The main topic of this thesis is to analyse and design passive WaveProbe couplers. Thus the task is to:

Problem Description GaAs MMIC technology is used increasingly in components located near the antenna of a radio device - both in infrastructure and equipment at the user end. This is because GaAs technology has good performance both in low noise, power amplifiers and antenna switches (high isolation). As more and more uses higher frequencies in the new wireless services, the use of GaAs technology increases and knowledge about the design of equipment-specific components will be in demand. The main topic of this thesis is to analyse and design passive WaveProbe couplers. Thus the task is to: - Learn to use the microwave simulation program (ADS and Momentum) for simulation and optimization. - Study the Triquint's MMIC phemt process. - Study the principles of the waveprobe couplers. - Simulate and design several waveprobes in MMIC using electromagnetic simulations. - Analyse, simulate and design new variants of waveprobes in MMIC using electromagnetic simulations. - Create several prototype layouts of the waveprobes. - Measure existing waveprobes Preliminary Specification: Bandwidth: 1-20GHz Insertion Loss: <0.2 db Coupling factor: > -15dB Directivity: > 12dB Assignment given: 01 February, 2011 Supervisor: Morten Olavsbråten, IET

Rameshwor Prasad Shah Design of a GaAs MMIC Passive Waveprobes Thesis for the degree of Master of Science Trondheim, July 2011 Norwegian University of Science and Technology Faculty of Information Technology, Mathematics and Electrical Engineering Department of Electronics and Telecommunication.

Abstract GaAs MMIC are taking big space in the wireless communication in today s world. The devices are to be characterized for higher frequencies. External passive waveprobes are used to characterize the devices. As the use and technology is increasing there will be more development on MMICs. It would be nice to have a passive waveprobe on the chip itself so as to minimize the losses created due to junctions. In this master s thesis several wave probes designing techniques are described. The wave probes are designed and simulated. Wave probes are loop couplers with a sufficient directivity and very small insertion loss. The proposed wave probes or coupler are realized using TriQuint GaAS phemt MMIC process. Simulated data for different wave probes are compared and it is found that a loop coupler at the edge of the transmission line radiating electromagnetic is best suited. The electromagnetic radiation causes coupling or cross talk. Radiation at the discontinuities or edge is high and this advantage is used to design a coupler with a high directivity. In the design process size of the device is tried to be kept as small as possible so as to make the cost low. EM-simulation of passive MMIC have been carried out and compared with the experimental result to validate the use of foundry models with real life measurements. i

Preface This thesis is submitted in partial fulfillment of the requirements for the degree of Master of Science Electronics and Telecommunication at University of Gävle, Sweden. The work was carried out in the period February 2011 to July 2011at Department of Electronics and Telecommunications, Norwegian University of Science and Technology (NTNU), Norway under the supervision of Associate Professor Morten Olavsbråten at NTNU. Acknowledgement First and foremost I would like to thank my supervisor Morten Olavsbråten for giving me the opportunity to work with the topic of MMIC wave probes and for his immense continuous support and valuable advices. I would also thank Professor Guennandi Kouzaev at Department of Electronics and Telecommunication, NTNU for his guidance and readiness to help during my work. Both the professors had vast and never ending insight in the topic which had been a great source of motivation and inspiration for this work. Further I would like to thank PhD students Juanito Marrento M and Dragan Mitrevski for their help in lab for measurement and the valuable input in my wok. Finally, special thanks to my friend and fellow student Hakan Yilmaz for his support and to the rest of my friends in Gävle and Trondheim. Rameshwor Prasad Shah Trondheim, Norway, July 2011 ii

Table of Contents Abstract... i Preface... ii Acknowledgement... ii Acronyms... viii 1 Introduction... 1 2 Theory... 3 2.1 Definition of directional coupler... 3 2.2 Principle of Operation... 5 2.3 Scattering Matrix... 7 2.4 Types of Directional Couplers... 9 2.4.1 Coupled-Line Directional Coupler... 9 2.4.2 Asymmetrical Coupled-Line Directional Coupler... 12 2.4.3 Branch-Line Couplers... 14 2.4.4 Lange Coupler... 15 2.5 Concept... 15 3 GaAs MMIC Technology... 17 3.1 Introduction to MMIC... 17 3.2 MMIC Advantage... 17 3.3 TriQuint 0.5µm TQPED Process... 19 4 Design and Simulation... 22 4.1 Microstrip Coupled LINe (MCLIN) Design... 22 4.2 Microstrip Asymetric Coupled LINe (MACLIN) Design... 27 4.3 Matching width of the transmission line... 30 4.4 Proposed Design of the Coupler... 31 4.4.1 Topology1... 31 iii

4.4.2 Topology 2... 33 4.4.3 Desired Design... 36 4.4.4 Topology 3... 37 4.5 Real time design... 38 4.5.1 Variation of length... 38 4.5.2 Variation of Width... 39 4.5.3 Change in position... 40 5 Measurements on Existing Physical Directional Coupler... 42 6 Discussion... 47 7 Conclusion... 49 7.1 Future Work... 49 References... 51 A. Appendix A... 53 B. Appendix B... 54 C. Appendix C... 57 D. Appendix D... 64 iv

List of Figures Figure 2-1 Example of directional coupler: (a) reflective wav caused by mismatching in antenna; (b) directional coupler to detect incident wave at port 2 and reflection wave at port 3... 4 Figure 2-2 Direction of E i (Electric field) and H i (magnetic field) corresponding to (a) the incident wave and (b) the reflected wave. The arrow from E i to H i is the direction of rotation for the right screw. The arrows of P i and P r show the direction of the wave... 5 Figure 2-3 Explanation of the directional coupler by using the characteristics of the difference in the relative directions of E and H between the incident and reflected waves: (a) direction of the incident wave and (b) direction of the reflected wave.... 6 Figure 2-4 different ports are connected by two passes and their phase difference is zero and π to the incident and the reflected wave, respectively: (a) P i appears at port 2; (b) P i disappears at port 3; (c) P r appears at port 3; (d) P r disappears at port 3.... 6 Figure 2-5 Coupled line Directional coupler.... 9 Figure 2-6 Even and odd mode excitation for a coupled line: (a) even excitation; (b) odd excitation.... 10 Figure 2-7 An asymmetrical coupled line couplers.... 12 Figure 2-8 Geometry of branch line coupler.... 14 Figure 2-9 Lange coupler using planar technology, [10]... 15 Figure 2-10 Position of coupling loop with respect to wave guide (left); instantaneous direction of the currents in the loop caused by a passive wave (right). [11]... 16 Figure 3-1 Cross section of TriQuint phemt process, [25]... 20 Figure 4-1 Parameters calculation of MCLIN using ADS LineCalc tool.... 23 Figure 4-2 Result of the MCLIN for (a) quarter wave, (b) decreased length of the transmission line and (c) the desired frequency range.... 25 Figure 4-3 Result for the MCLIN as a result of decreased line length and spacing between the lines.... 26 Figure 4-4 (a) power of harmonics for higher order; (b) Coupling factor for higher frequency.... 27 Figure 4-5 Results of MACLIN with (a)w2=50µm, (b) W2=25µm and (c) W2=10µm.... 29 Figure 4-6 Directivity properties of MACLIN on ADS Momentum.... 30 Figure 4-7 Reflection of the transmission line on TriQuint-tqped for (a) width calculated by linecalc and (b) by random method.... 30 v

Figure 4-8 Proposed design of the coupler.... 31 Figure 4-9 (a) Result of the momentum simulation of the design in Figure 4.8;(b) The transmission line on Metal 1 moved to the edge of the transmission line on Metal 2; Transmission line on metal 1 with width 5µm... 33 Figure 4-10 Structure with different feed line.... 34 Figure 4-11 (a) result for the design for structure in Figure 4.10; (b) Result for change in position of line on metal 1 from middle to the edge of the transmission line on metal 2; (c) Result for increase in length of the line on metal 1 keeping it at the edge of the line on metal 2... 35 Figure 4-12 The result for the desired directional coupler.... 36 Figure 4-13 Design layout for the meander line structure of the line on metal 1.... 37 Figure 4-14 Result of simulation for the structure in Figure 4.13.... 38 Figure 4-15 Plot of Table 4.3... 39 Figure 4-16 Interpretation of Table 4.4... 40 Figure 4-17 Graphical representation of Table 4.5.... 41 Figure 5-1 Analytical probe station... 42 Figure 5-2 (a) Chip with several couplers at the lower end (b) Magnified directional coupler on the chip.... 43 Figure 5-3 DC Probes with termination.... 44 Figure 5-4 Results of the measuring of the existing coupler in the lab for (a) coupled port; (b) Through port; (c) Isolated port.... 46 C-1 MCLIN Schematic.... 57 C-2 MACLIN Schematic... 58 C-3 Results of parameters (a) coupling, (b) isolation, (c) through with change in length of transmission line on metal one.... 60 C-4 Results of parameters (a) coupling, (b) isolation, (c) through with change in width of transmission line on metal one.... 61 C-5 Results of parameters (a) coupling, (b) isolation, (c) through with change in width of transmission line on metal one.... 63 D-1 Isometric view of the designed Waveprobes.... 64 D-2 Top view.... 65 D-3 Bottom view... 66 vi

