KR101829521B1 - Method for self-calibrating a rotary encoder - Google Patents

Abstract

A method for self-calibrating a rotary encoder including a single read head (111) and a circular scale (101) includes the steps of acquiring a calibration sample (150) by a readhead with respect to a rotation angle And estimating the spatial frequency (F) and spatial distortion parameters (?,?) Of the encoder from the calibration samples to self-calibrate the encoder.

Description

[0001] METHOD FOR SELF-CALIBRATING A ROTARY ENCODER [0002]

The present invention relates to a measuring apparatus, and more particularly to an absolute rotary encoder for measuring an absolute rotation angle.

Accurate estimation of the linear position and angle of rotation is an important challenge for industrial automation and similar applications. Devices such as computer numerical control (CNC) machines, drill bits, robotic arms or laser cutters, and assembly lines require accurate measurements. Feedback control is often used for precise measurements.

A typical encoder includes a scale and a read head. Optical encoders are typically used to measure an absolute or relative linear position or angle of rotation. Relative encoders need to measure the relative position or relative angle within the scale period and count the number of scale periods traversed to determine the absolute position or absolute angle. Absolute encoders do not require memory or power to remember the current position or current angle, and can also obtain the current position or current angle any time, especially at startup.

The optical encoder may be linear or rotary. The linear encoder measures the position, and the rotary encoder measures the angle. Conventional absolute rotary encoders typically employ a plurality of tracks to achieve a high resolution and apply a sine-cosine based interpolation method.

Single track absolute linear encoders utilizing a single scale and a single CCD / CMOS sensor are described in the present application. The encoder does not use the conventional sine-cosine-based interpolation. Instead, the encoder detects the edge or zero-crossing in the scan line to obtain high resolution absolute position information and aligns the model to the edge position. The encoder acquires a linear scale 1D image using a linear read head.

High precision rotary encoders are needed for precision machining and manufacturing facilities. However, some error can be introduced into the rotary encoder during manufacture. The error includes an error in a scale pattern, an installation error of a scale on a rotating shaft, a read head alignment error, and an error in noise in an electric circuit.

For rotary encoders, the spacing between scale lines varies depending on the circular nature of the scale. Another cause of the error is the eccentricity caused when the scale on the rotating disk is placed on the rotating shaft. In addition, out-of-plane motion (wobble) and poor alignment in the installation can also lead to variations in the distance between the read head and the scale. These factors affect the overall accuracy of the rotary encoder. The encoder can correct the manufacturing gap, the error in the scale pattern, the installation error of the scale on the rotating shaft, the read head alignment error, and the noise in the electric circuit. During operation, temperature fluctuations and mechanical vibrations can cause additional distortion and further reduce precision.

Because of the closer to the light source, the center of the sensor receives more light compared to its side. This causes vignetting, and the acquired 1D image has a darker center and a darker side. Vignetting leads to errors in the detected zero crossings (edges), thereby degrading the overall accuracy.

Some prior methods require a plurality of additional read heads to offset eccentric errors. For example, U.S. Pat. 6,215,119 and U.S. Pat. 7,143,518. The equally divided mean (EDA) method is described by Masuda et al. in "High accuracy calibration system for angular encoders ", J. Robotics and Mechatronics, 5 (5), 448-452, 1993. Rotary encoders using a plurality of read heads to reduce eccentricity errors increase the cost of the system and also make the system design cumbersome.

Conventional techniques also require accurate operation of the rotating part for self-calibration. For example, U.S. Pat. No. 5,138,564 discloses a method for moving an encoder at low speed and high speed for calibration. U.S.A. No. 6,598,196 drives the servo system in a predetermined orbit so that the encoder error occurs at the frequency outside the servo feedback loop. These requirements increase the calibration effort and time.

U.S.A. 7,825,367 describes a self-calibrating rotary encoder in which the angular difference is determined as a Fourier series. A sine-cosine interpolation-based rotary encoder is disclosed in U.S. Pat. 8,250, < RTI ID = 0.0 > 901. < / RTI > The voltage data corresponding to the sine and cosine of the rotation angle is fitted to the ellipse. The linear calibration parameters are obtained by transforming the ellipse into a circle.

