MICROSTRIP STRUCTURE DESIGN FOR BETTER DIPLEXER PERFORMANCE

Bachelor of Engineering (Electrical and Electronics Engineering)

Design and Development Project 2

SHEHAN DE FONSEKA

2012

DECLARATION I hereby declare that this report entitled Microstrip structure design for a better diplexer performance is the result of my own project work except for the quotations and citations which have been duly acknowledged. I also declare that it has not been previously or concurrently submitted for any other degree at Swinburne University of Technology (Sarawak Campus).

.............................................. Name: Shehan De Fonseka Date: 14th May 2012

Acknowledgement

First and foremost, I would like to thank my Project Supervisor, Mr.Ting Sie King for his valuable support and guidance in this endeavour.

Special Thanks goes to my partner, Abdul Rafay, for his effort throughout the semester. I would also like to extend my gratitude to Chamila Premarathne, Kennedy Mrema, Aaron Tan for their support during the fabrication process.

Last but not least, I would like to thank my parents, all my loved ones and especially my uncle Dr. Brian N DeFonseka for the motivation and strength they have given me in every possible way.

ii

Abstract A Diplexer design is proposed for the GSM frequency range with suppressed 2nd Harmonics with the use of a Dual Band Wilkinson Power Divider. The diplexer combines two bandpass filters which will allow two different pass bands to pass through. The design is fabricated on Microstrip and measured.

It further looks in to the aspect of using a dual band Wilkinson power divider instead of a common port junction in the design of a diplexer. There have been many prior designs of dual band Wilkinson power dividers, a one method has been used in this thesis to proceed with designing the diplexer.

The simulations have been carried out on Sonnet EM and Agilent ADS 2011 softwares.

iii

Table of Content
1 Chapter 1: Introduction ............................................................................................. 1 1.1 1.2 1.3 1.4 2 Project Background........................................................................................... 1 Problem Statement ............................................................................................ 1 Project Objectives ............................................................................................. 1 Work distribution .............................................................................................. 2

Chapter 2: Literature Review .................................................................................... 3 2.1 2.2 2.3 2.4 Introduction....................................................................................................... 3 S parameters ...................................................................................................... 3 Quarter Wavelength Transmission line ............................................................ 5 Microstripline.................................................................................................... 6 Coupled Microstrip Lines ........................................................................... 8

2.4.1 2.5

Power Dividers ................................................................................................. 9 Three Port Network .................................................................................. 10 T Junction Power Divider .................................................................. 11 The Lossless Divider ......................................................................... 11 The Resistive power Divider ............................................................. 12 The Wilkinson Power Divider ........................................................... 13

2.5.1

2.5.1.1 2.5.1.2 2.5.1.3 2.5.1.4

2.5.1.4.1 Even and Odd mode Analysis ....................................................... 15 2.5.1.4.2 Even Mode .................................................................................... 16 2.5.1.4.3 Odd Mode...................................................................................... 17 2.6 2.7 2.8 2.9 3 Few Important terms for a Wilkinson power divider ..................................... 18 Dual Band Wilkinson Power Divider ............................................................. 19 Band Frequencies ............................................................................................ 19 Compact Dual-band Power Divider using Branch lines and Resistors .......... 20

Chapter 3: Design Process and Simulation ............................................................. 22 3.1 RT Duroid 6010 .............................................................................................. 22
iv

3.2

Designing Compact Dual-band Power Divider using Branch lines and

Resistors ...................................................................................................................... 22 3.2.1 3.2.2 Simulation and Results ............................................................................. 24 Agilent ADS Momentum .......................................................................... 26 Working on ADS layout .................................................................... 26 ADS Component Library .................................................................. 27

3.2.2.1 3.2.2.2 3.2.3

Filters ........................................................................................................ 27 Low Pass Filter .................................................................................. 27 High Pass Filter ................................................................................. 28

3.2.3.1 3.2.3.2 3.2.4

The Diplexer Formation............................................................................ 29 Further Optimization ......................................................................... 31

3.2.4.1 3.2.5 3.2.6 3.2.7

Optimized Results ..................................................................................... 33 Metclad ..................................................................................................... 34 Simulation for Metclad ............................................................................. 34 Dilpexer Simulation........................................................................... 34 Simulation of the Low Pass Filter ..................................................... 35

3.2.7.1 3.2.7.2 4

Chapter 4: Fabrication and Measurement ............................................................... 36 4.1 Components Used ........................................................................................... 36 Vishay 180R Resistor (0402) .................................................................... 36 Panasonic ERJ3GEYJ101V (0603) .......................................................... 36

4.1.1 4.1.2 4.2 4.3 4.4 4.5 4.6 4.7 4.8

Test Fabrication on FR4 ................................................................................. 37 Toner Transferring .......................................................................................... 38 Fabrication on Metclad ................................................................................... 39 Measured Results for the Low Pass Filter ...................................................... 40 Measured Results for the Diplexer ................................................................. 41 Conclusion ...................................................................................................... 43 Recommendations & Future Work ................................................................. 43

References ....................................................................................................................... 44 Appendices...................................................................................................................... 45

