AbstractThe wide bandwidth in the unlicensed 60 GHz band enable short range wirelessdata transfer in the order of tens of gigabit per second. This, combined with therelatively lowcost, very high performance SiGe fabrication processes, has led to afeasibility study of a 60 GHz transceiver at Heidelberg University. Such a wirelesssystem is useful in very many applications, and is proposed as the readout of thefuture upgraded trackers in the ATLAS detector in the Large Hadron Collider. Thework presented here is the design of the power amplifier block in the transceiver.The power amplifier design consists of three cascaded stages of common emitters,using the heterojunction bipolar transistors in the 0.13 µm SiGe BiCMOS processfrom IHP. Class AB operation ensures a good tradeoff between efficiency and linearity. With load pull simulation, the output impedance was optimized for maximumoutput power. This was achieved without suffering from a bad output return lossand reflections, because conjugate matching was also attained at the output. Performance simulations yield a power gain of 21.5 dB, a bandwidth of 9 GHz and apeak power added efficiency of 19%. The output referred 1 dB compression pointwas simulated to 6.5 dBm, for which the amplifier consumes 24 mW.A layout of the power amplifier circuit measuring 0.15 mm2 is proposed. Verificationtests like electromagnetic simulation and corner and yield analysis remain, afterwhich the presented block design can be implemented in the top level design of thetransceiver chip.ivPrefaceThis work has been carried out at the University of Bergen in collaboration withHans Kristian Soltveit at the Physikalisches Institut, Heidelberg University. Radio frequency electronics is a fairly new field for the microelectronics group at theUniversity of Bergen, where a fellow student of mine, Magnus Pallesen, and I tookthe challenge of designing two blocks of an integrated 60 GHz transceiver. We hadno previous knowledge in microwave engineering, and much time was dedicated tolearning the theory and new design techniques. This thesis covers my work on themain and final project of my master’s degree, which started August 2015.AcknowledgmentsFirst and foremost I would like to thank my supervisor Kjetil Ullaland for his guidance and support. My gratitude also goes to Hans Kristian Soltveit for makingthis project possible, for his encouragement and for sharing his vast ASIC designexperience. I would like to acknowledge Yngve Thodesen at the Royal NorwegianNaval Academy, his help and guidance on RF design has been invaluable.I will also recognize my friends and fellow students for two good years at the university. Special thanks to Magnus Pallesen for good teamwork in mastering new fieldsof electronics. I am also thankful for the support and motivation I have receivedfrom family and friends throughout the work.Hans SchouBergen, June 2016vviContentsPreface v1 Introduction 11.1 Background and Motivation . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Wireless Systems . . . . . . . . . . . . . . . . . . . . . . . . . 21.1.2 Why 60 GHz? . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Power Amplifier Design Goals . . . . . . . . . . . . . . . . . . . . . . 41.3 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Process Technology 72.1 The SiGe HBT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.2 IHP SG13S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2.1 Bipolar Transistors . . . . . . . . . . . . . . . . . . . . . . . . 82.2.2 Passive Components in IHP SG13S . . . . . . . . . . . . . . . 103 Microwave Theory 113.1 Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113.1.1 Telegrapher Equations . . . . . . . . . . . . . . . . . . . . . . 11viiviii Contents3.1.2 Travelling Waves on a Transmission Line . . . . . . . . . . . . 123.1.3 Reflection Coefficients . . . . . . . . . . . . . . . . . . . . . . 143.2 Scattering Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 153.3 Smith Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163.3.1 Using the Smith Chart . . . . . . . . . . . . . . . . . . . . . . 163.4 Skin Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 Amplifier Design 214.1 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214.2 Amplifier Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234.3 Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.4.1 Tests for Stability . . . . . . . . . . . . . . . . . . . . . . . . . 254.4.2 Stability Circles . . . . . . . . . . . . . . . . . . . . . . . . . . 264.4.3 Interstage Stability . . . . . . . . . . . . . . . . . . . . . . . . 274.5 Impedance Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.5.1 Conjugate Matcing . . . . . . . . . . . . . . . . . . . . . . . . 294.5.2 Loadline Matching . . . . . . . . . . . . . . . . . . . . . . . . 294.5.3 Load Pull . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314.5.4 Impedance Transformation with Lumped Elements . . . . . . 314.5.5 Impedance Transformation with Transmission Lines . . . . . . 344.6 Amplifier Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 375 Schematic Design and Simulation 395.1 Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395.2 Transistor Characterization . . . . . . . . . . . . . . . . . . . . . . . 395.3 Three Stage Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Contents ix5.3.1 Output Stage . . . . . . . . . . . . . . . . . . . . . . . . . . . 415.3.2 Driver and PreDriver . . . . . . . . . . . . . . . . . . . . . . 455.3.3 Bias Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 485.4 Complete Amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495.4.1 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516 Design Realization and Layout 536.1 Replacing Inductors with Transmission Lines . . . . . . . . . . . . . . 536.2 MIM Capacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.3 Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.4 PostLayout Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 557 Discussion 597.1 Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597.2 Gain Flatness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607.3 Simulation Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . 