JP2016014688A - Rotary encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a rotary encoder that is advantageous in view of measurement accuracy.SOLUTION: A rotary encoder comprises: a cylindrical element including a cylindrical surface on which a plurality of marks are cyclically arrayed along the circumferential direction and rotating around the axis of the cylindrical surface as an axis of rotation; a detector including a light source for emitting light toward the cylindrical surface and a plurality of photoelectric conversion elements, the detector detecting a prescribed number of marks among the plurality of marks with the light emitted from the light source by using the plurality of photoelectric conversion elements; and a signal processing unit for obtaining the rotation angle of the cylindrical element on the basis of the output of the detector. The detector outputs a cyclic signal having a plurality of unequal cycles and corresponding to the prescribed number of marks. The signal processing unit obtains the rotation angle on the basis of the cyclic signal and data having a plurality of unequal cycles obtained on the basis of the geometric arrangement of the light source, cylindrical surface, and detector and a reference phase.

Description

  The present invention relates to a rotary encoder.

  Conventionally, an incremental rotary encoder or an absolute rotary encoder is used for the purpose of measuring an angle. Incremental encoders record slits with a constant period in a radial pattern on a disk, read the relative rotational movement of the slits optically or magnetically, and calculate an absolute angle in combination with an origin detection mechanism. In recent years, there are many high-resolution products in which the slit pitch is highly refined to about 80 microns, and the phase within one pitch is interpolated by about 10,000 divisions with an electric divider. Normally, the slits are not read one by one, but a plurality of slits are optically averaged, so that dust, scratches, defects, and variations in the pattern of the slits themselves are averaged, resulting in high accuracy. However, since absolute angle information cannot be obtained unless the origin is detected first, the application to the field of machine tools and robots has been particularly limited.

  Various types of absolute encoders have been devised as binary patterns, and there are a so-called gray code method for recording binary patterns on a plurality of tracks and a method for recording random numbers on one track. However, in the Gray code method, synchronization of detection of information on different tracks tends to be difficult due to mounting errors, and the resolution is not so high. Therefore, there is a method in which binary codes 1 and 0 are expressed by the size of transmittance, the width of slits, the presence or absence of slits or pits arranged at equal intervals, and the like as a cyclic code pattern and recorded on one track. Proposed. A cyclic code pattern refers to a pattern in which 1 and 0 are randomly arranged in one circle, and when there are M patterns adjacent to each other, there is absolutely no part of the same arrangement in the entire circle. Patent Document 1 discloses an absolute encoder that uses a kind of M-sequence code of a cyclic code. The absolute encoder described in Patent Document 1 is a method of modulating the ratio of the transmissive portion and the non-transmissive portion of the absolute code pattern, and the position of the code switching portion (edge portion) is set in the light receiving element array of high-definition pixels. It tries to improve the resolution by inserting.

  Patent Document 2 discloses an absolute encoder that uses an absolute code in which reflection slits periodically arranged at equal intervals are partially removed. The absolute encoder described in Patent Document 2 measures the absolute position by pattern matching (correlation calculation) with actual image information obtained by imaging a scale on which an absolute code is recorded with a light receiving element array and pre-calculated reference table data.

  However, when the one-track type absolute encoder as described above is applied to a rotary encoder having a cylindrical surface, a specific problem occurs. The first problem is that the distance d between the cylindrical surface and the light projecting / receiving unit (head unit) changes when the light beam is illuminated on the mark formed on the cylindrical surface and the light reflected by the mark is detected by the light receiving element array. As a result, the projection magnification varies. Regarding this first problem, Patent Document 3 discloses that a cylindrical surface and a head portion are attached at an appropriate interval d by installing a sensor for detecting the size of the interval.

