Dual Mode Dielectric Resonator Filters

By Samantha Kerr

Department of ITEE

Submitted for the degree of Bachelor of Engineering (Honours) In the Division of Electrical Engineering (Biomedical Minor)

-i-

Acknowledgements
In undertaking this thesis I have been fortunate to have had valuable help from many talented individuals. I would like to take this opportunity to thank all of the many people that were involved in the production of this thesis. First I would like to thank my thesis supervisors Aleksander Rakic and Christine Blair for their invaluable assistance, encouragement, patience and guidance throughout the year. They are unquestionably the best supervisors an undergraduate student could have for their thesis. I would also like to personally thank the entire staff at Filtronic who have also assisted me. In particular I would like to thank Max Francis who initiated my interest in this topic and helped make it all possible. Thanks are also due to Tuan Phuong and Warren Smith for their technical expertise, cheerfulness, advice and encouragement in making this square wheel work. Not to forget, Mostafa Abushannan for his mastery of HFSS. Thank-you to the mechanical workshop men, in particular Ivan McKenzie and Grant Black, for their willingness and on the spot assistance. I appreciate the help of the skillful drawing office, with special mention to Tony Racher and Matthew Viviers. I also gratefully acknowledge Vanessa Walker for her continued support, ideas and knowledge despite not ever meeting in person. I would also like to acknowledge my family and friends for their support. Thanks in particular goes to my Mum for her painstaking review of my thesis. I also greatly appreciate the continual dedicated effort of my boyfriend, Steven Humphries, for his assistance, inspiration, enthusiasm, compassion and understanding.

-ii-

Abstract
This thesis describes the development and realization of a novel 2GHz band filter using dual mode dielectric resonators for cellular communications. Dual mode filters are a new and exciting technology in the microwave filter industry. These filters capitalize on the inherent property of dual mode dielectric resonators to exhibit two resonances. The motivation for this project has been inspired by the need for high performance, small size and low cost filters to ensure less expensive, more efficient and quality wireless communication systems. To satisfy these challenging demands dual mode filters provide an attractive solution. The thesis initially discusses and evaluates recent developments produced within the microwave industry on this topic through a concise review of literature. The main thesis objective was to design and construct a 2GHz bandpass filter based on the previous work of a triple mode filter developed by Hunter and Walker in 2000. Additionally the design for the dual mode filter had to be simple to allow for easy adaptation and manufacturing. With this objective, the techniques and methodology to develop the dual mode filter are then discussed and evaluated. To achieve the desired requirements both simulations and practical work were undertaken. A 3-D finite element method (FEM) electromagnetic simulation tool, HFSS by Agilent, was used to determine the optimum geometry of a dielectric resonator. The FEM modelling software was then assessed. It was found to be useful in producing approximate results although it had some limitations. Using a technique similar to single mode dielectric resonator filter development, Networks, a lumped-element modeling tool was then used to design a 2-section, single cavity filter. Next suitable methods of intra-cavity coupling were selected from previous research and assessed. A 4-section filter was also constructed using the methodology above although methods for inter-cavity coupling were also considered. From an evaluation of the filter, the outcomes included such significant advantages as a major size reduction, low cost and easy tuning ability. Some limitations of the design included low Q, spurious cross couplings and other spurious modes close to the passband. In conclusion this project offers a new and relatively simple alternative to common single mode ceramic filters with many more attractive advantages.

-iii-

Table of Contents
Chapter 1 1.1 1.2 1.3 1.4 Chapter 2 : Introduction................................................................................................... 9 Motivation for the Project................................................................................. 10 Project Specifications........................................................................................ 10 Nature of Expected Outcomes of Project.......................................................... 11 Outline of Thesis............................................................................................... 11 : Dielectric Resonator Filters ..................................................................... 13

An Overview..................................................................................................................... 13 2.1 2.2 2.2.1 2.2.2 2.2.3 2.3 2.3.1 2.3.2 2.4 2.4.1 2.5 Chapter 3 3.1 3.2 3.3 3.4 Chapter 4 4.1 4.2 4.3 4.4 Chapter 5 Dielectric Materials........................................................................................... 14 Dielectric Resonators ........................................................................................ 14 Field Patterns ............................................................................................ 14 Resonant Frequency.................................................................................. 17 Quality Factor ........................................................................................... 17 Dielectric Resonator Filters .............................................................................. 18 Coupling.................................................................................................... 18 Frequency Tuning the Dielectric Resonator ............................................. 19 Development of Single Mode Dielectric Resonator Filters.............................. 19 Coupling Bandwidths................................................................................ 20 Dual Mode Filters ............................................................................................. 21 HFFS Simulations......................................................................................... 24 Introduction....................................................................................................... 25 Simulation of Single Cavity.............................................................................. 25 Laboratory Evaluation of HFSS Simulations ................................................... 29 Limitations of HFSS ......................................................................................... 30 Dual Mode Filter Design .............................................................................. 32 Design Procedure Outline ................................................................................. 33 Input and Output Coupling Mechanisms .......................................................... 33 Frequency Tuning ............................................................................................. 33 Intra-cavity Coupling ........................................................................................ 34 Design of a Single Cavity, ............................................................................ 38

Two-Section Filter ............................................................................................................ 38 -iv-

5.1 5.2 Chapter 6

Single Cavity Design Procedure ....................................................................... 39 Evaluation of Single Cavity Filter .................................................................... 40 Design of Four Section ................................................................................. 42

Dual Mode Filter............................................................................................................... 42 Two Cavity Design Procedure ...................................................................................... 43 6.1 6.2 6.3 6.4 Chapter 7 7.1 Chapter 8 8.1 8.2 Inter-cavity Coupling ........................................................................................ 43 Networks Modelling ......................................................................................... 44 Four Section Dual Mode Filter Frequency Response ....................................... 46 Evaluation of Two-Cavity Filter....................................................................... 48 Results of Thesis ........................................................................................... 50 Results............................................................................................................... 51 Conclusions................................................................................................... 53 Future Developments ........................................................................................ 54 Conclusions....................................................................................................... 55

Appendix A Measurement Techniques.......................................................................... 56 Appendix B HFSS Simulation Results .......................................................................... 61 Appendix C Coupling Measurements............................................................................ 64 Appendix D - Filter Models and Plot............................................................................... 67 Appendix E Photographs ............................................................................................... 76 References......................................................................................................................... 78

-v-

List of Figures
Figure 2.1 Conventional cross-section of a typical dielectric resonator structure [1]. ..... 14 Figure 2.2 Cohn model structure of cylindrical resonator [1]. ......................................... 15 Figure 2.3 Field patterns for (a) TE01 and (b) HE11 [4].............................................. 15 Figure 2.4 Frequency vs distance for two common tuning mechanisms [7]. ................... 19 Figure 2.5 Dual mode dielectric resonator filter [10]. ...................................................... 22 Figure 2.6 Triple mode resonator structure developed by Walker and Hunter [13]......... 22 Figure 2.7 Field symbolizations [13]. ............................................................................... 23 Figure 3.1 Triple Mode DR in Single Cavity based on Hunter and Walkers Design ..... 26 Figure 3.2 Coupling Orientations (a) vertical loop coupling (b) horizontal loop coupling ................................................................................................................................... 27 Figure 3.3 Dual Mode Dielectric resonator in Cavity....................................................... 28 Figure 4.1Transformer rods used for input and output coupling...................................... 33 Figure 4.2 Orientation of tuning discs within cavity ........................................................ 34 Figure 4.3 Intra-cavity Coupling Mechanisms [11].......................................................... 35 Figure 4.4 Coupling Mechanisms ..................................................................................... 36 Figure 5.1 Degree 2 Chebyshev filter arrangement .......................................................... 39 Figure 5.2 Single Cavity Filter Frequency Response ....................................................... 40 Figure 5.3 Optimised Lumped Element Model of Single Cavity Dual Mode Filter ........ 41 Figure 6.1 Goal post Inter-cavity Coupling ................................................................... 43 Figure 6.2 Four Section Dual Mode Filter Layout ........................................................... 44 Figure 6.3 Filter Response from Network Analyser Narrowband Plot ........................... 46 Figure 6.4 Filter Response from Network Analyser Wideband plot ................................ 47 Figure 8.1 Smith Chart View after obtaining 180 phase shift of f0 ................................. 56 Figure 8.2 Smith Chart View after obtaining 360 phase shift of f0 ................................. 57 Figure 8.3 Group Delay QE measurements ....................................................................... 59 Figure 8.4 Group Delay K12 measurements..................................................................... 60 Figure 8.5 H-field Contour Plot ........................................................................................ 61 Figure 8.6 E-field Contour Plot ........................................................................................ 61 Figure 8.7 18 18 Z HFSS Simulation Results ........................................................... 62 Figure 8.8 19 19 Z HFSS Simulation Results ............................................................ 62 -vi-

Figure 8.9 20 20 Z HFSS Simulation Results ............................................................ 63 Figure 8.10 Cavity Height vs Modal Frequencies ............................................................ 63 Figure 8.11 Coupling Screw 1/4" *25mm Tuning Range ................................................ 64 Figure 8.12 Milled Coupling Screw 1/4" *25mm Tuning Range..................................... 64 Figure 8.13 12mm Metal Tuning Discs from Chamfer Tuning Range............................. 65 Figure 8.14 M4 Coupling Screw Tuning Range ............................................................... 65 Figure 8.15 1/4" Coupling Screw on Chamfer Tuning Range.......................................... 66 Figure 8.16 Networks Filter Response of 4.2MHz Filter ................................................. 68 Figure 8.17 Networks Lumped Element Model for 4.2MHz Filter.................................. 68 Figure 8.18 Improved Frequency Response of Single Cavity without Peek Support ...... 69 Figure 8.19 Networks Filter Response 10MHz for Four Section Filter ........................... 71 Figure 8.20 Networks 9.5MHz Filter Response ............................................................... 73 Figure 8.21 Networks 13MHz Filter Response ................................................................ 75 Figure 8.22 Single Cavity - lid removed.......................................................................... 76 Figure 8.23 Four Section Filter lid removed................................................................... 76 Figure 8.24 Four Section Filter close-up of left cavity................................................... 77 Figure 8.25 Four Section Filter......................................................................................... 77

