high-frequency filtering - ulisboa · o objetivo desta tese é o desenvolvimento de dois filtros de...
TRANSCRIPT
High-Frequency Filtering
Miguel Coutinho Rodrigues Gurgo e Cirne
Thesis to obtain the Master of Science Degree in
Electrical and Computer Engineering
Supervisor: Prof. José António Beltran Gerald
Examination Committee
Chairperson: Prof. João Manuel Torres Caldinhas Simões Vaz
Supervisor: Prof. José António Beltran Gerald
Member of the Committee: Prof. Maria João Ramos Marques Coelho Carrilho do
Rosário
October 2014
ii
iii
To the ones I love
“This is not the end,
Not even the beginning of the end,
But the end of the beginning.”
- Sir Winston Churchill
iv
v
Acknowledgments
Acknowledgements
This thesis represents the result of my endeavors over the past years. This time was very fruitful,
particularly regarding personal experiences. The outcome of my work and personal formation would not
have been the same without the contribution of other people who, in one way or another, assisted me
throughout this journey and whom I would like to thank.
First, I want to acknowledge Professor José Gerald, my supervisor, who accepted me as his student,
giving me his support to carry out this dissertation in such an interesting field, despite the difficulties that
we expected in designing filters with such stringent requirements.
Professor Moisés Piedade, for his endless knowledge and the kindness of sharing it, contributing to
enhance the performance of the simulation results and leading me to establish an informal partnership
with AMRAD - Associação Portuguesa de Amadores de Rádio para a Investigação Educação e
Desenvolvimento.
Eng. Carlos Gorjão, an AMRAD associate, for sharing his experience with cavity filters, kindly providing
one operating filter, so that a proper simulation model could be constructed. Moreover, for his patience,
following all of the tuning process of the custom made filter that was designed.
My friends at Instituto Superior Técnico, for the endless good moments we collected, (we will never
remember all of them for being so many) hoping that many more are still to come. I met unique people
and some of them deserve a particular mention.
Silvério, meet him at day zero of the course was like a shot in the dark, but also the beginning of a deep
friendship that I am sure that will last a lifetime. Fernandes, who accompanied me in several classes,
works and reports, particularly during the last couple of years, and who I have the pleasure to share the
passion for fencing. Chico, who helped me with all of the programming and difficult disciplines of the
first years. Tavares, for always having a friendly word (and organizing such amazing events). Sérgio,
for lending me one of his house sofas during study nights and for pushing my patience boundaries to
the limit but, at the same time, making me embrace different work processes. Tomás, for his friendship
during class days and vacations; thank you all!
A special word to Diana, for her love, strength and unconditional support no matter what the distance or
the circumstances.
And finally, I would like to thank my family, my mother, my father and my brother, who made who I am,
for always being by my side, supporting me at any moment, during the last 23 years.
Pais, é a vocês que agradeço por ter alcançado tudo o que consegui e ainda sonho alcançar. É por
isso que vos dedico, com todo o orgulho, esta tese.
vi
vii
Abstract
Abstract
The most important objective of this thesis is the development of high frequency filters, whose structure
is designed according to the specifications of two real-life applications that are being affected by
interference. In the first case, LTE band 38, seen as a future high-capacity resource for the commonly
deployed FDD-LTE networks, faces the exhaustive use of the spectrum, making it difficult to avoid
interference from adjacent channels. In the second case, the installation of an amateur television
repeater jeopardizes the operation of a very narrow band moon bounce communicating service. Thus,
highly selective filtering is a common need and suitable filtering structures have to be designed.
For this purpose, after an analysis of the available solutions, combline cavity filters were chosen and a
relation between the structural dimensions of the filter and the ideal filter response is presented, followed
by the obtained results. Then, besides the simulation software validation with a real combline cavity filter
testing, a solution for each of the applications is presented.
For the LTE filter, an after-tuning insertion loss of about 25 dB, in the best obtained case, makes it
unusable, being it the reason to consider another set of specifications. In the amateur radio filter a near
flat response followed by 50 dB of attenuation in 30 MHz, makes the proposed structure a good
approximation of the required specifications.
Keywords
LTE, amateur radio, RF interference, combline filter, cavity filter
viii
ix
Resumo
Resumo
O objetivo desta tese é o desenvolvimento de dois filtros de alta frequência, cuja estrutura é desenhada
tendo em conta as especificações de duas aplicações que são atualmente afetadas por interferências.
Primeiro, a banda 38 do sistema LTE, vista como um recurso de alta capacidade para o futuro, depara-
se com a exploração exaustiva do espectro, existindo interferências de canais adjacentes. Segundo, a
instalação de um repetidor do serviço de televisão amador, põe em risco a operação do sistema de
reflexão lunar, este que opera com uma largura de banda bastante estreita. Assim, são necessários
filtros extremamente seletivos, tendo de ser desenvolvidas estruturas apropriadas para o efeito.
Depois de uma análise das soluções disponíveis, foram escolhidos os filtros de cavidade com
ressoadores alinhados como um pente (combline), sendo apresentadas as relações entre as suas
dimensões e as especificações de filtragem. Depois de apresentados os resultados do
dimensionamento dos filtros, e da validação do programa de simulação com recurso a um filtro real,
são indicadas soluções específicas para cada uma das aplicações.
Para o filtro de LTE, e após as operações de sintonia, foram registadas perdas de inserção a rondar os
25 dB, impossibilitando a sua viabilidade. Desta forma, um filtro com diferentes especificações foi
projetado. Para este, a ser usado num sistema de radioamador, foi obtida uma resposta bastante
próxima do pretendido, com uma ondulação mínima na banda de passagem e uma atenuação de 50
dB em 30 MHz, representando uma boa aproximação do especificado.
Palavras-chave
LTE, radioamador, interferência de RF, filtro combline, filtro de cavidade
x
xi
Table of Contents
Table of Contents
Acknowledgments ....................................................................................................................................v
Abstract................................................................................................................................................... vii
Resumo ................................................................................................................................................... ix
Table of Contents .................................................................................................................................... xi
List of Figures ......................................................................................................................................... xv
List of Tables ........................................................................................................................................ xvii
List of Acronyms .................................................................................................................................... xix
List of Symbols ...................................................................................................................................... xxi
1. Introduction .......................................................................................................................................... 1
1.1. Overview .................................................................................................................................. 2
1.2. Motivation ................................................................................................................................ 2
1.3. Objectives ................................................................................................................................ 3
1.4. Main Achievements of This Work ............................................................................................ 3
1.5. Contents .................................................................................................................................. 3
2. Fundamental Concepts and State of the Art .................................................................................... 5
2.1. High Frequency Filters Background ........................................................................................ 6
2.2. Applications ............................................................................................................................. 8
2.2.1. Front-End filter for LTE receiver ...................................................................................... 8
2.2.2. Amateur radio 23 cm band filter .................................................................................... 10
2.3. State of the Art ....................................................................................................................... 11
2.3.1. Combline and cavity filters ............................................................................................. 11
2.3.2. Interference in the TD-LTE system................................................................................ 13
2.3.3. Interference in the amateur radio 23 cm band .............................................................. 14
3. Design Methodology....................................................................................................................... 15
xii
3.1. Combline cavity filters design methodologies ....................................................................... 16
3.2. Combline cavity filters design ................................................................................................ 17
3.2.1. Filter generic specifications ........................................................................................... 18
3.2.2. Combline cavity filters structure ..................................................................................... 19
3.2.3. Combline cavity filters design equations of a Chebyshev approximation ..................... 20
4. Filtering Specifications and Design Results ................................................................................... 27
4.1. LTE Front-End filter ..................................................................................................................... 28
4.1.1 Previous considerations ........................................................................................................ 28
4.1.2 LTE filter specifications ......................................................................................................... 28
4.1.3 LTE combline cavity filter design results ............................................................................... 28
4.1.4 LTE combline cavity filter simulation model .......................................................................... 29
4.2 Amateur radio 23 cm band filter ................................................................................................... 30
4.2.1. Previous considerations ....................................................................................................... 30
4.2.2. Amateur radio 23 cm band filter specifications .................................................................... 31
4.2.3 Amateur radio 23 cm band combline cavity filter design results ........................................... 32
4.2.4 Amateur radio 23 cm band combline cavity filter simulation model ...................................... 32
4.3 Scattering coefficients .................................................................................................................. 33
5. Simulation Results .......................................................................................................................... 35
5.1. Simulation software presentation .......................................................................................... 36
5.2. Simulation parameters ........................................................................................................... 36
5.3. Real combline cavity filter structure measurement................................................................ 36
5.3.1. Real combline cavity filter frequency response ............................................................. 40
5.3.2. Real combline cavity filter simulation results and tuning process ................................. 40
5.3.3. Real combline filter analysis conclusion ........................................................................ 42
5.4. LTE front-end filter simulation................................................................................................ 43
5.4.1. LTE filter initial simulation results .................................................................................. 43
5.4.2. LTE front-end filter tuning process ................................................................................ 44
5.5. Amateur radio 23 cm band filter simulation ........................................................................... 47
5.5.1. Amateur radio 23 cm band filter initial simulation results .............................................. 48
5.5.2. Amateur radio 23 cm band filter tuning process ............................................................ 48
6. Conclusions and Future Work ........................................................................................................ 53
xiii
6.1. Conclusions ........................................................................................................................... 54
6.2. Future work ............................................................................................................................ 55
A. Annex A - LTE combline filter design results ............................................................................. 57
B. Annex B – Amateur radio 23 cm combline filter design results ................................................. 61
References ............................................................................................................................................ 65
xiv
xv
List of Figures
List of Figures
Figure 2.1 Chebyshev band pass filter schematic (extracted from [8]) ................................................... 6
Figure 2.2 Layout of a five-pole, hairpin-line microstrip band pass filter (extracted from [1]) ................. 7
Figure 2.3 Design dimensions of an iris-coupled band pass filter (extracted from [7]) ........................... 8
Figure 2.4 BS-to-BS interference scenario (extracted from [17]) .......................................................... 10
Figure 3.1 General structure of the microstrip equivalent model of a combline band pass filter (extracted
from [39]) ............................................................................................................................................... 16
Figure 3.2 LC equivalent circuit of a 3 resonator combline filter (extracted from [7])............................ 17
Figure 3.3 Generic band pass filter response ....................................................................................... 18
Figure 3.4 Side view of a combline filter block with 3 sections (extracted from [7]) .............................. 19
Figure 3.5 Cross-section of the combline filter block (extracted from [7]) ............................................. 19
Figure 3.6 Resonators and tuning screws positioning inside the cavity filter ........................................ 25
Figure 4.1 LTE ideal filtering specifications ........................................................................................... 28
Figure 4.2 LTE front-end filter simulation model ................................................................................... 30
Figure 4.3 Current frequency spectrum around 1296 MHz ................................................................... 31
Figure 4.4 Filter specifications for the ATV interference on the amateur radio 23 cm band ................. 31
Figure 4.5 Amateur radio 23 cm band filter simulation model ............................................................... 33
Figure 5.1 Real combline filter simulation model .................................................................................. 37
Figure 5.2 SMA female connector to be installed at each side of the cavity block to perform the
input/output connections (extracted from [42]) ...................................................................................... 38
Figure 5.3 Side view of the real combline cavity filter and identification of the main elements dimensions
............................................................................................................................................................... 38
Figure 5.4 Top view of the real combline cavity filter and identification of the main elements dimensions
............................................................................................................................................................... 38
Figure 5.5 Real combline cavity filter frequency response measured by the VNA ............................... 40
Figure 5.6 First simulation result of S12 for the real combline cavity filter model ................................... 40
Figure 5.7 Simulation model results for S12 with the new tuning screws depth ..................................... 41
Figure 5.8 Side view of the LTE front-end filter simulation model ......................................................... 43
Figure 5.9 First simulation results of the S12 parameter for the LTE filter model obtained through the
design equations ................................................................................................................................... 43
Figure 5.10 Detailed response behavior of the first simulation of the LTE filter model obtained through
the design equations ............................................................................................................................. 44
Figure 5.11 Simulation result of the S12 parameter after the first tuning iteration process made to the LTE
filter ........................................................................................................................................................ 44
xvi
Figure 5.12 Detailed simulation result of the S12 parameter after the first tuning iteration process made
to the LTE filter ...................................................................................................................................... 45
Figure 5.13 Final LTE front-end filter simulation model ........................................................................ 45
Figure 5.14 Side view of the final LTE front-end filter simulation model ............................................... 46
Figure 5.15 Top view of the final LTE front-end filter simulation model ................................................ 46
Figure 5.16 Final S12 simulation result of the LTE filter considering the changes from Table 5.7 ......... 46
Figure 5.17 Detailed picture of the simulation result for the S12 parameter from the LTE filter considering
the changes from Table 5.7 ................................................................................................................... 46
Figure 5.18 S11 simulation result of the final LTE filter obtained considering the changes from Table 5.7
............................................................................................................................................................... 47
Figure 5.19 Side view of the designed 23 cm amateur radio filter ........................................................ 47
Figure 5.20 First simulation results of the S12 parameter from the 23 cm amateur radio filter model
obtained through the design equations ................................................................................................. 48
Figure 5.21 Detailed first simulation results of the S12 parameter of the 23 cm amateur radio filter model
obtained through the design equations ................................................................................................. 48
Figure 5.22 Side view of the 23 cm amateur radio filter considering the changes from Tables 5.8 and
5.9 .......................................................................................................................................................... 49
Figure 5.23 Final simulation results of the S12 parameter of the 23 cm amateur radio filter model
considering the changes from Table 5.8 and 5.9 .................................................................................. 50
Figure 5.24 Detailed final simulation results of the S12 parameter of the 23 cm amateur radio filter model
considering the changes from Table 5.8 and 5.9 .................................................................................. 50
Figure 5.25 Final simulation results around 1296 MHz of the S12 parameter from the 23 cm amateur
radio filter model considering the changes from Table 5.8 and 5.9 ...................................................... 51
Figure 5.26 S11 simulation result of the final 23 cm amateur radio filter obtained considering the changes
from Table 5.8 and 5.9 .......................................................................................................................... 51
xvii
List of Tables
List of Tables
Table 3.1 Passband filter response relevant frequency levels .............................................................. 18
Table 3.2 Identification of combline cavity elements ............................................................................. 19
Table 4.1 LTE combline cavity filter main parameters values ............................................................... 29
Table 4.2 Spacing between resonators inside the LTE combline cavity filter ....................................... 29
Table 4.3 Amateur radio 23 cm combline cavity filter main parameters values .................................... 32
Table 4.4 Spacing between resonators inside the amateur radio 23 cm combline cavity filter ............ 32
Table 5.1 Simulation software analysis parameters .............................................................................. 36
Table 5.2 Real combline cavity filter main parameters measured values ............................................. 39
Table 5.3 Real combline cavity filter measured values for spacing between resonators ..................... 39
Table 5.4 Real combline cavity filter measured values for tuning screws depth ................................... 39
Table 5.5 New tuning screws depth inside the cavity filter simulation model ....................................... 41
Table 5.6 Comparison between real measured and simulated values of the tuning screws depth inside
the cavity................................................................................................................................................ 42
Table 5.7 Modifications of the LTE cavity filter parameters .................................................................. 45
Table 5.8 Modifications of the 23 cm amateur radio cavity filter parameters ........................................ 49
Table 5.9 New values of the coupling screws depth inside the 23 cm amateur radio filter .................. 49
Table A.1 Input values of the LTE filter design ..................................................................................... 58
Table A.2 Chebyshev combline filter design results for the LTE application ........................................ 59
Table B.1 Input values of the amateur radio filter design ...................................................................... 62
Table B.2 Chebyshev combline filter design results for the amateur radio application......................... 63
xviii
xix
List of Acronyms
List of Acronyms
2G Second-Generation
3G Third-Generation
3GPP 3rd Generation Partnership Project
4G Fourth-Generation
ACI Adjacent Channel Interference
AM Amplitude Modulation
ANACOM Autoridade Nacional de Comunicações
ATV Amateur Television
BS Base Station
CCI Co-Channel Interference
EME Earth-Moon-Earth
FM Frequency Modulation
FDD Frequency Division Duplex
FDD-LTE Frequency Division Duplex for Long Term Evolution
GNSS Global Navigation Satellite System
GSM Global System for Mobile Communications
HFSSTM High Frequency Structure Simulator
IARU International Amateur Radio Union
IF Intermediate Frequency
ITU International Telecommunication Union
LTE Long Term Evolution
PTFE Polytetrafluoroethylene
RF Radio Frequency
TD-LTE Time-Division for Long Term Evolution
TD-LTE-A Time-Division for Long Term Evolution Advanced
TD-SCDMA Time Division Synchronous Code Division Multiple Access
TDD Time Division Duplex
TEM Transverse Electromagnetic Mode
UMTS Universal Mobile Telecommunications System
VHF Very High Frequency
VNA Vector Network Analyzer
xx
xxi
List of Symbols
List of Symbols Permittivity Relative dielectric constant
θo Electrical length of resonator
λ Wavelength
ω Fractional bandwidth
b Cavity width Lumped capacitance
c Coupling screw depth
d Resonators diameter
dj Modified diameter
dterminals Terminal resonator diameter
e Isolation height
fO Centre frequency
f1 Upper passband edge frequency
f2 Lower passband edge frequency
fU Upper rejection frequency
fL Lower rejection frequency
h Cavity height
L Resonators height Attenuation at the upper rejection frequency
Lterminals Terminal resonator height Passband ripple
l Cavity length
n Number of filter sections
p Input/output coupling probe height
s Spacing between resonators
t Rectangular resonator thickness
W Rectangular resonator width
YA Characteristic admittance
xxii
1
1. Introduction
Chapter 1
Introduction
This chapter gives a brief overview of the environment and relevant topics of this thesis work. The
motivation and objectives are presented, followed by the detailed presentation of the thesis structure.
