JP2002213938A - Method of finding eccentricity of rotation body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately find an eccentricity generated in rotation of a rotation body, by removing an irregular and indeterminate deflection (non-periodic deflection) from measured data. SOLUTION: The distance up to the rotation body is measured in every prescribed angle while rotating the rotation body by one rotation, a tentative eccentric curve of the rotation body is found based on initial data provided by the measurement, and tentative shape data of the rotation body are found based on the tentative eccentric curve, a contradictory portion contradictory in a shape of the rotation body is found based on the tentative shape data of the rotation body, and correction data wherein a value in the start portion of the data are made consistent with a value in the end portion are found to eliminate the contradictory portion contradictory in the shape of the rotation body.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for removing irregular and indefinite fluctuations contained in measurement data in order to determine the amount of eccentricity generated in a rotating body.

[0002]

2. Description of the Related Art In a magnetic tape recording / reproducing apparatus or the like, generally, a helical scan system using a rotating drum which is provided with a plurality of magnetic heads arranged in a circumferential direction and rotated by a motor is used for recording data on a magnetic tape. Has been adopted.

[0003] Manufacturing of a drum assembly including the motor and the rotating drum generally requires high-precision production equipment specifications. In such a production facility, reproducibility at the μm level is required for the detection accuracy of the eccentricity and vibration level of the rotating shaft of the rotating drum.

[0004] The eccentricity of the rotating drum means a state in which the rotating drum rotates in an ellipse due to a deviation between the center of the rotating drum and the center of rotation.

That is, among the components of the drum assembly,
In the production work performed while correcting the eccentricity during rotation of the rotating drum, which is a rotating body, it is necessary to consider the eccentricity (aperiodic runout) that changes due to the rotation of the rotating drum, but the eccentricity is corrected by the operator during the work. It is difficult to produce the drum assembly as it is.

Further, it is impossible to predict the timing and magnitude of aperiodic run-out which occurs irregularly when the rotary drum rotates. Similarly, when the data obtained when the rotating drum is rotated is examined, the trajectory turns slightly trajectory every cycle.
That is, the rotating drum does not return to the same position even when rotated by 360 °. Therefore, at present, there is no way to cope with aperiodic vibration.

Therefore, during a manufacturing operation, conventionally, as shown in FIGS. 1 and 19, in a drum assembly production facility 1, a rotating drum (rotating body) 3, which is a part of the drum assembly, and a magnetic head (hereinafter, referred to as a part). , "Head") 2,
By moving the positional relationship with the microscope camera (hereinafter abbreviated as “camera”) 4 while looking at the positional relationship with 2,... Here, the movement between the adjustment portions is performed by rotating the rotating drum 3. However, if there is eccentricity during rotation of the rotating drum 3, high-precision work cannot be guaranteed.

Accordingly, the amount of eccentricity is measured in advance assuming that the rotary drum 3 draws a constant eccentricity, and the amount of eccentricity is used as a correction value of the distance to move the camera 4 back and forth. After measuring the amount of eccentricity,
The rotating drum 3 had to be rotated in the reverse direction by 360 °, and the work had to be performed in the orbit during measurement.

[0009] However, there is no technique for examining the amount of eccentricity in the case of including the non-periodic run-out in detail, and even if the approximate amount of eccentricity can be grasped, the accuracy sufficient to guarantee the quality in the work cannot be obtained. Since there is no means for avoiding the question of the reproducibility of the correction value, if the electrical evaluation in a later step becomes defective, the position adjustment of the heads 2, 2,. However, the correction value had to be changed and readjusted.

Further, in the conventional manufacturing operation, when the rotating drum 3 is mounted, in order to eliminate backlash after the mounting, the moving direction at the time of mounting is determined, and the rotating drum 3 is abutted in a fixed direction (eccentricity). It is fixed, and work is performed while taking into account the correction amount that has been checked in advance for each rotation angle. However, since the abutment condition changes depending on the operator and the lot of parts, accurate correction is not possible. Can not.

At the time of adjustment work using the camera 4, it is only necessary that the distance correction can be performed after the head 2 moves in front of the camera 4, but as shown in FIG. In order to measure the distance from the current position to the side surface of the rotating drum 3, the camera 4 and the length measuring sensor 6 must be attached to the same position. There was a problem that was reflected.

The measuring point of the length measuring sensor 6 is the head 2
A position as close as possible is good, but measurement is not possible due to the notch 3c in the portion where the head 2 is provided. 18 and 52, the side surface 3b of the rotating drum 3 is not flat but has many irregularities when viewed at the μm level, and is perpendicular to a plane orthogonal to the rotation axis. However, since it is slightly inclined, if the height or position of the measurement portion changes, the distance to the side surface of the rotating drum 3 will change.

As described above, there are many restrictions on the range that can be measured only to know the approximate amount of eccentricity, and the amount of correction accompanying eccentricity usually does not go as designed, and is empirically obtained at the production stage. At present, standard values were used while being repeatedly modified.

The bearing system used for the rotating part of the drum assembly production equipment 1 includes a ball bearing system which involves an aperiodic run-out but is inexpensive, and an air bearing which is expensive but generates almost no aperiodic run-out. There is a bear bearing method. Although the latter air bearing method has the advantage that eccentricity is unlikely to occur, it also affects the work layout and requires continuous air supply even when not in use, which is disadvantageous in terms of noise and energy saving. Also has the disadvantage of being large. Even if the above drawbacks are ignored, in production facilities that prioritize quality,
Although the latter uses the air bearing method, old production equipment and equipment that cannot adopt the air bearing method due to cost have to use the former ball bearing method at the expense of quality and workability. Was.

[0015]

SUMMARY OF THE INVENTION In view of the above-mentioned problems, the present invention eliminates irregular and irregular vibration (non-periodic vibration) components from measurement data, thereby achieving accurate rotation occurring when a rotating body rotates. It is an object to provide a method for obtaining a large eccentric amount.

[0016]

In order to solve the above-mentioned problems, the present invention measures the distance to the rotating body at predetermined angles while rotating the rotating body once, and calculates the distance from the initial data obtained by the measurement. A temporary eccentric curve of the rotating body is determined, and temporary rotating body shape data is determined from the temporary eccentric curve. From the temporary rotating body shape data, a portion inconsistent in shape of the rotating body is determined. In order to eliminate the contradictory part of the above, corrected data is obtained so that the value of the starting point and the value of the ending point of the data become the same value.

Therefore, an accurate eccentric amount can be obtained.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a method for determining the amount of eccentricity of a rotating body according to the present invention will be described below with reference to the accompanying drawings.

In the following embodiment, when manufacturing the drum assembly used as a rotating head in a magnetic recording / reproducing apparatus or the like, an accurate eccentricity of a rotating shaft is obtained and measured data is used. Eliminates the effects of periodic and irregular fluctuations (hereinafter referred to as "non-periodic touch") and removes the effects of rotating drums (hereinafter,
This is applied to a production facility that adjusts the position of a head when mounting the head on a "drum".

In the terms used in the following description, "data" refers to the distance to the drum side surface at a predetermined angle obtained when the drum is rotated once, and "AC component" refers to the distance. It shows a range that fluctuates in the data waveform, and “DC component” is an average value of all data, and when this value is subtracted from each data, the AC component remains. Indicates the difference between the uppermost point and the lowermost point of the sine wave, and "X distance" indicates the distance to the drum side surface.

First, at the time of manufacturing the drum assembly,
A description will be given of an outline of a position adjustment operation of a magnetic head (hereinafter, abbreviated as “head”) and a production facility 1 used for the operation.

That is, during the production of the drum assembly,
By using the production equipment 1, the heads 2, 2,..., As shown in FIG.
The work is performed on the base surface 3a while performing position adjustment with a predetermined angle division and an angle tolerance in seconds. At this time, the head 2 is enlarged and displayed on a monitor 5 via a microscope camera (hereinafter abbreviated as “camera”) 4.

In the above operation, when the drum 3 is rotated so that another arbitrary head 2 comes to the adjustment position after the specific head 2 is adjusted, the head 2 is monitored by the eccentricity of the drum 3. It looks off the center of the screen. Therefore, the center line (marked line) on the monitor screen is moved in accordance with the movement of the center position of the rotating drum 3, so that the head 2 is displayed at the center position.

The movement of the marked line is determined by checking the amount of eccentricity when the drum 3 is rotated, calculating the amount of left and right displacement in accordance with the rotation angle (angle division), and corresponding to the amount of eccentricity on the monitor 5. This is performed by moving the marking line to a position (a position indicating the center point of the drum 3).

The method of measuring the amount of eccentricity is as follows.
It is rotated, and the amount of displacement up to the side surface 3b is checked over the entire circumference to make data. This is processed to extract only the amount of fluctuation, and the component of one cycle is defined as the amount of eccentricity. Normally, the eccentricity returns to the original state when the drum 3 is rotated once.

The amount of eccentricity calculated as described above is used to know the amount of right and left deflection of the drum 3 on the monitor screen.

That is, as shown in FIGS. 2 and 3, the camera 4 and the length measuring sensor 6 are integrated.
Is closer to the camera 4 (length measuring sensor 6) side.