Table of Figure Figure 2.1- Two Port Network...............................................................................3 Figure 2.2- Microstripline......................................................................................6 Figure 2.3- Field lines of Microstrip.......................................................................7 Figure 2.4 - Coupled Microstriplines.......................................................................8 Figure 2.5 - Schematic Block Diagram of a n+1 port combiner and divider.................9 Figure 2.6 -Transmission Line model of lossless divider...........................................11 Figure 2.7 - A 3db Resistive power divider.............................................................12 Figure 2.8 Wilkinson Divider in microstrip format and the transmission line format.14 Figure 2.9 - Wilkinson Power Divider Schematic.....................................................14 Figure 2.10- Wilkinson Divider in Normalized and symmetric form..........................15 Figure 2.11- Even mode Excitation........................................................................16 Figure 2.12-Odd mode Excitation..........................................................................17 Figure 2.13 - Power Divider Topology...................................................................20 Figure 3.1 - Schematic Diagram............................................................................24 Figure 3.2 - Simulation Results.............................................................................25 Figure 3.3- Modified Layout of the power divider.26 Figure3.4- Low Pass Filter & Performance.............................................................27 Figure 3.5- Highpass filter & Performance.............................................................28 Figure 3.6- Filters connected in 90 degrees.............................................................29 Figure 3.7- Filters Connected in 60 degrees............................................................30 igure 3.8- Filters Connected in 45 degrees..............................................................30 Figure 3.9- Low Pass filter connected in 90 degrees with an extended transmission line ..........................................................................................................................31 Figure 3.10- Connected by the 14mm Transmission Line.........................................31 Figure 3.11- With two extra set of grooves at 0.2mm apart from each other...............32 Figure 3.12- Optimized Design Layout...................................................................33 Figure 3.13-Optimized Design Simulation Results...................................................33 Figure 3.14- Metclad Simulation of Diplexer..........................................................34
vi

Figure 3.15- Metclad Simulation for the Low Pass Filter.........................................35 Figure 4.1- Impedance vs Frequency curve for Vishay Thin Film Resistor.................36 Figure 4.2- Impedance vs Frequency curve for Panasonic Thick Film Resistor...........36 Figure 4.3 -Diplexer Fabricated on FR4.................................................................37 Figure 4.4 - The Fabricated lines of 0.35mm of the power divider............................37 Figure 4.5 Divider achieved by a toner transfer.................................................... 38 Figure 4.6- Another Toner Transfer attempt........................................................... 38 Figure 4.7 - Fabricated Diplexer on Metclad...........................................................39 Figure 4.8 - Awry lines achieved by toner transfer on Metclad..................................39 Figure 4.9 -Measured Results for the Low Pass Filter..............................................40 Figure 4.10- Taking Measurements.........................................................................40 Figure 4.11-Results for the low pass Filter..............................................................41 Figure 4.12-Results for the high pass filter..............................................................41 Figure 4.13- The close up of the fabricated Diplexer................................................42

List of Tables Table 1.1-Work Distribution ............................................................................................ 2 Table 2.1- GSM Frequencies ......................................................................................... 19 Table 3.1 Low Pass Filter .............................................................................................. 28 Table 3.2 High Pass Filter .............................................................................................. 29

vii

1
1.1

CHAPTER 1: INTRODUCTION
PROJECT BACKGROUND

In an era where advanced technology in Communication industry has revolutionized the whole world, the bar is set very high for any new product that enters the market. Mobile Devices, Wireless Devices and Computers are all getting compact and affordable. But with extra compactness can sometimes be a costly bargain in terms of the quality achieved. The occurrences of 2nd harmonics hinder the performance of a Microwave Diplexer to a considerable extent. Furthermore losses such as return loss and insertion loss need to be dealt with in the power distribution of the diplexer.

This thesis looks in to simulation and fabrication of Microwave Diplexer with better harmonic performances.

1.2

PROBLEM STATEMENT

A common port junction is essential to design a diplexer. its performance and compact size of the structure are of great concern in the power divider design.

1.3

PROJECT OBJECTIVES Study about Junctions and Power Dividers to join the two passband filters to form a diplexer. Study about methods to reduce the effect of 2nd harmonics that occur in a Diplexer along with high return loss and low insertion loss. Design and simulate a diplexer. Fabricate and Measure

1.4

WORK DISTRIBUTION Assigned Part Study and design a suitable power divider for the Diplexer design. Combine the Filters to form a Diplexer and simulate for optimum performance.

Group Member Shehan De Fonseka

Abdul Rafay

Study and design bandpass filters for the diplexer with good suppression of 2nd harmonics.
Table 1.1-Work Distribution

2
2.1

CHAPTER 2: LITERATURE REVIEW

The world of microwave engineering has been evolving since its inception. Whilst working with waves of which frequencies vary in a range of 300MHz to 300GHz of the Electromagnetic spectrum has been a different discipline for years, it has contributed greatly to the development of communication technology, especially with the proposals to use microwave technology in cellular phone systems in 1970(Pozar 2005). Four decades later, it has become an integral part of our daily lives. The microwave diplexer is such a component that has become an integral part of the communication industry. Thus it is important to achieve better performance while maintaining its compact size. 2.2 S PARAMETERS

S parameters refer to the Scattering Matrix, a concept developed in the decade 1950 onwards. The Scattering Matrix is a mathematical model which enables us to understand how RF energy propagates through a multiport network. For a RF signal that is incident on one port, a fraction of the signal reflects while the other scatters & exit the other ports. S-parameters are complex mainly because the network changes both the magnitude & the phase of the particular signals. S parameters describe the response of a N-port network ti voltage signals at different ports. S parameters are written as Sij where the first subscript (i) refers to the output port while the second subscript (j) refers to the input port. For a N port network, there will be N2 number of parameters. A two port network is shown below.

Figure 2.1- Two Port Network

2 = 21 1 + 22 2 a1 & a2 refer to the normalized incident wave variables while b1 & b2 refer to the normalized reflected wave variables. 1 + 1 0 20 20 1 0 0

1 = 11 1 + 12 2

2 =

1 =

2 + 2 0 20

2 =

1 =

2 2 0 20

1 1 0

2 0

1 0

By squaring the above equations for a and b gives: |1 |2 =

|2 |2 = |1 |2 = |2 |2 =

Therefore, the S parameters coefficients definitions can be given as:

Where an = 0 refers to a perfect impedance match at port n.

21 =

12 =

22 =

1 2

2 2

1 =0 1 =0

The parameter S11 and S22 are also termed the reflection coefficients while S12 and S21 are termed as transmission coefficients. In general, the S-parameters are complex values. Thus, it is convenient to express them in terms of amplitudes and phases where amplitudes are given in decibels (dB) which are shown as: For filter characterization, the two parameters can be defined as:
= 20 log| | = 20 log| | = 1, 2 ( ) = 1, 2 20 log| | , = 1, 2

(where LA refers to the insertion loss between ports n and m while LR refers to the return loss at port n.)