List of Tables Table 1-1 Design requirements... 2 Table 3-1 Chip cost against size [4]... 19 Table 4-1 Values from the Figure 4.5.... 29 Table 4-2 Tabulation of achieved and required parameters.... 37 Table 4-3 Coupling factor, directivity and insertion loss for the design in section 4.4.3 as for the variation of length of the transmission line on metal 1.... 39 Table 4-4 Coupler parameters for the change in width of the transmission line on metal 1... 39 Table 4-5 Coupler parameters for different position of the transmission line on metal one... 40 vii

Acronyms ADS CAD DC EM EMI FET GaAs GHz GPS GSG GSSG HBT InP LNA MACLIN MCLIN MESFET MIC MIM MMIC NiCr PA Advance Design System (Agilent Technologies) Computer Aided Design Direct Current Electromagnetic Electromagnetic Interference Field Effect Transistor Gallium Arsenide Giga Hertz Global Positioning System Ground Signal Ground Ground Signal Signal Ground Hetero-junction Bipolar Transistor Indium Phosphide Low Noise Amplifier Microsrtip Asymmetrical Coupled Line Microstrip Coupled Line Metal Semiconductor Field Effect Transistor Microwave Integrated Circuit Metal Insulator Metal Monolithic Microwave Integrated Circuit Nickel Chromium Power Amplifier viii

PCS phemt RF SiGe VSWR WLAN Personal Communication System pseudomorphic High Electron Mobility Transistor Radio Frequency Silicon Germanium Voltage Standing Wave Ratio Wireless Local Area Network ix

1 Introduction A wave probe is a loop coupler with a very tiny loop, the loop being significantly smaller than a quarter-wavelength of the highest frequency to be measured [12]. The loop has a very small insertion loss of 0.2dB and has a directivity that is significant for all load-pull measurement. The principle was published more than 60 years ago by H. C. Early [11]. The most important characteristic of the wave probe is its directivity. Directivity of a coupling structure is the quantitative measure of its ability to separate the waves travelling in the transmission line structure. A simple wave probe has a directivity of 12dB and has coupling higher at higher frequencies, which is beneficial for harmonic measurement [12]. As the directivity is good it can be used to characterize the power amplifiers The objective of this thesis is to design a passive wave probe on a GaAs MMIC for the measurement of reflections and monitoring power for the safety of the components located near the antenna of a radio device, both in infrastructure end and user end. The wave probes used externally has more loss due to junctions and brings to a large cost also. A small wave probe on a MMIC would overcome both the problems. Several wave probes are designed using coaxial cables and are discussed in [12, 14 & 15]. In MMIC technology cost is the most important factor, the smaller the size of the chip the less will it cost. Designing a quarter wave directional to get the ideal coupling factor of 3dB and isolation approximately zero would be impossible for a MMIC. A loop coupler as discussed in [12] is designed. Instead of wave guide and a cable two transmission line on two different layers are used. As the passive wave probe is a loop coupled line directional coupler so it will be referred as directional coupler in the rest of the text. The tasks to be performed for this thesis are stated as follows: - To learn to use the microwave simulation program (ADS and Momentum) for simulation and optimization. - To study the TriQuint s MMIC phemt process. - To study the principles of the wave probes couplers. - To similate and design several wave probes in MMIC using electromagnetic Simulation. - To analyze, simulate and design new variants of the wave probes in MMIC Using electromagnetic Simulation. - To create several prototype layouts of the waveprobes. 1

- To measure existing waveprobe. Some design criterion for the design of a passive wave probe coupler has been laid down. The frequency range and response properties would make the coupler useful in the current GaAs MMIC technology application. The design process TriQuint phemt process is used. The requirements are listed in the Table 1.1. Also the design is done so that it has a minimized size. Here only transmission line and vias are used so the design is independent from the coupling of different components. Table 1-1 Design requirements Bandwidth Insertion Loss Coupling factor Directivity 1-20 GHz <0.2dB >-15dB >12dB Chapter 2 explains the basic theory of couplers and concept of wave probes. Different couplers are introduced and their design methods are discussed. The principle of the four port coupler is formulated. Chapter 3 describes the MMIC technology. The advantages of using the MMIC technology are discussed. The effects of different passive components of TriQuint phemt have also been mentioned. In chapter 4 the design process is discussed. Different couplers are designed using ADS Momentum simulation. The result of the simulation is discussed and the best method is proposed. Another chapter 5 describes the measurement taken on an existing coupler on MMIC and the result is presented. Finally in chapter 6 discussions are made on the overall work, before conclusion is drawn in 7. 2

2 Theory 2.1 Definition of directional coupler The electromagnetic wave transmits from a transmitter to an antenna; sometimes, the reflected wave occurs when the antenna is mismatched with the feeder as shown in the Figure 2.1(a). To detect the component of the incident wave and the reflected wave, a four port network, called directional coupler, is used, as shown in Figure 2.1(b). When the incident power, P i, is supplied to port 1, the power appears at port 2 and 4 with the quantities of P i and P i, respectively and not at port 3 [1-3 & 12]. In the same way when reflected power, P r, is supplied to port 4, the power appears at port 1 and 3 with the quantities of P r and P r, respectively, and not at port 2. In this case, the quantity C defined as =10 =10 (2.1) is called coupling of the directional coupler, where coupling defines the quantity of the input power that is coupled to the output port. Here C is attenuated quantity from port 1 to port 2. Coupling is frequency dependent and when a coupling factor is defined it is usually for a given center frequency. (a) 3

Figure 2-1 Example of directional coupler: (a) reflective wav caused by mismatching in antenna; (b) directional coupler to detect incident wave at port 2 and reflection wave at port 3. The description presented above is the ideal case, where P i, does not appear at port 3. In the practical case, however, a very small amount of the power P i appears at port 3. In this case, the value of D defined as =10 = 10 (2.2) is called the directivity of the directional coupler, where directivity measures how good the power is directed in the wanted direction. In case of many applications like feed forward amplifiers directivity is of major concern as the reflection from the antenna might affect the functioning of the device [4]. In the ordinary devices, D is 20 to 30 db [5]. Isolation indicates the fraction of the input power at port 1 to the power at port 3 (isolated port) and is expressed as Or, =10 = 10 (2.3) = + (2.4) 4

2.2 Principle of Operation 2.2.1 Method of Using the Difference of the Relative Direction of E and H Between the Incident and the Reflected Waves. As Shown in Figure 2.2, the relative directions of E and H are opposite to the incident and the reflected waves. The characteristic can be used to construct the directional coupler. Figure 2-2 Direction of E i (Electric field) and H i (magnetic field) corresponding to (a) the incident wave and (b) the reflected wave. The arrow from E i to H i is the direction of rotation for the right screw. The arrows of P i and P r show the direction of the wave When the incident power, P i, is supplied to port 1 in Figure 2.3(a), the circuit is designed such that the quantity A proportional to the E i of P i appear at port 2 and 3 in the same phase and the quantity B proportional to the H i of P i appear to ports 2 and 3 in the opposite phase. If the design is made such that A=B, the incident wave does not appear at port 3. Next when the reflected wave is supplied to port 4, the phase of the magnetic field becomes opposite for the same electric field. Therefore the reflected wave appears at port 3 and does not appear at port 2, as shown in the Figure 2.3(b). 5

