Dedicated Digital Processors: Methods in Hardware/Software Co-Design

8

Digital Signal Processors

8.1 DIGITAL SIGNAL PROCESSING

To illustrate how to design and implement application-specific processing systems we consider a special application domain, the processing of continuous (‘analog’) signals. The importance of this area is obvious from the fact that the signals occurring in physical systems are primarily of this kind. Classically, such signals have been processed through electronic components such as networks of resistors, capacitors, operational amplifiers and non-linear circuits, implementing filters, modulators, and computational operations. Complex processing functions realized with such means suffer from the lack of reproducibility, precision, dependency on temperature, noise, etc. which are not present in digital processors (at the expense of much higher circuit complexities).

8.1.1 Analog-to-Digital Conversion

The first step to be taken to process a continuous input, e.g. a time-varying voltage e(t) ranging in some interval [U₁, U₂], by means of a digital processor is to perform a conversion of the input value at a given time t₀ into an encoded number (the A/D conversion). e(t₀) is called a sample of the time function e taken at t₀. Reciprocally, the digital processor outputs encoded numbers that have to be converted into a continuous signal with a proportional magnitude (the D/A conversion).

The most basic step towards performing such a conversion is the comparator that compares the input voltage e to a reference voltage r. The output ...