In transmission line, the reflection coefficient is defined as,

(where ZL refers to the input impedance while Z0 refers to the terminal impedance). Given this, the voltage standing wave ratio( VSWR) and the return loss can be related by,
= 1 + | | 1 | |

(Jia-Sheng Hong and MJ. Lancaster, 2001)

2.3

QUARTER WAVELENGTH TRANSMISSION LINE

Quarter wave transmission line or more commonly known as Quarter Wave transformer is used to match two transmission lines with different impedances. A transmission line length of usually
4

is used. The impedance of the quarter wavelength

transmission line is the square root of the multiplication of the matching impedances1 2 .

2.4

MICROSTRIPLINE

Microstrip is one of the most popular planar transmission structures used in Microwave circuit designing. It can be manufactured with photolithography technology and can easily be integrated with other passive or active microwave components.

Microstrip transmission line consists of a thin layer of conductor printed on top of a grounded dielectric substrate.

Figure 2.2- Microstripline

The important parameters related to the microstripline structure are conductor width (w), substrate thickness (h) , relative permittivity (r) & thickness of the conductor line. The thickness of the top conductor line is of less importance thus can be neglected. This is mainly due to the thickness which is usually in the range of 10-20ms. As the thickness of the conductor line increases, the number of electric field lines on the side planes of the top conductor layer would increase. This will affect the characteristic impedance of Z0 & the Effective dielectric constant (eff) of the microstripline. The Effective dielectric constant will be explained in due course. The microstripline cannot support a pure TEM wave. In actuality, what a microstripline can accommodate is a hybrid TM-TE wave. However in most of the practical applications the dielectric substrate is very thin thus giving the electric fields a quasi TEM nature. The analysis for determining the microstripline characteristics impedance & the propagation constant can be divided in to two groups, the quasi-static analysis method & full wave analysis method. Full wave analysis method considers a hybrid mode of propagation thus making the solutions provided analytically complex. This method shows the variation of characteristic impedance and phase velocity of a microstripline with frequency.
6

Quasi-static methods consider a pure TEM mode of propagation in microstripline. This allows us to obtain a good approximation for the phase velocity (Vp), the propagation constant () & the characteristic impedance ( Z0) . =

(2.1) (2.2)

= 0 Effective Dielectric Constant ( ) 0 = 2

(2.2)

Figure 2.3- Field lines of Microstrip

Since the conductor layer of microstripline lies between two different dielectric zones, air & the dielectric substrate, the parameter Effective dielectric constant is introduced. 1 r

Microstrip has some of its field lines in the dielectric region, concentrated between the conductor strip & the ground plane, while some fraction of it occupies the air region above. This parameter depends on the substrate thickness (h), & the conductor width (w). = 1 + 1 + 2 2 12 1 + 1 (2.3)

Characteristic impedance of a microstrip line can be calculated using,

0 =

120 1 (2.5) + 1.393 + 0.667 ln + 1.444

Z0 =

eff

60

ln

8h w w + for 1 w 4h h

(2.4)

whenever the relative permittivity & the characteristic impedance has been specified, the following equations will be used to do the calculations for the ratio between width & height. 8 = 2 2 2 (2.6)

where

r 1 0 r + 1 0.5 0.11 = + 0.23 + 60 2 r + 1 r

where

( 1) ln(2 1) +

1 2

ln( 1) + 0.39 60 2 0

0.61

(2.7)

2.4.1

COUPLED MICROSTRIP LINES

Coupled microstriplines are in laymans words, two microstrip lines placed close to each other in parallel. Coupled microstriplines are used to build Filters and Directional couplers. This coupled line structure has the ability to transmit two quasi-TEM wave modes, usually even & odd modes.

Figure 2.4 - Coupled Microstriplines

Even mode exists when there are charges of the same sign in both the microstriplines while Odd mode exits when the signs of the charges are opposite. Each of these modes will contain unique transmission line characteristics, namely the characteristic impedance (Z0e, Z0o) & the propagation velocity (Vpe,Vpo). Characteristic impedance in the even & odd modes depend on the dielectric constant (r), microstripline width to substrate height ratio ( ) & the separation between the microstriplines to substrate height ratio ( ).

2.5

POWER DIVIDERS

Power Dividers are passive microwave component that can be used for either power division or power combining. A power divider may divide an input signal in to two or more parts while the power combiner may combine two or more signals. Usually a three port network will be used to create a T junction & other types of power dividers. Power dividers can have more than three ports, directional couplers & hybrid use up four ports for instance. In this thesis, we focus on a three port network. Power dividers usually divide power equally (3db) however; unequal division can also be achieved.

Figure 2.5 - Schematic Block Diagram of a n+1 port combiner and divider

2.5.1

THREE PORT NETWORK

A three port network can be attributed to a T junction, which is the simplest type of a power divider. The scattering matrix of a 3 port network consists of nine parameters. S11 [S] = S21 S31 S12 S22 S32 S13 S23 S33

The T junction power divider consists of 3 ports, one port is used for the input signal while the other two allows the transmission of divided signals.

