CN105071807B - A kind of analog-digital converter adjustment scaling method and system - Google Patents

Abstract

The invention discloses a kind of analog-digital converter adjustment scaling method and system, method to comprise the following steps：S01, obtain sampling curve；S02, by sampling curve automatic segmentation and fitting a straight line is carried out, obtain final segmentation method；S03, the straightway obtained using segmentation are calibrated respectively.The output of analog-digital converter can be calibrated by this programme, keep accurate transformation result.This programme is applied to all analog-digital converters.

Description

Method and system for calibrating and calibrating an analog-to-digital converter

technical field

The invention relates to the field of circuit module calibration, in particular to an analog-to-digital converter calibration method and system.

Background technique

A circuit that converts analog signals into digital signals is called an analog-to-digital converter (A/D converter or ADC, Analog to Digital Converter). It is converted into a digital signal with discrete time and discrete amplitude. Therefore, A/D conversion generally goes through four processes of sampling, holding, quantization and encoding.

After the analog-to-digital converter has been used for a certain period of time, the conversion value will deviate, which will affect the output accuracy. At this time, the analog-to-digital converter needs to be calibrated.

The State Intellectual Property Office of the People's Republic of China published a patent document titled "Charge-Coupled Pipeline Analog-to-Digital Converter with Adaptive Error Calibration Function" on September 10, 2014 (publication number: CN104038225A), which includes a charge-coupled pipeline An analog-to-digital converter core, a common-mode error calibration block, a differential-mode error calibration block, an input common-mode voltage offset compensation block, a digital background calibration block, and an error calibration controller block. The invention can automatically detect the differential mode error, common mode error, input common mode voltage offset error caused by non-ideal characteristics in the fully differential structure charge-coupled pipeline analog-to-digital converter, and the temperature and voltage fluctuations caused by the normal operation of the circuit. errors and to calibrate for these errors. This scheme does not expose a suitable calibration algorithm.

Contents of the invention

The present invention mainly solves the technical problem in the prior art that lacks a method and device for reasonably calibrating the analog-to-digital converter, and provides an analog-to-digital converter calibration that can effectively calibrate the output accuracy of the analog-to-digital converter methods and systems.

The present invention mainly solves the above-mentioned technical problems through the following technical solutions: a method for calibrating and calibrating an analog-to-digital converter, comprising the following steps:

S01. Obtain a sampling curve;

S02. Automatically segment the sampling curve and perform straight line fitting to obtain the final segmentation method;

S03. Perform calibration using the straight line segments obtained by segmenting.

Preferably, obtaining the sampling curve is specifically: the analog voltage input port of the analog-to-digital converter inputs an analog voltage signal, and the analog voltage signal increases linearly from 0 to the upper limit of the input voltage of the analog-to-digital converter, and the analog-to-digital converter performs sampling , and save the sampled data.

Preferably, step S02 is specifically:

S021. The n sampling data points to be processed are ( _xi , y _i ), i=1, 2...n, set the maximum distance value d _max , the maximum residual square sum e _max and the minimum residual square sum e _min , the minimum distance d _min is 0;

S022, calculate the curvature at each sampling data point, namely

curvature

Δx _i is the small change of x _i , Δy _i is the small change of y _i , sort the n curvatures according to their size and store them in the array c[n]. In this array, only the K(i) obtained after sorting is stored. The value of i in is enough;

S023. It is assumed that the sampling curve needs to be divided into m segments, the initial value of m is 2, and each sampling curve contains more than 3 sampling data points;

S024, judge whether m is less than or equal to n/3, if yes, then enter step S025, otherwise enter step S029;

S025. Take the first m-1 numbers from the array c[n], query the sampling points corresponding to the m-1 numbers, use the m-1 sampling points as segmentation points, and divide all the sampling points into m parts , using the least squares method to fit m straight line segments;

S026, processing the transition time of adjacent straight line segments, the processing method is to set the y value of the transition time as the average number of the y values of the two straight line segments at the transition time;

S027, sequentially calculate the value of the evaluation function g when m=2, 3, ..., m _max ,

in, m _max is n/3, and takes an integer, m _min is 2, e is the sum of the squares of the difference between the points on the fitting line of the mth segment and the actual point, that is, the sum of squares of the residual, and d is the mth segment The maximum value of the distance from each actual sampling point corresponding to the fitted straight line to the fitted straight line;

S028, increase m by 1, return to step S024;

S029. Select the segmentation method corresponding to the maximum value in g as the final segmentation method.

