Weighted autocorrelation algorithm-based calibration method for time mismatch error in time-interleaved ADC (Analog to Digital Converter)

Abstract

The invention belongs to the technical field of analog-to-digital converter (ADC) digital calibration, and provides a method for calibrating a time mismatch error in a time-interleaved ADC based on weighted autocorrelation. The method specifically comprises the following steps: 1, symbol detection: calculating a weighted autocorrelation function of an output sequence of each channel, taking a channel 0 as a reference, solving an autocorrelation function difference between each channel and a reference channel, and judging the polarity (symbol) of a time mismatch error of each channel; 2, error correction: based on the polarity, adjusting the sampling time of each channel in a simulation domain by using a variable delay line (VDL) to realize time mismatch error compensation; and 3, iteratively executing the steps, wherein the final error calibration is generally completed in 10 iterations in a convergence manner. The method has the characteristics of high convergence speed (calibration can be completed by about 10000 sampling points) and high calibration precision; the method is suitable for a time-interleaved ADC architecture with any number of channels; moreover, pure digital calibration can be adopted for replacement, and the method has the advantages of being high in universality and easy to transplant.

Description

A calibration method for time mismatch error in time-interleaved ADCs based on weighted autocorrelation algorithm Technical Field

[0001] This invention belongs to the field of ADC digital calibration algorithms, and specifically relates to a calibration method for time mismatch error in a time-interleaved ADC based on a weighted autocorrelation algorithm. Background Technology

[0002] Analog-to-digital converters (ADCs) are used to quantize analog signals, converting them from the continuous-time domain to the discrete-time domain and from continuous amplitude to discrete amplitude, to obtain signals that are easy to process digitally. To improve the conversion rate, time-interleaved ADCs typically employ a structure consisting of a demultiplexer, a sample-and-hold circuit, multiple parallel sub-ADCs, and a multiplexer. The high-speed input signal is demultiplexed and distributed to multiple sub-channels. Each sub-channel is sampled and quantized in parallel at equal time intervals. Finally, the multiplexer outputs the signals alternately in time sequence, resulting in an equivalent high-speed digital sequence. The equivalent sampling rate is an integer multiple of that of a single sub-ADC.

[0003] In time-interleaved ADCs, factors such as differences in semiconductor processes, device inconsistencies, and environmental drift can lead to errors affecting the linearity of individual sub-channels. Furthermore, the entire time-interleaved system introduces non-ideal factors not present in other types of ADCs, resulting in imperfect matching between sub-channels and a decline in overall ADC performance. These sub-channel mismatch errors include offset mismatch error, gain mismatch error, and timing mismatch error. While offset and gain mismatch error calibration can typically be performed directly in the digital domain, timing mismatch error is more complex and more sensitive to high-frequency inputs. Even with high-precision multiphase clock signals, actual devices may still exhibit sampling time deviations of hundreds of femtoseconds or even larger between channels, significantly degrading the system's signal-to-noise ratio (SNDR) and spurious-free dynamic range (SFDR), with the impact further aggravated as the input frequency increases.

[0004] Existing time mismatch calibration methods based on statistics (such as autocorrelation algorithms) often require long sampling sequences (usually more than 100,000 points) to achieve satisfactory results and have a slow convergence speed. Summary of the Invention

[0005] To overcome the slow convergence problem of time mismatch calibration in existing technologies, this invention proposes a calibration method for time mismatch error in time-interleaved ADCs based on a weighted autocorrelation algorithm. This method performs calibration through an iterative process, significantly improving the convergence speed. The specific calibration method includes: performing 10 calibration iterations, each iteration consisting of two steps: sign detection and error correction. The sign detection step involves obtaining the autocorrelation results of each channel's data by calculating the weighted autocorrelation function, and calculating the difference of the autocorrelation function for each channel based on channel 0, thereby obtaining the sign of the time mismatch error for each channel. The error correction step uses a variable delay line (VDL) to adjust the time mismatch error within the analog domain. After more than 10 iterations, the time mismatch error calibration is finally completed.

[0006] Compared with existing technologies, this invention has the following advantages: It significantly improves the convergence speed of calibration while maintaining high-precision calibration, effectively enhancing the SFDR and SNDR of time-interleaved ADCs. Furthermore, this invention is applicable to time-interleaved ADCs with any number of channels, exhibiting good portability and broad application prospects. Attached Figure Description

[0007] Figure 1 is a schematic diagram of the overall architecture of the calibration method for time mismatch error in time-interleaved ADC based on weighted autocorrelation algorithm proposed in this invention;

[0008] Figure 2 is a schematic diagram of the specific weight allocation in the weighted autocorrelation algorithm proposed in this invention;

[0009] Figure 3 is a spectrum diagram of single-tone signal calibration in a specific example of the present invention;

[0010] Figure 4 shows the spectrum of a random band-limited signal before and after calibration in a specific example of the present invention.

