edward and S. Frankel, "Hairpin-line and hybrid hairpin/half-wave parallel

coupled line filters," Microwave Theory and Techniques, vol. MTT-22, no 11,
Novamber 1972, pp. 719-728.
[4] H. Gysel, "New theory and design for hairpin-line filters," IEEE Transactions
on Microwave Theory and Techniques, vol. MTT-22, no. 5, May 1974, pp.
523-531.
[5] K. Singh, R. Ramasubramanian and S. Pal, "Coupled Microstrip filters: Simple

Methodologies for improvred characteristics, communcation system group,"
India EESOF User Group Meeting, November 2005.
[6] J. Hong and M. Lancaster, "Coupling of Microstrip Square Open-Loop

Resonators for Cross-Coupled Planar Microwave Filters," IEEE Transactions
on Microwave Theory and Techniques, vol. 44, no.12, December 1996, pp.
2099-2109.
[7] D. Pozar, Microwave Engineering. 2nd edition, Wiley, 1998.
[8] I. Bahl, Lumped Elements for RF and Microwave Circuits, Artech House, Inc.,
Boston, 2003.
[9] J. Hong, "Couplings of asynchronously tuned coupled microwave resonators,"

IEE Proceedings Microwaves, Antennas and Propagation, vol.147, no.5,
October. 2000, pp. 354-358.
[10] CST Website, [Online]. Available: http://www.cst.com/content/products/mws/FIT.aspx
96
Chapter 5
Conclusions and Future Work
5.1 Conclusions
The thesis has search into coupled resonator circuits with multiple outputs.
Design techniques used for coupled resonator filters have been extended here to
multiple output circuits, and a three-port coupled resonator diplexer has been designed
in order to meet the specification of LTE-Advanced system, specifically band number 7
(i.e. UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz). The design approach allows
synthesis of diplexers from coupled resonator structures and hairpin microstrip
resonators have been used in the design.
The coupling matrix of a multiport circuit with multiple coupled resonators has
been used in the synthesis. A unified solution for the coupling matrix has been utilized
and it is generalized for both types of magnetic and electric couplings. Transmission and
reflection scattering parameters of a three-port coupled resonator circuit have been
found in terms of the general coupling matrix. A gradient based optimization technique
has been employed here to produce coupling matrix for the coupled resonator diplexer.
Coupled resonator diplexer has been successfully synthesized using optimization.
Unlike conventional diplexers, the proposed diplexer do not has any external junctions
for distribution of energy, such as waveguide manifolds, or T-junctions, or circulators.
Thus, the proposed diplexers can be miniaturized.
The synthesis procedure of coupling matrix of the proposed coupled-resonator

diplexer using optimization is based on minimization of a cost function that is evaluated
at frequency locations of reflection zeros.
Both unconstrained and constrained local optimization techniques have been

utilized to synthesis the coupling matrix, and a formula has been used to calculate the
external quality factors of diplexers with symmetrical channels. Setting the values of the
external quality factors at the outset of the optimization algorithm enhances the
convergence time. The optimization algorithm may converge to a local minimum. To
solve this problem, the optimization has been done in two stages. The first stage
assumed equally spaced reflection zeros and the resulting coupling coefficients have
97
been used as initial values for the second stage. The frequency locations of the reflection
zeros are then allowed to move in the second stage until equiripple level at the specified
insertion loss is achieved.
A comparison between diplexers with the same specification (channels

bandwidth, channels separation, and return loss) and different number of resonators has
been carried out. It has been found that the higher the number of resonators in the path
between output ports, the better the isolation.
LTE-Advanced 10-resonator diplexer has been designed using hairpin microstrip
resonators. EM CST simulator has been used to extract the desired design dimensions
according to a prescribed general coupling matrix and external quality factors. The
simulation results show a return loss better than 12 dB and isolation better than 60 dB in
the uplink channel and better than 35 dB in the downlink channel.
5.2 Future Work
In the future, the LTE-Advanced 10-resonator diplexer will be fabricated and