-vii-

List of Tables
Table 3-1 Resonant Frequency Calculations by Ansoft and Agilent versions of HFSS .. 28 Table 3-2 Laboratory and HFSS Derived Results for different sized DRs....................... 29 Table 3-3 Lab. and HFSS derived results for different sized DRs in different cavities ... 30 Table 4-1 Coupling Mechanisms Tuning Range...36

-viii-

Chapter 1: Introduction
1.1 1.2 1.3 1.4 Motivation for the Project Project Specifications Nature of Expected Outcomes of Project Outline of Thesis

-9-

1.1 Motivation for the Project

Dual mode dielectric resonator (DR) filters are a relatively new technology in the world of microwave filters. Consequently there are many potential development opportunities to take advantage of in this field. The motivation for this thesis therefore stems from the ground-breaking development of these new improved dual mode filters and the need for more efficient filters for commercial applications. The basis for this project has been initiated by a commercial company Filtronic, a market leader in the research, design and manufacture of Radio Frequency (RF) products for cellular wireless communication systems. The main concentration of research and design at Filtronic is in the area of electronic filters for mobile phone base stations. These base stations use filters, to define transmit and receive bands to prevent interference problems. The performance of these filters is paramount to a quality, and efficient system. At present the filters Filtronic develop are only single mode ceramic filters. Filtronic wishes to expand their research into dual mode filters to continue their market leadership in the communications industry. Commercial interest in this project illustrates that the development of dual mode filters is a tangible and significant problem for RF filter designers. Dual mode DR filters have been developed that are superior to single mode DR filters. The research and development of these new dual mode filters has been initiated as a response to the challenging requirements placed on current electronic filters due to the rapid development of the cellular telecommunications industry, the need for smaller and more efficient systems and the overcrowded EM spectrum. Dual mode dielectric resonator filters are desirable compared to single mode filters as they exhibit two resonances. This property can be exploited to develop filters that have the same advantages of single mode dielectric filters while also exhibiting higher performance capabilities, are smaller in size and mass than single mode filters, have fewer components and, therefore, are less expensive to manufacture. In conclusion, this research and the subsequent development of dual mode dielectric filters will revolutionise the performance of the filters used in todays communications industry.

1.2 Project Specifications

The ultimate aim of the project is to produce a 4-section filter, 2 GHz bandpass filter using dual mode resonators with the following specifications outlined below. The specifications include: Passband: 1960-1980 MHz Insertion Loss: > -0.4 dB min Return Loss: < -25 dB min Size: 170 (L) x 55 (W) x 55 (H) mm (for cavity size of 50 x 50 x 40).

-10-

1.3 Nature of Expected Outcomes of Project

As this project is the first research project of dual mode resonators undertaken by Filtronic there were many outcomes projected. The results obtained from the initial work undertaken in this project were to be a starting point for further research and development. Some outcomes to be determined by the project were: Acceptable methods of input and output coupling, coupling between dual mode cavities and different modes within one cavity, tuning mechanisms. Methods of eliminating spurious modes. The maximum Q that can be obtained. Possible adjustment of networks (lumped element) model to adapt for dual mode filters, to account for cross coupling, spurious modes Evaluation of simulation techniques based on practical data obtained, determine the accuracy of simulation software. Determine if usual techniques used in single mode DR filters apply.

The aim of this project was to produce a simple design for a dual mode filter that is easy to adapt. Overall this project would determine the difficulties involved in producing dual mode filters and also be assessed to establish if it is a viable technique, that is profitable and easy to implement.

1.4 Outline of Thesis

This thesis documents the specifications, design and implementation of a dual mode dielectric resonator filter, developed by Samantha Kerr. Following this introduction, in Chapter 2 an overview of dielectric resonator filter theory is presented. It describes basic dielectric resonator theory and then addresses filter development. It also discusses and evaluates recent developments produced within the microwave industry through a concise review of literature. Chapter 3 combines the knowledge base developed in Chapter 2 to determine the optimum geometry of a dielectric resonator using HFSS by Agilent, a 3-D finite element method (FEM) electromagnetic simulation tool. The FEM modelling software is assessed based on practical results from measurements taken on a single cavity. The limitations of HFSS for modeling are then discussed. Next in Chapter 4, the design methodology for the development of a dual mode filter is discussed. Suitable methods of intra-cavity coupling are selected from previous research and assessed for suitability.

-11-

The initial design techniques, developed in Chapter 4 are applied in Chapter 5 and Chapter 6 to yield 2-section and 4-section filters respectively. Methods for inter-cavity coupling are also considered in Chapter 6 with respect to the 4-section filter. Both filters responses are evaluated in each chapter. The 4-section filter response is also compared to the desired specifications recognised previously in this introduction. With the completion of the dual mode filter design, Chapter 7 discusses the overall results of the project. Based upon the performance of the initial design, Chapter 8 discusses areas in which the design may be improved or extended, and also presents the final conclusions.

-12-

Chapter 2:
2.1 2.2

Dielectric Resonator Filters An Overview

Dielectric Materials Dielectric Resonators 2.2.1 Field Patterns 2.2.2 Resonant Frequency 2.2.3 Quality Factor Dielectric Resonator Filters 2.3.1 Coupling 2.3.2 Frequency Tuning the DR Development of Single Mode DR Filters 2.4.1 Coupling Bandwidths Dual Mode Filters

2.3

2.4 2.5

-13-

2.1

Dielectric Materials

Knowledge of the waveguiding properties of dielectric materials is not a new idea. Despite this fact, dielectric resonator filters have only been recently developed due to the previously poor characteristics of ceramic materials. Now with the advanced development of improved dielectric materials and the rapid expansion of cellular and satellite communications, dielectric resonator filters have been the source of much research.

2.2 Dielectric Resonators

A dielectric resonator is the basic unit of a ceramic filter. It is described as a shaped piece of high dielectric constant material, commonly known as a puck. The puck is usually supported by a structure made of low dielectric constant material and is surrounded by a conducting enclosure, see Figure 2.1.
Figure 2.1 Conventional cross-section of a typical dielectric resonator structure [1].

Eo

conducting enclosure high Er puck support

At the resonant frequency most of the electromagnetic energy is stored within the dielectric resonator. The support is used to ensure that there is no contact between the puck and the enclosure. The enclosure acts as a shield to prevent radiation and due to the pucks remoteness the resonant frequency is controlled by its cross sectional area and permittivity constant [1]. The most important characteristics of a dielectric resonator include: its field patterns, Q factor, resonant frequency and spurious free bandwidth [1]. These factors are dependent on the dielectric material used, the resonators shape and the resonant mode used. These factors are discussed in more detail below.

2.2.1 Field Patterns

There are many field patterns or modes present in a dielectric resonator dependent on the shape of the dielectric resonator. Despite this, the most commonly used puck that is widely accepted in industry is the cylindrical shaped dielectric resonator operating in the -14-

TE01 mode [1]. Cohn first described this mode in 1968. He used a simple model that approximated the fields inside the dielectric by assuming a Perfect Magnetic Conductor (PMC) covered all surfaces except the end-caps. This allowed energy to leak out from the flat surfaces and thus the problem was reduced to a simpler waveguide problem [2]. The structure can be seen below in Figure 2.2. conducting enclosure l1 E0 z Er 2a l1
Figure 2.2 Cohn model structure of cylindrical resonator [1].

The mathematical analysis is quite complex and the theory is detailed in Reference [3]. From this theory there are three basic modes present in a dielectric waveguide: Transverse Electric (TE); Transverse Magnetic (TM); and Hybrid (HE) modes. TE and TM modes do not contain electric and magnetic fields in the axial direction respectively. Hybrid modes contain properties of both electric and magnetic fields that can propagate in all directions. As mentioned previously the majority of applications use TE01 (in single mode applications) and also HE11 (for dual mode operation), as they are the two lowest modes to resonate when a DR is placed within conducting boundaries. When designing, modes are chosen to best suit the application. The field patterns for these modes can be seen in Figure 2.3.

DR DR E H

Figure 2.3 Field patterns for (a) TE01 and (b) HE11 [4].

-15-

The dimensions of the DR and the exterior boundary conditions imposed, e.g. tuning elements, shields and supports, determine which mode has the lowest resonant frequency or is fundamental [4]. These fundamental modes are desirable as most of their E field is confined within the DR, indicating a high Q characteristic. Q factor will be described later in this report.

-16-

2.2.2.

Resonant Frequency

At the resonant frequency of a DR, the magnetic field energy equals the electric field energy and electromagnetic waves can be transmitted with minimal loss. Only at a resonant frequency can electromagnetic fields be sustained. The resonant frequencies of the dominant mode and the spurious modes of a DR are important in the design of a DR filter. Spurious modes within the passband can degrade a filters response. Therefore the calculation of these parameters is required to achieve a desired response in a microwave filter. The resonant frequency and field patterns of a DR can be determined approximately by simple models such as Cohns, as well as the Itoh and Rudokahs models [3]. These models have limitations with regards to the accuracy of resonant frequency. The more sophisticated techniques are required in the design of DR cavities and filters and take into account the influence of the surrounding environment. Some more rigorous techniques have been developed to calculate a solution by excessive approximations converging to an exact solution. Therefore field distribution and resonant frequency can be calculated to a desired accuracy. These rigorous techniques are mode matching, finite element/finite difference and method of moments [5].