U.S.A. No. 7,825,367 describes a self-calibrating rotary encoder. The rotary encoder includes a rotary disc with an angle code, a light source, and a linear sensor (CCD) for reading the angle code. The processing device obtains the readout value f (?) For a predetermined angularity. The difference between the readings f (? +?) And f (?) Within the read range on the linear sensor is g (?,?). The difference is determined as the Fourier series. Thus, the rotation angle [theta] at any position is obtained by analyzing the CCD image. Self-calibration is based on finding the angle of rotation at two different locations, and analyzing the differences for use in self-calibration.

Embodiments of the present invention provide an absolute rotary encoder of self-calibration, single track, single readhead. The encoder acquires measurements over one full revolution (360 °) or a portion of the full revolution. Thus, the encoder compensates for any errors or distortions introduced during and after manufacture as environmental or mechanical conditions change. The encoder can also compensate for vignetting due to illumination variations.

The embodiment does not require a plurality of read heads to solve eccentricity errors. This significantly reduces the cost and complexity of the encoder. Furthermore, the embodiments do not require the motors to be moved at various speeds or predetermined trajectories for calibration. In addition, the present invention also corrects other installation errors such as changes in gaps and shaft wobbles.

1A is a schematic diagram of a rotary encoder according to an embodiment of the present invention.
1B is a schematic diagram of a circular scale of a sector according to an embodiment of the present invention.
1C is a schematic diagram of a circular scale and linear read head according to an embodiment of the invention.
1D is a block diagram for calibrating the encoder of FIG. 1A in accordance with an embodiment of the present invention.
2 is a graph of a spatial frequency F (?) With a rotation angle according to an embodiment of the present invention.
3 is a graph of spatial frequency fluctuation caused by noise according to the embodiment of the present invention.
Fig. 4 is a graph of variation in the spatial distortion parameter? (?) With a rotation angle according to the embodiment of the present invention.
Fig. 5 is a graph of the variation in the spatial distortion parameter? (?) With the rotation angle according to the embodiment of the present invention.
6 is a graph of a quartic polynomial fit to? (?) According to an embodiment of the present invention.
Fig. 7 is a graph of the scanning line obtained by the 1D sensor and shows vignetting.
8 is a graph of a scaling factor according to an embodiment of the present invention.
9 is a graph of an offset factor according to an embodiment of the present invention.
10 is a graph of corrected sensor values after applying vignetting correction according to an embodiment of the present invention.

Embodiments of the present invention provide a single track absolute rotary encoder. The read head may be a linear charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) to obtain a 1D image of a rotating circular scale. The 1D image includes a linear pixel array. The scale includes reflective and non-reflective regions arranged in accordance with a de Bruijn sequence. The debris sequence fits well because the pattern itself is essentially circular.

Absolute circular scale

1 shows a small portion of a circular scale 100 of an absolute encoder according to an embodiment of the present invention. The details of the scale are described in US application 13/100092. The scale is used to determine the high resolution phase P (120).

The scale may include alternating reflected light 101 and anti-reflection (102) marks or bits. The mark can also alternately repeat opacity and transparency, depending on the relative position of the light source to the read head. Each mark is the width of the B micrometer, which is the scale interval. The width B of each mark is a half pitch. In one embodiment, B is 20 micrometers. Since the marks are of relatively small size, the exemplary marks shown in the figures are not to scale.

The read head 110 is set apart from the scale by a certain distance and also in parallel with the scale. The read head includes a sensor 111, a (LED) light source 112, and an optional lens. The sensor may be a detector array of N sensors, e.g., N may be 512. The array may be a complementary metal oxide semiconductor (CMOS) or a charge coupled device (CCD). The read head may also be connected to a digital signal processor 117 connected to the sensor and memory. It is understood that other types of processors may be used.