test. The work on coupled resonator diplexer can be further developed for more
complicated coupling structure; increase the number of resonators to improve the
isolation between the diplexer channels, or change the diplexer topology by using cross
coupling between resonators to achieve transmission zeros (elliptic response), and hence
sharp transition bands and selectivity can be improved. Further the structure dimensions
can be reduced by replacing the hairpin resonators with spirals resonators or
miniaturized hairpin resonators with folded lines.
98
Appendix A
Chebyshev Lowpass Prototype Filters
The g -values of lowpass prototype filters are used to calculate the coupling
coefficients and the external quality factors of coupled resonator bandpass filters. The
g -values for an N th order Chebyshev lowpass prototype filter with a passband ripple of
L Ar dB may be extracted from the following formulae,
g0 = 1
2  π 
g1 = sin  
γ  2N 
 (2i − 1)π   (2i − 3)π 
4 sin  . sin 
gi =
1  2N   2 N  for i = 2,3, K N (A.1)
g i −1  (i − 1)π 
γ 2 + sin 2 
 N 
 1 for N odd

g N +1 =  β 
coth 2   for N even
 4
Where
  LAr 
β = ln coth 
  17.37 
(A.2)
 β 
γ = sinh 
 2N 
The coupling matrix values and the external quality factors for a coupled
resonator bandpass filter with centre frequency of ω0 and passband edges of ω1 and ω2
may be found from the gvalues of the lowpass prototype filter as follows,
FBW
M i ,i +1 = for i = 1, K , N − 1
g i g i +1
(A.3)
g g g g
Qe1 = 0 1 , QeN = N N +1
FBW FBW
99
where FBW is the fractional bandwidth given by
ω 2 − ω1
FBW = (A.4)
ω0
ω 0 = ω1ω 2 (A.5)
100
Appendix B
Gradient based Algorithm

Step 1. Define M= maximum number of allowable iterations
N = number of variables
x ( 0) = initial estimate of x *
ε 1 = overall convergence criteria
ε 2 = line search convergence criteria
Step 2. Set k =0.
Step 3. Calculate ∇f ( x ( k ) ).
Step 4. Is ∇f ( x ( k ) ) ≤ ε 1 ?
Yes: Print ‘‘convergence: gradient’’; go to 13.
No: Continue.
Step 5. Is k > M?
Yes: Print ‘‘termination: k = M’’; go to 13.
No: Continue.
Step 6. Calculate s( x (k ) ).
From Quasi-Newton Methods
s ( x ( k ) ) = − B ( k ) ∇f ( x ( k ) )
B ( 0) = I
∆x ( k −1) ∆x ( k −1)T B ( k −1) ∆g ( k −1) ∆g ( k −1)T B ( k −1)
B (k ) = B k + −
∆x ( k −1)T ∆g ( k −1) ∆g ( k −1)T B ( k −1) ∆g ( k −1)
∆x ( k −1) = x ( k ) − x ( k −1)
∆g ( k −1) = ∇f ( x ( k ) ) − ∇f ( x ( k −1) )
or from Conjugate Gradient Methods
 ∇f ( x ( k ) ) 2 
S ( x ) = −∇f ( x ) + 
(k ) (k )  s ( x ( k −1) )
 ∇f ( x ( k −1) ) 2 
 
Step 7. Is ∇f ( x ( k ) ) s ( x ( k ) ) < 0 ?
Yes: Go to 9.
No: Set s( x ( k ) ) = −∇f ( x ( k ) ). Print ‘‘restart: bad direction’’; Go to 9.
Step 8. Find α (k ) such that f ( x ( k ) + α ( k ) s ( x ( k ) ))). → minimum using ε 2 .
Step 9. Set x ( k +1) = x ( k ) + α ( k ) s( x ( k ) ).
Step 10. Is f ( x ( k +1) ) < f ( x ( k ) ) ?
Yes: Go to 11.
No: Print ‘‘termination: no descent’’; go to 13.
Step 11. Is ∆x / x ( k ) ≤ ε 1 ?
Yes: Print ‘‘termination: no progress’’; go to 13.
No: Go to 12.
Step 12. Set k = k + 1 . Go to 3.
Step 13. Stop.
101

Mailadil T. Sebastian (Editor) - Microwave Materials and Applications, 2 Volume Set-Wiley (2017)