2.2.3 Quality Factor

The Quality factor (Q) is another important design parameter to consider in the design of DR filters. Q is determined by losses in a structure and in filter terms is a measure of bandwidth (as resonator bandwidth is inversely proportional to Q). It is also a major factor in insertion loss and the selectivity of a filter. In general, the Q factor relates to a resonant circuits capacity for electromagnetic storage with its energy dissipated as heat [6]. Q factor is a parameter often used to assess the performance of a dielectric resonator. It is determined by the equation below. Q = 2 Time average _ energy _ stored _ at _ resonant _ frequency Equation 1 Energy _ dissipated _ in _ one _ period _ at _ this _ frequency

From the determination of the Q factor, resonator bandwidth and loaded resonant frequency can also be calculated. In a resonant cavity acting as a load in a microwave circuit, several Q factors are defined. These Q factors are loaded Q (QE), unloaded Q (QU) and overall Q (QL). The loaded and unloaded Q accounts for external and internal losses respectively. The overall Q factor accounts for both internal and external losses. Furthermore the unloaded Q factor is due -17-

to losses in the resonator including power loss by conductors, dielectric fills and radiation. The loaded Q is a result of power loss due to an external load. A high Q is a desirable property of a DR, as it infers low losses.

2.3 Dielectric Resonator Filters

Dielectric resonators are the basic unit of a ceramic filter. Therefore the properties described previously are important in the design of a microwave filter. Many dielectric resonators are used to produce a filter. Equally as important in the design of microwave filters is the coupling, or the fields excitation relationships, between adjacent resonators and other microwave circuits (e.g. waveguides, loops, tuning screws etc.). Another factor that is vital in the design of ceramic filters is methods of tuning the individual resonant frequencies of DR to achieve a desired filter response. An accurate knowledge of these factors is required to develop dielectric resonator filters.

2.3.1 Coupling
A filter requires coupling at each end resonator to a terminating port. A probe that is codirectional to the external E-field of the resonator achieves this coupling. An alternative to this is to use a loop positioned with its area perpendicular to the H-field of the resonator. The choice of an appropriate structure to couple energy into and out of a DR depends on several variables: the mode to be excited; the transmission medium used (waveguide, microstrip, coaxial line, etc.); and also, the amount of coupling desired [3]. For example, modes with strong external magnetic fields, such as TE01, can be excited and achieve greater coupling through the use of a loop. This can be a circular sector of microstrip, a bent coaxial probe, or an offset straight microstrip [4]. Mutual coupling between resonators is also important as it determines the shape of the passband in a filter. In the case of single mode filters operating in the TE01 mode, design software usually works out the minimum coupling required to produce resonant frequency peaks close enough together to develop the passband. Coupling between resonators relies on the strongest external field, e.g. in the case of TE01, the magnetic field. One method of coupling is via irises, which are an aperture in the common wall between two cavities. Changing the dimensions of the iris is effective in varying the coupling between adjacent resonators. Coupling screws can also be used to introduce coupling between resonators.

-18-

2.3.2 Frequency Tuning the Dielectric Resonator

Another concept that is important in the design of DR filters is tuning. Resonators must be able to be tuned to a certain resonant frequency to produce a filter response. This is usually accomplished by approaching the DR with dielectric or metallic discs mounted on peek screws [6]. The effect of this is to perturb the outer electromagnetic fields of the DR to change the frequency. The screws are turned in or out to obtain a certain frequency. The method utilized will depend on the mode of the resonator to be tuned, the mechanical configuration of the device and the amount of tuning desired and its direction (i.e. to lower or higher frequencies) [3]. In a TE01 resonator, a metal disc approaching the resonator will raise the frequency and a dielectric disc will produce the opposite effect. This can be seen in Figure 2.4 below. When all screws are tuned to a certain configuration, a filter response can be attained. Coupling screws are tuned simultaneously to increase or decrease coupling between resonators to achieve the desired filter response.
1.2f0 Metal Disc Tuning

Tuning Disc Resonator

f0

Support

0.8f0

Dielectric Disc Tuning Distance from Resonator Figure 2.4 Frequency vs distance for two common tuning mechanisms [7].

2.4 Development of Single Mode DR Filters

Filtronic currently employs Networks, a computer analysis and modeling software, to develop filter responses. Cut and try prototyping is then used to obtain a physical filter based on the Networks output. Networks uses microwave filter theory to produce models for defined filter parameters, such as type (i.e. Chebychev), degree, Qu, centre frequency, bandwidth and required insertion and return loss. The number of resonators to use is determined by the degree of the filter. The minimum degree that meets the specifications is therefore desired and determined in the development of a filter. Networks produces a filter response, S11 and S21 plot, for the designer to determine if the filter meets the specifications, i.e. return loss and insertion loss.

-19-

Networks can also produce a model for the filter response. The output model can be selected from a range of identical mathematical representations of lumped element layouts. A useful layout commonly used is the Shunt Short Circuit Stub/J Inverter layout. This is a useful layout as Networks produces coupling bandwidth values when a synthesis is conducted that relate directly to the physical properties of a filter. The theory behind the calculation of coupling bandwidths by Networks for bandpass filters is explained below.

2.4.1 Coupling Bandwidths

A bandpass filter can be uniquely defined by its Qu, internal and external couplings, e.g. with tapped cavity filters, the external coupling, QE or K01, is determined by the tap point while the proximity and size of the resonators determines the internal couplings [8]. From this realization a filter can be developed based on these coupling values. The filters structures can be modified to obtain the correct coupling bandwidths and therefore the desired filter response. The concept of coupling bandwidths has been developed to simplify design methods. They mathematically represent admittance and impedance inverters (J or K) and are used as they can be measured using a Vector Network Analyser. Normalised coupling bandwidths can be calculated from low pass prototype g values (determined from filter degree and insertion loss requirements) as below. For theory underpinning low pass prototype and g values see [9]. To calculate normalized external couplings: q1 = g 0 .g1 (Usually q1 = q N ) To calculate the internal coupling: k i , i +1 , = 1 ( g i g i +1 for i = 1 to N-1
Equation 3

q1 = g N + g N +1

Equation 2

These coupling bandwidths can be given in units of frequency to easily relate to filter bandwidths as below: For the external couplings:

K 0 ,1 =

BW q1
-20-

K N , N +1 =

BW qN

Equation 4

Usually K0,1 = KN,N+1 = QE For internal couplings:

K i , j = k i , j BW
Equation 5

Networks calculates the coupling bandwidths, using this method described above, depending on the parameters it is passed. Networks also constructs a lumped element model, usually in the Shunt Short Circuit Stub/J Inverter layout. The J inverter is a model of the coupling between the resonators and is directly related to the coupling bandwidths calculated. The resonators are represented as shunted short circuit stubs. The coupling bandwidths that Networks calculates are then mirrored in a physical filter design. The transformer length and the tap height in a single mode dielectric resonator filter usually set input and output external Q values. The internal coupling bandwidths are set by the proximity of the resonators that is varied by using irises. Cut and try prototyping is then used to get the correct coupling bandwidths determined by Networks. The resonators are chosen to have a resonant frequency at the specified centre frequency. The cavity size is set to be twice the largest dimension of the dielectric resonator used [2]. The method above is used to develop single mode filters. To use this methodology, techniques to measure coupling bandwidths and external Q must be developed. These measurement techniques are outlined in Appendix A.

2.5 Dual Mode Filters

All the properties described above apply to both single and dual mode filters. Therefore standard design and synthesis methods can be utilised to develop such filters. Dual mode filters are a relatively new development of dielectric resonators. They are important in reducing the size of filters as they each dual mode resonator exhibits two resonances at the same frequency. In the past, the development of these filters was limited to high quality space applications because of high manufacturing costs involved. In 1982, Fedziuszko and Chapman presented the earliest model of a dual mode filter operating in HE11d mode to achieve low loss, high performance and a smaller volume. The HE11 mode was utilised as it can have a potentially high Q value as most of the Efield is confined to the dielectric [1]. Figure 2.5 below illustrates a typical dual mode filter configuration.

-21-

Iris Slots Dielectric Resonator

Figure 2.5 Dual mode dielectric resonator filter [10].

Traditionally dual mode filters have continued to operate with hybrid modes and also are usually constructed in a tubular cavity, where cylindrical resonators are loaded axially [11]. Although research has shown that planar structures, where a resonator is held perpendicular to either the top or bottom plate of the filter, have improved mechanical stability and are therefore more desirable [11]. Dual mode filters have been shown to have some disadvantages. The principle drawback is inferior spurious characteristics, although through careful resonator design this characteristic can be improved. The cost involved in producing specially designed resonators is also a disadvantage. Some researchers have tried to reduce costs in machining irises, which have also been found to be costly, by realising new configurations that do not require irises [12]. Despite these shortcomings, dual mode dielectric resonator filters are considered desirable due to their size and weight reductions compared to conventional single mode filters. Triple mode filters have also been developed to expand on the advantages of dual mode filters. Walker and Hunter in 2000 successfully produced a bandpass filter using triple mode TE01d dielectric resonators. They used the resonator structure consisting of a cube of ceramic material placed on an alumina support at the centre of a cubic cavity as shown below in Figure 2.6.

DIELECTRIC CUBE

ALUMINA SUPPORT
Figure 2.6 Triple mode resonator structure developed by Walker and Hunter [13].

-22-

This structure produces three degenerate modes orthogonal to each other due to the symmetry of the design. The field representation of the different fields can be seen in Figure 2.7. The resonant frequencies of the modes were determined, and the resonant frequencies of modes 1 and 3 were 2.6MHz lower than mode 2.

MODE 2

H FIELD MODE 3 E FIELD FIELDS FOR MODE 2

MODE 1

REPRESENTATION OF MODES 1, 2 AND 3

Figure 2.7 Field symbolizations [13].

Resonant frequencies of the modes were tuned independently via metal discs mounted on peek screws. Coupling loops were utilised to fine-tune the coupling. Coupling was also achieved by machining chamfers to create intra cavity coupling. This method of coupling by perturbing the stored electric energy of the resonator through cutting a portion of the DR was suggested by Wang in 1995. Other researchers have found this method of perturbing the field by creating chamfers to be more effective than using a cross coupling tuning screw at 45 degrees to the orthogonal modes [14]. This method is very effective as it avoids temperature compensation and dimensional problems. A problem associated with this design and triple mode filters in general is increased spurious resonant frequencies that are difficult to eliminate due to cross coupling between different modes. Converting the resonator structure into a dual mode TE01 resonator may reduce the complexity of this problem. The resonator structure developed by Walker and Hunter can be changed slightly to produce a dual mode TE01 cavity resonator. This can be accomplished by pushing mode 2 further away from the passband and using modes 1 and 3 as dual modes. This thesis project capitalises on this idea to produce an easy to produce, high Q, smaller and lighter dual mode filter. Furthermore the methods used by Walker and Hunter to couple and tune different modes can be used for this new resonator structure.

