RU2740511C1 - Angle in code transducer error correction device - Google Patents

Abstract

FIELD: measuring equipment.

SUBSTANCE: invention relates to measurement equipment and automatic control equipment, in particular to angle-to-code converters, and can be used in systems where it is required to measure position with high accuracy. Angle-code converter error correction device comprises four units for calculating correction coefficients (for first, second, third and fourth harmonics), based on conversion of uncorrected angle code (with shift to required number of digits from the left) into corresponding correction coefficients based on CORDIC-converters in "turn" mode, wherein for third harmonic formula of three-angle angle is used for sinus and formula of degree reduction for square of sine. Outputs of correction coefficients calculation units are added with angle-code converter code and output of last adder is corrected angle code.

EFFECT: high accuracy of angle converters into a code, including elimination of non-linearity with odd period (3rd harmonic).

1 cl, 1 tbl, 13 dwg

Description

The invention relates to measuring and automatic control technology, in particular, to angle-to-code converters, and can be used in systems where it is required to measure position with high accuracy. The invention is expediently used in the design of specialized integrated circuits for position sensors.

A known method of digital conversion of an angle into a code with a conversion error correction [1] is based on the fact that the values of the output code are formed by summing the output code of the main channel of the amplitude digital conversion of the tracking angle and the output code of the correcting channel, while the corresponding values of the output code of the correcting channel are formed when rotation of the rotor of the angle sensor of the main conversion channel with a constant angular velocity equal to half its maximum operating value, by analog-digital conversion of the tachometric signal of the main conversion channel, integrating the variable component of the converted tachometric signal and registering the integrated signal. The disadvantage of this technical solution is the requirement for an additional channel to correct the angle code.

There is a method of digital conversion of an angle into a code with a conversion error correction [2] based on the fact that two conversion channels (two sensors) with different spectra of spatial error are used to correct the conversion error. Differences of the first and second angle codes are calculated, according to which the amplitudes and phases of the spatial harmonics of the error of the first sensor are determined, the first correction is formed as the sum of the spatial harmonics of the error of the first sensor for the angle corresponding to the first angle code, the output code is generated by adding the first correction to the first angle code. The disadvantage of this solution is that to correct the conversion error, two sensors with different error spectra are required, which, if implemented at the level of an integral position sensor with an integrated sensor system, is technically impossible.

Known converter angle-code [3], implemented in a chip for processing signals from sine-cosine position sensors. In the processing path of this microcircuit, a direct algorithm for correcting the sine-cosine signal is implemented, including the correction of the offset (including the temperature sensor), amplitude imbalance and phase shift. The corrected signal is fed to the angle-code converter. The described design is widely used to correct the error of angle-code converters, but is applicable only for converters with independent conversion of a sine-cosine signal on discrete ADCs and, moreover, cannot compensate for the 3rd and 4th harmonics (associated with nonlinear distortions), but compensate only the first and second harmonics.

There are quite a few methods for correcting signals of converters of various types based on the correction of the conversion error by multiplying the sensor output by a correcting polynomial with different degrees, the higher the degree, the more accurately the conversion error can be minimized. This method is universal and applicable for converters of various types, including angle-code converters. A similar method is described, for example, in [4], [5] and is used in various modifications in a huge number of devices. The disadvantages of this method are either its resource intensity in a pipelined hardware implementation (it requires multipliers and the greater the degree of the polynomial, the more requirements for the bit width) or low performance in software implementation using a sequential calculator (for example, based on a microcontroller).

A known method of digital correction of the conversion error for sine-cosine converters angle-code [6] includes two ADCs that convert analog sine and cosine signals into a digital code, from the outputs of the ADC signals are fed to the input of the offset correction unit, amplitude and phase imbalance, from the output of the correction unit the signals go to the detector of displacement imbalance, gain and phase imbalance, signals proportional to the corresponding imbalances are sent from the detector output to the input of the correction unit. The corrected signal after the imbalance correction is fed to the harmonic distortion correction unit (3rd harmonic of the sine and cosine signal) and then to the angle-code converter. This solution is applicable only for converters in which sine-cosine signals are converted into a digital code and only then fed to an angle-code converter, and is not applicable for a whole class of tracking angle-code converters where there is no intermediate conversion of sine and cosine signals into a digital code [ 7, 8].

