Adaptive calibration detection method for analog-to-digital converter chip

Abstract

An adaptive calibration detection method for an analog-to-digital converter chip comprises the following steps: acquiring an input analog signal of the analog-to-digital converter chip, and sampling the input analog signal to obtain a sampling signal; performing spectrum analysis on the sampling signal to obtain a spectrum characteristic parameter of the sampling signal; judging whether the analog-to-digital converter chip has a nonlinear error, an offset error or a gain error or not according to the frequency spectrum characteristic parameters; if an error exists, calculating a calibration parameter based on a preset calibration model in combination with the spectrum characteristic parameter; according to the calibration parameters, the conversion process of the analog-to-digital converter chip is adjusted in real time so as to compensate the errors, automatic detection and calibration of various errors of the analog-to-digital converter chip are achieved, the calibration precision and efficiency are improved, and meanwhile the performance of the high-speed channel signal integrity chip can be accurately evaluated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of chip detection and calibration, and particularly relates to an adaptive calibration detection method for an analog-to-digital converter chip and a high-speed channel signal integrity chip evaluation system. BACKGROUND

[0002] Due to inevitable process deviations, environmental factors (such as temperature, power voltage fluctuations) and other factors during the manufacturing process of ADC chips, various errors may occur in actual work, such as non-linear errors, offset errors and gain errors. These errors can seriously affect the conversion accuracy and performance of the ADC, and further affect the reliability and accuracy of the entire electronic system. For example, in a communication system, if the ADC has a large error, it may cause signal demodulation errors, reducing communication quality and data transmission rate.

[0003] In order to solve these problems, a variety of calibration methods have been proposed in the prior art. For example, some methods achieve calibration through hardware circuits, such as using additional calibration circuits to compensate for the output of the ADC, but this method often increases the area and power consumption of the chip, and increases the cost. Some other methods use software calibration algorithms, such as a lookup table (LUT) based calibration method, which measures and stores the error values of the ADC under different input signals in advance, and then looks up the corresponding error compensation value according to the input signal in actual work to calibrate. However, this method requires a large amount of storage space to store the lookup table, and for complex error conditions, the establishment and maintenance of the lookup table is difficult, and the calibration accuracy is limited.

[0004] With the rapid development of deep learning technology, it has shown great advantages in signal processing, pattern recognition and other fields. Applying deep learning technology to the field of ADC calibration provides a new way of thinking and method for solving ADC error problems. At present, although there are some ADC calibration researches based on deep learning, there are still some deficiencies in practical application, such as complex model training, high hardware requirements, and calibration real-time and accuracy need to be improved.

[0005] Therefore, it is of great practical significance to develop an adaptive calibration detection method that can effectively improve the ADC calibration accuracy, real-time and accuracy, while reducing the cost and complexity. SUMMARY

[0006] The present application aims to provide an adaptive calibration detection method for an analog-to-digital converter chip and a high-speed channel signal integrity chip evaluation system to solve the above-mentioned problems existing in the prior art, to realize the automatic detection and calibration of various errors of the analog-to-digital converter chip, to improve the calibration accuracy and efficiency, and to accurately evaluate the performance of the high-speed channel signal integrity chip.

[0007] The adaptive calibration and detection method for an analog-to-digital converter chip of the present invention primarily includes the following steps: First, the analog input signal of the analog-to-digital converter chip is acquired and sampled to obtain a sampled signal. This step is fundamental to the entire calibration and detection process, as sampling converts the continuous analog signal into a discrete digital signal for subsequent processing. For example, in an audio signal acquisition system, the analog-to-digital converter chip needs to sample the analog audio signal captured by a microphone to provide the data foundation for subsequent audio processing.

[0008] Next, spectrum analysis is performed on the sampled signal to obtain its spectral characteristic parameters. Spectral analysis reveals the frequency composition and distribution of a signal. By extracting these spectral characteristic parameters, it is possible to determine whether errors exist in the analog-to-digital converter chip. For example, using RF signal sampling in communications systems, spectrum analysis can determine whether the RF signal exhibits frequency offset, harmonic distortion, or other issues during the analog-to-digital conversion process.