Figure 2-3 Explanation of the directional coupler by using the characteristics of the difference in the relative directions of E and H between the incident and reflected waves: (a) direction of the incident wave and (b) direction of the reflected wave. Figure 2-4 different ports are connected by two passes and their phase difference is zero and π to the incident and the reflected wave, respectively: (a) P i appears at port 2; (b) P i disappears at port 3; (c) P r appears at port 3; (d) P r disappears at port 3. 2.2.2 Different Ports Connected by a Multipass When the electric angle between ports 1 and 4 or 2 is π/2, the power of the incident wave supplied to port 1, P i, is added in the same phase and appears at port 2, as shown in Figure 2.4(a). On the other hand, passes I and II have a difference of π radian in electric angle, which results in the disappearance of P i at port 3, as shown in Figure 2.4(b). In the same way, the reflected power, P r, supplied at port 4 appear at port 3 as shown in Figure 2.4(c), and does not appear at port 3, as shown in Figure2.4(d). 6

2.3 Scattering Matrix The directional coupler of Figure 2.1 (b) can be expressed by scattering matrix. The scattering [S] matrix of a four port directional coupler matched at all ports can be expressed as in (2.5). 0 = 0 (2.5) 0 0 According to the principles of directional coupler the port 3 and 2 are isolated or have power zero for forward and backward signals respectively. Hence, = =0 (2.6) This results to the matrix 0 0 = 0 0 (2.7) 0 0 0 0 The self-products of the row of the [S] matrix of (2.3) yield the following equations: + =1 (2.8) + =1 (2.9) + =1 (2.10) + =1 (2.11) Which imply from equation (2.4) and (2.5) that =, and from equation (2.5) and (2.6) that =. Thus = = is chosen, where α is real. On the other hand S 12 and S 34 has a phase shift, so it is denoted as = = Ω assumed to have same phase shift, a case of symmetrical coupler where is real. The dot product of row 2 and 4 gives + =0 (2.12) 7

The phase constant for a symmetrical coupler Ω has to be π/2. / = (2.13) Thus the scattering matrix has the following form: 0 0 0 0 = (2.14) 0 0 0 0 The performance of the coupler can be described with the help of the following four main parameters as stated in earlier section. Coupling: =10 = 20 = 20 db (2.15) Throughput: =10 = 20 = 20 db (2.16) Isolation: =10 = 20 db (2.17) Directivity: =10 =10 =20 db (2.18) Where P 1 is the incident power at port 1, and P 2, P 3 and P 4 is the transmitted power at the coupled port, isolated port and through port respectively. 8

2.4 Types of Directional Couplers 2.4.1 Coupled-Line Directional Coupler Figure 2-5 Coupled line Directional coupler. Figure 2.5 shows a typical layout of directional coupler which is perfectly matched at all four ports. Along the coupled lines, an even and odd mode is propagating. Coupling is provided by the strong electrical field in the gap spacing between the lines, which is due to the odd mode, and the magnetic field lines surrounding the micro strips due to even mode as shown in Figure 2.6 For the even mode, the electric field has even symmetry about the center line, and no current flows between the two strip conductors. Whereas for the odd mode, the electric field lines have an odd symmetry about the center line, and a voltage null exists between the two strip conductors. The uniform and symmetrical coupler can be represented by the scattering matrix (2.7). 9

Figure 2-6 Even and odd mode excitation for a coupled line: (a) even excitation; (b) odd excitation. The four port impedance matrix regarding even and odd mode operation is: [6] = (2.19) Where, = = = = = = = = + (2.20) (2.21) = = = = = = = = + (2.22) (2.23) where, is the normalized electrical length of the line section for the even and odd modes respectively. The scattering coefficient for the two ports is: 10

= (2.24) = (2.25) = (2.26) = (2.27) Where Z is impedance, o and e denoted the odd and even mode respectively and is the electrical length. Complete derivation of the expressions for impedance matrix and scattering can be found in chapter of [6]. If the characteristic impedance, Z 0, and the voltage coupling coefficient, C, are specified, then the design equations (2.28) and (2.29) can be used for the even and odd mode characteristic impedance [7]. = (2.28) = (2.29) This type of coupler is best suited for weak couplings. 11

2.4.2 Asymmetrical Coupled-Line Directional Coupler Figure 2-7 An asymmetrical coupled line couplers. Symmetrical coupler has the capability to transfer full power between two identical guides which are in symmetry whereas only partial power can be transferred in an asymmetrical coupler. The propagation constant of the uncoupled guide determines the symmetry of the couplers. If the guides have same propagation constant then the two guides when brought together will act as a symmetrical coupler on the other hand if the propagation constant differs than the guides are asymmetrical [8]. Even though the phase velocity of the microstrip line varies as the width of the strip is changed, the variation is weak. A large enough velocity difference can achieve broadband by coupling one narrow and one wide strip. Asymmetrical couplers shown in Figure 2.7 can give high directivities. It is well known that a completely symmetrical, lossless four port is an ideal coupler if it is matched at all four ports from the above section. A similar result can be derived for a lossless four port which has only end to end symmetry. [9] Consider the coupler in Figure 2.7 whose port 1 and 4 are similar to each other but port 2 and 3 are similar to each other but different from 1 and 2 that makes the micro-strip line unsymmetrical. Thus from here it can be said that S 31 = S 24. Assuming reciprocity and complete matching at all four ports, the scattering matrix for the coupler in Figure 2.7 can be reduced to scattering matrix 12

0 = 0 0 0 (2.30) By transposing and taking the complex conjugate S t * and multiplying it with S unit matrix is obtained. = (2.31) This condition leads to + =0 (2.32) + =0 (2.33) + =0 (2.34) An isolation property can be achieved by multiplying (2.33) by S 31, substituting for from (2.32), and using (2.34) to express to get 2 =0 (2.35) From equation (2.35) it is concluded that port 1 is input port and one of the three remaining ports must be isolated port. For backward coupler = 0. It is assumed that all the fourports are matched. 13

2.4.3 Branch-Line Couplers Figure 2-8 Geometry of branch line coupler. Branch-line coupler are often made in microstrip or stripline form. The coupler is symmetrical with a quarter wave length (λ/4) lines and matched with the characteristic impedance Z 0. A basic structure is illustrated in Figure 2.8. In the figure Z 0 is the characteristic impedance, λ is the wavelength at the center frequency and Z 0i is the line impedance. The signal supplied at port 1 does not appear at port 4 because the electrical angles through passes 1 2 3 4 and 1 4 are different than π radian, which results in cancelling the waves at port 4. This is the brief concept. For known coupling factor in db and the characteristic impedance the line impedance can be calculated using formulas from [10]; =10 (2.36) = (2.37) Once the line impedance is known CAD tools can be used to find the dimension of the striplines by defining the substrate parameters and center frequency. 14

2.4.4 Lange Coupler Figure 2-9 Lange coupler using planar technology, [10] The lange coupler was first introduced by J. Lange [13]. This type of coupler has numerous tight coupled lines as shown in Figure 2.9 which allow wide bandwidth and gives 3dB coupling. The derivation of the equations is complicated and an approximated equation is given in [7]. The even and odd impedances as calculated by = ( )( ) = ( )( ) (2.38) (2.39) where, =10. The width and the length can be calculated by the help of CAD defining the appropriate substrate and desired center frequency. 2.5 Concept In 1946 H. C. Early, proposed a new type of direction coupler with a small loop, much smaller than the quarter wave (λ/4) that responded to both electrical and magnetic fields. Excellent directional characteristic over a frequency range was obtained when this loop was used in conjunction with special section of ridge wave guide as shown in Figure 2.10 [11]. The ends of the loop are connected to two 50Ω lossy cables. For the correct design relations are maintained. The power in cable A is proportional to the forward wave and in B is 15

proportional to reflected wave. This type of coupler exactly works on the principles explained in 2.2. These couplers are very small and are very advantageous to replace large size couplers. The same can be done in the case of coaxial cables and micro strip. These types of couplers have wide-band, high directivity, low VSWR, good isolation and EMI shielding. Figure 2-10 Position of coupling loop with respect to wave guide (left); instantaneous direction of the currents in the loop caused by a passive wave (right). [11] Similar coupler designed with coaxial cables is used as a wave probe for the load pull application [12]. The loop virtually introduces no insertion loss and has directivity that is sufficient for the load pull measurement application [12]. 16

