Level quantization effects in digital signal processing by discrete Fourier transform method

The paper is aimed at drawing the attention of developers of measuring and computing devices based on modern microelectronic elements to a crucial problem of digital signal processing - quantization of the signal levels. Study of the quantization effects and of distortions caused by those effects is carried out for spectral analysis based on discrete Fourier transform. The overflow effect (signal cutting per level due to shortage of the quantization device bits) is also considered. A formula is derived for transforming a real number to its nearest quantization level. Numerical modeling tests are performed on level-quantized digital realizations of harmonic oscillations (pure ones and mixed with noise) to construct characteristics - dependencies between program-assigned signal parameters and those measured in the course of digital processing under various signal quantization and cutting conditions.