[0009] Preferably, the spectral characteristic parameters are used to determine whether the analog-to-digital converter chip has nonlinear errors, offset errors, or gain errors. Different types of errors can cause the spectral characteristic parameters to exhibit different abnormalities. By setting corresponding thresholds, the error type can be accurately determined. For example, when the harmonic distortion in the spectral characteristic parameters exceeds a first preset threshold, the analog-to-digital converter chip can be determined to have a nonlinear error; when the signal-to-noise ratio is lower than a second preset threshold, an offset error can be determined; and when the spurious-free dynamic range is lower than a third preset threshold, a gain error can be determined.

[0010] If there is an error, the calibration parameters are calculated based on the preset calibration model and combined with the spectrum characteristic parameters. The preset calibration model adopted by the present invention is a neural network model based on deep learning, which is trained by a large number of input and output data of analog-to-digital converter chips containing different error types and degrees. During the training process, the input and output data are also normalized to improve the training effect and generalization ability of the model. Taking the sensor signal acquisition in the industrial control system as an example, when the analog signal output by the sensor has an error after being converted by the analog-to-digital converter chip, the neural network model based on deep learning can accurately calculate the calibration parameters according to the spectrum characteristic parameters to compensate for the error.

[0011] Preferably, the conversion process of the analog-to-digital converter chip is adjusted in real time based on calibration parameters to compensate for errors. Specifically, if a nonlinear error exists, the nonlinear compensation coefficient of the analog-to-digital converter chip is adjusted based on the calibration parameters; if an offset error exists, the offset compensation voltage is adjusted; if a gain error exists, the gain compensation factor is adjusted. For example, in an image acquisition system, when the analog-to-digital converter chip converts the analog signal output by the image sensor, nonlinear errors occur. By adjusting the nonlinear compensation coefficient, the converted digital signal can more accurately reflect the original image information.

[0012] After completing a calibration, the digital signal output by the analog-to-digital converter chip is continuously monitored, and the difference between the digital signal and the ideal output signal is calculated. This step is used to evaluate the effectiveness of the calibration. By comparing the difference between the actual output signal and the ideal output signal, it can be determined whether the calibration has achieved the expected accuracy. For example, in a digital audio playback system, continuously monitoring the difference between the audio digital signal output by the analog-to-digital converter chip and the ideal digital signal of the original audio file can be used to assess the distortion of the audio signal during the conversion process.

[0013] Preferably, the difference is used to determine whether the current calibration result meets the preset calibration accuracy requirement. If not, the system returns to reacquire the input analog signal and repeats the calibration process until the calibration accuracy requirement is met. For example, in a high-precision measurement system, the preset calibration accuracy requirement is an error of less than 0.1%. If the difference between the calculated digital signal and the ideal output signal exceeds 0.1%, the system will automatically return to recalibrate until the error meets the requirement.

[0014] Preferably, when performing spectrum analysis on the sampled signal and obtaining spectrum characteristic parameters, a fast Fourier transform (FFT) algorithm is specifically used to process the sampled signal to obtain a spectrum diagram of the sampled signal. The FFT algorithm is an efficient spectrum analysis algorithm that can quickly convert time domain signals into frequency domain signals. For example, for an analog signal containing multiple frequency components, after being sampled by an analog-to-digital converter chip, its spectrum diagram can be quickly obtained through the FFT algorithm, clearly showing the amplitude and phase information of each frequency component.

[0015] Harmonic distortion, signal-to-noise ratio, and spurious-free dynamic range (SFDR) are extracted from the spectrum as spectral characteristic parameters. Harmonic distortion reflects the harmonic content of the signal, signal-to-noise ratio indicates the ratio of signal to noise, and SFDR reflects the purity of the signal within the dynamic range. These spectral characteristic parameters can accurately reflect whether the analog-to-digital converter chip introduces errors during the conversion process, as well as the type and extent of the errors. For example, in an audio signal processing system, if the analog-to-digital converter chip exhibits nonlinear distortion, the harmonic distortion in the spectrum will be significantly increased; if noise interference is present, the signal-to-noise ratio will be reduced; and if the gain is inaccurate, the SFDR will be reduced.