Incidentally, the length measuring sensor is used when the drum 3 rotates.
This is a non-contact sensor for measuring the distance to the side surface (outer peripheral surface) 3b of the drum 3, and is required to have an accuracy capable of linearly detecting irregularities, notches, burrs and the like on the drum side surface 3b in addition to the distance. Further, since the surface roughness of the aluminum drum 3 is 1 μm in width and 2 μm to 3 μm in roughness, it is necessary to average and take in the spot measurement using a laser sensor as the length measurement sensor. .

Further, as shown in FIG. 4, another length measuring sensor 7 is provided as an auxiliary sensor at a position offset by 90 ° from the position where the camera 4 and the length measuring sensor 6 are arranged. Although it is possible to detect the right and left deflection of the drum 3, an accurate eccentric amount cannot be obtained unless the unevenness of the side surface shape of the drum 3 picked up by the length measuring sensor 7 is considered. Therefore, in the present invention, the length measuring sensor 6
The eccentricity is obtained by a calculation process using only.

When the rotation axis of the drum 3 itself moves, there is an aperiodic vibration in which the drum 3 swings right and left.
This aperiodic vibration must also be corrected taking into account the eccentricity. However, sudden fluctuations may occur in the rotation axis, and it was conventionally impossible to calculate only the amount of eccentricity from data including the aperiodic vibration. In addition to this, even if the aperiodic run-out when the drum 3 is rotated once can be recognized, the aperiodic run-out occurring in the next orbit will have a different size and an occurrence time, so that accurate eccentricity during work Correction of the amount was not possible.

Therefore, the drum assembly manufactured under such conditions is made into a finished product that can withstand actual use by repeating electrical adjustment work, that is, electrical evaluation and readjustment.

In recent years, in order to facilitate the above-mentioned electrical adjustment work, a rotating shaft using a bearing having an air bearing structure which does not generate aperiodic run-out has been adopted. Due to restrictions, air must be continuously applied even when not in use, which has many disadvantages in terms of noise and energy saving.

Existing production equipment employs a rotating shaft that uses a ball bearing that causes aperiodic runout. Such production facilities are large in number, and it is difficult to reinvest for the purpose of modification (for example, using an air bearing for a bearing) to improve quality and workability. In addition, there is no technical progress in creating new production equipment if only expensive parts are used, although there are advantages.

Therefore, by applying the method for determining the amount of eccentricity of the rotating body of the present invention to existing production equipment, even if an air bearing which is expensive and has problems in terms of noise and energy saving is not used, non-periodic run-out can be achieved. It is possible to obtain an accurate eccentric amount by correcting the influence on the eccentric amount.

There are two types of configurations of the production equipment to which the present invention is applied. That is, the aperiodic vibration correction is performed by using one length measuring sensor having a conventionally used configuration shown in FIG. 2 and the aperiodic vibration generated from now is predicted in addition to the function shown in FIG. The configuration is such that it can be performed.

The present invention is introduced into existing production equipment in two stages. First, in a state as shown in FIG. 2, the rotating body is introduced for the purpose of measuring the shape (unevenness) and detecting aperiodic run-out. Then, a calculation logic is incorporated to check whether or not the aperiodic vibration is within an allowable range. Next, when it is confirmed that the aperiodic vibration is at a problematic level, the configuration is switched to the configuration shown in FIG.

Incidentally, it is inevitable that incorrect data is mixed in the actual measured value. Among them, there is a case where an unmeasurable part appears in a part of data due to measurement restrictions. In this case, the conditions are worse,
Conventionally, this has been dealt with by changing the measurement points. However, by applying the present invention, it has become possible to solve this problem and also to increase the reliability of the calculation results. .

Next, the definition of the aperiodic vibration will be described with reference to an example.

In the case of the non-periodic vibration, it is necessary to recognize the angle while distinguishing the orbit. That is, as shown in FIG. 5 and FIG.
This means that the distance to the drum 3 after rotation does not match the distance before rotation. When the displacement of the distance to the side surface 3b is measured while rotating the drum 3, the change in the distance is mostly due to the influence of the eccentricity, as shown in FIG.

As shown in FIGS. 7 and 9, if the fluctuation of the displacement of the distance to the side surface 3b, which is similar to a sine wave, is caused by the eccentricity of the drum with respect to the rotating shaft, the same applies even if the rotation is different. If it is an angle split, the distance to the side surface 3b must be the same. However, there are times when the same measurement data cannot be obtained. For this reason, there is a difference between the distance at the 0 ° position before the rotation of the drum 3 and the distance at the 0 ° position in the number of revolutions. Note that the angle division means that the heads 2, 2,... Are attached at positions where the drum 3 is rotated at a specified rotation angle from a reference point.
It is the angle from this reference point to the two mounting positions of each head.

As shown in FIG. 1, the eccentricity factors during rotation of the drum 3 include a deviation between the rotation base 13 on which the drum 3 is mounted and the center of the drum 3, and another factor of the rotation of the motor 14. The center shift between the shaft 14a and the rotating base 13 is considered.

Further, the eccentricity data obtained from the measured data of the distance to the side surface 3b of the drum 3, as shown in FIG.
Instead of combining the eccentricity data of the two different phases described above, eccentricity data of only the deviation between the rotation shaft 14a of the motor 14 and the center of the drum 3 is obtained. That is, the rotating shaft 1
4a and the rotating base 13 may be considered as one rotating body,
What is necessary is just to consider the displacement between the drum 3 and the rotating shaft 14a of the motor 14 when the drum 3 is placed on the rotating body. However, since the purpose is to eliminate eccentricity at the time of rotation, the rotating shaft 1 of the drum 3, the rotating base 13 and the motor 14
It is necessary to mount the three members 4a and 4a so that their central axes are aligned to one. However, in such equipment without play, workability deteriorates.

Even with the same angle division, the distance to the side surface 3b of the drum 3 in FIG. 9 differs before and after the drum 3 makes one rotation. This deviation is defined as a first aperiodic component.
When the drum 3 is rotated once, it should return to the same position and posture as before rotation, but it may come to a slightly shifted position due to aperiodic vibration, and the rotation angle is 0 °. The linear change connecting the displacement between the time and 360 ° is referred to as the first non-periodic vibration component of the rotation (as shown in FIG. 7, the irregular displacement generated during one rotation is not considered). ). This first non-periodic component causes a calculation error when determining the eccentricity of the drum 3 and must be removed first. Note that the displacement amount means a change in the X distance when the Δθ angle changes based on an arbitrary angle θ.

That is, as shown in FIGS. 10 and 11, the rotation period of the drum 3 (up to 360 °) is different from the period (0 ° to indefinite) at which aperiodic vibration occurs. Even after one rotation, the period of the aperiodic vibration is still in the middle. When the angle of drum 3 is 0 ° and 360 °
When rotated, the distance to the side surface 3b must be the same, but the distance will be different due to aperiodic runout. It can be understood that the eccentricity cannot be calculated only by measuring the displacement of the distance to the side surface 3b for one rotation of the drum 3 due to the addition of the aperiodic vibration.

As shown in FIGS. 7 and 12, the non-periodic vibration causes an irregular and indefinite change (sudden change). That is, in addition to the first non-periodic vibration, which has a period in a long range, there is a factor that largely fluctuates irregularly. Therefore, this sudden displacement is defined as a second aperiodic component. The second non-periodic component refers to a phenomenon in which, when the drum 3 rotates, the center of rotation is not determined by physical fluctuation of the bearing. This happens occasionally,
It may be mistaken for the drum shape, but due to irregular occurrence, it draws a different trajectory depending on the lap.

.. Are often seen in a rotating shaft using a ball bearing. As shown in FIG. 13, the shape of the ball bearings 10, 10,... Used in the bearing of the rotating shaft is approximately spherical. , It is necessary to know that the ball bearings 10, 10,... Used are not all the same size.

Therefore, the inner frame 11 rotating via the ball bearings 10, 10,... With respect to the outer frame 12 does not draw a fixed trajectory, but turns to one side suddenly. The non-periodic fluctuation that results in a movement showing a large change causes a change having a certain period. This run-out is often seen when a ball bearing is used for the bearing of the rotating shaft. In a bearing using a ball bearing, the inner frame also rotates in accordance with the rotation of the ball bearing. The center of rotation of the inner frame is eccentric due to the positional relationship between the ball bearing and the frame supporting it, and the sphericity of the ball bearing itself is low (it is not a perfect sphere but a distorted sphere)
Eccentricity.

Here, as one example, when a ball bearing is used as a bearing, the manner in which eccentricity occurs when the inner frame rotates will be described below. That is, depending on the position where the inner frame 11 and the outer frame 12 are engaged with the ball bearing 10,
The inner frame may run eccentrically.

Now, a ball bearing 1 having a diameter of 7.5 mm
The outer diameter is 30m for 0, 10 ... (only one is shown)
Assuming an inner frame 11 of m and an outer frame 12 having an inner diameter of 45 mm (FIG. 14), when the ball bearing 10 makes one rotation,
The ball bearing rolls on the inner periphery of the outer frame 12 by 7.5π. At this time, the outer frame 12 has an inner circumference of 45π.
Therefore, the ball bearing 10 is about 60 ° from the initial position.
(FIG. 15).

The inner frame 11 is provided with the ball bearing 10.
Rotates on the outer periphery of the inner frame 11 by 7.5π. Similarly, since the inner frame 11 has an outer peripheral length of 30π, it is rotated by about 90 ° by the rotation of the ball bearing 10, but in addition, about 60 ° of the movement of the ball bearing 10 is added. Eventually, it will rotate about 150 ° (FIG. 15).

