Survey of Fully Digital FFT Hardware

Although numerous methods of calculating digital Fourier spectra are known today, they are all variations of the Danielson‐Lanczos (symmetry of cosine functions) or of the Cooley‐Tukey algorithm (successive decomposition of one‐dimensional data strings into two‐dimensional arrays). Hardware implementation of an FFT algorithm can result in higher spectrum generation speed, higher real‐time bandwidth of data analysis, faster refresh rate of control functions in servo‐applications, or in instruments of limited flexibility geared for special applications. The cost and performance characteristics of any FFT hardware depend on how the major tasks are implemented: complex multiplication, phasor generation, data and phasor indexing, and memory management. For completeness, the less glamorous, but equally important, parts of signal analysis systems should also be considered: overall system control, data acquisition, user interface, retrieval of results, etc. Only in the context o{ the complete system can the FFT hardware be fully evaluated.