The objective of the invention is to improve the accuracy of angle-to-code converters with obtaining correction coefficients using only the angle code, namely, to eliminate nonlinearities with periods multiples of 1, 2, 3, 4 relative to the signal period (first, second, third and fourth harmonics).

The problem is solved due to the fact that the angle code of the transducer is summed with a set of correction values obtained from this angle code:

- for the 1st correction value by converting CORDIC in the mode of rotation with a shift by a fixed value, the signal is taken from the SIN output of the CORDIC block;

- for the 2nd correction value by shifting the angle code to the left by one digit and converting CORDIC in the mode of rotation with a shift by a fixed value, the signal is taken from the SIN output of the CORDIC block;

- for the 3rd correction value by converting CORDIC in the rotation with a shift by a fixed value, the signal is taken from the SIN output of the CORDIC block, multiplying by 3, followed by subtracting from this value another value obtained by shifting the angle code one bit to the left and converting CORDIC in the mode of rotation with a shift by a fixed value, the signal is taken from the COS output of the CORDIC block, subtracting this value from the value corresponding to the maximum for the CORDIC block, shifting to the right by one bit, multiplying by the SIN output of the first CORDIC block, followed by a shift by 2 bits to the left , the signal is taken from the output of the reader;

- for the 4th correction value by shifting the angle code to the left by two digits and converting CORDIC in the mode of rotation with a shift by a fixed value, the signal is taken from the SIN output of the CORDIC block.

There is a causal relationship between the set of essential features of the claimed invention and the achieved technical result, namely, the digital angle code correction unit makes it possible to correct the nonlinearities of the angle code having a harmonic character with periods that are multiples of the angle code period, namely to eliminate the first, second, third and fourth harmonics.

Mathematically, the technical essence of the proposed invention is described by the system of expressions:

ϕ ₁ = AMP1⋅sin (DOFFSET1 + ANGLE [N: 0])

ϕ ₂ = AMP2⋅sin (DOFFSET2 + ANGLE [N-1: 0])

ϕ ₃ = 3⋅NMAX⋅АМР3.1⋅sin (DOFFSET3.1 + ANGLE [N: 0]) - 4⋅M ₃

ϕ ₄ = AMP4⋅sin (DOFFSET4 + ANGLE [N-2: 0])

ANGLE_CORR = ANGLE [N: 0] + ϕ ₁ + ϕ ₂ + ϕ ₃ + ϕ ₄

The technical essence of the proposed invention is illustrated by drawings, where FIG. 1 contains a block diagram of the angle code correction unit of the angle-code converter, FIG. 2 contains a block diagram of the formation of the first correction value, FIG. 3 contains a block diagram of the formation of the second correction value, FIG. 4 contains a block diagram of the formation of the third correction value, FIG. 5 contains a block diagram of the formation of the fourth correction value, FIG. 6-13 are graphs showing the operation of the device.

The structure of the error correction circuit is shown in FIG. 1.and contains:

1 - angle-code converter;

2 - block of formation of the 1st correction value;

3 - block of formation of the 2nd correction value

4 - block of formation of the 3rd correction value;

5 - block of formation of the 4th correction value;

6 - the first adder;

7 - the second adder;

8 - the third adder;

9 - fourth adder.

From the output of the converter angle-code 1, the signal is fed to the blocks for forming the correction values ϕ ₁ -ϕ ₄ 2-5, from the output of each of which the signals are fed to the inputs of the corresponding adders 6-9.

The block diagram of the block for generating the 1st correction value is shown in FIG. 2 and contains:

10 - adder;

11 - shift register of the initial value;

12 - CORDIC block in rotation mode;

13 - register of the amplitude of the output signal.

The angle code is shifted by a fixed value by adding the initial value 11 to the shift register. Then the shifted angle code is fed to the input of the CORDIC block in the "rotation" mode, the amplitude of the CORDIC block output signal is set by register 13. The SIN output of block 12 is the first correction value ϕ ₁ angle code.

The block diagram of the block for generating the 2nd correction value is shown in FIG. 3 and contains:

14 - adder;

15 - shift register of the initial value;

16 - CORDIC block in rotation mode;

17 - register of the amplitude of the output signal.

The angle code without the most significant bit is shifted by a fixed value by adding on the adder 14 with the shift register of the initial value 15. After that the shifted angle code is fed to the input of the CORDIC block in the "rotation" mode, the amplitude of the CORDIC block output signal is set by register 17. The SIN block output 16 is the second angle code correction value ϕ ₂ .