If the inner frame 11 and the outer frame 12 are assumed to have a perfect circular inner or outer diameter shape, the inner frame 11 rotates once if the ball bearing 10 is rotated 2.4 times. Become. In order to return the ball bearing 10 to the original position, the inner frame 11 has only to make 2.5 turns.

By the way, as shown in FIG. 17, if a small diameter is mixed in the ball bearings 10, 10,... Or if the inner frame 11 or the outer frame 12 in contact with the ball bearing 10 has a dent, Inner frame 1 at the recess
1 will be eccentric. Since the cycle of the eccentricity is a period in which the ball bearing 10 makes one rotation, the inner frame 1
One rotation is 2.5 rotations.

Under the same conditions as above, when the sphericity of the ball bearing is poor, the center of the inner frame 11 is eccentric every time the ball bearing 10 rotates, as shown in FIG.

That is, while the inner frame 11 makes one revolution, the ball bearing 10 rotates 2.4 times, so that abnormal data generated from one ball bearing 10 appears two or three times. That is, one of the second aperiodic components is:
The rotation of the inner frame 11 becomes about 1/3 rotation.

Here, the inner frame 11, the ball bearing 10
The point at which the outer frame 12 and the outer frame 12 contact each other varies depending on the position of the ball bearing 10, so that abnormal data having the same value does not always occur, but only the period of the abnormal data will be considered.

Due to the above factors, different eccentricity data can be obtained for each rotation of the inner frame 11, but an approximate cycle is determined by the outer or inner diameter of the inner frame 11 and the outer frame 12 and the diameter of the ball bearing 10. In the above case, when the inner frame 11 makes five turns, the ball bearing 10 makes two turns and returns to the initial position. From this, about five revolutions of the inner frame 11 are the cycle of the aperiodic run-out of the first and second aperiodic components by the ball bearing 10.

Next, the main part of the present invention will be described in detail. The present invention relates to a method for correcting a run-out (aperiodic run-out) that occurs irregularly and irregularly due to the eccentricity of a rotary body. The method mainly includes a non-periodic run-out generated in a rotary body using a ball bearing structure as a rotary shaft bearing structure. It is intended to correct various inconveniences caused by the above.However, even if a rotating body that does not use a bearing structure using ball bearings is affected by similar abnormal data, this is corrected, The present invention can also be used as a correction system for accurately correcting a target eccentricity.

As shown in FIG. 2, it is assumed that the work for adjusting the mounting position of the product part is performed at a fixed and accurate distance from the camera 4. If the length measurement sensor 6 controls the distance to the drum 3 to be kept constant, the distance to the drum 3 is corrected each time an aperiodic run-out occurs. There is no sex.

However, the distance to the drum 3 could not be corrected,
When the amount of eccentricity at the time of rotation must be obtained in advance, for example, when it is desired to maintain a certain distance without being affected by unevenness in a narrow range of the side surface 3b of the drum 3 as shown in FIG. Is not a perfect circle, but when it is desired to mount the heads 2, 2,... On the locus of the perfect circle, or when the length measuring sensor 6 cannot be mounted at the position viewed by the camera 4 (when the length measuring sensor 6 is As shown in FIG. 19, when a notch (hole) or the like is formed at a place where the distance is to be measured, an accurate distance cannot be measured. Used when correction using an accurate amount of eccentricity is required.

By the way, eccentricity can be roughly divided into those that are caused by a deviation between the center of the rotating body and the object to be rotated, and those that are combined with the original eccentricity and the irregularly and irregularly generated vibration. It is. Furthermore, this is affected by burrs and flaws on the surface of the object to be measured. Therefore, FIGS. 10 and 11
In the case of the drum 3, as shown in FIG.
The side surface 3b may draw a different trajectory. Among them, the effects of burrs and scratches can be removed as high frequency components of the measurement data.

Hereinafter, an outline of a method for removing irregularly and irregularly generated shakes, that is, aperiodic shakes will be described.

First, the aperiodic vibration is divided into two components, and the difference that does not return to the original after one rotation of the rotating body is defined as the first aperiodic vibration. In addition, a shake that occurs irregularly and irregularly due to a physical factor of the rotation axis is defined as a second aperiodic shake.

The first aperiodic runout is used to correct the rising or falling tone of the data in order to enable mathematical calculations and finally to obtain an accurate eccentricity. This is not limited to a simple tilt correction in which only the starting point and the end point are matched, but a tilt correction value that does not cause inconsistency in the shape of the rotating body (drum 3) obtained by calculation is calculated.

The (side surface) shape of the rotating body is a concavo-convex shape obtained by subtracting the amount of eccentricity from the measurement data. If the data indicating the shape of the rotating body is approximated by a straight line, the slope component of the straight line should be zero. However, the approximation straight line is inclined by the first non-periodic fluctuation. The first non-periodic vibration is determined so as to eliminate this tilt component. The above is the preparation for performing the mathematical calculation. It should be noted that the linear approximation refers to connecting a portion having a large amount of distribution data with a straight line.

Next, a missing portion of data which has not been measured due to a hole or the like on the surface of the rotating body is corrected. Incidentally, even though it is not measured, since an indefinite value is actually included, this causes an error in the calculation result.

When differentiating all the data including the data missing portion, high frequency pulses are obtained before and after the hole as shown in FIG. When this differential data is subjected to Fourier expansion and only the primary component is integrated, a temporary eccentric curve is restored. This means that a missing data portion is filled with this curve, which is obtained by Expression 13 described later. The Fourier expansion and the Fourier transform described later are a calculation method for decomposing data repeated at a constant cycle into a composite waveform of a sine wave and a cosine wave. In particular, the 1 Hz component (A · sin (θ))
It can be regarded as the amount of eccentricity of the drum.

By the above processing, data necessary for calculating the amount of eccentricity is obtained, and the amount of eccentricity of the rotating body can be calculated. Here, the final eccentricity is calculated without considering irregular and irregular second aperiodic shakes, but elements including the second aperiodic shakes are discarded during the data loss correction process. There is no problem.

As shown in FIG. 13, the second aperiodic vibration
It is easy to understand when assuming a ball bearing using a metal ball as the bearing. Multiple distorted ball bearings 1
The inner frame 11 rotating through 0, 10,... Rotates while swinging in an arbitrary direction. This is referred to as a second aperiodic vibration.
Also, since the inner or outer diameters of the inner frame 11 and the outer frame 12 that are in contact with the ball bearings 10, 10,... Are not true circles, the inner frame 11 and the outer frame 12 are combined with the slow run-out corresponding to the first non-periodic run-out. Aperiodic vibration occurs. Here, if the 2 to 5 Hz component is extracted using Fourier expansion, the second non-periodic vibration is obtained.

Therefore, from the above, since data = eccentricity + aperiodic runout + shape, the shape can be calculated. By the above processing, the eccentric amount and shape of the rotating body can be accurately obtained, so even if new measurement data is imported,
It is possible to recognize an unknown aperiodic shake. Also,
If the above calculation logic is applied, it is also possible to perform a real tendency management of data fluctuation (application example 2) described later.

Hereinafter, the purpose of the aperiodic shake detection and the calculation formula will be described.

As shown in FIG. 21, when adjusting the position of the head of the drum assembly performed using the production equipment 1,
Regardless of the angle at which the drum 3 is located, the amount of eccentricity must be ascertained and the distance between the drum 3 and the camera 4 must be kept constant.

That is, the head 2 from the side surface 3b of the drum 3
In the case of adjusting the amount of protrusion, the head 2 is first enlarged and displayed by the camera 4 having the lens 8 and the CCD 9. At this time, the protrusion amount of the head 2 is
Since the distance to the camera 4 can be ascertained by the degree of focus of the image, the head 2 is fixed at the specified focus position. Therefore, it is necessary to keep the distance between the camera 4 and the drum 3 constant at any angle division.

When the degree of eccentricity of the drum 3 is checked, it can be calculated by using a Fourier transform as a mathematical processing method. However, since the actual data includes aperiodic fluctuations, the calculation formula using the Fourier transform cannot be applied.

A general Fourier transform calculation method will be described. An arbitrary waveform is composed of a combination of a DC component having no fluctuation and a plurality of frequency components, and an n-frequency component is expressed by the following equation (1). As shown in the Fourier expansion formula, An
Sin (n · θ) + Bn · cos (n · θ), and represents an arbitrary waveform as a composite waveform of a DC component and each frequency component.

[0075]

(Equation 1)

However, if the non-periodic vibration is included, the components of the sin and cos waveforms are shifted in phase. Therefore, the Fourier expansion formula including the non-periodic vibration shown in the following Expression 2 is obtained (0 ° and 360 °). ° data values are shifted).

[0077]

(Equation 2)

Here, there is no method for obtaining the an and bn components (only the aperiodic component) that offset the angle. In addition, it is necessary to acquire data for one cycle of the aperiodic shake, but since the cycle of the aperiodic shake is not determined, the number of data to be taken is not determined.

Therefore, by removing the non-periodic run-out from the measurement data for one rotation of the drum 3, it is possible to use the above-described equation (1) which is the general equation of the Fourier transform, although it is the measured eccentricity in the orbit. Can be done.

