0 valutazioniIl 0% ha trovato utile questo documento (0 voti)

9 visualizzazioni112 pagineLý thuyết trường điện từ và anten

Nov 22, 2018

© © All Rights Reserved

PDF, TXT o leggi online da Scribd

Lý thuyết trường điện từ và anten

© All Rights Reserved

0 valutazioniIl 0% ha trovato utile questo documento (0 voti)

9 visualizzazioni112 pagineLý thuyết trường điện từ và anten

© All Rights Reserved

Sei sulla pagina 1di 112

Faculty of Engineering

Electrical Engineering Department

LTE-Advanced

Advanced

System

By

Hazem M. Abukaresh

Supervisor

Dr

Dr. Talal F. Skaik

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.

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.

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

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

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

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

5.1 Conclusions .................................................................................................................... 97

5.2 Future Work ................................................................................................................... 98

Appendix A .......................................................................................................................... 99

ChebyshevLowpass Prototype Filters ............................................................................... 99

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

optimization process................................................................................................................. 64

Figure 3-10: Topology of Diplexer with 10 resonators ....................................................... 65

Figure 3-11: Diplexer prototype response with 10 resonators from the first stage of

optimization process................................................................................................................. 66

Figure 3-12: Diplexer prototype response with 10 resonators from the second stage of

optimization process................................................................................................................. 68

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 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

optimization process ........................................................................................................ 63

Table 3-3: Coupling matrix of diplexer with 10 resonators from the first stage of

optimization process ........................................................................................................ 66

Table 3-4: Coupling matrix of diplexer with 10 resonators from the second stage of

optimization process ........................................................................................................ 67

Table 3-5: Coupling matrix of diplexer with 12 resonators from the first stage of

optimization process ........................................................................................................ 70

Table 3-6: Coupling matrix of diplexer with 12 resonators from the second stage of

optimization process ........................................................................................................ 70

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

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.

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.

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].

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].

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].

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].

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

2 1850 MHz – 1910 MHz 1930 MHz – 1990 MHz 20 MHz FDD

3 1710 MHz – 1785 MHz 1805 MHz – 1880 MHz 20 MHz FDD

4 1710 MHz – 1755 MHz 2110 MHz – 2155 MHz 355 MHz FDD

7 2500 MHz – 2570 MHz 2620 MHz – 2690 MHz 50 MHz FDD

8 880 MHz – 915 MHz 925 MHz – 960 MHz 10 MHz FDD

9 1749.9 MHz – 1784.9 MHz 1844.9 MHz – 1879.9 MHz 60 MHz FDD

10 1710 MHz – 1770 MHz 2110 MHz – 2170 MHz 340 MHz FDD

11 1427.9 MHz – 1447.9 MHz 1475.9 MHz – 1495.9 MHz 23 MHz FDD

12 698 MHz – 716 MHz 728 MHz – 746 MHz 12 MHz FDD

13 777 MHz – 787 MHz 746 MHz – 756 MHz 21 MHz FDD

14 788 MHz – 798 MHz 758 MHz – 768 MHz 20 MHz FDD

15 Reserved Reserved – –

16 Reserved Reserved – –

17 704 MHz – 716 MHz 734 MHz – 746 MHz 18 MHz FDD

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

21 1447.9 MHz – 1462.9 MHz 1495.9 MHz – 1510.9 MHz 33 MHz FDD

22 3410 MHz – 3500 MHz 3510 MHz – 3600 MHz 10 MHz FDD

33 1900 MHz – 1920 MHz 1900 MHz – 1920 MHz N/A TDD

34 2010 MHz – 2025 MHz 2010 MHz – 2025 MHz N/A TDD

35 1850 MHz – 1910 MHz 1850 MHz – 1910 MHz N/A TDD

36 1930 MHz – 1990 MHz 1930 MHz – 1990 MHz N/A TDD

37 1910 MHz – 1930 MHz 1910 MHz – 1930 MHz N/A TDD

38 2570 MHz – 2620 MHz 2570 MHz – 2620 MHz N/A TDD

39 1880 MHz – 1920 MHz 1880 MHz – 1920 MHz N/A TDD

40 2300 MHz – 2400 MHz 2300 MHz – 2400 MHz 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.

spectrum in the band that are attributed to paired (FDD operation) and unpaired

(TDD operation) blocks.

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].

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

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.

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].

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.

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.

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.

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.

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.

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

R. Marsden, E. Sexton and A. Siong, DotEcon, June 2010, [Online].

Available:

http://dotecon.com/publications/dp1001.pdf

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.

World to the Global Network" , 2011, [Online]. Available:

http://www.triquint.com/prodserv/brochures/handset_brochure.pdf

interdigital and parallel-coupled filters,” IEEE Trans. Microw. Theory Tech.,

1965, pp. 328–334

multiplexers,” IEEE Trans. Microw. Theory Tech., 1968, 16, pp. 147–157.

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

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

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.

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].

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].

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].

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].

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.

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.

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.

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.

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.

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.

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.

24

2.5 Coupled Resonator Diplexer

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.

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].

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.

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].

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

parameters, such as impedance matrix, admittance matrix, and scattering matrix. Figure

2-18 shows a typical three-port network [38].