In order to have minimum reflection at each of the three ports, its important for the network to be matched. A matched network is where Sij = 0 when i = j ,thus the diagonal components from the upper left corner to the bottom right corner will equal to zero as shown below. In this case, S11=S22=S33=0. 0 [S] = S21 S31 S12 0 S32 S13 S23 0

Another property of the 3 port network is the property of reciprocity. If the material that is used to make the divider is passive & has the same electrical properties throughout (isotropic), then the transmission between any two ports suffer the same power loss as each other despite different propagation directions. This makes the parameters of the S matrix symmetric for Sij=Sji. An ideal passive network is considered to be of lossless nature, without heat or radiation generated when propagating along. Under these circumstances, power incident on one port should be the same as the sum of powers leaving the other ports. When a matched network is completely lossless, it has to fulfil certain conditions:
S13 S23 = 0

(2.8) (2.9) (2.8) (2.10) (2.11)

S23 S12 = 0

|S12 |2 + |S13 |2 = 1
10

S12 S13 = 0

|S12 |2 + |S23 |2 = 1

In an ideal scenario the sum of squares of all S parameters ought to equal one, however in reality it can never be done. What can be done is that to bring the sum of the Squares of the parameters to be close as possible to 1. It is not possible to design an ideal lossless network. When looking at the above equations, its possible to observe that in order to satisfy the conditions laid by the equations, at least two of S12, S13, S23 must equal zero. But this disagrees with the conditions laid down based on Reciprocity condition. Thus it is not possible to achieve all these properties at the same time. In order to realize a practical design, we need to relax at least one of them(Pozar 2005).

|S13 |2 + |S23 |2 = 1

(2.12)

2.5.1.1 T JUNCTION POWER DIVIDER T Junction power divider is the simplest of 3 port network power dividers that can be used in any form of transmission medium. The T junction power divider, in the absence of transmission line loss, cannot be matched at the same time in all ports. The variants of T junction power dividers are discussed below. 2.5.1.2 THE LOSSLESS DIVIDER

Figure 2.6 -Transmission Line model of lossless divider

Lossless divider can be modelled with a slight variation of T junction which has come to known by the popular term of Y junction. The fields and higher order modes associated at such a junction leads up to some energy being stored at the Junction. This
11

can be attributed to a lumped susceptance of value B. For the divider to be matched to the characteristic impedance of Z0, we can write the following equation. 1 1 + Z1 Z2

Yin = jB + there is no lumped susceptance, thus

(2.13)

Assuming the transmission line to be of low loss or lossless nature, we can consider

Yin =

1 1 + Z1 Z2

(2.14)

Although in reality, the susceptance will not equal zero, its effect can be minimized by adding a reactive element to the power divider. Z1 and Z2 can be selected as to have different power ratios, if the output lines are matched, it can be match with the input line. However, there will be no isolation between the ports.

2.5.1.3 THE RESISTIVE POWER DIVIDER Resistive power divider is a lossy, yet can be matched at all ports. This is achieved by using lumped element resistors in the transmission line. This explanation is for an equal power divider, but it must be mentioned that different power ratios can also be achieved.

Figure 2.7 - A 3db Resistive power divider

12

Considering the impedance of an output port, it can be seen that the total impedance of an output port is: Z= Z0 + Z0 3 (2.15)

Thus the input impedance of the divider will equal, Zin = Z0 2Z0 + = Z0 3 3

This shows that the inputs are matched. While the outputs are also matched, the network is also symmetric from all ports. Therefore S11=S22=S33=0. Assuming voltage at the input port to be V1, we calculate the voltage V at the centre of the junction, 2Z0 2 3 = V1 V = V1 Z0 2Z0 3 + 3 3 Z0 + Z0 3 Z0 = 3 1 V = V1 4 2

The output voltages can be calculated as, V2 = V3 = V

Therefore, S21=S31=S23=0.5, which means that the output power is -6db below the level of the input power. The scattering matrix for this can be written as, [S] = 1 0 1 2 1 1 0 1 1 1 0

The power delivered to the output of the divider can be found out to be ,P2=P3=0.25Pin by using the basic formula for power. Thus it can be seen that half of the power delivered is dissipated through the resistor. 2.5.1.4 THE WILKINSON POWER DIVIDER In the previous types of power dividers we discussed, we saw that each of them had significant short comings. For instance, the lossless power divider, though lossless, suffered from the inability of being matched at all ports not did it have the ability to isolate output ports. The resistive divider, though can be matched at all ports, suffered from the inability to be lossless and to achieve high isolation between the output ports

13

Figure 2.8 Wilkinson Divider in microstrip format and the transmission line format

The Wilkinson power divider is the best option to replace the above mentioned. The Wilkinson Power Divider was first proposed by Ernest J. Wilkinson in 1960(Wilkinson 1960). Like the resistive divider, Wilkinson power divider is matched and reciprocal, therefore Lossy. However, when matched properly, the power can be divided without any loss with only the reflected power being dissipated. Wilkinson power divider can be made with different power ratios, but in analysing the Wilkinson power divider, its easier to consider the equal power division or the 3db case.

Figure 2.9 - Wilkinson Power Divider Schematic

The isolation resistor employed between the two output ports is one of the most important parts of the Wilkinson power divider concept. It is the single component that enables this model of power divider achieve what its predecessors could not. Dissipation of energy, when signal enters through any of the output ports, happens only through the isolation resistor. This does not affect the efficiency of the Wilkinson network. Furthermore, the isolation resistors allows the divider achieve perfect isolation
14

when operating at its working frequency. Two methods are usually used to analyse the Wilkinson power divider, one is the original analysis proposed by Ernest J. Wilkinson, while the other is the even and odd mode analysis(Pozar 2005). The latter will be used to analyse in this thesis.

2.5.1.4.1 Even and Odd mode Analysis In order to proceed with the even and odd mode analysis, a plain circuit is derived from the original one. Thus analysis becomes much less complicated. The derived circuit is given below.

Figure 2.10 - Wilkinson Divider in Normalized and symmetric form

For simplicity, we normalize impedances to the characteristic impedance Z0. This network is symmetric across the centre plane. The two source resistors have been given normalized values of two so that they will form a resistor of value one when in parallel to represent the impedance of the matched ports. The quarter wave transformer lines have a normalized characteristic impedance of Z, while the shunt resistors contain a value of r. For an equal split power divider, Z=20.5 & r = 2. In the even mode, Vg2 =Vg3=2V0 while in the odd mode, Vg2 =-Vg3=2V0. The superposition of these two modes we can obtain an excitation of Vg2 = 4 V0, Vg3=0. This will be useful in finding the S parameters.