3 GaAs MMIC Technology 3.1 Introduction to MMIC The acronym MMIC stands for Monolithic Microwave Integrated Circuit. The word monolithic (from Greek) means as a single stone and thus describes that MMICs are fabricated from a single piece of semiconductor substrate [16]. Microwave is the ac signal between the frequency ranges of 300 MHz to 300 GHz. Integrated Circuit contains not only a diode or transistor but the whole electronic circuit on the same semiconductor material. Thus MMIC is a microwave circuit in which the active and passive components are fabricated on the same semiconductor substrate which is measuring a few millimeters per side at the most. MMICs are a logical development of hybrid microwave integrated circuit (MICs) in which active devices like FETs, diodes and passive components like resistors, capacitors are mounted on a dielectric substrate such as alumina [17]. MMICs have different layers of metal, dielectrics, insulators and other alloys on the same substrate in which both the active and passive devices are fabricated. Military and space applications have been a major driving force behind MMIC technology. It is used for transmitting and receiving microwave signals, ranging for low frequency (in gigahertz) equipment like cellular phones, wireless local area networks (WLANs) and global positioning system to the hundreds of gigahertz equipment like Earth observation radiometers and security scanners. As such MMIC has become very important for communication industry with in military, space and civil. Gallium Arsenide (GaAs) semiconductor is one of the most commonly used substrate to develop MMIC circuits. Indium Phosphide (InP) and Silicon Germanium (SiGe) are other substrate used. GaAs has low field mobility and high saturated electron drift velocity which gives an advantage for faster device like cellular phones and other mobile technologies. It also can be made with high resistivity, making it suitable substrate for microwave passive components [19]. On the other hand, Silicon is four times cheaper as compared to GaAs HBT and has same gain and noise properties as GaAs [20]. Though of silicon being cheaper GaAs are used for higher frequencies because of its performance value. 3.2 MMIC Advantage MMIC was developed as to combine high performance microwave transistors with low loss passive components and transmission lines. Complex circuits with multiple interconnections 17

using just a few photolithographic process steps could be formed on this developed technology. Miniaturization of microwave circuits is of importance for portable devices like cellular phones, smart cards and radio transceivers for notebook computers. MMIC as stated earlier reduces the size and mass of the chip as compared to MIC as the circuit is fabricated on the same substrate. There are no active devices mounted from the top as in MIC. The transistors dimensions are in the order of microns, hence the resulting size of the chip is only few millimeters (mm) which is smaller than the hybrid microwave integrated circuits (MICs) with packaged transistors mounted on it. The small dimension of the MMICs means they have less weight then the equivalent hybrid MICs. Miniaturization and lightness of MMIC gives the commercial advantage to the portable products like cell phones. MMICs are more reliable then hybrid circuits, as they do not have junctions in the same substrate like hybrid MICs and are therefore less dependent on the reliability of each and every component. Temperature cycling, shock and vibration have effect on reliability of hybrid circuits because there are chip attachment and wire binding. The fact that the fabrication process is carefully controlled and qualified the reliability is assured. To have a better performance and high reliability manufacturing tolerances must be considered in the simulations. MMICs and MMIC process have high reliability levels that they are used in space-borne applications [21-22]. PCS, GPS, WLAN and mobile technologies are few civil applications of MMICs. These devices have large high volume commercial markets where reliability and manufacturing at consumers prices are of equal importance. More on reliability can be found in [23-24]. The performance of the MMICs is related to the use of active devices. The yield would be very poor because of the multiplicative effect of the yields of individual transistors on the same chip. Therefore in case of LNA and PA, where performance is prime concern, for the very best noise figure and power efficiency it is necessary to use discrete transistors before and after the MMIC. On the other hand passive components give satisfying results regardless of some components like inductors can take large space. Hybrid MIC designer has the independence of choosing the best transistor from different manufacturers for the job to be done. In case of millimeter-wave circuits (i.e., >30GHz) MMIC can be employed with good performance and yield. Tuning by adjusting the lumped circuit is not possible in case of MMIC as all the passive components are formed on the substrate. Unlike hybrid MIC 18

parasitic caused by junction, solder pads and packing is eliminated, which results in better performance of the MMIC. Cost is the main concern when it comes to any device manufacturing. In case of MMIC, the smaller the size of the chip the lower is the price, as many chip can be fabricated on the same wafer each with similar performance and no hand tuning. If the chip size is large or small amount of chip is to be produced than the hybrid MIC is the option though there is question for reliability and performance of the chip. Table 3.1 shows the approximate chip fabrication cost against chip size for a high yield MESFET process using ion implementation [19]. Table 3-1 Chip cost against size [4] Chip Size (mm 2 ) Typical Yield (%) Working circuits per Bare chip cost ($) at 6 wafer $5k per wafer 1 Χ 1 80 12800 0.4 2 X 2 70 2800 1.8 5 X 5 45 288 17 7 X 7 30 98 51 10 X 10 20 32 156 The main drawback of MMIC is that because of its microscopic features involvement on the surface the equipment is installed in the clean room environment to prevent dust and moisture to affect the features. This makes the fabrication process time consuming and expensive. MMICs have over all more advantage to the MICs when it comes to reliability, size and weight, cost for complex circuits. 3.3 TriQuint 0.5µm TQPED Process Design process of the MMIC starts with the choice of the substrate, the type of transistors and passive component. Then the choice of the foundry process is required. There is several foundry processes but TriQuint MMIC phemt process is studied as it is required in this thesis. A cross section of this process is shown in Figure 3.1. 19

Figure 3-1 Cross section of TriQuint phemt process, [25] The entire production process description available from TriQuint is included in [25]. The targeted applications for this process are, according to TriQuint; Low noise amplifiers Linear, low loss and high isolation RF switches. Converters Integrated RF front ends Power detectors and couplers. The process can also be suitable for other applications. The main features of the process are; Depletion mode transistors with -0.8V pinch off and enhancement mode transistors with +0.35V threshold voltage. Two global, and one local, connection layers. High precision NiCr resistors with very low temperature dependence. High value Metal Insulator Metal (MIM) capacitors. 20

There are three metal layers and each have different resistivity and current density capacity. Metal at layer two has the highest current density capacity and lowest resistivity with metal at layer zero has the poorest current density capacity and highest resistivity. These parameters should be considered while doing the layout of the circuit. All of the passive components are designed to have high q-values. It is possible to use bulk resistors for high resistance in a small are, though this is not recommended as they have high temperature coefficient, low accuracy and poor current density capacity. Further information on the process data is available in the process specific design manual from TriQuint, which is only available to the users who have signed the non-disclosure agreement. 21

4 Design and Simulation In this chapter different coupler are designed to achieve the desired goal expected in chapter 1. An asymmetrical loop coupler is most emphasized. Also different topology is tried to decrease the size of the device to make it more price compatible. 4.1 Microstrip Coupled LINe (MCLIN) Design The co-axial directional coupler as discussed in section 2.5 is the form of coupled line directional couplers. As for the design requirements the coupling is -15dB and the circuit is matched to impedance of Z = 50Ω. First the coupling co-efficient linear value is evaluated. 10logC= 15dB ~C= 10 =10 0.178 Using the equations (2.28) and (2.29) the even and odd mode is computed as Z, =Z 1+C 1 C 59.75Ω Z, =Z 1 C 1+C 41.8Ω These values fed into linecal tool of ADS gives the width of the transmission lines, space between the two transmission lines and the length of the lines as in Figure 4.1. Substrate definition is set using the values from Appendix A and a centre frequency of 10GHz is considered to get the shortest λ/4 lines. The length, width and the spacing between the line is 2940.73µm, 53.3µm and 75.22µm respectively. The component used from ADS is MCLIN (Microstrip Coupled LINe). Appendix B1 contains the details of MCLIN. 22

