JP2000046748A - Method and apparatus for inspecting conductor pattern and production of multilayered substrate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a conductor pattern inspecting method and apparatus capable of performing inspection certainly at a high speed, and a method for producing a multilayered substrate. SOLUTION: A circuit board 51 to be inspected is illuminated by polarized illumination light having a wavelength of 300-550 nm, and the optical image based on reflected light blocking the polarizing component of the polarized illumination light from the illuminated circuit board to be inspected is detected by a detector to be converted to an optical image signal. The converted optical image signal is binarized on the basis of a predetermined threshold value to be converted to a binary image signal showing a conductor pattern, the positional shift quantity of the conductor pattern is calculated on the basis of the binary image signal showing the converted conductor pattern, and alignment is performed on the basis of the positional shift quantity of the conductor pattern to be compared to extract the defect of the conductor pattern.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an appearance inspection of an electric circuit board, and more particularly to filling a through hole formed for electrically connecting front and back wirings of a green sheet used for a ceramic substrate. Pattern inspection method and apparatus for inspecting a conductor pattern composed of a conductive paste filler containing metal fine particles to be formed and wiring pads of the conductive paste containing metal fine particles printed on the surface of the green sheet, and a multilayer ceramic The present invention relates to a method for manufacturing a multilayer substrate such as a substrate.

[0002]

2. Description of the Related Art Conventionally, there is disclosed a through hole filling state inspection technique for inspecting metal fine particles (paste) filled in a large number of through holes (through holes) formed in a green sheet used for a ceramic substrate. 30700
No. 6 (prior art 1). In the related art 1, first, an image is detected from above or obliquely above the circuit board while being illuminated from obliquely above. The image detected from above is binarized to extract a region of the filling, and based on the area and perimeter, a foreign matter adhesion / mixing defect, a scattering defect, a bleeding defect, and a defect without a through hole are detected. The image detected from obliquely above is binarized, an image region corresponding to the wall portion of the through hole is extracted, and a lack defect is detected based on the area. A technique for inspecting wiring pads formed on the surface of the green sheet is disclosed in Japanese Patent Application Laid-Open No. 1-259245 (prior art 2). In the prior art 2, first, a binary reference pattern is enlarged and reduced, and edges of the enlarged pattern and the reduced pattern are obtained. An overlapping portion between the detected binary image captured and binarized and each edge is detected as a defect.

[0003]

Since the filling material is filled with a paste containing fine metal particles in the through holes by screen printing, the boundary between the paste and the base material (green sheet), that is, the edge portion has many irregularities. Further, the green sheet is not flat, but has irregularities such as dents. In the case of the above prior art 1, based on the area and the perimeter of the paste, foreign matter adhesion / mixing defects, scattering defects, blurring defects, etc. are detected. If the edge of the paste has irregularities, the perimeter becomes longer, causing oversight and false information. In addition, since the insufficient filling defect is detected by making the wall portion of the through-hole bright and visible, the dent of the green sheet may be detected brightly and become false information. Further, the fine metal particles contained in the paste glow sharply, which causes a false alarm.

On the other hand, the wiring pads on the surface of the green sheet are formed by screen printing a paste containing fine metal particles. Due to variations in screens and printing conditions, the shape and size of the wiring pads vary from one green sheet to another. Also, the size of each wiring pad of one green sheet varies depending on the hole shape of the screen and printing conditions. In the case of the above-described prior art 2, since the variation in the size of the wiring pad cannot be dealt with, a defect is overlooked or a false report occurs.

[0005] An object of the present invention is to solve the above problems.
Surplus defects (bleed,
It is possible to reliably and quickly inspect spatters, protrusions, missing defects (chips, pinholes), missing defects, foreign matter-incorporated defects, defects without conductor patterns, and displacement errors while minimizing false alarms. An object of the present invention is to provide a conductor pattern inspection method and apparatus.

Another object of the present invention is to provide a conductive pattern including a conductive paste containing metal fine particles and a conductive pattern such as a wiring pad, even if a variation occurs between sheets or within a sheet surface. A conductor that enables reliable and high-speed inspection of surplus defects, missing defects, insufficient defects, foreign matter-containing defects, defects without conductor patterns, positional displacement defects, etc., even if irregularities occur on the surface, without erroneous detection. An object of the present invention is to provide a pattern inspection method and an apparatus therefor. It is another object of the present invention to provide a method of manufacturing a multilayer substrate capable of manufacturing a multilayer ceramic substrate at a high yield so that it is not necessary to form a repair thin film.

[0007]

In order to achieve the above object, the present invention provides a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductive pattern. On the other hand, 550 nm or less (for example, 400 to
550 nm), and a detector receives an optical image based on reflected light obtained by shielding the polarized component of the polarized illumination light from the illuminated circuit board to be inspected from the illuminated circuit board to be inspected, and forms an optical image signal. And converting the converted optical image signal into a binary signal at a predetermined threshold value to indicate the conductor pattern.
The converted conductor pattern is converted into a digitized image signal, the amount of displacement of the conductor pattern is calculated based on the binarized image signal indicating the converted conductor pattern, and the binarized image signal indicating the converted conductor pattern and the conductor A conductor pattern defect is extracted by aligning and comparing a binary image signal as a reference for determining a pattern defect determination criterion based on the calculated positional shift amount of the conductor pattern. This is a pattern inspection method.

[0008] The present invention also provides a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern, to a thickness of 550 nm or less (for example, 400 to 550 nm). Illuminating polarized illumination light having a wavelength, receiving an optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the illuminated circuit board to be inspected, and converting it into an optical image signal by receiving a detector; The converted optical image signal is binarized at a predetermined threshold value and converted into a binarized image signal indicating the conductor pattern, and a conductor pattern is generated based on the binarized image signal indicating the converted conductor pattern. Is calculated and the size (including variation) of the conductor pattern is calculated, and 2
Positioning the digitized image signal and a binarized image signal as a criterion for determining a defect determination criterion of the conductor pattern matched to the calculated size of the conductor pattern based on the calculated displacement amount of the conductor pattern The method for inspecting a conductor pattern is characterized in that a defect of the conductor pattern is extracted by performing a comparison.

Further, the present invention irradiates polarized illumination light to a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern. An optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board to be inspected is received by a detector and converted into an optical image signal, and the converted optical image signal is converted to a predetermined threshold. The value is converted into a binary image signal indicating the conductor pattern by a value, and the displacement amount of the conductor pattern and the size (variation) of the conductor pattern are determined based on the converted binary image signal indicating the conductor pattern. ), And a binarized image signal indicating the converted conductor pattern and a standardized binarized image signal for determining a defect determination criterion of the conductor pattern according to the size of the calculated conductor pattern. And the above calculation A method of inspecting a conductive pattern and extracting a defect of the conductive pattern by Align compared on the basis of the positional deviation amount of conductors patterns. Further, the present invention illuminates polarized illumination light on a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a substrate in a predetermined conductor pattern, and illuminates the illuminated substrate. The detector receives an optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the inspection circuit board, converts the optical image signal into an optical image signal, and converts the converted optical image signal into first and second different optical signals. The first and second binary signals which are binarized at the threshold value and indicate the conductor pattern
A converted image pattern, and calculates a displacement amount of the conductor pattern based on the first or second binarized image signal indicating the converted conductor pattern; The position of the binarized image signal is compared with the binarized image signal of the first reference which determines the defect judgment criterion of the outer frame with respect to the conductor pattern based on the calculated positional deviation amount, and is compared. A defect of the conductor pattern existing outside the frame is extracted, a second binarized image signal indicating the converted conductor pattern, and a binarized image of a second reference for determining a defect judgment criterion of the inner frame with respect to the conductor pattern. A conductor pattern inspection method characterized in that a defect of a conductor pattern existing in an inner frame is extracted by aligning and comparing a signal with a signal based on the calculated displacement amount.

Further, the present invention illuminates a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material into a predetermined conductor pattern, and irradiates polarized illumination light to the circuit board. An optical image based on the reflected light obtained by blocking the polarized light component of the polarized illumination light from the circuit board to be inspected is received by a detector and converted into an optical image signal, and the converted optical image signal is converted into a first signal different from the first image signal. And a second threshold (TH
a or THh, THb or THl), and the first and second binarized image signals Ga (x, y) or Fa (x, y), Gb (x, y) indicating the conductor pattern.
Or Fb (x, y), and calculating the amount of displacement of the conductor pattern and the size of the conductor pattern based on the first or second binarized image signal indicating the converted conductor pattern, Defect determination of the outer frame with respect to the first binary image signal Ga (x, y) or Fa (x, y) indicating the converted conductor pattern and the conductor pattern adjusted to the calculated size of the conductor pattern The first to determine the criteria (enlarge)
A defect of the conductor pattern existing outside the outer frame is extracted by aligning and comparing the reference binarized image signal Re ′ (x, y) based on the calculated positional deviation amount, and the converted image signal is converted. The second binarized image signal Gb (x, y) or Fb (x, y) indicating the conductor pattern obtained and the defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the calculated size of the conductor pattern. The second (decrease) to decide
A defect of the conductor pattern existing in the inner frame is extracted by aligning and comparing the reference binary image signal Rc ′ (x, y) based on the calculated positional deviation amount. This is a method for inspecting a conductive pattern.

