Design And Implementation Of Timing Skew Mismatch Calibration Algorithm In TIADC

With the rapid development of the electronic information technology,the application field of analog-digital converter(ADC)has been continuously expanded.Meanwhile,the requirements of each application system for ADC have been continuously improved.TIADC can improve the speed and maintain the precision of the single chip ADC,but process deviation will cause mismatch between each channel.These mismatches seriously affect the overall performance of the system.Among them,calibration method for offset mismatch and gain mismatch has been relatively mature,time skew mismatch is still research difficulties,and also the research emphasis in this thesis.Firstly,an adaptive background calibration method is adopted to correct time skew mismatch in TIADC system.In view of the problem that the convergence speed and stability performance restrict each other,the estimation process of the time skew mismatch is improved by means of variable step size method.The self-correlation function of the error factor is used to adjust the step size,it can reduce the sensitivity of the algorithm to the noise signal.The convergence speed of the algorithm is accelerated when the calibration precision is ensured.This thesis estimates the time skew mismatch using the improved adaptive method.Meanwhile,first derivative of sub-channel ADC is obtained by differential FIR filter.Then,first derivative is multiplied with the estimated mismatch and the product is subtracted from the output of sub-channel ADC to realize the real-time calibration of time skew mismatch.In addition,the error convergence of the algorithm is analyzed.The symbol function is used to represent the calibration direction and correct the error convergence.In this thesis,a four-channel,14 bit,320MSPS TIADC system is realized with four AD9255 to verify the effectiveness of the time skew mismatch calibration algorithm.The TIADC output before and after calibration is tested.The result shows that the convergence speed of the algorithm is greatly improved,and the number of the required sampling points is reduced from 200 k to 30 k.When the input frequency is 132.407 MHz,SFDR has raised 36.2231 dB and reached 79.2343 dB.Therefore,the algorithm has a good calibration effect.