Yu.А. Bryuchanov - Dr.Sc. (Eng.), Professor, Head of the Department of Electronic Systems Dynamics, Yaroslavl State University. E-mail: bruhanov@uniyar.ac.ru
Yu.А. Lukashevich - Electronic Engineer of the Department of Electronic Systems Dynamics, Yaroslavl State University. E-mail: dcslab@uniyar.ac.ru

One of effective ways of quantization noise reduction is use of upsampling. Thus transformed signal is sampled with a frequency of fд exceeding the minimum requirements of the sampling theorem fд = 2fв, where fв - the top frequency in a range of an analog signal, that a range of quantization noise is distributed in the excess range of frequencies. As the noise level of doesn't depend on sampling frequency, the frequency (digital) filtration of the transformed signal provides reduction of quantization noise level, and together with it and nonlinear distortions at analog-to-digital transformation of signals. Necessary decrease in fд at the exit of the digital filter is carried out by the compressor of frequency of sampling. Work purpose - the analysis of influence of upsampling of a signal in the analog-to-digital converter with any number of levels quantization (step) (categories in representation of numbers) on nonlinear distortions of harmonious signals. For detection of regularities calculations of nonlinear distortions in analog-to-digital converters with use of the low-pass Battervort filter at periodic influences x'(n)=A cos n and different values of coefficient of redundancy Кизб = ωд/(2ωc), where ωc is the cut frequency of low-pass filter are executed. We believe that for representation of numbers arithmetics with the fixed point and fractional numbers is used. Calculations show that use of upsampling provides considerable reduction of nonlinear distortions at analog-to-digital signals transformation. Extent of influence depends on the size of Кизб. So, at increase in coefficient of redundancy from 5 to 30 (i.e. by 6 times) the coefficient of nonlinear distortions of Кни depending on number of categories and a way of approximation of numbers decreases by 1,7-4,6 times, and at increase in Кизб from 30 to 50 (т.е by 1,7 times) coefficient Кни decreases approximately by 60 times. The received dependences of coefficient of nonlinear distortions of Кни on coefficient of redundancy of Кизб, number of categories and way of approximation of numbers allow to choose the sampling frequency, the necessary parameters of quantizers at creation of systems of transmission of messages with digital processing of signals.