RU2007025C1 - Method for error correction during analog-to-digital conversion - Google Patents

Abstract

FIELD: measuring devices. SUBSTANCE: two previous values of already corrected code of input signal are stored and each subsequent signal is corrected according to equation given in invention specification. EFFECT: excluded dynamic error during correction of digitized signals that vary in time. 3 dwg

Description

 The invention relates to measuring technique and can be used in information-measuring systems and information-computer complexes (IVC) for correction of errors of analog-to-digital conversion associated with non-linearity and with changing parameters of the conversion function.

 Known iterative methods for correcting errors of an analog-to-digital converter [1], as well as the use of piecewise-linear approximation of sections of the converter characteristics in correction algorithms [2]. The disadvantage of these methods is the fact that the correction algorithms do not take into account the change in time of the input signal, which limits their scope.

 Closest to the invention is a method for correcting errors of analog-to-digital conversion [3], where the correction algorithm uses a method for solving a nonlinear problem, known as Newton’s tangent method. This method provides correction of errors in analog-to-digital conversion with high accuracy and speed, however, the correction of the input signal in time does not take into account the correction algorithm.

 The aim of the invention is to increase the accuracy of error correction for the case of analog-to-digital conversion of time-varying signals.

The difference from the prototype of the proposed correction method is that the two previous values of the corrected code of the input signal are stored, which allows you to enter into the algorithm for calculating the corrected code of the input signal a dynamic correction for the change in the input signal in time. Dynamic correction is the result of a linear extrapolation of the time dependence of the input signal for a given j-th error correction interval containing n _j cycles by using the results of n (j-1) and n (j-2) correction cycles.

 The formation of a correction algorithm with dynamic correction is shown in FIG. 1. The implementation of the attached correction method is illustrated in CPI, the block diagram of which is shown in FIG. 2.

 The CPI includes a computing complex 1 with random access memory 2 and a calculator 3, a general bus type 4 trunk, an analog-to-digital converter (ADC) 5, an input switch 6 of analog signals, and a precision digital-to-analog converter (DAC) 7.

 The ADC, DAC and switch are controlled by the CPM via the “common bus” trunk, with the help of which information is also exchanged between the calculator and the measuring transducers.

 CPI implementing this method operates as follows.

 The input signal, for example, may vary as shown in FIG. 1. Since the IVK repeatedly connects the input signal due to its change in time, to distinguish n correction cycles at different values of the input signal, the index j of measurements of these correction cycles is introduced, that is, they continue to operate with nj error correction cycles of the input signals converted using the ADC .

The proposed method uses the well-known principle of linear extrapolation of the time dependence of the input signal, which can be written as
x $\genfrac{}{}{0ex}{}{\text{ck}}{\text{nj}}$ = x _nj + {[x _{n (j-1)} x _{n (j-2)} ] / [n (j-1) T _o ]} · njT _o .

-Y_(i-1)j] /[Y

] } ·K
i = 1,2,3, . . .(1)
For each j, the correction of errors in the analog-to-digital conversion of the input signal obeys the iteration formula
x _ij = x _{(i-1) j} + {[

-Y _{(i-1) j} ] / [Y

]} · K
i = 1,2,3,. . .

= F(

) - результат j-го аналого-цифрового преобразования значения

входного сигнала;
Y_(i-1)j= F(x_(i-1)j/β);
Y

= F[(x_(i-1)j+K)/B]
- результаты аналого-цифрового преобразования первого и второго промежуточных сигналов;
К - код величины образцового сигнала;
β- коэффициент передачи ЦАП.j = 1,2,3,. . . where i is the numbering of the error correction cycles of the analog-to-digital conversion;
j is the numbering of the sequence of nj error correction cycles of the analog-to-digital conversion;
x _ij , x _{(i-1) j} - corrected codes of input signals in the i-th and (i-1) -th correction cycles for the j-th conversion of the input signal;

= F (

) is the result of the j-th analog-to-digital conversion of the value

input signal;
Y _{(i-1) j} = F (x _{(i-1) j} / β);
Y

= F [(x _{(i-1) j} + K) / B]
- the results of analog-to-digital conversion of the first and second intermediate signals;
K is the code value of the reference signal;
β- DAC transmission coefficient.

The intervals of the time sequence are determined by the duration njT _o of correction cycles, where T _o is the duration of one correction cycle.

x_nj - x_(n-1)j

≅Δ₃ , где Δ₃ - заданная величина погрешности коррекции;
x_nj - скорректированный код входного сигнала в конце цикла коррекций.Stopping the correction cycle occurs on the basis of

x _nj - x _{(n-1) j}

≅Δ ₃ , where Δ ₃ is the specified value of the correction error;
x _nj is the corrected input signal code at the end of the correction cycle.