2
1.1. Overview
Radio spectrum is nowadays considered one of the most important resources, as the need for available
frequency bands follows the ever growing development of wireless technologies. The regulation of
consigned frequencies is imposed by national governing bodies, with the International
Telecommunication Union (ITU) playing a worldwide coordinator role.
Microwaves may be used to describe electromagnetic waves with frequencies ranging from 300 MHz to
300 GHz [1]. The microwave applications can be referred to as the electronic systems that explore the
usage of frequency spectrum in that frequency range, whether they are used for communications, radar
systems, satellite links or amateur radio operations, among others.
In the operation of these systems, additional frequency signals are generated outside the band of
interest for various reasons, such as interference or nonlinearity of the components. These additional
signals are unwanted as they affect the satisfactory functioning of the equipments and may cause
unexpected malfunctions to neighbor systems.
Particularly focusing on communication systems, filters play an important role since the early years, as
highly selective filtering devices became essential, being the cellular phones one of the biggest
application domains. Here, filters are needed between the receiving antenna and preamplifier to select
the signals from the correct band in order to be amplified, as well as in the transmit path, as regulations
forbid the emission of power outside the licensed band [2].
Emerging applications, such as the latest 4th generation (4G) of mobile networks or the spectrum
overcrowd of applications, continue to challenge the filters domain with ever more stringent
requirements: higher performance, smaller size, lighter weight, and lower cost.
Generally, microwave band pass filters have been indispensable devices of modern microwave
communication systems and their performance will directly affect the quality of communication between
final users, whether they are humans or machines. As one of these microwave filters, combline filters
are broadly used as band pass filters in modern microwave systems. Their compactness, excellent stop
band, selectivity performance and ease of integration [3], make them a viable option when high rejection
levels are required.
1.2. Motivation
With recent developments in mobile networks, satellite links, digital TV broadcasting and other forms of
wireless data communication, such as WLAN or Bluetooth, frequency control becomes a key issue since
a lot of systems are crowding the available frequency bands.
In addition to that, the deployment of the LTE mobile telecommunication system, commonly recognized
as the 4th generation network in many countries, including Portugal, also represented a motivation point,
as the exhaustive use of the spectrum makes it difficult to avoid interferences from adjacent channels.
In other words, and being more specific, every TD-LTE communication channel requires a fairly high
isolation from adjacent frequency bands, which can either belong to the same operator or another.
This issue was first addressed by Fragoso, in [4]. In his work, an all-digital filter solution was studied,
with particular emphasis in developing a universal solution that was able to accommodate any of the
3
TD-LTE assigned frequency bands. Most of this filtering specification, which will be presented in Section
4.1.2, stay the same in this work, only the filter technology will now be completely analog.
1.3. Objectives
The objective of this thesis is the development of resonant cavity filters that suit the desired
specifications of a system, i.e., to modulate the dimensions of a structure whose frequency response
has their characteristics, such as central frequency, bandwidth, and selectivity, as close as possible to
the ideal behavior.
These dimensions are going to be obtained from the ideal filter specifications through a set of pre-
established equations, which can be found in the related literature, meaning that the method can be
applied to any frequency band.
The frequency response simulation and test of additional adjustments introduced into the structure will
be done in Ansoft’s High Frequency Structure Simulator (HFSS) [5] that uses the Finite Element Method1
to model electromagnetic fields of three dimensional structures. By designing and defining the
construction material of all the cavity elements, a model is generated. Then a numerical mesh of the
geometric model is generated and a local function will represent the electromagnetic field in each
element. Maxwell’s equations are then transformed into matrix equations and solved by traditional
numerical methods, allowing the analysis of a variety of parameters.
1.4. Main Achievements of This Work
The main achievements of this work were:
• Design of two combline cavity filters structural dimensions that suit the specific filtering
specifications of two applications;
• Simulation and tuning of the above mentioned filters, with the introduction of some changes to
the improve their frequency behavior;
• Evidence of trustable simulation results through the simulation of a real combline filter.
1.5. Contents
This thesis is organized in six chapters, each one highlighting a specific part of the work, from the
introduction of high frequency filters to the analysis of the final results.
The present chapter presents a general presentation of the environment where the work carried out in
this thesis belongs and a brief overview over the motivation and objectives of this thesis.
A brief presentation about some of the most relevant high frequency filtering technologies is given in
Chapter 2, with some examples being highlighted. Moreover, both studied applications are introduced
and a state of the art of the main addressed topics of this thesis is presented.
1 The finite element method (FEM) is a numerical technique for finding approximate solutions to boundary value problems for partial differential equations. It uses variational methods to minimize an error function and produce a stable solution.
4
In Chapter 3, a complete presentation of the design methodology of a combline cavity filter that performs
a Chebyshev response is given, including all of the required equations and expressions. For a better
understanding, the structural organization of a combline filter is also presented, together with an ideal
filter response, which can be used to define any wanted filter specifications.
Chapter 4 presents the filtering specifications of the two studied applications, as well as the
corresponding combline filter dimensioning results. Based on these results, a preview of the simulation
models is also given. Once the scattering parameters will be the guide line for the simulation results
analysis, a brief introduction of the required parameters is also presented.
A crucial part of this thesis work is presented in Chapter 5. In the first section of this chapter the
simulation software validation is performed. This validation was done with a real combline cavity filter
that was measured and tested with a VNA and later simulated, in the chosen software. Then, simulation
results of the designed filters are presented and an individual analysis, including tuning operations, is
performed to each of the applications specific filter. The introduced changes are highlighted and its
beneficial effects are presented.
Finally, Chapter 6 concludes the report, recapping the most important conclusions from the simulation
results and suggesting future works possibilities.
5
2. Fundamental Concepts and State of the Art
Chapter 2
Fundamental Concepts and
State of the Art
This chapter provides some fundamental concepts associated with high frequency filters as well as the
applications that are going to be studied over the next chapters. To conclude, a state of the art of the
addressed topics of this thesis is presented.
6
2.1. High Frequency Filters Background
Filters can perform their functions in several frequency levels. In applications where the filtering stages
are located after the carrier frequency is mixed with a local oscillator, resulting in a down conversion of
the signal that is going to be processed afterwards, filters are classified as of Intermediate Frequency
(IF) filters. One of the main reasons to apply down conversion is to improve filters selectivity. Despite
the filters that are going to be studied in this dissertation could benefit from this operation, this technique
was not adopted, as it has already been used in a previous related work, performed by Fragoso, in [4].
On the other hand, when a filter is directly applied as a part of the front-end equipment, without any type
of signal processing between electromagnetic receivers (commonly antennas) and the filter itself, they
are called of Radio Frequency (RF) filters. This classification may well be interpreted as high frequency
filtering, since this RF acronym is used when referring to frequency oscillations between 3 kHz and 300
GHz [6].
The classification of filter types may differ according to the author or even with the publication focus.
The classification may be done according to the response function, support technology, application
domain or even according to them being digital or analogical.
In fact, grouping filters in terms of response functions ends up being a common situation regardless of
other distinctions that may exist. Low pass, high pass, band pass, band reject and all pass are the filters
more usual types.
For this thesis purpose, and since the filters structural design that is going to be presented follows the
same publication, band pass filter topologies are here divided according to the work done by Natarajan,
in [7], which is the basis for the following section:
• Lumped and semi-lumped RF filters;
• Microstrip and Stripline filters;
• Microwave cavity filters.
In one hand, filters that use discrete inductors and capacitors are called lumped RF filters. An example
of such filters is given in Figure 2.1, where a schematic circuit is represented.
Figure 2.1 Chebyshev band pass filter schematic (extracted from [8])
With the advancements in chip ceramic capacitors, lumped RF filters are designed up to 100 MHz. In
lumped RF filters, inductors and capacitors are networked in the form of series and shunt sections.
On the other hand, in semi-lumped RF filters, discrete capacitors are substituted by distributed
capacitance, meaning that the required capacitance is developed in the design of the filters instead of
using a discrete capacitor.
7
Substrate filter is the generic name for microstrip and stripline filters. Although some substrate filters
could have higher losses than cavity filters, they are widely used, as they are highly compact in size.
The designed filter pattern is printed on a substrate, which is a dielectric material. This type of design
has a particular interest for mobile devices, where the size of the final product imposes reduced
implementation areas for filters. An example of a microstrip filter layout is given in Figure 2.2, as follows.
Figure 2.2 Layout of a five-pole, hairpin-line microstrip band pass filter (extracted from [1])
Finally, microwave cavity filters are mainly characterized by the metallic enclosure inside of which the
resonator elements are mounted, having one side connected to the block and the other end free.
Cavities can have the shape of a cylinder but could also have a rectangular (sometimes also square)
cross section, both having resonating elements inside. Regarding the resonators, they normally have a
cylindrical shape and the necessary distributed capacitance is developed at their free end, with the help
of a screw.
Cavity filter topologies are often analyzed in terms of sections. These sections are no more than the
number of resonating elements that are mounted inside a cavity block, which develop an inductive
coupling between them and the already mentioned distributed capacitance. This number is also related
with the structure complexity and filtering specifications, as more stringent requirements lead to the need
of more installed resonators.
To provide a connection to the rest of the system devices, the end cavity sections are coupled to coaxial
connectors.
According to an article from three of the most influent authors in the microwave filters domain [9], Levy,
Snyder and Matthaei, microwave filters can be further divided into 8 types or topologies: interdigital,
parallel-coupled, ceramic resonators, suspended substrate stripline designs, waveguide, coaxial,
evanescent-mode and combline filters. Nevertheless, Natarajan, in [7], divides the microwave cavity
filters into combline and iris-coupled filters (Figure 2.3), this later being no more than a composition of
cylindrical sections and resonators, coupled by irises instead of screws.
8
Figure 2.3 Design dimensions of an iris-coupled band pass filter (extracted from [7])
Iris-coupled filters use λ/4 resonators instead of the λ/8 resonators used in combline filters, with the
wavelength being given, in both types, by the central frequency of the wanted filter response.
The combline cavity filter structure is introduced with further detail in Chapter 3 since this was the chosen
option to perform the filtering operations required by both applications. This option was made since a
straight forward design approach is presented in the literature and a good compromise between
complexity, simulation capability and realization can be achieved.
2.2. Applications
The objective of this work is to develop and study high frequency filters for two particular operating
systems. These systems are now briefly introduced and the scope of application is presented.
2.2.1. Front-End filter for LTE receiver
Representing something that we today take as granted, mobile communication networks are nowadays
coping with an increasingly higher number of connected users, with comprehensively impact on the
amount of carried traffic, which is growing exponentially and that has no prediction of slowing down.
After the 2nd generation (2G) of mobile telecommunication systems, also known as Global System for
Mobile Communications (GSM), worldwide deployment, whose capabilities introduced the first data
services to users in parallel with the core objective of providing voice and texting services to everyone,
everywhere, it was the introduction of 3rd generation (3G) networks, technically known as Universal
Mobile Telecommunications System (UMTS), that exponentially boosted the demand for data services.
It was this new mobile communication standard that definitely marked the transition of mobile networks
from voice-dominated to data-dominated traffic networks [10].