As a preference, using the fitting straight line in this paragraph to calibrate is specifically:

A. Obtain the ideal sampling values of the starting point x _sp and the end point x _fp of this straight line from the storage unit, set as Ref1 and Ref2;

B. Take the actual sampling values of these two points and set them as Avg1 and Avg2;

C. Calculate the calibration gain coefficient CalGain and the calibration offset CalOffset, namely CalOffset=Avg1×CalGain-Ref1;

D. Calculate the calibrated sampling value newResult of each point on the line from the calibration gain coefficient CalGain and calibration offset CalOffset, that is, newResult=oldResult×CalGain-CalOffset, where oldResult represents the value directly obtained by the A/D converter before calibration sample value;

E. Store the calibration gain coefficient CalGain and calibration offset CalOffset of each straight line segment corresponding to the range of this segment y _i , so as to facilitate query and calibration.

Preferably, when the A/D converter performs conversion, query the fitted straight line segment where it is located according to the conversion value, use the calibration gain coefficient CalGain and calibration offset CalOffset of this segment to calibrate the result, and output it after calibration.

An analog-to-digital converter calibration calibration system, comprising:

obtaining a sampling curve unit for obtaining a sampling curve of the A/D converter;

The storage unit is used to store the ideal conversion value corresponding to the input analog voltage, and the calibration gain coefficient CalGain and the calibration offset CalOffset and the range of this segment _yi of each straight line segment after segment fitting;

Automatic segmentation unit, used to reasonably segment the A/D sampling curve, so that the fitting result is close to the actual sampling curve;

The calibration unit is used for calibrating each section of straight line fitted after segmentation.

The substantive effect brought by the invention is that the analog-to-digital converter can be calibrated and the accuracy of the output result can be improved.

Description of drawings

Fig. 1 is a schematic diagram of an analog-to-digital converter calibration calibration system of the present invention;

Fig. 2 is a flow chart of an analog-to-digital converter calibration method of the present invention;

In the figure: 1. A/D converter, 2. Obtaining sampling curve unit, 3. Storage unit, 4. Automatic segmentation unit, 5. Calibration unit.

Detailed ways

The technical solutions of the present invention will be further specifically described below through the embodiments and in conjunction with the accompanying drawings.

Embodiment: An analog-to-digital converter calibration system of this embodiment includes:

Obtaining a sampling curve unit 2 for obtaining a sampling curve of the A/D converter;

The storage unit 3 is used to store the ideal conversion value corresponding to the input analog voltage, and the calibration gain coefficient CalGain and the calibration offset CalOffset and the range of this segment _yi of each straight line segment after segment fitting;

Automatic segmentation unit 4 is used to reasonably segment the A/D sampling curve so that the fitting result is close to the actual sampling curve;

The calibration unit 5 is used to calibrate each segment of the straight line fitted after segmentation.

As shown in Figure 1, the A/D transfer module 1 is respectively connected to obtain the sampling curve unit, the automatic segmentation unit and the calibration unit, the storage unit is respectively connected to the acquisition sampling curve unit, the automatic segmentation unit and the calibration unit, and the automatic segmentation unit is also connected to the calibration unit. Connect to calibration unit.

The calibration method of A/D converter is:

A. Automatically fit the A/D sampling curve into segments;

B. Calibrate the straight line segments obtained by segmenting.

The specific adjustment and calibration method is shown in Figure 2.

Obtaining the sampling curve The method of obtaining the sampling curve by the unit is:

A. When the unit starts to work, the analog voltage input port of the A/D converter will input an analog voltage signal with such characteristics, that is, the voltage increases linearly from 0 to the maximum analog voltage value that the converter can input. And its time-varying speed can meet the sampling rate requirements of the converter to ensure the accuracy of the obtained sampling curve;

B. The A/D converter performs sampling, and the acquisition sampling curve unit saves the obtained data.

The acquisition sampling curve unit and the automatic segmentation unit can regularly check whether the A/D converter is idle, and if it is idle, the acquisition sampling curve operation or segmentation operation is performed; otherwise, the last segmentation result will continue to be used during calibration.

The automatic segmentation unit is segmented based on the least square method and optimization algorithm. The principle of the least squares method is: known sampling data points ( _xi , y _i ), i=1, 2...n), set the mathematical model fitted by it as y=ax+b, and use the sum of squares of the residual The minimum is to get the parameters a and b in the model. The calculation method is to take the partial derivative of the expression of Q to a and b respectively, and make it equal to zero. Solve these two equations to get the values of a and b, then you can get

The optimization algorithm is to select the optimal segmentation by combining the three indicators of the number of segments, the sum of the squares of the residuals, and the distance from the actual sampling point to the fitted straight line segment.