[0011] The present invention also relates to a method for receiving an optical image obtained from a circuit board to be inspected, which is formed by printing a conductive paste containing metal fine particles on a base material into a predetermined conductive pattern, and receiving the optical image with a detector. And converting the converted optical image signal into first and second thresholds (T
(Ha or THh, THb or THl) and convert them into first and second binarized image signals representing the conductor pattern, and at least a reference binarized image signal representing the reference conductor pattern. Expanded, shrunk as it is 3
Forming at least three types of reference binary image signals, and combining the created at least three types of reference binary image signals with the first or second binary image signal indicating the converted conductor pattern. The position shift amount of the conductor pattern and the size of the conductor pattern are calculated by calculating the degree of coincidence by relatively shifting, and the first binarized image signal Ga (x, y) or the first binary image signal indicating the converted conductor pattern is calculated. A first (enlarged) reference binary image signal Re ′ that determines Fa (x, y) and the calculated size of the conductor pattern or a defect determination criterion of the outer frame with respect to the conductor pattern adjusted to the size. (X, y)
Are compared with each other based on the calculated positional deviation amount to extract a defect of the conductor pattern existing outside the outer frame, and a second binarized image signal Gb indicating the converted conductor pattern is obtained. (X, y) or Fb (x,
y) and a second (reduced) reference binarized image signal Rc ′ (x, y) that determines a defect determination criterion of the inner frame with respect to the conductor pattern that is matched with the calculated conductor pattern size.
A defect of the conductor pattern existing in the inner frame by extracting and comparing the position of the conductor pattern based on the calculated positional deviation amount.

Further, the present invention provides a method for receiving an optical image obtained from a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material on a predetermined conductive pattern by a detector to receive an optical image. An image signal, the converted optical image signal is binarized at first, second, and third thresholds (THa, THb, THc) different from each other, and the first and second signals indicate the conductor pattern And a third binarized image signal indicating a defect in the conductor pattern, and the reference binarized image signal indicating the reference conductor pattern is at least expanded and contracted as it is. Three types of reference binary image signals are formed, and the created at least three types of reference binary image signals and a first or second binary image signal indicating the converted conductor pattern are formed. Relative to each other Calculating the magnitude of the displacement amount and the conductor patterns of the conductor pattern by obtaining 致度,
Defects of the outer frame with respect to the first binarized image signal Ga (x, y) or Fa (x, y) indicating the converted conductor pattern and the conductor pattern adjusted to the calculated size of the conductor pattern A defect of the conductor pattern existing outside the outer frame is obtained by aligning and comparing the binarized image signal Re '(x, y) of the first standard for determining the judgment standard based on the calculated positional deviation amount. And a second binarized image signal Gb indicating the converted conductor pattern
(X, y) or Fb (xy) and a second reference binary image signal Rc ′ ( (x, y) is registered based on the calculated positional deviation amount and compared to extract a defect of the conductor pattern present in the inner frame, and to indicate a defect in the converted conductor pattern. 3 binarized image signal Gc
(X, y) and the first reference binary image signal Re ′
(X, y) or a third reference binarized image signal D ′ (x, y) that determines a defect determination criterion matched to the size of the calculated conductor pattern and the calculated positional shift amount The method for inspecting a conductor pattern is characterized in that a defect in the conductor pattern is extracted by performing alignment and comparison based on the pattern.

Further, the present invention is characterized in that in the above-described method for inspecting a conductor pattern, the conductor pattern is a filler formed by filling and printing a conductive paste in a through hole formed in a base material. Further, the present invention is characterized in that in the above-described method for inspecting a conductor pattern, the conductor pattern is a wiring pad formed by printing a conductive paste on a base material.

Further, according to the present invention, there is provided a through-hole forming step of forming a large number of through-holes in a base material, and a conductive paste containing fine metal particles is filled into each of the through-holes formed in the through-hole forming step by printing. A conductor filling step, a conductor pattern filled in the conductor filling step, a conductor pattern inspection step of inspecting by the conductor pattern inspection method, and a metal fine particle on a circuit board determined to be good by the conductor pattern inspection step. A printed wiring forming step of forming a wiring conductor having a wiring pad by printing a conductive paste containing the wiring pad, and a wiring pad inspection step of inspecting the wiring pad formed in the printed wiring forming step by the method for inspecting a conductor pattern. Laminating and sintering circuit boards determined to be non-defective in the wiring pad inspection process to obtain a multilayer board.
And a sintering step.

Further, according to the present invention, a wavelength of 550 nm or less (400 to 550 nm) is applied to a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern. An illumination optical system for illuminating polarized illumination light having: a detector for receiving an optical image based on reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board to be inspected illuminated by the illumination optical system, A detection optical system that converts the optical image signal into an optical image signal, and a binarization unit that binarizes the optical image signal converted by the detector of the detection optical system with a predetermined threshold value and converts the binarized image signal into a binarized image signal indicating the conductor pattern. 2 showing a digitizing circuit and a conductor pattern converted by the binarizing circuit
Calculating means for calculating the amount of displacement of the conductor pattern based on the binarized image signal; a binarized image signal indicating the conductor pattern converted by the binarization circuit; and a criterion for determining a defect determination criterion of the conductor pattern. A conductor for comparing the binarized image signal with the binarized image signal based on the amount of misalignment of the conductor pattern calculated by the calculation unit, and comparing the two with each other to extract a defect in the conductor pattern. This is a pattern inspection device. The present invention also provides an illumination optical system for illuminating polarized illumination light on a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a substrate in a predetermined conductor pattern. A detection optical system for receiving, by a detector, an optical image based on reflected light obtained by blocking a polarization component of the polarized illumination light from a circuit board to be inspected illuminated by an illumination optical system and converting the optical image signal into an optical image signal; The first and second optical image signals converted by the detector of the system are different from each other.
A binarization circuit which binarizes at the threshold value and converts them into first and second binarized image signals indicating the conductor pattern; and a first or second circuit indicating the conductor pattern converted by the binarization circuit. Calculating means for calculating the amount of displacement of the conductor pattern and the size of the conductor pattern based on the second binarized image signal; and a first binarized image showing the conductor pattern converted by the binarization circuit The position shift calculated by the calculating means is the signal and the first reference binary image signal which determines the defect determination criterion of the outer frame with respect to the conductor pattern adjusted to the size of the conductive pattern calculated by the calculating means. A defect of the conductor pattern existing outside the outer frame is extracted by performing alignment and comparison based on the amount, and a second binarized image signal indicating the conductor pattern converted by the binarization circuit and the calculation means Conductor pattern calculated by And a second reference binarized image signal for determining a defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size of the conductor pattern based on the displacement amount calculated by the calculation means. And a comparing means for extracting a defect of the conductor pattern existing in the inner frame.

Further, the present invention provides an illumination optical system for irradiating polarized illumination light to a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern. And a detection optical system that receives an optical image based on reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board to be inspected illuminated by the illumination optical system, and converts the optical image into an optical image signal by a detector. First and second binarized image signals indicating the conductor pattern by binarizing an optical image signal converted by a detector of the detection optical system with first, second, and third thresholds different from each other. And a binarization circuit for converting the signal into a third binarized image signal indicating a defect in the conductor pattern; Type of binary image signal And relatively shifting the created at least three types of reference binarized image signals and the first or second binarized image signal indicating the conductor pattern converted by the binarization circuit. Calculating means for calculating the degree of misalignment of the conductor pattern and the size of the conductor pattern by determining the degree of coincidence, a first binarized image signal indicating the conductor pattern converted by the binarization circuit, and the calculation A first reference binarized image signal for determining a defect determination criterion of the outer frame with respect to the conductor pattern adjusted to the size of the conductor pattern calculated by the means based on the displacement amount calculated by the calculation means; The defect of the conductor pattern existing outside the outer frame is extracted by performing alignment and comparison, and a second binarized image signal indicating the conductor pattern converted by the binarization circuit is calculated by the calculation means. A second reference binarized image signal that determines a defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size of the body pattern is aligned and compared based on the positional shift amount calculated by the calculation unit. To extract a defect of the conductor pattern present in the inner frame, and to convert the third binarized image signal indicating the defect in the conductor pattern converted by the binarization circuit and the first reference binary. A digitized image signal or a third reference binarized image signal that determines a defect determination criterion corresponding to the size of the conductor pattern calculated by the calculation unit based on the positional deviation amount calculated by the calculation unit A conductor pattern inspection apparatus, comprising: comparison means for extracting defects in the conductor pattern by performing alignment and comparison.

Further, the present invention illuminates linearly polarized light with light having a wavelength of blue to ultraviolet light on a circuit board on which a large number of through holes filled with paste are formed, It is configured such that the polarization of the illumination light is shielded and an image from the circuit board is detected upward. With this blue to ultraviolet polarized light illumination / polarized light detection system, the number of bright spots due to the fine metal particles in the paste can be reduced. Further, the detected image is binarized by the first threshold to obtain a binarized image.
The threshold value is a value that separates the paste as the filler from the green sheet surface as the base material. With respect to the detected binary image Ga (x, y) or Gb (x, y), an enlarged or reduced reference pattern image Re (x, y) in which a dead zone is provided at an edge of a reference pattern prepared in advance or By aligning the position with Rc (x, y), a mismatched pixel between the detected binary image and the reference pattern image is extracted. Unmatched pixels are bleeding, scattering, inclusion of foreign matter, and defects without through holes. Also, at the time of alignment, the reference pattern image and the detection 2
If the displacement amount of the value image exceeds the reference value, it is a through-hole displacement defect. As described above, according to the above-described configuration, the shape of the paste filled in the through hole is compared with the reference shape having no defect, instead of determining the defect based on the feature amount such as the area and the perimeter as in the related art. Inspection does not depend on the unevenness of the edge of the paste,
Defects such as bleeding, scattering, inclusion of foreign matter, no through-hole, and misalignment of the through-hole can be detected.