15

2.5.1.4.2 Even Mode Since Vg2 =Vg3=2V0, V2e= V3e there will be no current flowing through the r shunt resistor. So there will be a short circuit across the path of the shunt resistor.

Figure 2.11- Even mode Excitation

The network can be bisected with having open circuits at as shown above because no current will flow through the transmission line between the two inputs at port 1 or the shunt resistors r/2.
e = Zin

e e V2 = V0 since Zin =1

Z2 2

Let x=0 at port1 & x=- at port 2, Therefore, the transmission line voltage; V(x) = V + ejx + ejx ; where is the reflection coefficient Equalling to the x values above, e V2 = V = jV + (1 ) = V0 4 ( + 1) e V1 = V(0 ) = V + (1 + ) = jV0 ( 1) = 2 2 2 + 2
4

is seen from port 1, looking towards the resistor of normalized 2

e V1 = jV0 2

16

2.5.1.4.3 Odd Mode For the odd mode excitation, Vg2 =-Vg3=2V0, V2o= -V3o . The parallel connected transmission line is /4 and short circuited at port 1 one while the r/2 impedance can be seen when looking in to port 2.

Figure 2.12-Odd mode Excitation

Port 2 can be matched for odd mode for a value of r=2. Then,
o = V0 V2

All the power in the odd mode is delivered to r/2 with no power being delivered to the port 1.Finding the input impedance when port 2 and 3 are matched with identical loads, 1 Zin = (2)2 = 1 2

o V1 =0

S11=0 (Zin=1 for port1 ) S22=S33=0 (matched in both modes) S12=S21= S13=S31=
o Ve 1 +V1 o Ve 2 +V2

(symmetry at ports 2 and 3)

S23=S32=0 (short/open circuit at the midplane) When the outputs are matched, there is no power dissipating through the resistor r ,therefore the power divider is lossless when perfectly matched. Furthermore since S23=S32=0, there is good isolation between the ports 2 and 3.
17

2.6

FEW IMPORTANT TERMS FOR A WILKINSON POWER DIVIDER Insertion Loss(S21=S31) For a power divider, insertion loss refers to the additional loss due to the splitting of the signal. The additional losses are usually caused by reflections, dielectric absorption, radiation effects and conductor losses. Conductor losses in high frequency devices are mainly caused by the skin effect. Losses due to reflections also increase with increase of frequency.

Power Divider Isolation(S23=S32) In an ideal power divider, the output ports can be expected to be mutually isolated. Isolation means that a signal entering port 3 does not leak in to port 2 or vice versa. Isolation values above 15db are usually considered to be good. The achievable isolation usually depends on the design itself. Usually, when larger the bandwidth and higher the frequency, the difficult it is to provide good isolation.

Return Loss(S11=S22=S33) The return loss and VSWR both pretty much refer to the same thing. The return loss is an indication on how well the network is matched. Return loss and isolation go hand in hand such that, it is difficult to attain good isolation without good return loss performance.

18

2.7

DUAL BAND WILKINSON POWER DIVIDER

For the purpose of building a microwave diplexer, we are required to understand the concepts of the Power Dividers. In order for us to design a microstrip structure for better diplexer performance, the power divider junction is an essential part of the project. In the previous chapters, basics in to power dividers & junctions were given. It was rather obvious that the Wilkinson power divider would be the best choice for a design of a microwave diplexer. However, the Wilkinson Power Divider is designed for a single band operation. With the advancement in the communication field, multi band structures have a very higher demand. Thus, there has been much research done about creating a Dual Band Wilkinson Power Divider with good performance. In this thesis, we look in to two methods that has been researched about before, and published.

The equations derived in these researches have been used for calculation and design of this part of the project. Simulation results achieved by using Agilent ADS and Sonnet EM have also been attached.

2.8

BAND FREQUENCIES

The frequency bands selected for this project is GSM 900/1800 bands frequently used in Asia. System P-GSM-900 DCS-1800 Band 900 1800 Uplink (MHz) 890-915 1710-1785
Table 2.1

Downlink (MHz) 935-960 1805 -1880

The operating frequencies used in this thesis are 900MHz and 1800MHz.

19

2.9

COMPACT DUAL-BAND POWER DIVIDER USING BRANCH LINES AND RESISTORS

A Wilkinson power divider structure with the use of cascaded Branch lines and two resistors is have been reported (Law & Cheng 2008). The design of the structure basically contains cascaded branch line sections (ZA and ZB) and resistors (RA and RB). The divider caters to upper (f2) and lower band (f1) frequency ratio of up to 3. ,1 (f = f1 , f2 ) = 2Zs (3.19)

The conditions stated in 3.19 and 3.20 will be used to derive the required circuit parameters where Zs refer to the source impedance and ZL refer to the load impedance.

,1 (f = f1 , f2 ) = ZL

(3.20)

Figure 2.13 - Power Divider Topology

The Design formulas obtained in the paper are as follows: = 2 (3.21)

= ( k =

2 2 1) + 1 . 1 1 2

(3.21) (3.22)

( 1) tan 2 + ( 1)2 tan 4 + 4 2

20

2 =

1 1 ( 1) ( 1) 20 . (1 ) 1 +

20

(3.23)

1 =

(3.24) (3.25) (3.26)

= (). 0 = 0

Z0 refers to the characteristic impedance of the Microstripline.

21

3
3.1

CHAPTER 3: DESIGN PROCESS AND SIMULATION

The Substrate RT Duroid 6010 is used in the power divider design. The RT Duroid has a relative dielectric constant of 10.2 while the substrate thickness is given as 1.27mm. The characteristic impedance is taken to be 50.

