The Approaches for Reducing the Complexity of Quadrature Demodulation Implementation

The usage of quadrature demodulation for measuring parameters of sinusoidal and polyharmonic signals is well known. The wide usage of quadrature demodulation is primarily for frequency measurement due to its wide capabilities for ensuring high metrological characteristics. At the same time, this method has a number of disadvantages associated with the complexity of implementation, high dynamic component of the error, and the impossibility of simultaneous measurement of parameters of several spectral components. This paper is devoted to solving the first issue: ensuring the simplicity of the method implementation under conditions of limited performance of the element base. The main problem of the method implementation is associated with the need to use two filters at the demodulator output. The paper proposes the most computationally simple type of output filter - a moving average filter. The main feature of such a filter is the presence of zeros in the magnitude response at certain frequencies associated with the filter length. Since the interference of the demodulator output signal has spectral components at specific frequencies, which are known as a result of frequency measurement by the demodulator itself, the moving average filter is theoretically capable of completely suppressing them. This can be achieved by adaptively adjusting the filter length to the current value of the input signal frequency. However, the accuracy of this approach is limited due to the discrete nature of the filter length parameter, which greatly limits the extreme achievable precision of the demodulator. The paper proposes the so-called “vernier” approach to adjusting the filter length. It consists of approximating the signal at the last sample time interval and integrating at this interval of the approximation curve. Stepwise, piecewise linear functions are considered as approximating polynomials. The good achievable precision of measuring the considered signal parameters is estimated. Simulation models are performed in Simulink program that allow estimating the ultimate errors of the proposed technique and the features of practical implementation.