JPS58146805A - Measuring device of flatness - Google Patents

Abstract

PURPOSE:To measure flatness with high accuracy without errors, by detecting the interference fringes associated to a plane to be inspected and a reference plane with image sensors, calculating the flatness and correcting the same in accordance with the detection signal corresponding to the correction region on the reference plane. CONSTITUTION:The interference luminous fluxes generated by a plane to be inspected and a reference plane which are disposed in parallel are made incident to image sensors 34, via a total reflection mirror 32 rotated by a pulse motor 42. The video outputs thereof are taken via a head amplifier 46, a peak holding amplifier 55, one of LPFs 56a-56c, a sample holding amplifier 57, and an AD converter 58 into a microcomputer 48. On the other hand, the output of a rotary encoder 41 operating cooperatively with the motor 42 is read into the microcomputer 48 as well. Flatness is calculated from the positions of the bright fringes in the three positions A-C of a mirror 32 and the pitch of the fringes. The results obtained by the comparison between the values detecting the positions of the thin lines provided on the reference plane and the reference value are subjected to correction for deviation and oscillations of the optical systems, whereby the flatness is measured accurately without errors.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention proposes a flatness measuring device W that can measure the flatness of an object plane ultra-precisely and automatically.

The slider UkJK of the magnetic head of the magnetic recording device is 0.
The trend is for extremely precise flatness of 0.05 m or less to be required. Although there are various methods for non-contact measurement of flatness that utilize optical interference, in order to easily and stably perform measurements with the above-mentioned accuracy, it is necessary to improve the precision of mechanical parts, ultra-fine adjustment, and measurement accuracy. There are many problems with maintenance, etc., and there are no examples of this being put into practical use.

The present invention 111-1 was developed in view of the above circumstances, and its purpose is to provide an apparatus that can automatically measure flatness with an accuracy of 001 μm.

First, flatness measurement using the measuring device of the present invention l1AJl!
11, or this principle itself is described in Japanese Patent Application No. 53-68674 (Japanese Unexamined Patent Publication No. 54-15) proposed by the present inventor.
The measurement principle is the same as that of the "Flatness Measuring Method" in Japanese Patent Publication No. 9258). That is, by using, for example, a Fizeau optical interference system, interference fringes l between the test plane and a reference plane such as an optical flat are obtained as shown in FIG.

This pattern of interference fringes 1 is captured at three different positions in the fringe direction by a one-dimensional image sensor 34 whose longitudinal direction is aligned with the width direction of the interference fringes, and a predetermined brightness is detected. Line (or dark line) position a.

The stripe width direction positions Pa, Pb, Pc and stripe pitch P in b and c are detected from the image sensor output, and the flatness F
Calculate as = mu x ze 2 n p... (1), where λ: interference light source wavelength n: interferometry & (however, n here 1) JP = Pb yuan earth mu... (2) It is something.

It is an object of the present invention to provide a flatness measuring device that eliminates errors caused by external vibrations, errors caused by optical system deviation over time, etc., and ensures highly accurate measurement.

DESCRIPTION OF THE PREFERRED EMBODIMENTS The apparatus of the present invention will be specifically explained below based on drawings showing embodiments thereof.

FIG. 2 is an external view of the apparatus of the present invention, and FIG. 3 is a layout diagram of its optical system. This optical system consists of a case 1 in which the inside is shielded from light except for the vertically movable sample stage 11 on which the object to be measured is placed with its measurement target plane 10 m facing upward.
It is contained within 2.

He-Ne radar (wave f/c O, 63
The laser beam is mounted vertically in the cylindrical part 12a of the case 12 provided at the rear of the 21-piece sample stage 11 with the output side facing upward.
D filter 22. The beam passes through a pinhole 24 having a diameter of 23 degrees and a diameter of 20 tones, and reaches total reflection @25 provided above. A total of 25 total reflection mirrors are arranged at an angle of 45° with respect to both the horizontal and vertical planes to horizontally reflect the narrow laser beam exiting the pinhole 24 toward the front of the case 12. Above the sample stage 11 in the optical path, a total reflection mirror 26 is arranged at an angle of 45 degrees with respect to both the horizontal and vertical planes so as to project a laser beam vertically toward the sample stage 11. The radar beams are all directed toward the dimple stage 11 through an achromat 27 as an objective lens, a half mirror 28, and an optical shaft 29 arranged behind the total reflection mirror 26 and the sample stage 11. The half mirror 28 is tilted 45@ with respect to both the horizontal plane and the vertical plane, and the interference fringe pattern caused by the light from the optical flat 29 side, that is, the reference plane 29- of the optical cut 29 and the test plane 10a, is reflected in the case 12. It is arranged to reflect horizontally to the left.