[0016] Preferably, the error determination is based on the following criteria: when the harmonic distortion in the spectral characteristic parameters exceeds a first preset threshold, the analog-to-digital converter chip is judged to have a nonlinear error. This is because nonlinear errors typically cause the signal to generate additional harmonic components, increasing the harmonic distortion. For example, when the output signal of a power amplifier is converted by an analog-to-digital converter chip, if the chip has nonlinear characteristics, the harmonic distortion of the output signal may exceed the normal range, thereby being judged to have a nonlinear error.

[0017] When the signal-to-noise ratio (SNR) of the spectral characteristic parameters falls below a second preset threshold, the analog-to-digital converter chip is determined to have an offset error. This offset error can introduce a DC bias into the signal during the conversion process, reducing the effective amplitude of the signal and thus causing a decrease in the SNR. For example, in a bioelectric signal acquisition system, if the analog-to-digital converter chip has an offset error, the SNR of the collected bioelectric signal will be reduced, affecting the signal quality and subsequent analysis.

[0018] When the SFDR in the spectral characteristic parameters is less than a third preset threshold, the analog-to-digital converter chip is determined to have a gain error. Gain error can cause inaccurate amplification or reduction of the signal amplitude, affecting the signal's performance within the dynamic range and reducing the SFDR. For example, in a communication system, if the gain of the analog-to-digital converter chip is inaccurate, the SFDR of the received signal may decrease, making signal demodulation difficult and affecting communication quality.

[0019] Preferably, when determining whether the current calibration result meets the preset calibration accuracy requirements, an alarm signal is issued if multiple consecutive calibrations fail to meet the requirements. This alarm mechanism can promptly alert the user to system anomalies and the need for further inspection and maintenance. For example, in an industrial automation control system, if the calibration of an analog-to-digital converter chip fails to meet the preset calibration accuracy requirements five times in a row, the system will issue an alarm signal, notifying engineers to conduct troubleshooting to ensure normal system operation.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention uses a large amount of input and output data from analog-to-digital converter chips containing different error types and degrees to train, enabling it to learn complex error patterns and compensation strategies, thereby more accurately calculating calibration parameters and effectively compensating for nonlinear errors, offset errors, and gain errors, significantly improving the calibration accuracy of the analog-to-digital converter chip. Compared to traditional calibration methods, such as those based on lookup tables, the method of the present invention can adapt to more complex error situations and achieve higher calibration accuracy. For example, in a medical imaging device with extremely high precision requirements, the calibration method of the present invention can improve the conversion accuracy of the analog-to-digital converter chip by an order of magnitude, providing doctors with clearer and more accurate imaging diagnostic information.

[0022] The present invention utilizes a sliding window algorithm to monitor the digital signal output by the analog-to-digital converter chip. This allows real-time tracking of signal changes, prompt detection of errors, and subsequent calibration. Furthermore, the entire calibration process responds quickly, completing calibration parameter calculation and error compensation in a short period of time. This approach meets the needs of applications with high real-time requirements, such as communication systems and industrial control systems. For example, in high-speed communication systems, where real-time signal processing is crucial, the present invention's method can complete a calibration in microseconds, ensuring stable transmission of communication signals.

[0023] The calibration method of the present invention can automatically determine the calibration effect and make adjustments based on the difference between the digital signal output in real time by the analog-to-digital converter chip and the ideal output signal. If the calibration effect does not meet the preset accuracy requirements, it can automatically return to recalibrate until the requirements are met. This adaptability makes the calibration process more intelligent, capable of adapting to error variations under different working environments and input signal conditions, and improving the reliability and stability of the system. For example, in an industrial field environment, factors such as temperature and humidity may change at any time, affecting the performance of the analog-to-digital converter chip. Using the adaptive calibration method of the present invention, the chip can maintain a high conversion accuracy under different environmental conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is the system block diagram of the patent of this invention. DETAILED DESCRIPTION

[0025] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0026] A method for adaptive calibration and detection of analog-to-digital converter chips primarily includes the following steps: First, the analog input signal of the analog-to-digital converter chip is acquired and sampled to obtain a sampled signal. This step is fundamental to the entire calibration and detection process, as sampling converts the continuous analog signal into a discrete digital signal for subsequent processing. For example, in an audio signal acquisition system, the analog-to-digital converter chip needs to sample the analog audio signal captured by the microphone to provide the data foundation for subsequent audio processing.