The Long Term Evolution (LTE) system, or what we recognize today as the 4th generation of mobile
telecommunication systems, started its history in 2004, when the standardization organization 3rd
Generation Partnership Project (3GPP) defined its initial targets. The evolution of 3G networks into 4G
has been driven by the appearance and development of new massif services for mobile devices and
was enabled by the advancement of the available technology. Furthermore, the market in which mobile
systems are deployed and operate also suffered an evolution throughout the years, especially in terms
of competition between mobile operators and challenges from other mobile technologies [11].
9
LTE subscriptions are globally growing strongly, with around 40 million added in the second quarter of
2014, becoming a total universe of 280 million LTE subscriptions [12].
It is important to note that this history of mobile networks represent a time frame of about 25 years,
considering that the first commercially 2G mobile systems became available around 1991 [13].
LTE has 2 different multiple access techniques: Frequency Division Duplex, which applied to LTE is
commonly called FDD-LTE, and in which there are different allocated frequency bands for up and
downlink channels; and TDD-LTE (that is also seen referenced as TD-LTE), short for Time Division
Duplex applied to LTE, that separates up and downlink channels into different time-slots. Each access
mode has its own allocated frequency bands, imposed by 3GPP.
This option for LTE operation (time division duplex) is nowadays available, as of data from July 2014, in
41 commercially available networks (17 of them in Asia, as the continent with the biggest number), in
25 countries, with an additional 76 networks planned [14]. Comparing those numbers with the total of
291 FDD-LTE networks deployed (with Europe leading the distribution with 107 networks) in 105
countries, we can see that FDD-LTE took the leading spot in the solution option for LTE installation
worldwide, as many operators see TD-LTE only as a future option for increasing internet access speed
[15].
Being a recently established system, LTE encountered its frequency slots among a lot of pre-established
networks. Following the trend of the rest of the European counterparts, Portuguese operators got the
right to install and operate their LTE networks on the three available FDD-LTE bands: the 800 MHz,
1800 MHz and 2600 MHz bands, with Vodafone being the only one that has purchased a TD-LTE
reserved frequency, in the 2600 MHz band.
In Portugal’s case, ANACOM is responsible for spectrum management and supervision, and was the
entity responsible for the LTE bands public auction, promoted during November 2011 and whose final
report [14] was presented in January 2012.
Knowing that interoperability should be guaranteed and frequency spectrum is scarce, every allocated
frequency band cannot be re-allocated or used until the system that runs on top of it is turned off. It is
also obvious that national spectrum management organizations make an effort to accommodate as
many systems as possible, yet guaranteeing minimal interference.
Since the allocated LTE bands have no frequency gaps in between them, it is of the operator’s concern
to isolate their allocated bands from neighbor interferers as well as preventing an emission outside their
system’s reserved frequency bands, which can be subject to penalizations by the governing body. This
is saying that there is a big interest, and growing need, for front-end filters that protect receivers, namely
base-station antennas, from adjacent channel interference (ACI) and output filters that limit the
bandwidth of transmitter noise [18] and prevent emission of signals outside the operator’s reserved
frequency band. Additionally, co-channel interference (CCI) also occurs. In this case, channels from the
same operator interfere between them and it is something that also needs to be prevented. With so
many limitations to be considered, it is comprehensible that filters for both communication paths tend to
follow more and more stringent requisites, including better performance, miniaturization and reduced
costs.
10
Figure 2.4 BS-to-BS interference scenario (extracted from [17])
The isolation of channels in the TD-LTE 2600 MHz band is a work proposed by a Portuguese mobile
operator, initially requiring a 80 dB rejection at each side of the pass band. It was first addressed by
Fragoso in [4], in which by implementing a down conversion, a digital filtering of the signal followed by
an up conversion back to the original frequency, reached an attenuation that was very close to the
desired one. However, the total amount of signal processing and its very difficult digital implementation,
with many additional filters required, made the solution quite complex.
Thus, in this thesis, a different high frequency filtering technology is used to perform the same task. This
option was considered as resonant cavity filtering has already proven to be useful in this particular
application area [18, 26, 27].
2.2.2. Amateur radio 23 cm band filter
The last few years of the 19th century set the stage for rapid development of commercial radio and
established the foundations of amateur radio. The works of Ørsted, Ampere, Faraday, Henry, Hertz as
well as the famous Maxwell, set the physical foundations that allowed the exploitation of the frequency
spectrum for communication purposes. But it is with the work and experiments of Guglielmo Marconi
that the real radio usage started. His patent submission and production, in 1896, of a radio apparatus
working with long waves to communicate at large distances – the wireless telegraph – and in 1901,
when he demonstrated that a trans-Atlantic communication (using Morse code) was possible, were the
marks from which radio communications really started to flourish.
With a growing interest in the area, and as governments realized that many applications would benefit
from wireless communicating technologies, interferences had to be reduced to minimal values. With that
in mind, amateur radio operators started to require a license to operate and be obliged to follow stringent
rules.
Amateur radio operations stayed in the low frequencies through a lot of the early years, and after the
World Wars, radio enthusiasts started using new frequency bands such as the 3.5 MHz, 30 MHz as well
as the 6 and 2 meter bands [19]. Note that it is common, in this domain, to use the wavelength
nomenclature in submultiples of meter instead of the equivalent frequency in Hertz.
By the 60’s, the first two-way contact via the moon was made in the United States by the Rhododendron
Swamp VHF Society in Massachusetts to the Eimac Radio Club in California, using one of the most
significant amateur radio bands: the 23 cm band, approximately at 1296 MHz.
11
In Portugal’s case, amateur radio comes to light with the development and deployment of wireless
telegraphy around 1925, when José Joaquim de Sousa Dias Melo installed one of the first experimental
transmitting stations in Lisbon [20].
Besides Morse code and voice conversation, amateur radio enthusiasts also have television
broadcasting services, also known as Amateur Television (ATV), that follow the same rules as the rest
of the amateur radio activity, what means that it is also limited to certain frequency bands. The vast
majority of ATV activity is on three amateur-reserved frequencies: the 70 cm, 33 cm and the 23 cm
band.
Modern Amateur Television uses a transmission format fully compatible with video equipment designed
for the individual consumer market. For transmitting, video is amplitude modulated and audio is
frequency modulated (though there is a version where both are frequency modulated: FM-ATV),
meaning that a ATV picture displays full motion, has a simultaneous sound channel, is usually in color,
and has excellent detail just like commercial television [21].
Focusing particularly on this 23 cm band, and by analyzing the IARU – Region 1 (where Portugal is
included) band plan, it is clearly visible that a lot of systems are supported by this frequency band:
packet radio, FM, AM and digital ATV channels, amateur satellite service, earth-moon-earth (also known
as moon bounce) links and continuous-wave, among others [22].
Our attention in this thesis will be on the moon bounce reserved frequency band (1296 MHz – 1296.15
MHz) and on the ATV system, as this latter service is the cause of interference in an AMRAD associate
moon bounce communicating system. This is a problem that can be solved with a filter installation and,
since AMRAD has some expertise in cavity filters, a partnership was established to understand these
filters construction and beneficiate of their knowledge to design filtering structures that can suit the
studied applications.
Thus, besides the filtering structure for the LTE mobile communication system, the same design
procedure will be applied for this application.
2.3. State of the Art
This thesis addresses the design, simulation and study of resonant cavity filters, for two particular
applications. Both filters are meant to be installed as front-end equipments, helping in the rejection of
neighbor signals.
Generically, all filter technologies can be designed and later finely tuned to the required frequency
response characteristics. Still, it is important to notice that there are still some limitations inherent to
each of the technologies which can limit, for example, some characteristics of the frequency response
or insertion losses.
2.3.1. Combline and cavity filters
The chosen option to directly perform band pass filtering at high frequencies is the combline microwave
cavity filter. The combline filter, which was first introduced by Matthaei et al., in [3], is one of the most
commonly used band pass filters in many communication systems and other microwave applications.
12
This technology, according to Boukari et al., in [23], remains the unchallenged technology in satellite
communications, where dual and multimode microwave cavity band pass filters come very close to
satisfying the often contradictory requirements of good selectivity, low in-band insertion loss, linear
phase, small size and light weight.
Being a widely used technology for base station’s band pass filters, the work from Hesselbarth, in [24],
states that for base station diplexers, filters based on cavity resonators are often chosen because of
their low dissipation loss and high quality factor.
In [25], Höft et al., present a base-station band pass filter using compact stepped combline resonators.
The band pass filter consists of 4 resonators, has a center frequency of 2.0175GHz, a bandwidth of
15MHz and cross-coupling by a cascaded quadruplet for improved blocking performance. The objective
is the application in base-stations for TD-SCDMA mobile communication systems, which is declared as
Chinese 3G standard. It shows a 50 dB attenuation level at a bandwidth of 75 MHz.
Also regarding this communication standard, in [26], Wang et al., present a narrowband band pass
combline filter with coaxial input. A novel configuration using identical posts with capacitance-loading
and symmetrical structure is introduced. By optimizing the offsets between resonators to achieve the
best coupling, the design procedure of the combline filter can be simplified. Moreover, fabrication cost
can also be lowered, since the identical and symmetrical structure is suitable for wire-cutting process.
Simulated results show that with five resonators centered at 2.0175 GHz, return loss lower than -18 dB
can be achieved over a relative bandwidth of 0.75%. Tested results from the sample filter have also
been given, which agree with the simulated results.
In [27], Yangping et al., present an 8th order cavity filter with two symmetrical transmission zeros in the
stop band. Generalized Chebyshev synthesis method is used to satisfy the IMT-Advanced system
demands. A co-simulation with Ansoft HFSS is presented. The effectiveness of the co-simulation
method is validated by the excellent consistency between the simulation and the experiment results. In
fact, the specifications of a frequency range between 1850 MHz and 2025 MHz, an insertion loss of 0.6
dB, a pass band ripple of 0.2 dB, a return loss below 20 dB and a stop band rejection of 20 dBc at a 10
MHz distance from central frequency are accomplished by using a CQ topology structure, which
produces two symmetrical transmission zeros.
Inside this very topology there are also cases where novel structures have been presented. An example
is given in the work developed by Höft and Yousif, in [28], where two different band pass filter topologies
are presented. The first one has an inline arrangement of six orthogonal coaxial cavity resonators. Its
distributed cross-coupling consists of four interlaced triplets, which generate two transmission zeros
above the pass band. The second filter design has a folded arrangement of two inline triplets with strong
cross-coupling. Measurement results of the two corresponding filter prototypes confirm the predicted
performance, with the inline arrangement having a 100 MHz passband and a 60 dB attenuation level at
a 200 MHz bandwidth and the orthogonal arrangement having an 80 MHz passband and a 50 dB
attenuation level making a 200 MHz bandwidth.
Working at a center frequency of 2.5 GHz, in [29], Zakaria et al., present a low-loss coaxial cavity band
pass filter with post-manufacturing tuning capabilities with driving screws. The coaxial cavity filter based
upon TEM mode of propagation has a bandwidth of 168 MHz in the 3 dB attenuation level. The author
13
adds that this type of microwave filter would be useful in any microwave systems where low insertion
loss and high selectivity are crucial, such as in a base station, radar and satellite transceivers. The
approach of this work is very similar to the one that is going to be explained in the present thesis, and
served as a basis for choosing this topology to perform the high frequency filtering addressed in the
present thesis. Additionally, this work from Zakaria et al., demonstrates that a good approximation is
obtained between filter specifications and physical dimensions of the cavity, as well as an accurate
simulation when compared to the prototype frequency response results.
Since one of the applications here studied would be a front-end filter of a TD-LTE base-station, it is also
relevant to make notice of the work developed by Chu et al., in [30]. Here the design of the RF front-end
transceiver of a base-station for 2600 MHz TD-LTE-A communication, with bandwidth of 100 MHz is
provided. Tested with real modulated signals, it shows good accuracy of modulation and demodulation.
The study of configurability of combline filters is also of great interest since a multi-standard product can
well have much more success in the market than a fixed solution. In [31], an article presented by Kwak
et al., a combline structure is studied to see if it can be used for a reconfigurable filter that covers a
frequency range of 2.0 GHz to 2.7 GHz and a bandwidth ranging from 50 MHz to 80 MHz. Since the
bandwidth, as well as the center frequency, should vary in a wide range, the coupling between the
load/source and the first/last resonator must be changeable or, in other words, the objective was to
make proof that by exclusively changing the coupling and tuning screws depth inside the cavity the
structure would present different characteristics. Four characteristics were studied: center frequency of
2.15 GHz with a bandwidth of 80 MHz; center frequency of 2.025 GHz with a bandwidth of 50 MHz;
center frequency of 2.65 GHz with a bandwidth of 80 MHz; center frequency of 2.675 GHz with a
bandwidth of 50 MHz. In the end, although the design results do not perfectly meet the specifications,
the amount of the discrepancy is allowable since the filter can be tuned after fabrication.
2.3.2. Interference in the TD-LTE system
In the TD-LTE system, interference problems were already addressed by several authors, mainly from
China, since this is one of the countries where operators are taking this option of time-division for their
LTE networks deployment.
In [32], Tingting et al., present a study of adjacent channel coexistence interference between LTE
systems. It mainly researches macro cell and micro/pico cell coexistence scenario in LTE systems and
it introduces in detail some key technology like network topology, power control and the Adjacent
Channel Interference Ratio model. Using the uncertainty analysis method or system level simulation
method, they conclude that an additional isolation needed in the BS interfering BS case is 40.2 dB. To
realize it they suggest that the first option can be a 5 MHz increase in the frequency protection
bandwidth, which leads to the need of an additional isolation value of 37.9 dB, a reduction of 2.3 dB. As
a second option they point out the installation of a RF filter that would reach 40 dB of attenuation at a
frequency point which is 5 MHz away from the passband.
Also regarding adjacent channel interference, in [33], Lan and Harada, analyze the ACI between two
coexisting systems with different duplex modes, i.e., frequency division multiplexing and time division
multiplexing, deployed in adjacent frequency bands. The coexistence analysis results of the two systems
14
are presented and the macrocell performance degradation caused by ACI, when it operates with
microcell/picocell on adjacent channel in some typical interference scenarios, is evaluated. The focus
of the analysis is on the 2.6 GHz band (2500-2690 MHz). In some countries, there will be at least two
operators that will deploy LTE systems on the same geographical area, possibly with FDD and TDD
respectively. The study conclude that to guarantee that the downlink throughput loss is smaller than an
acceptable 5% decrease, at least about 77 dB and 50 dB Adjacent Channel Interference Ratio are
needed for macro cell & microcell and macro cell & pico cell coexistences, respectively. As a solution
the authors state that the interference limiting techniques include antenna techniques, use of guard
bands, additional front-end filters and channel or deployment restrictions, with all solutions being
associated with an increased level of complexity or lower spectrum efficiency.