The light beam reflected from the bar 7mi 228 is transmitted through a half mirror 31 that is inclined at 45 degrees with respect to both the left and right directions and the front and back directions of the case 12, and then passes through a total reflection mirror 32 that is mounted so as to be horizontally rotatable around a vertical axis. Finally, Half Mira 3
1 and reaches a focus 7-door 33 provided on the front of the case 12. This is for visual observation and photographing of 33 interference fringes. A one-dimensional image sensor f34 is located behind the vertical axis of the total reflection mirror 32.
is installed with its longitudinal direction vertical, and when the total reflection mirror 32 is in a rotational position tilted by 45 degrees with respect to both the left-right direction and the front-back direction of the case 12, the image sensor f34 When the field of view position is at position b shown in FIG. 1, that is, approximately at the center of the interference fringes obtained by these optical systems, and the total reflection mirror 32 is rotated clockwise (or counterclockwise) in plan view, Field of view position #- of image sensor 34
This results in deviation in the IC position direction (also in the LIT position direction).

The total reflection mirror 32#i is attached to an appropriate frame 32m, and the rotary shaft 32 is a round shaft extending downward from the center of the lower side.
b, this rotary shaft 32b is fitted with a KFi gear 32c, and a rotary encoder 41 supported by a frame (not shown) is connected to the lower end. A pulse motor 42 with a vertical output shaft is mounted on a pedestal on the side of the rotary encoder 41, and a cogged belt 13 is hooked and rotated between the gear 42& fitted to the output shaft and the gear 32c. The total reflection mirror 32 is driven by the pulse motor 42.
This is done so that it can be rotated horizontally. Limit switches 43 and 44 are provided in the rotation range of the total reflection mirror 32 and the frame body 32m, and limit switches 43 and 44 are provided to specify the completion of the rotation, and the completion of the rotation is determined by the rotation of the image sensor 34 as described above. From the state where the visual field position is 5 positions, it is rotated counterclockwise and clockwise by approximately equal angles (10'' or less) to the 9th position.

As can be understood from the above configuration, the interference optical path section, more specifically, the light on the detection side of the beam combining optical system is transferred to the image sensor 34 using a total reflection mirror 32 (or a prism) capable of rotating the beam.
The structure is such that the position at which the stripes should be read is changed by rotating the total reflection @32.

A fine correction line 29b is formed on the surface of the optical flat 290 by a method such as metal vapor deposition, so that the reflectance of this part is lower than that of other parts of the reference plane 29. . The correction thin line 29b is linear, and as shown by the two-dot chain line in FIG.
It is observed that the longitudinal direction of the image sensor 34 appears in a direction perpendicular to the longitudinal direction of the image sensor 34, and the bit address to be described later is 52. The position on the optical flat 29 is determined so as to be a larger position). The length of the correction thin line 29b is the total reflection mirror 3
Even when 2 is at the limit of the rotation range on both sides, the size is set so that it is captured by the image sensor, and the width is determined from the pitch of 7 otodiodes of the image sensor f34 (1j20/1 purchase in the example). It is set a little thicker (for example, 30 houses) so that its presence is detected by at least 1 bit of the image sensor 4j34.

Although the above-mentioned configuration uses a Fizeau interference optical system, other optical systems for obtaining interference fringes, such as a Michelson interference optical system, can be used as well. Others 14゜14i1 Screw for adjusting the posture of optical flat 29, 3#''i case 12
This is an external power source for the laser.