The steps of automatic segmentation are:

A. For n sampling data points ( _xi , y _i ) that need to be processed, i=1, 2...n, first give a maximum distance value d _max , maximum residual square sum e _max and minimum residual square sum e _min , the minimum distance d _min is 0;

B. Calculate the degree of curvature at each sampled data point, ie

curvature

Compare the size of these n curvature values, sort them, and store them in the array c[n]. In this array, it is only necessary to store the i value in K(i) obtained after sorting;

C. Assuming that the sampling curve needs to be divided into m segments, because the environmental impact sampling curve is nonlinear, so m≥2, and it does not make much sense to fit every two points, so m≤n/3, m is taken Integer, set the maximum value as m _max , when m=2, obtain the i value corresponding to the maximum curvature from the array c[n], that is, the first number in the array, and use this point as the segment point, and divide The sampling points before and after this sampling point are respectively fitted by the least squares method to obtain two straight line segments. Similarly, when m=3, take the first two numbers from the array c[n], and use these two points as segments point, use the least squares method to fit the sampling points to obtain three straight line segments until m takes the maximum value m _max ;

D. Process the transition moment between two adjacent straight lines. For example, the transition time of the straight lines L ₁ and L ₂ is x _p , after the piecewise straight line fitting, there are two values of the sampling value y corresponding to this time and , can make

E. Select the optimal segment by combining the number of segments, the sum of squared residuals, and the distance from the actual sampling point to the fitted straight line segment, and introduce the efficacy function g _i , whose value is the efficacy coefficient, specifying g _i ∈( 0, 1), when satisfied, g _i =1; when dissatisfied, g _i =0; in other cases, take the number between 0-1, because the above three objectives should be considered comprehensively, you can follow Constitute an evaluation function for optimal selection, where

e is the sum of the squares of the difference between the points on the fitted straight line and the actual point’s ordinate difference, that is, the sum of the squares of the residual error, and d is the maximum distance from each actual sampling point corresponding to the fitted straight line to the straight line. value, calculate the value of the evaluation function g when m=2, 3, ..., m _max in turn, compare and select the maximum value g _max , this situation is the optimal segmentation situation, that is, choose this segmentation method.

The method for the calibration unit to calibrate the segmented straight line segments is: to calibrate by using the linear gain error and the offset error.

The calibration unit calibrates the segmented straight line segments as follows:

A. Obtain the ideal sampling values of the starting point x _sp and the end point x _fp of this straight line from the storage unit, set as Ref1 and Ref2;

B. Take the actual sampling values of these two points and set them as Avg1 and Avg2;

C. Calculate the calibration gain coefficient CalGain and the calibration offset CalOffset, namely CalOffset=Avg1×CalGain-Ref1;

D. Calculate the calibrated sampling value of each point on this straight line from the calibration gain coefficient CalGain and the calibration offset CalOffset, that is, newResult=oldResult×CalGain-CalOffset, where _oldResult represents the value directly obtained by the A/D converter before calibration sample value;

E. Store the calibration gain coefficient CalGain and calibration offset CalOffset of each straight line segment corresponding to the range of this segment y _i , so as to facilitate query and calibration.

When the A/D converter is converting, query its fitted straight line segment according to the conversion value, use the calibration gain coefficient CalGain and calibration offset CalOffset of this segment to calibrate the result, and output it after calibration.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs can make various modifications or supplements to the described specific embodiments or adopt similar methods to replace them, but they will not deviate from the spirit of the present invention or go beyond the definition of the appended claims range.

Although terms such as sampling curve, segment, and evaluation function are frequently used in this paper, the possibility of using other terms is not excluded. These terms are used only for the purpose of describing and explaining the essence of the present invention more conveniently; interpreting them as any kind of additional limitation is against the spirit of the present invention.

