|
Methods for processing a frequency-modulated pulse signal are presented in this paper. The methods allow acquiring information about the frequency of a pulse signal with constant sampling frequency. These methods, depending on the time instant for which the measurement result is determined, can be divided into two groups. In measurements on-line the measured value is determined based only on the position of pulses preceding the moment of measurement. The method in which the measured value is determined from the last period and that in which the measured value is determined from extrapolation of two last periods (Fig. 1) are described. In measurements off-line, first there is saved the location of all pulses in the time of recording, and after there is determined the value measured in the assumed moments of the measurement. Therefore, in this case the measured value is determined not only by the position of pulses before the moment of measurement, but also those occurring after this time. The method in which the measured value is determined from the current period and that in which the measured value is determined from interpolation of two periods (Fig. 2) are described in the paper. The accuracy of conversion is determined by a relative error or the number of bits, and thus by the resolution. The noise added to the useful signal has a significant impact on the processing accuracy in real measurements. Therefore the ENOB (Effective Number of Bits), the number of digital signal bits above the noise, is more useful than the resolution. The paper presents examples of simulation results for speed measurements. As it can be seen from the results, increase in the standard generator frequency allows increasing the effective number of bits only to a certain limit. Further increase in the frequency, due to the noise being a result of the method used, doesn`t increase the ENOB value ratio.
|
