CN-121984502-A - Time interleaving ADC offset correction method

Abstract

The invention relates to the field of algorithms, in particular to a time interleaving ADC (analog to digital converter) channel mismatch correction method, which comprises the following steps: and iterating the offset of each channel to obtain the offset of each channel between standard values, and returning the offset value to the data of the channel so as to finish offset correction. For an ideal single-channel non-offset ADC, the quantized average value of output data of the single-frequency sine wave is quantized to be approximately 0, and the actual offset O i of the ADC of a certain channel is calculated through an iterative formula, wherein the essence of the offset O i is that the quantized average value of the output data of the channel, O i is returned to the data of the channel, and the quantized average value of the output data of all the channels is adjusted to be 0, so that the offset of the time interleaving ADC can be completed. The mismatch correction method between the time interleaving ADC channels has little complex multiplication and division calculation, reduces the calculation complexity and has good correction effect.

Inventors

• YUAN YONGBIN
• LIAO YONG
• LIANG JIANGSHAN
• BAI JINSONG

Assignees

• 上海源斌电子科技有限公司

Dates

Publication Date

20260505

Application Date

20251231

Claims (4)

1. 1. A method for correcting mismatch between channels of a time-interleaved ADC, wherein the method comprises iterating the offset of each channel to obtain the offset of each channel from a standard value, and returning the offset value to the data of the channel, thereby completing the offset correction, the method comprising: Inputting a single-tone signal into the ADC, down-sampling the output data of each sub-channel ADC, and quantizing the data to obtain data D i (n) of each channel, wherein i represents the channel number, and n represents the data number of the channel; initializing offset O i of each channel, and setting an offset initial value of each channel to 0; iterating through the offset O i for each channel; The next iteration offset O i (n+1) of a certain channel is only dependent on the current channel offset O i (n), the current channel quantized data D i (n) and the iteration step a, O i is returned to the data of the channel, and the average value after the output data of all channels are quantized is adjusted to 0, so that the offset of the time interleaving ADC can be completed.
2. 2. The method for correcting mismatch between channels of a time-interleaved ADC according to claim 1, wherein the offset initial value of each channel is set to 0, and each channel can be corrected independently without setting a specific reference channel.
3. 3. The method of claim 1, wherein the offset O i is determined by the following equation: O i (n+1)＝O i (n)+α*[D i (n)-O i (n)]。
4. 4. The method according to claim 1, wherein the current D i (n) and the current O i (n) are used as initial values of the next iteration, and the next iteration result O i (n+1) is determined.

Description

Time interleaving ADC offset correction method Technical Field The invention belongs to the field of algorithms, and particularly relates to a time interleaving ADC offset correction method. Background The speed-precision limit of single channel ADCs has approached the process bottleneck, and the industry has generally adopted an M-channel parallel "time-interleaved" structure (TI-ADC) to boost the total sampling rate to m·fs while maintaining relatively relaxed power consumption and area of the single channel. Due to device mismatch, bias voltage drift, PVT variations, etc., there is a direct current bias (offset) difference for each sub-channel. Deregulation manifests as fixed frequency spurs (spurs) in the output spectrum, directly reducing SFDR and the number of significant bits (ENOB), becoming one of the most significant performance bottlenecks of TI-ADCs. The mismatch of the mismatch appears as a random error increment, which is believed to originate from the offset of the comparator. For single channel operating ADCs, the DC component due to misalignment is easily removed and most communication systems ignore this effect. But for interleaved ADCs, the corruption by mismatch is significant. Disclosure of Invention The invention provides a time interleaving ADC offset correction method, which can reduce the influence of offset on ADC linearity in the time interleaving ADC. The common mode value of the offset error only affects the direct current component output by the ADC, and the offset differential mode value only introduces spurious components on the frequency spectrum, thereby affecting the dynamic performance of the ADC. Correction of the mismatch of the offset is mainly focused on the correction of the relative offset of the two channels without having to calculate the absolute offset error. The zero input method can calculate the mismatch by only one group of data, and is simple and direct. However, zero input is not necessarily required to obtain a mismatch. As long as the input signal is a periodic signal with incoherent frequency with the sampling frequency of the system, the sampling point of each channel ADC is random relative to the phase of the input signal from the statistical rule, the probability that each point of the input signal is taken by two channels is the same, so that the output results of the two channels are respectively accumulated and averaged, and when the accumulated times N are large enough, the average results are accumulatedThe common mode value of each channel can be reflected. If there is no relative misalignment between the two channelsIs equal (because gain mismatch amplifies positive and negative input signals equally, positive and negative outputs are cancelled during accumulation, thusNo effect on the value of (2). The offset correction method of the time interleaving ADC provided by the invention iterates the offset of each channel to obtain the offset of each channel between standard values, and returns the offset value to the data of the channel, thereby completing offset correction. The method comprises the following steps: The input tone signal enters the ADC, down-samples the output data of each sub-channel ADC, and quantizes the data to obtain data D i (n) for each channel, where i represents the channel number and n represents the data number for that channel. Offset O i for each channel is initialized, and the initial value of offset for each channel is set to 0. Iteration is carried out on the offset O i of each channel, and a specific iteration formula of each channel is as follows: Oi(n+1)＝Oi(n)+α*[Di(n)-Oi(n)] Where a represents the iteration step. The next iteration offset O i (n+1) for a channel depends only on the current channel offset O i (n), the current channel quantized data D i (n) and the iteration step a. And returning the O i to the data of the channels, and adjusting the average value of the quantized output data of all the channels to 0, so that the offset of the time interleaving ADC can be completed. The time interleaving ADC inter-channel mismatch correction method is characterized in that the offset initial value of each channel is set to 0, a specific reference channel is not required to be set, and each channel can be corrected independently. The time interleaving ADC inter-channel mismatch correction method is characterized in that the current D i (n) and the current O i (n) are used as initial values of the next iteration, and a next iteration result O i (n+1) is determined. The iteration principle is that for an ideal single-channel non-offset ADC, the quantized average value of output data of the single-frequency sine wave is quantized to be approximately 0, and the actual offset O i of the ADC of a certain channel is calculated through an iteration formula, wherein the essence is that the quantized average value of the output data of the channel is calculated, O i is returned to the data of the channel, and the quantized