Biography
Enrollment Date: 2008
Graduation Date：2012
Degree：M.S.
Defense Date：2012.05.24
Advisors：Fule Li
Department：Institute of Microelectronics，Tsinghua University
Title of Dissertation/Thesis：Research on the Calibration Algorithm of Pipeline ADC Based on Code Occurrence
Abstract:
High speed high resolution Pipeline ADC is widely used in Wireless communication and image processing domain. Under the limitation of manufacturing technique, pipeline ADC usually need some kind of calibration algorithm to improve its performance. Digital background calibration technique can measure the error and correct it without interrupting the normal converting of ADC, so it has a great advantage in contrast with the foreground calibration. However, most such calibration algorithms need modification of analog circuits, which make the sensitive analogue components complicated [4] or interfere the signal [5].
In recent years, code-occurrence-based digital calibration algorithm can overcome these problems easily without modifying the analogue structure or any test signal, which can fully develop the analogue circuit’s potential. Unfortunately some back points still involve with this algorithm, such as high cost of hardware and oversensitive to the input signal [6, 8].
The code-occurrence-based digital calibration algorithm mention in this paper has intensively considered about the costly hardware and signal oversensitivity problem. The algorithm only calculates the 7 kinds of code occurrences instead of counting every code so as to save reduce the hardware cost. Besides, in order to higher the adaptability of the algorithm to input signal, we add a determining condition to the input signal. Only if the input signal satisfies the condition, the algorithm can adjust result. Otherwise, data will be discarded. So this will prevent deteriorate of AD’s performance.
In the end, we use a 13bit 50M pipeline ADC under UMC18 technique to test and verify the algorism. According to the test result, performance of the pipeline ADC has improved greatly. INL (Integral non-linearity) has dropped from 105LSB to 0.8LSB; DNL (Differential non-linearity) has dropped from 2LSB to 0.3LSB; SNDR (Signal-to-Noise and Distortion Ratio) has grown for 38dB to 66.5dB; SFDR (Spurious Free Dynamic Range) has grown from 45dB to 83dB; THD (Total Harmonic Distortion) has dropped from -35dB to -80dB. Test-results have completely verified the algorism.