The outline of the aperiodic shake detection in the present invention is as follows. That is, the eccentricity and the shape of the drum 3 are obtained as temporary values by a general formula. However, since the aperiodic run-out is not taken into account, inconsistency occurs in the drum shape values. Therefore, conditions are set to cancel this contradiction. Then, the inclination of the graph is corrected using this condition. Next, after the above processing, if any part of the data that could not be measured exists, it is restored using a virtual curve. Then, abnormal values that occur irregularly in the measured revolutions are found and removed. Finally, from the ideal eccentric curve, an accurate eccentric amount and the shape of the drum 3 are obtained.

FIG. 22 shows the flow of the processing procedure based on the above. The outline of each processing flow is described below.

In processing 1 and processing 2, data [raw data] obtained by measuring the distance to the side surface 3b by rotating the drum 3 by the length measuring sensor 6 is acquired, and D is obtained from the data.
This is a process for obtaining data [data 1] from which the C component, noise, and the like have been removed.

Processes 3 to 5 are processes for obtaining a provisional eccentric curve from the above [Data 1].
The difference [difference 1] between the data of 360 ° and 360 ° was obtained, and the processing 4
In order to eliminate the difference between the data of 0 ° and 360 °, [Data 2] is obtained by performing a simple linear correction, and in processing 5, a temporary eccentric curve is calculated as [Data 3].

Processes 6 and 7 are processes for obtaining the temporary shape of the drum 3. In process 6, the temporary shape is obtained as [data 4]. In process 7, the [data 4] is linearly approximated. Things.

The difference [Difference 2] between 0 ° and 360 ° in [Data 4] linearly approximated in Process 7 prior to Process 8;
That is, the gradient component is obtained, and if there is a difference, the process proceeds to processing 8;

Processes 8 and 9 are processes for obtaining the first component of the non-periodic vibration included in the temporary shape of the drum 3 obtained by the process 6. In process 8, when there is a gradient component (when [difference 2] is present in [data 4]), the [difference 1] is corrected, and the process returns to process 5. In the process 9, when there is no [difference 2], a correction value when the tilt component of the drum shape is eliminated is taken out as the first non-periodic vibration [difference 4].

Process 10 is a process for obtaining [data 6] by linearly correcting [data 1] in order to eliminate the first aperiodic vibration. The linear correction means that data is gradually corrected from the base point so that the end point and the start point have the same value. As you approach the end point, the amount of correction increases.

Here, it is checked whether or not there is a data missing portion in the middle of the measurement data. If there is a data missing portion, the process 11 is performed. If there is no data missing portion, the process 11 is not performed. Is passed as it is to the processing 12 as [data 9].

In the process 11, the data missing portion is filled with a virtual curve to perform restoration, and this is set as [data 9].

The process 12 is a process for obtaining a correct eccentricity correction value. The amount of eccentricity is extracted from the [data 9] obtained from the above process 10 or the process 11, and is obtained by the [data 1].
0].

Processes 13 and 14 are processes for obtaining the non-periodic shake, and 2 to 5 from [Data 9].
The component of Hz is taken out, this is taken as [Data 11] (second aperiodic vibration), and [Data 6] which is linearly corrected by the above [Difference 4] (First aperiodic vibration) is [Data 11].
And this is used as [Data 12] (aperiodic fluctuation).

Process 15 is a process for obtaining the drum shape. [Data 1] is obtained by subtracting [Data 10] (the amount of eccentricity) and [Data 12] (Aperiodic vibration) from [Data 1].
3] (drum shape).

The process 16 is a process for predicting the aperiodic vibration. When a series of totalization by the above processes 1 to 15 is completed, the aperiodic vibration in an arbitrary round can be calculated. Become.

The process 17 is a process for obtaining a final eccentricity correction amount.

Hereinafter, the processes 1 to 17 will be described in detail.

In process 1, the drum 3 is rotated and its rotation angle (hereinafter, simply referred to as “angle”) is changed from 0 ° to 360 °.
The raw data (the distance to the side surface 3b of the drum 3) is fetched by the length measuring sensor 6 at a constant interval up to. The required data is in units of 1 °, but in order to average the data later, a large amount of data is taken in at as small an interval as possible before and after the required angle. FIG. 23 shows an example of raw data obtained by this processing.

Process 2 is a process of calculating a moving average of the raw data and leveling the raw data to obtain [data 1]. In this calculation, data values in units of 1 ° are used. However, there is a possibility that components such as noise may be included when capturing the raw data. Therefore, a unique weighting method is applied to the raw data to calculate the average at each angle. I take the. That is, as shown in FIG. 24, the raw data value at the target angle of 1.00 ° is 10
0%, and the raw data before and after that (0.99 ° and 1.01
°) 17% and its neighbors (0.98 ° and 1.02 °) 4%.
The average of these seven data points is taken with the weight of 1% next to it (0.97 ° and 1.03 °). Similarly, this process is repeated for each subsequent angle (2.00 °, 3.00 °,..., 360 °). In addition, the method of determining the ratio of each of the above data is such that if the data value of the target angle is abnormal, it is canceled out, and an empirical ratio that is considered to be unaffected by the surrounding data values is set It is.

Process 3 is a process for measuring the length of [Data 1].
Data when the base point is 0 ° and the end point 3 when the drum 3 has made one round
This is a process of comparing the data at the time of 60 °, finding the difference, and setting this as [difference 1] (see FIG. 24).

In the process 4, as shown in FIG. 25, [Data 2] is obtained by linearly correcting [Data 1] so as to eliminate the [Difference 1].

That is, as shown in FIGS. 26 and 27, if the base point and the end point of the graph showing the distance data to the side surface of the drum 3 due to the rotation deviate from each other, the mathematical calculation by the Fourier transform performed later will be omitted. Problems arise. Small fluctuations in the data can be removed as high-frequency components, but if the data gradually increases or decreases or if the final value deviates from the base point value, an error occurs in the calculation (the reason is that the processing 8 described later will be described). Described in). However, the correction in this process 4 is
The simple correction is simply applied to the entire data so that 0 ° and 360 ° have the same value. It should be noted that the correction in the process 4 has no mathematical significance, but rather has a meaning in performing a moderate correction. Drum 3 in later processing
However, if the correction in the process 4 is incorrect, the shape of the drum 3 will be inconsistent. The method of correcting the data at the angle θ here is to obtain an equation for changing the inclination of the straight line of Expression 3 below at θ °.

[0101]

(Equation 3)

Process 5 is for obtaining a temporary eccentric curve of the drum 3. That is, a 1 Hz component is extracted from [Data 2] by Fourier transform, and a provisional eccentric curve is obtained [Data 3]. By the way, the 2 to 5 Hz component becomes a part of what constitutes the non-periodic vibration. The reason for the boundary of 5 Hz is that five peaks of non-periodic vibrations, which are considered to be one of the non-periodic vibrations at the experimental stage, affect the amount of eccentricity during one rotation of the drum. (See FIGS. 30 and 31).

The process 6 determines the difference between the above [Data 3] and [Data 1] by using the temporary shape of the drum 3 [Temporary data 4].
It is taken out as. FIG. 32 is a graph of the temporary shape [temporary data 4] of the drum 3. Note that this [temporary data 4] includes errors and noises due to aperiodic vibrations in addition to the shape of the drum 3.

In the process 7, the [temporary data 4] is set to a minimum of 2
A straight line is approximated by a multiplication method or the like, and Equation 4 which is an equation of the following straight line is obtained. Incidentally, when the dispersed data is plotted in a two-dimensional graph of the X-axis and the Y-axis, if there is a correlation between the amount of displacement in the X direction and the amount of displacement in the Y direction, a straight line connecting the mode values can be drawn. , The least squares method is one of the calculation methods that can obtain the straight line.
The mode value is a value in which the data is distributed more in the decomposed data.

[0105]

(Equation 4)

When the above equation 4 is graphed, the graph is inclined. That is, when Equation 4 rises and the value of the θ-axis (the rotation angle of the drum 3) increases, the value of the y-axis (= S0 (θ)) also increases, or when the value of the θ-axis increases, the y-axis increases. The value decreases. This means that the component a in Equation 4 is not 0. However, Equation 4 gives
Since the shape of the drum 3 is approximated by a straight line, the graph representing this straight line should not be inclined.

Here, the unevenness of the shape of the drum 3 is several μm.
Therefore, the treatment is performed under the condition that there is no unevenness which affects the component a of the equation (4). By the way, it can be said that the coefficient a component of Equation 4 represents the degree of inclination of a straight line when the shape of the drum 3 is approximated by a straight line (see FIG. 33).

This is because, as shown in FIG. 34, when the [Data 2] is acquired, a non-periodic shake (a shift from the base point 0 °) still remains at the end point of 360 °. Since the non-periodic fluctuation changes the original data value into an abnormal value like noise, it is assumed that 0 which should match.
This is because the difference between the data values of (° and 360 °) (data values at the same position) has occurred, and the chain of data has failed.

As shown in FIG. 33, the fact that the drum shape is inclined means that the calculated eccentricity curve corresponds to the mode value of [Data 1] (the most data value among the data fluctuations).
You have not passed through. That is, the calculated amount of eccentricity (eccentric curve) is wrong.

