Research On Reconstruction And Application Of Nonuniformly Sampled Signals

With the development of high speed data acquisition technique, the research of nonuniform sampling has become an important component of digital signal processing subject. Periodic nonuniform sampling is a special case of nonuniform sampling pattern, with the well-known time-interleaved analogue-to- digital converter (TIADC) technique as its practical module in engineering. TIADC technique is an effective way to implement ultra-high speed data acquisition system with relatively slow circuits by using multiple ADCs operated in parallel mode. However, TIADC systems degenerate into periodic nonuniform sampling because of time-skew mismatch errors cased by the imperfect environment. The mismatch errors must be corrected in practical use, since they degrade the performance of TIADC system significantly. In fact, the calibration of TIADC system's time-skew error is equal to the reconstruction of periodic nonuniformly sampled signals. Based on the above background, the reconstruction of periodic nonuniformly sampled signals and its application in TIADC system are researched in the thesis.Basic theories of periodic nonuniform sampling are studied firstly. The time domain module and the digital spectral representation of the periodic nonuniformly sampled signal are derived and the influence of sampling nonuniformity on the sampled signal is analyzed. The reconstruction of periodic nonuniformly sampled signal is researched then. To get the time-skew error parameters, the improved time-skew error estimation algorithm based on sine wave fitting and interpolated fast Fourier transform (IpFFT) is proposed. To reconstruct periodic nonuniformly sampled signal, two different methods, polynomial approximation and spline interpolation, are presented. It is noticed that the polynomial approximation algorithm is simplified by making use of the periodic nonuniformity. Furthermore, iterative computation can be operated to improve the polynomial approximation algorithm's reconstruction performance when the sampled signal has a high frequency.To demonstrate the algorithms researched above, a prototype 12bit 400MSPS 4-channle TIADC system constituted of 4 ADS6124s is design and realized. The improved time-skew error estimation algorithm is used to measure the system's mismatch errors. Polynomial approximation algorithm is used to reconstruct the system's sampled signals. Finally, the system's dynamic parameters, before and after reconstruction, are tested and analyzed. The test results show that the system's performance can be improved significantly by reconstruction. The system's signal-to-noise and distortion ratio (SINAD) is increased by 14~18dBc, and effective number of bits (ENOB) is increased by 2~3 bits. The average SINAD of the system is about 50dBc. The test results verify the time-skew error estimation algorithm and the reconstruction algorithm for periodic nonuniformly sampled signals researched in the thesis.