A fast digital calibration algorithm for pipeline ADC

In this paper, a pseudo-random sequence into the digital background calibration algorithm applied in the calibration of gain error in pipeline ADC is presented. The LMS algorithm is used to establish adaptive process for gain error estimation and real-time calibration of the signal. In view of the traditional PN injection amplitude algorithms cannot satisfy redundancy calibration and the problems of slow convergence speed, the calibration scheme of reduce PN injection amplitude and using alternating sequences of plus or minus 1 instead of pseudo random sequence is proposed. Under the input signal frequency of 9.8999MHz, the sampling frequency of 100MHz, the first stage of 2.3% gain error, simulation results show that the gain speed is about 200000 clock cycles. Compared with the traditional algorithm, it is improved about 100 times. SFDR and SNR respectively by 67.2dB and 53.2dB before calibration is improved to 112dB and 85.9dB after calibrated. Effective number of bits is increased from 8.495bits before calibration to 13.978bits after calibrated. System accuracy can be greatly improved after calibration.