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

S ij = S ji i≠ j (2.5)

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].

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].

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

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

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].

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

es es − 2i1 R1

a1 = b1 =

2 R1 2 R1

a2 = 0 b2 = ix Rx (2.11)

a2 = 0 b3 = i y R y

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

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

may have different resonant frequencies, and extra entries mii are added to the diagonal

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

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

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 )

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

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

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].

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

is is − 2i1G1

a1 = b1 =

2 G1 2 G1

a2 = 0 b2 = v x Gx (2.21)

a2 = 0 b3 = v y G y

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

From previous sections formulations, the most notable is that the formulation of

normalized impedance matrix [Z ] is identical to that of normalized admittance matrix

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.

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

Figure 2-22: The frequency response of a sixh-order type I Chebyshev low-pass filter [40].

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

2.7 Summary

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.

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.

http://www.rfsolutions.com/duplex.htm

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.

Proceeding of the IEEE MTT-S International Microwave and Optoelectronics

Conference (IMOC), Nov. 2009, pp. 215 - 217.

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.

Evaluation of Suitable Characteristic Polynomials,” IEEE MTT-S International

Microwave Symposium, San Francisco, 11-16 June 2006, pp. 111-114.

microwave multiport networks,” IEEE MTT-S Int. Microwave Symp. Digest,

USA, Vol. 2, June 2004, pp. 455–458.

formed by coupled resonators with distributed coupling,” IEEE Antennas and

Propagation Society Int. Symp., USA, Vol. 1A, 2005, pp. 89–92.

Magazine, IEEE, Vol.12 , Issue: 6, 2011, pp. 101 – 109.

Progress in Electromagnetics Research Letters, Vol. 31, April 2012, pp. 113-

120.

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

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

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 .

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.

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.

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.

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.

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.

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].

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].

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 )

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

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

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,

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.

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.

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.

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.

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

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

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 .

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

59

3.7.1 Diplexer with n=8, x=0.263157, and r=3

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 .

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

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:

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

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

shown in Figure 3-8.

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

shown in Figure 3-9.

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

64

Figure 3-10: Topology of Diplexer with 10 resonators.

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

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

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 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],

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

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

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.

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.

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.

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

Resonator Based Diplexers,” Passive RF and Microwave Components, 2012,

pp. 67- 75.

Coupled Resonator Microwave Circuits Using Coupling Matrix

Optimization,” IET Journal of Microwaves, Antenna & Propagation, vol.5,

no.9, June 2011, pp. 1081- 1088.

Junctions” Journal of Electromagnetic Analysis and Applications, 2011, pp.

238-241.

Methods and Applications. 2ed Edition, John Wiley & Sons, Inc. 2006

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

using Coupling Matrix Optimisation”, PhD Thesis, March 2011, School of

Electronic, Electrical and Computer Engineering, the University of

Birmingham.

employing determinants for the synthesis of microwave coupled filters,”

IEEE MTT-S International Microwave Symposium, USA, 2004, pp. 1369-

1372.

John Wiley & Sons, Inc. 2011.

75

Chapter 4

Diplexer Design

4.1 Introduction

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].

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

Cd

ε re = (4.1)

Ca

1

Zc = (4.2)

c Ca C d

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 + 0.041 − (4.3)

2 2 w h

η 8h w

Zc = ln + 0.25 (4.4)

2π ε re w h

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

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

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 operation frequency in GHz as follows:

300

λg = mm (4.11)

f (GHz ) ε re

2π

β= (4.12)

λg

and if the microstrip is quarter-wavelength ( l = λ g / 4 ) or half-wavelength ( l = λ g / 2 ),

and half-wavelength microstrip lines are used in the design of microstrip diplexers [1].

78

4.2.2 Resonators

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

distributed line resonators can be shaped into various configurations such as hairpin,

spirals and open loop resonators [1].

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

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.

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.

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.

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],

Qu = ω (4.13)

Average power lost in the resonator

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

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

2

δ= (4.18)

ωµσ

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]:

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].

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],

where FBW is the fractional bandwidth. The actual external quality factor Qe is related

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

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).

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

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

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.

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.

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.

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

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.

Parameter Final Value (mm) Parameter Final Value (mm)

L1 11.935 L6 13.365

W1 4 W6 2

D1 1 D6 1

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

W3 2 L9 14.152

D3 1 W9 2

S 34 3.341 D9 1

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

Wiley & Sons, Inc.,NY, 2011.

[2] R. Collin, Foundation for Microwave Engineering, 2nd edition, IEEE Press,

Wiley, New York, 1992.

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.

Methodologies for improvred characteristics, communcation system group,"

India EESOF User Group Meeting, November 2005.

Resonators for Cross-Coupled Planar Microwave Filters," IEEE Transactions

on Microwave Theory and Techniques, vol. 44, no.12, December 1996, pp.

2099-2109.

[8] I. Bahl, Lumped Elements for RF and Microwave Circuits, Artech House, Inc.,

Boston, 2003.

IEE Proceedings Microwaves, Antennas and Propagation, vol.147, no.5,

October. 2000, pp. 354-358.

96

Chapter 5

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.

diplexer using optimization is based on minimization of a cost function that is evaluated

at frequency locations of reflection zeros.

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.

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.

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

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

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

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