7 - Hardware Approach to Digital Signal Processing

Abstract

The discrete Fourier transform (DFT) is a mathematical operation that is performed on a finite length of contiguous discrete time samples to produce an equivalent number of frequency samples. Signal levels, noise levels, and harmonic content can all be calculated from the DFT output. The usage of coherent sampling eliminates leakage and the requirement for windowing. If noncoherent sampling is chosen, the input frequency selection has fewer restrictions. The fast Fourier transform (FFT) is simply an algorithm that is used to greatly reduce the number of mathematical calculations needed to perform the DFT output spectrum. The primary application of coherent sampling using FFT is in the testing analog-to-digital converters (ADCs) using sine wave inputs. One of the fundamental processes in digital signal processing (DSP) is the development of digital filters. Analog filters are limited in their performance and are susceptible to passive component fluctuations over time and temperature. The characteristics of digital filters can be changed by software control. This means that digital filters have a great advantage in the processing of signal from sensors, digital audio, and mobile radio. The procedure for designing digital filters is very similar to the design of analog filters. This chapter describes the architecture of the Analog Devices ADSP-2101. Many other DSP architectures follow similar design blocks. The ADSP processor contains three independent units: the arithmetic logic unit (ALU), the multiplier–accumulator (MAC), and the shifter.