The measuring device main body 2, which consists of a case 12 and an optical system inside the case 12, has each part rounded to be as solid as possible to eliminate measurement errors caused by vibration, and the bottom surface is equipped with anti-vibration rubber 1111.
Please take precautions such as installing 5. Furthermore, the accuracy of the mounting position of the optical system and the working accuracy of the movable parts shall be on the order of 10 tones. J! Furthermore, the half mirrors 28, 31 and the optical flat 29 are provided with several degrees of chi/( on their back surfaces to avoid rear surface interference.

FIG. 4 is a block diagram schematically showing the main parts of the electronic circuit of the device of the present invention. Of this electronic circuit, the above-mentioned image sensor 34. Rotary encoder 41. ) Gurus motor 42. Its drive circuit 53. Drive circuit 45 and head amplifier 46 for limit switches 43 and 44 and image sensor 34
, and some other operating members, except for the pilot lamp, are installed inside and outside the case 47 as a control I14.

48 is a microcomputer which is the center of IIJ# calculation of the device of the present invention, and includes a CPU (central processing unit), R
The start switch 49 and stop switch 50 on the front panel of the control unit 4 are equipped with OM (read-only memory), RAM (random access memory), input/output interface, clock pulse oscillation circuit, etc. (hereinafter referred to as a microcomputer) 48 is a strange push-button illuminated switch that instructs the microcomputer 48 to start, stop, or interrupt measurement.When pressed, it gives the above commands to the microcomputer 48, and also sends signals from the microcomputer 48 side. Lights up at. The digital switch 51Fi thumbwheel switch, which is also placed in front of the control unit 4, consists of two digits, and the contents of the setting are data for specifying the field of view from which to obtain the image sensor output for which the flatness of the object to be inspected is to be calculated. , in other words, positions a, c
This is given to the microcomputer 48 as rounded data indicating the rotation angle position of the seven-way total reflection mirror 32 whose field of view is . The rotary encoder 41 is configured to output 16 pulses per 1° rotation of the total reflection mirror 32, and its output is read into the microcomputer 48 via the encoder interface 52. Counterclockwise rotation 1iii
The operating states of the limit switch 43 that defines the J range limit and the limit switch 44 that defines the clockwise rotation range are both read into the microcomputer 48.

Operation status of limit switches 43 and 44 11 This is the start/stop control information of the pulse motor 42 by the microcomputer 48 ◎In other words, in the initial state, the total reflection @32 is in the state where the limit switch 43 is activated, and it starts in this state. When the switch 49 is pressed, the microcomputer 48 issues a signal to the motor drive circuit 53 to rotate the pulse motor 42 in the normal direction and rotate the total reflection mirror 32 clockwise. Limit switch 44
When the stop switch 50 is activated or the stop switch 50 is pressed, the above-mentioned signal for forward rotation is cut off, the motor drive circuit 53 and the hepal smoke 42 are reversed, and the total reflection mirror 32 is rotated counterclockwise. When the rotation is performed and one limit switch 43 is activated, the signal to the warp drive circuit 53 is cut off.

The microcomputer 48 is equipped with a ninth kakunta for specifying the field of view position of the image sensor 34 or the rotational position of the total reflection mirror.
This kakunta is pressed so that it is cleared by the origin pulse generated by the rotary encoder 41. The rotation axis 32b is set so that this origin pulse is obtained at a position where the total reflection mirror 32 is slightly rotated clockwise from the limit switch 43 operating position.
and the rotary encoder 41 are connected. Then, the total reflection of the microcomputer 48 makes the count value of the above kakunta 90! !
32 is the field of view position where the output of the image sensor at that time should be used for flatness calculation, and the setting position by the digital switch 51 is X, then the kakunta count value is 90-x, 9O-)-1. Two rotational positions A,
Let C be visual field positions a and c whose image sensor outputs at that time should be used for flatness calculation.

Similarity X is set as a value representing the same dimensions a-c in units of ■, and since the optical system is configured so that the rotary encoder pulse corresponds to 0.2 points in the field of view, x - (X/
2) Calculated as +0.2. When the content of the above-mentioned Katank becomes 90+x, control is performed to reverse the pulse motor 42 without waiting for the limit switch 44 to operate.

A nine-push button type switch 61 arranged on the side of the switches 49 and 50 is provided for instructing the correction mode, and when pressed, the microcomputer 48 performs a correction mode operation to be described later.

It is desirable that the number of interference fringes captured by the image sensor 34 be about 100 bits.