-23-

Chapter 3:

HFSS Simulations
3.1 3.2 3.3 3.4 Introduction Simulation of Single Cavity Laboratory Evaluation of HFSS Simulations Limitations of HFSS

-24-

3.1 Introduction
To determine the optimum geometry of the dielectric resonator puck simulations in HFSS were undertaken. Agilents version of HFSS is a simulation tool that uses the finite element method to calculate a full three-dimensional electromagnetic field within an arbitrary structure. From the generation of an EM field solution for a structure, HFSS can then compute S-parameters. For this project, HFSS was useful to calculate the resonant frequencies of a dielectric resonator within a cavity from S11 parameters.

3.2 Simulation of Single Cavity

Simulations were undertaken of a dielectric resonator placed within a single cavity. To reveal the resonant frequencies of a dielectric resonator within a cavity, the following procedure was undertaken. First a model had to be constructed with important parameters such as ports, boundary conditions and material properties inserted. To obtain S11 parameters to determine the resonant frequencies, a fast frequency sweep was initiated over a frequency range where the resonant frequencies were expected. The convergence criterion was set to a very small number; this was in order to control the accuracy of the solution through the number of adaptive passes undertaken. Next the simulation was run to allow for two adaptive passes. The results were then displayed on a plot of S11 parameters across the range indicated. The resonant frequencies are recorded as the frequency at which the S11 parameter dips. Further adaptive passes were run until the resonant frequencies stabilized. This was been found to be the best method for obtaining the most accurate estimate of resonant frequencies within an arbitrary structure. The initial cavity is shown below in Figure 0.1. It is consistent with the structure presented in 2000 by Walker and Hunter [13]. It is constructed from a 19*19*19mm dielectric resonator puck (Er = 43, tan = 10-5) suspended in an air-filled cavity of dimensions 50*50*50mm. The material properties of the dielectric resonator were set from Walker and Hunters model.

-25-

Figure 0.1 Triple Mode DR in Single Cavity based on Hunter and Walkers Design

Different probes were then utilized and analysed to obtain the best method of coupling. The input coupling mechanism was achieved by a coaxial cable, with impedance set to 50 Ohms through its dimensions to allow maximum transmission. The inner conductor was then extended within the cavity to form a magnetic loop or electric probe to obtain the lowest S11 parameters or best coupling. The only difference between mechanisms was a slight change of frequency and a significant change of S11 parameter depths as expected. Both magnetic and electric coupling mechanisms generated enough coupling to determine resonant frequencies. As both coupling mechanisms were sufficient to calculate resonant frequencies, the easiest to construct probe was used. This was the magnetic loop oriented orthogonal to the direction of the magnetic field. To ensure the detection of all resonant modes present in the cavity (as they are positioned orthogonal), two identical structures were produced with magnetic loops in both the vertical and horizontal position, see Figure 0.2.

-26-

Figure 0.2 Coupling Orientations (a) vertical loop coupling (b) horizontal loop coupling

The initial geometry of the Hunter and Walker puck was a cube inciting the triple degenerate TE01 mode. The HFSS simulation revealed the triple mode resonant frequency at 2.036 GHz with the first spurious at 2.63 GHz. To ensure these values were correct, they were compared to simulations made on Ansofts version of HFSS with an eigenmode solver. Ansoft simulations calculated the triple mode resonant frequency at 2.02 GHz and the spurious at 2.62 GHz. This revealed a maximum discrepancy of less than one percent between the versions, with the Agilent version having a slightly higher resonant frequency calculation. The Ansoft version has been found to give approximately correct results when tested practically. The method to calculate resonant frequencies with the Agilent version of HFSS was therefore found to be a valid approximation to the structures resonant frequencies. The next step was to vary the dielectric resonator geometry to obtain a dual mode rather than a triple mode resonance. First, Hunter and Walkers puck was modified by inserting a hole through the middle, see Figure 3.3.

-27-

Figure 0.3 Dual Mode Dielectric resonator in Cavity

This was for two reasons: for mechanical reasons, to allow the puck to be supported sufficiently in a fixed centred position; and also to split the degenerate triple mode into a dual mode and single mode dielectric resonator. Simulations on both Ansoft and Agilents version of HFSS calculated the single mode, dual mode and spurious to be the values listed below in Table 0-1. E and H field contour plots of the dielectric resonator were also performed, with the loop in the vertical position, to reveal that the coupling mechanism and the dual mode resonant frequency were correct. These can be seen in Figure 0.5 and Figure 0.6 in Appendix B.
Table 0-1 Resonant Frequency Calculations by Ansoft and Agilent versions of HFSS

HFSS Version Ansoft Agilent

Single mode Frequency (GHz)

2.06 2.09

Dual mode Spurious Frequency (GHz) Frequency (GHz)

2.25 2.28 2.75 2.93

Unfortunately in this process the single mode has been moved to below the dual mode, the optimum position would be above dual mode with the spurious modes. Further simulations were undertaken to determine the best geometry of the dual mode puck with the above considerations in mind. The results from the different dielectric resonator shapes are shown in Figure 0.7, Figure 0.8 and Figure 0.9 in Appendix B. As the single mode did not move to a frequency above the dual mode resonance the best possible geometry was the 14*19*19mm puck with the single and spurious spaced as far from the dual mode as possible to create minimum interference. Three different dielectric resonator shapes were then sent to Wolverhampton, in the UK, to be processed and constructed. These were the 14*19*19mm, two 19*19*19mm pucks one of those with a chamfer designated by Walker and Hunters design. In the meantime, further simulations were performed to determine the effect of a reduced cavity height on the resonant frequencies. The 50*50*50mm cavity was found to have the -28-

best spurious response. This can be seen in graphical form in the Figure 0.10 in Appendix B.

3.3 Laboratory Evaluation of HFSS Simulations

An evaluation of the simulations in comparison to practical results was first undertaken to determine the accuracy and therefore the effectiveness of HFSS as a design tool. An aluminium cavity (50*50*50mm) was built with the dielectric resonator supported by a PEEK screw through the middle. The 19*19*19mm without the chamfer and also the 14*19*19mm dielectric resonators were examined. The input coupling was made from an SMA connector pin soldered to a piece of wire acting as an electric probe. The electric probe was oriented vertically to couple to the dual mode and horizontally to couple to the single mode.

Table 0-2 below shows the resonant frequencies obtained from the Network Analyser in a laboratory measurement and also the HFSS derived result of the modes with the 19*19*19mm and 14*19*19mm dielectric resonator inserted into the cavity.
Table 0-2 Laboratory and HFSS Derived Results for different sized DRs

Single mode Frequency (GHz) 19*19*19mm Dielectric Resonator Lab. Measurement HFSS Derived Result 14*19*19mm Dielectric Resonator Lab. Measurement HFSS Derived Result 2.04 2.09

Dual mode Spurious Frequency (GHz) Frequency (GHz)

2.22 2.28

2.72 2.93

2.17 2.20

2.59 2.66

3.10 3.11

Further measurements were taken of resonant frequencies with the same dielectric resonators inserted in a 40*50*50mmm cavity. The results from both cavities and dielectric resonators using a electric probe input coupling can be seen below in Table 0-3.

-29-

Table 0-3 Lab. and HFSS derived results for different sized DRs in different cavities

19*19*19mm Dielectric Resonator 50*50*40mm Lab. Measurement Cavity HFSS Derived Result 50*50*50mm Lab. Measurement Cavity HFSS Derived Result 14*19*19mm Dielectric Resonator 50*50*40mm Lab. Measurement Cavity HFSS Derived Result 50*50*50mm Lab. Measurement Cavity HFSS Derived Result

Single mode Dual mode Spurious Frequency (GHz) Frequency (GHz) Frequency (GHz) 2.05 2.10 2.04 2.09 2.23 2.3 2.22 2.28 2.73 2.88 2.72 2.93

Single mode Dual mode Spurious Frequency (GHz) Frequency (GHz) Frequency (GHz) 2.18 2.22 2.17 2.20 2.6 2.68 2.59 2.66 3.02 3.22 3.10 3.11

3.4 Limitations of HFSS

This method and HFSS software in general have some limitations. The main problem with HFSS is that the program is very computer intensive and progress becomes very slow when the computer memory limits are approached. In fact, progress is discontinued and convergence errors arise if there is insufficient memory, which was a regular occurrence. The method for obtaining resonant frequencies, mentioned above, is also very tedious. The overall process of modeling EM fields within the simplest cavity is very slow. Another major disadvantage with HFSS is that the results are dependent on the input coupling mechanism. Therefore awareness of the expected field distribution in a structure to enable sufficient coupling to modes is essential. Additionally, from these results it can be seen that the resonant frequencies HFSS derives are only approximate. They are consistently higher than the laboratory measurements. Although the results are only approximate, they are accurate enough for modeling. A source of inaccuracy in the HFSS calculated resonant frequencies may have been due to the model cavity not being an accurate representation of the real cavity. Some errors could have been generated as the PEEK screw support was not included and the actual cavity didnt have straight 90 corners, as they are too difficult to manufacture. Further simulations should be undertaken to determine if these inconsistencies were a sources of error. -30-

In conclusion, Agilents version of HFSS is a suitable modeling tool for dielectric resonator structures. It must be understood that the HFSS results are only approximate and resonant frequencies are calculated to be above real measurement values. Some disadvantages of modeling with HFSS are the procedure for calculating resonant frequencies is tedious and the dependence of the solution on the coupling mechanism. Ansoft version of HFSS seems to be a more accurate modeling tool and also quicker. Further work should be conducted to evaluate the software programs for modeling dielectric resonator structures.

