Patent Grant
7825367
The present invention relates to a rotary encoder, and more particularly, to an incremental encoder and an absolute encoder used for a surveying apparatus and the like capable of self-calibrating a measured value.
A surveying apparatus is equipped with a rotary encoder for measuring a horizontal angle and a vertical angle. The rotary encoder is a device that measures a horizontal angle and a vertical angle by a rotary disk that rotates in conjunction with a telescope and a detector that reads a scale provided along the circumferential direction of this rotary disk.
Rotary encoders include incremental encoders and absolute encoders. An incremental encoder is provided with a rotary disk with an equal-interval and equal-width scale (slits). This incremental encoder irradiates light emitted from a light source onto the rotary disk and detects brightness/darkness according to a number of the scale by a photodetector so as to measure a rotation angle of the rotary disk.
An absolute encoder is, for example, as shown in
However, in the conventional rotary encoder used for a surveying apparatus, since it is difficult to attach the rotary disk 1 to a rotating shaft without eccentricity and to provide the scale and angle codes completely at equal intervals, errors have occurred in measured values. In order to eliminate such errors, self-calibration by a pair of measurements has been carried out in accordance with the ISO and JSIM standards. For this self-calibration, collimators are disposed around the surveying apparatus, the collimators are collimated while rotating the surveying apparatus by an appropriate angle, and a scale or angle code error is measured.
Patent Document 1: Japanese Published Unexamined Patent Application No. 2002-13949
However, in the above-described self-calibration by a pair of measurements, not only is collimation required by hand, but also about one hour of time is required, so that there has been a problem of a great burden on a worker. In addition, an inexperienced worker may increase the collimation error, so that there has also been a problem in lack of accuracy of self-calibration. Furthermore, there has also been a problem that equipment that requires several collimators is expensive. Moreover, there has also been a problem that it is difficult to carry out self-calibration for a vertical encoder.
The present invention has been made in view of the aforementioned problems, and an object thereof is to provide a rotary encoder capable of carrying out self-calibration for a horizontal encoder as well as a vertical encoder without requiring special equipment and simply and inexpensively with a high degree of accuracy while reducing the burden on a worker.
In order to solve the aforementioned problems, as claim 1 of the invention, in a rotary encoder including: a rotary disk with an angle code; a light source that irradiates the angle code; a detector that reads the angle code; and a processing unit that acquires a reading value f(θ) based on the angle code read by the detector, every time the rotary disk is rotated by a predetermined angle, where a rotation angle of the rotary disk is provided as θ and an arbitrary angle from the rotation angle θ within a reading range on the detector is provided as φ, the process sing unit acquires reading values f(θ+φ) and f(θ) within the reading range on the detector, and self-calibrates the reading value f(θ) based on a change in a difference g(θ,φ) between the reading values f(θ+φ) and f(θ) when the rotary disk is rotated one turn.
As claim 2 of the invention, in the claim 1 the invention, the rotational position of the rotary disk where the reading values f(θ+φ) and f(θ) are acquired is provided as a position where 360° is roughly equally divided and a position where the reading range on the detector is overlapped at an adjacent rotational position thereto, an average gm(φ) of the differences g(θ,φ) between the reading values f(θ+φ) and f(θ) traversing one turn and deviations h(θ,φ) between the differences g(θ,φ) and the average gm(φ) at respective rotational positions are taken, and by arranging while overlapping the deviations h(θ,φ) in parts where the reading ranges are overlapped between adjacent rotation angle positions, self-calibration of the reading value f(θ) of 0 to 360° is carried out.
As claim 3 of the invention, in the claim 1 of the invention, the rotational position of the rotary disk where the reading values f(θ+φ) and f(θ) are acquired is provided as a position where 360° is roughly equally divided, based on a change in the difference g(θ,φ) between the reading values f(θ+φ) and f(θ), the difference g(θ,φ) is determined as a Fourier series, and self-calibration of the reading value f(θ) of 0 to 360° is carried out based on the result.
As claim 4 of the invention, in the claim 3 of the invention, in the Fourier series, one or more arbitrary higher harmonics are determined.
As claim 5 of the invention, in the claims 1, 2, 3, or 4 of the invention, for the arbitrary angle θ, one or more arbitrary positions are used.