Since the dimension λ force between adjacent bright lines (or dark lines) is measured with an accuracy of 1/100, the accuracy in this case is theoretically (λ/2) x (1/100) ◆ 0.003 degrees. -, and a practical accuracy of 0.011 ws can be easily achieved. Therefore, in the example, for an optical system configuration that captures 10 interference fringes, M of 1024 bits is required.
Use the OS sensor.

Now, when the microcomputer 48ti enters the data reading process MK, it issues a start pulse and a clock pulse to the drive circuit 45. The drive circuit 45 sets this to a required level to overcome the image sensor 346C. Then the image sensor 34
is sequentially scanned from the 1st bit to the 1024th bit, and the received light amount KJ6 is output as a Hernohitheo output.This video output is amplified by the head amplifier 4.61 and sent to the beak hole F amplifier 55 of the control section 4. ．． Low pass filter 56a.

56b or #-156c, sample hold amplification [15
The data is input to the microcomputer 48 as digital data via the width converter 58. The number, hold period, and AI/D conversion period are matched with the clock pulse period applied to the drive circuit 45.

The cut-off frequencies of the low-pass filters 56, 56b, and 56c are 500Hz, 250Hz, and 100Hz.
This is different from the above, and one of them is selected by the flJ exchange switch 59. Which one to select may be determined depending on the frequency of the noise.

Based on the data captured in this way, the flatness F is calculated according to equation (1), and the result is displayed in 71 units on a 94-digit display 60 installed in front of the control unit 4. It is set to display up to three decimal places.

Figure 5 shows the microcomputer 48 when operating in correction mode.
3 is a flowchart showing control details.

Operation in this correction mode is necessary [6].In that case, the switch 6 must be pressed without setting the test object lO.
1 is pressed. When the microcomputer 48 detects this, it rotates the pulse motor 42 in the forward direction. Then, read the setting value X of the digital switch 51 (preset for the next measurement) and select P@wm9 G-x.
, p, sg90+X are calculated. P defining these PIePl and B positions.

=90 is stored in a predetermined register. When the pulse motor 42 starts rotating, an origin pulse is input from the rotation encoder 41 with a slight delay. By inputting this origin pulse, a counter for counting output pulses of the rotary encoder 41 is reset in order to specify the visual field position of the image sensor 34. Then, the output pulses of the rotary encoder 41 are counted, and when the count reaches P, which is stored in the register, data is read from the image sensor 34, and the bottom address P, bottcm is calculated (described later). The bottom address captures the correction thin line 29b (the image sensor 34 captures the reference plane 2 of the optical flat 29).
Only the part of the thin line 29b that captures the reflected light from the image sensor 9a has a low reflectance and is dark) These are pit addresses 1 to 1024 of the image sensor 34.

Next, K Id P(, = Pl bottom - PIB is calculated [see Fig. 6]. Here, p, Fi are the standards written in advance in the ROM of the microcomputer 48, for example, at the time when the device of the present invention is assembled, etc.) value,
The content is the pit address of the image sensor 34 where the thin line 29b is captured at that time, that is, the pit address where the thin line 29b would originally be captured under ideal conditions where the optical system is sufficiently adjusted and there is no vibration. It has become.

Next, IPc, I is compared with the tolerance E (for example, 2 bits) written in the ROM in advance, and if it is less than that, r/i0 is used, and if it exceeds E, the calculated value Pc1 is used as correction data. It is stored in RAM etc. as Pct.

As we continue to count the output pulses of the rotary encoder 41, when the count content of the kakunta reaches P, we calculate the bottom address P, bottom in the same manner as in lit, and then P. , and store it in RAMK. Furthermore, when the count content of Kakunta reaches P3, the PCI is similarly obtained and stored in the RAM. The same value is used for PB and E #- of one prefecture to calculate PC2+P(3.Then, the pulse motor 42 is reversed, and the limit switch 43
When the pulse motor 42 is activated, the reverse rotation of the pulse motor 42 is completely stopped, and a series of correction mode operations are completed.

Next, the operation and control during measurement will be explained. FIG. 7 is a flowchart of the main program showing the control contents of the microcomputer 48. When the start switch 49 is operated, the pulse motor 42 is rotated in the forward direction.