2.3.3. Interference in the amateur radio 23 cm band
Publications around this topic are not just related with interference caused by the ATV system, as
addressed in this thesis, but also with the on-going deployment of the European global navigation
satellite system (GNSS): Galileo.
In [34], this topic is presented and discussed. It is stated that the Galileo’s deployment may lead to some
limitations on continuous transmissions such as beacons, television repeaters and FM repeaters below
1300 MHz. They add that the Galileo signal at the earth’s surface will be very weak and spread over a
wide bandwidth, and will only be a source of interference to EME stations with large antennas. As a
typical 23cm EME system uses a large antenna (greater that 3m), the satellite will only be present in the
beam for a short time.
Nevertheless, and to mitigate this possible interference, new frequency allocations are already being
proposed, as exemplified by a publication done by AMSAT-UK, in [35].
Apart from this, amateur radio is an area where experimentalism has accompanied its evolution over
the years. Problems such as interference tend to be common and solutions to mitigate them often come
from this strand of experimentalism. Many radio amateurs create their own products, customized to their
particular needs and operations interest.
Many custom-made products can be found on forums so scientific publications about this topic are
scarce, as many radio amateurs share their knowledge between them and their solutions are inspired
in the commercially available systems.
15
3. Design Methodology
Chapter 3
Design methodology
This chapter presents some design methodologies, including an ideal filtering specification definition
and the combline filter structural organization. To conclude, the combline filter design method used to
define the simulation models is given, with all of the equations and expressions given.
16
3.1. Combline cavity filters design methodologies
Combline filters are constructed by capacitor-loaded resonators, as shown in Figure 3.1. Here
resonators are oriented so that the short circuits are all on one side of the filter (like a comb), and all of
the capacitors at the other side. This gives a great size reduction compared with the conventional
quarter-wavelength resonators. This kind of filter was invented in the Stanford Research Institute in the
1960’s [36].
Working as front-end equipments, microwave filters with cavity resonators, for example combline filters
or interdigital filters, are frequently used for base-station applications because they have a high quality
factor and their simple and cheap manufacture [37].
The main advantages of such devices are their compactness, ease of design, excellent stop band
behavior and high selectivity, besides post-manufacturing tuning capabilities, which sometimes is not
available to other technologies [7].
The design of combline filters has different starting approaches presented in the literature. One of the
most significant is the stripline equivalent model, which is depicted in Figure 3.1 and that was first
presented by Matthaei et al., in [38]. The following explanation was adapted from that publication.
Figure 3.1 General structure of the microstrip equivalent model of a combline band pass filter (extracted from [39])
In Figure 3.1 the resonators consist of line elements which are short-circuited at one end, with a lumped
capacitance between the other end of each resonator line element and ground. Lines 0 and n+1 are
not resonators but simply the input and output coupling connectors.
In combline cavity filters, coupling between resonators is achieved by mean of the fringing fields that
exist between resonator lines. With the lumped capacitors present, the resonator lines will be less
than λ/4 long at resonance. In fact, to make the lines λ/8 long, as wanted for compactness reasons, it is
usually desirable to have large values in the lumped capacitances.
This reduction in the line length also implies a change in the frequency where the unwanted second
passband occurs. The combline cavity filters response present a sequence of resonant frequencies: the
main one, located where the first resonant frequency is located and where the wanted filtering operation
is designed and the other resonant frequencies, located at the subsequent harmonics of the first central
frequency, separated by a constant distance in the frequency spectrum. Thus, when the resonators
equivalent lines have a length given by one eighth of the desired filter central frequency wavelength, the
17
second passband is located somewhat over four times the central frequency of the principal passband,
potentially causing no harm, in terms of interferences, to the system where the filter is going to be
installed.
Note that, with this model, the actual cavity dimensions have to be obtained a posteriori, using equations
that establish equivalences between necessary coupling values and the physical dimensions of the
cavity components. This type of approach is further explained in [37].
Another common equivalent model for combline filter design is the LC circuit. An example of a 3
resonators combline filter is presented in Figure 3.2.
Figure 3.2 LC equivalent circuit of a 3 resonator combline filter (extracted from [7])
Here the resonator inductance and distributed capacitance form the shunt sections of the band pass
filter. Though both inductive and capacitance couplings exist between resonator sections, only an
inductance is shown between the shunt sections, as the predominant coupling between resonator
sections is inductive [25]. Maximum values of the inductive coupling are located at the short-circuited
end of resonators and maximum values of capacitive coupling are measured at the free end of
resonators. It is common to use this approach by starting with a low-pass prototype filter followed by an
adequate mapping function. This type of approach is further explained in [8].
As seen, both approaches require an analysis that is somewhat complex and that is unnecessary when
more practical and simple approaches can be found in the literature. In fact, the used approach is no
more than a practical realization of an enclosed structure whose components will approximate or try to
recreate the capacitance, inductance and line values given by these two approaches.
In other words, the combline cavity filter synthesis mainly consists in determining the coupling values
between resonators and between resonators and the cavity block itself. These couplings are created by
the resonators structural dimensions, the distances that separate them as well as the position they have
inside the cavity block.
A simple and straight forward practical approach is given by Natarajan, in [7]. That synthesis is
presented in Section 3.2.3, with the introduction of some adjustments, for a better understanding.
3.2. Combline cavity filters design
This design process of resonant cavity filters is a direct conversion of a filtering specification into
mechanical dimensions of a cavity block, resonators and position of additional tuning components using
design equations. The publication of Matthaei et al., in [38], provides design equations and graphs for
computing the mechanical dimensions both for maximally flat (also known as Butterworth type) and
18
Chebyshev types of filters. With that publication as basis, a very practical and direct approach is
presented in [7], by Natarajan.
This was the chosen option for synthesis since it does not require the use of equivalent circuits or models
like the ones already presented. The modelling technique here presented helps the designer to produce
a structure which frequency response approximates a desired filter function.
3.2.1. Filter generic specifications
Before start, it is here presented the used nomenclature for band pass filter specifications. A generic
band pass response is represented in Figure 3.3:
Figure 3.3 Generic band pass filter response
For a correct design, it is important to have all of these values defined since they are, together with the
passband ripple in the Chebyshev response approximation, the initial inputs of the design equations. In
Table 3.1, the frequency points from Figure 3.3 are identified:
Parameter Description
fo Centre frequency
f1 Lower passband edge frequency
f2 Upper passband edge frequency
fU Upper rejection frequency
fL Lower rejection frequency
Table 3.1 Passband filter response relevant frequency levels
Note that the difference between the upper rejection frequency (fU) and the upper passband edge
frequency (f2) can be referred to as the rejection band, once this is the bandwidth where the defined
attenuation level shall be guaranteed.
19
3.2.2. Combline cavity filters structure
Graphically, the designed combline filters will have a similar aspect to the one that is depicted in Figures
3.4 and 3.5.
Figure 3.4 Side view of a combline filter block with 3 sections (extracted from [7])
Figure 3.5 Cross-section of the combline filter block (extracted from [7])
In these figures, resonators are still in their rectangular shape, despite their final format is to be
cylindrical. Furthermore, it can be seen in Figure 3.4 the presence of two elements that will be left as
free variables during simulation: the input and output coupling probes, located before the first resonator
and after the last resonator, as we go from left to right in Figure 3.4, and whose dimensions (height and
diameter) variation give some control over the return losses; the coupling screws, located between each
pair of resonators, inserted into the cavity to control the bandwidth of the structure frequency response.
For a better understanding, the cavity structural dimensions are resumed and identified in Table 3.2.
Parameter Description
b Cavity width
L Resonators height
W Rectangular resonator width
t Rectangular resonator thickness
h Cavity height
l Cavity length
s Distance between resonators
Table 3.2 Identification of combline cavity elements
20
3.2.3. Combline cavity filters design equations of a Chebyshev approximation
To transform the frequency response specifications into the mechanical dimensions of the combline
filter block, the practical design equations from [7] were used, leading to a Chebyshev approximation.
This section is based on it, with some nomenclature changes introduced to avoid ambiguity with other
entities commonly used in filters domain.
We start by calculating the number of filter sections, which gives the number of resonators inside the
cavity, given by (3.1):
=cosh
10 − 110 − 1!"cosh #$%$′'(((((((() (3.1)
Where:
• : Attenuation at the upper rejection frequency;
• $%/$′: Suggested low-pass to band pass transformation, as stated in (3.2);
• : Passband ripple.
It is important to make notice that the number of sections must always be an integer number, as it will
define the + index that will repeatedly be used hereinafter.
The suggested low-pass to band pass transformation is given by (3.2):
$%$′ = 2$ × ./01 − .01.01 (3.2)
Where:
• $: Fractional bandwidth, given by (3.3);
• ./: Upper rejection frequency;
• .: Centre frequency.
The fractional bandwidth is given by equation (3.3):
$ = .201 − .01.01 (3.3)
Where:
• .2: Upper passband edge frequency;
• .: Lower passband edge frequency.
This particular design uses one eighth wavelength resonators. This implies that the resonators’ length
is given by (3.4):
33 = 48 × .601 (3.4)
Where:
• 4: Speed of light in vacuum.
21
The resonators diameter is approximated by the mean value of the modified diameters given by equation
(3.5):
733 = 187933
9: (3.5)
Where:
• 79: Modified diameter of the resonator indexed by j.
This modified diameter is given by (3.6), as follows:
7933 = 1.05 × 2= >?933 + A33B,+ = 1AD (3.6)
Where:
• ?9: Width of the j-th rectangular resonator;
• A: Thickness of rectangular resonator.
Note that the modified diameter value is increased by 5%, as it is recommended in [7], in order to
minimize cut-and-try efforts in tuning operations.
The thickness of rectangular resonators is given by equation (3.7) setting b, the cavity’s width, as 14
mm.
A33 = AE × E33 (3.7)
Where:
• A/E: Ratio between rectangular resonators thickness and cavity width, which is set as 0.2.
In this particular approach, this t/b value of 0.2 was set due to the Lagrange interpolations that are going
to be made in equations (3.19) and (3.20).
The width of each rectangular resonator (indexed by letter j) is given by equation (3.8):
?933 = ?9E × E33, + = 1AD (3.8)
Where:
• ?9/E: Normalized width of each rectangular resonator.
The normalized width of each rectangular resonator is given by equation (3.9):
?9E = 12 #1 − AE' F12 #9 ' − >GH%B9,9 − >GH%B9,9I J , + = 1AD (3.9)
Where:
• 9/: Normalized capacitance per unit length between each line and ground, given by the set of
equations (3.10);
• GH%/: Normalized even-mode fringing capacitance, given by the set of equations (3.20).
22
The normalized capacitance per unit length between each line and ground is given by the set of
equations (3.10):
9 =
KLLLLLMLLLLLN
376.7 × RS√ U1 − VWXR Y , + = 0376.7 × RS√ #RZ9R − 1 + WXR − #[,2R ' tan _Z`' + , + = 1376.7 × RS√ #RZ9R − #[9,9R ' tan _Z` − #[9,9IR ' tan _Z`' , + = 2AD − 1376.7 × RS√ #RZ9 R − 1 + WX9R − #[9,9R ' tan _Z`' + 9I , + =
376.7 × RS√ U1 − VWX9R Y , + = + 1
(3.10)
Where:
• R: Characteristic admittance;
• : Relative dielectric constant (commonly air, except in cases where any other type of dielectric
is present between resonators and cavity block);
• WX/R: ratio calculated with equation (3.11);
• RZ9/R: Normalized characteristic admittance;
• [/R: ratio calculated by equation (3.17);
• _: Electrical length of resonator.
Following the approach of Natarajan, in [7], some values are set as constants, as follows: characteristic
admittance R of 0.02 S (given by the inverse of a characteristic impedance of 50 Ω), resonators line
admittance RZ9 of 1/70 S (to ensure a high unloaded quality factor, as recommended in [7]), yielding a
normalized characteristic admittance RZ9/R of 0.714, and an electrical length of the resonator _ of
0.786 radians (once this particular design uses one eighth wavelength resonators, 360 degrees is
divided by 8, yielding 45 degrees of electrical length).
The ratioWX/R: is given by equation (3.11):
WXR = $ aERbcc$′WXR = $ aERbccI$′ (3.11)
Where the ratio E9/R is given by equation (3.12):
E9R = RZ9R dcot _Z` + _ csc2 _Z`2 e , + = 1AD (3.12)
For Chebyshev filters having resistor terminations at both ends and with responses with a pre-defined
passband ripple,c = 1 and$% = 1, the element values may be computed by the following set of
equations (3.13):
23
c = 2fgc6 = 4f6f6E6c6 , .Dij = 2ADcI = k 1,.DiD77coth2 l4 ,.Dimnm
(3.13)
Where the unknown variables are given by equations (3.14), (3.15) and (3.16):
f6 = sin p=q2j − 1r2 s , .Dij = 1AD (3.14)
l = ln#coth `u17.37 ' (3.15)
g = sinh # l2' (3.16)
The ratio [/R is given by equation (3.17):
[9,9IR = $$′ #E9R' #E9IR 'c9c9I , .Di+ = 1AD − 1
(3.17)
Where:
• $′: Normalized prototype parameter in radians, previously set as 1.
The normalized even-mode fringing capacitances GH%/ are obtained from a graph reproduced in [7].
For a better computerization the author provides curve-fitting equations for this graph, obtained by
Lagrange interpolation, with a maximum error of 5%. Those equations are here reproduced in set of
equations (3.19), providing expressions to obtain the spacing between resonators normalized by the
cavity widthv9,9I/E, whose values are used in equations (3.20), which lead to theGH%/ values, as the
main goal is. Note that these equations are fitted for the curve where t/b is 0.2.