A mark or bit on the exemplary scale 100 may be disposed on the rotatable disk 130 or the shaft. The only requirement here is that the marks are placed sequentially for a particular code or aperiodic sequence.

As shown in FIGS. 1B and 1C, the mark is arranged as a sector of the circle 130 of the scale. The sensor 111 of the read head 110 includes a linear sensor array 114. Here, the read head is disposed at the center in the tangential direction with respect to the rotation center 116 and at a position offset from the rotation center 116 by the offset 115. [ It is therefore noted that sensor pixels near both ends of the linear read head observe a wider sector portion than sensors near the center of the read head. This leads to distortion of the signal on the 1D sensor.

Calibration

The DSP performs the calibration of the encoder as shown in Figure 1d. Calibration can be performed periodically or continuously offline during operation of the encoder.

During calibration, a calibration sample 150 is acquired by the sensor 111 of the readhead 110, with respect to the rotation angle of the circular scale 100, with respect to the full rotation or any partial rotation of 360 degrees. The partial rotation may be useful when the scale reciprocates in the circumferential direction instead of performing the full rotation. Note that the calibration sample can also be acquired for multiple rotations.

The frequency F, and the distortion parameters [alpha] and [beta] 161 are estimated 160 from the calibration samples. The frequency F and the distortion parameters alpha and beta can be stored directly in the memory, for example as a look-up table, and are sufficient to accurately determine the phase of the encoder during operation. Lookup table lookup may be advantageous if the search is fast, or if it requires less time and less memory than evaluating parametric functions.

For convenience, the present invention uses a parametric function 171 to model 170 the variation in the frequency F and the distortion parameters alpha and beta, and also to model the variation in memory 17 for use during on- .

Real-time operation

During real-time operation, the phase 195 of the encoder is determined 190 from the modeled variation in the test sample 151 and the frequency F and the distortion parameters alpha and beta. It should be understood that the variation may be obtained from the raw parameters stored in memory as a look-up table during operation. It is also understood that the parameters can be acquired during the real time operation of the encoder.

The details of the structure and calibration of the encoder will now be described in more detail.

De Bruijn Sequence

To achieve 100% information density on the scale, a bit sequence is used. Each subsequence is of a finite length and is unique, e.g., a debrunse sequence 103. Each unique sequence corresponds to a coarse phase angle. It is an object of the present invention to self-calibrate the encoder to obtain a fine or precise angle.

The debris sequence B (k, n) of the k -th n -th order is a predetermined sequence of alphabets (number of angles) with a size k, with respect to which all possible subsequences of length n in the alphabet are It appears exactly once as a sequence of consecutive characters. There is a separate debris sequence B (k, n) of (k! ^{K (n-1)} ) / k ⁿ where each B (k, n) has length k ⁿ . If the sequence is truncated from the front or back, then the resulting sequence also has uniqueness with the same n. It should be noted that any non-periodic sequence with non-repeating subsequences may be used.

The detector array requires at least n bits of field of view (FOV) to enable decoding. For half pitch B = 20 micrometers, using a 16th order debris sequence requires that the FOV be 16 x 20 = 320 micrometers on the scale. In one embodiment, the field of view is designed to be 1 to 2 mm to have the desired precision.

With respect to Nyquist sampling, each bit of the sequence, i.e., each half pitch of the scale, is mapped to at least two pixels in the linear detector array. This requires at least 16 x 2 = 32 pixels, which is significantly below the number of pixels in conventional sensors. In order to process optical aberrations such as defocus blur or diffraction, the number of pixels per half pitch can be increased.