Then, read the setting value X of the digital switch 51,
Calculate P, = 90-x, P@= 90+x. P,=90 defining these P, , P, and B positions are stored in a predetermined register. When the pulse motor 42 starts full rotation, an origin pulse is input from the rotation encoder 41 with a slight delay. This origin pulse input causes the image sensor 3 to
In order to specify the visual field position of No. 4, a kakunta that suppresses the output pulse of the rotary encoder 41 is reset. Then, data is read from the image sensor 34 (described later).
When a pulse from the rotary encoder 41 is input, data is read from the image sensor 34 again, and then the kakunta is incremented. This process is repeated until the counted contents of the kakunta match Pl of the register, and when Pl is reached, the peak address is calculated (described later).

Peak address is an image sensor that captures bright lines.
34 bit addresses 1 to 1024, and the above-mentioned P
, ΔP. And the calculated peak address (there are as many as the number of bright lines) pH.

Pl2...Pa1... are stored in the RAM. In this case, it is stored in two separate @ areas, one t-IP, noP
, and the other is used for correspondence with the next peak address. Each time a pulse from the rotary encoder is input, data is read from the image sensor 34 and the kakunta is incremented. First, calculate the peak address using the data read from the image sensor, and compare this with the peak address obtained in the previous peak address calculation, first #-1Pu-Pu...Phi...
A history regarding peak addresses corresponding to newly appearing bright lines, disappearing bright lines, etc. is stored in the RAM together with the addresses. The peak address is then stored in the RAM and used as comparison data with the next calculated peak address. This process is continued until the content of the kakunta becomes P. Bright lines are identified by the sequential comparison of peak addresses as described above, so that even if the interference fringe pattern is complex, adjacent bright lines that differ by λ/2 wavelength will not be mistakenly confused.

Then, when the counter content becomes P, the peak address Pal I P! is reached as in the case of PK. t...Pti...
is calculated and stored in RAM. Thereafter, the same processing as between 21 and 17 is performed until the counter content becomes P, and when it becomes PI, the peak address P11+PM2...P
si... is calculated and stored in RAM. Then, the pulse motor 42 is reversed until the limit switch 43 is activated.

Among the peak addresses at 7i1r, , P, , P, the flatness is calculated for those related to the common bright line, and this is displayed on the display device 60.

Next, we will discuss the process of reading data from the image sensor in Part 8.
The explanation will be based on the flowchart shown in the figure. Rotary encoder 4
To give a start pulse to the left drive circuit 45 where a predetermined condition for starting data reading is met, such as input of 1 pulse, the output port connected to the left drive circuit 45 is set to high level, a predetermined counter [3 is set], and then a clock pulse is input. The drive circuit 4
Send to 5. The counter is decremented by this tarok pulse, and after the counter content becomes 0, the output port is set to low level. That is, the start pulse is caused to fall. The processing up to this point is to compensate for the time delay (equivalent to three tarok pulses) from when the image sensor 34 used receives the start pulse until it outputs the first bit.

Next, the starting address K of the area in the RAM prepared for the data reading is set in the address counter, and bit &1024 of the image sensor 34 is set in the counter. The output of the image sensor 34 is converted into A/'I) in synchronization with the clock pulse, the converted data is stored in the address, the address counter is incremented, and the counter is decremented. A counter handles this kind of processing. Continue until it becomes 0, then K - K + 102 in RAM
Store data at address 3.

Figure $9 is a 70-chart showing the processing content for calculating the peak address. First, the first address counter is set to K (RAM storing 1024 data to be processed).
The first address of the area) is set to +50, and another counter is set to 900. This is a measure to exclude the top 51 bits and the bottom 124 bits of the 1i1024 data from data processing targets. Next, the starting address of the RAM area for storing the calculated peak address is set as W in the second address counter. Furthermore, a beak counter for counting the number of calculated peaks is set to 0.

Next, the address of the first address counter (first tfK+50
Data DATA O at number l1l (+, ) and data DATA 1 at the next address (initially +51 address) are read out and compared. If both are equal (meaning that the light reception levels of both pits are equal), the content of the first address counter is incremented by +1, and the counter is incremented by -1.