In the process 8, first, a gradient component of the entire data is obtained. That is, from Equation 4, θ = 0.
S0 when (initial value) and when θ = 360 (final value)
Find the difference in (θ) values. Since Equation 4 is an approximate straight line of the shape of the drum 3, S0 of the base point (0 °) and the end point (360 °)
(Θ) values must match. If the tentative shape of the drum 3 includes an inclination component (a component) (when a is not 0), it means that there was an error in calculating the shape of the tentative drum 3 [temporary data 4]. Become.

Here, [data 2] is created by adding the difference [temporary difference 2] between the base point and the end point of the slope line to the difference [difference 1] between the raw data of the start point and the end point of [data 1]. It is not enough to eliminate the inclination of the shape of the drum 3 (contradiction in shape). As shown in FIG. 26, if the last data is an abnormal value, the contradiction of the shape of the drum 3 [temporary data 4] represented by Expression 4 will not disappear even if the calculation is performed with some correction. .

Therefore, in this case, as shown in FIG. 35, [Temporary difference 2] is corrected to [Difference 1], and data that is greatly deviated from the unevenness of the shape is removed. Returning to the process 5, the process is repeated from the point of obtaining the temporary eccentric curve again. By repeating this correction until the a component of Equation 4 becomes 0, inconsistency in the shape of the drum 3 is eliminated.

The processing in the above processing 8 is specifically described as follows. That is, [difference 1] is corrected by [temporary difference 2], the shape of the drum 3 is obtained again, and its inclination component is examined. As a result, if an a component that is not a problem is obtained from the difference between 0 ° and 360 °, the drum 3
The difference between the shape data [temporary data 4 = data 4] and the length measurement data [data 1] is determined as the difference due to the aperiod at 360 °, and as the slope of the measurement data due to the aperiodic deflection [difference 2]. Incidentally, in the experiment, a difference at 360 ° of less than 0.1 μm (a <0.00028) is regarded as a difference that does not cause a problem.

By the way, f (360) representing [data 1]
°) to the value of f (0 °), the drum 3
Tried to eliminate the tilt component of the shape of f (360 °)
Is an abnormal value, the calculated eccentric curve is rather greatly inclined, and a and b in Expression 4 of the shape of the drum 3
This greatly affects the components (FIGS. 28 and 29).
And FIG. 34). In this case, the processing of processing 5 to processing 8 is repeated using an eccentric curve (drum shape greatly shifted) including a large error.

In particular, if there are abnormal values near the data base point and end point, it takes time to calculate a true eccentric curve.
To avoid that, start by removing this outlier. In the following, a process for removing the abnormal value (suddenly changed data) in the process 8 will be described.

When the displacement of adjacent data at the two locations near the starting point and the ending point of [Data 1] exceeds the maximum displacement of Equation 5 described below, one of the data is regarded as an abnormal value. Become. Processing 8 is performed after removing data near the start point or end point where adjacent data does not fit within the maximum displacement amount.

Here, as shown in FIG. 36, when the eccentricity is replaced by a sin curve, the theoretical maximum displacement is sin
Displacement (differential value) at (0 °) and sin (180 °)
Is maximized. Also, if PP = maximum value−minimum value,
The maximum displacement is represented by the following equation (5), which indicates the maximum change point of the sine wave.

[0118]

(Equation 5)

Here, the abnormal value may be sudden data in which the measured distance changes in the plus direction due to a flaw or the like, and sudden data in the minus direction due to tension or dust.

When the entire data is moving in the minus direction due to the aperiodic fluctuation, as shown in FIG.
When sudden positive data is entered near 360 °, the data at 0 ° and 360 ° have the same apparent value, and it may appear that there is no aperiodic vibration.

In this case, as shown in FIGS. 28 and 29, the eccentric curve obtained by performing the Fourier transform as it is has an end point and a start point, but is an approximate curve (curve) far from the measured data in the middle. (Approximately drawn curve). The point to be noted when performing curve approximation by Fourier transform is that abnormal values existing in the middle of continuous data are absorbed to some extent, but abnormal values near the starting point and end point are greatly affected. is there. Note that the curve approximation means connecting a portion having a large value from distributed data with a smooth curve.

Even if sine waves having the same phase are combined, the values of 0 °, 180 °, and 360 ° do not change as shown in FIG. That is, if the data at the start point and the end point are different, the high-frequency component (sin (n · θ) or the sin wave (sin
Since the component of (θ + a)) is complicated, it takes time to calculate this component, and it is difficult to distinguish and recognize the aperiodic component from the calculation result.

Further, if many high-frequency components are synthesized to reproduce a complicated waveform, the low-frequency components desired to be obtained as the amount of eccentricity are reduced. It will not be possible to remove the data, and it will be difficult to restore the missing data.

Process 9 is a process for obtaining the first non-periodic shake. That is, when [Data 5] is calculated by linearly approximating the drum shape again from the [Data 4] that does not include a tilt component in the drum shape, a ≒ 0 is obtained from the above-mentioned Expression 4, and thus the following DC of the drum shape is obtained. Equation 6 representing the components is obtained.

[0125]

(Equation 6)

The b component (DC component) in Equation 6 should be b ≒ 0 when the shape data [data 4] is calculated. However, since the processing is not repeated until the a (slope) component becomes completely zero in the processing 8, b =
It is not 0. Further, in order to set a = 0, the processing must be repeated many times, and the efficiency is poor. Therefore, the b component that will be generated here is also added to the correction value (if the b component is ignored, , Equation 4 will not pass through the position of the center of gravity of [Data 4]).

To correct [difference 1], it is sufficient to correct the increment of θ from 0 to 360 in equation (4). Therefore, as shown in FIG. A correction value [difference 3] of [difference 1] for eliminating inconsistency in the drum shape is obtained by Expression 7 which is a calculation formula of a correction value for performing the correction.

[0128]

(Equation 7)

[Difference 1], which is a difference between the data values of 0 ° and 360 ° in [Data 1], and [Difference 3], which is a correction value calculated to eliminate inconsistency in the drum shape, are added.
This is defined as the first component [difference 4] of the aperiodic vibration.

The method of correcting [Difference 1] is to apply a linear correction in the same manner as in the above-mentioned formula 3, and use this as the first non-periodic vibration correction value. FIG. 34 shows a comparison between [Data 1] and data obtained by correcting the first non-periodic shake.

The following summarizes the processings 8 and 9 described above.

That is, the data is gradually increased or decreased so that the numerical values of 0 ° and 360 ° of [Data 1] match,
[Difference 1] is eliminated. However, in this state, when the drum shape is obtained, a contradiction occurs in which the starting point and the ending point of the shape waveform do not coincide. Therefore, [Difference 1] is corrected and [Difference 4] is corrected so as to eliminate this contradiction. calculate. Then, [Data 1] is gradually increased or decreased using [Difference 4], and the cycle of the first non-periodic vibration is extracted.

Process 10 is a process for linearly correcting [data 1] so as to eliminate the first aperiodic component [difference 4] in order to obtain data from which the first aperiodic vibration has been removed. The data obtained by this is [Data 6].

As shown in FIG. 38, [difference 4] is the difference at the end point of 360 °, so the correction value at the angle θ ° is
The equation for removing the first non-periodic vibration shown in Expression 8 is as follows.

[0135]

(Equation 8)

The process 11 is a process for restoring a data missing portion (filling with a virtual curve).

The purpose of the present invention is to examine the amount of eccentricity when the drum 3 rotates, provided that data can be obtained from all measurement points. If this condition is satisfied, the calculation of the Fourier transform is performed. Expressions can be applied. However, as shown in FIG. 19, the side surface 3b of the drum 3 has a notch (a notch 3c for arranging the heads 2, 2,...) Or a groove (a groove for obtaining an air film effect, a groove due to a scratch, etc.). )
Is notched, and in these notches and grooves, a part of the data cannot be measured or the data becomes incorrect. Therefore, the eccentricity must be calculated after restoring the data missing part. Processing 1
1 is a process for this.

The outline of the calculation method for restoring the data missing part is as follows. That is, when the waveform including the data missing portion is differentiated, the data missing portion can be extracted as a high frequency. Only the first-order component is extracted from this differential waveform,
By integrating this, a waveform having no data missing part can be restored. When a DC component is added to this integrated waveform, a restored waveform in which data missing portions of the original waveform are filled can be created. However, if a high-frequency component is added to the first-order component of the differentiated waveform, and the data including the data missing portion of the original waveform are integrated, a composite waveform of eccentricity and drum shape is likely to be reproduced. If the part is not filled, the drum shape data cannot be reproduced.

Specifically, prior to the process 11, it is checked whether or not a portion that could not be measured as data (missing data) is included. If there are grooves, small scratches, dust, or the like on the side surface of the drum 3 and continuous measurement data includes a depressed portion or a protruding portion, a small data missing portion will be present. When the arrangement portion of the heads 2, 2,... Is measured, the shape data is entered as shown in FIG. Therefore, if the amount of eccentricity is calculated including these indefinite data, an incorrect eccentricity curve will be calculated.

Therefore, in process 11, (1) f
Differentiating (θ), (2) Extracting the primary component of differentiation, (3)
(4) Find the maximum and minimum values of f (θ), (5) Search for missing data, (6) Offset the integrated value, (7) Find the missing data (8) Confirm the authenticity of the restoration curve by filling in the integral value and setting it as g0 (θ). Here, if the deviation between g0 (θ) and f (θ) is large, the processing is performed in the order of starting from (1) with g0 (θ) as f (θ).