As claim 6 of the invention, in the claim 1, 2, 3, 4, or 5 of the invention, the reading values f(θ+φ) and f(θ) are acquired by averaging a large number of times of reading values.
According to the claim 1 of the invention, by the present invention, every time the rotary disk is rotated by a predetermined angle, where a rotation angle of the rotary disk is provided as θ and an arbitrary angle from the rotation angle θ within a reading range on the detector is provided as φ, reading values f(θ+φ) and f(θ) within the reading range on the detector can be acquired, and the reading value f(θ) is self-calibrated based on a change in a difference g(θ,φ) between the reading values f(θ+φ) and f(θ) when the rotary disk is rotated one turn, and therefore, self-calibration of a horizontal encoder and a vertical encoder can be carried out without requiring special equipment and simply and inexpensively with a high degree of accuracy while reducing the burden on a worker, and a highly-accurate measured angle value can be acquired.
According to the claim 2 of the invention, by the present invention, furthermore, the rotational position of the rotary disk where the reading values f(θ+φ) and f(θ) are acquired is provided as a position where 360° is roughly equally divided and a position where the reading range on the detector is overlapped at an adjacent rotational position thereto, an average gm(φ) of the differences g(θ,φ) between the reading values f(θ+φ) and f(θ) traversing one turn and deviations h(θ,φ) between the differences g(θ,φ) and the average gm(φ) at respective rotational positions are taken, and by arranging while overlapping the deviations h(θ,φ) in parts where the reading ranges are overlapped between adjacent rotation angle positions, self-calibration of the reading value f(θ) of 0 to 360° is carried out, and therefore, self-calibration can be carried out more quickly.
According to claim 3 of the invention, by the present invention, the rotational position of the rotary disk where the reading values f(θ+φ) and f(θ) are acquired is provided as a position where 360° is roughly equally divided, based on a change in the difference g(θ,φ) between the reading values f(θ+φ) and f(θ), the difference g(θ,φ) is determined as a Fourier series, and self-calibration of the reading value f(θ) of 0 to 360° is carried out based on the result, and therefore, not only can further highly-accurate self-calibration be carried out, but also a highly-accurate measured angle value can be acquired.
According to the claim 4 of the invention, furthermore, in the Fourier series, one or more arbitrary higher harmonics are determined, and therefore, calibration accuracy can be improved without increasing the time required for calibration very much.
According to the claim 5 of the invention, furthermore, for the arbitrary angle φ, one or more arbitrary positions are used, and therefore, calibration accuracy can further be improved.
According to the claim 6 of the invention, furthermore, the reading values are acquired by averaging a large number of times of reading values, and therefore, influence of noise and the like can be reduced, and self-calibration can be carried out with a higher degree of accuracy.
Hereinafter, the present invention will be described in detail based on the drawings. First, description will be given of a first embodiment of the present invention based on
In the plan view of
However, the reading values f(θ) and f(θ+φ) are periodic functions with a period of 360° and can therefore be expressed by the following formulas by use of Fourier series.
f(θ)=θ+Σn=1M/2An*sin(nθ+αn) (1)
f(θ+φ)=(θ+φ)+Σn=1M/2An*sin {n(θ+φ)+αn} (2)
Here, n means a natural number (1, 2, 3, . . . M/2) expressing the order of a higher harmonic, An means amplitude of the n-th higher harmonic, an means an initial phase of the n-th higher harmonic, and M means a scale number, which is an even number.
Therefore, by a comparison between the respective reading values f(θ) and f(θ+φ) of the point A and point B, a difference g(θ,φ) therebetween shown by the following formula is determined.
g(θ,φ)=f(θ+φ)−f(θ) (3)
This difference g(θ,φ) can be expressed as in the following formula based on formulas (1) and (2).
g(θ,φ)=f(θ+φ)−f(θ) =(θ+φ)+Σn=1M/2An*sin {n(θ+φ)+αn}−θ+Σn=1M/2An*sin(nθ+αn) (4)
When the calculation is further continued, g(θ,φ) finally becomes as in the following formula.
In this manner, g(θ,φ) can also be expressed by a Fourier series with a period of 360°.