3.2

DESIGNING COMPACT DUAL-BAND POWER DIVIDER USING BRANCH LINES AND RESISTORS

The second design is reported in the following using Cascaded sections of branch lines and resistors(Law & Cheng 2008). The design equations are referred to 3.21 until 3.26. The source impedance and load impedance will be equal to the characteristic impedance of the microstripline. = = 50 = ( 2 2 1) + 1 . = (2 1)(2 + 1) . = 1 1 2 2 6 2 4 + (2 1)2 tan + 8 6 6 = 0.7953 4 1 1 ( 1) 20 = 202.96 = 2 =2

k =

(2 1) tan 2 =

1 =

( 1) 20 . = 107.79 (1 ) 1 + = 63.06 = 79.29

The centre frequency of the frequencies f1(900MHz) and f2(1800MHz) is calculated as:

22

(1 + 2 ) = 1350 = 1.35 2

The width and length for the designed parameters are then calculated based on the formulas 2.3-2.7 listed under section 2.4. 10.2 1 79.29 10.2 + 1 0.5 0.11 = 3.368 = + 0.23 + 2 10.2 + 1 60 10.2 8 3.368 = 23.368 = 0.2763 < 2 2 WZA = 0.351mm 1 + 12 0.2763 1 = 6.386

therefore, the width required for ZA transmission line is calculated as:

The effective dielectric constant is calculated as: = 1 + 10.2 10.2 1 + 2 2

The length can be calculated using, = 2

2 1.35 109 = 6.386 2 3 108 1 = 21.984

Through the same method, the design parameters for ZB transmission line are found. WZB = 0.6996mm = 21.624

23

3.2.1

SIMULATION AND RESULTS

As it was stated in the previous section, the first step for the simulation is to design the proper schematic. Once all parameters are entered, one can proceed with the simulation, however the desired results cannot be achieved at once. In order to achieve the desired results, tuning is required until the proper results are obtained. The tuning was done wisely with identification of different portions of the circuit rather than trial and error The schematic for this method is given below

Figure 3.1 - Schematic Diagram 24

4.3.2- Simulation Results

Figure 3.2 - Simulation Results

Thus the results have been grouped in a table below.

Loss in dB S22=S33(Return Loss/Transmission) S23=S32 (Isolation) S21=S13 (Insertion Loss) S11 (Input matching)

900MHz -47

1800MHz -45.8

-25 -3.02 -24.8

-30.9 -3.02 -31.1

These results give much clearer dual band characteristics around 0.9GHz and 1.8GHz. Although the isolation and input matching occurs at around 0.8GHz instead of 0.9GHz, the values are not that far off.

25

3.2.2

AGILENT ADS MOMENTUM

Momentum is a part of Agilent Advanced Design Systems that provides simulation tools for design and evaluation of modern communication systems. Momentum is an electromagnetic simulator which computes S-parameters for generalplanar circuits, such as microstrip, slotline, stripline, coplanar waveguide, and other topologies.

Momentum is based on a numerical discretization technique called the method of moments. This technique solves Maxwell's electromagnetic equations for planar structures embedded in a multilayered dielectric substrate. The simulation modes available in Momentum (microwave and RF) are both based on this technique, but use different technologies to achieve their results.

Momentum Microwave (Full Wave Mode) and Momentum RF differ from each other by the Green Function formulations that are being used. Microwave mode uses full wave Green functions, whereas RF mode simplifies the Maxwells equations when characterizing the substrate. This will result in calculating L and C values that are real (instead of complex values that are frequency dependent as in Microwave mode) and frequency independent. Therefore, Momentum Microwave provides the most accurate results although Momentum RF simulation can help to reduce simulation time significantly.

3.2.2.1 WORKING ON ADS LAYOUT The designed power divider is translated in to the ADS Layout with certain modifications to accommodate the resistors.

Figure 3.3- Modified Layout of the power divider 26

3.2.2.2 ADS COMPONENT LIBRARY High Frequency model resistors, ERJ3GEYJ 101 for 100 Ohm and ERJ3GEYJ 181 for the 180 Ohm resistors have been used for simulations in the formation of the Diplexer here onwards.

3.2.3

FILTERS

Conventional 3rd order Open Loop filters with grooves and stubs have been used to maximize the filters designed for the Diplexer. 3.2.3.1 LOW PASS FILTER The low pass filter has a dimension of 42.4mm by 25.6mm.

Figure3.4- Low Pass Filter & Performance

27

Modified Open-Loop Lowpass Filter Filter order Passband Centre Frequency Fractional Bandwidth Insertion Loss (S21) Return Loss (S11) 3 0.86GHz 1.00GHz 0.9GHz 15.6% 0.911dB 13.65dB
Table 3.1

The harmonics have been suppressed below -20db. 3.2.3.2 HIGH PASS FILTER The High Pass filter has a dimension of 42.4mm by 11mm.

Figure 3.5- Highpass filter & Performance

28

Modified Open-Loop Highpass Filter Filter order Passband Centre Frequency Fractional Bandwidth Insertion Loss (S21) Return Loss (S11) 3 1.749GHz 2.019GHz 1.8GHz 15% 0.001dB 8.888dB
Table 3.2

3.2.4

THE DIPLEXER FORMATION

Although a Y or T junction is commonly used for the formation of a Diplexer, in this project we have tried a Dual Band Wilkinson Power divider for a junction. The Filters designed above has to be connected to the Wilkinson power divider, so that best results can be obtained by simulation. Several connection modes have been used in deciding the optimum diplexer performance in this thesis.

Figure 3.6- Filters connected in 90 degrees

29

Figure 3.7- Filters Connected in 60 degrees

Figure 3.8- Filters Connected in 45 degrees

The figures 3.7 and 3.8 go on to show that none of the connections have the performance required from the Diplexer. However, while simulating for differently connected diplexers, it was found out that better performance may be obtained by increasing the length the transmission line connecting the Low Pass Filter to the Dual Band Wilkinson power divider.

30

Figure 3.9- Low Pass filter connected in 90 degrees with an extended transmission line

The transmission line which was originally the size of 8mm, has been extended by another 8mms, immediately there is a significant improvement on the performance of the Diplexer. With this design in mind, the filters have been connected in 60 degrees and 45 degrees, however the best results were produced when connected in 90 degrees. Thus the diplexer above given was chosen for further optimization including changing the length of the transmission line connecting the Low Pass Filter from the Dual Band Wilkinson Power divider.