[0027] Next, spectrum analysis is performed on the sampled signal to obtain its spectral characteristic parameters. Spectral analysis reveals the frequency composition and distribution of a signal. By extracting these spectral characteristic parameters, it is possible to determine whether errors exist in the analog-to-digital converter chip. For example, using RF signal sampling in communications systems, spectrum analysis can determine whether the RF signal exhibits frequency offset, harmonic distortion, or other issues during the analog-to-digital conversion process.

[0028] The spectral characteristic parameters are used to determine whether the analog-to-digital converter chip has nonlinear errors, offset errors, or gain errors. Different types of errors can cause the spectral characteristic parameters to exhibit different abnormalities. By setting corresponding thresholds, the error type can be accurately determined. For example, when the harmonic distortion in the spectral characteristic parameters exceeds a first preset threshold, the analog-to-digital converter chip can be determined to have a nonlinear error; when the signal-to-noise ratio is lower than a second preset threshold, an offset error can be determined; and when the spurious-free dynamic range is lower than a third preset threshold, a gain error can be determined.

[0029] If there is an error, the calibration parameters are calculated based on the preset calibration model and combined with the spectrum characteristic parameters. The preset calibration model adopted by the present invention is a neural network model based on deep learning, which is trained by a large number of input and output data of analog-to-digital converter chips containing different error types and degrees. During the training process, the input and output data are also normalized to improve the training effect and generalization ability of the model. Taking the sensor signal acquisition in the industrial control system as an example, when the analog signal output by the sensor has an error after being converted by the analog-to-digital converter chip, the neural network model based on deep learning can accurately calculate the calibration parameters according to the spectrum characteristic parameters to compensate for the error.

[0030] The ADC chip's conversion process is adjusted in real time based on calibration parameters to compensate for errors. Specifically, if nonlinear errors occur, the ADC chip's nonlinear compensation coefficient is adjusted based on the calibration parameters; if offset errors occur, the offset compensation voltage is adjusted; if gain errors occur, the gain compensation factor is adjusted. For example, in an image acquisition system, nonlinear errors may occur when the ADC chip converts the analog signal output by the image sensor. By adjusting the nonlinear compensation coefficient, the converted digital signal can more accurately reflect the original image information.

[0031] After completing a calibration, the digital signal output by the analog-to-digital converter chip is continuously monitored, and the difference between the digital signal and the ideal output signal is calculated. This step is used to evaluate the effectiveness of the calibration. By comparing the difference between the actual output signal and the ideal output signal, it can be determined whether the calibration has achieved the expected accuracy. For example, in a digital audio playback system, continuously monitoring the difference between the audio digital signal output by the analog-to-digital converter chip and the ideal digital signal of the original audio file can be used to assess the distortion of the audio signal during the conversion process.

[0032] The difference is used to determine whether the current calibration result meets the preset calibration accuracy requirements. If not, the system returns to reacquire the input analog signal and repeats the calibration process until the calibration accuracy requirements are met. For example, in a high-precision measurement system, the preset calibration accuracy requirement is an error of less than 0.1%. If the difference between the calculated digital signal and the ideal output signal exceeds 0.1%, the system will automatically return to recalibrate until the error meets the requirement.

[0033] Preferably, when performing spectrum analysis on the sampled signal and obtaining spectrum characteristic parameters, a fast Fourier transform (FFT) algorithm is specifically used to process the sampled signal to obtain a spectrum diagram of the sampled signal. The FFT algorithm is an efficient spectrum analysis algorithm that can quickly convert time domain signals into frequency domain signals. For example, for an analog signal containing multiple frequency components, after being sampled by an analog-to-digital converter chip, its spectrum diagram can be quickly obtained through the FFT algorithm, clearly showing the amplitude and phase information of each frequency component.