The second problem is that periodic distortion occurs. The periodic distortion is caused when the light beam illuminating the peripheral portion of the cylindrical surface is projected onto the linear light receiving element array as compared with the light beam illuminating the front surface of the cylindrical surface, particularly when the radius R of the cylindrical surface is small. This is a phenomenon that enlargement magnification becomes large. When periodic distortion exists, the period of the light and dark pattern on the light receiving element array extends toward the periphery. In Patent Document 3,
There is no suggestion regarding periodic distortion.

  In the case of a one-track type absolute rotary encoder, a code error or phase detection (interpolation) error occurs if the intensity and phase information of the projection pattern are not detected accurately at the center and the periphery of the light receiving element array. Therefore, conventionally, only products in which a scale is formed on a cylindrical surface of a relatively large R (about 50 mm) that is regarded as a flat surface have been put into practical use.

JP 60-89713 A JP-T-2004-529344 Special table 2008-539409 gazette

  Therefore, in the case of an absolute rotary encoder with a scale on the cylindrical surface, the change in mark projection magnification due to the change in the distance between the cylindrical surface and the head and cancellation of the influence of periodic distortion can be used to reduce the size of the cylindrical surface. It is required to be possible. In addition, a so-called modular encoder with a separate cylindrical surface and head part has a tolerance for mounting errors and also has a function of outputting information that the mounting state is appropriate for practical use. is important. Therefore, an absolute rotary encoder incorporating such a function is desired.

  An object of the present invention is to provide a rotary encoder that is advantageous in terms of measurement accuracy.

  The present invention is a rotary encoder, comprising a cylindrical surface in which a plurality of marks are arranged with a period along the circumferential direction, a cylindrical body that rotates about the axis of the cylindrical surface as a rotation axis, and toward the cylindrical surface A light source that emits light; a plurality of photoelectric conversion elements; a detector that detects a predetermined number of marks among the plurality of marks by the light emitted from the light source by the plurality of photoelectric conversion elements; A signal processing unit that obtains a rotation angle of the cylindrical body based on an output of the detector, and the detector outputs a periodic signal corresponding to the predetermined number of marks having a plurality of non-uniform periods. The signal processing unit outputs the periodic signal and data having a plurality of non-uniform periods and a reference phase obtained based on a geometric arrangement of the light source, the cylindrical surface, and the detector. Get the rotation angle based on It is characterized in.

  According to the present invention, for example, a rotary encoder that is advantageous in terms of measurement accuracy can be provided.

It is the top view and side view of an absolute rotary encoder. It is a figure explaining the flow of signal processing of an absolute rotary encoder. It is a graph which shows the relationship between the radius of a cylindrical surface, and the projection phase on a light receiving element array. It is a figure explaining determination of the integer part of an absolute code.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a diagram illustrating a configuration of an absolute rotary encoder having a cylindrical body in which a plurality of marks are formed on a cylindrical surface, and FIG. 2 is a diagram illustrating a flow of signal processing. As shown in FIG. 1, a head portion HEAD is disposed on a cylindrical surface having a radius R with a distance d. A light source (point light source LED) that emits light toward the cylindrical surface and a detector (light receiving element array PDA) that detects the light reflected by the marks arranged on the cylindrical surface are installed in the head section Head. Yes.

  The cylindrical body rotates relative to the head part Head with the central axis (rotating axis) O of the cylindrical surface as the center of rotation. Non-reflective slits are arranged radially at equal intervals on the cylindrical surface, and reflective slits or semi-reflective slits are arranged in the circumferential direction at regular intervals between the non-reflective slits. The reflection slit and the semi-reflection slit constitute a plurality of marks detected by the light receiving element array PDA. Of course, the reflective slits are arranged radially at equal intervals on the cylindrical surface, and non-reflective slits or semi-reflective slits are arranged between the reflective slits in the circumferential direction with a constant period and gap. A plurality of marks can also be configured. The semi-reflective slit can be realized by attaching a thin film to the reflective slit, reducing the size of the reflective slit, partially blocking with a hatching pattern, etc., but any method can be used as long as the amount of reflected light can be reduced. In FIG. 1, the reflective slit and the semi-reflective slit constituting the mark are schematically shown by thick lines. Although details are not described in FIG. 1, the amount of reflected light of the slit corresponds to the amplitude of the periodic signal of the light receiving element array PDA.