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Mailadil T. Sebastian (Editor) - Microwave Materials and Applications, 2 Volume Set-Wiley (2017)

Caricato da

Copyright:

Formati disponibili

Islamic University of Gaza

Deans of Graduate Studies

Coupled Resonator Diplexer for LTE

A Thesis Submitted in Partial Fulfillment of the Requirements for the

The synthesis procedure of the proposed coupled-resonator diplexer is based on optimization of

My acknowledgement also goes to the department of Electrical and Communication

Chapter 2: Overview of Diplexers ...................................................................................... 16

Chapter 3: Synthesis of Coupled Resonator Diplexer .................................................. 48

Chapter 4: Diplexer Design .............................................................................................. 76

Chapter 5: Conclusions and Future Work..................................................................... 97

Appendix B ........................................................................................................................ 101

Table 1-1: Operation bands for LTE-Advanced ............................................................... 7

1.1 Technology Background

Communication is everywhere, used by everybody, and at any time - Cellular

The objective of this Thiess is to analyze and design a coupled resonator

1.2 Evolution of wireless standards

With the invention of microprocessors and the release of the cellular

The Second generation (2G) mobile communication systems were created to

3.5 G is a grouping of disparate mobile telephony and data technologies

The fourth generation (4G) is often ccalled LTE-Advanced (LTE-A)

1.3.1 The 2.6 GHz Spectrum Band “Band 7”

5 824 MHz – 849 MHz 869 MHz – 894MHz 20 MHz FDD

6 830MHz – 840 MHz 875MHz – 885 MHz 35 MHz FDD

41 3400 MHz – 3600MHz 3400 MHz – 3600MHz N/A TDD

• ITU Option 3 allows freedom of choice about the respective amounts of

Figure 1-4: ITU Options for the 2.6GHz Band [24].

LTE-Advanced transceiver configured for multi-mode and multi-band as an

Diplexers are indispensable components in RF/microwave wireless

A diplexer is a three-port network which usually consists of two filters used to

1.3 Thesis Motivation

The goal of this thesis is to design microstrip diplexer using 10 coupled

Conventional diplexers consist of two channel filters connected to a junction for

1.4 Thesis Overview

Chapter 3 conducts the synthesis of diplexer that specifies LTE-Advanced band

[2] N. Boucher, Cellular Radio Handbook – A Reference for Cellular System

[7] E. Dahlman, S. Parkvall, J. Sköld and P. Beming, 3G Evolution HSPA and

[19] Discussion Paper, Fixed or flexible? A survey of 2.6GHz spectrum awards,

[20] Next Generation Mobile Networks Initial Terminal Device Definition,

[22] Mobile Device Products, TriQuint Semiconductor "Connecting the Digital

[23] G. Matthaei, and G. Cristal, “Multiplexer channel-separating units using

[24] J. Wenzel, “Printed-circuit complementary filters for narrow bandwidth

[25] J. Reed, Software Radio: A Modern Approach to Radio Engineering. 1st

[29] T. Skaik and M. Lancaster, “Coupled Resonator Diplexer without External

Full duplex systems allow simultaneous radio transmission and reception

2.2 TDD Systems

2.3 FDD Systems

Figure 2-2: The principle of operation in FDD systems [2].

2.4 Conventional Diplexer

Figure 2-3: Conventional duplexer [3].

The most commonly used distribution configurations are E- or H-plane n-

Figure 2-5: Configuration of circulator-coupled diplexer [4,15].

In manifold configurations, channel filters are connected by transmission lines:

2.4.1 Literature Review

Many structures of diplexers have been proposed in literature. In [16], a novel

Figure 2-8: Fabricated microstrip diplexer [13].

Figure 2-9: Fabricated microstrip hairpin diplexer [17].

Figure 2-10: The Microstrip Diplexer [18].

In [19] a microstrip diplexer is proposed which has good performance on the

Figure 2-11: The microstrip diplexer [19].

In [20] a compact diplexer based on folding microstrip lines is presented. The

Figure 2-12: Circuit layout of the diplexer [20].

Figure 2-13: a compact diplexer using a square open loop [21].

The synthesis procedure of the proposed diplexer in this thesis is based on