The Load-pull Method of RF and Microwave Power Amplifier Design

John F. Sevic

• Publisher: Wiley
• Year: 2020
• Print ISBN: 9781118898178
• eBook ISBN: 9781119078036
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Electrical Engineering & Telecommunications
• Index: Technology & Engineering

Using the load-pull method for RF and microwave power amplifier design

This new book on RF power amplifier design, by industry expert Dr. John F. Sevic, provides comprehensive treatment of RF PA design using the load-pull method, the most widely used and successful method of design. Intended for the newcomer to load-pull, or the seasoned expert, the book presents a systematic method of generation of load-pull contour data, and matching network design, to rapidly produce a RF PA with first-pass success. The method is suitable from HF to millimeter-wave bands, discrete or integrated, and for high-power applications. Those engaged in design or fundamental research will find this book useful, as will the student new to RF and interested in PA design.

The author presents a complete pedagogical methodology for RF PA design, starting with treatment of automated contour generation to identify optimum transistor performance with constant source power load-pull. Advanced methods of contour generation for simultaneous optimization of many variables, such as power, efficiency, and linearity are next presented. This is followed by treatment of optimum impedance identification using contour data to address specific objectives, such as optimum efficiency for a given linearity over a specific bandwidth. The final chapter presents a load-pull specific treatment of matching network design using load-pull contour data, applicable to both single-stage and multi-stage PA's. Both lumped and distributed matching network synthesis methods are described, with several worked matching network examples.

Readers will see a description of a powerful and accessible method that spans multiple RF PA disciplines, including 5G base-station and mobile applications, as well as sat-com and military applications; load-pull with CAD systems is also included. They will review information presented through a practical, hands-on perspective. The book:

• Helps engineers develop systematic, accurate, and repeatable approach to RF PA design
• Provides in-depth coverage of using the load-pull method for first-pass design success
• Offers 150 illustrations and six case studies for greater comprehension of topics

1
Historical Methods of RF Power Amplifier Design

1.1 The RF Power Amplifier

The RF power amplifier (PA) is a fundamental component of anything communicating wirelessly, being found in the ever ubiquitous mobile phone, the laptop, pad and tablet computers, as well as being found in many other applications, such as medical diagnostic tools, RF heating equipment, and navigation systems. Compared to adjacent sections of the radio, such as the receiver, and digital signal processing functions, such as the baseband modem, the RF PA can draw substantially more power, and thus is exposed to increased scrutiny due to its impact on battery life or base-station operating expense. The RF power transistor represents a significant cost factor because of die size, nature of the semiconductor material, and package cost, particularly for wireless infrastructure applications. Application-specific constraints also exist, particularly in 4G and 5G wireless systems, due to the time-varying envelope whose fidelity must be maintained, mandated by strict air-interface standards for spectral emissions.

Defining an RF PA is usually arbitrary, and is almost always relative. For example, the 2 W provided by a mobile RF PA might be considered high power compared to the CMOS transceiver IC driving it, whereas the 1 kW infrastructure RF PA may also be considered high power. Nevertheless, in contrast to the small-signal amplifier, the RF PA is unambiguously and uniquely distinguished by three key properties.

The RF PA operates in the large-signal regime where the voltage and current traverse the entire active region of the transistor, instantaneously probing, and sometimes dwelling in, cut-off, and saturation. In contrast, the small-signal amplifier operates in the weakly nonlinear regime, exhibiting limited instantaneous excursions over the load-line, with negligible harmonic content. Because of this, superposition holds for the small-signal amplifier, and its optimum terminating impedances and performance over the entire Smith chart are uniquely and accurately described by its four s-parameters alone.

In contrast, because superposition does not hold over the large-signal operating regime of the RF PA, extrapolation over the entire Smith chart is not possible. The strong nonlinearities exhibited by the RF PA leads to the phenomenon of mixing products, a superset of intermodulation distortion, whereby spectral content at integer multiples of the stimulus appears at both the input and output. How these individual mixing products are terminated establishes the performance of the RF PA, particularly its linearity and efficiency, although these cannot be a priori known. RF PA design requires that measurements be made within a predefined region of the Smith chart to uniquely identify optimum terminating impedances and performance, but only in the area bounded by predefined region, since superposition does not hold.

