Introduction

Digital techniques have a number of advantages over their analog counterparts for electronic signal processing. Digital signals are represented by only two states, high and low (`1' and `0') and are hence less sensitive to noise than analog signals. Further, the design and analysis of digital circuits is less problematic than analog circuits, because digital circuits require less precision in models for evaluation and simulation.

As a consequence of these advantages, the main thrust of electronic circuit integration has been in digital solutions. Much effort has been devoted to the development of digital, rather than analog, circuits, tools, and applications. However, some form of analog signal processing will always be required in electronic systems. Firstly, the real world is analog by nature. Signals are continuous in time and amplitude, and analog techniques are required for the acquisition and conditioning of real world signals. Secondly, the effective bandwidth which can be processed with analog techniques is much higher than with digital techniques. Analog techniques will always be required for high performance, high bandwidth applications.

Analog circuits are often necessary in larger integrated systems in order to process analog inputs from sensors and perhaps to produce analog outputs. Although the data processing in such integrated systems is performed by digital circuits, signal conversions from analog to digital form and vice versa must take place. There is thus a need for D/A and A/D converters, analog filters, amplifiers, etc., and for this we need basic analog building blocks. In this note we describe a few important analog building blocks and some analysis techniques.

In contrast to digital circuits, where we distinguish between two discrete signal values, high and low (``1'' and ``0''), the signals in analog circuits are currents and voltages on a continuous scale. Such analog currents and voltages will usually consist of a fixed quiescent operating point on which are superposed some (small) fluctuations representing the signal:



Here V and I represent the quiescent operating values, while and represent the signal values, which are functions of time t. The following convention will be used: Capital letters will be used for quiescent values, and lower-case letters will be used for signal values.

Analysis of linear analog circuits usually proceeds in two stages:

1. First, all quiescent values are determined by using the usual transistor equations with signal values at zero. In so doing we determine the quiescent operating point of the analog circuit.
2. Second, the so-called small-signal equivalent diagram of the circuit is formed. This diagram is used to determine dependences between the different signal values. The small-signal equivalent diagram is a linearised model of the circuit valid for reasonably small signal values.

In analyzing small-signal equivalent diagrams we use the Laplace transformation in the usual manner. Signals are regarded as complex entities, and dependences between different signals (impedances and transfer functions) are regarded as functions of the complex frequency .

Analog design differs from digital design not only by the analysis techniques, but also from the required precision of the implemented structures. To achieve the potential high bandwidth requires precision components, and as a consequence, analog designs are sensitive to variations in the process parameters. Further, the combination of analog and digital circuitry on the same die introduces several problems related to noise. An important part of analog design is concerned with reducing the influence of random process variations and noise. This results in layout considerations and structures, which have no counterpart in digital designs.

Chapter 2 gives an introduction to the practical limitations of MOS structures. The chapter discusses, how performance is affected by variations in process parameters. Methods are presented to minimise the consequences of undesired effects by better layout. In Chapter 3, small signal models of MOS transistors are discussed. Several different small signal models and their applications are reviewed. The MOS current mirror is reviewed in Chapter 4 using the models presented, and the consequences of parameter variations are analysed. Suggestions are given for the layout of structures, which are less sensitive to process variations. The differential pair is the subject of Chapter 5, and are analysed similar to the current mirror. Finally, Chapter 6 considers techniques to reduce noise coupling between the analog and digital domains of a mixed analog-digital integrated circuit.