-31-

Chapter 4 :

Dual Mode Filter Design

4.1 4.2 4.3 4.4 Design Procedure Outline Input and Output Coupling Mechanisms Frequency Tuning Intra-cavity Coupling

-32-

4.1 Design Procedure Outline

In order to produce a filter some means of creating input, output and intra-cavity coupling mechanisms, that were tunable, had to be developed and evaluated. An assessment was undertaken by measuring the range of coupling bandwidths achieved by each method, and how simple the design was to use and manufacture. Also a method for effective tuning of the two resonant frequencies had to be investigated.

4.2 Input and Output Coupling Mechanisms

With the objective to create a product that is easy to manufacture, the input coupling mechanism was changed. A transformer was used to replace the electric wire probe. The advantages of using a transformer as opposed to the electric probe include: it is more mechanically stable; and easier to implement and repeat. The transformer was a rod made of brass, smaller than the length of the cavity, with a hole at the bottom to enable fixing to the cavity and a hole through the middle to enable insertion of an SMA pin, see Figure 0.1 below. It was placed vertically within the cavity to couple to the electric field of the dual mode resonance.
Transformer

SMA connector

DR

Figure 0.1Transformer rods used for input and output coupling

Using this method, the input and output external Q could be modified by either changing the tap height, the transformer length or inserting a screw close to the transformer. The tap height was the height at which the SMA pin was connected to the transformer.

4.3 Frequency Tuning

-33-

From Walker and Hunters investigation into tuning mechanisms for a triple mode filter the best option was to use a dielectric tuner, Er = 44, mounted on a peek screw. This method was found to be beneficial as it does not decrease the unloaded Q of the mode and also offers different effects on the resonant frequencies depending on the diameter of the tuner. When tuned in, the tuner perturbs the fields, splits the dual mode and will drive the modes down in frequency as for a single mode TE01 resonator. The available dielectric tuners were of diameter 19mm and were sufficient to have a tuning range of approximately 61 MHz. Although there is some differential tuning of the other mode of 29MHz, the modes can be tuned almost independently. The tuning discs were oriented as shown in Figure 0.2 below to offer maximum effect on the tuning of the dual modes.
Figure 0.2 Orientation of tuning discs within cavity DR Dielectric Tuners

SMA connector

4.4 Intra-cavity Coupling

There were several methods described in the research literature to create the coupling between the dual modes in each cavity. In [11], most of the coupling was created by chamfers (shown in Figure 0.3) at some edges of the puck depending on what modes were to be coupled. Fine-tuning of the chamfers was achieved by coupling loops and metal tuning screws. They were found to decrease the coupling created by the chamfer if placed next to or 180o away from the chamfer (position 1 and 3) or alternatively increase the coupling when in position 2 and 4.

-34-

#2 #1 #4

#3 1-3 Coupling

Fine tuning of 1-3 Coupling 2

3 1

1-3 Coupling Figure 0.3 Intra-cavity Coupling Mechanisms [11]

These coupling mechanisms were attempted in the single test cavity (50*50*50 mm) with the 19*19*19 mm puck with a chamfer. The amount of coupling was measured using the methods described in Appendix A by measuring the coupling bandwidth, K12. As a starting point from Walker and Hunters design the chamfer was set at 1.7mm. With this chamfer and both the transformers in place, K12 was measured to be approximately 9.1MHz. Next the influences of the different coupling mechanisms were measured with respect to coupling bandwidth and input external Q. The mechanisms measured were an M4 screw above the DR chamfer through the top of the lid, a normal *40 TPI screw suspended from the middle of the lid, a milled *40 TPI screw from the center of the lid, metal tuning disc from a chamfer made on the cavity, a *40 TPI screw from the chamfer and a wire loop mounted on peek placed in the centre of the lid. See Figure 0.4 below.

-35-

Figure 0.4 Coupling Mechanisms

The measurements can be seen in graphical form from the various mechanisms in Figure 0.11 - Figure 0.15 in Appendix C.

Table 0-1 below demonstrates the range of tuning depending on the mechanism utilized.
Table 0-1 Coupling Mechanisms Tuning Range Coupling Tuning Mechanism Tuning Range (MHz) M4 Screw near DR Chamfer -2.27 " Screw at Chamfer -5.13 Milled Screw -6.78 1/4" Screw from top -7.02 Wire Coupling Loop 9.53/-8.00 Metal Tuning Discs at Chamfer -1.09 *sign of coupling can be negative or positive depending on position in cavity.

The coupling mechanism with the widest range was the wire loop mounted on peek screw. Although this mechanism by far had the best tuning range, it had some limitations. -36-

These being that the loops are difficult to reproduce, therefore difficult to manufacture and also they are difficult to tune. The difficulties in tuning arise as different coupling is produced from a change in depth and also loop orientation. The best alternative was therefore the metal screw placed in the middle of the lid. For fine-tuning, the M4 screws from the top of the lid placed closest to the dielectric resonators chamfer were the best option. As the devices for frequency and coupling tuning and input and output coupling mechanisms were investigated and the best methods found, the next stage was the design of a single cavity, two-section filter.

-37-

Chapter 5 : Design of a Single Cavity, Two-Section Filter

5.1 5.2 Single Cavity Design Procedure Evaluation of Single Cavity Filter

-38-

5.1 Single Cavity Design Procedure

A similar technique used in developing single mode filters, as outlined in Chapter 2, was undertaken to produce a single cavity two-section filter with dual mode resonators. First, a lumped element Networks model was set-up to gain a degree 2 Chebychev filter response with a K12 of 9.1MHz. This was performed to calculate the unknown input and output external Q values to obtain the filter response, as K12 was already set from the chamfer on the dielectric resonator. The center frequency was set to the resonant frequency of the dual mode. The bandwidth of the filter was then modified until the criteria of K12 = 9.1MHz was met. The Networks component list, frequency response (Figure 8.16) and lumped element model (Figure 0.17) can be seen in Appendix D for the 4.2 MHZ bandwidth filter that met the above requirements. The input and output external Q values were found to be 8.6MHz. The following layout of the single cavity filter was used, see Figure 5.1, based on the previous laboratory evaluation of the input/output coupling, intra-cavity coupling and frequency tuning mechanisms. A photo of the single cavity filter can be seen in Appendix D in Figure 0.22.
Output DR

Input

Dielectric

tuners

Figure 5.1 Degree 2 Chebyshev filter arrangement

As the original input QE was measured to be 22.7MHz with a tap height of 25mm, the tap height had to be reduced to obtain the desired 8.6MHz determined from the model. Assuming the relationship between QE and tap height was linear, the new tap height to obtain 8.6MHz was calculated to be 10mm. The tap height was adjusted and the filter was tuned to obtain the response, obtained form the vector network analyzer, shown in Figure 5.2.

-39-

CH1 S 11 CH2 S 21 PRm 1 Cor

log MAG log MAG

5 dB/ 10 dB/ 3

REF 0 dB REF 0 dB

20 Sep 2001 20:44:07 3_:-27.604 dB 3_ -.6508 dB 2 213.550 000 MHz 1_:-11.632 dB 2.209 GHz 2_:-14.012 dB 2.216 GHz

MARKER 3 2.21355 GHz

2 1 2

3 PRm Cor

1_ -1.2172 dB 2.209 GHz 2_ -.9231 dB 2.216 GHz

START 2 170.000 000 MHz

STOP 2 270.000 000 MHz

Figure 5.2 Single Cavity Filter Frequency Response

5.2 Evaluation of Single Cavity Filter

This response was expected except for the addition of a zero on the low side. Spurious coupling, from the transformers, may have caused this. The return loss gained was at 27.7dB, further below the specifications at -25dB. Conversely, the insertion loss was quite high at -0.48 dB. This was due to the losses in the filter. Some sources of loss may be attributed to: cavity not being silver plated, dirty cavity; or low unloaded Q of cavity. To determine the source of the zero and also the unloaded Q, Networks was used in its optimization mode. In this function of Networks, the plot of the filters response from the network analyser was imported as a device under test (DUT) and a model was set up to

-40-

match the DUT. From optimization, a lumped element model was obtained and the unloaded Q deduced. The optimised lumped element model can be seen below, in Figure 5.3.

Figure 5.3 Optimised Lumped Element Model of Single Cavity Dual Mode Filter

From this model, it can be seen that there is a cross coupling from the input transformer to one of the dual modes. This cross coupling is very difficult to eliminate. As it is causing a steeper slope on the low side, an advantage, this amount of spurious cross coupling was accepted. The other parameter calculated by Networks was the unloaded Q. Unfortunately it was found to be 3500, which was extremely low. This was a valid reason for the high insertion loss. An investigation was then undertaken to determine the reason for the low Qu. To begin, the unloaded Q was determined from the network analyzer to establish if it was an error in the Networks model. The Qu was resolved by the network analyzer after replacing the transformers by an e-probe and achieving critical coupling. The Qu from this method was found to be 5732. This indicated that the Networks model was correct. The low Qu was found to be from the insertion of the PEEK screw support through the middle of the dielectric resonator. When the screw was replaced by an alumina support, the Qu increased to 11940. The alumina support was then glued to the dielectric resonator with araldite for mechanical stability. A plot of the single cavity, 2-section filter with DR glued to alumina support obtained form the vector network analyzer can be seen in Figure 0.18 in Appendix D. Although the return loss is not as low, due to the wrong tap heights for the input and output transformers the insertion loss increased to 0.34dB, a great improvement. -41-

Chapter 6: Design of Four Section Dual Mode Filter

6.1 6.2 6.3 6.4 6.5 Two Cavity Design Procedure Inter-cavity Coupling Networks Modelling Four Section Dual Mode Filter Frequency Response Evaluation of Two-Cavity Filter

-42-

6.1 Two Cavity Design Procedure

The same process for the design of the single cavity dual mode filter was undertaken for the four-section, two cavity filter. The only deviation from the original design process was to determine and evaluate methods of obtaining inter-cavity coupling.