Further, the present invention illuminates linearly polarized light with light having a wavelength of blue to ultraviolet light on a circuit board on which a large number of through holes filled with paste are formed, The polarization of the illumination light is shielded, and an image from the circuit board is detected obliquely upward. The detected image is binarized by a first threshold, and the binarized image Ga (x,
y) or Gb (x, y). First threshold value TH
a or THb is a value that separates the paste as the filler from the green sheet surface as the base material. Also,
The detected image is binarized by the second threshold value THc to obtain a binarized image Gc (x, y). The second threshold value THc is a value that separates pixels corresponding to wall surfaces of through holes that are not filled with the paste (insufficient defects) with respect to the green sheet surface as the base material. Detection binarized image Ga
The amount of displacement between (x, y) or Gb (x, y) and a previously prepared defect-free reference pattern image R (x, y) is obtained, and the detected binary image Gc (x, y) is calculated. Alignment with the reference pattern image R (x, y) is performed. For the detected binary image Gc (x, y), the enlarged reference pattern image Re '
By performing masking using (x, y), an insufficient defect is detected. By using the reference pattern image as a mask, the surface of the green sheet, which is the base material, is outside the inspection target area, and a false report due to the dent of the green sheet does not occur.

As described above, according to the above configuration,
Surplus defects (bleed,
Spatters, protrusions), missing defects (chips, pinholes), missing defects, foreign matter inclusion defects, defects without conductor patterns, displacement errors, etc. can be reliably and rapidly inspected while minimizing false alarms. . Further, according to the above configuration, even if the conductive pattern such as the filling or the wiring pad made of the conductive paste containing the metal fine particles varies from sheet to sheet or within the sheet surface, there is a surplus defect, a defect defect, or a shortage. Defects, foreign matter contamination defects, defects without conductor patterns, position deviation defects, etc.
Moreover, the inspection can be performed at high speed.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of a method for inspecting a filled state of through holes, a pattern inspection method, and an inspection apparatus therefor according to the present invention will be described with reference to the drawings. First, a method and an apparatus for optically detecting a filling material filled in a through hole of a green sheet and a wiring pad printed on the surface will be described. First, in the inspection of the through hole filling material, as shown in FIG. 1, bleeding 4 of the surface of the filling material 3 with respect to the through hole 2 formed in the base material 1 of the green sheet 4 It is necessary to optically inspect for defects such as no through-hole 7, through-hole displacement 8, and insufficient filling 9. By the way, the conductive paste which is a wiring material printed on the through holes 2 and the surface of the green sheet 1 contains metal fine particles. Therefore, when the conductive paste is irradiated with the illumination light to detect the regular reflection light, the regular reflection component of the illumination light is detected from a part of the surface of the metal fine particles, and a bright spot is generated.

Therefore, in order to eliminate bright spots due to regular reflection,
As described in JP-A-5-307006, polarized illumination / polarized light detection is performed. However, polarized light illumination / polarized light detection is extremely effective, but it cannot completely eliminate the bright spot. The reason is that the direction of polarization of the light illuminated on the paste is disturbed due to multiple reflections and the like, and a portion that passes through the polarizing plate on the detection side is present on the paste. Accordingly, the present invention focuses on the spectral reflection characteristics when the metal fine particles are copper (Cu) (in fact, copper is most used as a wiring material), and focuses on green (570).
490 nm) or blue (490-460 nm) or indigo-purple (460-400 nm) by performing polarized illumination / polarized light with a wavelength limited, it is possible to almost completely remove the bright spot. Note that ultraviolet light of 300 to 400 nm may be used as the illumination light, such as excimer laser light. FIG. 2 shows the spectral reflection characteristics of four samples of conductive paste containing copper fine particles (four samples arbitrarily selected, such as a new one and an old one). As shown in FIG. 2, the spectral reflection characteristic of the conductive paste containing the copper fine particles has a lower reflectance as the wavelength becomes shorter from about 550 to 600 nm. For this reason, polarized illumination with a wavelength limited to about 550 nm or less.
By the polarization detection, the brightness of the luminescent spot can be reduced to half or less and detected.

Next, an embodiment of the inspection of the paste filled in the through holes according to the present invention will be described with reference to FIGS. First, a detection image of each filling state will be described with reference to FIGS. From diagonal direction,
When the linearly polarized illumination light 11 is illuminated on the substrate circuit 1 on which the conductive paste containing the copper fine particles is printed, the substrate, that is, the green sheet 1 is composed of fine particles, and is irregularly reflected, and the reflected light is polarized. The direction is disturbed uniformly.
However, since the reflected light from the surface of the paste (metal fine particles) filled in the through holes is specularly reflected light,
The polarization direction is the same as the polarization direction of the polarizing plate 64, as shown in FIG. Therefore, as shown in FIG. 7, by arranging the polarization direction of the polarizing plate 65 so as to be orthogonal to that of the polarizing plate 64, reflected light from the surface of the filler 3 made of Cu metal fine particles can be reflected by the polarizing plate 65. Can be shielded from light. For this reason, when the detection is performed with the detection direction 12 substantially vertical, the filling 3 can be detected dark and the substrate 1 can be detected bright as shown as an image signal in FIG. That is, as shown in FIG. 3, the detector 63 outputs a clear detection image signal G having a black-and-white contrast in which the image portion corresponding to the filler 3 is black and the image portion corresponding to the substrate 1 is white.
(X, y) is obtained. Then, the detected image signal G
By binarizing (x, y) with the threshold value THa and converting it into a detected binary image signal Ga (x, y), the filler 3 and the substrate 1 can be separated. Based on the detected binary image signal Ga (x, y), the image portion corresponding to the filling 3 is largely detected for the blurred defect 4 with respect to the normal filling state, and the defect 7 without the through hole is detected. Is
There is no image portion corresponding to the filling 3, and the scattering filling 5 is detected for the scattering defect 5.

Regarding the foreign matter adhesion / mixing defect 6, in reality, there are many examples of green sheet debris, and since many sheet debris are thin, they often do not shine sufficiently brightly. Therefore, the foreign matter defect 6 can be identified without being overlooked by binarizing with the threshold value THb (<THa) to obtain the detected binary image signal Gb (x, y). For the insufficient filling defect 9, the illumination light 11 is obliquely incident on the wall surface of the through-hole 2, so that the green sheet surface near the through-hole wall surface has a green color near the normally filled through-hole. It is detected brighter than the sheet surface. Therefore, the threshold value THc (> THa) is used for binarization and detection 2
By obtaining the quantified image signal Gc (x, y), it is possible to identify the through hole of the insufficient filling defect 9. In addition,
As described in Japanese Patent Application Laid-Open No. 5-307006, when the obstruction is detected from obliquely above, the insufficient filling defect 9 can be detected brighter.

Next, the detected binary image signal G according to the present invention will be described.
FIG. 4 shows an embodiment of a defect determination algorithm for extracting a defective portion from a (x, y), Gb (x, y), and Gc (xy).
This will be described with reference to FIG. In each of the three binarization circuits 20a, 20b, 20c,
Is binarized by each of the threshold values THa, THb, and THc to obtain three binarized image signals Ga (x, y) and Gb (x, y). ), Gc (x,
y) is obtained. As described above, in the binarization circuit 20c, the detected image signal G (x, y) is binarized by the threshold value THc to obtain the detected binarized image signal Gc (x, y). The vicinity of the through hole wall surface of the defect 9 is sorted out. That is, as the binary image signal Gc (x, y), a signal in which the bright portion near the wall surface of the through hole is “1” and the other portions are “0” is obtained. Although the generation of bright spots in the paste portion is suppressed by polarized light illumination / polarized light detection of blue or the like whose wavelength is shortened to about 550 nm or less, the bright spot removing circuit 23c considers the case where bright spots are generated by any chance. , A small area of "1" is erased, and a binarized image signal Gc '(x, y) from which the bright spot is removed is obtained. In the binarization circuit 20a, the detected image signal G (x, y) is binarized with the threshold value THa to obtain the detected binarized image signal Ga (x, y), so that the base, that is, the filler 3 is removed. Be separated. That is, as the binarized image signal Ga (x, y), a signal in which the paste portion is “0” and the green sheet portion is “1” is obtained. In order to remove minute scattering that is not a defect, a minute "0" region is erased by an isolated point removing circuit 23a, and a binarized image signal Ga '(x, y) from which isolated points are removed.
Get.