The use of these equations implies the calculation of the normalized capacitance per unit length between
adjacent resonators9,9I/, given by equations (3.18):
, = 376.7 × RS√ − 9,9I = #376.7 × RS√ ' #[9,9IR ' tan _Z` ,.Di+ = 2AD − 1,I = 376.7 × RS√ − I (3.18)
The values of the capacitances per unit length between adjacent resonators are now substituted into
the variable x of the set of equations (3.19), obtained by Lagrange interpolation, to computev9,9I/E:
KLMLN51.38qx − 0.105rqx − 0.022r − 133.87qx − 0.195rqx − 0.022r + 104.46qx − 0.195rqx − 0.105r, for0.022 ≤ x < 0.1952.1061qx − 0.375rqx − 0.195r − 9.3897qx − 0.73rqx − 0.195r + 8.3074qx − 0.73rqx − 0.375r, for0.195 ≤ x < 0.730.4105qx − 1.04rqx − 0.73r − 1.7281qx − 1.6rqx − 0.73r + 1.4831qx − 1.6rqx − 1.04r, for0.73 ≤ x < 1.60.0311qx − 2.23rqx − 1.6r − 0.1684qx − 3.55rqx − 1.6r + 0.1628qx − 3.55rqx − 2.23r, for1.6 ≤ x < 3.550.0233qx − 3.9rqx − 3.55r − 0.1538qx − 5.2rqx − 3.55rr + 0.1385qx − 5.2rqx − 3.55r, for3.55 ≤ x < 5.20.4105qx − 1.04rqx − 0.73rr − 1.7281qx − 1.6rqx − 0.73r + 1.4831qx − 1.6rqx − 1.04r, for5.2 ≤ x < 7
(3.19)
The normalized even-mode fringing capacitances GH%/ are now calculated with the set of equations
(3.20), which were also obtained with curve-fitting Lagrange interpolation, by substituting the variable y
by each of the v9,9I/E previously obtained from (3.19):
24
0.395qy − 0.2rqy − 0.4r − 6.278qy − 0.02rqy − 0.4r + 5.3947qy − 0.02rqy − 0.2r, for0.02 ≤ y < 0.42.2778qy − 0.7rqy − 1r − 6.2222qy − 0.4rqy − 1r + 3.5qy − 0.4rqy − 0.7r, for0.4 ≤ y < 15.04qy − 1.25rqy − 1.5r − 10.48qy − 1rqy − 1.5r + 5.44qy − 1rqy − 1.25r, for1 ≤ y < 1.5 (3.20)
The spacing between adjacent resonators v9,9I is calculated by multiplying the results produced by
equations (3.19) by the cavity width, already set as 14 mm, according to equation (3.21). This product
is corrected by the 0.95 factor in order to minimize cut-and-try efforts in tuning operations:
v9,9I33 = 0.95 × v9,9IE × E33, + = 1AD − 1 (3.21)
The total length of the filter cavity is calculated by taking into account all of the resonators diameters,
the spacing between them and considering a gap of 5 mm at each terminal side of the cavity, in order
to install the input and output connectors. This 5 mm gap is considered to be a common value that
allows the installation of the input and output SMA connectors.
So, since each filter specification will produce different results for the resonators dimensions, the cavity
length is given by equation (3.22):
33 = q2 × 5r + > × 733B +8v9,9I339: (3.22)
As an additional data, we shall calculate the air gap that ensures the existence of a certain distributed
capacitance between each resonator and the tuning screw above it, which is the master key to the filter
tuning. This is given by equation (3.23):
cf33 = /3 × = × #7332 '29 (3.23)
Where:
• : Permittivity given by equation (3.24);
• 9: Distributed capacitance, as expressed in equation (3.25).
The permittivity is given by expression (3.24):
/3 = × (3.24)
Where:
• : Permittivity of vacuum, which is approximately 8.854 pF/m;
• : Relative dielectric constant of the gap material (commonly air, which implies a relative
dielectric constant of approximately 1.0006).
The distributed capacitance is given by equation (3.25):
9 = RS × #RZ9R ' × cot _Z`2=.X01 (3.25)
The cavity height is obtained by adding 1 mm to the resonators height, according to equation (3.26).
ℎ33 = 33 + 1 (3.26)
It will be inside this 1 mm space that the tuning screw will be placed. Then, by tightening the screw until
it reaches a certain depth, ensuring a gap between it and the resonator top given by equation (3.23),
the necessary distributed capacitance will be created. For a better understanding a single resonator and
its tuning screw are used, in Figure 3.6, to demonstrate what was explained.
25
Figure 3.6 Resonators and tuning screws positioning inside the cavity filter
As a final note, it is important to clarify that this particular design was originally intended to have hollow
resonators to create, together with the tuning screws, the necessary distributed capacitance calculated
in equation (3.25). However, due to the software design limitations, it won’t be simulated as such.
Furthermore, it will be shown in Chapter 5, during the simulation software validation, that this
modification will not be relevant.
The resonators will be simulated as dense cylinders, and the distributed capacitance will be controlled
by the air gap between the top of each resonator and the cavity top, as already stated before.
26
27
4. Filtering Specifications and Design Results
Chapter 4
Filtering specifications and design
results
This chapter presents the filtering specifications for the two studied applications as well as the design
results for the combline cavity filter structural dimensioning. With these results a preview of the
simulation models is also presented. The scattering coefficients are briefly introduced in the last part of
the chapter.
28
4.1. LTE Front-End filter
4.1.1 Previous considerations
For this application, the filter specifications were initially proposed by a Portuguese mobile operator and
slightly modified according to the results obtained by the first approach, performed in the work of
Fragoso, in [7], where a digital filtering structure was used. The referred modification is in the attenuation
level where, instead of an 80 dB rejection at a 5 MHz distance from the upper passband frequency, only
60 dB are now specified.
4.1.2 LTE filter specifications
For this particular application, the ideal filter specifications are the ones present in Figure 4.1.
Figure 4.1 LTE ideal filtering specifications
As a reference for future considerations, the percentage of passband bandwidth as a function of the
central frequency is about 0.77 %.
Regarding the passband ripple, once it is a Chebyshev filtering function that is going to be implemented,
the initial requirements did not specify any value, leaving this value up to the designer to set. Still, it is
important to notice that this filter is intended to be used in cell phone systems, whose performance can
be affected by small changes in the signal power.
4.1.3 LTE combline cavity filter design results
Taking into account the specifications presented in Figure 4.1, and setting a passband ripple level of 0.1
dB and an extra 5 dB attenuation, to ensure the desired 60 dB while tuning the filter, the design equations
presented in Section 3.2.3 produced the results shown in Tables 4.1 and 4.2.
The passband ripple is a value that was left to be settled by the designer, as already mentioned. A 0.1
dB level was chosen as a compromise between the filter structural complexity and a frequency response
29
as flat as possible. Smaller values for the ripple could be considered but that would produce a structure
with a higher number of sections and, therefore, higher complexity.
Parameter Description Value
n Number of sections 11
b Cavity width 14 mm
L Resonators height 14.5068 mm
h Cavity height 15.5068 mm
l Cavity length 274.8911 mm
d Resonators diameter 6.773 mm
Table 4.1 LTE combline cavity filter main parameters values
Spacing between resonators - s Value [mm]
1-2 18.1201
2-3 19.1061
3-4 19.2788
4-5 19.3348
5-6 19.3542
6-7 19.3542
7-8 19.3348
8-9 19.2788
9-10 19.1061
10-11 18.1201
Table 4.2 Spacing between resonators inside the LTE combline cavity filter
Also as a result of the design equations from last chapter, a 0.3629 mm gap between the top of the
resonators and the tuning screws should be guaranteed. This value will be used as an initial value for
the screws depth, as it will be modified according to the first simulation results.
The detailed results of the design equations for the LTE front-end filter are presented in Annex A.
4.1.4 LTE combline cavity filter simulation model
With the results shown in last section, a simulation model was constructed and its design is presented
in Figure 4.2.
As an initial approximation, the coupling screws depth inside the cavity will be set as 10 mm and the
input and output coupling probes have 2.5 mm of radius and height.
Despite the structure symmetry, and for an accurate referral, the resonators can be numerated from left
to right, following the yy axis orientation of Figure 4.2.
30
Figure 4.2 LTE front-end filter simulation model
The simulation results for this initial approximation and its after tuning values are presented in Chapter
5.
4.2 Amateur radio 23 cm band filter
4.2.1. Previous considerations
For this application, the filter specifications were proposed by AMRAD, a Portuguese amateur radio
association.
According to the IARU Region 1 (where Portugal is included) 1240 – 1300 MHz band plan partition,
which can be consulted with further detail in [40], the 1291.494 – 1296 MHz frequency band is reserved
to all modes operation. In the same document, the 1296 – 1296.15 MHz frequency band is used to both
moon bounce and PSK312.
The fact that the all modes operation band is located just next to the moon bounce band, with no
frequency gap separating them, is the main cause of interference, particularly when spurious emissions
exist. In this particular case, an amateur TV transmitter, centered at 1294 MHz with a 6 MHz bandwidth,
is the cause of the existence of spurious emissions that interferes with the very weak signals of an earth-
moon-earth communication channel, belonging to this narrowband system centered at 1296 MHz.
In Figure 4.3, the current situation is depicted.
2 PSK31 or "Phase Shift Keying, 31 Baud" is a computer-soundcard-generated radioteletype mode, used primarily by amateur radio operators to conduct real-time keyboard-to-keyboard chat. [41]
31
Figure 4.3 Current frequency spectrum around 1296 MHz
The oscilloscope is centered at 1296 MHz and it is visible a slight perturbation on the left side of the
spectrum, centered near 1294 MHz. This is the perturbation that this filter should eliminate.
4.2.2. Amateur radio 23 cm band filter specifications
For this particular application, the ideal filter specifications are shown in Figure 4.4.
Note that, in these specifications, the rejection band is not specified. This value was left as a variable,
but is to be chosen taking into account the complexity of the final structure and the fact that the interfering
source is located at 1294 MHz.
Figure 4.4 Filter specifications for the ATV interference on the amateur radio 23 cm band
As a reference for future considerations, the percentage of passband bandwidth as a function of the
central frequency is about 0.62 %.
Besides this specification, a passband ripple of 0.04 dB and a S11 (scattering parameters will be defined
in Section 4.3) minimum value of -20 dB within the passband were specified.
32
4.2.3 Amateur radio 23 cm band combline cavity filter design results
With the specifications just presented in the last section, as well as setting a passband ripple level of
0.01 dB and an extra 10 dB of attenuation (to guarantee a security margin), as well as fixing a rejection
band of 21 MHz, the design equations presented in Section 3.2.3 produced the results shown in Tables
4.3 and 4.4.
Here, the passband ripple, when compared to the LTE front-end filter presented in Section 4.1.2, can
be closer to a completely flat response since the complexity of the combline cavity is lower. Regarding
the rejection band, the 21 MHz value is, as seen before, a compromise between complexity and the
specifications fulfillment.
Parameter Description Value
n Number of sections 5
b Cavity width 14 mm
L Resonators height 28.8462mm
h Cavity height 29.8462 mm
l Cavity length 117.8417 mm
d Resonators diameter 6.7414 mm
Table 4.3 Amateur radio 23 cm combline cavity filter main parameters values
Spacing between resonators - s Value [mm]
1-2 17.9009
2-3 19.1665
3-4 19.1665
4-5 17.9009
Table 4.4 Spacing between resonators inside the amateur radio 23 cm combline cavity filter
Also as a result of the design equations from last chapter, a 0.1808 mm gap between the top of the
resonators and the tuning screws should be guaranteed. This value, as in the LTE application, will be
used as an initial value for the screws depth, as it will suffer modifications according to the first simulation
results.
The detailed results of the design equations for the amateur radio 23 cm band filter are presented in
Annex B.
4.2.4 Amateur radio 23 cm band combline cavity filter simulation model
With the results shown in last section, a simulation model was constructed and its design is presented
in Figure 4.5.
As an initial approximation, the coupling screws depth inside the cavity will be set as 25 mm and the
input and output coupling probes have 2.5 mm of radius and height.
33
Once again, resonators can be numerated for a better referral. It will follow an ascendant order from left
to right of Figure 4.5.
Figure 4.5 Amateur radio 23 cm band filter simulation model
4.3 Scattering coefficients
As already mentioned in Section 4.2.2, the scattering parameters will be used as a measuring constant
during the simulation results demonstration of this thesis. Following the same presentation approach as
Matthaei, in [3], the scattering parameters are now introduced.
The performance of any linear two-port network with termination can be described in terms of four
scattering coefficients: S11, S12, S21 and S22. Regarding S11 and S22 they represent the reflection
coefficients at ends 1 and 2, giving information about the input and output return losses, respectively.
The S12 and S21 parameters are defined as the transmission coefficients from ends 2 to 1 and 1 to 2,
respectively, both giving information about the insertion losses. Furthermore, if the structure is reciprocal
S12 will be equal to S21.
In the simulation results that are going to be presented in Chapter 5 only the S12 simulation result will
be shown. This is due to the already mentioned symmetry, which was guaranteed by the design
equations, producing almost exactly equal results for S11 and S22 and for S12 and S21.
34
35
5. Simulation Results
Chapter 5
Simulation results
This chapter starts by presenting the simulation software validation. This validation is done using a real
cavity filter frequency response results and comparing it with the results obtained from a simulation
model. Then, the initial and after-tuning simulation results of the two designed filters for the studied
applications are presented, as well as the description of the tuning procedures for each of the filters.
36
5.1. Simulation software presentation
To access the accuracy of the simulation, particularly the frequency response of a combline cavity
structure that performs the band pass filtering, software developed by ANSYS that performs an analysis
based on the Finite Element Method was used. HFSSTM version 13.0, [5], short for High Frequency
Structural Simulator, allowed the simulation and fine tuning of the structures conceived by the design
equations presented in Section 3.2.3.
HFSSTM offers multiple solver technologies based on finite element, integral equation and advanced
hybrid methods that solve a wide range of microwave, RF and high-speed digital applications [5].
By defining the structure and its composing materials, the software automatically generates an
appropriate, efficient and accurate mesh for solving the problem. In the present work, a constant analysis
of the scattering parameters will be presented, since this is an indicator of the filters frequency response.
5.2. Simulation parameters
For an accurate and congruent study, every analysis along this thesis was made with the following
simulation parameters.
HFSSTM simulation parameter Value
Solution Type Driven Terminal
Setup passes 4
Percent refinement 30
Delta S 0.02
Basis Order First Order
Step Size 0.1 MHz
Sweep Type Fast
Table 5.1 Simulation software analysis parameters
Apart from these values, three other parameters are to be defined according to the analyzed case. The
Solution Frequency parameter was configured according with the central frequency of the analyzed
model, as it will be used to correctly simulate the S-parameters. The two remaining simulation
parameters are the Sweep Start and Sweep Stop values, which are defined according to the portion of
the spectrum that the user wants to analyze.
