Research Of Key Signal Processing Algorithms In Modern Spectral Analyzer And Its FPGA Implementation

With the rapid development of microelectronic, signal processing and computer technologies, the demand for signal measurement and analysis in frequency-domain is increasingly high, so a series of drawbacks showed in traditional analog spectral analyzers. Therefore, real-time digital spectrum analyzer with high resolution and wide bandwidth has been the key research topic. This paper has done the following research:1. The paper mainly researched digital down-conversion, which is the most key component in modern spectrum analyzers, determined its system scheme, and discussed the design of digital mixer circuit in DDC and the implementation of decimation filter in FPGA. In digital mixer module, a dual-way quadrature local signal was generated by NCO. The paper also studied implementing structure of NCO based on lookup table; we added jitter in FPGA implementation to improve spurious harmonics by 8dB. In decimator filter module, the implementing scheme of multistage decimator filtering was given in the consideration of the parallel pipeline structure of FPGA. Meanwhile, the implementing structure of cascade integrator comb filter (CIC) and polyphase filter in FPGA were discussed. A multistage CIC filter with variable decimation factors was designed, and its Attenuation can reach to -70dB. And finally, according to the character of parallel MAC structure and the switch structure of traditional polyphase filter, the implementation structure of polyphase filter was improved. It can consume few hardware resources and achieve high processing speed when the decimation times are high. The system is suitable for the filtering of frequency conversion from middle-frequency to base-band frequency.2. The paper adopted improved 2-based butterfly algorithm unit, and then implemented a 1024-point fast FFT processor with FPGA. The processor has 10-stage pipeline structure and the input port of rotating factor in every stage is fixed to improve the rate of processor. Its medium result was stored in dual-port RAM. The clock frequency can be operated in 100MHz, the time to finish a FFT circle is less than 63μs after pipelining, and the calculation error is less than 1%. All this can satisfy the demand for real-time spectrum analysis.3. This paper adopted DDC circuits and fast FFT processor to design a FPGA-based spectrum analysis system, described how to acquire, store and display in modern spectrum analyzers in detail, and implemented the design of RAM, FIFO of asynchronous, SRAM controller and VGA display controller.4. This paper tested all modules and the entire system in hardware, and the result verified the accuracy and validity of our system. The whole digital processing procedure is achieved in one FPGA, which is called SOPC system.