In the binarizing circuit 20b, the detected binary image signal Gb (x, y) obtained by binarizing the detected image signal G (x, y) with the threshold value THb has a binary value of the threshold value THa. Similarly to the image signal Ga (x, y), the paste portion is “0”,
The difference is that the green sheet portion becomes “1”, but the sheet waste (foreign matter) becomes “1”. Therefore, in the bright spot removing circuit 23b, the non-defective minute “1” area is erased, and the binary image signal Gb ′ (x,
y) is obtained. On the other hand, the reference pattern R (x,
Prepare y). The reference pattern R (x, y) may be generated from the design data, or may be created by drawing an image of a non-defective sample.
The base portion is formed by the binary image signal of “1”. The reference pattern R (x, x) at the same location as the detected image G (x, y)
y), the position of the detected image signal G (x, y) and the position of the reference pattern image signal R (x, y) are accurately aligned due to manufacturing variations of the substrate 1 to be inspected. There may not be. For this reason, the detected binary image signal Ga
Alignment is performed by calculating the amount of displacement between (x, y), Gb (x, y), Gc (x, y) and the reference pattern image signal R (x, y). That is, the mismatch counter circuit 75
An image signal read from the image memory 74 by shifting the reference pattern image signal (positioning pattern image signal) R (x, y) by ± several pixels in the x and y directions (for example, if the image signal is shifted by ± 3 pixels, 7 × 7 = 49 binary images)
And a detected binarized image signal (for example, an image binarized by the threshold value THa or the threshold value THb) Ga ′ (x, y)
Alternatively, the number of pixels that do not match Gb ′ (x, y) is counted. The through-hole position shift detection circuit (PC) 76 converts the reference pattern image having the minimum number of mismatched pixels counted by the mismatch counter circuit 75 into the detected binary image Ga ′ (x,
y) or Gb ′ (x, y) and the reference pattern image signal R (x, y).
Binarized image signal Ga '(x, y) or G
The position shift amount of the through hole at b ′ (x, y) is calculated. Therefore, the through-hole position shift detection circuit (PC) 76 outputs the image signal Re '(x, x) indicating the permissible limit stored in each of the image memories 83,84,85.
y), Rc ′ (x, y) and D ′ (x, y) are read out from the image memory 2 by performing position shift correction based on the calculated through-hole position shift amount 111.
4a, 24b, and 24c, the detected binary image signals Ga '(x, y), Gb' (x, y), G
c ′ (x, y) and an image signal Re ′ (x, y) indicating an allowable limit read from each of the image memories 83, 84, 85.
y), Rc ′ (x, y) and D ′ (x, y) will be aligned.

The dimensions of the through-holes 2 formed in the green sheet substrate 1 slightly vary, and as a result, the diameter of the filler 3 also varies. Therefore,
The detected image signal G (x,
y) is detected and the detected binary image signal Ga ′ (x,
y), Gb ′ (x, y), and Gc ′ (x, y) also show variations in the size of the filler 3. Therefore, allowable determination criteria Re ′ (x, y) and Rc ′ that can be determined based on the reference pattern indicating the filler 3 in the reference pattern image signal R (x, y) and the determination pattern signal D (x, y).
(X, y) and D ′ (x, y) are converted to the detected binary image signal G
a '(x, y), Gb' (x, y), Gc '(x, y)
It is necessary to match the fluctuation of the size of the packing 3 appearing in the above. For this reason, the expansion circuit 71a forms a binarized image signal obtained by expanding the reference pattern image signal R (x, y) by 1 to 2 pixels corresponding to the one in which the diameter of the filler 3 is larger, and stores it in the memory. 74. Similarly, the expansion circuit 71b,
Each of 71c and 71d is an enlarged reference pattern image signal Re (x, y) indicating an allowable limit such as bleeding or scattering defect, and a reduced reference pattern image signal Rc (x) indicating an allowable limit such as no through-hole or sheet dust defect. , Y), the judgment pattern image signal D (x, y) indicating the permissible limit of the insufficient filling defect or the like
1 to 2 corresponding to the larger diameter of the packing 3
A binarized image signal in which pixels are expanded is formed, and the memories 83, 84, and 8 are switched via switching circuits 82b, 82c, and 82d.
5 is stored. The through circuit 72a stores the reference pattern image signal R (x, y) in the memory 74 as it is.
To memorize. Similarly, the through circuits 72b, 72c, 7
Each of 2d is an enlarged reference pattern image signal Re (x, y) indicating an allowable limit such as bleeding or scattering defect, and a reduced reference pattern image signal Rc (x, y) indicating an allowable limit such as no through hole or sheet dust defect. ), The judgment pattern image signal D (x, y) indicating the permissible limit of the insufficient filling defect or the like is directly transferred to the memory 8 via the switching circuits 82b, 82c, 82d.
3, 84, and 85 are stored. As described above, since the through circuits 72a to 72d output the input image signals as they are, they do not exist as circuits.

The shrinking circuit 73a forms a binarized image signal obtained by shrinking the reference pattern image signal R (x, y) by one or two pixels corresponding to a smaller filling material 3 diameter. Stored in the memory 74. Similarly, the contraction circuit 73
b, 73c, and 73d represent enlarged reference pattern image signals Re (x,
y), a reduced reference pattern image signal Rc (x, y) indicating an allowable limit such as no through hole or sheet dust defect, and a determination pattern image signal D indicating an allowable limit such as a defective filling defect.
With respect to (x, y), a binarized image signal is generated by shrinking one to two pixels corresponding to a smaller diameter of the filler 3 and the memory 8 is switched via the switching circuits 82b, 82c, 82d.
3, 84, and 85 are stored. Therefore, the image memory 74 provides a binarized image signal in which the diameter of the filler is expanded to be larger, a binary image in which the diameter of the filler is contracted to be smaller, with respect to the reference pattern image signal R (x, y), Three types of alignment patterns are generated without changing the reference pattern.
Then, the non-coincidence counter circuit 75 generates an image shifted to ± several pixels by reading the image memory 74 from the image memory 74 while shifting the pixel to ± several pixels in the x and y directions (for example, when the image is shifted to ± 3 pixels). , 1 in total
47 binarized images), detected binarized image Ga ′ (x,
y), and check the positional deviation from Gb ′ (x, y). Accordingly, the through-hole position shift detection circuit (PC) 76 determines the through-hole position shift amount indicating the minimum number of mismatches obtained from the mismatch counter circuit 75 and the type of the alignment pattern (the filling in the detected image signal G (x, y)). Objects 3 having a large diameter, a normal diameter, and a small diameter 3 are detected. Therefore, as described above, the through-hole position shift detection circuit (PC) 76 outputs the image signals Re ′ (x, y) and Rc ′ (x, x) indicating the permissible limit stored in each of the image memories 83, 84, 85. y),
When D ′ (x, y) is read, the detected binary image signal read from each of the image memories 24 a, 24 b, and 24 c is corrected by correcting the position shift based on the calculated through-hole position shift amount 111. Ga '(x,
y), Gb '(x, y), Gc' (x, y) and image signals Re '(x, y), Rc' (x) indicating the permissible limit read from each of the image memories 83, 84, 85. , Y),
D ′ (x, y) can be aligned.
Further, the through-hole displacement detection circuit (PC) 76
Filling material 3 in detected image signal G (x, y)
The switching circuit 82 is controlled by a signal 112 corresponding to three types, that is, a large one, a normal one, and a small one.
b, 82c, 82d, the detection 2 stored in each of the image memories 24a, 24b, 24c.
Binarized image signals Ga '(x, y), Gb' (x, y), G
The image signals Re '(x, y), Rc' (x,
y) and D '(x, y) are stored in image memories 83 and 84,
85 can be stored.

Thus, even if the size of the filler varies from sheet to sheet or within the sheet surface,
Certainly, the detected binary image signals Ga '(x, y), Gb'
(X, y) and Gc ′ (x, y) are replaced with image signals Re ′ (x, y), Rc ′ (x, y), and D ′ indicating allowable limits based on the reference pattern image signal and the determination pattern image signal. (X,
y) and the detected binary image signals Ga ′ (x, y), Gb ′ (x, y), Gc ′
Image signals Re ′ (x, y), Rc ′ (x,
y) and D '(x, y) can be selected. When the calculated through-hole position shift amount is equal to or larger than a specified value, the through-hole position shift detection circuit (PC) 76 determines that the position is a through-hole position shift defect 106 and outputs it. The expansion in the expansion circuits 71a to 71d and the contraction in the contraction circuits 73a to 73d are simple, for example, when an operator of 4 (or 8) 3 × 3 pixels is operated. As shown in FIG. 5A, if there is “0” corresponding to the conductive paste in any of the 4 × or 8 × 3 × 3 pixels, the center pixel is changed from “1” to “0”. As a result, the outer periphery of the binarized pattern of the conductive paste indicated by “0” is expanded by one pixel. Also, as shown in FIG.
If there is a “1” corresponding to the green sheet surface in either 4 or 8 connection of 3 × 3 pixels, the conductivity indicated by “0” is obtained by converting the center pixel from “0” to “1”. The outer periphery of the binarized pattern of the conductive paste is contracted by one pixel. If it is desired to further expand or contract one pixel, the operator of 4 connection (or 8 connection) of 3 × 3 pixels may be applied again.