Then, similar processing is executed for the next address.

On the other hand, if DATA O > DATA 1 (K+, meaning the +51st bit is darker than the 500th bit), the address of the first address counter is +1, and the counter is -l. do. Then, read out DATA 1 (data at address K+52) lk and DATA O (data at address K+51) and compare them. If DATA I > DATA O is not true, the data tends to become darker as the address increases. , so repeat the same process.

During this process, if DATA O<DATA I, the process enters the peak identification step described below.

In other words, when the first address counter is +50, DATA is incremented one after another.
If O<DATA 1, the brightness tends to increase as the address increases and data processing progresses. In this case as well, increment the first address counter by 1,
Process to increment the counter by 1 DATA O> DATA
Repeat until the value becomes 1 (that is, until the address of the brightest bit becomes the content of the first address counter), and use the content of the @1 address counter at that step as the peak address to the second address counter. RA instructed by
Store in the address of M (first address #iK'). Then, the content of this second address kakunta is incremented by +1, and the bi-nokakunta is incremented by +1. And DATA O, DATA I
Iteratively compares the size of . The above processing is executed until the content of kakunta becomes 0.

The peak address obtained as described above is positional information with one end of the image sensor 34 as the center position of the bright line of the interference fringes captured by the image sensor 34.

The bottom address calculation in the correction mode is the above-mentioned peak address calculation and DATA O.

The same procedure is performed by reversing the relationship of DATA 1. That is, in FIG. 8, DATA O> DATA I? and DATA O< I) ATA 1 ? Processing is performed with the conditions reversed.

In other words, after the condition of [)ATA O> DATA I is satisfied (the larger the address, the darker the address), DATA O<
When the address of the dark bit of CLK (for which the DATA I condition is met becomes the content of the first address kakunta), the content of the first address kakunta is stored in a predetermined register as the bottom address. Since there is only one bottom address corresponding to one thin line 29b, it goes without saying that the processing regarding the second address kakunta and peak kakunta shown in FIG. 8 is unnecessary.

In this embodiment, the flatness is determined using the bright line, but when the dark line is used, the bottom address calculation as described above will be performed.

Now, in this example, the peak addresses pHrP12...Ptl..., P□, P7.・
...Pl1'..., P31 r Pst...Psi
... is the data for flatness calculation, P, =
A bright line that appears or disappears midway between 90-x-P3=9'0+x (for example, the bright line represented by the bottom solid line in FIG. 1) #-i is excluded from the processing target.

After calculating the peak address as described in 1-t0tI,
It is identified by comparing it with the peak address calculated one output pulse ago from the rotary encoder, and its history information is attached to the RAMK.

That is, the latest peak address is P1.1. P4.
Let z...Pj, m (m is the count value of the peak kakunta), and let the peak address one output before the rotary encoder be P jl, 1 + P j-1, *.... Then, as shown in Figure 1O, first (town, x = PI, t
)/4=±1/4 of the fringe pitch P is calculated as Po, and the condition for fringe identification is that it is less than this.

Then, calculate the difference PieΔP between PH,+ and P5-s, le Pj-'+'', and identify the combinations of stripes smaller than Po as the same.PL' and PI-t,
If s is identified, P 3-r, * is stored as having disappeared, or is left as unnecessary for flatness calculation. The following k (-2, 3 m)th peak Pj, k
Pl-t with K, k-sPj-1,k + Pj”
The difference between l and the dog is ΔPk-1, ノPk, ノPk,
is determined, and combinations of stripes for which this value is less than or equal to P0 are identified as the same stripe.

Therefore, the peak address PIS # PIt...P, Bi..., Pl1 * Pl1...Pyl... and P3
□, P8. ...Pal'..., excluding those that have a history of occurrence and disappearance midway through, a set of three numbers combined in order of magnitude will be identified as a peak address related to the same bright stripe. . Therefore, using a set of three data, the flatness F = (F+
+ Ft + ...+ Fm-t )/(m 1
), and m is the denominator Lr of the multiplier of Fle Ft... in the above formula common to visual field positions a, b, and c.
1 corresponds to P in formula (1), and the molecule corresponds to formula (1) or (2
) corresponds to ΔP in the equation. In calculating this P, only data on the field of view position is used, and almost simultaneously it
Since the data obtained by resetting is used, there is no need for correction due to deviation of the optical system, disturbance, etc. On the other hand, since the data at each of the visual field positions a, b, and c are used to calculate ΔP, if there is a deviation in the optical system at each visual field position or an influence of vibration due to the visual field position, these correction values will be Of course, the calculation of Kii has no substantial influence.