  In this embodiment, two types (plural types) of slits, ie, reflective slits and semi-reflective slits, constitute two types (plural types) of marks, and a plurality of two types of marks are arranged to form an M-bit absolute code. Yes. In this embodiment, the two types of marks have the same shape and different reflectances, and each of the two types of marks has a uniform reflectance within the mark. The light receiving element array PDA includes a plurality of photoelectric conversion elements arranged at a pitch smaller than the mark period. The plurality of photoelectric conversion elements detect a predetermined number of marks (reflection slits and semi-reflection slits) among the plurality of marks formed on the cylindrical surface. The brightness distribution of incident light to the light receiving element array PDA is as shown in GRPH0 in FIG. A plurality of mark signals corresponding to light and dark are output as electrical signals from the photoelectric conversion elements of the light receiving element array PDA. In the present embodiment, a plurality of electrical signals output from the light receiving element array PDA are once stored in a register and serially transferred using the externally applied clock signal as a trigger.

  The serially transferred signal waveform GRPH1 is the same as GRPH0 indicating the light amount distribution incident on the light receiving element array PDA. The signal waveform GRPH1 is shown as the amplitude of a sinusoidal waveform having a narrow center portion and a wide peripheral portion modulated by an absolute code. However, the actual waveform has a triangular wave or a trapezoidal shape due to a change in the interval between the cylindrical surface and the light receiving element array PDA, and the sine wave is distorted. The output waveform from the light receiving element array PDA is handled as “waveform data” corresponding to the number of effective channels of the light receiving element array PDA, and the “waveform data” is successively generated by the subsequent calculation unit (signal processing unit) CULC. Suppose that it is converted.

  Signal processing in the calculation unit CULC for calculating the absolute rotation angle of the cylindrical body will be described with reference to FIG. The calculation unit CULC calculates the absolute rotation angle of the cylindrical body at the position of the light receiving element array PDA based on the serial transfer waveform GRPH1 output from the light receiving element array PDA. Reflected light from the slit on the cylindrical surface is projected on the light receiving element array PDA as a bright and dark pattern. However, due to the geometric arrangement of the light source LED, the cylindrical surface, and the light receiving element array PDA, a predetermined number of mark areas on the cylindrical surface are distorted and projected onto the light receiving element array PDA. As a result, the interval between the light and dark patterns increases from the center to the periphery of the light receiving element array PDA. In addition, the light and dark phases output from the photoelectric conversion elements of the light receiving element array PDA shift. That is, the signal of the predetermined mark detected by the light receiving element array PDA has a non-uniform cycle.

  The phase shift (correction data) indicating the non-uniformity of the pattern projected on the light receiving element array PDA is calculated in advance using parameters relating to the geometric arrangement of the light source LED, the cylindrical surface, and the light receiving element array PDA. be able to. Then, the calculation unit CULC corrects the output from each photoelectric conversion element so as to reduce the non-uniformity of the period using the phase shift data. Parameters regarding the geometrical arrangement are: radius R of the cylindrical surface, distance d between the cylindrical surface and the point light source and the light receiving element array, photoelectric conversion element number N (expressed as ± with 0 as the center), light receiving element array Pitch PT, and the number of slits / circumference U formed on the cylindrical surface. Now, the reference phase φ of the light / dark pattern incident on the Nth photoelectric conversion element is expressed by a function φ (N). When the slit formed on the cylindrical surface is positioned in front when viewed from the head part Head, the reference phase of the light / dark pattern of the slit is defined as zero. That is, at phase zero, the photoelectric conversion element number No. The amount of reflected light incident on 0 is maximized.