[0034] Harmonic distortion, signal-to-noise ratio, and spurious-free dynamic range (SFDR) are extracted from the spectrum as spectral characteristic parameters. Harmonic distortion reflects the harmonic content of the signal, signal-to-noise ratio indicates the ratio of signal to noise, and SFDR reflects the purity of the signal within the dynamic range. These spectral characteristic parameters can accurately reflect whether the analog-to-digital converter chip introduces errors during the conversion process, as well as the type and extent of the errors. For example, in an audio signal processing system, if the analog-to-digital converter chip exhibits nonlinear distortion, the harmonic distortion in the spectrum will be significantly increased; if noise interference is present, the signal-to-noise ratio will be reduced; and if the gain is inaccurate, the SFDR will be reduced.

[0035] The error determination criteria include: when the harmonic distortion in the spectral characteristic parameters exceeds a first preset threshold, the analog-to-digital converter chip is judged to have a nonlinear error. This is because nonlinear errors typically cause the signal to generate additional harmonic components, increasing the harmonic distortion. For example, when the output signal of a power amplifier is converted by an analog-to-digital converter chip, if the chip has nonlinear characteristics, the harmonic distortion of the output signal may exceed the normal range, thus being judged as a nonlinear error.

[0036] When the signal-to-noise ratio (SNR) of the spectral characteristic parameters falls below a second preset threshold, the analog-to-digital converter chip is determined to have an offset error. This offset error can introduce a DC bias into the signal during the conversion process, reducing the effective amplitude of the signal and thus causing a decrease in the SNR. For example, in a bioelectric signal acquisition system, if the analog-to-digital converter chip has an offset error, the SNR of the collected bioelectric signal will be reduced, affecting the signal quality and subsequent analysis.

[0037] When the SFDR in the spectral characteristic parameters is less than a third preset threshold, the analog-to-digital converter chip is determined to have a gain error. Gain error can cause inaccurate amplification or reduction of the signal amplitude, affecting the signal's performance within the dynamic range and reducing the SFDR. For example, in a communication system, if the gain of the analog-to-digital converter chip is inaccurate, the SFDR of the received signal may decrease, making signal demodulation difficult and affecting communication quality.

[0038] When determining whether the current calibration result meets the preset calibration accuracy requirements, an alarm signal is issued if multiple consecutive calibrations fail to meet the requirements. This alarm mechanism can promptly alert users to system anomalies and the need for further inspection and maintenance. For example, in an industrial automation control system, if the calibration of an analog-to-digital converter chip fails to meet the preset calibration accuracy requirements five times in a row, the system will issue an alarm signal, notifying engineers to conduct troubleshooting to ensure normal system operation.

[0039] Example 1:

[0040] Step 1: Signal acquisition and preprocessing: At the signal sampling front end of the communication system, the analog-to-digital converter chip receives the analog signal from the antenna. First, the input analog signal is preprocessed by filtering out high-frequency noise through a low-pass filter, and then the signal is amplified by an amplifier to a range suitable for the analog-to-digital converter chip to process. For example, an analog signal with a center frequency of [specific frequency] is filtered by a low-pass filter with a cutoff frequency of [specific frequency] and then input to a gain of

[0041] Amplification is performed by an amplifier of [specific multiple].

[0042] Step 2: Spectral Analysis and Error Determination: Sample the preprocessed signal, setting the sampling frequency to [specific frequency] to ensure accurate signal characteristics. Use the Fast Fourier Transform (FFT) algorithm to process the sampled signal, generating a spectrum. Spectral characteristic parameters such as harmonic distortion, signal-to-noise ratio (SNR), and spurious-free dynamic range (SFDR) are extracted from the spectrum. Assume the first preset threshold is [Harmonic Distortion Threshold], the second preset threshold is [SNR Threshold], and the third preset threshold is [SFDR Threshold]. If the harmonic distortion detected is [Actual Value 1] and exceeds the first preset threshold, a nonlinear error is determined. If the SNR is [Actual Value 2] and is lower than the second preset threshold, an offset error is determined. If the SFDR is [Actual Value 3] and is lower than the third preset threshold, a gain error is determined.