3.2.4.1 FURTHER OPTIMIZATION

Figure 3.10- Connected by the 14mm Transmission Line

31

By changing the connected Transmission line length, it was found at that better results can be obtained at a Distance of 14mms. Despite the being reasonably closer to the diplexer performance required, there is a surge in the 2nd Harmonics. Since the Wilkinson Power Divider itself is a separate microwave device with harmonics, it will add up to the existing harmonics thus hindering the better performance. In order to investigate how the performance can be made better, several changes were made particularly to the low pass filter such as changing the port positions, adding up extra stubs, adding more grooves..etc.

Figure 3.11- With two extra set of grooves at 0.2mm apart from each other

32

3.2.5

OPTIMIZED RESULTS

Figure 3.12- Optimized Design Layout

Figure 3.13-Optimized Design Simulation Results

The final design produces an Insertion loss of -0.02db and -0.295db at Low Pass and High Pass frequencies respectively while the return loss is at -8.712 and -5.456 at Low Pass and High Pass frequencies.
33

3.2.6

METCLAD

Metclad is another type of substrate with a dielectric constant of 10.8 and a thickness of 1.55mm. Metclad has been used to fabricate the final design in this theses instead of RT Duroid 6010 due to financial constraints.

3.2.7

SIMULATION FOR METCLAD

Following is the simulation results for the Diplexer with the use of Metclad as the substrate. In this design, the width of the port 1 entrance transmission line has been doubled from 0.35mm to 0.7mm due to practical requirenment of port connection. This will have a slight effect on the results.

3.2.7.1 DILPEXER SIMULATION

Figure 3.14- Metclad Simulation of Diplexer

34

3.2.7.2 SIMULATION OF THE LOW PASS FILTER

Figure 3.15- Metclad Simulation for the Low Pass Filter

35

4
4.1

CHAPTER 4: FABRICATION AND MEASUREMENT

4.1.1

Vishay SMD wraparound thin film resistor with tolerance of 2%, up to 50GHz

Figure 4.1- Impedance vs Frequency curve for Vishay Thin Film Resistor

4.1.2

PANASONIC ERJ3GEYJ101V (0603)

Panasonic SMD wraparound thick film resistor with 5% tolerance, can withstand up to 20GHz.

Figure 4.2- Impedance vs Frequency curve for Panasonic Thick Film Resistor 36

4.2

TEST FABRICATION ON FR4

The testing for the Diplexer design was carried out by way of toner transfer method on to a FR4 substrate.

Figure 4.3 -Diplexer Fabricated on FR4

The fabricated line widths of the power divider were quite close to the actual design value of 0.35mm.

Figure 4.4 - The Fabricated lines of 0.35mm of the power divider

37

4.3

TONER TRANSFERRING

Toner transferring was done by printing the design on a Pulsar toner paper via a Laser Printer available from the library. First and Formost, the quality of the print was not extremely good. To obtain better results, it would be helpful to have access to a better Laser Printer. Using the laminator to transfer toner on to the Metclad was proven trickier than transferring on to FR4. The methods of Laminating, Ironing and a Combination of that were tried out, neither provided a satisfactory transfer. Shown in figure 4.6 is how the Power divider looked after a toner transferring.

Figure 4.5 Divider achieved by a toner transfer

It is important to note that the toner transfer selected to proceed with, was not the best toner transfer that was made comparatively but was used due to the time constraints.

Figure 4.6- Another Toner Transfer attempt

38

4.4

FABRICATION ON METCLAD

Toner transferring using the Laminator on to the Metclad was very difficult in comparison with the toner transfer on to the FR4. Metclad requires to cool down once the toner transfer is made before being put in to cold water. This will allow the unwoven fibreglass structure to absorb the toner on to the copper layer.

Figure 4.7 - Fabricated Diplexer on Metclad

When taking a closer look, it was obvious how the power divide has not been properly transferred on to the board with lines being awry.

Figure 4.8 - Awry lines achieved by toner transfer on Metclad

39

4.5

MEASURED RESULTS FOR THE LOW PASS FILTER

Figure 4.9 -Measured Results for the Low Pass Filter

Measured results show a return loss (S11) of -11dB while the insertion loss (S21) is around -2.4dB. Although the Return Loss value is similar to the simulated value, the insertion loss is quite high. 2nd Harmonics have been suppressed below 20dB (-21dB). The insertion loss can also considered low once cable loss is deducted. The centre frequency is shifted to around 0.95GHz instead of the 0.9GHz mark.

Figure 4.10- Taking Measurements

40

4.6

MEASURED RESULTS FOR THE DIPLEXER

Figure 4.11-Results for the LowPass Filter

The measured Lowpass filter clearly shows that there is something wrong with the fabricated Diplexer. There is a high amount of losses occurring as seen by the S21 Insertion loss being at -4.3443dB. The return loss values are in higher scale than that of the Insertion loss which is unacceptable.

Figure 4.12-Results for the high pass filter

41

The measured result for the high pass filter is quite far off from the simulated results as well. As in the previous case, the Insertion Loss -8.0134 dB which is far away from the desired 0.5 dB. Heavy losses achieved in the diplexer can be attributed to the inaccurate fabrication of the power divider of the Diplexer.

Figure 4.13- The close up of the fabricated Diplexer

As portrayed in figure 4.13, there are several instances where the power divider has not maintained the width of 0.35 uniformly. This is the main cause of the inaccurate results obtained by fabrication.

42

4.7

CONCLUSION

This thesis looked in to microstrip structure design for a better performing diplexer, simulations with good performance were achieved by using a dual band Wilkinson power divider instead of a common port junction design. The fabricated results were quite far off from the simulated results due to complications in Fabrication. Inability to maintain the width of the Power Divider resulted in high losses, thus Insertion loss being further away from 0.5 dB. In a broader sense performance could have been bettered if not for the fabrication limitations that prevented the designs to have a minimum line width of 0.3mm.

