TIADC Data Acquisition System Based On Channel Multiplexing Calibration Algorithm

The rapid development of phased array radars,5G wireless communication test systems,broadband digital receivers and digital oscilloscopes is pushing the system bandwidth to higher levels,significantly increasing the urgent demand for high-speed and high-resolution ADCs(Analog-to-Digital Converters)in data acquisition systems.TimeInterleaved ADC(TI-ADC)technology is an effective way to increase the sampling rate,but this will inevitably introduce channel mismatch errors,including offset mismatch,gain mismatch and sampling time mismatch,and seriously reduce the performance of the TI-ADC sampling system.Therefore,the calibration technology for channel mismatch has become the key to the practical application of the TI-ADC sampling system,and is an important technical means in the design of high-speed and high-resolution data acquisition systems.Starting from the channel mismatch problem of the TI-ADC acquisition system,this thesis conducts research and analysis on the calibration techniques of three main mismatch errors.An adaptively calibration structure featured by channel multiplexing that can simultaneously calibrate the offset,gain,and timing mismatch errors existing in TI-ADC channels in pure digital domain is proposed,which focuses on the problem of limited input signal bandwidth and high calibration hardware overhead.In error compensation aspect,these three mismatch errors are compensated simultaneously by adaptive method.The gain mismatch compensation employed here can help to overcome the deviation that would exist when the first-order Taylor compensation was employed alone for the timing mismatch,which can significantly improve the calibration effect in high-frequency band in stead of employing complex high-order Taylor compensation structure.In error estimation aspect,some innovative ways are introduced in this thesis.For the offset mismatch,the average value of all channel data is used as the reference to detect the offset errors of each channel,which help to avoid the problems such as downsampling and cross-clock domain operations that would be involved when any fixed channel data was used as the reference.For the gain mismatch,the mean squares instead of the absolute value of total channel data are used as the reference to detect the mismatch error,which can enhance the mismatch information and finally improving the accuracy of the error detection.For the timing mismatch,all the adjacent channel correlation values are exploited as the reference in the form of averaging.Compared with setting one fixed channel as the reference,there is no need for additional zero padding(zero padding would be necessary when fixed channel was used as the reference because you cannot compensate the reference)and a better calibration effect can be achieved at the same time.The calibration architecture proposed in this thesis can be adapted to both low frequency and high frequency(over Nyquist Frequency)signals with good scalability and low hardware overhead.Focuses on the failure of the calibration algorithm for special frequency point in TIADC,this thesis introduces a module that can judge the input of special frequency points,and uses the corresponding low-pass filter or band-pass filter to eliminate the spurious input of the special frequency.In this way,error correction is realized and the limitations of the calibration algorithm are overcome.This thesis uses the TI-ADC chip designed by the cooperative unit with TSMC-28 nm process,completes the all-digital channel mismatch error calibration algorithm based on Xilinx FPGA,and designs and implements a 14 bit,640MHz four-channel TIADC acquisition system.The test results show that the input signal bandwidth that can be collected by the system reaches 600 MHz,and the calibration technology can increase the SNDR of the collection system by more than 25 d B.