Therefore, as shown in
θ=f(θ)−Σn=1M/2An*sin(nθ+αn)=f(θ)−E(θ) (9)
Here, E(θ)=Σn=1M/2An*sin(nθ+αn) means an error function. Here, in a Fourier series of the error function E(θ), it is unnecessary to determine excessively high-order higher harmonics, and in a rotary encoder of opposing detection, an eccentric error has been cancelled, and thus it is often sufficient to determine the second higher harmonic (n=2). Even when the third and higher harmonics are determined, this only increases the calculation amount and storage capacity and calibration accuracy is not improved very much in most cases.
In order to realize a higher calibration accuracy, it is necessary to eliminate influence of noise and the like. Therefore, at the time of angle measurement, by carrying out data acquisition from the CCD linear sensor 3 a large number of times and taking an average, the reading values f(θ) and f(θ+φ) are determined. As is apparent from the above description, according to the present embodiment, the error function E(θ) of a measured angle value is determined without particularly requiring an experienced engineer or expensive equipment, and a higher calibration accuracy can be realized.
The above self-calibration is carried out by the CPU 6. Procedures of this self-calibration are shown in
First, when a self-calibration program is started, the process proceeds to step S1 to acquire reading values f(θj) and f(θj+φ) directly under the two points A and B on the CCD linear sensor 3. After carrying out this reading a predetermined number of times, the process proceeds to step S2 to calculate an average of the reading values f(θj+φ) and f(θj). Furthermore, the process proceeds to step S3 to calculate and store a difference g(θj,φ)=f(θj+φ)−f(θj) between the two reading values f(θj+φ) and f(θj).
Next, the process proceeds to step S4 to check whether all possible data has been acquired, that is, to check whether g(θj,φ) for each predetermined angle θj has been acquired. For example, it is checked for θ whether g(θj,φ) has been acquired at intervals of 5° traversing 360°. If all possible data has not been acquired, the process proceeds to step S5, the rotary disk 1 is manually rotated by a predetermined angle (automation by a stepping motor or the like is also possible), and the process returns to step S1. Thereafter, steps S1 to S5 are repeated to determine and store g(θj,φ) for each predetermined angle θj.
When it is judged in step S4 that the predetermined number of times of data acquisition has been completed, the process proceeds to step S6 to calculate, based on g(θj,φ) for each predetermined angle θj, coefficients of the Fourier series of g(θ,φ) shown in formula (5), that is, B0, Bn, and βn. Furthermore, the process proceeds to step S7 to calculate, based on these B0, Bn, and βn, coefficients of a Fourier series of an error function E(θ) of the reading value f(θ), that is, An and αn. Furthermore, the process proceeds to step S8 to determine and store these An and αn. This allows accurately determining the true angle θ based on the reading value f(θ) by use of formula (9) in the subsequent measurement.
Next, description will be given of a second embodiment of the present invention based on
f(θ)=θ+Σn=1∞An*sin(nθ+αn)=θ+E(θ) (11)
f(θ+φ)=(θ+φ)+Σn=1∞An*sin {n(θ+φ)+αn}=(θ+φ)+E(θ+φ) (12)
Here, E(θ) means an error function.
Here, a difference g(θ,φ) between both reading values f(θ+φ) and f(θ) shown by the following formula is calculated.
g(θ,φ)=f(θ+φ)−f(θ) (13)
When formulas (11) and (12) are substituted for formula (13), the following formula is obtained.
g(θ,φ)=φ+E(θ+φ)−(θ) (14)
Now, as shown in
gm(φ)={Σj=0N−1g(θj,φ)}/N≈φ (15)
Therefore, h(θj,φ) defined by the following formula (16) is further calculated.
h(θj,φ)=g(θj,φ)−gm(φ) (16)
When formulas (14) and (15) are substituted for formula (16), the following formula is obtained.
h(θj,φ)≈E(θj+φ)−E(θj) (17)
Here, where E(θj) is provided as Cj when E(θj) is considered as a constant assuming that the position of θj is set to a base point, formula (17) can be expressed as in the following formula.
h(θj,φ)≈E(θj+φ)−Cj (18)
Although this Cj is an unknown, since the reading ranges are overlapped at the adjacent rotational positions θj−1 and θj, h(θj−1,φ′) and h(θj,φ) must be coincident in the overlapping range. An offset adjustment amount ΔCj therebetween is expressed by the following formula.