[0043] Step 3: Calibration Process: Because multiple errors are identified, calibration parameters are calculated based on a pre-trained deep learning-based neural network calibration model and the acquired spectral characteristic parameters. For nonlinear errors, the calculated calibration parameters correspond to the nonlinear compensation coefficient adjustment value [specific value 1]; for offset errors, the offset compensation voltage adjustment value [specific value 2]; and for gain errors, the gain compensation multiplier adjustment value [specific value 3]. Based on these calibration parameters, the corresponding parameters within the ADC chip are adjusted in real time to compensate for errors during the conversion process.

[0044] Step 4: Calibration Performance Monitoring: After completing a calibration, continuously monitor the digital signal output by the analog-to-digital converter chip. A sliding window algorithm with a width of [specified window size] is used to process the digital signal to improve real-time monitoring and accuracy. The digital signal is synchronized with the ideal output signal, and the difference between the two is calculated using a mean square error (MSE) algorithm. If the difference exceeds the preset calibration accuracy requirement (e.g., the MSE threshold is [specified value]), the input analog signal is reacquired and the calibration process is repeated. During the calibration process, detailed records are kept of each calibration time, calibration parameters, and error data before and after calibration. For example, for this calibration, the pre-calibration harmonic distortion is [pre-calibration value 1], the signal-to-noise ratio is [pre-calibration value 2], and the SFDR is [pre-calibration value 3]. After calibration, the values ​​are [post-calibration value 1], [post-calibration value 2], and [post-calibration value 3], respectively. A calibration log is generated based on this data and regularly analyzed to continuously optimize the calibration model. If five consecutive calibrations fail to meet the requirements, an alarm is issued, notifying communications system maintenance personnel for inspection. After the calibration is completed and the accuracy requirements are met, the calibration parameters are stored in the non-volatile memory of the analog-to-digital converter chip and can be directly called the next time it is used.

[0045] Example 2:

[0046] Step 1: Signal Input and Processing: In industrial automation production lines, analog signals collected by sensors are transmitted to an analog-to-digital converter chip. Similarly, the input analog signal is first filtered, using a bandpass filter to remove noise unrelated to the sensor signal's frequency range. For example, if the sensor output signal has a frequency range of [specify frequency range], a bandpass filter with a center frequency of [specify frequency] and a bandwidth of [specify bandwidth] is used for filtering. The signal is then amplified by an amplifier. The amplification factor is set based on the strength of the sensor output signal and the input requirements of the analog-to-digital converter chip, assuming it is [specify factor].

[0047] Step 2: Error Detection and Analysis: Sample the processed signal at [specific sampling frequency], use the Fast Fourier Transform algorithm to generate a spectrum of the sampled signal, and extract the spectral characteristic parameters. Errors are determined based on preset thresholds. If an offset error is detected, it may be caused by factors such as electromagnetic interference in the industrial environment. Calibration parameters for the offset error are calculated based on the calibration model, assuming that the offset compensation voltage needs to be adjusted by [specific value].

[0048] Step 3: Calibration and Subsequent Operations: Based on the calculated calibration parameters, the offset compensation voltage of the analog-to-digital converter chip is adjusted to complete the calibration. The output digital signal is then continuously monitored and processed using a sliding window algorithm with the window sliding step set to [specified step size]. The difference from the ideal output signal is calculated using a mean square error algorithm to determine the calibration effect. If the calibration meets the preset accuracy requirements, normal operation continues; if not, recalibration is performed. During the calibration process, relevant data is recorded to generate a calibration log. Analysis of the log revealed that during certain production periods, changes in equipment operating conditions led to increased sensor signal interference and frequent errors. Based on these analysis results, the calibration model was optimized, adjusting parameter weights for different interference scenarios to improve calibration effectiveness. Once calibration accuracy is achieved, the calibration parameters are stored in non-volatile memory for future use.

