Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
By
Hazem M. Abukaresh
Supervisor
Dr
Dr. Talal F. Skaik
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.
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.
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
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
Appendix A .......................................................................................................................... 99
ChebyshevLowpass Prototype Filters ............................................................................... 99
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
viii
Chapter 1
Introduction
1
The main 2G standards are Interim standard 95 (IS-95) in USA, Personal Digital
Cellular (PDC) in Japan, IS-136 (Digital AMPS) in USA and Global System for Mobile
Communications (GSM) which has been specified and developed by many European
countries working in co-operation with each other as illustrated in Figure 1-1 [1].
Figure 1-1: illustrated many European countries working in co-operation with each other to
develop GSM system [3].
2
2.5G systems address the data capacity limitations correlating with the 2nd
generation systems. GSM introduced General Packet Radio Service (GPRS) which can
be used for services like Wireless Application Protocol (WAP) access, Multimedia
Messaging Service (MMS) and for internet services including email and World Wide
Web access [4]. IS-95 is upgrade to IS-95B which provides users with an efficient
mechanism for wireless internet browsing and email access [1]. NTT DoCoMo (Nippon
Telegraph and Telephone DoCoMo) developed their own proprietary packet based
technology called i-mode on its PDC network which supports games, color, graphics
and interactive web page [1]. IS-136 is developed to IS-136 HS.
During the transition from the second generation to the third generation a 2.75G
has appeared which is considered as a pre 3G that allows GSM operators to use existing
GSM radio bands to offer wireless multimedia IP-based services and applications [5].
In this generation GPRS network evolved to Enhanced Data Rates for GSM Evolution
(EDGE) networks which improved data transmission rates and services [5].
Third Generation (3G) Technology was developed in order to face up to the new
requirements of services of what were coming, as high-quality images and videos or to
provide access to the Web with higher data rates. The data-handling capabilities of 2G
systems are limited and are necessary for other mobile technology. 3G systems provide
a platform that is common for multiple wireless standards and technologies [6]. 3G
systems supported technologies is to create a single global standard that allows for
global roaming. The five 3G standards included in IMT-2000 terrestrial radio interfaces
are: The IMT-2000 defined highlights five distinct mobile/terrestrial radio interface
standards [7,8]:
1- IMT-MC: CDMA Multi-Carrier (known as cdma2000 or IS-2000)
2- IMT-DS: CDMA Direct Spread (known as Wideband Universal Mobile
Telecommunications System (UMTS) or WCDMA-FDD)
3- IMT-TC: CDMA TDD (WCDMA-TDD)
4- IMT-SC: TDMA Single Carrier (known as UWC-136 and EDGE)
5- IMT-FT: TDMA Multi-Carrier (well known as DECT, Digital Enhanced
Cordless Telecommunication)
3
Figure 1-2 show the evolution of mobile cellular standards until 3G which include
the above standards. CDMA 2000 3X Radio Transmission Technology (3xRTT) is
upgrade to Evolution for Data Optimized (EV-DO) [9,10].
Figure 1-2: Evolution towards third generation mobile systems in terms of data rate support [1,7].
3.75G 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].
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.
8
Figure 1-5: transceiver configuration example of multi-mode/multi-band [22].
9
1.4 Overview of Diplexers and their Application
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].
The objective of this research is to analyze and design microstrip diplexer. The
design is based on synthesis of coupling matrix of a 3-port coupled resonator circuit
using optimization. Diplexer meets the specification of LTE-Advanced system,
specifically band number 7 (i.e. UL: 2.50–2.57 GHz and DL: 2.62–2.69 GHz), and it is
to be used inside base transceiver station. Microstrp diplexer consists of 10 “U” shape
resonators that provide very small insertion loss and good isolation between the two
ports. The size of the proposed diplexer may be miniaturized in comparison to
conventional diplexers since a novel method presented in [29,30] will be used in the
synthesis.
11
Chapter 1 introduces cellular system evolution from 1980-2012 and beyond,
presents overview of LTE-Advanced system and its bands the licensing of the 2.6 GHz
spectrum band and assess the viability and implications of the three ITU band plan
Options. Finally the chapter provides a general overview about diplexers and their
applications.
Chapter 2 explains the conventional diplexers synthesis and the different types
of external junctions for distribution of energy. It discusses the theory of coupling, the
calculation of coupling coefficients between two resonators, and the calculation of
external coupling coefficients.
Chapter 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.
[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.
[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
[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.
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
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].
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].
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.
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.
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
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].
26
Figure 2-15: Diplexer with an extra resonator as common junction [35].
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
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].
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
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].
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
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
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 )
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
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
a2 = 0 b2 = v x Gx (2.21)
a2 = 0 b3 = v y G y
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
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.
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].
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
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.
[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.
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.
[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.
[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
[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 .
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.
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.
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.
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.
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].
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
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.
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 .
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
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
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
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
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.
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
[6] S. Rao, Engineering Optimization Theory and Practice. 3rd Edition, John
Wiley & Sons, Inc. 2009
75
Chapter 4
Diplexer Design
4.1 Introduction
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
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
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
300
λg = mm (4.11)
f (GHz ) ε re
2π
β= (4.12)
λg
78
4.2.2 Resonators
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
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.
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
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
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).
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.
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.
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.
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
[2] R. Collin, Foundation for Microwave Engineering, 2nd edition, IEEE Press,
Wiley, New York, 1992.
[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.
[8] I. Bahl, Lumped Elements for RF and Microwave Circuits, Artech House, Inc.,
Boston, 2003.
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.
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.
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
101