The block diagram of the block for generating the 3rd correction value is shown in FIG. 4 and contains:

18 - adder 1;

19 - register 1 shift of the initial value;

20 - block 1 CORDIC in rotation mode;

21 - register of amplitude 1 of the 3rd correction value;

22 - adder 2;

23 - register 2 shift the initial value;

24 - block 2 CORDIC in turn mode;

25 - register of amplitude 2 of the 3rd correction value;

26 - subtractor 1;

27 - register of the maximum value of the amplitude NMAX;

28 - shift register 1 bit to the right;

29 - multiplier 1;

30 - multiplier 2;

31 - shift register 2 bits to the left;

32 - compensating multiplier;

33 - subtractor 2.

The angle code is shifted by a fixed value by adding on the adder 18 with register 1 the shift of the initial value 19. After that the shifted angle code is fed to the input of the CORDIC 20 block in the "rotation" mode, the amplitude of the SIN output signal of the CORDIC block is set by register 21. The angle code without the senior bit, is shifted by a fixed value by adding on the adder 22 with the shift register the initial value 23. After that the shifted angle code is fed to the input of the CORDIC 24 block in the "rotation" mode, the amplitude of the COS output signal of the CORDIC block is set by the register 25. The COS output is fed to the subtractor 26 where it is subtracted from the maximum value register NMAX 27. From the output of the subtractor, the signal is fed to the shift register one bit to the right 28 and then to the multiplier 30 where it is multiplied with the SIN output of the CORDIC 20 block, from the output of the multiplier the signal is shifted to the register 31 by two bits to the left. The SIN output of the CORDIC 20 block is multiplied by 3 at the multiplier 29 and then goes to the input of the compensating multiplier 32 (to compensate for the amplitude difference between the two branches of the block associated with the integer representation of the data) and then to the input of the subtractor 33. The signal from the output is fed to the second input of the subtractor 33 shift register 31. The output of the subtractor 33 is the third correction value ϕ ₃ of the angle code.

The block diagram of the 4th correction value generating unit is shown in FIG. 5 and contains:

34 - adder;

35 - shift register of the initial value;

36 - CORDIC block in rotation mode;

37 - register of the amplitude of the output signal.

The angle code without the two most significant bits is shifted by a fixed value by adding on the adder 34 with the shift register of the initial value 35. After that the shifted angle code is fed to the input of the CORDIC 36 block in the "rotation" mode, the amplitude of the CORDIC block output signal is set by the register 37. Output The SIN of block 36 is the fourth angle code correction value ϕ ₄ .

The proposed design makes it possible to correct the angle error caused by the imperfection of the sensor system and the analog processing path and having a periodic nature.

FIG. 6-8 are graphs showing the operation of the error correction system to compensate for 1st order nonlinearity (first harmonic) caused, for example, by DC bias voltage in the sine and / or cosine channels. The graphs are shown for cases where a DC bias voltage is present (signal amplitude 1 V, converter resolution 13 bits or 0.05 °):

- 20 mV only for the sine channel;

- 40 mV only on the cosine channel;

- 20 mV each on sine and cosine channels.

The results of the correction system are shown in Table 1. The system compensates for the first-order nonlinearity error by more than 20 times (from 2-5 ° to 0.1 ° for a 13-bit converter).

FIG. 9, 10 are graphs showing the operation of the error correction system to compensate for 2nd order nonlinearity (second harmonic) caused, for example, by the difference in amplitudes in the sine and cosine channels. The results of the correction system for the case of differences in the amplitudes of the sine and cosine channels 0.9 and 1.1 are shown in Table 1. The system compensates for the second-order nonlinearity error by more than 20 times (from values of the order of 6 ° to a level of 0.2 ° for a converter with 13 bit resolution).

FIG. 11 shows the graphs of the operation of the correction system to compensate for the 4th order nonlinearity. Such nonlinearities are caused by nonlinear distortions in the analog path of angle-code converters. The figure shows the conversion error caused by harmonic distortion in the sine and cosine channels at the 2% level, which is ± 1.14 °. The selected correction factors can reduce the error by almost 10 times.

FIG. 12 shows the results of the operation of the correction system to compensate for the nonlinearity caused by all the factors described above (offset, amplitude difference, nonlinear distortion) and including harmonics of the first, second and fourth orders. By selecting the coefficients of the 1st, 2nd and 4th channels, it was possible to reduce the conversion error by 20 times.