[0049] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive calibration and detection of analog-to-digital converter chips, characterized in that: The following steps are involved: Acquire an input analog signal of an analog-to-digital converter chip, sample the input analog signal to obtain a sampled signal; perform spectrum analysis on the sampled signal to obtain spectrum characteristic parameters of the sampled signal; and determine whether the analog-to-digital converter chip has a nonlinear error, an offset error, or a gain error based on the spectrum characteristic parameters; If there is an error, a calibration parameter is calculated based on a preset calibration model and combined with the spectrum characteristic parameters; and the conversion process of the analog-to-digital converter chip is adjusted in real time according to the calibration parameter to compensate for the error; After completing one calibration, continuously monitoring the digital signal output by the analog-to-digital converter chip, and calculating the difference between the digital signal and the ideal output signal; According to the difference, determining whether the current calibration effect meets the preset calibration accuracy requirement; If not, return to reacquire the input analog signal and perform the calibration process again until the calibration accuracy requirement is met.

2. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: The performing spectrum analysis on the sampled signal to obtain the spectrum characteristic parameters of the sampled signal specifically includes: The sampling signal is processed by a fast Fourier transform algorithm to obtain a spectrum diagram of the sampling signal; and harmonic distortion, signal-to-noise ratio, and spurious-free dynamic range are extracted from the spectrum diagram as the spectrum characteristic parameters.

3. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: The determining, based on the frequency spectrum characteristic parameters, whether the analog-to-digital converter chip has a nonlinear error, an offset error, or a gain error specifically includes: When the harmonic distortion in the spectrum characteristic parameters exceeds a first preset threshold, it is judged that the analog-to-digital converter chip has a nonlinear error; when the signal-to-noise ratio in the spectrum characteristic parameters is lower than a second preset threshold, it is judged that the analog-to-digital converter chip has an offset error; when the spurious-free dynamic range in the spectrum characteristic parameters is less than a third preset threshold, it is judged that the analog-to-digital converter chip has a gain error.

4. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: The preset calibration model is a neural network model based on deep learning, and the neural network model is trained by a large number of input and output data of analog-to-digital converter chips containing different error types and degrees; when the neural network model is trained by a large number of input and output data of analog-to-digital converter chips containing different error types and degrees, the input and output data are also normalized.

5. The method for adaptive calibration and detection of analog-to-digital converter chip according to claim 1, characterized in that: The step of adjusting the conversion process of the analog-to-digital converter chip in real time according to the calibration parameters to compensate for the error specifically includes: If there is a nonlinear error, the nonlinear compensation coefficient of the analog-to-digital converter chip is adjusted according to the calibration parameters; if there is an offset error, the offset compensation voltage of the analog-to-digital converter chip is adjusted according to the calibration parameters; if there is a gain error, the gain compensation multiple of the analog-to-digital converter chip is adjusted according to the calibration parameters.

6. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: When continuously monitoring the digital signal output by the analog-to-digital converter chip, a sliding window algorithm is used to process the digital signal to improve the real-time performance and accuracy of monitoring; the difference between the digital signal and the ideal output signal is calculated, and the digital signal and the ideal output signal are synchronously processed; and a mean square error algorithm is used to calculate the difference between the synchronized digital signal and the ideal output signal.

7. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: When executing the calibration process, the time, calibration parameters, and error data before and after each calibration are recorded; a calibration log is generated based on the recorded data, and the calibration log is regularly analyzed to optimize the preset calibration model. Before obtaining the input analog signal of the analog-to-digital converter chip, the input analog signal is preprocessed, and the preprocessing includes filtering and amplification.

8. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: When judging whether the current calibration effect meets the preset calibration accuracy requirements, if multiple consecutive calibrations do not meet the requirements, an alarm signal will be issued.

9. The method for adaptive calibration and detection of analog-to-digital converter chips according to claim 1, wherein: After the calibration is completed, the calibration parameters are stored in the non-volatile memory of the analog-to-digital converter chip so that they can be directly called the next time they are used.

Images