4.8

RECOMMENDATIONS & FUTURE WORK

It is important to research more about different Fabrication methods which will help achieve much more accurate results on different substrates in future. Better quality printing and laminating can help achieve a good toner transfer for better fabrication.

Furthermore, the use of dual band Wilkinson power divider in a Diplexer can be researched by looking in to different designs of the Wilkinson Power Divider. This project has taken a step forward by using a Wilkinson power divider instead of a common port junction to form a diplexer, with further research and better fabrication options, better results can be achieved.

It is also recommended to obtain professional software such as Microwave Office or Agilent ADS, which will provide accurate simulations and enable students working on RF/Microwave related projects, achieve better results with higher productivity.

43

References
Chatim, RH 2005, MODIFIED WILKINSON POWER COMBINER FOR APPLICATIONS IN THE MILLIMETER-WAVE RANGE, Master Thesis thesis, DEPARTMENT OF RF-TECHNIQUES / COMMUNICATION SYSTEMS, UNIVERSITY OF KASSEL,GERMANY, pp. 74. Harty, DD 2010, NOVEL DESIGN OF A WIDEBAND RIBCAGE-DIPOLE ARRAY AND ITS FEEDING NETWORK, thesis, WORCESTER POLYTECHNIC INSTITUTE, pp. 97. Jianpeng, W, Jia, N, Yong-Xin, G & Dagang, F 2009, 'Miniaturized Microstrip Wilkinson Power Divider With Harmonic Suppression', Microwave and Wireless Components Letters, IEEE, vol. 19, no. 7, pp. 440-442, Law, C & Cheng, K 2008, 'Compact dual-band power divider design using branch-lines and resistors Only,' Microwave Conference, 2008. APMC 2008. Asia-Pacific, 1-4. Lei, W, Zengguang, S, Yilmaz, H & Berroth, M 2006, 'A dual-frequency wilkinson power divider', Microwave Theory and Techniques, IEEE Transactions on, vol. 54, no. 1, pp. 278-284, Mohra, ASS 2008, 'Compact dual band Wilkinson power divider,' Radio Science Conference, 2008. NRSC 2008. National, 1-7. Pozar, DM (ed.) 2005, Microwave Engineering, John Wiley & Sons,Inc. Wilkinson, EJ 1960, 'An N-Way Hybrid Power Divider', Microwave Theory and Techniques, IRE Transactions on, vol. 8, no. 1, pp. 116-118, Zhang, Y, Tang, X, Fan, Y, Ooi Ban, L, Leong Mook, S & Mouthaan, K 2008, 'A Miniaturized Wideband Wilkinson Power Divider,' Electronics Packaging Technology Conference, 2008. EPTC 2008. 10th, 271-274. (Chatim 2005; Harty 2010; Jianpeng et al. 2009; Lei et al. 2006; Mohra 2008; Zhang et al. 2008)

44

Appendices Appendix A
Agilent Advanced Design System
Agilent ADS is a software that provides an integrated design environment to RF electronic designers. Agilent ADS supports designing of a schematic, layout, time domain and frequency domain simulation and electromagnetic field simulation. Agilent ADS is one of the most popular EM simulator softwares, and one that is being used in the industry regularly.

Getting started with Agilent ADS is very easy, the user is required to create a new project in the user directory provided upon installation.

Figure 1

Once the new project is created, the user may be able to select from a drop down menu on the top left hand corner to select different mediums, components, functions. etc.

45

Figure 2

Once the user has selected what is required, he place components on the schematic. Furthermore, once a circuit has been built, if you are required to produce the layout, there is a simple function of generating the layout, which will automatically generate the layout. This may be done vice versa as well, where you create the layout 1st and generate the layout later.

Figure 3 46

Appendix B
Sonnet Lite EM Simulator
The sonnet lite is a free limited feature version of Sonnets more professional suites. Sonnet provide full wave simulations for planar 3D circuits. It also allows for design and analysis in coupled transmission lines, microstrip matching networks, for planar couplers and splitter analysis etc. This software also can be integrated with Agilent ADS through a small bridging software, thus allowing to import and export layouts and analyse. To get started, go to Edit Project, select New Geometry. A plain page with a square box with dots will appear.

Figure 4

Then, go to Circuit and select Units. Here the units used for the length and frequency are defined, values for Dielectric Layers and Box Size can be assigned as these are listed under Circuit as well.

47

Figure 5

Figure 6

Figure 7

48

During the design, the ports must be touching the box wall as the box wall act as ground for the port, and thus the distance between the resonator and the box wall should be two or three times the substrate height. This is to avoid the interference of the simulation box with the structure.

Appendix C Fabrication Process

1. Import the SONNET file in Corel Draw. Make sure in Corel Draw that the page size is set as Letter if your toner transfer paper is letter size. Initially only the outline of the image will be imported we have to fill it with black colour and remove the outline. 2. Print the layout onto the toner transfer paper. Always print directly from Corel Draw. Never save file in .pdf format and print, because the dimensions gets disturbed. 3. From the toner transfer paper, the toner is transferred to the dielectric board using the laminating machine. 4. Go through the laminating machine for several times 5. After laminating, cool the dielectric board and then soak it in water for 5 minutes. Peel the toner transfer paper gently. 6. Let the surface dry naturally. Do not dry with tissues or clothes. This will cause the toner to wear off and also some particles from the tissue might appear on the board. 7. After the board is dried, run it under the water to remove the chemical layer left from the toner paper. 8. Put a layer of green film onto the layout side and laminate again, at least 20 times. Make sure there are no wrinkles. This will create an additional green layer on the toner which will protect the toner from etching chemical. Peel the green film after laminating.

49

9. The board is then etched inside an etching machine with ferric chloride at a temperature of about 150 o C . 10. Its very important to check the board every few minutes to avoid over etching. After the area besides the coated tracks is totally etched, wash it with water and dry it. 11. Remove the coating with acetone. 12. Solder the ports. The best option is to use solder paste, it will make soldering much easier.

50