FIG. 13 shows the results of the correction system to compensate for third-order nonlinearity (caused, for example, by mechanical factors) at the 5% level. By choosing the coefficients, it was possible to reduce the conversion error by 7 times.

The data presented in Table 1 show the effectiveness of the proposed correction system - the conversion error is reduced by 7-20 times or more.

The algorithm for obtaining the correction coefficients assumes an initial run of the angle-code converter (angular position sensor) for the entire revolution with the construction of an error curve by comparison with a reference sensor. Correction factors are determined from the error curve. If there are several harmonics, the Fourier transform can be used to determine the amplitude coefficients of each AMP1-AMP4 harmonic.

To increase the accuracy of calculating the correction factors, you can use CORDIC blocks with a resolution significantly exceeding that required for error correction, followed by a shift to the right by the required number of bits.

The technical and economic effect of the proposed invention is to improve the accuracy of the angular position sensors. The invention is most effective in the integrated implementation for the correction of the conversion error in the integrated circuits of angular position sensors.

Information sources

1. RF patent 2488958

2. RF patent 2235422

3. Article "Integrated position processor for precision control systems for the movement of moving units and mechanisms." Journal "Components and Technologies", No. 7, 2016, pp. 81-85.

4. Article "A Temperature Compensation Algorithm Based on Curve Fitting and Spline Interpolation". Chemical Engineering Transactions, 2016, no. 51, pp. 1345-1350.

5. Article "PDF-Based Progressive Polynomial Calibration Method for Smart Sensors Linearization". IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, 2009, Vol. 58, No. 9, pp. 3245-3252.

6. US patent 7250881

7. Utility model of the Russian Federation 167428

8. RF patent 2659468

Claims (1)

An angle code conversion error correction device comprising a first adder for adding an angle code with a first correction value, a second adder for adding an angle code with a second correction value, a third adder for adding an angle code with a third correction value, a fourth adder for adding an angle code with a fourth correcting value value, the first input of the first adder is connected to the output of the angle-code converter, the output of the first adder is connected to the first input of the second adder, the output of the second adder is connected to the first input of the third adder, the output of the third adder is connected to the first input of the fourth adder, characterized in that the code with the output of the converter is added to the first shift register of the initial value on the fifth adder, after which it is fed to the input of the first CORDIC block in rotation mode, the signal from the first amplitude register is sent to the input of the amplitude setting of the first CORDIC block, the sine output of the first CORDIC block is the first cor reciting value and enters the second input of the first adder, the code from the converter output is shifted one bit to the left and added to the second shift register of the initial value on the sixth adder, after which it enters the input of the second CORDIC block in the rotation mode, to the input of setting the amplitude of the second CORDIC block a signal is received from the second amplitude register, the sine output of the second CORDIC block is the second correcting value and is fed to the second input of the second adder, the code from the converter output is added to the third shift register of the initial value at the seventh adder, and then enters the input of the third CORDIC block in the rotation mode , the signal from the third amplitude register is sent to the input of the amplitude setting of the third CORDIC block, the signal from the sine output of the third CORDIC block is fed to the first multiplier, where it is multiplied by 3, the output of the first multiplier goes to the second multiplier where it is multiplied by the value of the register of the maximum amplitude value, the output of the second multiplier goes to the first input of the first subtractor, the code from the converter output is shifted one bit to the left, then it is added to the fourth shift register of the initial value on the eighth adder, after which it is fed to the input of the fourth CORDIC block in the rotation mode, to the input of setting the amplitude of the fourth block CORDIC, the signal from the fourth amplitude register is received, the signal from the cosine output of the fourth CORDIC block is subtracted from the register of the maximum amplitude value, then it is shifted one bit to the right and then multiplied by the sine output of the third CORDIC block, the output of the multiplier is shifted two digits to the left and then goes to the second the input of the first subtractor, the output of the subtractor is the third correcting value and is fed to the second input of the third adder, the code from the output of the converter is shifted two digits to the left and added to the fifth shift register of the initial value on the ninth adder, after which it is fed to the input of the fifth CORDIC block in In the rotation mode, the signal from the fifth amplitude register arrives at the input of the amplitude setting of the fifth CORDIC block, the sine output of the fifth CORDIC block is the fourth correcting value and goes to the second input of the fourth adder, the output of the fourth adder is the corrected code of the angle-code converter.

Images