This parameter setting has taken into account the HFSSTM tutorial, in [43], where a band pass interdigital
filter was simulated.
5.3. Real combline cavity filter structure measurement
In order to obtain a better understanding of the software environment and some experience in it, an
already constructed and operating combline cavity filter, courtesy of an AMRAD associate, was used in
a reverse engineering exercise.
Despite this combline cavity filter did not have the exact same structural organization comparing to the
ones that are going to be proposed for the considered applications, its dimensions were measured with
37
a caliper rule and together with a definition of the construction material an analysis model could be
defined in HFSSTM. In fact, one of the biggest interest points in studying this operating cavity filter is to
correctly define which materials are used in the producing process of such components, so that the
simulation models could be as close as possible to reality.
The studied filter performs a band pass filtering at a central frequency of 2.8 GHz. The dispersion
parameter S12 was measured in a Vector Network Analyzer (VNA), and will be the comparison mean
between real and simulated systems.
The software model is represented in Figure 5.1 and the measured values are presented in Table 5.2,
5.3 and 5.4.
Figure 5.1 Real combline filter simulation model
It is important to make note of following considerations:
• Resonators, in their cylindrical shape, will be numerated for an accurate referral (R1 to R6), in
ascendant order from left to right of Figure 5.1. Resonators 1 and 6 are equal but different from
2, 3, 4 and 5, whose shape is equal between them. All of the resonators are centered with
respect to the cavity width (b) and concentric with the input and output coupling probes (in the
case of resonators 1 and 6) or tuning screws (in case of resonators 2 till 5);
• The top of the resonators in the real filter have a small hollow reentrance where the tuning screw
fits for a correct tuning. However, the resonators will not be modelled with this hollow top once
the software is only capable of simulating pre-determined geometrical forms. Moreover, through
simulation results, this small negligence will prove to be harmless to the obtained results. All of
the resonators are silver plated;
• The tuning screws located above resonators 2 till 5 will adopt the designation of the resonator
that is below them (as an example, screw above resonator 2 will be designated screw 2).
Furthermore, screws are represented with a cylindrical shape without thread, due to the
software limitations in terms of available geometrical shapes. They are modeled only with its
depth (c) inside the cavity for representation simplicity and since the part of the screw that stay
midway from outside to the inside of the cavity has no influence in the simulation results. In
reality they penetrate in the cavity block through a hole that is drilled on top of the cavity block,
being adjusted from the exterior, by means of a screwdriver, that modifies the proximity between
38
resonators and the top of the resonator. Screws are made of brass and despite each one having
its own depth inside the cavity (c) they have a common diameter of 5.52 mm;
• Input and output connections are provided through a SMA female jack. In the simulation
software, this terminal jack is represented by an outer isolation made of PTFE (commonly known
as Teflon) and an inner conductor made of copper;
Figure 5.2 SMA female connector to be installed at each side of the cavity block to perform the input/output
connections (extracted from [42])
• The filter outer structure is a compact block made of brass, with a side cover plate that unravels
the inside structure and its resonating elements. This enclosing structure provides a robust
seating structure for the resonant elements, with almost 10 mm at the top and 2 mm at the side
to prevent external vibrations that would affect the desired behavior;
• The input and output probes, represented by its height inside the cavity (p) are of cylindrical
shape and are silver plated. They have 2.5 mm of diameter.
A more detailed picture of the combline filter is presented in Figures 5.3 and 5.4, together with the
identification of the measured dimensions. The resonators are numerated, as mentioned, from the left
to the right.
Figure 5.3 Side view of the real combline cavity filter and identification of the main elements dimensions
Figure 5.4 Top view of the real combline cavity filter and identification of the main elements dimensions
39
The measured values of the real combline cavity filter, obtained with the caliper rule, are organized into
Tables 5.2, 5.3 and 5.4.
Parameter Description Value [mm]
b Cavity width 16.2
L Resonators height 20
Lterminals Terminal resonators height 23.65
h Cavity height 24.85
l Cavity length 150.2
d Resonators diameter 5.87
dterminals Terminal resonators diameter 8.7
p Coupling probes height 1.2
e Teflon isolation height 10
Table 5.2 Real combline cavity filter main parameters measured values
Spacing between resonators - s Value [mm]
1 (terminal) and 2 11.43
2 and 3 22.16
3 and 4 22.56
4 and 5 22.16
5 and 6 (terminal) 11.43
Table 5.3 Real combline cavity filter measured values for spacing between resonators
Tuning screws depth – c Value [mm]
2 3
3 2.4
4 3.3
5 3.4
Table 5.4 Real combline cavity filter measured values for tuning screws depth
40
5.3.1. Real combline cavity filter frequency response
By connecting the input and output ports of the combline cavity filter to a VNA, the measured scattering
parameter S12 is as the one depicted in Figure 5.5, as follows:
Figure 5.5 Real combline cavity filter frequency response measured by the VNA
5.3.2. Real combline cavity filter simulation results and tuning process
Without any modification, the model simulation produced the following result for S12:
Figure 5.6 First simulation result of S12 for the real combline cavity filter model
2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60Freq [GHz]
-180.00
-160.00
-140.00
-120.00
-100.00
-80.00
-60.00
-40.00
dB
(St(
Cyl
ind
er1
7_T
1,C
ylin
der1
8_T
1))
HFSSDesign1XY Plot 1 ANSOFT
m1
Curve Info
dB(St(Cylinder17_T1,Cylinder18_T1))Setup1 : Sw eep
Name X Y
m1 2.7698 -92.8208
41
It is clearly visible that the measured and simulated results do not perfectly match.
As an effort to approximate the simulated and real measured results, an iterative cut-and-try process
has been started in order to find what the simulation model would require in its dimensions to achieve a
result as close as possible to the one that was measured in the VNA.
The modifications were only introduced in the tuning screws 2, 3, 4 and 5, as these elements are not
just those responsible for the cavity response tuning but also where there have been some difficulties
to make a perfect measure with the caliper rule, as they did not have a clear access through the
removable side cover plate.
Moreover, it was noticeable beforehand, comparing measured and simulated results from Figures 5.5
and 5.6, respectively, that the simulation model would benefit from a lower tuning screws depth inside
the cavity. This modification leads to lower levels of distributed capacitance between resonators and the
tuning screws, causing an increase of the filter response central frequency.
After some iteration steps, the tuning screws depths inside the cavity that led to a scattering parameter
S12 that better approximated the measured result from Figure 5.7 are described in Table 5.5.
Tuning screws depth – c Value [mm]
2 2.85
3 2.3
4 3.21
5 3.36
Table 5.5 New tuning screws depth inside the cavity filter simulation model
After applying these modifications to the simulation model tuning screws depth, it produced the following
result from Figure 5.7.
Figure 5.7 Simulation model results for S12 with the new tuning screws depth
Comparing Figures 5.5 and 5.7, it is noticeable that the shape of the frequency response is pretty similar.
Moreover, the fact that the markers from both graphs present a difference of about 1 dB, and knowing
42
that the real and simulated filters will never be exactly equal due to, for example, production error
margins or measurement precision, show that the results are pretty close to each other and, more
importantly, that HFSSTM simulations produce a trustable prediction of the real frequency behavior for
these cavity filter structures.
It is also been proven here that the small difference in the resonators modelling (the small hollow
resonator top where the tuning screw fit was not modelled due to software design limitations) does not
prevent the simulation result to approximate the real measurement.
5.3.3. Real combline filter analysis conclusion
This part of the work is also important to raise awareness for the fact that these types of filtering
structures (combline cavity filters) have a fairly high sensitivity to structural modifications. This
characteristic is well proven by the minor changes that needed to be introduced in the tuning screws
depth, comparing to the measured values of the real one. For a better understanding it is compiled in
Table 5.6 the values already shown in Tables 5.4 and 5.5, together with the variations that they were
submitted to.
Tuning screws depth – c Simulation [mm] Real [mm] Variation [mm] Percent Variation [%]
2 2.85 3 -0.15 5.3
3 2.3 2.4 -0.1 4.3
4 3.21 3.3 -0.09 2.8
5 3.36 3.5 -0.14 4.2
Table 5.6 Comparison between real measured and simulated values of the tuning screws depth inside the cavity
In the work of Liu et al., in [44], where despite being a case where the gap between resonators and the
top of the cavity was the one responsible for creating the needed distributed capacitance (in the present
case, as well as in the case of the combline filters that are going to be presented afterwards, tuning
screws are those responsible for creating the needed distributed capacitance, leading to a simpler after
production tuning), it is stated that both the resonant frequency of a cavity and the unloaded quality
factor are found to be functions of the cavity size, post sizes and the gap between the post top and
cavity ceiling. When the gap between the post and the cavity top wall is very small, a gap change in the
order of micrometers can result in a frequency change in the order of GHz. In the studied case, the
frequency changes did not reach those levels but it has been verified that, by introducing modifications
in the order of micrometers to the tuning screws depth, visible changes were seen in the simulation
results.
Having this real cavity filter as an example, the simulated models that are going to be presented next
will use the same tuning and coupling screws diameter, the same input and output ports (SMA female),
despite being positioned at each side of the cavity instead of in one of the cavity tops, the same
construction materials for the cavity block and screws (brass), and silver plating of the resonators and
input and output coupling probes will also be used.
43
5.4. LTE front-end filter simulation
With the results shown in Section 4.1.3, a simulation model was constructed and its design is presented
in Figure 5.8, as it has already been introduced in another perspective.
It is important to recall that, despite not being depicted in Figure 5.8 to avoid it to be overcrowded, the
structure dimensional nomenclature follows what was presented in Figures 5.3 and 5.4.
Figure 5.8 Side view of the LTE front-end filter simulation model
5.4.1. LTE filter initial simulation results
A first simulation using the exact values given by the equations was done. The results proved to be
detuned from the expected 2585 MHz of central frequency. Moreover, neither the passband ripple nor
the rejection band was exactly accomplished.
It is, nevertheless, a quite good result in terms of response shape, despite the 10 dB losses that occur
on the passband, and that need to be accounted for if this filter is to be installed in a real system.
Figure 5.9 First simulation results of the S12 parameter for the LTE filter model obtained through the design
equations
2.00 2.20 2.40 2.60 2.80 3.00Freq [GHz]
-250.00
-200.00
-150.00
-100.00
-50.00
0.00
dB
(St(
Cyl
ind
er1
4_
T1,
Cyl
ind
er1
5_
T1
))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
44
Figure 5.10 Detailed response behavior of the first simulation of the LTE filter model obtained through the design
equations
5.4.2. LTE front-end filter tuning process
Analyzing these results, and taking into account that the wanted central frequency of the filtering
response is lower than the one here obtained (2585 MHz instead of the actual 2640 MHz), it was
concluded that the modifications should include the tuning screws tightening, causing a decrease of the
air gap between tuning screws and resonators top, producing bigger values of distributed capacitance
and, therefore, lowering the frequency response. Moreover, some modifications to the coupling screws
could be tried, as the wanted bandwidth is not exactly obtained.
Since the exact values that produce the best results are completely unknown, an iterative trial and error
method had to be used. Several options were simulated, all of them consisting of combinations of higher
screws depths inside the cavity.
The best results were obtained with:
• Air gap between tuning screws and resonators top: 0.345 mm;
• Coupling screws depth inside the cavity: 14.5 mm.
The simulation result for S12 is represented in Figure 5.11.
Figure 5.11 Simulation result of the S12 parameter after the first tuning iteration process made to the LTE filter
2.56 2.58 2.60 2.62 2.64 2.66 2.68 2.70 2.72Freq [GHz]
-100.63
-90.00
-70.00
-50.00
-30.00
-10.00
0.35
dB(S
t(C
ylin
der
14_T
1,C
ylin
der1
5_T
1))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
2.00 2.20 2.40 2.60 2.80 3.00Freq [GHz]
-225.00
-200.00
-175.00
-150.00
-125.00
-100.00
-75.00
-50.00
-25.00
0.00
dB
(St(
Cyl
ind
er1
4_T1
,Cyl
ind
er15
_T1
))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
45
Figure 5.12 Detailed simulation result of the S12 parameter after the first tuning iteration process made to the LTE
filter
From this result it is clearly noticeable that, despite the correct tuning that has been reached, the filter
response still does not show the correct bandwidth and rejection values. Moreover, at this point, both
the tuning and coupling screws did not have much more further space to continue with the iterative
process. Thus, a second hypothesis had to be considered.
Since lower frequencies lead to combline cavity structures with higher values of height (given that
resonators height is obtained with respect to the filter specification central frequency), a 0.5 mm increase
in the cavity height value (h) was imposed, leading to a height of exactly 16 mm instead of the previous
15.5068 mm. Moreover, the resonators height (L) was also increased to 14.52 mm instead of 14.5068
mm obtained from the design equations.
After a further iterative process, the values that were modified (comparing to what is stated in Chapter
4) to produce the best simulation results, are compiled in Table 5.7.
Parameter Description Value
L Resonators height 14.52 mm
h Cavity height 16 mm
Air gap Gap between resonators and tuning screws 0.35 mm
Coupling screw depth Coupling screws depth inside cavity 12.5 mm
Table 5.7 Modifications of the LTE cavity filter parameters
The simulation model has a very similar aspect since the changes, despite being effective, are minimal,
in the order of a few millimeters or even micrometers.
Figure 5.13 Final LTE front-end filter simulation model
2.48 2.50 2.53 2.55 2.58 2.60 2.63 2.64Freq [GHz]
-64.61
-55.00
-45.00
-35.00
-25.00
-15.00
-5.00
1.37
dB
(St(
Cyl
inde
r14_
T1,C
ylin
der
15_
T1))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
46
Figure 5.14 Side view of the final LTE front-end filter simulation model
Figure 5.15 Top view of the final LTE front-end filter simulation model
The best results are now presented for S12, in Figures 5.16 and 5.17, and S11, in Figure 5.18.
Figure 5.16 Final S12 simulation result of the LTE filter considering the changes from Table 5.7
Figure 5.17 Detailed picture of the simulation result for the S12 parameter from the LTE filter considering the changes from Table 5.7
2.00 2.20 2.40 2.60 2.80 3.00Freq [GHz]
-225.00
-200.00
-175.00
-150.00
-125.00
-100.00
-75.00
-50.00
-25.00
0.00
dB
(St(C
ylin
der1
4_T
1,C
ylin
der
15_T
1))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
2.50 2.53 2.55 2.58 2.60 2.63 2.65 2.67Freq [GHz]
-87.50
-75.00
-62.50
-50.00
-37.50
-25.00
-12.50
0.00
dB
(St(
Cyl
ind
er1
4_T
1,C
ylin
der1
5_
T1))
HFSSDesign1XY Plot 3 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
47
The more detailed figure of S12 shows that the filtering response inside the passband (2575 – 2595 MHz)
is not completely perfect since the passband ripple is nowhere near the expected 0.1 dB and about 25
dB of insertion loss is registered. It was, however, the best result possible, given what is exposed next,
in the S11 simulation result analysis.