Since the scale is circular, the reflective and non-reflective areas are equi-angular and not equi-distant when using linear sensors (see FIG. 1C). Due to the circular scale, the width of the reflection / non-reflection area increases at both ends of the sensor. Therefore, the spatial frequency F is not constant along the sensor.

z (i) is the detected zero crossing (edge position), P is the phase angle, and F is the frequency. Let k (i) be the number of bits between two consecutive zero crossings z (i) and z (i + 1). In the present invention,

, The i-th zero crossing of the rotary encoder is defined as a three-dimensional model of c (i)

Where the parameters of the three-dimensional model include phase P, spatial frequency F, and spatial distortion parameters alpha and beta. This model is generated in consideration of the error due to the uneven spacing of the scale lines on the circular disk 130. Using N zero crossings, N equations are obtained. For example, the twenty zero crosses z (1), ... , z (20) are present, the corresponding c (1), ... , c (20) can be known. These equations represent linear systems in the unknowns P, F, alpha and beta. In the present invention, the linear system is solved to obtain the values of P, F, alpha and beta.

The rotation angle? Is obtained by using? = P / F x 360 / K + Coarse_Position, where K is the number of gradations in the scale and Coarse_Position is a phase angle based only on the basic code subsequence of the picture. For example, K may be 1024.

Self-correction

The estimated parameters F,? And? (161) are expressed from the viewpoint of the actual rotation angle? As F (?),? (?) And? The embodiment examines variations in three parameters F (?),? (?), And? (?). Due to imaging noise, there is a small variation (normal variation) in these parameters.

In a state where there is no mechanical error and no eccentricity error in installation, the variation in these parameters varies from 0 to 360 degrees, so that the normal fluctuation in one full revolution or part of revolution will be. However, due to the eccentricity, the wobble, or the gap change (the distance between the read head and the scale), the spatial parameters F (?),? (?), .

FIG. 2 shows an example 200 of an estimate F (?) With a rotation angle over one full revolution.

3 shows frequency variation 300 in more detail. The high frequency variation 301 is due to noise. The low frequency variation 302 is due to the eccentric wobble and gap variation. The object of the present invention is to correct these variations.

During the self-calibration process, these variations are modeled using parametric functions. The shaft with the scale is rotated in full rotation (360 °) or partial rotation (<360 °), and the encoder scale is sampled at several positions. For example, the scale may be rotated every 2 DEG, and a sensor image corresponding to the angle is stored in the memory. For all these angles, an estimate of the frequency and distortion parameters is stored along with the estimated encoder phase P.

Curve Fitting

Appropriate parametric functions or splines are used to model variations in frequency and distortion parameters using least square fitting.

4 shows the variation in? (?) 400 that can be modeled using the fourth order polynomial model for the rotation angle?.

T ₁ , t ₂ , t ₃ , t ₄ and t ₅ are model parameters. The model parameters are estimated using the least squares fitting of the estimate [alpha] ([theta]). FIG. 5 shows a variation 500 with respect to? (?). Note that the model order or format need not be the same for all three parameters. For example, the frequency F ([theta]) can be modeled using a spline basis function, and [alpha] ([theta]) and [beta] ([theta]) can be modeled using a polynomial function. 6 shows an estimate? (?) Fitted to the quartic polynomial 600 concerning the full rotation. After curve fitting, the model parameters are stored in the memory of the DSP 115.

action

During the encoder operation, the last encoder position can be used to determine the value of the frequency and spatial distortion parameters with respect to the current position. These values are used to determine the phase P. Alternatively, the frequency and distortion parameters may be determined iteratively with the phase. This is useful at startup, and the last encoder phase is unknown or invalid. A first estimate of the current rotation angle, frequency and distortion parameters may be obtained as described above. Using the estimated rotation angle [theta], the parameters F, alpha and beta relating to the current position are recalculated using their respective models. A new value of these parameters is then used to recalculate phase P.

U.S. granted to Nakamura. 7,825,367, the autocorrelation is based on the rotation angle at two different positions, and the difference in the self-calibration is analyzed and used for calibration. Nakamura does not describe spatial frequency and distortion parameters. The encoder according to the present invention is based on the fundamental frequency and distortion parameters used for modeling the zero crossing at a specific rotation angle, not based on the actual rotation angle as in Nakamura.