6.2 Inter-cavity Coupling

In Hunter and Walkers design, a long and narrow vertical iris achieved inter-cavity coupling. This method is undesirable as too much spurious coupling is created. As a Chebychev filter is to be realized, only one mode should couple between the cavities. Instead of this method, a goal-post inter-cavity coupling design was implemented, see Figure 6.1 below.
PTFE

Wire

Transformer Rods Figure 6.1 Goal post Inter-cavity Coupling

A piece of wire was inserted into a tube of PTFE and set up across two transformers in different cavities. This can allow a short, narrow iris preventing unwanted spurious coupling. The amount of coupling can then be adjusted by the usual change of tap height, insertion of a screw in the lid above the transformers or a reduction of transformer height. The ideal method would be to have no iris and instead drill a single hole the diameter of PTFE shroud. This cannot be implemented, as it is not easily manufactured. The filter arrangement can be seen below, Figure 6.2. It has two identical mirrored cavities joined by a narrow, short iris with goal post transformers.

-43-

Input
Dielectric tuners

Output DR

Dielectric tuners

goal post coupling

Figure 6.2 Four Section Dual Mode Filter Layout

6.3 Networks Modelling

A networks model was set up to determine the coupling bandwidths required to realize a degree 4 Chebychev filter. A 10MHz bandwidth filter was chosen as it had coupling bandwidths near the values of the previous single cavity filter. The lumped element model components list and response (see Figure 0.19) can be seen in Appendix D. Using the single mode cavity measurements for indications of the input, output and intracavity coupling values, the filter was then designed to match the Networks model coupling bandwidths. First the input and external coupling parameters were determined. Using the same linear relationship between QE and tap height, a tap height of 12mm was calculated for a desired QE of 13.27MHz. Next the intra-cavity coupling bandwidths (K12 and K34) had to be increased from the original 9.1 MHz to 10.4 MHz to fit the Networks model. This was to be achieved by using the coupling loop. Although previously it was decided that the metal screw in the center of the cavity was the most effective coupling tuning mechanism, it could not be used in this case as it can only decrease the coupling. Another method that could be used to increase the coupling would be to use an M4 Screw placed in a position 180o away from the chamfer. This unfortunately would create too much spurious frequencies due to it coupling to the input transformer. The coupling loop was therefore the only available option to achieve this increased coupling despite the difficulties in tuning. The ideal method would be to use another dielectric resonator with an increased chamfer; this would avoid all of the problems experienced above.

-44-

Finally the inter-cavity coupling bandwidth was set up. This was to be achieved by the goal post coupling mechanism. The iris width was set as the PTFE tube width and the depth was set as the tap height minus the PTFE tube width. As measurements couldnt be taken for this inter-cavity coupling parameter on the single cavity, another method to determine approximate coupling bandwidth values for K23 had to be devised. Since the goal post topology is similar to the input and output transformer coupling methods, the coupling bandwidths were to be approximated by using the tap height vs QE relationship. To obtain the required 7.7MHz of coupling, a tap height of 6.97mm was calculated. Unfortunately this tap height was too low to be manufactured. To solve this problem, the iris was cut down to the minimum height that could be manufactured which was 20mm. This would allow the minimum tap height to be 22.1mm.This would produce a larger K23 than required, to reduce this coupling the transformers were increased to the maximum height, 49mm. These heights were kept as starting points for the initial twocavity filter. K23 could be modified later by a change of transformer and tap height or the insertion of a screw positioned above the transformer. The two-cavity filter was then manufactured from aluminium with the tap heights and coupling arrangements determined above. Next measurements of coupling bandwidths were made. K23 was measured with and without the goal post coupling mechanism. There was no coupling measured through the iris only, without the goal post coupling mechanism in place. The values of the coupling bandwidths with the goal post mechanism implemented were: QEin = QEout = 11.79 MHz K12 = K34 = 13.46 MHz K23 = 7.34MHz At this stage as K23 is difficult to tune, due to the transformer lengths being at their maximum and a screw cannot be inserted above the transformer, a new model to match this value was developed. A 9.5MHz filter best matched the inter-cavity coupling bandwidth. The model and frequency response can be seen in Figure 0.20 in Appendix D. As the current K12 is too large, this will be decreased by the coupling loop as it has the greatest tuning range. The input and output external Q values are approximately the same for the new model; these will be altered later by increasing the tap height. Unfortunately K12 was found to be too difficult to set with the coupling loops, so another model was developed on Networks to match K12. The final model developed to match K12 was a 13MHz filter, see Figure 0.21 in Appendix D for Networks component list and response. Again the QE input and output values and K23 needed adjustment. QE needed to be increased to 17.25MHz from the initial 11.79MHz. To achieve this the transformer length was decreased by 1mm. Another method that could increase the QE would be to use a higher tap height. The reduction of transformer length was preferred as a coupling screw can always be inserted above the transformer to reduce the coupling bandwidth to its former value. The inter-cavity coupling, K23, needed to be increased from 7.34MHz to 10.03MHz. This coupling

-45-

bandwidth was also increased by a reduction of the goal post transformers length. After these final adjustments a filter response was obtained.

6.3 Four Section Dual Mode Filter Frequency Response

The narrowband and wideband responses obtained form the vector network analyzer can be seen in Figure 6.3 and Figure 6.4 respectively. Photos of the four section filter can be seen in Figure 0.22 in Appendix E.
Figure 6.3 Filter Response from Network Analyser Narrowband Plot

2: dB 5 1: -5

1:Reflection 2:Transmission

Log Mag 5.0 dB/ Ref 0.00 dB 2 Log Mag 10.0 dB/ Ref 0.00 dB 1 3

C C

2 -10 -15 -20 -25 -30 -35 Center 2 220.000 MHz 1:Mkr (MHz) dB 1: 2215.2333 -23.036 2: 2228.6000 -22.524 3> 2220.8667 -22.843 Span 100.000 MHz 2:Mkr (MHz) 1: 2215.2333 2> 2228.6000 3: 2220.8667 dB -0.853 -0.553 -0.508 1 3 2

-46-

1:Reflection 2:Transmission dB 10 3 5 -10 -20 5 -30 -40 -50 -60 -70 Center 2 220.000 MHz 1:Mkr (MHz) dB 1: 2215.2333 -18.912 2: 2228.6000 -19.675 3> 2027.6000 -1.963 4: 2641.0667 -1.472 5: 1746.3333 -7.627 3

Log Mag 5.0 dB/ Ref 0.00 dB Log Mag 10.0 dB/ Ref 0.00 dB

2 1

2 4 1 2

Span 1 000.000 MHz 2:Mkr (MHz) dB 1: 2215.2333 -5.021 2: 2228.6000 -4.459 3: 2027.6000 -33.108 4: 2641.0667 -32.498 5> 1746.3333 -11.082

Figure 6.4 Filter Response from Network Analyser Wideband plot

-47-

6.4 Evaluation of Two-Cavity Filter

The response attained was a good achievement for a first attempt of a dual mode filter, although there are some limitations to this design. The response is irregular; the low side rejection is lower than that on the high side. This response was to be expected as the single cavity filter exhibited a similar response due to the zero on the low side. Once again, spurious coupling from the transformers is likely to have caused this. From the wideband response the first spurious on the high side was at 2.64GHz, closer than the measured 2.72GHz in the single cavity. The low side of the filter showed some more spurious responses. The first was experienced at 1.75GHz, this was probably the transformers resonating. The next, causing the steep rejection on the low side was at 2.027GHz. This is probably due to spurious coupling between modes, or the transformers. These spurious couplings will have to be eliminated to gain a symmetrical filter response. From an evaluation of the filter based on the initial specifications it became apparent that some of the specifications were unrealistic at this stage in the design. With this in mind, the filter was assessed against the initial specifications: Specifications Passband: 1960-1980 MHz Insertion Loss: > -0.4 dB min Return Loss: < -25 dB min Size: 170 (L) x 55 (W) x 55 (H) mm (for cavity size of 50 x 50 x 40).

The specified band for the filter was too low for the dielectric resonator chosen; another dielectric resonator would have to be designed to suit this band. Application to this specification was not considered essential at this stage in the design process. Also the bandwidth of the filter was decreased from the initial 20MHz to 13MHz, this was due to the difficulties experienced in increasing the K23 coupling. An increase in the coupling could have easily been achieved by an increase of chamfer, without increasing spurious coupling. A new resonator would have to be manufactured to achieve this increased coupling and therefore produce a 20MHz bandwidth filter. The filter failed on the specifications regarding return loss and insertion loss. The return loss gained was -22.8dB, unfortunately above the specifications at -25dB. Also, the insertion loss was quite high with a minimum of 0.489 and a maximum of -0.844 dB on the low side, due to the high rejection. The minimum insertion loss is still quite high; silver-plating the cavity and using aluminium transformers rather than brass could reduce these losses. The high losses could also be due to a low unloaded Q of the cavity. The unloaded Q was calculated to be 5500. This is very likely to be the cause for the unexpected low insertion loss and high return loss.

-48-

The filter was predominantly within specifications with respect to the size. Although the cavity size height was larger, the overall filter bodys length was much smaller than the specification. This size reduction is a significant achievement. Despite the limitations identified, with respect to the specifications, there are some clear advantages of this filter compared to the single mode equivalent. These are: a significant size reduction, the dual mode filter is almost half the size of an equivalent single mode filter. low cost, due to the reduction of resonators and cavity size easy tuning ability, as there are less components to tune.