Figure 4-1 Parameters calculation of MCLIN using ADS LineCalc tool. Applying the values for length, width and spacing between the transmission lines on a schematic design in ADS and simulating the circuit in Appendix C2 the result as shown in Figure 4.2(a) is achieved. Port 1, 2, 3 and 4 are input port, coupled port, isolated port and through port respectively. It is clear that at the centre frequency a coupling of -15.421dB is obtained which is the desired value as per the requirement and a directivity of -5.697dB. At higher frequency there is resonance. This can be shifted to higher frequency by decreasing the length of the transmission line from 2940µm to 1500µm as shown in Figure 4.2(b). the resonance has been shifted from 18 GHz to 35 GHz. The concern is only up to 20 GHz. In Figure 4.2(c) coupling at 20GHz is found to be -15.543dB and a directivity of 4.043dB. 23

db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-10 -20-30 m1 m2 freq= 9.000GHz freq= 9.000GHz db(s(2,1))=-15.421 db(s(3,1))=-21.118 m1 m2-40 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (a) 0-10 db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) -20-30 -40-50 0 5 10 15 20 25 30 35 40 freq, GHz (b) 24

db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-10 -20-30 m1 freq= 20.00GHz db(s(2,1))=-15.543 m2 freq= 20.00GHz db(s(3,1))=-19.586 m1 m2-40 -50 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (c) Figure 4-2 Result of the MCLIN for (a) quarter wave, (b) decreased length of the transmission line and (c) the desired frequency range. The isolation is very poor in this case. Isolation can be improved by decreasing the spacing between the lines and by also decreasing the length of the line. Doing this, a very small directional coupler of dimension of 250µm can be archived. The result of such directional is shown in Figure 4.3. Here the line length is 250µm and the spacing between the lines is 5µm. As the spacing between the lines decreases the coupling becomes stronger. 25

db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-10 -20-30 -40-50 -60-70 m1 freq= 20.00GHz db(s(2,1))=-14.849 m2 freq= 20.00GHz db(s(3,1))=-37.091 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 Figure 4-3 Result for the MCLIN as a result of decreased line length and spacing between the lines. These simulation and result of MCLIN provide a good concept of designing the directional coupler. Coupling factor increases at higher frequency this helps in better harmonics measurement as, the power at high order harmonics are very low. Figure 4.4 has the plot for the power levels of harmonics and the coupling that is archived from design. (a) 26

(b) Figure 4-4 (a) power of harmonics for higher order; (b) Coupling factor for higher frequency. 4.2 Microstrip Asymetric Coupled LINe (MACLIN) Design As discussed in previous section line length, spacing between the lines and the width are the parameters those can be changed to get a better coupling and directivity. Though the width of the line is not changed as it is matched to Z 0 =50Ω. The liberty is given to vary only the line length and the spacing between the lines. And these are the two parameters when varied give better result. In case of GaAs MMIC there are different layers of metals. The objective of the thesis work is to couple out the power that is transmitted on metal two from metal one. Hence, there is specific spacing between the lines and cannot be changed. There is restriction of varying the space. MACLIN gives us the independence of varying the width of the lines. The detailed properties of MACLIN are contained in Appendix B2. Keeping one width (W1) constant and varying the other width (W2) with lengths 250µm and spacing as 5.5µm (constant) the results are displayed as in Figure 4.5. 27

db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-10 -20-30 -40-50 -60-70 m1 freq= 20.00GHz db(s(2,1))=-15.546 m2 freq= 20.00GHz db(s(3,1))=-33.670 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (a) db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-20 -40-60 -80 m1 freq= 20.00GHz db(s(2,1))=-15.070 m2 freq= 20.00GHz db(s(3,1))=-39.178 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (b) 28

db(s(4,1)) db(s(3,1)) db(s(2,1)) db(s(1,1)) 0-10 -20-30 -40-50 -60-70 m1 freq= 20.00GHz db(s(2,1))=-14.710 m2 freq= 20.00GHz db(s(3,1))=-37.616 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 Figure 4-5 Results of MACLIN with (a)w2=50µm, (b) W2=25µm and (c) W2=10µm. Table 4-1 Values from the Figure 4.5. Width (µm) Coupling (db) Directivity (db) 50-15.546 18.124 25-15.070 24.108 10-14.710 22.906 (c) Coupling decreases with the decrease of width but is not significant, directivity on the other hand is observed to initially increase and then decrease after a certain value. Figure 4.6 illustrates the Table 4.1 29

30 Directivity of Maclin 25 Directivity (db) 20 15 10 5 0 0 10 20 30 40 50 60 Width (µm) Figure 4-6 Directivity properties of MACLIN on ADS Momentum. 4.3 Matching width of the transmission line From the Figure 4.1 the width of the transmission line is 53.304µm, which is supposed to be matched to Z0=50Ω. The width was used in ADS momentum and simulated for the transmission line length of 500µm on a TriQuint_tqped. The result as in Figure 4.7(a) shows that the line is not perfectly matched. Hence a random method is used and it matches exactly with the width of 87.5µm and the result shown in Figure 4.7(b). S11 S11 freq (1.000GHz to 20.00GHz) (a) freq (1.000GHz to 20.00GHz) (b) Figure 4-7 Reflection of the transmission line on TriQuint-tqped for (a) width calculated by linecalc and (b) by random method. 30

4.4 Proposed Design of the Coupler 4.4.1 Topology1 Figure 4-8 Proposed design of the coupler. A design was proposed at the beginning of the project which was based on the simple coaxial cable directional coupler as discussed in section 2.5. Here a transmission line of 500µm in length was constructed on the metal layer two of the GaAs TriQuint_tqped with a width of 87.5µm (matched to Z0=50Ω) is taken as shown in Figure 4.8 (Blue colored transmission line). Another transmission line of width 25µm and a length of 50µm are taken on metal layer one of the GaAs TriQuint_TQPED to couple out the power on the transmission line on metal layer two. The green line is the transmission line on layer one. Two feed lines of width 10µm was constructed on metal layer zero (red line), which was connected to the transmission line on layer one by a passive component TQPED Metal 0 to Metal 1 vias as shown in Figure 4.8. These feed lines are connected to two different ports on Metal 2 by TQPED Metal 0 to Metal 2 vias. In the Figure 4.8 the ports number 1,2,3and 4 are input, coupled, isolated and through ports respectively. The distance between the main transmission line to the ports 2 and 3 is kept enough so that there is no coupling between the line and the ports on Metal 2. The transmission line on metal 2 is fed with power which develops an electromagnetic coupling onto the line on metal 1 as the space between the two lines is very small. 31

This structure was run under ADS Momentum Analysis simulation as it can perform electromagnetic (EM) simulations. The normal ADS schematic simulation can only simulate the EM effects for components like Maclin and Mclin that can be constructed on the same layer. In case of MMIC there are different layers. The result obtained is shown as in Figure 4.9. db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 m1 freq= 20.00GHz db(s(2,1))=-14.632 m2 freq= 20.00GHz db(s(3,1))=-18.758 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (a) db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 m1 freq= 20.00GHz db(s(2,1))=-15.218 m2 freq= 20.00GHz db(s(3,1))=-19.242 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (b) 32

db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 -60 m1 freq= 20.00GHz db(s(2,1))=-21.240 m2 freq= 20.00GHz db(s(3,1))=-33.591 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (c) Figure 4-9 (a) Result of the momentum simulation of the design in Figure 4.8;(b) The transmission line on Metal 1 moved to the edge of the transmission line on Metal 2; Transmission line on metal 1 with width 5µm Keeping all the parameters constant in Figure 4.8 the coupling in case of Figure 4.9(a), is - 14.632 db and the power at isolated port is -18.758 db at 20 GHz. Here the directivity is very low (18.758-14.632 = 4.126dB). The desired coupling (>-15dB) is achieved but the directivity is found to be very poor. The transmission line on metal 1 is shifted towards the upper edge of the transmission line on metal 2. The result is observed to be -15.218dB as coupling and -19.242dB as isolation as in Figure 4.9(b). Here also, the directivity is 4.024dB which is not improved. The target is to improve the directivity as for other parameters are changed. The width of the transmission line on metal 1 is changed to 5µm and the line is positioned at the upper edge of the transmission line on metal 2. Here a desired directivity of 12.351dB (>10dB) is achieved at the highest frequency but at lower frequency the directivity seems to be still poor. 4.4.2 Topology 2 From topology 1 the desired results are not obtained so the structure of the feed line is changed here and then the different parameters are changed and again simulated to find the results. From previous topology it is discovered that when the width of the transmission line on metal 1 is decreased better directivity is achieved. Therefore, in the structure in Figure 4.10 the width of the transmission line on metal 1 is kept as 5µm and the other parameters are same as in topology 1. 33