(1) Differentiation of f (θ) is to differentiate [Data 6] from which the first non-periodic vibration has been removed as original data f (θ). Since the data is represented by the above-mentioned formula 1 by Fourier transform, f (θ)
Is the derivative of equation 1, which is as shown in equation 9 below and in FIG.

[0142]

(Equation 9)

Here, according to the sampling theorem, the frequency component (n = α) for which the summation of Expression 9 is performed is limited to a frequency component equal to or less than half the number of samples. Since there are 360 data,
The frequency component α ≦ 180. Note that the sampling theorem refers to the number of samples required for the frequency of data. For example, when sampling a 9 Hz signal 10 times,
When 10 samples are taken at the timing of the 0.9 cycle, the 1.8 cycle,..., The 9 cycle, the curve connecting them is exactly 1
I get the same curve as the Hz signal. To avoid this, sampling at least twice (18 or more) the frequency may be performed.

(2) Extracting the primary component of the differential means extracting the primary component from the differential waveform equation (9).
The curve obtained by Expression 9 reproduces the displacement of the measured amount at each angle of the original data f (θ). As a result, a portion corresponding to the groove comes out as a large displacement as shown in FIG.

That is, the distance to the groove is measured as infinity (this results in the loss of data), the displacement increases instantaneously, and an irregular waveform is drawn by the head existing before the end of the groove, or , And then returns to the original smooth waveform.

Here, when the original eccentric amount sinθ is differentiated and its displacement is examined, a sine wave differential expression shown in the following Expression 10 is obtained.

[0147]

(Equation 10)

Further, among the components constituting the displacement f '(θ), the primary component derives an eccentric curve. Therefore, from Expression 9 representing the differential waveform, f ′ (θ) shown in Expression 11 below is obtained.
Take out the primary component of

[0149]

[Equation 11]

On f ′ (θ), grooves, burrs, and the like are represented by pulses (high-frequency components). Therefore, by extracting only the primary components, these extra data can be removed. .

(3) Integrating the first-order component means f ′ (θ)
Is to integrate Equation 11, which is the primary component of Equation 11 is obtained by differentiating only the amount of eccentricity, so that the original eccentricity curve can be restored by performing the back calculation. Eccentric curve F0 (θ) = ∫ ｛f '
Since the primary component of (θ) is｝ dθ, an eccentric curve shown in the following Expression 12 is obtained.

[0152]

(Equation 12)

(4) To find the maximum and minimum values of f (θ) means to calculate the maximum and minimum values of the original data f (θ),
To examine it. The curve obtained by the above equation (12) has a different DC component, but is similar to data in which the groove portion of the original data f (θ) is filled, and has the same phase as the PP value (maximum and minimum range). Becomes The original data f (θ)
Is represented by [the maximum value of the expression 12] − [the minimum value of the expression 12].

(5) Searching for a missing data portion means finding a missing data portion. The original data f (θ) is
Comparing with the eccentricity curve F (θ) restored by the above equation 12, it can be seen that there is no coincidence between the two data due to the influence of the groove, burr, drum shape and the like.

Therefore, a process for clearly separating the groove or burr portion from the portion representing the drum shape is performed by the following method.

That is: 1. The displacement of θ ° to (θ + 1) ° of the eccentric curve obtained by equation 12 is obtained. 2. The displacement of 1 ° at this angle is the maximum value of the irregularity standard in the drum manufacturing. 2 μm is added to obtain the maximum displacement amount. 3. If the original data f (θ) is within the maximum displacement amount, it is determined that there is no groove or burr. 4. Next, (θ + 1) ° to (θ) θ + 2) ° displacement is obtained,
Repeat steps 2 and 3 until 360 ° is reached. 5. If there is a data change that exceeds the maximum displacement,
Judge that there is a groove or a burr there. 6. If there is a groove or a burr, check the displacement next to
A process of repeating the comparison of the above 2 and 3 with a displacement amount of 2 ° up to (θ + 2) ° is performed. By keeping the reference point when checking the displacement amount, the displacement amount compared with the data of the data missing part such as the groove becomes an abnormal value, and the point where this returns to normal indicates the end of the data missing part. Then, when the displacement returns to the normal amount, the reference point is changed, and the processes (2) and (3) are repeated (see FIG. 42).

(6) To offset the integrated value means to obtain the DC component of the integrated value of the restored eccentric curve equation (12). In the process (5), the average value of the difference between the value of the eccentric curve F0 (θ) and the value of the original data f (θ) is obtained in a range excluding a portion where it is determined that there is no groove or burr. Formula 1
Since the value F0 (θ) obtained in step 2 is a curve including only the amount of eccentricity excluding the DC component, the average value of the difference is calculated using the temporary DC component (dc
As a result, a curve close to the eccentric curve F (θ) of the original data f (θ) is reproduced. That is, the eccentric curve F (θ) = [Equation 12] + DC component, which is represented by the following Equation 13.

[0158]

(Equation 13)

Note that f (θ) is a raw data curve including a data missing portion, F0 (θ) is a temporary eccentric curve (only AC component),
The dc component is the average (F0) of the difference between F0 (θ) and f (θ).
(Tentative DC component of (θ)) and F (θ) are eccentric curves (including DC components).

(7) Filling the missing data with an integral value means
As shown in FIG. 43, in the process (5), a curve g0 (θ) obtained by replacing the portion of the original data f (θ) determined to be a data missing portion with the value of F0 (θ) in the above equation 13 is used. That is to ask. A curve g0 (θ) in which the data missing part of f (θ) is filled with a temporary eccentric curve is defined as [temporary data 9].

The data of f (θ) is a waveform including components other than the eccentric component. However, since the purpose of this processing is to obtain the eccentricity, here, dare to include the data missing portion up to the shape component. Instead of filling in a waveform resembling (θ), only the eccentric component is used.

(8) To confirm the authenticity of the restoration curve,
This means that the authenticity of F0 (θ) obtained by integrating the primary components of f ′ (θ) is determined. A curve g0 (θ) in which a data missing part of f (θ) is filled with F0 (θ) data.
The waveform must be restored so that the boundary of the data missing part of the original data f (θ) is smoothly connected.

In order to determine the authenticity (there is no deviation in data) of the restoration curve F0 (θ), F0 (θ) and f0 (θ)
(Θ), the distribution of this difference is linearly approximated,
Obtain a tilt component and an offset component. If the inclination component and the offset component are 0, it can be determined that g0 (θ) matches f (θ).

Then, the difference between the eccentric curve F0 (θ) and the [temporary data 9] g0 (θ) which fills in the data missing part of f (θ) is obtained (comparison of data other than the data missing part). Is [data 7]. [Data 7] is linearly approximated to calculate [Data 8]. The [Data 8] is represented by the following Expression 14 obtained by linearly approximating the difference between the above Expression 13 and [Temporary Data 9].

[0165]

[Equation 14]

When both the slope component c and the offset component d in the above equation 14 are not 0, g0 (θ) is calculated as f0
As (θ), the processing from (1) onward is repeated.

It should be noted that the above equation (14) gives S1 (θ) = g0
It is a linear equation obtained by linearly approximating the difference between (θ) and F0 (θ), and is a tentative curve obtained by restoring a data missing portion of g0 (θ) = f (θ).

In the process (8), as shown in FIG. 43, the reason why the repetition process is performed until the equation (14) becomes 0 is that, as shown in FIG. 44 and FIG. Then, since the differential value in the above (1) is affected by the data missing part, the correct restoration curve F (θ)
Is not obtained. Also, when the data of the data missing portion has abnormally far values, an error occurs in the approximate curve to be obtained.

When the width of the data missing portion is wide, the reproduced waveform obtained by the Fourier transform calculation reproduces the data missing portion by synthesizing a plurality of high-frequency components, so that the synthesis ratio of the primary component serving as the basic waveform decreases. Will be done. Therefore, P of the eccentric curve F0 (θ) obtained by the Fourier transform
The P value becomes smaller. However, by repeating the processing of (1) to (8) so that Equation 14 becomes 0,
It can approach the true F (θ).

However, if the range of the data missing part is widened,
Many repetitions are required until Equation 14 converges to zero. When the slope component c of Expression 14 becomes 0, g0
(Θ) is a restoration curve in which the data missing portion is correctly corrected. This [g (θ) = data missing portion corrected data] is referred to as [data 9].

Note that g (θ) [data 9] indicates a correct eccentric curve [F (θ) + DC component] for the data missing part of f (θ).
This is the curve filled with.

By the process 12, a correct eccentricity correction value when the drum 3 rotates is obtained. That is, in the process 12, FIG.
As shown in FIG. 7, only the eccentric component (primary component) is extracted from the above [Data 9] by Fourier transform, and the correct eccentric amount [Data 10] of the drum 3 is obtained.

In the process 13, the [Data 9] (g (θ)) is extracted to the 2 to 5 Hz component by the Fourier transform, and is set as the second component [Data 11] of the non-periodic vibration.

When the main shaft (drum 3) makes one rotation, as shown in FIG. 47, the vibration is smaller than the eccentricity (1 Hz).
In addition, a vibration larger than the shape (20 to 30 Hz) may occur. This fluctuation occurs irregularly and has no periodicity.