At RF and microwave frequencies, the wavelength of the stimulus is on the order of the size of the physical dimensions of the network embedding RF power transistor die or package. This distributed-parameter environment, coupled with the extremely low impedance required by modern high-power RF power transistors, poses a challenge for synthesis and design of matching networks. To establish optimum performance in this environment, composed of multiple parasitic resonances, frequency-dependent losses, and distributed effects, considerable time-intensive experimentation is necessary to identify the optimum terminations that establish absolute performance. This is followed by design of a suitable matching network replicating the required impedance terminations at the relevant mixing product frequencies, usually at the harmonic and baseband frequencies. In contrast, small-signal amplifier design is only concerned with

-parameters at the fundamental frequency. This uncertainty, and need to present a specific impedance at many discrete frequencies, means matching network design for the RF PA involves multiple iterations and extensive trial-and-error design.

Management of heat, and its consequences, is one of the major challenges RF power transistor and PA designers face. Heat results in reduced gain, reduced effective power density, and impaired linearity. Moreover, each of these manifestations of heat generation is accentuated when the wavelength of the source signal is on the order of the physical dimensions of the die, as is often the case for high-power applications. In addition to the aforementioned effects on performance, reliability can be impaired or catastrophic failure can be induced due to localized self-heating, particularly under mismatch conditions, as commonly encountered in the handset PA environment.

Because of these three properties, RF PA design has often been described as a black art. Those endowed with these mystical powers of PA design methods rely upon different techniques of physical impedance synthesis to search for the optimum terminating conditions, while using indirect measurements to infer that an operating target or trade-off condition has been achieved. With a DC current meter and an RF power meter, one can infer optimality, without actually having to see the time-domain voltage and current signals of the RF PA. How these physical impedances are synthesized, and how the resultant measured data is used for matching network design, forms the essence of all RF PA design methods.

1.2 History of RF Power Amplifier Design Methods

Until the launch of practical commercial automated load-pull in the 1980s, two broad classes RF PA design methods were commonplace. The first class was based on a physical implementation of a variable impedance network. The input and output of the transistor were terminated by this variable impedance network, and while manually adjusting impedance, measurements showed performance that allowed identification of optimum impedance terminations to be located. The most common of these methods was the shunt-stub coaxial tuner, augmented by copper tape and an X-Acto knife for tuning in micro-strip media. Once optimum performance was identified, the transistor was removed and the input and output terminating impedances were measured with VNA, and then replicated by an appropriately designed matching network. Repeating this process several times facilitated construction of data contours superimposed on a Smith chart or rectangular impedance plane as an aid in identifying performance maxima and minima and the terminating impedances endowing the target performance.

The second class relied on analytical expressions, possibly supplemented with measured data or transistor data-sheet parameters, to approximate the optimum terminating impedance terminations. The most important of these methods is the Cripps method [1]. This method identifies an optimum load impedance, followed by design of a test-fixture whose impedances replicate those identified as optimum by the Cripps analysis.

Both classes require the impedance terminating the input and output of the transistor be known, at the fundamental frequency, and possibly the harmonic and baseband frequencies. The terminating impedances serves as an aid in contour construction, design of matching networks, or, often times, both. The key benefit of automated load-pull over these methods is a priori empirical determination of optimal impedance terminations and the ability to rapidly measure fully de-embedded performance. Automated load-pull resulted in a dramatic reduction in design cycle-time and post-fabrication tuning while simultaneously providing accessibility to RF PA design to a much larger group of engineers than before. In fact, the standard set by automated load-pull is so high that it has been difficult to displace it by newer methods, particularly EDA and time-domain methods, because the incremental improvement provided against the additional cost often does not lead to an appreciable difference in performance or reduction in design cycle-time.

1.2.1 Copper Tape...

Table of contents

• Cover
• Table of Contents
• List of Figures
• List of Tables
• Acronyms, Abbreviations, and Notation
• Preface
• Foreword
• Biography
• 1 Historical Methods of RF Power Amplifier Design
• 2 Automated Impedance Synthesis
• 3 Load-Pull System Architecture and Verification
• 4 Load-Pull Data Acquisition and Contour Generation
• 5 Optimum Impedance Identification
• 6 Matching Network Design with Load-Pull Data
• Index
• End User License Agreement