  For example, as shown in FIG. 3A, PT = 0.0125 mm, R = 6.25 mm, d = 3 mm, and U = 512. Then, the pitch on the circumference of the slit in the cylindrical surface is about 0.077 mm. At this time, the reference phase φ of the pattern corresponding to the photoelectric conversion element number is as shown in FIG. The horizontal axis represents the photoelectric conversion element number, and the vertical axis represents the phase φ. The graph is bent rather than straight. The number of slits on the cylindrical surface in this calculation example is U = 512 = 2 ＝ 9 and 9 bits, so it is only necessary to correctly receive the light / dark pattern in the region of 9 cycles. However, for the convenience of calculating the phase, information on the front and rear light / dark patterns is also necessary, so that data of a region for ± 1 period is further used. Then, the reference phase value in the region of φ = ± 5.5 on the vertical axis in FIG. 3A is held in the register RGST (SIN +) in FIG. The register RGST (SIN−) holds a value obtained by adding a half period, that is, +0.5 to this reference phase. The register RGST (cos +) holds a value obtained by adding ¼ period, that is, +0.25 to this reference phase. The register RGST (cos−) holds a value obtained by adding 3/4 period, that is, +0.75 to this reference phase.

  In another example, as shown in FIG. 3B, PT = 0.0125 mm, R = 25 mm, d = 3 mm, and U = 2048. Then, the slit pitch on the cylindrical surface is about 0.077 mm as before. At this time, the reference phase φ of the pattern corresponding to the photoelectric conversion element number is as shown in FIG. The horizontal axis represents the photoelectric conversion element number, and the vertical axis represents the phase φ. The graph is close to a straight line but slightly bent. The inclination is different from that shown in FIG. The number of slits on the cylindrical surface in this calculation example is U = 2048 = 2 ^ 11 and 11 bits, so it is sufficient to receive light and dark patterns for 11 periods correctly, but information on the front and back is also necessary for the convenience of calculating the phase. Therefore, it is assumed that data of a region for ± 1 period is further used. Then, the reference phase value in the region of φ = ± 6.5 on the vertical axis in FIG. 3B is held in the register RGST (SIN +) in FIG. A value obtained by adding +0.5 to the reference phase is held in the register RGST (SIN−), and a value obtained by adding +0.25 to the reference phase is held in the register RGST (cos +). -) Holds a value obtained by adding +0.75 to the reference phase.

  The arithmetic unit AVE calculates the sum signal of the photoelectric conversion elements in the region where the relative phase of each photoelectric conversion element and its surroundings spans one cycle (-π to + π), converts it to a waveform like GRPH2, and the maximum value of the waveform Memorize Max and Min value Min. Further, the comparator COM quantizes (binarizes) the waveform of GRPH2 calculated by the arithmetic unit AVE with an intermediate intensity and converts it into a data string such as GRPH3. This data string is a temporary integer part of the absolute code. The calculation unit CULC converts the first temporary absolute code into first angle data having a mark period as a minimum unit.

  Further, gain correction data such as GRPH4 is generated by the arithmetic unit GAIN, and the waveform data of GRPH1 obtained from the light receiving element array PDA is multiplied by the gain correction data multiplier MULT0 of GRPH4. By this multiplication, waveform data such as GRPH5 in which the influence of the amplitude modulation of the absolute code is removed and the amplitude is normalized is generated.

  The waveform data of GRPH5 is distributed to four signals. Multiplier MULT1 multiplies the first of the distributed signals by the waveform data {(1-sin2πφ) / 2} held in register RGST (SIN−). The multiplier MULT2 multiplies the second of the distributed signals by the waveform data {(1-cos2πφ) / 2} held in the register RGST (cos−). The multiplier MULT3 multiplies the third of the distributed signals by the waveform data {(1 + sin2πφ) / 2} held in the register RGST (sin +). The multiplier MULT4 multiplies the fourth of the distributed signals by the waveform data {(1 + cos2πφ) / 2} held in the register RGST (cos +). The adders SUM1 to SUM4 calculate the sum of the output values of the respective elements, so that A (+), B (+), A (-), B (- ) Four-phase signal (voltage).