The result obtained for the figure above is shown in Figure 4.11(a). Here it is seen that the coupling is -19.417dB and the directivity is 10.708db and the directivity is same for the lower frequency also which is not as in topology 1. Figure 4-10 Structure with different feed line. db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 -60 m1 freq= 20.00GHz db(s(2,1))=-19.417 m2 freq= 20.00GHz db(s(3,1))=-30.125 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (a) 34

db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 -60-70 m1 freq= 20.00GHz db(s(2,1))=-22.018 m2 freq= 20.00GHz db(s(3,1))=-36.312 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 -40-50 -60 (b) m1 freq= 20.00GHz db(s(2,1))=-15.439 m2 freq= 20.00GHz db(s(3,1))=-29.939 0 2 4 6 8 10 12 14 16 18 20 freq, GHz m1 m2 (c) Figure 4-11 (a) result for the design for structure in Figure 4.10; (b) Result for change in position of line on metal 1 from middle to the edge of the transmission line on metal 2; (c) Result for increase in length of the line on metal 1 keeping it at the edge of the line on metal 2 The metal 1 transmission line position is changed from middle to the upper edge of the metal 2 transmission line. This arrangement gives much better directivity of around 14.294dB but the coupling factor is low as seen in Figure 4.11(b). From the results of Figure 4.11(a) and 4.11(b) it is easily verified that the directivity is better in case the coupling line width is 35

smaller and the position of the coupling line is at the edge of the line carrying the signal. Here still one problem persists that is the coupling factor is poor. To improve the coupling factor the length of the coupled line is increased to 150µm keeping all other parameters as before. The result obtained for this change in length gives better coupling of -15.439dB and a directivity of 14.5dB as seen in Figure 4.11(c). From the result of Figure 4.11(a), 4.11(b) and 4.11(c), it is found that the three parameters width, length and the position of the coupled line varies the coupling factor and the directivity of the device in this case. 4.4.3 Desired Design 0 m3 freq= 20.00GHz db(s(4,1))=-0.198 m3 db(s(1,1)) db(s(4,1)) db(s(3,1)) db(s(2,1)) -10-20 -30-40 -50-60 m1 freq= 20.00GHz db(s(2,1))=-15.439 m2 freq= 20.00GHz db(s(3,1))=-29.939 0 2 4 6 8 10 12 14 16 18 20 freq, GHz Figure 4-12 The result for the desired directional coupler. The directional coupler designed with the transmission line on metal 1, to have length of 150µm, width of 5µm and the positioned at the edge of the transmission line on metal 2 acquires the result shown in Figure 4.12. The above result is tabulated in Table 4.2 m1 m2 36

Table 4-2 Tabulation of achieved and required parameters. Factors Coupling Factor(dB) Directivity(dB) Insertion loss (db) Achived -15.439 14.5 0.198 Required > - 15 >10 <0.2 Conclusion Approximately Fufill Fulfill The Table 4.2 clearly states that all the specifications are achived. The coupling factor has a slight deviation which is considerable. The 3-D view of the design is appended in Appendix D 4.4.4 Topology 3 Meander line: Meander line is done to get small physical length with large electrical length [26]. The greatest challenge in MMIC technologies is to get the cost per chip reduced. Therefore the smaller the chip is the lower is the cost as discussed in section 3.2. Meander trace would decrease the length fo the transmission line and hence help in achieving smaller device. The topology 3 uses meander trace for the length of 150µm. The design layout is shown in Figure 4.13 and the result is presented in Figure 4.14. Figure 4-13 Design layout for the meander line structure of the line on metal 1. 37

db(s(4,1)) db(s(3,1)) db(s(2,1)) 0-10 -20-30 m1 freq= 20.00GHz db(s(2,1))=-14.853 m3 freq= 20.00GHz db(s(4,1))=-0.263 m3 m1 m2-40 -50 m2 freq= 20.00GHz db(s(3,1))=-22.309 0 2 4 6 8 10 12 14 16 18 20 freq, GHz Figure 4-14 Result of simulation for the structure in Figure 4.13. The coupling has improved slightly to -14.853 but on the other hand diretivity is poor (6dB) and insertion loss is more. This can be as a result of the coupling in the meander line structure itself. Hence this topology is not adviseable. 4.5 Real time design A good design is always verified by checking for the tolerance. The tolerance specified for the width of transmission line by TriQuint is 0.1µm [25]. In the followin sections different parameter are varied and then the effect on the coupling factor, directivity and the insertion loss is analysed. 4.5.1 Variation of length Length of the transmission line on metal one is varied keepin its position at just under the edge of the line on metal two. The width is kept to be 5µm. The result for the coupling, directivity and insertion loss is tabulated in Table 4.3. ADS result is attched to the appendix. Figure 4.15 illustrates the Table 4.3, it is observed that all the parmeters coupling, directivity and insertion loss have a linear nature. Thus with the tolarance of 0.1µm can the device can be easily fabricted to get the same results.the variation in length effects the coupling factor but tolerance of 0.1µm can be achieved. The graphical result from ADS is attached to appendix C3. 38

Table 4-3 Coupling factor, directivity and insertion loss for the design in section 4.4.3 as for the variation of length of the transmission line on metal 1. Length (µm) Coupling (db) Isolation (db) Directivity (db) Through (db) 50-22.275-37.748 15.473-0.057 100-18.009-32.409 14.4-0.118 150-15.428-29.845 14.417-0.200 Length Variation 20 15 10 Power (db) 5 0-5 -10-15 0 50 100 150 200 Couplin (db) Through (db) Directivity (db) -20-25 Length (µm) Figure 4-15 Plot of Table 4.3 4.5.2 Variation of Width As discussed in section 4.2 directivity is directly effected by the variation of the width of the transmission line. For increase in width the directivity falls. Table 4.4 and its interpration Figure 4.16 verifies the fact. The width of the transmission line on metal one is varied. The result is quite linear thus the tolarance of 0.1µm is satisfactory. The graphical result from ADS is attached to appendix C4. Table 4-4 Coupler parameters for the change in width of the transmission line on metal 1. Width (µm) Coupling (db) Isolation Directivity (db) Through (db) (db) 4-15.754-30.667 15.473-0.185 39

5-15.428-29.845 14.4-0.200 6-14.726-27.585 14.417-0.236 Width Variation 20 15 10 Power (db) 5 0-5 -10 0 2 4 6 8 Coupling (db) Directivity (db) Through (db) -15-20 Width (µm) Figure 4-16 Interpretation of Table 4.4 4.5.3 Change in position Electromagnetic radiation is maximum at the discontinuities like joints and edges of the Transmission lines. Thus the results in the Table 4.5 verifies the statement. The directivity is maximum and the insertion loss is the least when the tramission line on metal one is kept just uner the edge of the transmission line on metal two carrying rf signal. From the Figure 4.17 it is observed that the variation of position with 0.1µm doesn t have much difference on the parameters but if the line on metal one is moved away from the line carrying RF signal then there is drop of coupling and the directivity.the position is changed by sliding the transmission line on metal one under the transmission line on metal two. The graphical result from ADS is attached to appendix C5. Table 4-5 Coupler parameters for different position of the transmission line on metal one. Position Coupling (db) Isolation Directivity (db) Through (db) (db) Edge -15.754-29.845 14.091-0.200 Middle -13.249-24.753 11.504-0.330 Outside -20.010-30.709 10.699-0.078 40

Position Variation Power (db) 20 15 10 5 0-20 -5 0 20 40 60-10 -15-20 -25 Position Coupling (db) Directivity (db) Through (db) Figure 4-17 Graphical representation of Table 4.5. The varition of the length, width and position of the transmission line on the metal one of the GaAs MMIC the three factors coupling, directivity and isolation vary. But if the design is made with the least width, reasonable length of around 150µm and placed near the edge of the transmission line on metal two the desired goal can be acquired with a tolarance of 0.1µm. 41

5 Measurements on Existing Physical Directional Coupler Measurement on the physical directional couplers those were available in the microwave lab designed by Professor Morten Olavsbråten earlier in 2009 were carried out using probe station in microwave lab at NTNU. The probe station has been connected to a network analyzer to get S-parameters readings. The network analyzer has been interfaces with the personal computer and thus results can be seen on the pc also. The probe station has three probes the right and left probes are to measure the S-parameters and at center is the dc feed probe. The dc probe has 4pins GSSG (Ground Signal Signal Ground) and the other probe has three pins GSG (Ground Signal Ground). Picture of the probe station can be seen in Figure 5.1. Before taking the measurements the probe is calibrated to open, short and matched load. The directional coupler to be measured is seen in Figure 5.2. Figure 5-1 Analytical probe station 42

(a) (b) Figure 5-2 (a) Chip with several couplers at the lower end (b) Magnified directional coupler on the chip. 43