Vignetting correction

As shown in FIG. 7, the vignetting correction can be performed by acquiring the measurement 700 during the rotation of the scale.

8, the maximum pixel value m ₁ (p) is a scaling factor 800, and as shown in FIG. 9, the minimum pixel value m ₂ (p) is The offset factor is 900. These are used for vignetting correction as follows.

As shown in Fig. 10 for all positions, the sensor value i (p) is modified as follows (1000).

This correction ensures that the minimum intensity of each pixel is set to zero and the maximal intensity of each pixel is set to 255 when the encoder is rotated. This eliminates the effect of vignetting.

The steps of the method for performing self-calibration and vignetting correction may be performed in a DSP or similar microprocessor connected to a memory and input / output interface as is known in the art.

Claims (20)

CLAIMS 1. A method for self-calibrating a rotary encoder including a single read head and a circular scale,
Acquiring a calibration sample for the rotation angle of the circular scale by the read head;
Estimating a spatial frequency and spatial distortion parameter of the rotary encoder from the calibration sample to self-calibrate the rotary encoder
/ RTI &gt;
Wherein the spatial frequency and the spatial distortion parameter are modeled taking into account errors due to non-uniform spacing of the scale lines on the circular scale,
Wherein the spatial frequency and the spatial distortion parameter are estimated using zero-crossing and the number of bits between the zero crossings,
Wherein the zero crossing indicates an edge position in a scanning line of the read head
Way.
delete The method according to claim 1,
Measuring the number of bits between two successive zero crossings on the circular scale;
When the zero crossing is represented by c (i)

,

A step of modeling
Further comprising:
(I) is a zero crossing, k (i) is the number of bits between the two consecutive zero crossings z (i) and z (i + 1), i is an integer parameter, P is a phase Is a phase value, F is the spatial frequency, and? And? Are the spatial distortion parameters
Way.
The method according to claim 1,
Obtaining a test sample of the circular scale;
Determining a phase angle of the rotary encoder using the spatial frequency and the spatial distortion parameter;
&Lt; / RTI &gt;
The method according to claim 1,
Wherein the rotary encoder is an absolute rotary encoder having an aperiodic sequence.
The method according to claim 1,
Wherein the circular scale is in a de Bruijn sequence format.
The method according to claim 1,
Wherein the mark on the circular scale is arranged as a sector,
Wherein the read head is offset from the center of rotation of the circular scale and disposed at the center of the tangential direction
Way.
The method according to claim 1,
Modeling a variation in the spatial frequency and the spatial distortion parameter using a parametric function;
Obtaining a test sample of the circular scale;
Determining a phase angle of the rotary encoder using the modeled spatial frequency and the spatial distortion parameter
Further comprising:
Wherein the spatial frequency and the spatial distortion parameter are modeled as a function of the actual rotation angle
Way.
The method according to claim 1,
Wherein data of the read head is acquired for a rotation angle of 360 degrees or less.
8. The method of claim 7,
Wherein the spatial frequency is F (?), The spatial distortion parameters are? (?) And? (?), The fourth order polynomial is

T ₁ , t ₂ , t ₃ , t ₄ and t ₅ are the parameters of the quadratic polynomial estimated using least square fitting
Way.
The method according to claim 1,
And storing the spatial frequency and the spatial distortion parameter in a memory as a look-up table.
The method according to claim 1,
Wherein the spatial frequency and the spatial distortion parameter are acquired during a real time operation of the rotary encoder.
The method according to claim 1,
The read head including a linear pixel array,
Further comprising the step of measuring the intensity of the pixel value of the pixel in the linear pixel array so as to acquire the maximum intensity as a scaling factor and obtain the minimum intensity as an offset factor,
The intensity of the pixel value is modified using the scaling factor and the offset factor
Way.
14. The method of claim 13,
The intensity i (p) of the pixel value is

Lt; / RTI &gt;
Wherein m ₁ (p) is the maximum intensity and m ₂ (p) is the minimum intensity
Way. delete delete delete delete delete delete

Images