  The arc tangent calculator ATAN uses the difference signal between the sum signals A (+) and A (-) and the difference signal between the sum signals B (+) and B (-) to calculate the arc tangent calculation ATN (or Refer to the table by division). Then, the calculation unit CULC can calculate the phase PHS of the periodic signal having the same resolution as a so-called incremental encoder. Note that φ corresponds to a value obtained by dividing the phase in which the serial transfer waveform moves by one period of light and dark by 2π. This phase PHS information is very accurate even if the original sine wave signal output from the light receiving element array PDA itself is close to a triangular wave or trapezoidal, and the phase is usually more than 1000 divisions. Interpolation is possible. For example, if the arctangent value is accommodated in the K-bit register RGST7, the phase PHS of the periodic signal that is the interpolation unit can be expressed in K bits. The calculation unit CULC calculates the second angle data having a minimum unit smaller than the first angle data with the resolution of the length of the division into which the mark period is divided from the information of at least one phase PHS of the periodic signal whose amplitude is normalized. Is calculated.

  A method for generating the M-bit integer part of the absolute code from the temporary integer part of GRPH3 will be described. First, it is assumed that the value obtained by converting the phase PHS of the periodic signal into a value after the decimal point in decimal notation is θ and has a value in the region of 0 ≦ θ <1. FIG. 4 shows an area of 7 bits (for 7 periods of light and dark) when the slit formed on the cylindrical surface rotates in one direction and the value of θ fluctuates in increments of 0.25. The two graphs for the six values of θ in FIG. 4 are a light receiving element array output waveform graph (corresponding to GRPH1 in FIG. 2) and a temporary integer part graph (corresponding to GRPH3 in FIG. 2). Then, a photoelectric conversion element having a reference phase corresponding to the value of θ is selected, and a value obtained by extracting data from the photoelectric conversion element is described below with a numerical value of 0 or 1.

  As shown in FIGS. 4C and 4D, when 0 ≦ θ <0.5, the code “01001” is adopted by discarding the front and rear of the 7-bit data. Further, as shown in FIG. 4E, when 0.5 ≦ θ <1, the code “01001” is adopted by discarding the previous two bits of 7-bit data.

  (A) and (B) in FIG. 4 correspond to the case of moving in the opposite direction from (C). Therefore, since 0.5 ≦ θ <1, the code “10011” is adopted by discarding the previous two bits of 7-bit data. If the phase of (C) is zero, it is the negative side, so it is correct to use the previous code. Such an acquisition rule is an example, but the true integer part of the M-bit of the absolute code is determined by extracting the data from the temporary integer part according to the prescribed rule and arranging it according to the phase θ of the interpolation part. it can.

  Although the integer part of M bits to be finally output may be the cyclic code recorded in the slit formed on the cylindrical surface, it is preferably converted into a general binary code. The calculation unit CULC combines the M-bit integer part M-CODE and the K-bit interpolation part K-CODE obtained as described above to generate data representing the absolute rotation angle of the cylindrical body, and generates a serial signal. Output as.

  The above operation is established when the geometrical arrangement of the head portion HEAD and the cylindrical surface, which is the basis for the calculation of the correction data for the non-uniformity of the period, is the same as the reference value at the time of design. However, in actuality, the distance d (relative position) between the head portion HEAD and the cylindrical surface is not the same as the reference value at the time of design, and an error occurs. Whether or not the error of the distance d between the head portion HEAD and the cylindrical surface is within an allowable range can be determined using signals of a predetermined number of marks corrected by the calculation unit CULC. In the present embodiment, as a corrected mark signal used to determine whether or not the error is within an allowable range, a four-phase signal (A + corresponding to a four-phase incremental encoder) stored in the register RGST6 in FIG. , B +, A-, B-). For example, when the sum of squares of four-phase signals (A +, B +, A-, B-) exceeds a specified value, the light-emitting diode (LED) lights continuously, flashes slightly below the specified value, and even less It is possible to display the degree of error by a method such as not lighting up. Of course, the sum of squares may be directly output from the output terminal.

