Design and Implementation of Wavelet Time-Frequency Analysis for Multi-Domain Oscilloscope

Wavelet time-frequency analysis is widely used in electronic measurement and fault diagnosis areas. For multi-domain oscilloscope (MDO) applications, the ability to analyze the spectrum of wideband signals is demanded. Unfortunately, limited by the digital signal processing capability of the MDO, the frequency resolution is too low to inspect the wideband signal accurately. Moreover, the frequency domain is independent of the time domain, which means the information in the time-frequency domain is ignored. In this work, an optimized wavelet structure is designed based on the coefficient approximation extraction algorithm. The coefficient decimation scheme provides a flexible way for frequency resolution adjustment. The wavelet transform is performed in the Field programmable gate array device(FPGA), and the time resolution of the wavelet transform can be manipulated by controlling the input data accessing. These wavelet results are uploaded to the Industrial Personal Computer and provide a time-frequency perspective in real time. The wavelet scale parameter and coefficients are stored on the hardware side, these parameter can be accessed flexibly, which accelerate the wavelet transform process. By omitting the nonzero values of the parameters, resource consumption has been reduced by 30%. The time-sharing multiplexing technology is applied in the wavelet transform unit to save the multiplier resource. The designed wavelet structure is evaluated on a verification platform, the BPSK and AM modulation signal are used for the verification. The input signal has a bandwidth of 8 GHz and a sampling rate of 20 GSPS. The trend of the frequency and phase shift can be detected accurately, the temporal resolution of time-frequency analysis is 1us, and the frequency resolution is up to 1 MHz.