-49-

Chapter 7: Results of Thesis

7.1 Results

-50-

7.1 Results
This thesis revealed the difficulties in producing a dual mode filter. It also established that dual mode filters are more profitable to implement than single mode filters, although further work needs to be undertaken to gain a commercially viable design. The design procedure, using simulations and then undertaking practical work, was effective. Although the method Agilent HFSS used to calculate resonant frequencies is very tedious, the simulations provided an accurate basis for further practical work. The simulation process may have been quicker if Ansofts version of HFSS (with an eigenvalue solver) was employed to solve resonant frequencies. The Ansoft version has demonstrated greater accuracy. In future work this software, Ansoft HFSS, should be considered for determining resonant frequencies rather than Agilent HFSS. The usual design process in developing single mode filters has also been found to be useful in producing dual mode filters. This process uses Networks to calculate a desired response and the required coupling bandwidths and then a filter is developed to approximate these coupling bandwidths. The manufactured filter coupling bandwidths are then measured and improved if they do not match. This method is fairly quick and accurate to produce a desired filter response. A major outcome of this research was to determine acceptable methods of input and output coupling, coupling between dual mode cavities and different modes within one cavity and tuning mechanisms. The most effective of the many methods experimented have been identified. These were: For tuning dielectric discs mounted on PEEK screws For intra-cavity coupling a chamfer on dielectric resonator and also for fine tuning a tuning screw mounted above center of dielectric resonator. For inter-cavity coupling goal post transformer arrangement For input and output coupling transformers These mechanisms were chosen as they are simple to tune and easy to manufacture. The unloaded Q, Qu, was also determined for both the single and two-cavity filter. It was found that PEEK inserted through the dielectric resonator drastically reduced the Qu, instead an alumina support was used. The Qu for the single cavity was low but acceptable, unfortunately the Qu for the two cavity filter was calculated to be significantly lower. To improve the Qu, the filter and transformers should be silverplated. A more accurate representation of the maximum Qu that could be achieved using dual mode dielectric resonators could be determined. An adjustment of the networks (lumped element) model to adapt to dual mode filters was undertaken for the single cavity filter. This should be undertaken for the two-cavity filter to determine sources of spurious cross coupling. Further work could include adjusting networks model to account for dual mode resonators in general not just for specific cases.

-51-

Unfortunately no methods of eliminating the spurious modes present were determined. Further research should be undertaken in this area. Overall, this project determined that the advantages of dual mode filters far outweigh the traditional single mode filters. Dual mode filters are definitely the way of the future. Although, more research needs to be undertaken to understand dual mode filters completely, once understood they will be a very profitable venture.

-52-

Chapter 8:Conclusions
8.1 8.2 Future Developments Conclusions

-53-

8.1 Future Developments

To produce a commercially viable dual mode filter further research and development of the initial design must be undertaken. The limitations of the initial design must be first addressed. This includes determining methods of eliminating spurious coupling, improving the low Qu. Further simulations for an improved DR could also be undertaken. Sources of spurious coupling must be identified. An optimization of the two-cavity filter would achieve this. Transformers and the goal post inter-cavity coupling should be included in the Networks model. Once identified, these sources of spurious coupling can be eliminated. Mechanisms to eliminate unwanted coupling should also be investigated. The source of low Qu should also be investigated. First the cavity and transformers should be by silver-plated and change in Qu observed. If the Qu doesnt improve significantly, other sources of loss should be investigated, identified and eliminated. Greater degree filters should also be designed and constructed. The development of a 6 section filter would be the next stage in the design process, once the above problems are addressed. This would be advantageous, as filters in industry can require greater than 4sections. A different approach to the design could include further simulations using Ansoft version of HFSS to determine another improved resonator shape. This improved DR would have an improved frequency response, i.e. the single mode resonance positioned on the high side of the dual mode resonant frequency. This dielectric resonator could have a different hole diameter, a flatter and/or a wider shape. The cavity could also be decreased to obtain a smaller cavity for this new and improved resonator. A procedure to improve repeatability of filter would be to develop a jig to accurately position the dual mode dielectric resonator. This was found to be a small source of discretion in the development stages. With these suggested improvements in place and using the design procedure developed, there is great potential for the production of a commercially viable dual mode filter.

-54-

8.2 Conclusions
The main project objective to design and construct a 2GHz bandpass filter was achieved. To achieve this objective both simulations and practical work were undertaken. HFSS, by Agilent, a 3-D FEM simulation tool was used to determine the optimum geometry of a dielectric resonator. The FEM modelling software was assessed based on practical results from measurements taken on a single cavity. It was found to be useful in producing approximate results although the main limitations were: it was a slow process and had dependencies on coupling mechanisms. Techniques similar for developing single mode dielectric resonator filters were used to create dual mode single and two-cavity filters. Methods of input/output coupling, intracavity coupling, inter-cavity coupling and tuning were determined and evaluated. Successful methods were implemented in the design of the filters. The responses from each of the filters were then evaluated. Although the two-cavity filter did not meet the specifications, positive outcomes were: a significant size reduction; low cost; and easy tuning ability. The limitations of the design included low Q, spurious cross couplings and other spurious modes close to the passband. Further research and design is required for the commercial development of dual mode filters. This includes identifying and eliminating limitations of the initial design. In conclusion, this thesis has offered a simple methodology and adaptable mechanisms to produce dual mode filters with many more advantages compared to traditional single mode ceramic filters.

-55-

Appendix A Measurement Techniques

Measurement of Coupling Bandwidths
Coupling bandwidths can be measured on the Vector Network Analyser in a number of different ways. The methods described are the Reflected Phase and the Reflected Group Delay methods.

Reflected Phase Method

Calibrate a VNA for S11 in Smith Chart mode. Connect the input of the filter and short out all resonators, this is achieved with the tuning screws. Adjust the delay or phase offset to set the center frequency, which should be in a tight ball at either the 0 or 180 points of the Smith Chart. Remove the short of the first resonator and adjust its tuning screw to obtain a 180 phase shift of f0. See Figure 0.1.
+90

f2

f0

Initial locus in ball

f1
-

Figure 0.1 Smith Chart View after obtaining 180 phase shift of f0

90

The locus will expand. The frequencies at points +90 and -90 are measured. QE is then given by

QE = f (+90) f (-90)

Equation 6

Repeat this process with the second resonator, to produce another 180 shift of f0. Each time the locus expands. The 3 points where the locus of S11 crosses the real axis of the chart are measured, i.e. frequencies f1, f2, f3, see Figure 0.2. -56-

+90

f1 & f2 f0 0

-90
Figure 0.2 Smith Chart View after obtaining 360 phase shift of f0

The coupling bandwidth is given by: K 23 = f 2 f 1

Equation 7

The third resonator is now restored and tuned to produce another 180 phase shift. There are now 5 crossing points, in increasing frequency: f2, f1, f0, f3, f4. The next coupling bandwidth is determined by: K 23 = f 3 f 1
Equation 8

The fourth resonator is restored next and tuned to produce another 180 phase shift. In this process, 7 crossing points are now produced: f3, f2, f1, f0, f4, f5, f6. The coupling bandwidth is determined by:
K 34 = ( f 4 f1 )( f 6 f 3 ) K12
Equation 9

Again the process is repeated for further coupling bandwidths although the equations become more complicated. Consult reference [15] for further equations.

Reflective Group Delay Method

-57-

Group delay can also be used to calculate coupling bandwidths. This methods similarity to the reflective phase delay method, involves shorting out all resonators and then consecutively restoring and tuning individual resonators. The difference is that group delay is measured at each stage rather than frequency crossings. This method is very useful in the design stage for calculating the correct tap point from QE for filters with transformer coupled inputs.

For the first resonator:

QE =

2 td1

Equation 10

Where t d 1 = measured group delay of S11 at f0. See Figure 0.3. This can also be expressed as
Q E ( MHz ) = 636.6 td1
Equation 11

Where t d 1 units are in nsec

-58-

20

gd (ns) 10

0 f0 frequency
Figure 0.3 Group Delay QE measurements

Again, the tuning screw is used to move S11 through 180 each time a resonator is restored.

The coupling bandwidth for the second resonator is:

-59-

K12 =

636.6 td1 td 2

Equation 12

See Figure 0.4.

Figure 0.4 Group Delay K12 measurements

50

gd (ns) 25

0 f0 frequency This process is continued for further coupling bandwidth measurements. For more measurement details with regards to this method refer to Reference [8].

-60-

Appendix B HFSS Simulation Results

Figure 0.5 H-field Contour Plot Figure 0.6 E-field Contour Plot

-61-

18*18*Z HFSS Sim ulation Results

21.5

21

20.5

single 20 dual spurious

19.5

19

18.5 2 2.5 F r eq uency ( G Hz ) 3 3.5

Figure 0.7 18 18 Z HFSS Simulation Results

19*19*Z HFSS Simulation Results
20

19

18

17 Z (mm) single 16 dual spurious 15

14

13

12 2 2.5 Frequency (GHz) 3 3.5

Figure 0.8 19 19 Z HFSS Simulation Result

-62-

20*20*Z HFSS Simulation Results 23 22 21 20 19 Z (mm) 18 17 16 15 14 13 12 1.8 2.3 Frequency (GHz) 2.8 3.3 single dual spurious

Figure 0.9 20 20 Z HFSS Simulation Results

Cavity Height vs Modal Frequencies

3.5 3 Frequency (GHz) 2.5 2 1.5 1 0.5 0 0 10 20 30 40 50 60 Cavity Height (mm)
Figure 0.10 Cavity Height vs Modal Frequencies

single dual spurious

-63-

Appendix C Coupling Measurements

Coupling Screw 1/4" *25mm
12 10 K12 (MHz) 8 6 4 2 0 0 5 Tuning (mm)
Figure 0.11 Coupling Screw 1/4" *25mm Tuning Range

10

15

Milled Coupling Screw 1/4" *25mm

10 9 8 7 6 5 4 3 2 1 0 0 5 10 Tuning (mm)
Figure 0.12 Milled Coupling Screw 1/4" *25mm Tuning Range

K12 (MHz)

15

20

-64-

12mm Metal Tuning Discs from Chamfer

9.8 9.6 K12 (MHz) 9.4 9.2 9 8.8 8.6 8.4 0 5 10 tuning (mm)
Figure 0.13 12mm Metal Tuning Discs from Chamfer Tuning Range

15

20

M4 Coupling Screw
9 8 7 K12 (Mhz) 6 5 4 3 2 1 0 0 5 10 Tuning (mm)
Figure 0.14 M4 Coupling Screw Tuning Range

15

20

-65-

1/4" Coupling Screw on Chamfer

10 9 8 7 6 5 4 3 2 1 0 0 5 10 tuning (mm)
Figure 0.15 1/4" Coupling Screw on Chamfer Tuning Range

