11.1 Introduction

Analogue sampled-data signal processing has been dominated for the past decade by the switched-capacitor technique. It has proved to be a reliable workhorse in a wide range of signal processing applications, both linear and non-linear, over a frequency range spanning audio and video. So why should a new sampled-data technique such as switched-currents be needed?

Following its inception in 1972,[1] switched-capacitors gained favour as a technique for implementing active filters capable of greater precision and compactness than earlier active-RC filters, especially in low frequency applications (e.g. telephony). Its repertoire has grown to include more general signal processing[2] and at the height of its popularity it was used to implement complete sampled-data subsystems[3]. But by the mid-1980's, systems which had previously been made using switched-capacitors were starting to find solutions using digital signal processing. Although often more dissipative and less silicon-area-efficient, the digital approach offered easier computer-aided design with shorter time-to-market. Test methods were simpler and as VLSI feature sizes diminished, complexity rose (and continues to rise) so that complete systems can now be integrated on a single silicon chip. This has changed the role of switched-capacitors. Single chip digital systems require analogue circuits at their interface with the outside world and switched-capacitors are now commonly found performing A-D and D-A conversion, sample and holding, filtering etc. at these interfaces.

Switched-capacitors have never fitted standard VLSI processing. Their need for linear floating capacitors has led to special process options, e.g. double polysilicon, being added to the digital VLSI...