Next, an image signal Re 'indicating an allowable limit is shown.
(X, y), Rc ′ (x, y), and D ′ (x, y) as the basis of the enlarged reference pattern image signal Re (x, y),
The reduced reference pattern image signal Rc (x, y) and the determination pattern image signal D (x, y) will be described. The enlargement processing circuit 29 determines an inspection criterion (for example, as shown in FIG. 6A, the upper limit Dmax of the size of the bleeding for the through-hole, with respect to the reference pattern image signal R (x, y). The upper limit Dmax of the range in which the scattering must not occur as shown in (1) is determined. 78. The reduction processing circuit 30 outputs the reference pattern image signal R
For (x, y), an inspection criterion (for example, FIG. 6C)
As shown in (1), the lower limit Dmin of the range in which there is no through-hole with respect to the through-hole or where foreign matter such as sheet waste exists is determined. ) To reduce the size of the reference pattern image signal Rc (x,
y) is formed and stored in the memory 80. The determination pattern image signal D (x, y) serves as an inspection area mask for determining a filling defect, and can be set by storing an area pattern in which an insufficient defect can be detected in the memory 86. it can.

As described above, each of the enlarged reference pattern image signal Re (x, y), the reduced reference pattern image signal Rc (x, y), and the determination pattern image signal D (x, y) serving as the inspection determination reference. Can be stored in each of the memories 78, 80, and 86 and acquired. Further, as shown in FIG. 4B, the edge extraction circuit 87 applies a contour to the enlarged reference pattern image signal Re ′ (x, y) adjusted to the size of the filler of the detection image signal read from the memory 83. The enlarged edge reference pattern image signal R obtained by extracting a portion (edge portion) and setting the contour portion (edge portion) to “0”
Form e'E (x, y). Similarly, the edge extraction circuit 88 outputs the reduced reference pattern image signal R adjusted to the size of the filler of the detected image signal read from the memory 84.
A contour portion (edge portion) is extracted from c ′ (x, y) to form a reduced edge reference pattern image signal Rc′E (x, y) in which the contour portion (edge portion) is set to “0”. . Then, as shown in FIG. 4B, in a comparison circuit (logical product circuit) 31a, a signal obtained by inverting the detected binary image signal Ga ′ (x, y) binarized by the threshold value THa, and an enlarged reference pattern By comparing the image signal Re '(x, y) with, for example, a logical product, the blur defect 102a and the scattering defect 102b exceeding the allowable limit are detected as "0", for example. In FIG. 4B, the defect is detected as “0” for convenience of display, but may be detected as “1”. Further, in the comparison circuit (logical product circuit) 31d, the detected binary image signal Ga '(x, y) or the defect signal 10
3 is compared with a signal obtained by inverting the enlarged edge reference pattern image signal Re'E (x, y), for example, by taking a logical product, the relationship between the blur defect 102a and the through hole is obtained. The defect signal 102a ′ existing on the enlarged edge is set to, for example, “0”, indicating a blur defect and a scattering defect.
And is detected separately. That is, in the comparison circuit (logical product circuit) 31d, the signal indicating the scattering defect located outside the enlarged edge is erased, and the blurred edge defect 102
a 'will be detected separately. Accordingly, the defect signal 102 detected from the comparison circuit (logical product circuit) 31a
Is compared with the defect signal 104 detected from the comparison circuit (logical product circuit) 31d to determine whether the bleeding defect 10
2a and the scattering defect 102b can be output separately. When the scattering defect 102b exists on the enlarged edge, the scattering defect 102b is blurred by the defect 10b.
It will be classified in the same way as 2a.

Further, in the comparison circuit (logical product circuit) 31b, the detected binary image signal Gb '(x, y) binarized by the threshold value THb and the reduced reference pattern image signal Rc'
By comparing the signal obtained by inverting (x, y) with, for example, a logical product, the defect 103a without a through-hole and the foreign matter defect 103b such as sheet dust on the filler are detected as "0", for example. In FIG. 4B, the defect is detected as “0” for convenience of display, but may be detected as “1”. Further, in the comparison circuit (logical product circuit) 31e,
Detected binary image signal Gb ′ (x, y) or defect signal 1
02 and the reduced edge reference pattern image signal Rc'E
By comparing with a signal obtained by inverting (x, y) and taking a logical product, for example, a signal indicating a foreign substance such as a sheet dust on the filler is erased, thereby indicating that there is no through hole present on the reduced edge. The edge signal 103a 'is detected as, for example, "0" separately from the defect signal 103 indicating a defect without a through-hole and a foreign matter defect. Therefore, the defect signal 103 detected from the comparison circuit (logical product circuit) 31b
And the defect signal 105 detected from the comparison circuit (logical product circuit) 31e, it is possible to separately output the defect 103a without through-hole 103a and the defect 103b shown in FIG. 6C. Becomes In this case, when the foreign matter defect 103b exists on the reduced edge, the foreign matter defect 103b is replaced with the through-hole-free defect 1b.
It will be classified in the same way as 03a. By the way, in the case of a defect without a through-hole, a signal of the reduced reference pattern image signal Rc ′ (x, y) itself is output from the comparison circuit 31b. Therefore, the comparison circuit (logical product circuit)
At 31e ', a signal obtained by inverting the defect signal 103 obtained from the comparison circuit 31b and the reduced reference pattern image signal R
By comparing the signal with the signal obtained by inverting c '(x, y) and checking the coincidence (for example, by taking a logical product), a signal indicating a defect without a through-hole where no inconsistent portion is generated is erased and an inconsistent portion is generated. A signal indicating the foreign matter defect 103 is output. Therefore, the comparison circuit (AND circuit) 3
By comparing the defect signal 103 detected from 1b with the defect signal 105 'detected from the comparison circuit (logical product circuit) 31e', the defect 103a without through-hole is compared.
And defect output 103b. Further, since the feature amount such as the area differs due to the difference between the shape and the sheet waste without the through hole, the feature amount such as the area for each defect pattern is calculated with respect to the defect signal 103 so that there is no through hole. Defect 103a
And the sheet dust defect 103b can be output separately.

Further, in the comparison circuit (logical product circuit) 31c, the detected binary image signal Gc '(x, y) binarized by the threshold value THc and the enlarged reference pattern image signal Re'
By comparing the signal obtained by inverting (x, y) with a logical product, for example, the insufficient filling defect 101 becomes, for example, “0”.
Is detected as In FIG. 4B, the defect is detected as “0” for convenience of display, but may be detected as “1”. At this time, the enlarged reference pattern image signal Re ′
Instead of (x, y), it is conceivable to use a crescent-shaped or half-moon-shaped judgment pattern image signal D '(x, y). The judgment pattern is larger than the enlarged reference pattern.
Since the pattern area is reduced, there is an effect of reducing false reports that a bright point of the paste is erroneously determined to be a defect with insufficient filling. In addition,
Detected binary image signal Gc ′ binarized by threshold value THc
(X, y) and an enlarged reference pattern image signal Re ′ (x,
After the logical product of y) and the inverted signal, the shrinkage circuit 89 shrinks the “0” portion, so that a shallow insufficient defect can be erased and not detected.

Next, a first embodiment of an apparatus for inspecting a paste (filling material) filled in a through hole according to the present invention will be described with reference to FIG. An X stage 53 is mounted on the stage base 52, and a Y stage 54 is mounted on the X stage 53, so that the XY stages 53 and 54 can be independently moved in the X and Y directions. Is constituted. On the Y stage 54, a circuit board (green sheet substrate 1 on which the conductive paste 3 is printed) 51 to be inspected is positioned and mounted. The X stage 53 and the Y stage 54 are respectively moved in the X and Y directions by being driven by an X axis driving unit 55 and a Y axis driving unit 56 as external driving devices. Further, an X stage 53 and a Y stage 54
Each position is measured by an X-axis length measuring device 57 and a Y-axis length measuring device 58. On the other hand, the inspection position on the circuit board 51 is illuminated from an oblique direction. The light generated by the fiber light source 60 is guided by a light guide 59, and is transmitted by a wavelength limiting filter 67 to a wavelength of 400 to 55.
0 nm, and as a linearly polarized illumination light by the polarizing plate 64 in the optical path, through the condenser lens 61, the circuit board 51
Illuminate the inspection location above. The image at the inspection position is detected by a line sensor 63 from above vertically through a polarizing plate 65 and a detection lens 62. At that time, the polarizing plate 6
5 is provided so that the polarizing direction is orthogonal to the polarizing plate 64. Illumination systems 61, 64, 67 and image detection systems 62, 63,
The plane formed by the 65 optical axes is set perpendicular to the detection target surface, and the image detection area of the line sensor 63 is also set perpendicular to the plane. The line sensor 63 is characterized in that it has a high sensitivity.
Delay and Integration) line sensors are preferred.
The order of the filter 67, the polarizing plate 64, and the condenser lens 61 may be interchanged, and the detection lens 62 and the polarizing plate 65 may also be interchanged. The fiber light source 60 is preferably a high-brightness xenon lamp or a mercury lamp.

