CICC 2001 Ed Sessions: Session 2

IC Design for Wireless Applications

Moderator: Ranjit Gharpurey, Texas Instruments Inc.
Assistant: Vincent von Kaenel, Sibyte, Inc.

E2-1 - 8:00-9:50

Theoretical and Practical Aspects of RF Circuit Design
Oliver Werther, Microtune Corporation

RF circuits are the key components in highly integrated receivers/tuners. The RF performance of amplifiers and mixers is critical to achieve the required dynamic range. This tutorial begins with an overview of important RF parameters (NF, IP3, CTB etc...) . Next, design tradeoffs of popular circuits are discussed. The second part of the tutorial addresses important practical issues. This includes a discussion of simulation techniques to verify the stability of amplifiers and mixers, influence of package parasitics and methods to improve stability. In addition de-embedding techniques (for example baluns), accurate at frequencies over several GHz, will be discussed. These techniques can be applied in many applications and result in very accurate evaluation of noise gain and linearity performance of balanced RF circuits.

A key goal is to develop a good understanding and appreciation of the important issues of the design and evaluation of front-end circuits designed at GHz frequencies.

E2-2 - 10:10-12:00

RF CMOS Devices and Circuits
Tajinder Manku, University of Waterloo

Designing good RF CMOS circuits within the GHz range requires one to have a fairly good understanding of MOS devices at these frequencies. In this short course, the design of critical RF components within a CMOS environment will be reviewed. In the first part of the lecture, the various RF parameters are reviewed (noise parameters, IP3, blocking, network properties, etc.). Next, the basic physics of CMOS devices are reviewed. Topics that will be covered are linearity, small signal parameters, and noise. This will lead to describing small and large signal models of CMOS devices at GHz frequencies. Examples of various circuit topologies including LNAs and up/down conversion mixers will be presented. Designing for noise matching, power matching, gain, and linearity will be addressed.

E2-3 - 1:00-2:50

RF Power Amplifier Fundamentals
Peter Baltus, Philips Semiconductors

The trend for single-chip integration of cellular handset transceivers has thus far largely excluded the RF power amplifier. As a result, the power amplifier is becoming an important contributor to size, cost, and power dissipation of handsets.

This lecture will cover typical power amplifier requirements such as efficiency, power handling, linearity, and gain for second- and third- generation cellular handset power amplifiers. Meeting these requirements with a small and cheap power amplifier will be examined with emphasis on system-level tradeoffs such as linearization techniques, antenna interface, and battery interface.

In addition, some relevant selected topics at the circuit level (tradeoffs between active and passive devices and packaging) and technology level (robustness and thermal properties) will be included.

E2-4 - 3:10-5:00

The Design of Low Phase Noise Oscillators
Thomas H. Lee, Stanford University

This presentation covers the basics needed to design low noise oscillators, by making use of the recently developed linear time-variant model of phase noise. Because the subject of phase noise has been made exceptionally complex through study from widely divergent (and occasionally wrong) perspectives over many decades, this course begins by asking a few fundamental questions: How does one determine if a system (any system) is linear and/or time-invariant? What, if anything, does an impulse response reveal? We then examine simple abstractions for oscillators, and study both theoretical and practical implications. To keep complexity under control, we'll initially ignore amplitude variations. Once we've extracted as much design insight as possible from that simplification, we'll augment the analysis to accommodate dynamic effects of the amplitude control mechanisms always present in practical oscillators.

The design insights provided by the model allow us to appreciate why tuned oscillators are so superior to relaxation oscillators. Furthermore, they provide a clear understanding of how 1/f device and circuit noise transforms into phase noise. More important, theory (and experiment) show that it is possible to suppress by large factors the contribution of 1/f noise by exploiting certain symmetry concepts. Finally, the theory points the way to possible significant improvements, as yet unachieved.