The directional coupler as in Figure 5.2(b) is a four port device but the probe station has just two probes. The device in this case is a passive device and the DC feed probe is not needed for feeding the DC. Thus here the DC feed probe is also used here to measure the signals at 2 and 3. As seen in Figure the coupler has two pads on port 1 and port 4 where three pins of the probes can be easily situated. The center pin as stated earlier is used for the S-parameter measurement and the outer two pins are used as a ground which is connected to ground through substrate via. A special way of measurement is performed. The two ports of the probe station are calibrated to open, short and matched load. The measurement for the through port (S 41 ) is measured by terminating the port 2 and port 3 with a terminator via a DC feed probe which no more is a DC feed probe but acts like a terminator as can be seen in Figure 5.3. Similarly coupled port (i.e.s 21 ) is measured by terminating port 3 and 4. Also the isolated port (S 31 ) is measured by terminating port 2 and port 4.The result are shown in Figure 5.4. Figure 5-3 DC Probes with termination. 44

db(datasetname..s(2,1)) -10-15 -20-25 -30-35 m1 freq= 20.00GHz db(datasetname..s(2,1))=-14.819 m1-40 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (a) db(datasetname..s(1,1)) db(datasetname..s(2,1)) 10 0-10 -20-30 m1 freq= 20.00GHz db(datasetname..s(2,1))=-0.568 m1-40 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (b) 45

-20 m1 db(datasetname..s(2,1)) -25-30 -35-40 -45-50 m1 freq= 20.00GHz db(datasetname..s(2,1))=-20.910 0 2 4 6 8 10 12 14 16 18 20 freq, GHz Figure 5-4 Results of the measuring of the existing coupler in the lab for (a) coupled port; (b) Through port; (c) Isolated port. From the figure it is found that the coupling factor is -14.819dB, transmission loss is 0.568dB and isolation port has a reading of -20.910Db thus resulting a directivity of -6.091dB. The directivity seen here is too less. 46

6 Discussion A concept used by Early more than 60 years before explained in paper [11] has been developed to produce waveprobes. Waveprobes are loop couplers which have satisfactory directivity for the load pull measurement. There are coaxial or transmission line couplers where the length of the loop of the couplers is smaller than the quarter wave length for a certain frequency. The task to be performed was to design a similiar coupler on a GaAs MMIC technology as it is now being used vigoursly in wireless communication. Firstly, a deep knowledge was acquired on the coupler theory and the different types of coupler. Several directional coupler has been designed on MMICs but the problem is the size. The size of the chip is directly proportional to the cost. Thus the coupler had to be as smaller as possible. Secondly the MMIC technology was studied to know the effects of different components and their behaviour. A good knowledge on the TriQuint phemt foundry was gained. After these studies the rigrous design process were simulated ADS Momentum. ADS Momentum was used because it has the EM-simulation and as coupler was to be designed EM-simulation was the must. Several designs were simulated and analysed. Finally the target was achived with the transmission line on metal one having width to be 4µm length to be 150µm and the position of the line was just under the upper edge of the transmision line on. metal two as shown in Figure 4-10.Then also several analysis was done to get much better result. It was found that when the line on metal one was moved to the centre of the upper metal line the couplin was increased as there is more capacitance because of less EM-fields at the centre and also the isolation was poor. Hence it was advisable to keep the line at the edge. The minimum line width specified by TriQuint is 2µm. A design was made using width of 4µm and a desirable directivity was obtained. Design with lower width gave better directivity but the coupling was very low. The length of the line was found to be large so a smaller length was analyzed and it was found that coupling had become poor as the resistance had decreased. Also menderline structure was tried; to get the length decreased but the results were not satisfactory as it can be seen in Figure 4.14. the different structure of feed probes also effected the results. The result of Figure 4.9 and 4.12 shows the effect for the two different structure of the feed lines. So while designing it has to be kept in mind to minimize the coupling between the ports and the feed line. 47

Also the designs were compared so that the design can be within the tolerance limit and it was found from Figures 4.15 to-4.17 that a real time design is possible. Finally a waveprobe (coupler) present in the NTNU lab was measured with the help of a probe station. The results were analysed and found that it had very poor directivity and a higher insertion loss. Few drawbacks those were faced during the design were the simulation was done in Momentum which was time consuming. So the design shuld be mades as symmetrical as possible so the mesh created is less which will lead to faster solution as it will have less cell to calculate. Momentum optimaztion for TQPED was complex and not possible to be done on ADS 2009. 48

7 Conclusion Several waveprobe couplers were designed and among those one satisfactory waveprobe coupler on a GaAs MMIC was realized. Three important parameters that affected the design were length, width and the position of the transmission line on metal one. Coupling factor is directly proportional to the length of the line. Directivity is inversely proportional to the width. Finally position of the line at the edge gives better directivity so the position should be kept at the edge to get better directivity cause directivity is of high importance in case of the coupler. So the length, width and the position of the line on metal 1 must be considered while designing a coupler. At high frequency (20GHz) the coupling is maximum (-15dB) and directivity of more than 12dB is obtained. Directivity is seen to be same throughout the range of 1-20GHz as shown in Figure 4-12. This type of coupler will be of great use in load pull measurements where higher harmonics are to be considered as the coupling factor is higher at higher frequency or in other words higher order harmonics. It can be also used as power detector because it has a satisfactory directivity. The most important benefit of having such device is that it is on-chip device which helps to get rid of the losses produced by external device because of its junctions. Measurement of previous manufactured coupler was performed and the directivity was found to be very poor. The design mentioned in the thesis easily overcome the problem of directivity. This coupler is much smaller in size designed on GaAs MMIC and has better performance. 7.1 Future Work Though the targets were approximately fulfilled for this work but there are always scope for the developement of the device. The design was mainly designed with preassumption by a CAD so not much of mathematical calculations were analysed to make the design for much better performance. Hence some task for future work is proposed which was not carried out due to the time constraint and lack of resources. Some of the future works are listed as follows: The size of the device could be decreased more having a good directivity and coupling factor. Size plays a important role in designing of MMICs. Wide-band coupler could be designed using different toplogies. 49

This coupler could be used to design Butler Matrix explained in [27]. The coupling could be improved more for the coupler to act as a splitter. 50

References [1] Jean Verspecht, Jean-Pierre Teyssier, Fabien de Groote. Affordable Large-Signal Network Analyzer Technology. Power Amplifier Symposium USA, 2007 [2] Jean Verspecht, Jean-Pierre Teyssier, Fabien de Groote, Tony Gasseling, Oliver Jardel. Introduction to Measurements for Power Transistor Characterization. IEE Microwave Magzine, Vol 9, No. 3, pp 70-85, USA June 2008. [3] Jean Verspecht, Jean-Pierre Teyssier, Fabien de Groote, Jad Faraj. High power on wafer capabilities of a time domain load-pull setup. 71 st ARFTG Conference Digest, USA, 2008 [4] Peter B. Kenington. High-Linearity RF Amplifiers Design. Artech House, 2000 [5] Yoshihiro Konishi. Microwave Electronic Circuit Technology. Maracel Dekker, Inc, 1998. [6] Gunter Kompa. Practical microstrip design & Application. Artech House, 2005. [7] David M. Pozar. Microwave Engineering. John Wiley & Sons, Inc, 1998. [8] S. E. Miller. Coupled wave theory and wave guide applications. Bell Syst. Tech. Vol33. pp.661-719. [9] Perttik. Ikalainen and Geroge L. Matthaei. Wide-Band, Forward-Coupling Microstrip Hybrids with High Directivity. MTT-35, NO 8, August 1987. [10] Guillermo Gonzalez. Microwave Transistor Amplifiers, pp 327-333. Prentice-Hall, 1997. [11] H. C. Early. A Wide-Band Directional Coupler for Wave Guide. 1946 [12] Fabien De Groote, Jean-Pierre Teyssier, Tony Gasseling, Olivier Jardel, and Jan Versecht. Introduction to Measurements for Power Transistor characterization. IEEE microwave magazine, June 2008 [13] J. Lange. Interdigitated Stripline Quadrature Hybrid. IEEE Trans. Microwave Theory Tech. MTT-17, pp 1150-1151, 1869. 51

