| With the rapid development of modern electronics,time-frequency measurement and control,and other related technologies,the performance requirements of frequency sources in many practical engineering applications are becoming more and more demanding.Phase noise,as an important index to measure the performance of frequency sources,determines the market competitiveness of the frequency source to a certain extent.Therefore,the research on frequency source phase noise measurement technology has a pivotal position.Traditional phase noise measurement techniques are developed for two signals with the same nominal frequency.For phase noise measurement of signals with different frequencies,normalization is required in the circuit,which will inevitably introduce additional noise.Aiming at the above problems,this thesis designs a method for direct phase comparison in the digital linear region.By estimating the power spectrum of the extracted phase difference information,the phase noise measurement between the inter-frequency signals can be achieved without the need for normalization.The research contents of this thesis are as follows:（1）Aiming at the problem that traditional phase comparison can only occur between two signals with the same nominal frequency,this thesis explores the changing law of the phase difference between the signals and finds that for the two signals whose nominal frequency is a multiple relationship,the phase difference have the laws of monotonic and periodic change.Using these laws,a method for extracting phase difference data in the linear region of the measured signal is designed to realize the phase comparison between different frequency signals.（2）Aiming at the problem of non-uniform distribution of sampling points of nonlinear signals,this thesis analyzes the sampling process of sinusoidal signals and the law of change of tangents at sampling points.Finds that in a section of the signal zero-crossing point,the quantization blur area between adjacent sampling points is close to uniform distribution and the phase step is obvious.Combining the edge effect to extract the sampling point data can effectively offset part of the sampling error and improve the measurement accuracy.（3）Based on the research on the principle of linear phase comparison,a phase noise measurement method that is simple in structure and easy to implement is designed in this thesis.It uses ADC as a linear phase detector and uses FPGA to set the phase comparison time flexibly to extract the sampling point in linear region of the measured signal.The measurement of phase noise from far carrier frequency to near carrier frequency is realized.（4）The non-parametric spectrum estimation algorithm is simulated and analyzed in MATLAB software,and the spectrum estimation algorithm used in this system is determined.At the same time,in order to effectively improve the resolution and reduce the variance,after multiple comparison experiments,the reasonable length of window and coincidence degree are set.To solve the problem of fixed resolution in the FFT operation,the segmentation and splicing of the frequency spectrum are designed,the phase noise measurement with high-resolution and low-consumption is achieved.（5）This measurement system is used to conduct self-calibration and mutual comparison experiments on different frequency sources.In the case of self-calibration,the global frequency stability of OCXO 8607 is analyzed first,and its second-level frequency stability can be reached4.106×10-13,thus the feasibility and high precision of the method of extracting sampling point data in the linear region of the measured signal proposed in this thesis is verified.The phase fluctuation data is estimated with the power spectrum,and the measured noise floor of the system can be reached-160.3dBc/Hz@（29）10kHz.In the case of cross-comparison,the noise of the rubidium clock at the offset carrier frequency of 1MHz is-135.9dBc.Finally,the possible errors in the measurement system are analyzed. |
PDF Full Text Request