  When the degree of error is large, the calculation unit CULC changes the interval value, corrects the non-uniformity value of the period using the changed interval value, and changes the mark value using the corrected non-uniformity value. The signal correction is repeatedly performed until the error is within the allowable range. Specifically, the value of the interval d is changed to change the waveform data in the registers RGST (SIN-), RGST (COS-), RGST (SIN +), RGST (COS +), and the result is the sum of squares value. And change to d so that the value of the sum of squares is maximized. By doing so, the signal can be automatically detected in an optimum state. In other words, even if there is a placement error between the head section and the cylindrical surface, stable operation is possible, resulting in an absolute rotary encoder that is highly tolerant of placement errors. it can.

  In this embodiment, since the influence of the non-uniformity of the mark period caused by distorted projection of the mark area on the detector can be canceled, an absolute having a cylindrical surface with a desired diameter from a large diameter to a small diameter The measurement accuracy of the rotary encoder can be improved. Further, since information about whether or not the distance between the cylindrical surface and the head portion is appropriate is obtained, the head portion can be easily corrected to an appropriate position with respect to the cylindrical surface.

[Other Embodiments]
The present invention is not limited to the configuration of the embodiment described above, and various modifications and changes can be made within the scope of the gist. In the above embodiment, the slit is formed on the outer surface of the cylindrical surface, and the head part Head is disposed on the outer surface of the cylindrical surface. However, a slit may be formed on the inner surface of the cylindrical surface, and the head part Head may be disposed inside the cylindrical surface. In this case, contrary to the above-described embodiment, the light-dark cycle gradually narrows toward the periphery. Therefore, the phase output from each light receiving element is configured to be shifted at unequal intervals. The correspondence is solved by the same method as the above embodiment.

  In the above embodiment, the absolute code is embedded by disposing the non-reflective slits at equal intervals and disposing the reflective slit or the semi-reflective slit therebetween. However, conversely, the absolute code may be embedded by arranging the total reflection slits at equal intervals and arranging a non-reflection slit or a semi-reflection slit between them. In this case, the light quantity distribution is obtained by inverting the waveform of GRPH0 upside down, and can be similarly configured by changing the binarization processing or the like accordingly.

  In the above embodiment, the amount of reflected light is binary of 100% and 50%. However, it is also possible to use a cyclic code in which the amount of reflected light is set to other values, for example, 100% and 70%, or two values or more, for example, four values of 100%, 75%, 50% and 25%. In that case, information of 2 bits or more per 1 slit can be embedded. Further, the light receiving element array PDA may acquire and calculate a signal with a pattern having a light and dark number equal to or greater than the number of bits of the absolute code as the number of light receiving elements. In that case, the so-called “incremental effect of the grating of the incremental encoder” of the periodic signal is increased, and further high accuracy can be obtained. In addition, a method of reducing the influence of partial read errors (adding redundancy) by appropriate signal processing can be applied.

  In the above-described embodiment, the distance d between the light source (light emitting element) LED and the cylindrical surface and the distance d between the cylindrical surface and the light receiving element array PDA are configured to be equal. . However, it should be noted that the projection magnification changes. Further, the present invention can be modified to a transmissive encoder configured to receive light transmitted through a mark formed on a cylindrical surface. In that case, the light emitting element LED and the light receiving element array PDA may be separately arranged inside and outside the cylindrical surface. However, it should be noted that the projection magnification changes.