For j = 1 and j = 2, respectively, x _n1 and x _n2 do not have a dynamic correction. Starting from j = 3, dynamic corrections Δ _g are introduced into the expressions for the adjusted input signal codes, taking into account the change in the input signal over time:
x $\genfrac{}{}{0ex}{}{\text{ck}}{\text{nj}}$ = x _(nj) + {[x _{n (j-1)} -x _{n (j-2)} ] / [n (j-1) T _o ]} · njT _o , where [x _{n (j-1)} - x _{n (j-2)} ] / [n (j-1) T _o ] characterizes the gradient of the input signal;
Δ _d = {[x _{n (j-1)} -x _{n (j-2)} ] / [n (j-1) T _o ]} · njT _o .

To calculate the dynamic correction from the calculator 3 CPI in the memory 3 receives codes x _nj , where they are stored in two intervals nj, which allows us to solve the problem of extrapolating the input signal.

 An analysis of the dependence of nj on j shows that one can take nj = n (j-1). This can be shown by the example of an ADC characteristic of the form y = x + cx (1-x); c = 0.5 (Fig. 3).

= 0,4,

= 0,6 и

= 0,8 количество циклов коррекции равно двум (на фиг. 3 не показано преобразование второго промежуточного сигнала). В силу этого формулу 1 можно использовать в более простой записи:
x_nj ^ск = x_nj + [x_n(j-1) - x_n(j-2)] . (56) 1. Алиев Т. М. и др. Автоматическая коррекция погрешностей цифровых измерительных приборов. М. : Энергия, 1975.For three examples of input signal conversion at

= 0.4,

= 0.6 and

= 0.8, the number of correction cycles is two (in Fig. 3 the conversion of the second intermediate signal is not shown). By virtue of this, formula 1 can be used in a simpler notation:
x _nj ^ck = x _nj + [x _{n (j-1)} - x _{n (j-2)} ]. (56) 1. Aliev T.M. et al. Automatic correction of errors of digital measuring instruments. M.: Energy, 1975.

 2. Copyright certificate of the USSR N 984030, cl. H 03 M 1/06, 1982.

 3. Copyright certificate of the USSR N 1714808, cl. H 03 M 1/10, 1992.

Claims (1)

METHOD FOR CORRECTION OF ERRORS OF ANALOG-DIGITAL CONVERSION, which consists in generating a code proportional to the input analog signal, with its subsequent storage, performing n correction cycles, in the first of which the first intermediate analog signal proportional to the stored code is generated, and then the second intermediate analog signal is generated by adding to the first intermediate analog signal an exemplary signal, then sequentially generate codes proportional to the first and second the intermediate analog signals, with their subsequent storage, calculate the corrected code of the input analog signal using codes proportional to the input analog signal and two intermediate analog signals, remember it and compare it with the stored code proportional to the input analog signal, if the received difference does not exceed the predetermined value, generating an output code equal to the corrected code; otherwise, the following correction cycles are performed in which the first intermediate system the signal is generated proportional to the corrected code stored in the previous correction cycle, the second intermediate signal is generated by adding an exemplary signal to the first intermediate signal, the corrected code is calculated and stored, compared with the corrected code calculated in the previous cycle, the correction cycles are stopped based on the received difference under the conditions the first cycle, characterized in that, in order to improve the accuracy of correction, the calculation of the corrected code is carried out in according to the formula
x _ij = x _{(i-1) j} +

K + [x _{n (j-1)} -x _{n (j-2)} ]
where i = 1,. . . , n is the numbering of the error correction cycles of the analog-to-digital conversion;
j is the numbering of the sequence of nj error correction cycles of the analog-to-digital conversion of the input signal, where the intervals of the time sequence are determined by the duration njT ₀ correction cycles;
T ₀ - the duration of one correction cycle;
x _ij ; x _{(i-1) j} - corrected codes of the input signals in the i-th and (i - 1) -th correction cycles for the j-th conversion of the input signal;

= F (x $\genfrac{}{}{0ex}{}{\text{\_}}{\text{j}}$ ) is the result of the j-th analog-to-digital conversion of the value x _{j of the} input signal;
Y _{(i-1) j} = F

- the result of analog-to-digital conversion of the first intermediate signal;
Y _{(i-1) j} = F

- the result of analog-to-digital conversion of the second intermediate signal;
K is the code value of the reference signal;
β is the transfer coefficient of an analog-to-digital converter;
[x _{n (j-1)} -x _{n (j-2)} ] - dynamic correction, taking into account the change in time of the input signal.

Images