Strictly speaking, the first non-periodic run-out obtained in the process 9 is more correctly corrected by obtaining data for obtaining the eccentric curve F (θ) than by obtaining the inclination of data. This is because the non-periodic component of 2 to 20 Hz which could not be removed by the inclination correction of the data (correction of the first non-periodic shake) still remains. Further, the magnitude of the first aperiodic fluctuation varies depending on the magnitude of the indeterminate second aperiodic vibration. The final aperiodic vibration is an irregular and irregular vibration generated due to these two factors.

The amount of eccentricity F (θ) is obtained from [Data 9] corrected by the first non-periodic vibration which may be wrong, but f (θ) is differentiated in the above process 11 (2). At this point, there is no problem because the second aperiodic runout has already been removed. The second non-periodic vibration obtained here is calculated in order to grasp a data component generated irregularly. This is because when the drum shape is obtained in the subsequent process 15 or when the aperiodic runout is predicted in the process 16,
If the data contains irregular and indeterminate components,
This is because these values cannot be calculated.

Incidentally, the second non-periodic run-out in an experimental facility using a ball bearing for the bearing structure of the main shaft (drum) had a large undulation of about five times (see FIG. 48).

The magnitude and the number of undulations of the aperiodic vibration vary depending on the circulation.

The process 14 is a process for combining the first aperiodic vibration and the second aperiodic vibration into an aperiodic vibration.
As shown in the figure, aperiodic vibration [data 12] = linear correction of [difference 4] +
[Data 11]

Here, since the unevenness of the drum shape is guaranteed to be suppressed within a range of several μm, the processing is performed under the condition that there is no shape data that affects the calculated aperiodic runout. .

By the way, for example, when measuring a shape having irregularities of 2 to 20 Hz, this shape component is processed as aperiodic vibration. If a waveform having the same tendency is observed even when the measurement is performed while changing the circuit, it is possible that unevenness due to the drum shape exists. If the tendency is grasped, it becomes possible to remove the non-periodic fluctuation value due to the drum shape.

In the process 15, as shown in FIG. 50, the drum shape is changed to the drum shape [data 13] = [raw data] -eccentricity-aperiodic vibration = [data 1]-[data 10]-[data 12] It is what is sought.

However, the above [Data 13] includes a noise component which is not removed even by the moving average. Therefore, a small burr-like shape cannot be detected because it cannot be distinguished from noise. If the bearing has any damage, it will be recognized as irregular vibration during rotation. Also in this case, it is possible to make the waveform of the repeated tendency into the shape of the drum by measuring while changing the circuit.

Step 16 predicts the occurrence of aperiodic run-out on the basis of the drum shape obtained in step 15 described above.

This prediction is to grasp the amount of aperiodic vibration which is a vibration that occurs irregularly and irregularly.
51, the correction value in the X direction to keep the distance between the drum 3 and the camera 4 constant, and the correction value in the Y direction to indicate the center position of the drum 3 in the head position adjustment equipment shown in FIG. Required.

In order to calculate the aperiodic run-out, the processing 1
Through processing 15 is performed. Next, the actual work is started using this result. The aperiodic run-out occurring at this time is different from the aperiodic run-out calculated in advance. Therefore, it is necessary to know the current value of the aperiodic vibration correction at the time of starting the actual work.

The non-periodic shaking in the X direction can be detected by the length measuring sensor 6, as shown in FIG.
Since the eccentric amount and the shape at the current rotation angle of the drum 3 have already been calculated, the calculated distance and the length measurement sensor 6 are used.
The difference from the currently measured distance results in a sudden aperiodic vibration occurring in the X direction.

The non-periodic vibration in the Y direction can be detected by the length measuring sensor 7 as shown in FIG.
The difference between the calculated distance in 90-degree offset angle division and the distance currently measured by the length measuring sensor 7 results in a sudden aperiodic vibration occurring in the Y direction.

Here, the length measuring sensor 7 is used only for detecting the non-periodic vibration, but instead of calculating the eccentric amount described above, the eccentric amount in the Y direction is derived by the length measuring sensor 7. Can not.

Since the side surface 3b of the drum 3 is a curved surface,
The movement in the Y direction slightly fluctuates in the distance in the X direction, but this is not a problem and is ignored. The basis is shown below.

That is, when a shift occurs in the Y direction, a change in the distance X to the drum with respect to the shift is calculated using the following equation (15) for calculating the distance to the side of the cylinder (drum). , The distance in the X direction. Where y is Y
Direction deviation, r is the radius of the upper surface of the cylinder (radius of the drum), Δ
Let x be a distance correction amount in the X direction due to a shift in the Y direction.

[0192]

(Equation 15)

In the above equation (15), for example, when the radius of the drum is 10 mm, y = 30 μm and Δx = 0.045
μm, so there is no problem that the effective diameter is less than 0.1 μm.

When the length measuring sensor 6 is installed on the right or left side of the camera 4 as shown in FIG. 52, the above-mentioned fluctuation amount becomes large.

For example, when the diameter of the drum 3 is 20 mm, the heads 2, 2,.
If the length measuring sensor 6 is to be installed on a parallel plane having the same height as that of the camera 4, the measuring point is placed at a position 2 mm to the right or left of the line of sight of the camera 4. In this case, when the influence of −30 μm shake in the Y-axis direction in the X direction is applied to Equation 15 above, A1 = 10000−10000 · cos (sin ⁻¹ (2000
/10000)=202.041 A2 = 10000-10000 · cos (sin ⁻¹ (2030
/10000)=208.213 Error in X direction due to shake in Y direction: Δx = A2-A1 =
6.172 μm, and the aperiodic deflection in the X direction can be calculated by the following formula: measurement value of length measurement sensor 6−eccentricity−shape value−6.172.

When the length measuring sensor 6 is installed above or below the line of sight of the camera 4, the distance to the measurement point of the line of sight of the camera 4 cannot be measured. This is because the side surface 3b of the drum 3 is not vertical but is inclined as shown in FIG. There is no effect on the eccentricity calculated by extracting only the AC component.

However, if the height of the measurement point is different between when the amount of eccentricity is measured and when the head 2 is projected by the camera 4, the aperiodic runout is predicted because the drum shape is different from the previously checked drum shape. It is not possible.

In the process 17, the final eccentricity correction amount is obtained. However, the correction of the eccentricity amount in the actual work is performed by correcting the eccentricity amount including the aperiodic run-out in the calculated eccentricity amount. Will go on. Since the amount of eccentricity that has swung right and left on the screen can be recognized only by the calculation result of the eccentricity correction, the correct eccentricity state is examined in consideration of the aperiodic shake amount, and the non- The period deviation is detected separately to derive an eccentricity correction value according to the current situation.

As described above, when the aperiodic run-out correction method is used for the operation of adjusting the position of the magnetic head during the manufacture of the drum assembly and the production equipment used for the same, some holes or the like on the side surface of the drum in the measurement range are obtained. Even if there is, the influence of the eccentricity can be corrected without being affected by this, so that the measurement range is hardly restricted, and the drum assembly can be manufactured with specifications close to the design instructions. In addition, since not only holes but also abnormal portions of data can be detected, an incorrect correction value due to this can be eliminated,
It can also be used for detection when an abnormality occurs.

Further, since the shape value of the drum can be accurately detected, for example, when a laser sensor is used as the length measuring sensor, even the surface roughness of the drum side surface is detected.
In such a case, it is possible to calculate not the average value of the fluctuation range but the curved line connecting the outermost protrusions around which the magnetic tape is wound as the drum side surface.

Furthermore, old production equipment used for adjusting the position of the magnetic head during the production of the drum assembly uses a bearing such as a ball bearing that causes aperiodic run-out. Is wasteful, and it is also possible to extract old equipment corresponding to this.

Finally, an application example of the processing in the aperiodic shake detection method described above will be described.

As an application example 1, as shown in FIGS. 53 to 57, when there is data missing due to a groove, the shape of the missing part is predicted from the entire flow, and a virtual shape without a groove is restored. It is made possible.

That is, in the production of the drum assembly shown in FIG. 1, the protrusion reference surface for adjusting the protrusion amount of the heads 2, 2,... However, if there is no appropriate reference surface around the head 2, for example, the side surface 3 closest to the head 2
If there is a groove or the like in b, a reference point for measurement cannot be obtained, and adjustment of the protrusion amount cannot be performed. Therefore, by assuming a virtual surface at a portion corresponding to the groove or the like, it becomes possible to adjust the amount of protrusion.

The method of obtaining the virtual plane is as follows.

As shown in FIG. 53 and FIG. 54, the data missing portion is filled with an eccentric curve in the above-mentioned processes 1 to 11, and an eccentric component is removed from this data as shown in FIG. And

The data immediately before the missing portion is extracted by the same amount as the width of the missing portion, and the frequency component of the first half waveform of this data is examined. This frequency component is obtained using a Fourier transform, and a high-order component is obtained to such an extent that the extracted data can be restored by calculation. Then, the waveform reproduced by the above calculation is used as the first half pattern.

Similarly, the data immediately after the missing portion is extracted by the same amount as the width of the missing portion, and the frequency component of the latter half waveform of this data is examined. This frequency component is obtained using a Fourier transform, and a high-order component is obtained to such an extent that the extracted data can be restored by calculation. Then, the waveform reproduced by the above calculation is used as the latter half pattern.

Then, as shown in FIG. 56, the shape of the missing portion is predicted and restored from the data obtained by combining the former half pattern and the latter half pattern. At the beginning of the missing part, the composite value is obtained by increasing the ratio of the first half pattern and increasing the ratio of the second half pattern toward the end of the missing part.