617.4 CAD Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617.5 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618 Conclusion 63A Circuit Schematics 65B Ideal Simulation Results 69C Layout 73D Post Layout Simulation Results 75E Test Benches 79x ContentsGlossary 79Bibliography 87CHAPTER 1IntroductionThe demand for wireless electronics is ever increasing as more and more gadgets areconnected wirelessly, and technological devices are being improved constantly. Thisdrives the research and development of tomorrow’s electronics, and the performanceof wireless technology is advanced in all possible aspects. To mention a few, thedata transfer rates are improved for faster communication, and better efficiencyallows for smaller batteries in handheld devices. There is always a tradeoff betweenperformance and cost, but positive trends are apparent regarding this as well.Wireless technology is usually utilized to allow a device to be mobile. However, theadvancements made in radio frequency electronics has in some specific applicationsenabled wireless data transmission to outperform the traditional wired connections.The work presented in this thesis is a part of a feasibility study of a 60 GHz wirelesssystem for applications requiring short range multigigabit data transfer.1.1 Background and MotivationThe European Organization for Nuclear Research (CERN1) is upgrading the ATLASsilicon microstrip trackers in the Large Hadron Collider (LHC). The trackers aredetecting the trajectory, momentum and energy of new particles created in a collisionof particles21. A full readout of the upgraded detector will require a data transferrate of 50100 Tbs. The optical links of current trackers restrict the data bandwidthdue to limitations in power budget, mass and physical measures 5. The feasibilityof 60 GHz wireless data transfer is being studied at Heidelberg University 28, andthis work is a part of the development of a prototype in that study.1An acronym for the French Conseil Europen pour la Recherche Nuclaire.12 Chapter 1. IntroductionFigure 1.1: A crosssection of the trackers layered around the collision point. Theantennas illustrate the radial readout of each tracker 28.Figure 1.1 illustrates the detectors placed in layers around the collision point, and theproposed radial wireless readout of the tracker data. Each layer detects particlesas they traverse the trackers towards the outer enclosure. Data recorded by thedetectors must be read out, for which the proposal is a wireless radial readout. Thedistance between the tracker layers are approximately 10 cm. The silicon trackersattenuate millimeter waves,2 which is why the signal is brought through subsequenttrackers by additional sets of antennas.1.1.1 Wireless SystemsIn order to communicate wirelessly, a radio transmitter and receiver (combined theyare called a transceiver) are necessary, both of which are shown in Figure 1.2. Totransmit, the data signal is modulated and mixed to a carrier frequency suppliedby a local oscillator. The mixed signal is then amplified by a power amplifier andfiltered before it is transmitted by the antenna.An antenna picks up the signal that has weakened in strength during transmissionto the receiver. The signal must be amplified and discriminated from noise by a lownoise amplifier. Then the signal is filtered and downconverted with a mixer. Oncethe signal is demodulated, it should be the same as the data signal that was provided2Millimeter waves range from 30 GHz to 300 GHz, where the wavelengths range from 10 mmto 1 mm, respectively.1.1. Background and Motivation 3Figure 1.2: The basic building blocks of a transmitter (top) and a receiver 28.to the transmitter to begin with. The transmission depends on the performance ofall the system blocks not to lose too much information.1.1.2 Why 60 GHz?With the increasing number of wireless devices and their rising performance requirements, the usable frequency spectrum is very crowded. Large parts of the spectrumare licensed, and the allocated bands for specific applications require very advancedmodulation techniques to enable high data transfer rates. Communication is alsoheavily encoded to reduce interference between different channels in the same band1.Advances in technology increase the useable spectrum by enabling wireless systemsto utilize frequencies far into the GHzrange, at fairly low cost. The globally unlicensed 60 GHz band varies in some regions, as shown in Figure 1.3, but most ofthese regions have a total bandwidth of 7 GHz. This enables very high data transferrates. Dependent on the modulation scheme, 7 GHz bandwidth is capable of tensof gigabits per second.3Higher carrier frequency increases the free space path loss and the signal attenuationin materials. This means shorter transmission range and that the receiver should316QAM (quadrature amplitude modulation), with its spectral efficiency of 4 bps per hertz, iscapable of 28 Gbps. Onoff keying can provide 3.5 Gbps at the same bandwidth.4 Chapter 1. IntroductionFigure 1.3: The unlicensed spectrum around 60 GHz for different regions 10.be in the line of sight from the transmitter. These disadvantages are consideredadvantages in applications with short transmission ranges; high attenuation reducesthe likelihood of interference of separate systems. Short range transmission is alsoless susceptible to eavesdropping since the receiver has to be fairly close to thetransmitter.The short wavelength of 60 GHz (5 mm in vacuum), enables very small antennadimensions. In some applications, the antenna can be integrated on chip, acquiringa remarkably small form factor of the wireless system.1.2 Power Amplifier Design GoalsThe goal of this work is to design a power amplifier, shown as the last active blockof the transmitter in Figure 1.2. It is responsible for providing the antenna with asignal strong enough to be picked up by the antenna in the receiver. To do so, thepower amplifier in this work must be designed to meet the requirements listed inTable 1.1, which is the main goal. The power gain and 1 dB compression point isspecified for the center frequency, 61.5 GHz. The power amplifier should be ready tobe integrated on the transceiver chip with the other system blocks. Electromagneticsimulation should verify the performance of the power amplifier. The fabricationtechnology chosen for the chip is the 0:13 µm SiGe BiCMOS process provided byIHP Microelectronics.The obtained bandwidth is defined by where the reflection coefficients, S11 and S22are within specification. Inside the passband, the gain should be as flat as possible,and outside it, the gain should be kept at a minimum. The input and output shouldboth be matched to 50 Ω.1.3. Outline of the Thesis 5Table 1.1: The required performance specifications for the PA.Specification Goal UnitPower gain (S21) 20 dB1 dB compression point >5 dBmPower consumption