With the above configuration, the circuit board 51 is translated in a direction perpendicular to the detection longitudinal direction of the line sensor 63, that is, the X stage 53 is driven, and the circuit board 51 is synchronized based on the output signal of the X axis length measuring device 57. By generating a pulse, an image signal is obtained from the line sensor 63. The image signal of the line sensor 63 is A / D-converted by the image detection circuit 66 to become a multi-valued digital image signal. Further, the correction circuit 68 performs dark level correction, stage speed fluctuation correction, and shading correction, and corrects variations in dark current of elements, sensitivity unevenness, illumination unevenness, and brightness unevenness due to stage speed unevenness. Thereafter, binarization processing is performed by the binarization circuits 20a, 20b, and 20c to obtain a binary image signal. Note that each of the threshold values THa, THb, and THc is determined based on the histogram of the detected multilevel image. The multivalued image is taken into the control PC 22 via the image input board 21 and
C22 calculates the histogram and calculates each threshold TH
a, THb and THc are determined. The determined threshold values THa, THb, and THc are used as the respective binarization circuits 20a and 20a.
b, 20c. Each binarization circuit 20a,
The detected binary image signal Ga obtained from 20b and 20c
The bright point / isolated point removing circuits 23a, 23b, 23c remove the bright points or the isolated points from (x, y), Gb (x, y), and Gc (x, y). For example, the bright point / isolated point can be removed using the operator shown in FIG. n × n pixels (FIG. 8 shows a case where n = 5)
When all the shaded pixels are “1”, the operator can remove the bright spot by setting the value of the internal pixels to “1”. Similarly, all the hatched pixels are “0”.
In the case of, the isolated point can be removed by setting the value of the internal pixel to “0”.

Detected binary image signals Ga '(x, y), Gb' (x, y), Gc '(x,
y) is temporarily stored in the image memories 24a, 24b, 24c. The image memory 24 includes two memories each having a capacity enough to store an image for one scan of the stage (the number of pixels of the line sensor × the detection length in the stage traveling direction).
Used as a double buffer. While storing the detected image for one scan on one surface, the image detected by the previous scan on the other surface is read out, and the subsequent matching processing is performed. On the other hand, the data processing PC 25 reads the compressed reference pattern stored in the hard disk and sends the data to the decompression circuit 26. The data processing PC 25 communicates with the control PC 22 via a network, and operates according to an instruction from the control PC 22. The decompression circuit 26 decompresses the compressed data, generates a reference pattern image signal R (x, y) to be compared with the detected binary image signal, and stores it in the image memory 27. The compression method may be any reversible compression method, for example, a run-length encoding method. Image memory 2
7 has two memories each having a capacity enough to store an image for one scan of the stage (the number of pixels of the line sensor × the detection length in the stage traveling direction), similarly to the image memory 24.
Used as a double buffer. While storing the expanded reference pattern at the same location as the detected binary image on one side, the image is read out on the other side to perform the subsequent matching processing. In addition,
The expansion circuit 71a and the through circuit 72 apply the reference pattern image signal R (x, y) read from the image memory 27.
a, the through hole is made larger by the size matching circuit consisting of the contraction circuit 73a,
Three types of reference pattern image signals, which are made smaller, are created and stored in the image memory 74 as alignment patterns.

Based on the cut-out address signal 110 from the control PC 22, the detected binarized image signal Ga '(x, y) is cut out from the image memory 24a for each local area, and at the same time, three sizes of the image data are read out from the image memory 74. Binary binary image signal R (x, y) having the same
The position shift amount is obtained by a position shift detection circuit 28 including a mismatch counter circuit 75 and a through-hole position shift detection circuit 76. That is, the position shift detection circuit 28 applies the reference pattern binarized image signal R (x, y) in the x and y directions to the detected binarized image signal Ga ′ (x, y), The number of mismatched pixels is counted while shaking the size of. Then, the position shift detection circuit 28 compares the numbers of mismatched pixels and determines the position shift amount 111 and the size 112 of the pattern (through-hole pattern) at the same time from the displacement in the x and y directions up to the minimum. know. At this time, the position shift detection circuit 28 determines that the position shift between the detected binary image signal and the reference pattern binary image signal is large and exceeds the measurement range (in the x and y directions of the reference pattern binary image). If the displacement exceeds the specified width, the read address of the image memory 74 is shifted, the number of unmatched pixels is calculated again, and the minimum value is obtained.
It is determined as the through-hole position shift defect 106 and stored in the determination result memory 32 according to the coordinate address measured by the coordinate length measurement circuit 35. In addition, the displacement detection circuit 28
Is configured only by the mismatch counter circuit 75, and the control PC 22 compares the numbers of mismatch pixels based on the number of mismatch pixels counted by the mismatch counter circuit 75, and determines the smallest one in the x and y directions up to the smallest one. From displacement, displacement amount 111
At the same time, the size 112 of the pattern (through-hole pattern) may be known. Further, the position shift detection circuit 28 shifts the read address of the image memory 74 when the position shift between the detected binary image signal and the reference pattern binary image signal is large and exceeds the measurement range, and again determines the number of mismatched pixels. Is calculated, and the control PC 22 obtains the minimum value. When the obtained positional deviation exceeds a specified value,
The determination may be made as the through-hole position shift defect 106 and stored in the determination result memory 32 in accordance with the coordinate address measured by the coordinate measuring circuit 35.

The enlargement processing circuit 29 enlarges the reference pattern to a size determined by the inspection judgment criterion, and generates an enlarged reference pattern binary image signal Re (x, y) for defect judgment. The reduction processing circuit 30 performs reduction processing on the reference pattern to a size determined by the inspection determination reference, and generates a reduced reference pattern binary image signal Rc (x, y) for defect determination.
At this time, the displacement detection circuit 28 or the control PC 22
Is based on the size 112 of the detected pattern obtained by the alignment process between the detected binary image and the reference pattern binary image,
One of the size matching circuits including the expansion circuit 71b, the through circuit 72b, and the contraction circuit 73b is selected, and the expansion circuit 71
It is also conceivable that by selecting any one of the dimension matching circuits composed of c, the through circuit 72c, and the contraction circuit 73c, the amount of expansion and contraction is adjusted to match the size of the detection pattern. That is, if the pattern of the detected image is large, the criterion is set large, and if the pattern of the detected image is small, the criterion is set small. The detected binary image signal Gc ′ (x, y) read from the image memory 24 c and the enlarged reference pattern binary image signal Re ′ for defect determination created by the enlargement processing circuit 29.
The comparison circuit 31c compares the shape with (x, y), detects the insufficient filling defect 101, and determines the determination result according to the coordinate address measured by the coordinate measuring circuit 35 and the address of the defect position in the image signal. It is stored in the memory 32.

The detected binary image signal Ga '(x, y) read from the image memory 24a and the enlarged reference pattern binary image signal R for defect determination created by the enlargement processing circuit 29
e ′ (x, y) is compared in shape by a comparison circuit 31a,
The bleeding, scattered defect 102 is detected, and the determination result memory 32 is determined according to the coordinate address measured by the coordinate measuring circuit 35.
To memorize. Detection 2 read from image memory 24b
The comparison circuit 31b compares the shapes of the digitized image signal Gb '(x, y) and the reduced reference pattern binary image signal Rc' (x, y) created by the reduction processing circuit 30 for defect determination. The defect 103 having no foreign matter and no through hole is detected and stored in the determination result memory 32 according to the coordinate address measured by the coordinate measuring circuit 35. At this time, it is natural that the comparison circuits 31a to 31c compare the shapes in consideration of the displacement amount obtained by the displacement detection circuit 28. Defect information (position, type) obtained by each of the comparison circuits 31a to 31c
Are sequentially stored in the determination result memory 32, and the defect information is sent to the control PC 22 collectively for each scanning of the stage or at the end of one substrate inspection. The control PC 22 can output the defect information to a printer or a monitor, or can output the defect information to a host computer via a network.

Next, a second embodiment of the inspection apparatus for the paste (filling material) filled in the through holes according to the present invention will be described with reference to FIG. In the second embodiment, the illumination optical systems 61c, 64c, 67c and the detection optical systems 62b, 63b for detecting missing defects are made independent, and detection is performed diagonally from above, thereby increasing the detection sensitivity of missing defects. Since the number of detection optical systems is two, the number of defect determination processing circuits is also two. Except for the two systems, the configuration and operation of the apparatus are the same as those of the first embodiment shown in FIG. However, the illumination optical systems 61a, 64a, 67a for the detection optical systems 62a, 63a, 65; 61b, 64b, 67b
Were configured to face each other. The determination algorithm of each system will be described with reference to FIG. 10 and FIG. 10A and 10B show, as in FIGS. 4A and 4B, scattering, bleeding, inclusion of foreign matter, no through-hole, and through-hole displacement defects 102, 103, 104, and 10 based on image signals detected from above by the line sensor 63 a.
5 and 106 are detected. Note that the data processing PC is 25a, the decompression circuit is 26a, the image memory is 27a, and the displacement detection circuit is 28a. FIG. 10 shows through-hole defects and sheet dust defects 103.
In order to separate the sheet dust defect 103b from the signal, the logical product of the signal obtained by inverting the defect signal 103 by the comparison circuit (logical product circuit) 31e 'and the signal obtained by inverting the reduced reference pattern image signal Rc' (x, y) is obtained. Indicates a negative case. FIG.
An algorithm 1 detects an insufficient filling defect 101 from an image signal detected obliquely from above by the line sensor 63b. That is, the detected image signal G ′ (x, y) detected by the line sensor 63b is
0c, the detected binary image signal Gc is detected by the threshold value THc.
(X, y), and further, the bright point is removed by the bright point removing circuit 23c, and the detected binary image signal Gc ′ (x,
y) is stored in the image memory 24c. On the other hand, the detected image signal G '(x, y) is converted into a detected binary image signal Ga (x, y) for detecting a position shift by the threshold value THa in the binarization circuit 20d, and is converted to the image memory 24d. It is memorized.