The thousand-faced skin FFi calculated as described above is displayed on the display device 60. Note that the maximum value and minimum value among F1 to F, , are preferably not used when calculating F. In this case, an error is displayed when the number of tfm is 2 or less. Also, if the number is 16 or more, the data is deemed unreliable or abnormal and an error is displayed.

As described above, the flatness measuring device according to the present invention obtains interference fringes related to a test plane and a reference plane, and measures the flatness of the test plane based on the interference fringes. It includes an optical system arranged to generate interference fringes, an image sensor for capturing the interference fringes, and a data processing unit that calculates flatness based on the image sensor output, as described above. The open surface is locally provided with correction regions having different reflectances, and the data processing unit compares the image sensor output information specifying the correction region with reference information to obtain correction information, and Since it is configured to refer to the correction information when calculating the degree, it is possible to eliminate the decrease in measurement accuracy due to errors due to external vibrations, deviations of the optical system over time, etc. Flatness measurement can be achieved. Moreover, this measurement is performed automatically, and the operation of the correction mode is extremely simple.

Furthermore, in order to obtain the above-mentioned accuracy, a 10-piece ceramic auger is sufficient for the precision of mechanical parts or assembly, and there is no need for ultra-fine adjustment or troublesome maintenance during use. Furthermore, the field scanning IF of the one-dimensional image sensor for reading the interference fringe pattern is constructed by rotating an optical means (a total reflection mirror in the embodiment) interposed in the middle of the optical path, so the image sensor is firmly fixed. noise due to noise, vibration, etc., and therefore errors;
The ¥ factor can be eliminated. Furthermore, since the field of view scanning of the optical means is performed by rotational movement, the 0T moving part thereof can be easily manufactured with high precision, and the rotational movement thereof can also be performed stably.

Furthermore, the measurement area can be easily set using a number setting means such as a digital switch, and the f! Not only is it extremely convenient for selecting and instructing the four regions, but it also allows for high-precision measurement because it can be corrected according to the measurement of the m region set by this number setting means.
This is preferable in that it enables

As detailed above, the present invention makes it possible to provide a flatness measuring device that is innovative in terms of measurement accuracy, convenience of use, and manufacturing, and the process by which the present invention contributes to ultra-precision flatness measurement technology. is huge.

[Brief explanation of drawings]

The drawings show an embodiment of the present invention, in which Fig. 1 is an explanatory diagram of the measurement principle, Fig. 2 is an external view of the apparatus of the present invention, Fig. 3 is a layout diagram of the optical system, and Fig. 4 is an electronic diagram. A block diagram of the circuit, FIG. 5 is a flowchart in the correction mode, FIG. 6 is a schematic diagram for explaining the correction mode, and FIGS. 7 to 10 are 70-charts of calculations and control by the microcomputer. 21... He-Ne radar 28, 31... Half mirror 29... Optical 7 rat 29b... Thin line for correction 32... Total reflection mirror 34... Image sensor 41... Rotary encoder 42. ...Pulse motor 48...Microcomputer 51...Disicle switch 56a, 56b, 56c...Low pass filter patent applicant Sumitomo Special Metals Co., Ltd. agent Patent attorney Noboru Kono Inu J 4 J/,
J4a b
c No. 67 Calculation 8

Claims (1)

[Claims] l. Obtaining interference fringes related to a test plane and a reference plane;
An apparatus for measuring the flatness of a test plane based on the interference fringes, comprising: an optical system including the reference plane and configured to generate interference fringes; an image sensor for capturing the interference fringes; and an image sensor output. a data processing unit that calculates flatness based on the reference plane; correction areas having different reflectances are locally provided in the reference plane, and the data processing unit includes an image sensor that specifies the correction area. A flatness measuring device characterized in that the output information is compared with reference information to obtain correction information, and the correction information is referred to when calculating the flatness.