[14] W. L. Firestone. Analysis of Transmission Line Directional Couplers. IEEE Journal, 1954, pp 1529-1538. [15] R. C. Knechtli. Further Analysis of Transmission Line Directional Couplers. Processing of the IRE, 1955, pp 867-869. [16] Marsh, Steve. Practical MMIC Design. Artech House, 2006, page 2. [17] Peter H Ladbrooke. MMIC Design: GaAs FETs and HEMTs. Artech House, 1989. [18] Houman Mohebbi. Compact 3dB quadrature coupler implemented in MMIC. IET NTNU. 2007. [19] I. D. Robertson and S. Lucyszyn. RFIC and MMIC design and technology. TIEE, London. 2001. [20] James M. Moniz. Is SiGe the future of GaAs for RF Applications? Gallium Arsenide Integrated Circuits (GaAs IC) Symposium, 1997. IEEE. [21] Roesch, W.J. ; Rubalcava, T; Hanson, C. Lifetesting GaAs MMICs under RF stimulus. Microwave and Milimeter-Wave Monolithic Circuits Symposium, 1992. Digest of papers, IEEE 1992. [22] Cazaux, J. L. MMIC for Space-Bourne Applications: Status and Prospectives. GaAs IC symposium, 1994. [23] Sammy Kayali, George Ponchak, and Roland Shaw. GaAs MMIC reliability Assurance Guideline for space applications. JPL Publicatio, NASA, California Institute of Technology, Pasadena, California, 1996. [24] Ira Deyhimy. Gallium arsenide joins the giants. IEEE Spectr., 32(2:33-40,1995. [25] Triquint Semiconductor. TQPED Design Manual. Verson 2.2 [26] Frank M. Caimi. Meander Line Antennas. August 2002. [27] Simon Schroter. Butler Matrix with Lumped Element Directional Couplers. Uni Ealangen-Nurmberg, Germany, 12 December 2007. 52

A. Appendix A A1 TQPED EM Simulator Parameters EM Simulator Parameters for TQPED, no MIM GDSII layer T Mat. Cond. Er L Tan Re(Mu) Im(Mu) um S / m PSN 0.8 SiNx 6.8 0.0004 1 0 17 ME2 4.0 Gold 4.1e7 ILD2 3.2 BCB 2.8 0.0006 1 0 16 VIA2 1.2 Gold 4.1e7 15 ME1 2.0 Gold 4.1e7 ILD1 1.0 BCB 2.8 0.0006 1 0 14 VIA1 0.6 Gold 4.1e7 9 ME0 0.4 Gold 2.632e7 GaAs 85 GaAs 12.9 0.0006 1 0 EM Simulator Parameters for TQPED, with MIM layer T Mat. Cond. Er L Tan Re(Mu) Im(Mu) um Sie / m PSN 0.8 SiNx 6.8 0.0004 1 0 17 ME2 4.0 Gold 4.1e7 ILD2 3.2 BCB 2.8 0.0006 1 0 16 VIA2 1.2 Gold 4.1e7 15 ME1 2.0 Gold 4.1e7 ILD1 1.0 BCB 2.8 0.0006 1 0 14 VIA1 0.3 Gold 4.1e7 23 MIM 0.2 Gold 2.632e7 CSN 0.1 SiNx 6.8 0.0004 1 0 9 ME0 0.4 Gold 2.632e7 GaAs 85 GaAs 12.9 0.0006 1 0 53

B. Appendix B B1: MCLIN (Microstrip Coupled Lines) Symbol Illustration Parameters Name Description Subst Substrate instance name Units Default None MSub1 W Line width mil 25.0 S Space between lines mil 10.0 L Line length mil 100.0 Temp Physical temperature (see Notes) C None W1 (for Layout option) Width of line that connects to pin 1 mil 0.0 W2 (for Layout option) Width of line that connects to pin 2 mil 0.0 W3 (for Layout option) Width of line that connects to pin 3 mil 0.0 W4 (for Layout option) Width of line that connects to pin 4 mil 0.0 Range of Usage 0.01 H W 100.0 H 0.1 H S 10.0 H 1 Er 18 T 0 Simulation frequency (GHz) where Er = dielectric constant (from associated Subst) H = substrate thickness (from associated Subst) T = conductor thickness (from associated Subst) W Š 0, S 0, L 0 for layout W1 0, W2 0, W3 0, W4 0 54

B2: MACLIN (Microstrip Asymmetric Coupled Lines) Symbol Illustration Parameters Name Description Subst Substrate instance name Units Default None MSub1 W1 Width of conductor 1 mil 25.0 W2 Width of conductor 2 mil 10.0 S Conductor spacing mil 5.0 L Conductor length mil 100.0 Temp Physical temperature (see Notes) C None WA (for Layout option) Width of line that connects to pin 1 mil 0.0 WB (for Layout option) Width of line that connects to pin 2 mil 0.0 WC (for Layout option) Width of line that connects to pin 3 mil 0.0 WD (for Layout option) Width of line that connects to pin 4 mil 0.0 55

Range of Usage 1 Er 18 T 0 0.01 H W1 100.0 H 0.01 H W2 100.0 H 0.1 H S 10.0 H Er = dielectric constant (from associated Subst) H = substrate thickness (from associated Subst) T = conductor thickness (from associated Subst) Simulation frequency (GHz) W1 > 0, W2 > 0, S > 0, L > 0 for layout WA 0, WB 0, WC 0, WD 0 56

C. Appendix C Simulation Setups and Results: C1: MCLIN Simulation S-PARAMETERS S_Param SP1 Start=1.0 GHz Stop=20 GHz Step=1.0 GHz MSub MSUB MSub1 H=85 um Er=12.9 Mur=1 Cond=4.1e7 Hu=3.9e+34 T=4 um TanD=0.0006 Rough=0 um Term Term2 Num=2 Z=50 Ohm Term Term1 Num=1 Z=50 Ohm MCLIN CLin1 Subst="MSub1" W=53 um S=75 um L=2940.73 um Term Term4 Num=4 Z=50 Ohm Term Term3 Num=3 Z=50 Ohm C-1 MCLIN Schematic. 57

C2: MACLIN Simulation. S-PARAMETERS S_Param SP1 Start=1.0 GHz Stop=20 GHz Step=1.0 GHz MSub MSUB MSub1 H=85 um Er=12.9 Mur=1 Cond=4.1e7 Hu=3.9e+34 T=4 um TanD=0.0006 Rough=0 um Term Term2 Num=2 Z=50 Ohm Term Term1 Num=1 Z=50 Ohm MACLIN CLin1 Subst="MSub1" W1=87.5 um W2=50 um S=5.5 um L=250 um Term Term4 Num=4 Z=50 Ohm Term Term3 Num=3 Z=50 Ohm C-2 MACLIN Schematic 58

C3: Result for change in length -15-20 -25 db(s(10,9)) db(s(6,5)) db(s(2,1)) -30-35 -40-45 -50 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (a) -25-30 -35 db(s(11,9)) db(s(7,5)) db(s(3,1)) -40-45 -50-55 -60-65 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (b) 59

0.00-0.05 db(s(12,9)) db(s(8,5)) db(s(4,1)) -0.10-0.15-0.20 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (c) C-3 Results of parameters (a) coupling, (b) isolation, (c) through with change in length of transmission line on metal one. C4: Variation of width -10-15 -20 db(s(10,9)) db(s(6,5)) db(s(2,1)) -25-30 -35-40 -45 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (a) 60

-25-30 -35 db(s(11,9)) db(s(7,5)) db(s(3,1)) -40-45 -50-55 -60 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (b) 0.00-0.05 db(s(12,9)) db(s(8,5)) db(s(4,1)) -0.10-0.15-0.20-0.25 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (c) C-4 Results of parameters (a) coupling, (b) isolation, (c) through with change in width of transmission line on metal one. 61

C5: Change in Position -10-15 -20 db(s(10,9)) db(s(6,5)) db(s(2,1)) -25-30 -35-40 -45 0 2 4 6 8 10 12 14 16 18 20 (a) freq, GHz -20-25 -30 db(s(11,9)) db(s(7,5)) db(s(3,1)) -35-40 -45-50 -55-60 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (b) 62

0.00-0.05-0.10 db(s(12,9)) db(s(8,5)) db(s(4,1)) -0.15-0.20-0.25-0.30-0.35 0 2 4 6 8 10 12 14 16 18 20 freq, GHz (c) C-5 Results of parameters (a) coupling, (b) isolation, (c) through with change in width of transmission line on metal one. 63

D. Appendix D 3 D View of the designed Waveprobe D-1 Isometric view of the designed Waveprobe. 64

D-2 Top view. 65