The data processing PC 25b reads the compressed reference pattern stored in the hard disk and sends the data to the decompression circuit 26b, and also reads the compressed determination pattern stored in the hard disk and sends the data to the decompression circuit 26c. Send. The decompression circuit 26b decompresses the compressed data, generates a reference pattern image signal R (x, y) for detecting a displacement, and stores the generated signal in the image memory 27b. The expansion circuit 26c expands the compressed data, generates a reference pattern image signal D (x, y) to be compared with the detected binary image signal Gc ′ (x, y), and generates an image memory 27.
c. The reference pattern image signal R (x, y) stored in the image memory 27b has a through-hole made larger by a dimensional matching circuit including an expansion circuit 71a, a through circuit 72a, and a contraction circuit 73a. The image memory 7 generates three types of reference pattern image signals, which are made smaller, and as an alignment pattern.
4 is stored. Then, based on the cut-out address signal 110 from the control PC 22, the detected binary image signal Ga (x, y) is cut out from the image memory 24d for each local area, and at the same time, three kinds of sizes are read out from the image memory 74. The reference binarized image signal R (x, y) is cut out for each of the same regions, and the position shift amount is obtained by a position shift detection circuit 28b including a mismatch counter circuit 75 and a through-hole position shift detection circuit 76. That is, the displacement detection circuit 28b
Indicates that the reference pattern binarized image signal R (x, y) is displaced in the x and y directions with respect to the detected binary image signal Ga (x, y) while the size of the reference pattern is displaced. Count the number of pixels. Then, the displacement detection circuit 28
b compares the number of unmatched pixels,
From the displacements in the x and y directions,
The size 112 of the pattern (through-hole pattern) can also be known.

When reading the determination pattern image signal D (x, y) indicating the permissible limit stored in the image memory 27d, the position shift detecting circuit 28b is based on the calculated through-hole position shift amount 111. By performing the position shift correction, the detected binary image signal Gc ′ (x, y) read from the image memory 24c and the image memory 2
The determination pattern image signal D (x, y) indicating the permissible limit read from 7d can be aligned. Further, the position shift detection circuit 28b determines whether the diameter of the through hole 2 (filler 3) in the detected image signal G ′ (x, y) is large, normal, or small. By switching the switching circuit 82d with the signal 112 corresponding to the type, an allowable limit suitable for the size of the through hole 2 appearing in the detected binary image signal Gc '(x, y) stored in the image memory 24c is indicated. The determination pattern image signal D (x, y) can be stored in each of the image memories 85. Thereby, even if the size of the filler varies from sheet to sheet or within a sheet surface, the detected binary image signal Gc ′ is surely ensured.
(X, y) can be aligned with the image signal D (x, y) indicating the permissible limit based on the judgment pattern image signal, and the filling appearing in the detected binary image signal Gc ′ (x, y) can be obtained. It is possible to select an image signal D (x, y) indicating an allowable limit suitable for the size of the object 3.

Next, in the comparison circuit (logical product circuit) 31c, the detected binary image signal Gc '(x, y) binarized by the threshold value THc and the crescent-shaped or half-moon-shaped judgment pattern image By comparing the signal D '(x, y) with the inverted signal and taking, for example, a logical product, the insufficient filling defect 101 is detected as "0", for example. In FIG. 11, the defect is detected as “0” for convenience of display.
It may be detected as "1". Since the determination pattern has a smaller pattern area than the enlarged reference pattern, it has an effect of reducing false reports that a bright point of the paste is erroneously determined to be a defect with insufficient filling. Note that the detected binary image signal Gc ′ (x,
After performing a logical product of y) and a signal obtained by inverting the determination pattern image signal D ′ (x, y), the shrinkage circuit 89 shrinks the “0” portion, thereby eliminating the shallow insufficient defect and not detecting it. It is also possible to do so.

Next, an embodiment of an inspection of a wiring pad formed by printing a conductive paste containing metal fine particles on the surface of a green sheet according to the present invention will be described with reference to FIGS. 12 and 13. FIG. FIG. 12 shows a defect target in a wiring pad formed by printing a conductive paste containing metal fine particles. Defects include the wiring pads 12
There are a pinhole 121, a defect 122, a protrusion 123, and a scatter 124 in 0. The pinhole 121 and the defect 122 are concave defects, and the protrusion 123 and the scattering 124 are convex defects. For a detected image F (x, y) detected from such a wiring pad 120, a concave defect determination pattern Rc (x, y) and a convex determination pattern Re (x, y) determined based on the inspection criterion are used. By fitting, the pinhole 14 inside the concave defect determination pattern Rc (x, y) is obtained.
2b and the defect 142a are determined as defects, and the scattering 141b and the protrusion 1 outside the convex determination pattern Re (x, y) are determined.
41a will be determined as a defect. Next, a defective part is extracted from the wiring pad according to the present invention from the detected binary image signal detected by the illumination optical system and the detection optical system shown in FIG. 14 which are similar to the illumination optical system and the detection optical system shown in FIG. FIG. 1 shows an embodiment of a defect determination algorithm to be performed.
3 will be described. Detected image signals F (x, y) obtained by imaging the substrate 51 to be inspected are converted into binary circuits 20a 'and 20a.
In each of the two threshold values THh and THl,
Into two values, two binary image signals Fa (x, y), Fb
(X, y) is obtained. By binarizing with the threshold value THh, the base, that is, the wiring pad is separated. 2 where the paste part is “0” and the green sheet part is “1”
Two binary image signals Fa (x, y) and Fb (x, y) are obtained. In order to remove minute scattering that is not defective with respect to the binary image signal Fa (x, y), a small isolated point removing circuit 2
A small area of "0" is erased by 3a ', and a binarized image signal Fa' (x, y) from which isolated points are removed is obtained and stored in the image memory 24a. That is,
The small isolated point removing circuit 23a 'removes isolated points by erasing minute "0" areas.

In the image signal Fb (x, y) binarized by the threshold value TH1 (<THh), the paste portion is “0”, as in the binarized image signal Fa (x, y) of the threshold value THh. And the green sheet portion becomes "1", but since the threshold value TH1 is low, the pinhole can be reliably set to "1". Although the generation of bright spots in the paste portion is suppressed by polarized light illumination / polarization detection of light having a wavelength of 400 to 550 nm such as blue, a small pinhole removing circuit is considered in the event that bright spots are generated. 23b 'eliminates the minute area of "1" to obtain a binarized image signal Fb' (x, y) from which the bright spot has been removed, and stores it in the image memory 24b. That is, the small pinhole removing circuit 23
b ′ is obtained by erasing a minute “0” area.
It removes bright spots and minute pinholes that are not defective. On the other hand, the reference pattern image signal R (x,
Prepare y). The reference pattern image signal R (x,
y) may be generated from design data or may be created by sampling an image of a non-defective sample, and is a binary image signal in which the paste portion is “0” and the base material portion is “1”.
The detected image signal and the reference pattern image signal at the same location are compared and inspected. However, the position of the detected image signal and the position of the reference pattern image signal may not be correctly aligned due to manufacturing variations of the substrate 51 to be inspected. is there. Therefore, detection 2
Binarized image Fb ′ (x, y) and reference pattern image signal R
Align (x, y). That is, the mismatch counter 7
5, an image (for example, ± 3 pixels) in which the alignment pattern (three types of reference pattern image signals R (x, y) (expanded, unmodified, contracted)) is shifted by ± several pixels in the x and y directions.
When shifted to the pixel, 49 binary images in total) and detection 2
The amount of misalignment is calculated by counting the number of pixels that do not match the binarized image signal (for example, the binarized image signal Fb ′ (x, y) with the threshold value THh). Then, the positional deviation detection PC 76 detects the reference image having the minimum number of mismatched pixels counted by the mismatch counter 75 as the image whose position is most matched with the detected image. At this time, the reference pattern image signal is converted into a binary image expanded by the expansion circuit 71a, a binarized image contracted by the contraction circuit 73a, and an image obtained by generating three alignment patterns of the reference pattern as they are in the through circuit 72a. When each image is read from the image memory 74 and stored in the memory 74, an image shifted in the x and y directions by ± a few pixels is generated (for example, ± 3 pixels).
When the pixel is shifted to the pixel, a total of 147 binary images are detected. Expansion and contraction can be easily performed, for example, by applying an operator of 4 connections (or 8 connections) of 3 × 3 pixels. Thereby, the size of the wiring pad 120 becomes
Even if there is a variation in each sheet or in a plane within the sheet, the comparators 31a to 31e surely detect the binarized image signals Fa '(x, y) and Fb'.
(X, y) and the defect determination reference pattern Re '(x, y),
Rc '(x, y) can be compared for alignment with the size adjusted. Note that the position shift detection PC 7
In the case of No. 6, when the amount of displacement is equal to or more than a specified value, it is determined to be a pad displacement defect 106 and output.