  In the present embodiment, as shown in FIG. 2, the calculation unit CULC holds the reference phase waveform in four registers, and multiplies it by the periodic signal (amplitude normalized) obtained from the light receiving element array PDA. Thus, a four-phase phase difference signal equivalent to the incremental encoder was generated. However, an equivalent function may be realized by another algorithm or flow. For example, a method of adding / subtracting / multiplying a signal from the light receiving element array with a parallel analog circuit, a method of adding / subtracting / multiplying / filtering with a serial analog circuit, a signal of the light receiving element array is immediately AD converted, and digital information is obtained by an FPGA or the like There is a method of arithmetic processing. In the present embodiment, the diverging light is directly illuminated onto the slit of the cylindrical surface, but the projection optical system can be modified such as a method of collimating and illuminating parallel light, or a method of collimating only in one direction with a cylinder lens.

  LED: Light source. PDA: Detector. CULC: Signal processor.

Claims (11)

A rotary encoder,
A cylindrical body that includes a cylindrical surface in which a plurality of marks are arranged with a period along the circumferential direction, and that rotates about the axis of the cylindrical surface;
A light source that emits light toward the cylindrical surface;
A detector that includes a plurality of photoelectric conversion elements, and detects a predetermined number of marks among the plurality of marks by the light emitted from the light source by the plurality of photoelectric conversion elements;
A signal processing unit for obtaining a rotation angle of the cylindrical body based on the output of the detector;
With
The detector outputs a periodic signal corresponding to the predetermined number of marks having a plurality of non-uniform periods;
The signal processing unit is based on the periodic signal and data having a plurality of non-uniform periods and a reference phase obtained based on a geometric arrangement of the light source, the cylindrical surface, and the detector. Obtaining the rotation angle,
A rotary encoder characterized by that.   The signal processing unit includes a multiplier that performs multiplication between the periodic signal and the data, and an adder that obtains a sum of output values of the multiplier, and obtains the rotation angle based on the sum. The rotary encoder according to claim 1.   The signal processing unit obtains a plurality of phase signals having different phases from a plurality of the sums of the plurality of data having different reference phases, and obtains the rotation angle based on the plurality of phase signals. The rotary encoder according to claim 2.   The rotary encoder according to any one of claims 1 to 3, wherein the light source is a point light source. The plurality of marks includes a plurality of types of marks, and is arranged along the circumferential direction with the period and the gap,
The signal processing unit
Generating a data string composed of the predetermined number of data by quantizing the amplitude of the periodic signal for each of the plurality of periods, and obtaining first angle data based on the data string;
Based on the periodic signal and the data, a phase of the periodic signal related to at least one of the plurality of periods is obtained, and based on the phase, second angle data having a resolution higher than the resolution of the first angle data. And
Based on the first angle data and the second angle data, data representing an absolute rotation angle of the cylindrical body is generated.
The rotary encoder according to any one of claims 1 to 4, wherein the rotary encoder is provided.   Before the quantization, the sum of the periodic signals is obtained with respect to each of the plurality of photoelectric conversion elements and the surrounding photoelectric conversion elements over one corresponding period among the plurality of periods. Item 6. The rotary encoder according to Item 5.   The rotary encoder according to any one of claims 1 to 6, wherein the data is obtained and held in advance based on the geometric arrangement.   8. The signal processing unit according to claim 7, wherein the signal processing unit determines whether an error from the reference value of the geometric arrangement is within an allowable range based on the periodic signal and the data. Rotary encoder.   If the signal processing unit determines that the error is not within the allowable range, the signal processing unit changes the reference value and changes the data based on the changed reference value. The rotary encoder according to claim 7, wherein the rotary encoder is executed so as to be within.   The rotary according to any one of claims 1 to 9, wherein the plurality of photoelectric conversion elements detect light emitted from the light source and reflected by the predetermined number of marks. Encoder.   The rotary according to claim 1, wherein the plurality of photoelectric conversion elements detect light emitted from the light source and transmitted through the predetermined number of marks. Encoder.