Figure 5.18 S11 simulation result of the final LTE filter obtained considering the changes from Table 5.7
The reason to present here the S11 simulation result is to evidence the low levels of return losses that
are registered in this case. Despite other values of tuning and coupling screws depth produced slightly
better results for S12 than the one presented in Figure 5.16, the return losses reached values of 20 dB,
or even more, inside the 20 MHz of the wanted passband. Here, the return losses inside the passband
are lower than 2 dB.
Once this filter is to be installed in a system where the 2 communication directions share the same
access, it is of maximum importance to guarantee that the frequency band where the information is
transported stays as unchanged as possible until it reaches the transmitting antenna.
5.5. Amateur radio 23 cm band filter simulation
With the results shown in Chapter 4, a simulation model was constructed and its design is presented in
Figure 5.19.
It is important to recall that, despite not being depicted in Figure 5.19 to avoid it to be overcrowded, the
structure dimensional nomenclature follows what was presented in Figures 5.3 and 5.4.
Figure 5.19 Side view of the designed 23 cm amateur radio filter
2.00 2.20 2.40 2.60 2.80 3.00Freq [GHz]
-5.00
-4.00
-3.00
-2.00
-1.00
0.00
dB(S
t(C
ylin
de
r14
_T1
,Cyl
ind
er1
4_
T1)
)
HFSSDesign1XY Plot 2 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder14_T1))Setup1 : Sw eep
48
5.5.1. Amateur radio 23 cm band filter initial simulation results
The first simulation results, using the exact values given by the equations of Chapter 3, resulted in a
filter response slightly detuned from the expected central frequency and with a bigger passband than
the 8 MHz wanted.
Figure 5.20 First simulation results of the S12 parameter from the 23 cm amateur radio filter model obtained
through the design equations
Figure 5.21 Detailed first simulation results of the S12 parameter of the 23 cm amateur radio filter model obtained
through the design equations
5.5.2. Amateur radio 23 cm band filter tuning process
By the analysis of the results shown in Figure 5.20, it was determined that the tuning procedure should
include modifications in both the tuning and coupling screws. These modifications are related to the
need of lowering the central frequency and the passband width of the filter response, respectively.
1.00 1.10 1.20 1.30 1.40 1.50Freq [GHz]
-160.00
-140.00
-120.00
-100.00
-80.00
-60.00
-40.00
-20.00
0.00
dB
(St(
Cyl
ind
er1
4_
T1
,Cyl
ind
er1
5_
T1
))
HFSSDesign1XY Plot 1 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
1.30 1.31 1.33 1.34 1.35 1.36Freq [GHz]
-50.52
-45.00
-35.00
-25.00
-15.00
-5.00
0.40
dB(S
t(Cyl
inde
r14_
T1,C
ylin
der1
5_T1
))
HFSSDesign1XY Plot 1 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
49
However, for this application, it has been decided to directly proceed to the cavity height oversize. This
decision was made taking into account that, for the LTE filter, the simple tuning of the original designed
structure did not fulfill the expected performance.
Thus, an increase of more than 1 mm was imposed to the cavity height. An increase of the resonators
height was also tried but, since it did not produced any major improvements to the filtering response,
was later discarded.
The new values are described in Tables 5.8 and 5.9, with the rest of the parameters presented in
Chapter 4 staying unchanged.
Parameter Descript ion Value
h Cavity height 31 mm
Air gap Gap between resonators and tuning screws 0.15 mm
Coupling probes height Input and output coupling probes height 2.8 mm
Table 5.8 Modifications of the 23 cm amateur radio cavity filter parameters
Coupling screws depth - c Value [mm]
1 20
2 6
3 5
4 20
Table 5.9 New values of the coupling screws depth inside the 23 cm amateur radio filter
The changed simulation model considering the modifications stated in Tables 5.8 and 5.9 can be
identified in Figure 5.22, where the new simulation model is depicted. One of the more noticeable
changes, comparing it to the original designed filter, is in the coupling screws depth.
Despite not being depicted in Figure 5.22, coupling screws are identified from 1 to 4, from the left to the
right of the figure and are located between each pair of resonators.
Figure 5.22 Side view of the 23 cm amateur radio filter considering the changes from Tables 5.8 and 5.9
50
The new simulation results, considering these changes, are presented in Figures 5.23 till 5.26.
Figure 5.23 Final simulation results of the S12 parameter of the 23 cm amateur radio filter model considering the
changes from Table 5.8 and 5.9
From the obtained results it is noticeable that the specified attenuation level of 60 dB is not fulfilled. Still,
once this filter was meant to stay at a lower complexity level than the LTE already presented, some
trade-offs had to be accepted.
Figure 5.24 Detailed final simulation results of the S12 parameter of the 23 cm amateur radio filter model
considering the changes from Table 5.8 and 5.9
Despite the combline cavity filter seems a little detuned, it is important to recall here that this filter is
meant to protect a moon bounce radio amateur communication system, centered at 1296 MHz having
a very narrow bandwidth (the maximum bandwidth value is, according to [40], 500 Hz). Furthermore, a
10 dB of insertion losses in the passband must be accounted for.
1.25 1.27 1.30 1.33 1.35Freq [GHz]
-75.00
-62.50
-50.00
-37.50
-25.00
-12.50
0.00
dB(S
t(C
ylin
de
r14
_T1
,Cyl
inde
r15_
T1)
)
HFSSDesign1XY Plot 1 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
1.290 1.295 1.300 1.305 1.310 1.315Freq [GHz]
-25.00
-20.00
-15.00
-10.00
-5.00
0.00
dB
(St(
Cyl
ind
er1
4_
T1
,Cyl
ind
er1
5_
T1
))
HFSSDesign1XY Plot 1 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
51
Figure 5.25 Final simulation results around 1296 MHz of the S12 parameter from the 23 cm amateur radio filter
model considering the changes from Table 5.8 and 5.9
Figure 5.26 S11 simulation result of the final 23 cm amateur radio filter obtained considering the changes from
Table 5.8 and 5.9
Given that the initial specifications provided a minimum level of -20 dB for the S11 parameter, it is possible
to say, from the analysis of Figure 5.26, that this specification is accomplished by the combline cavity
filter, making it a good approximation of the wanted filtering specifications.
To conclude, it is important to say that despite the amount of simulation results that can be reached,
using different configurations of the model, whose combinations are almost outnumbered, even when
some dimensions are fixed, the real frequency behavior of these structures are exclusively accessed
through prototypes, where modifications are easily done and results are immediately analyzed.
1.291 1.292 1.294 1.296 1.298 1.300 1.302 1.304Freq [GHz]
-14.98
-13.75
-12.50
-11.25
-10.00
-8.75
-7.50
-6.25
-5.24d
B(S
t(C
ylin
de
r14
_T
1,C
ylin
de
r15
_T
1))
HFSSDesign1XY Plot 1 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder15_T1))Setup1 : Sw eep
1.25 1.27 1.30 1.33 1.35Freq [GHz]
-8.75
-7.50
-6.25
-5.00
-3.75
-2.50
-1.25
0.00
dB
(St(
Cyl
ind
er1
4_
T1
,Cyl
ind
er1
4_
T1
))
HFSSDesign1XY Plot 2 ANSOFT
Curve Info
dB(St(Cylinder14_T1,Cylinder14_T1))Setup1 : Sw eep
52
53
6. Conclusions and Future Work
Chapter 6
Conclusions and future work
In this chapter, a summary of all the work carried out during this dissertation is provided, together with
the indication of the obtained results and its implications. In the end, some future works suggestions are
given.
54
6.1. Conclusions
The main objective of this thesis was the study of high frequency filtering in the signal processing of two
applications that are being affected by interference. To overcome this situation, the installation of front-
end filters, capable of perform a highly selective operations, are one of the possible solutions and its
design and performance was here addressed.
Motivated by a previous work regarding the isolation from interference caused by adjacent channels of
the LTE system band 38, where it was suggested that the use of cavity filters could be a viable
hypothesis, the study of this filtering technology was now presented. Nevertheless, as the previous
design experience led to a very complex solution, and the filtering specifications stayed almost
unaltered, a second application was considered, mainly due to its less demanding requirements,
promising a more realizable filter. This second application was the amateur radio 23 centimeters band,
more specifically, the reserved set of frequencies for moon bounce communications, whose
performance of a user is being affected by the emitting power of an amateur television service repeater.
Thus, in this thesis, an initial research of cavity filters was made in Chapter 2, with special attention
being given to publications where a direct transformation of filtering specifications into structural
dimensions was presented. Combline cavity filters were chosen due to its guarantee of producing steep
rejection bands and the existence, among the cited publications, of a straightforward approach, able of
converting filter specifications into dimensions of the cavity itself and all of its composing parts, such as
resonators or spacing between them. The elected design methodology, which implements a Chebyshev
approximation for the filtering response, is presented in Chapter 3, with all of the equations and definition
of parameters being also given. The design approach here presented was adapted from the publication
of Natarajan, in [7].
In Chapter 4, both specific filter requirements are presented followed by the results of the filtering
structural design. Regarding the requirements, on one hand, the LTE filter requested 60 dB of
attenuation in a 5 MHz rejection band, assuring the existence of 20 MHz of passband centered at 2585
MHz. No specific requirements were made on the maximum insertion losses, and a 0.1 dB of passband
ripple was chosen to have a response as flat as possible. On the other hand, the amateur radio filter,
besides its lower central operating frequency of 1300 MHz, was conceived to have 60 dB of attenuation
in a 21 MHz of rejection band, together with a passband of 8 MHz and a maximum ripple of 0.01 dB,
trying to assure the maximum value 0.04 dB that was specified. Regarding the return losses, a maximum
value of -20 dB was specified, but once this value is not used by the design equations, it had to be
verified afterwards, in the simulation results analysis.
From the design equations results some conclusions can be pinpointed. First, the LTE filter is a quite
complex filtering structure. In fact, the LTE filter was designed to have 11 resonators inside the cavity
block and the amateur radio filter only 5. This number of resonators is mainly given by the rejection band
that is requested; as narrower the rejection band is, the greater the number of resonators has to be. The
structure length is also directly influenced by the number of resonators, the bigger the number of
resonators that have to be installed, the bigger the structure will be. Second, as the resonators height is
given by one eighth of the filter central frequency wavelength, and the cavity structure height is directly
55
influenced by this value, the bigger the application working frequency is, the lower the filtering structure
will be. Finally the concept of distributed capacitance was introduced as the key to a correct filter tuning.
The filter response central frequency will decrease if a bigger value of distributed capacitance is created
between the tuning screw and the resonator that is concentrically placed below it, and the other way
around. This distributed capacitance increase is developed by the tuning screws fastening, and vice
versa.
The simulation models are presented in detail in Chapter 5. Here, the first part is dedicated to the
simulation software validation that was performed using a real combline filter, measuring it and
constructing a simulation model, reaching error margins of about 5% for the tuning screws depth inside
the cavity. Then, both custom designed filters were simulated. Despite both initial approximations
presented slightly detuned frequency responses with a good approximation, the tuning process that was
performed afterwards, where some structural modifications were introduced in order to obtain a correct
filter central frequency tuning, led to different outcomes. For the LTE filter, the after tuning conclusion,
which included a 1 millimeter increase of the cavity height, helping the filter response central frequency
to be lower than initially obtained, is that the best reached result did not perfectly fit the specifications,
having about 25 dB of insertion loss, producing a filtering response that was far from what was expected.
Moreover, high sensitivity was registered in every screw, independently of them being of tuning or
coupling control. Any minimal alteration to their values produced noticeable changes, sometimes
meaning differences of several MHz. This fact was also verified in the amateur radio filter but in a minor
scale, due to its lower structural complexity. In that case, a good approximation was reached, and
despite it does not completely fulfill the initial requirements, it is one promising first step that can assure
a good study base for the next stage, the structure milling and production. The after tuning results,
indicated that 50 dB of attenuation at a 30 MHz distance from the central frequency can be reached. It
is important to recall that the tuning process of this filter also included a 1 millimeter increase of the
cavity height, for the same reasons as in the LTE filter.
Finally, as already mentioned in Chapter 5, and as a final remark, it is important to remember that the
presented simulation results are no more than a software emulation of a real structure. In reality, many
variables here neglected like vibrations, operating temperature or lack of precision in the filter
production, can influence its behavior. Due to this fact, final conclusions about this technology appliance
in the requested filtering operations can only be done by prototype testing results.
6.2. Future work
To be able to advance into the production stage of any of the filters and deliver them into the market,
some issues could require some further attention.
Regarding the filtering technology that was used in this thesis (combline cavity filters), simulations of the
very same structures in other software environments could be considered, preferentially in those where
a closer to reality models. This would be useful to confirm the obtained results from HFSSTM having,
therefore, a much higher level of confidence if real prototypes are to be constructed and produced
afterwards. Other design approaches can also adopt different design methodologies, with some
examples given in Chapter 3. Future works could also try to impose different combline cavity filter
56
structural constants. In this work, a constant value of 14 mm for the cavity width was considered, given
that the author of the followed design approach suggested it as a standard value. However, other values
are absolutely valid, as long as attention is given to the graphic interpolation that was performed, and
that is mentioned in Section 3.2.3.
Furthermore, and particularly regarding the LTE filter, which motivated this work, future works could
consider the modelling and simulation of the same filtering specifications into other cavity filtering
structures, such as interdigital or iris-coupled filters, as they also promise to achieve quality factors in
the order of what is requested. Particularly for the second option mentioned (iris-coupled filters), a very
practical, straight-forward, approach is presented in the same publication that was used in this thesis
for the combline filter design. Furthermore, given that both studied applications require filtering
passbands that are around 0.7 % of the response center frequency (filter specifications are presented
with detail in Sections 4.1.2 and 4.2.2), and once iris-coupled filter are commonly applied in the design
of band pass filters having a bandwidth of less than 1% of center frequency [38], they might have a
better performance than those presented here. This iris-coupled topology was not already considered
in this thesis due to the already cited simulation software limitations in terms of available geometrical
forms. In the chosen simulation software (HFSSTM) the design of the irises was completely impossible
since their position, almost embedded in the cylindrical sections, requires a fine trimming of the cavity
block, which in practical terms is easy, but in this simulation software was unattainable.