Claims (4)

1. An analog-to-digital converter calibration method is characterized by comprising the following steps:

s01, acquiring a sampling curve;

s02, automatically segmenting the sampling curve and fitting a straight line to obtain a final segmentation method;

s03, respectively calibrating by utilizing the straight line segments obtained by segmentation;

the acquisition of the sampling curve specifically comprises the following steps: an analog voltage input port of the analog-to-digital converter inputs an analog voltage signal which linearly increases from 0 to the upper limit of the input voltage of the analog-to-digital converter, and the analog-to-digital converter samples and stores sampled data;

step S02 specifically includes:

s021, the n sampling data points to be processed are (x)_i，y_i) I is 1, 2 … n, and a maximum distance value d is set_maxMaximum residual sum of squares e_maxSum minimum residual sum of squares e_minMinimum distance d_minIs 0;

s022, calculating the curvature at each sampled data point, i.e.

Curvature

Δx_iIs x_iBy a small amount of change, Δ y_iIs y_iThe small variation of (c) is stored in an array c [ n ] by sorting n curvatures according to size]In the array, only the value i in the K (i) obtained after sorting is stored;

s023, setting a sampling curve to be divided into m sections, wherein the initial value of m is 2, and each section of the sampling curve contains more than 3 sampling data points;

s024, judging whether m is less than or equal to n/3, if yes, entering a step S025, otherwise, entering a step S029;

s025, taking the front m-1 sampling points from the array c [ n ], inquiring sampling points corresponding to the m-1 sampling points, taking the m-1 sampling points as segmentation points, dividing all the sampling points into m parts, and fitting by using a least square method to obtain m straight line segments;

s026, processing transition time of adjacent straight-line segments, wherein the processing mode is that the y value of the transition time is set as the average of the y values of the two straight-line segments at the transition time;

s027, m 2, 3, …, m_maxThe value of the evaluation function g of time,

<mrow> <mi>g</mi> <mo>=</mo> <mroot> <mrow> <msub> <mi>g</mi> <mn>1</mn> </msub> <msub> <mi>g</mi> <mn>2</mn> </msub> <msub> <mi>g</mi> <mn>3</mn> </msub> </mrow> <mn>3</mn> </mroot> <mo>,</mo> </mrow>

wherein,m_maxis n/3 and is an integer, m_mmThe sum of squares of difference values of the vertical coordinates of points on the mth section of fitting straight line and actual points, namely the sum of squares of residual errors, is 2, and d is the maximum value of distances from each actual sampling point corresponding to the mth section of fitting straight line to the fitting straight line;

s028, m is increased by 1, and the step returns to the step S024;

s029, selecting the segmentation method corresponding to the maximum value in g as the final segmentation method.

2. The method for calibrating an analog-to-digital converter according to claim 1, wherein the calibration using the fitted straight line of the mth segment specifically comprises:

A. obtaining the starting point x of the m-th fitting straight line from the storage unit_spAnd end point x_fpSetting the ideal sampling values of Ref1 and Ref 2;

B. taking actual sampling values of the two points, and setting the actual sampling values as Avg1 and Avg 2;

C. calculating the calibration gain factor Calgain and the calibration offset Caloffset, i.e.CalOffset＝Avg1×CalGain-Ref1；

D. Calculating a new result of a sampling value of each point on the m-th section of fitting straight line after calibration by using the calibration gain coefficient Calgain and the calibration offset CalOffset, namely calculating the new result of each point on the m-th section of fitting straight line after calibration, namely calculating the new result which is equal to old result multiplied by Calgain-CalOffset, wherein the old result represents a sampling value directly obtained by the A/D converter before calibration;

E. the calibrated gain coefficient Calgain and the calibrated offset Caloffset of each section of the fitted straight line and y thereof_iThe range of the data is correspondingly stored, so that the data is conveniently inquired and correctedThe method is accurate.

3. The method for calibrating the analog-to-digital converter according to claim 2, wherein when the analog-to-digital converter performs conversion, the fitted straight line segment where the analog-to-digital converter is located is searched according to the conversion value, and the result is calibrated by using a calibration gain coefficient CalGain and a calibration offset CalOffset corresponding to the searched straight line segment and is output after calibration.

4. An analog-to-digital converter calibration system using the method of claim 1, comprising:

the acquisition sampling curve unit is used for acquiring a sampling curve of the analog-digital converter;

a storage unit for storing an ideal conversion value corresponding to the input analog voltage, a calibration gain coefficient Calgain and a calibration offset Caloffset of each section of straight line segment after the piecewise fitting, and the section y_iA range of (d);

the automatic segmentation unit is used for reasonably segmenting the analog-digital sampling curve so as to enable the fitting result to be close to the actual sampling curve;

and the calibration unit is used for calibrating each segment of straight line which is fitted after segmentation.