ΔCj=Cj−Cj−1 (19)
When h(θj,φ) adjusted by use of this offset adjustment amount ΔCj is provided as h′(θj,φ), h′(θj,φ) is as in the following formula.
h′(θj,φ)=h(θj+φ)+ΔCj (20)
Thereafter, ΔCj is calculated for each angle θj, and when formulas (18) and (19) are substituted for formula (20), the following formula is obtained.
h′(θj,φ)=E(θj+φ)+Cj−1 (21)
Hereon, the following formula is obtained for the arbitrary θ.
h′(θ,0)=E(θ)+Cj−1 (22)
Although this h′(θ,0) includes an offset amount Cj−1, usually, Cj−1 may be ignored to make h′(θ,0) roughly equal to the error function E(θ). Therefore, when h′(θ,0) adjusted by adding the offset adjustment amount ΔCj to h(θj,φ) determined by formula (16) is determined as the error function, the true angle θ is determined by the following formula.
θ=f(θ)−h′(θ,0) (23)
Next, procedures of this self-calibration are shown in
First, when a self-calibration program is started, the process proceeds to step S11 to acquire reading values f(θj+φ) and f(θj) directly under a large number of points on the CCD linear sensor 3 while varying φ. After carrying out this reading a predetermined number of times, the process proceeds to step S12 to calculate an average of the reading values f(θj+φ) and f(θj). Furthermore, the process proceeds to step S13 to calculate a difference g(θj,φ)=f(θj+φ)−f(θj) between the two reading values.
Next, the process proceeds to step S14 to check whether all possible data for each predetermined angle has been acquired, that is, to check whether g(θj,φ) according to φj for each predetermined angle has been acquired. For example, it is checked for θ whether g(θj,φ) has been acquired at intervals of 5° traversing 360°. If all possible data has not been acquired, the process proceeds to step S15, the rotary disk 1 is manually rotated by a predetermined angle (automation by a stepping motor or the like is also possible), and the process returns to step S1. Thereafter, steps S1 to S5 are repeated to determine and store g(θj,φ) for each predetermined angle θj.
When it is judged in step S14 that all possible data acquisition has been completed, the process proceeds to step S16 to calculate an average gm(φ) of g(θj,φ) (see formula (15)). Next, the process proceeds to step S17 to calculate h(θj,φ)=g(θj,φ)−gm(φ) (see formula (16)). Next, the process proceeds to S18 to calculate the offset adjustment amount ΔCj between h(θj−1,φ′) and h(θj,φ) (see formula (19)). Next, the process proceeds to step S19 to add the offset adjustment amount ΔCj to h(θj,0) and determine and store the error function h′(θ,0).
This allows accurately determining the true angle θ based on the reading value f(θ) by use of formula (23) in the subsequent measurement. In the case of the present embodiment, since the error function h′(θj,0) can be determined by a small calculation amount, the time required for self-calibration can be reduced.
However, it is possible to variously modify the aforementioned embodiments. For example, it is not always necessary to determine the reading values f(θ) and f(θ+φ) directly under the foot A of a vertical line drawn from the center O of the rotary disk 1 to the CCD linear sensor 3 and the point B on the CCD linear sensor 3, and reading values at a plurality of positions arbitrarily determined on the CCD linear sensor 3 may be determined. In addition, it may be possible to select a third point on the CCD linear sensor 3 in
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/JP2006/316690 | 8/25/2006 | WO | 00 | 5/15/2008 |
| Publishing Document | Publishing Date | Country | Kind |
|---|---|---|---|
| WO2007/086160 | 8/2/2007 | WO | A |
| Number | Name | Date | Kind |
|---|---|---|---|
| 6928386 | Hasser | Aug 2005 | B2 |
| Number | Date | Country |
|---|---|---|
| 11-118520 | Apr 1999 | JP |
| 2002-013949 | Jan 2002 | JP |
| 2006043403 | Apr 2006 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 20100181468 A1 | Jul 2010 | US |