57
A. Annex A - LTE combline filter design results
Annex A
LTE combline filter design results
This chapter presents the detailed results of the LTE combline filter dimensioning, using the design
equations from Chapter 3.
58
For an accurate display of the filter structural dimensions, detailed results of the LTE combline filter
dimensioning, according to the filter specifications of Section 4.1.2 and obtained with the design
equations that are presented in Section 3.2.3, are now presented
Constant Denomination Value 65 0.1 ./ 2.5950e+09 . 2.5850e+09 . 2.5750e+09
b 14
Rejection band 5e+06
Table A.1 Input values of the LTE filter design
The following results are indexed by its numeration along Chapter 3.
Chapter 3 - Equation Result
(1) 11
(2) 1.5
(3) 0.0077
(4) 14.5068
(5) 6.7730
(6)
6.7169 6.7958 6.7855
6.7833 6.7826 6.7824 6.7826
6.7833 6.7855 6.7958 6.7094
(7) 2.8
(8)
7.2484 7.3664 7.3511 7.3478
7.3468 7.3465 7.3468 7.3478
7.3511 7.3664 7.2372
(9)
0.5177 0.5262 0.5251 0.5248
0.5248 0.5247 0.5248 0.5248
0.5251 0.5262 0.5169
(10)
6.9551 4.8065 5.3107 5.3226
5.3249 5.3256 5.3258 5.3256 5.3249
5.3226 5.3107 4.8025 6.9551
(11) 0.0059 0.0059
(12) 0.9181
(13) 1.0000 1.2031 1.4523 2.1515 1.6332 2.2378
1.6559 2.2378 1.6332 2.1515 1.4523 1.2031
(14) 0.1423 0.4154 0.6549 0.8413 0.9595
1.0000 0.9595 0.8413 0.6549 0.4154 0.1423
59
Chapter 3 - Equation Result
(15) 5.1573
(16) 0.2366
(17) 0.0054 0.0040 0.0038 0.0037 0.0037
0.0037 0.0037 0.0038 0.0040 0.0054
(18)
6.9551 4.8065 5.3107 5.3226 5.3249
5.3256 5.3258 5.3256 5.3249
5.3226 5.3107 4.8025 6.9551
(19) 0.4536 1.3624 1.4365 1.4495 1.4537 1.4552
1.4552 1.4537 1.4495 1.4365 1.3624 0.4536
(20) 0.4427 0.6662 0.6737 0.6750 0.6754 0.6755
0.6755 0.6754 0.6750 0.6737 0.6662 0.4427
(21) 18.1201 19.1061 19.2788 19.3348 19.3542
19.3542 19.3348 19.2788 19.1061 18.1201
(22) 274.8911
(23) 0.3629
(24) 1.0006
(25) 8.7955e-13
(26) 15.5068
Table A.2 Chebyshev combline filter design results for the LTE application
60
61
B. Annex B – Amateur radio 23 cm combline filter design results
Annex B
Amateur radio 23 centimeter band
combline filter design results
This chapter presents the detailed results of the amateur radio combline filter dimensioning, using the
design equations from Chapter 3.
62
For an accurate display of the filter structural dimensions, detailed results of the amateur radio 23 cm
band filter dimensioning, according to the filter specifications of Section 4.2.2 and obtained with the
design equations that are presented in Section 3.2.3, are now presented
Input values of the equations are:
Constant Denomination Value 70 0.01 ./ 1.304e+09 . 1.3e+09 . 1.296e+09
b 14
Rejection band 21e+06
Table B.1 Input values of the amateur radio filter design
The following results are indexed by its numeration along Chapter 3.
Chapter 3 - Equation Result
(1) 5
(2) 6.25
(3) 0.0062
(4) 28.8462
(5) 6.7414
(6) 6.6761 6.7970 6.7860 6.7970 6.6509
(7) 2.8
(8) 7.1873 7.3682 7.3518 7.3682 7.1497
(9) 0.5134 0.5263 0.5251 0.5263 0.5107
(10) 6.8828 4.7437 5.3089 5.3221 5.3089
4.7303 6.8828
(11) 0.0075 0.0075
(12) 0.9181
(13) 1.0000 0.7563 1.3049 1.5773 1.3049
0.7563 1.0000
(14) 0.3090 0.8090 1.0000 0.8090 0.3090
(15) 7.4599
(16) 0.8171
(17) 0.0057 0.0039 0.0039 0.0057
(18) 0.6512 0.0428 0.0297 0.0297
0.0428 0.6512
(19) 0.4221 1.3459 1.4411 1.4411
1.3459 0.4221
63
Chapter 3 - Equation Result
(20) 0.4238 0.6646 0.6741 0.6741 0.6646 0.4238
(21) 17.9009 19.1665 19.1665 17.9009
(22) 117.8417
(23) 0.1808
(24) 1.0006
(25) 1.7490e-12
(26) 29.8462
Table B.2 Chebyshev combline filter design results for the amateur radio application
64
65
References
References
[1] Hong, J., Lancaster M. J., “Microstrip Filters for RF/Microwave Applications”, John Wiley &
Sons, Inc., 2001.
[2] Qingming Chen; Fang Li; Hongbin Cheng; Qing-Ming Wang, "Characteristics of dual mode
AlN thin film bulk acoustic wave resonators," Frequency Control Symposium, 2008 IEEE
International , vol., no., pp.609,614, 19-21 May 2008
[3] Matthaei, G. L., “Combline band pass filters of narrow or moderate bandwidth" Microwave J.,
82-91, 1963.
[4] Fragoso, M., “All Channels LTE Filtering System”, M.Sc. Thesis, Instituto Superior Técnico,
Lisbon, Portugal, 2013
[5] ANSYS, HFSS – High Frequency Structural Simulator,
(http://www.ansys.com/Products/Simulation+Technology/Electronics/Signal+Integrity/ANSYS
+HFSS)
[6] Encyclopædia Britannica, “The radio-frequency spectrum”, Alfred O. Hero III, 2014
(http://www.britannica.com/EBchecked/topic/585825/telecommunications-media/76251/The-
radio-frequency-spectrum)
[7] Natarajan, D., “A Practical Design of Lumped, Semi-Lumped and Microwave Cavity Filters”,
Springer, 2013
[8] Puglia, K.V., “A General Design Procedure for Band pass Filters Derived from Low Pass
Prototype Elements: Part I”, Microwave Journal, December 2000.
[9] Levy, R.; Snyder, R.V.; Matthaei, G., "Design of microwave filters," Microwave Theory and
Techniques, IEEE Transactions on , vol.50, no.3, pp.783,793, Mar 2002
[10] Almeida, D. X., “Inter-Cell Interference Impact on LTE Performance in Urban Scenarios”,
M.Sc. Thesis, Instituto Superior Técnico, Lisbon, Portugal, 2013.
[11] Dahlman, E., Parkvall, S., Sköld, J., “4G LTE/LTE-Advanced for Mobile Broadband”,
Academic Press, Oxford, United Kingdom, 2011.
[12] Ericsson, “ERICSSON MOBILITY REPORT – Interim Update”, August 2014
(http://www.ericsson.com/res/docs/2014/ericsson-mobility-report-august-2014-interim.pdf )
[13] 4G Americas, “Understanding 1G vs. 2G vs. 3G vs. 4G”
(http://www.4gamericas.org/index.cfm?fuseaction=page§ionid=361)
[14] GTI – Global TD-LTE Initiative, “TD-LTE Global Market”, 2014 (http://www.lte-
tdd.org/Resources/lgmd/2014-07-21/4003.html)
66
[15] GTI – Global TD-LTE Initiative, “LTE Industry Briefing”, July 2014 (http://www.lte-
tdd.org/d/file/Resources/pub/2014-07-24/99f1e19aae4a637d9a6d64d5817e8edb.pdf)
[16] ANACOM, Relatório final do leilão, 2012, (http://www.anacom.pt/render.jsp?categoryId=
344542)
[17] Yang Lan; Harada, A, "Power Allocation-Based Adjacent Channel Interference Reduction for
Coexisting TD-LTE and LTE-FDD Networks," Vehicular Technology Conference (VTC
Spring), 2013 IEEE 77th , vol., no., pp.1,5, 2-5 June 2013
[18] Höft, M., "Tunable Capacitive Coupling for Cavity Resonator Filters," German Microwave
Conference, 2009 , vol., no., pp.1,4, 16-18 March 2009
[19] ARRL – American Radio Relay League, Inc., “Amateur Radio: 100 Years of Discovery”, Jim
Maxwell, W6CF, Jan. 2010.
[20] ARAS - Associação de Radioamadores de Amadora-Sintra, “História do Radioamadorismo
Nacional - O Sr. José Joaquim de Sousa Dias de Melo, e o Progresso Das Comunicações”
(http://www.qsl.net/cs1aas/historia/melo.html)
[21] Arizona Amateurs on Television, “Welcome to Fast Scan Amateur TV”
(http://www.qsl.net/ki7cx/aatv_f.htm)
[22] Hellenic Amateur Radio Station SV8KOA, “IARU Region 1 HF - VHF - UHF BAND PLAN”,
(http://www.sv8koa.com/bandplan/)
[23] Boukari B., Vicente C.P., Cogollos S. “High selective H-plane TE dual mode cavity filter design
by using nonresonating nodes”, Microwave and Optical Technology Letters, Vol. 56, Issue 1,
pages 161–166, January 2014
[24] Hesselbarth, J., "Surface-mount cavity filter technology," Microwave Conference, 2007.
European, vol., no., pp.442, 445, 9-12 Oct. 2007
[25] Höft, M.; Burger, S.; Magath, T.; Bartz, O., "Compact combline filter with improved cross
coupling assembly and temperature compensation," Microwave Conference, 2006. APMC
2006. Asia-Pacific, vol., no., pp.781, 784, 12-15 Dec. 2006
[26] Zheyu Wang; Qingyuan Wang; Yanfen Zhai, "Design of an economical compact combline
filter," Microwave and Millimeter Wave Technology, 2008. ICMMT 2008. International
Conference on , vol.1, no., pp.308,310, 21-24 April 2008
[27] Yangping Zhao; Taijun Liu; Yan Ye; Like Cen; Haili Zhang; Xian Liu, "8th-order cavity filter
design with co-simulation by HFSS and Designer," Electronics, Communications and Control
(ICECC), 2011 International Conference on , vol., no., pp.2401,2403, 9-11 Sept. 2011
[28] Höft, M.; Yousif, F., "Orthogonal Coaxial Cavity Filters With Distributed Cross-Coupling,"
Microwave and Wireless Components Letters, IEEE , vol.21, no.10, pp.519,521, Oct. 2011
[29] Zakaria, Z.; Sabah, A; Sam, W.Y., "Design of low-loss coaxial cavity band pass filter with post-
manufacturing tuning capabilities," Business, Engineering and Industrial Applications
(ISBEIA), 2012 IEEE Symposium on , vol., no., pp.733,736, 23-26 Sept. 2012
67
[30] Yingying Chu; Jianyi Zhou; Fei Huang; Zhiqiang Yu; Weicheng Huang, "A design on radio
frequency front-end for LTE-A base-stations," Microwave and Millimeter Wave Circuits and
System Technology (MMWCST), 2013 International Workshop on , vol., no., pp.376,379, 24-
25 Oct. 2013
[31] Changsoo Kwak; Manseok Uhm; Inbok Yom, "Feasibility study on combline filter for tunable
filters," Microwave Conference Proceedings (APMC), 2013 Asia-Pacific, vol., no., pp.927, 929,
5-8 Nov. 2013
[32] Tingting C.; Jun D., "Coexistence Study and Interference Analysis in LTE Networks," Control
Engineering and Communication Technology (ICCECT), 2012 International Conference on,
vol., no., pp.751, 754, 7-9 Dec. 2012
[33] Yang L.; Harada, A, "Interference Analysis and Performance Evaluation on the Coexistence
of Macro and Micro/Pico Cells in LTE Networks," Vehicular Technology Conference (VTC
Spring), 2012 IEEE 75th , vol., no., pp.1,5, 6-9 May 2012
[34] Southgate – Amateur Radio News, “Potential Interference To Galileo From 23cm Band
Operations”, December 2011 (http://www.southgatearc.org/articles/galileo.htm)
[35] AMSAT-UK – Radio Amateur Satellites, “23 cm band and WRC-2018”, August 2014
(http://amsat-uk.org/2014/08/13/23-cm-band-and-wrc-2018/)
[36] Zhang Hualiang, “Compact, reconfigurable and dual-band microwave circuits”, PhD Thesis,
Hong Kong University of Science and Technology, January 2007.
[37] Serebryakova, E., “High-Power Comb-Line Filter Architectures for Switched-Mode RF Power
Amplifier Systems”, PhD Thesis, Fakultät für Elektrotechnik und Informationstechnik,
Technischen Universität Ilmenau, October 2012.
[38] Matthaei, G., Young, L., Jones, E.M.T., “Microwave Filters, Impedance Matching Networks
and Coupling Structures”, Dedham, MA: Artech House, 1964.
[39] Abidin, N., “Tunable Combline Band pass Filter”, B.Eng. Thesis, Faculty of Electronic and
Computer Engineering - Universiti Teknikal Malaysia Melaka, May 2008.
[40] International Amateur Radio Union - Region 1, “VHF MANAGERS HANDBOOK”, Version
6.11, August, 2013 (http://www.mrasz.hu/szakagak/urh/pdf/szabalyzatok/VHF_Handbook_
V6_12.pdf)
[41] Wikipedia, “PSK31”, (http://en.wikipedia.org/wiki/PSK31)
[42] Cziezerski, C., Solder Fumes, online blog (http://solderfumes.blogspot.pt/2012/02/
rearranging-enclosure.html)
[43] ANSYS, “Getting Started with HFSS: A Band pass Filter”, Canonsburg, May 2010
(http://www0.egr.uh.edu/courses/ece/ECE6351-5317/SectionJackson/5113/HFSS%20band
pass%20filter.pdf)
[44] Xiaoguang Liu; Katehi, L.P.B.; Chappell, W.J.; Peroulis, D., "High-Q Tunable Microwave
Cavity Resonators and Filters Using SOI-Based RF MEMS Tuners," Microelectromechanical
Systems, Journal of , vol.19, no.4, pp.774,784, Aug. 2010