K12 (MHz)

15

20

-66-

Appendix D - Filter Models and Plot

Networks Component List for 4.2 MHz Single Cavity Filter Model

N0: BPF1 Lumped Element BandPass Filter : Shunt SC Stub / J Inverter Prototype : Gen. Chebyshev Degree : 2 Return Loss : 25.00000 dB Band Edges : Lower Band Edge: 2217.900 MHz Upper Band Edge: 2222.100 MHz Bandwidth : 4.200000 MHz Centre Freq. : 2220.000 MHz Q Model : Q : 1000000. Frequency : 2220.000 MHz Coupling Bandwidths and External Q (MHz): QEin 8.603030 1-2 9.101215 QEout 8.603030 Components : J1 20.00000 mMhos CS1 Tx1 Length: 33.76041 mm Impedance: 0.1521804 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J2 -21.15816 mMhos CS2 Tx2 Length: 33.76041 mm Impedance: 0.1521804 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J3 20.00000 mMhos

-67-

1 singlecavitynwk.nwk : Graph : S21 Mag dB / S11 Mag dB S21 dB 0 -10 -20 -30 -40 -50 -60 -70 -80 -90 -100 2.2 2.21 2.215 2.22 2.225 2.23 2.24

0 -3 -6 -9 -12 -15 -18 -21 -24 -27 -30 GHz

S11 dB

Figure 0.16 Networks Filter Response of 4.2MHz Filter Figure 0.17 Networks Lumped Element Model for 4.2MHz Filter

-68-

CH1 CH2 PRm 1 Cor

S 11 S 21

log MAG log MAG

5 dB/ 10 dB/

REF 0 dB REF 0 dB

25 Sep 2001 15:58:48 1_:-18.41 dB 1_ -.3366 dB 2 259.139 999 MHz 1

PRm Cor

CENTER 2 240.000 000 MHz

SPAN 100.000 000 MHz

Figure 0.18 Improved Frequency Response of Single Cavity without Peek Support

-69-

Networks Component List for 10MHz Four Section Filter Model

N0: BPF1 Lumped Element BandPass Filter : Shunt SC Stub / J Inverter Prototype : Gen. Chebyshev Degree : 4 Return Loss : 25.00000 dB Band Edges : Lower Band Edge : 2215.000 MHz Upper Band Edge : 2225.000 MHz Bandwidth : 10.00000 MHz Centre Freq. : 2220.000 MHz Q Model : Q : 1000000. Frequency : 2220.000 MHz Coupling Bandwidths and External Q (MHz): QEin 13.274765 1-2 10.409025 2-3 7.715166 3-4 10.409025 QEout 13.274765 Components : J1 20.00000 mMhos CS1 Tx1 Length: 33.76041 mm Impedance: 0.2348203 Ohms Dielectric constant: 1.000000 Loss tangent: 0.0000000 J2 20.00000 mMhos CS2 Tx2 Length: 33.76041 mm Impedance: 0.1443783 Ohms Dielectric constant: 1.000000 Loss tangent: 0.0000000 J3 20.00000 mMhos CS3 Tx3 Length: 33.76041 mm Impedance: 0.1290049 Ohms Dielectric constant: 1.000000 Loss tangent: 0.0000000 J4 -21.15816 mMhos CS4 Tx4 Length: 33.76041 mm Impedance: 0.2348203 Ohms Dielectric constant: 1.000000 Loss tangent: 0.0000000 J5 20.00000 mMhos

-70-

1 2 Untitled : Graph : S21 Mag dB / S11 Mag dB S21 dB 0 0 S11 dB

-10

-5

-20

-10

-30

-15

-40

-20

-50

-25

-60 2.17 2.19 2.2 2.21 2.23 2.25

-30 2.27 GHz

Figure 0.19 Networks Filter Response 10MHz for Four Section Filter

-71-

Networks Component List for 9.5MHz Four Section Filter Model N0: BPF1 Lumped Element BandPass Filter : Shunt SC Stub / J Inverter Prototype : Gen. Chebyshev Degree : 4 Return Loss : 25.00000 dB Band Edges : Lower Band Edge: 2215.250 MHz Upper Band Edge: 2224.750 MHz Bandwidth : 9.500000 MHz Centre Freq. : 2220.000 MHz Q Model : Q : 1000000. Frequency : 2220.000 MHz Coupling Bandwidths and External Q (MHz): QEin 12.611027 1-2 9.888574 2-3 7.329408 3-4 9.888574 QEout 12.611027 Components : J1 20.00000 mMhos CS1 Tx1 Length: 33.76041 mm Impedance: 0.2230792 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J2 20.00000 mMhos CS2 Tx2 Length: 33.76041 mm Impedance: 0.1371593 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J3 20.00000 mMhos CS3 Tx3 Length: 33.76041 mm Impedance: 0.1225546 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J4 -21.15816 mMhos CS4 Tx4 Length: 33.76041 mm Impedance: 0.2230792 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J5 20.00000 mMhos

-72-

1 twocav2.nwk : Graph : S21 Mag dB / S11 Mag dB S21 dB 0 0 S11 dB

-10

-5

-20

-10

-30

-15

-40

-20

-50

-25

-60 2.17 2.19 2.2 2.21 2.23 2.25

-30 2.27 GHz

Figure 0.20 Networks 9.5MHz Filter Response

-73-

Networks Component List for13MHz Four Section Filter Model N0: BPF1 Lumped Element BandPass Filter : Shunt SC Stub / J Inverter Prototype : Gen. Chebyshev Degree : 4 Return Loss : 25.00000 dB Band Edges : Lower Band Edge: 2.213500 GHz Upper Band Edge: 2.226500 GHz Bandwidth : 0.01300000 GHz Centre Freq. : 2.220000 GHz Q Model : Q : 1000000. Frequency : 2.220000 GHz Coupling Bandwidths and External Q (GHz): QEin 0.017257 1-2 0.013532 2-3 0.010030 3-4 0.013532 QEout 0.017257 Components : J1 20.00000 mMhos CS1 Tx1 Length: 33.76041 mm Impedance: 0.3052672 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J2 20.00000 mMhos CS2 Tx2 Length: 33.76041 mm Impedance: 0.1876924 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J3 20.00000 mMhos CS3 Tx3 Length: 33.76041 mm Impedance: 0.1677068 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J4 -21.15816 mMhos CS4 Tx4 Length: 33.76041 mm Impedance: 0.3052672 Ohms Dielectric constant:1.000000 Loss tangent: 0.0000000 J5 20.00000 mMhos

-74-

1 Untitled : Graph : S21 Mag dB / S11 Mag dB S21 dB 0 0 S11 dB

-10

-5

-20

-10

-30

-15

-40

-20

-50

-25

-60 2.17 2.19 2.2 2.21 2.23 2.25

-30 2.27 GHz

Figure 0.21 Networks 13MHz Filter Response

-75-

Appendix E Photographs

Figure 0.22 Single Cavity - lid removed Figure 0.23 Four Section Filter lid removed

-76-

Figure 0.24 Four Section Filter close-up of left cavity Figure 0.25 Four Section Filter

-77-

References
[1] I.C. Hunter, Theory and Design of Microwave Filters, IEE, Stevenage, U.K., 2001. [2] S. Cohn, Microwave Bandpass Filters Containing High-Q Dielectric Resonators, IEEE Transactions on Microwave Theory and Techniques, Vol. 16, No. 4, April 1968, pp. 218- 227. [3] D. Kajfez and P. Guillon, Dielectric Resonators, Noble Publishing Corporation, Atlanta, U.S.A., 1998. [4] Trans-Tech (a subsidiary of Alpha Industries), Dielectric Resonators Application Notes Nos. 821, 831, 851, Alpha Industries, http://www.alphaind.com/default.htm (current Mar. 25, 2000). [5] M. Sadiku, Numerical Techniques in Electromagnetics, CRC Press, Florida, U.S.A., 2000. [6] S.Gerin, Design of a Bandpass Filter using Ceramic Resonators, undergraduate thesis, ISEN, Dept. Electrical Engineering, 1997. [7] D. Bradford, Dielectric Bandpass Filter, undergraduate thesis, Griffith University, School of Microelectronic Engineering, 1998. [8] P. Martin and J. Ness, Coupling Bandwidth and Reflected Group Delay Characterisation of Microwave Bandpass Filters, Filtronic Lecture Notes, Filtronic, Australia, Brisbane, 1999. [9] G. Matthei, L. Young and E. Jones, Microwave Filters, Impedance-Matching Networks, and Coupling Structures, Artech House, Dedham, Massachusetts, 1964. [10] S. Fiedziuszko and R. Chapman, Minature Filters and Equalizers utilising Dual Mode Dielectric Resonator Loaded Cavities, IEEE Transactions on Microwave Theory and Techniques Digest, 1982, pp. 386- 388. [11] C. Wang, H. Yao and K.A. Zaki, Mixed Modes Cylindrical Planar Dielectric Resonator Filters with Rectangular Enclosure, IEEE Transactions On Microwave Theory and Techniques, Vol. 43, No. 12, Dec. 1995, pp. 2817-2823. [12] K.A. Zaki, C. Chen and A.E. Atia, Canonical and Longitudinal Dual-Mode Dielectric Resonator Filters Without Iris, IEEE Transactions On Microwave Theory and Techniques, Vol. 35, No. 12, Dec. 1987, pp. 1130-1135.

-78-

[13] V. Dassonville and I. Hunter, Triple Mode TE01d Resonator Transmission Filters, Filtronic PLC Engineering Conference, Filtronic, U.K., Yorkshire, 2000, pp234238. [14] K.A. Zaki, R.Levy, J. Liang, Dual-Mode Dielectric-Loaded Resonators with Cross-Coupling Flats, IEEE Transactions On Microwave Theory and Techniques Digest, 1995, pp. 509-512. [15] J.B. Ness, A Unified approach to the Design, Measurement and Tuning of Coupled Resonator Filters, IEEE Transactions On Microwave Theory and Techniques, Vol. 46, No. 4, April. 1998, pp. 343-351.

-79-