Improvement And Implementation Of ADC Test Algorithm Based On Maximum Likelihood Estimation

In recent years, ADC (analog-to-digital converter) has become an indispensable part of modern advanced electronic equipment or electronic system. With the increasing of the ADCs' precision, the existing standard test method is becoming more and more inefficient. The ADCs' test algorithm based on the maximum likelihood estimation is an expansion of sine-wave fitting algorithm. It can achieve a higher test accuracy and get more ADC parameters. But the test time required for the algorithm is too long, which limits the application of the algorithm in the actual test.This thesis is focused on the research of the ADCs' test algorithm based on the maximum likelihood estimation to reduce the testing time and enhance the practicability of the algorithm. The algorithm consists of two main parts:the estimation of the initial value and the solution of the likelihood equation. In this thesis, the requirements for the estimating method of the initial value are low computing cost, un-repeated sampling, and sufficient accuracy. Firstly, according to this requirement, a three spectral line interpolation DFT algorithm is selected to estimate the frequency of the input sine excitation. This method has a good adaptability to the noise environment and the non-coherent sampling condition. In the simulation environment, the relative error of the estimation accuracy is 10-6.Then, the other initial value is obtained by the three-parameter sine-wave fitting algorithm, the fitting process is fast and absolutely convergent. The initial range of the transfer levels are limited by the range of each code width. Finally, the particle swarm optimization algorithm (PSO) is used to solve the likelihood equation. The standard particle swarm optimization algorithm has been improved, so that it can achieve a better convergence accuracy in the case of high dimension and multi peak value. MATLAB is also used as to realize the algorithm, and the parallel computing is used to improve computing efficiency.Based on the improved algorithm, this thesis carries out a series of simulation by MATLAB. To verify the validity of the algorithm, an 8bit ADC hardware circuit boards based on MAX 1195 is designed and a test system is set up. The test result shows, the number of iterations is about 1500?1700 times in the improved algorithm of this thesis, compared with the 3000?3500 times iteration of the traditional maximum likelihood estimation algorithm, the computing efficiency increases more than one fold; The estimated value of SINAD is 46.79dB and ENOB is 7.48, the accuracy is equivalent to the result obtained by the traditional maximum likelihood estimation algorithm.