Finally, as shown in FIG. 57, by adding an eccentric component to the restored shape waveform, the raw data without any missing portion can be restored.

Further, only the calculation logic in the present invention can be used for simple data analysis and the like.
That is, as shown in FIG. 58, the application example 2 is used as a data tendency management tool, and shows a transition of daily data that changes every moment with a smooth curve to make it easy to understand the tendency of the change. is there. This method is used when it is not determined how the data will fluctuate, the difference between adjacent data is large, and the tendency management is difficult.

Although the approximate curve is obtained by using the Fourier transform, the fluctuation of the rising or falling tone of the data is not linearly trended but the future trend is grasped while grasping the daily fluctuation. The point is where you can read. It is not an approximate straight line often used in the XY correlation diagram, for example, a graph representing the fluctuation of the stock market, but a daily curve connected by a smooth curve. The curve can be represented by a low frequency component Δdθ of the curve = Δf ′ (θ).

Incidentally, the specific shapes and structures of the respective parts shown in the above-described embodiments are merely examples for embodying the present invention. The scope should not be construed as limiting.

[0214]

As described above, the method of obtaining the amount of eccentricity of the rotating body according to the present invention measures the distance to the rotating body at predetermined angles while rotating the rotating body once, and obtains the initial value obtained by the measurement. A temporary eccentric curve of the rotating body is obtained from the data, and temporary rotating body shape data is obtained from the temporary eccentric curve.From the temporary rotating body shape data, a part inconsistent in shape of the rotating body is obtained. Correction data in which the value of the starting point portion and the value of the ending point portion of the data are set to be the same value so as to eliminate the portion inconsistent in shape is obtained, so that an accurate eccentric amount can be obtained. .

According to the second aspect of the present invention, when there is a missing portion in the initial data, the missing portion is filled with a virtual curve to create interpolation data, and the initial data or the interpolation data is Fourier-transformed. By calculating the eccentricity correction value, a specific low-frequency component is extracted from the initial data or the interpolation data, and the low-frequency component is added to the correction data to obtain the eccentricity. A large amount of eccentricity can be obtained.

According to the third aspect of the present invention, irregularly and irregularly occurring vibrations are predicted from the obtained eccentricity and the shape data of the temporary rotating body. Thus, actual work such as mounting the magnetic head on the rotating drum can be efficiently performed.

According to the fourth aspect of the present invention, since the virtual curve is obtained by calculus by separating the order components, even if there is a data missing portion such as a notch, the interpolation shape of the portion is accurately calculated. can do.

According to the fifth aspect of the present invention, since the virtual curve is obtained by predicting the shape of the missing portion from the entire flow of the remaining data, there is a data missing portion such as a notch. Also, it is possible to accurately calculate the interpolation shape assuming that there is no such part.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a system configuration of a production facility for performing a position adjustment operation of a magnetic head.

FIG. 2 is a schematic diagram showing a positional relationship between a drum and a length measuring sensor when viewed from a side.

FIG. 3 is a graph showing data of an eccentric curve and an eccentric amount at an arbitrary angle.

FIG. 4 is a schematic diagram showing a positional relationship between two length measurement sensors.

FIG. 5 is a perspective view schematically showing a positional relationship between a drum and a length measuring sensor from a side.

FIG. 6 is a graph showing change data of a distance in the X direction due to eccentricity.

FIG. 7 is a graph showing trajectory data of a distance in the X direction due to a difference in orbit.

FIG. 8 is a schematic diagram illustrating the principle of eccentricity of the drum.

FIG. 9 is a graph showing eccentricity data including a first aperiodic component.

FIG. 10 is a schematic diagram showing a state in which a drum center position moves due to eccentricity.

FIG. 11 is a schematic diagram showing a state in which a drum center position moves due to eccentricity including aperiodic run-out.

FIG. 12 is a graph showing data of a movement locus of an axis when the drum rotates.

FIG. 13 is a schematic view showing a bearing using a ball bearing.

FIG. 14 shows a state in which the ball bearing rotates together with FIGS. 15 and 16, and FIG. 14 is a schematic view showing an initial state.

FIG. 15 is a schematic diagram showing a state when the ball bearing makes one rotation.

FIG. 16 is a schematic view showing a state in which a ball bearing returns to an original position.

FIG. 17 is a graph showing a non-periodic chart of a ball bearing.

FIG. 18 is a schematic diagram showing a drum shape in an emphasized manner.

FIG. 19 is a schematic view showing a relationship between a notch of a drum and a head.

FIG. 20 is a graph showing an eccentric curve including a data missing portion and a differential curve thereof.

FIG. 21 is a schematic diagram showing a state of a head position adjusting operation.

FIG. 22 is a diagram showing a processing flow in the present invention.

FIG. 23 is a graph showing data (raw data) obtained by measuring a distance to a drum side surface.

24 is a graph showing data (data 1) obtained by removing a noise component from the raw data shown in FIG.

FIG. 25 is a graph showing data (data 2) after performing processing for matching the start point (0 °) and the end point (360 °) of data 1 shown in FIG. 24.

FIG. 26 is a graph showing data when the final data is an abnormal value.

FIG. 27 is a graph showing data in the case of an abnormal shape.

FIG. 28 is a graph showing data when a start point and an end point are combined for calculation processing.

FIG. 29 is a graph showing data of a provisional shape when a start point and an end point are combined for calculation processing.

FIG. 30 is a graph showing data (data 3) of a provisional eccentric curve when a start point and an end point are combined.

FIG. 31 is a graph showing temporary second non-periodic vibration data when the start point and the end point are combined.

FIG. 32 is a graph showing temporary drum shape data (data 4).

FIG. 33 is a graph showing data of an approximate straight line of a temporary drum shape.

FIG. 34 is a graph showing data when correction for removing a tilt component is performed to obtain a correct shape value.

FIG. 35 is a graph showing a difference in a temporary drum shape depending on the number of corrections.

FIG. 36 is a graph showing data of a maximum change point of a sine wave.

FIG. 37 is a graph showing how sine waves are combined.

FIG. 38 is a graph showing data (data 6) obtained by performing the first aperiodic shake correction on data 1 shown in FIG. 24.

FIG. 39 is a graph showing data including a missing portion and data representing a provisional eccentric curve in a drum to which a magnetic head is attached at a distance of 180 ° from each other.

40 is a graph showing data obtained by correcting a data missing portion in FIG. 39.

41 is a graph showing the data of the eccentric curve after correcting the data missing part in FIG. 40.

FIG. 42 is a graph showing a method for detecting a data missing portion.

FIG. 43 is a graph showing an eccentric curve including a data missing portion and its Fourier transform value.

FIG. 44 is a graph showing a state of data correction when a data missing portion is narrow.

FIG. 45 is a graph showing a state of data correction when a data missing portion is wide.

FIG. 46 is a graph showing eccentricity curve data (data 10) after the first aperiodic shake correction.

FIG. 47 is a graph showing data of an actual eccentric state of a drum.

FIG. 48 is a graph showing data of a second aperiodic vibration (data 11).

FIG. 49 is a graph showing data (data 12) of an aperiodic deflection curve.

FIG. 50 is a graph showing drum shape data (data 13) obtained by subtracting data 10 and data 12 from data 1;

FIG. 51 is a schematic view showing a difference between a displacement in the Y direction and a distance in the X direction.

FIG. 52 is a schematic diagram showing a state where a length measurement sensor is arranged beside a camera.

53 shows, in conjunction with FIGS. 54 to 57, a state of data correction at the time of restoring the drum shape when there is a data missing portion in the application example 1, and FIG. 53 is a graph showing raw data. .

FIG. 54 is a graph showing data after correction with an eccentric component.

FIG. 55 is a graph showing data after correction with a temporary shape waveform.

FIG. 56 is a graph schematically showing how a drum shape is restored by data synthesis.

FIG. 57 is a graph showing data after restoring the amplitude that changes in a curve.

FIG. 58 is a graph showing a data trend curve in application example 2.

Claims (5)

[Claims] 1. A method according to claim 1, further comprising: measuring a distance to the rotating body at predetermined angles while rotating the rotating body once; obtaining a temporary eccentric curve of the rotating body from initial data obtained by the measurement; From the tentative rotator shape data, and from the tentative rotator shape data, find a part inconsistent in shape of the rotator. A method for determining the amount of eccentricity of a rotating body, wherein correction data is obtained so that the value of a portion and the value of an end point portion have the same value. 2. When there is a missing portion in the initial data, an interpolation data is created by filling the missing portion with a virtual curve, and an eccentricity correction value is calculated by Fourier transforming the initial data or the interpolation data. The eccentricity of the rotating body according to claim 1, wherein a specific low-frequency component is extracted from the initial data or the interpolation data, and the eccentricity is calculated by adding the low-frequency component to the correction data. How to ask. 3. The amount of eccentricity of a rotating body according to claim 2, wherein irregularly and irregularly occurring runout is predicted from the obtained amount of eccentricity and the shape data of the provisional rotating body. How to ask. 4. The method for determining the amount of eccentricity of a rotating body according to claim 2, wherein the virtual curve is determined by calculus by separating order components. 5. The method according to claim 2, wherein the virtual curve is obtained by predicting the shape of the missing portion from the entire flow of the remaining data.