[0011] Figure 5 shows the statistical results of calibrating a series of signals with normalized frequencies in the range of [0, 0.5] in a specific example of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] This invention proposes a calibration method for time mismatch error in a time-interleaved ADC based on a weighted autocorrelation algorithm. The specific calibration method includes: performing 10 calibration iterations, each iteration consisting of two steps: sign detection and error correction. The sign detection step involves calculating the autocorrelation results of each channel's data using a weighted autocorrelation function, and calculating the difference in the autocorrelation function for each channel based on channel 0 to obtain the sign of the time mismatch error for each channel. The error correction step involves adjusting the time mismatch error in the analog domain using a variable delay line (VDL). After more than 10 iterations, the final calibration of the time mismatch error is completed.

[0014] As an optional implementation, the specific steps of the calibration method of the present invention are as follows:

[0015] S1. For the i-th iteration (where i ranges from 0 to 9), the weighted autocorrelation function of each channel is calculated using the data from each channel and its adjacent channels. Specifically, the weight w[r] is the triangular window coefficient, and L is the sequence length, which is 1000 in this invention. The specific value of w[r] is shown in Figure 2.

[0016] S2. For the i-th iteration, use the weighted autocorrelation results to calculate the autocorrelation function difference for each channel.

[0017] S3. For the i-th iteration, use the autocorrelation function difference of each channel to obtain the sign of the time mismatch error of each channel relative to channel 0 (the first channel).

[0018] S4. For the i-th iteration, combine the sign of the time mismatch error of each channel relative to channel 0, and use the variable delay line (VDL) to fine-tune the error and adjust the time mismatch error of each channel.

[0019] S5. After 10 cycles, complete the time mismatch error calibration for all channels.

[0020] The above calibration method was used to calibrate a 4-channel time-interleaved ADC with time mismatch error, and the following results were obtained:

[0021] B1. The input signal is a single-tone signal, and the above-mentioned technology is applied for calibration. The spectrum results before and after calibration are shown in Figure 3. After calibration, the SFDR is improved by 39.52dB and the SNDR is improved by 19.96dB.

[0022] B2. The input signal is a random band-limited signal, and the above-mentioned technology is applied for calibration. The spectrum results before and after calibration are shown in Figure 4. The spurious signals caused by time mismatch are significantly suppressed after calibration, indicating that the time mismatch error has been effectively calibrated.

[0023] B3. The calibration technique described above was applied to calibrate a series of signals with normalized frequencies in the range of [0.0, 0.5]. The calibration results are shown in Figure 5. The SFDR improvement range is 19.47 dB to 39.78 dB, and the SNDR improvement range is 7.08 dB to 23.09 dB.

[0024] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A calibration method for time mismatch error in a time-interleaved ADC based on a weighted autocorrelation algorithm, characterized by: (1) Calculate the weighted autocorrelation function of the output sequence of each channel; (2) Using channel 0 as a reference, calculate the difference between the weighted autocorrelation function of each channel and the weighted autocorrelation function of the reference channel, and determine the polarity (sign) of the time mismatch error of each channel accordingly; (3) Based on the polarity, use a variable delay line (VDL) to fine-tune the sampling time of each channel in the analog domain; (4) Iterate through steps (1) to (3) until convergence.

2. The time mismatch error calibration method in a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, The weighted autocorrelation function is obtained through the following process: The original ADC sampling sequence is grouped into several overlapping sub-sequences; the autocorrelation results of each sub-sequence are calculated and then superimposed, which is equivalent to applying a triangular window weight to the original sequence autocorrelation function, resulting in the specific expression of the weighted autocorrelation function: Where m is the channel index (starting from 0), w[r] is the triangular window coefficient, when r<=L / 2, w[r]=r+1, otherwise w[r]=Lr, where L is the sequence length.

3. The calibration method for time mismatch error in a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, The weighted autocorrelation function can be calculated with only about 1,000 sampling points in each iteration to obtain the accuracy required for calibration.

4. The time mismatch error calibration method in a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, The weighted autocorrelation difference between each channel and the reference channel (channel 0, the first channel) is defined as... in The sign of the time mismatch error for each channel is determined based on the sign of the autocorrelation difference for each channel. The sign of the time mismatch error is opposite to the sign of the weighted autocorrelation difference.

5. The calibration method for time mismatch error in a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, In claim 1, the analog domain fine-tuning can be replaced by a derivative filter in the digital domain to compensate for time mismatch errors.

6. The calibration method for time mismatch error of a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, Each iteration uses approximately 1000 sampling points, and the total number of iterations is preferably 10, accumulating approximately 10000 sampling points to complete the final calibration.

7. The calibration method for time mismatch error of a time-interleaved ADC based on a weighted autocorrelation algorithm according to claim 1, characterized in that, In the hardware implementation, multipliers and adders can be reused. Compared with the traditional autocorrelation implementation, only one additional multiplier and one adder are needed to meet the computational requirements.

Images