The enlargement reference pattern Re (x, y) serving as the first inspection criterion is subjected to enlargement processing by the enlargement processing circuit 29 with respect to the binarized reference pattern R (x, y) at a size determined by the inspection criterion. And stored in the image memory 78. Further, the reduction reference pattern Rc (x, y) serving as the second inspection criterion is created by performing reduction processing on the reference pattern R (x, y) by the reduction processing circuit 30 with a size determined by the inspection criterion. And stored in the image memory 80. Then, the expansion reference pattern Re (x, y) and the reduction reference pattern Rc
By extracting pattern edges for each of (x, y), an enlarged reference edge pattern and a reduced reference edge pattern are created and stored in the image memories 87 and 88.
At this time, the switching circuits 82b and 82c are switched based on the wiring pad size data 112 detected by the displacement detection PC 76 based on the result of the alignment between the detected image and the reference pattern. The expansion or contraction circuit 7 further expands the expansion reference pattern read from the
3b, the contraction reference pattern read from the image memory 80 is further expanded by the expansion circuit 71c or contracted by the contraction circuit 73c, and the inspection determination standard is adjusted to the size of the wiring pad. In particular, when the size of the wiring pad 120 varies from sheet to sheet or within a sheet surface, it is necessary to match the inspection criteria to the size of the wiring pad as described above. That is, if the pattern of the detected image is large, the inspection criterion needs to be large, and if the pattern is small, the inspection criterion needs to be small.

Further, a judgment pattern D (x, y) for judgment may be provided together with the reference pattern R (x, y). The determination pattern D (x, y) is a concave defect determination pattern Rc.
This is a binary image pattern in which (x, y) and the convex defect determination pattern Re (x, y) are superimposed. Therefore, the logical product of the binarization reference pattern R (x, y) and the binarization determination pattern D (x, y) in the logical product circuit 90 is obtained.
The convex defect judgment pattern Re of the judgment pattern D (x, y)
(X, y) is extracted, and a convex defect determination pattern (enlarged reference pattern) Re (x, y) can be obtained. Also,
In the AND circuit 91, the binary reference pattern R (x,
y) and a binarization determination pattern D (x,
By taking a logical product with y) and negating, the concave defect determination pattern Rc (x, y) of the determination pattern D (x, y) is extracted and the concave defect determination pattern (reduced reference pattern) Rc
(X, y) can be obtained. There is an advantage that the degree of freedom of the inspection criterion can be greatly improved by preparing such a determination pattern D (x, y) in advance. The defect 141 including the projection defect 141a and the scattering defect 141b is converted into an image signal Fa '(x, y) binarized by the threshold value THh read from the image memory 24a.
And a convex defect determination pattern Re ′ (x, y) read by shifting (correcting the position shift) from the image memory 83 based on the position shift amount 111 detected by the position shift detection PC 76. By taking the logical product and negating, it is detected as "0". Note that the defect signal 141
May be detected as “1”. Further, in a comparison circuit (logical product circuit) 31d, a logical product of a signal obtained by inverting the image signal Fa '(x, y) and a signal obtained by inverting the edge signal of the convex defect determination pattern read from the image memory 87 is obtained. , The projection defect 143 can be output because the projection defect exists on the edge of the convex defect determination pattern. Therefore, the comparison circuit (logical product circuit) 31
By comparing the defect signal 141 output from a with the defect signal 143 output from the comparison circuit (logical product circuit) 31d, it is possible to output the projection defect 141a and the scattering defect 141b separately.

The defect 142 composed of the defect defect 142a and the pinhole defect 142b is obtained by converting the image signal F binarized by the threshold value TH1 read from the image memory 24b.
The concave defect determination pattern Rc ′ (x, y) read by shifting (correcting the position shift) from the image memory 84 based on b ′ (x, y) and the position shift amount 111 detected by the position shift detection PC 76. ) Is negated by taking the logical product with the inverted signal, and is detected as “0”. In addition,
The defect signal 142 may be detected as “1”. Furthermore,
In the comparison circuit (logical product circuit) 31e, the image signal F
By taking the logical product of b ′ (x, y) and the signal obtained by inverting the edge signal of the concave defect determination pattern read from the image memory 88, the defect exists on the edge of the concave defect determination pattern. For that reason, the missing defect 144 can be output. Therefore, the comparison circuit (AND circuit) 3
By comparing the defect signal 142 output from 1b with the defect signal 144 output from the comparison circuit (logical product circuit) 31e, the defect defect 142a and the pinhole defect 14 are compared.
2b can be output separately. FIG. 14 shows a schematic configuration of an embodiment of a wiring pad inspection apparatus according to the present invention. That is, in a wiring pad inspection apparatus,
The components denoted by the same reference numerals as those in FIG. 7 have the same configuration. The binarization circuit 20a 'outputs the detected image signal F (x, y)
Is binarized by a threshold value THh to obtain a binarized image signal Fa.
(X, y). Isolated point removal circuit 23
a ′ is a detected binary image signal F obtained by removing non-defect minute scattering (isolated point) from the binary image signal Fa (x, y).
a '(x, y) is formed and temporarily stored in the image memory 24a. The binarization circuit 20b 'binarizes the detected image signal F (x, y) with a threshold value THl and converts the binarized image signal F (x, y) into a binarized image signal Fb (x, y). The luminescent spot removing circuit 23b 'detects the binarized image signal F obtained by removing non-defective minute luminescent spots from the binarized image signal Fb (x, y).
b '(x, y) is formed and temporarily stored in the image memory 24b. Each of the image memories 24a and 24b has two memories each having a capacity enough to store an image for one scan of the stage (the number of pixels of the line sensor × the detection length in the stage traveling direction), and is used as a double buffer.
While storing the detected image for one scan on one surface, the image detected by the previous scan on the other surface is read out, and the subsequent matching processing is performed.

On the other hand, the data processing PC 25 sends the compressed reference pattern data stored in the hard disk to the expansion circuit 26a and the compressed determination pattern data to the expansion circuit 26b. The data processing PC 25 communicates with the control PC 22 via a network, and operates according to an instruction from the control PC 22. Expansion circuit 26
In a and 26b, the compressed data is decompressed, and a reference pattern and a judgment pattern to be compared with the detected binary image are generated.
These are stored in the image memories 27a and 27b, respectively. The image memories 27a and 27b are configured in the same manner as the image memories 24a and 24b, and store the expanded reference pattern and determination pattern at the same location as the detected binary image on one surface, read the image on the other surface, Is performed. The position shift detection circuit 28 detects the position shift between the detected binary image Fb ′ (x, y) read from the image memory 24b and the reference binary image R (x, y) read from the image memory 27a. The amount and the data of the size of the wiring pad 120 on the substrate 51 to be inspected are calculated and transmitted to the control PC 22. As shown in FIG. 13, the synthesizing circuit 40 synthesizes, for example, the binarization reference pattern R (x, y) and the binarization determination pattern D (x, y), and forms the convex defect determination pattern Re (x, y). ) And a concave defect determination pattern Rc ((x, y)).
The combining circuit 40 detects the detected binary image Fb ′ calculated by the displacement detection circuit 28 for the convex defect determination pattern Re (x, y) and the concave defect determination pattern Rc (x, y).
Expanded, contracted, or processed as it is to match the wiring pad size 112 at (x, y), and the convex defect determination pattern Re ′ (x, y) and the concave defect determination pattern Rc ′ (x, y) Is obtained and stored in the image memory. Further, from the final image memory of the synthesizing circuit 40, the expansion or contraction or the convex defect determination pattern Re '(x, y) as it is in a state in which the position is adjusted to the displacement amount 111 calculated by the displacement detection circuit 28. ) And the concave defect determination pattern Rc ′ ((x, y)).

Next, in the comparison circuit 31a, the detected binary image Fa '(x,
y) and the binarized image Re ′ (x, y) for judging a convex defect created by the synthesizing circuit 40 are compared in shape to determine the projection defect 141.
A defect 141 consisting of a and a scatter defect 141b is detected and stored in the determination result memory 32 together with its position coordinates.
At the same time, in the comparison circuit 31b, the detected binary image Fb '(x, y) read from the image memory 24b and the concave defect determination binary image Rc' created by the synthesis circuit 40.
By comparing the shape with (x, y), a defect 142 including a defect defect 142a and a pin pole defect 142b is detected and stored in the determination result memory 32 together with its position coordinates. Further, as described with reference to FIG. 13, the defect 141 can be separated into a projection defect 141a and a scattering defect 141b. As for the defect 142, the defect 142a
And a pin pole defect 142b. At this time, each of the comparison circuits 31
It is natural that the shapes are compared in consideration of the displacement amount obtained in step 8. The defect information (position and type) obtained by each comparison circuit 31 is sequentially stored in the judgment result memory 32, and the defect information is collectively stored in the control PC 22 at each stage scan or at the end of one substrate inspection. send. Control PC 22
Outputs defect information to a printer or monitor, and to a host computer via a network.

As described above, according to the embodiment of the inspection of the filling state of the through hole and the inspection of the wiring pad according to the present invention, the bright spot of the copper paste is detected by the polarized light illumination with the wavelength limited to 400 to 550 nm. Furthermore, by creating a plurality of reference patterns having different sizes and performing alignment, it is possible to allow variation in the size of the detection conductor pattern, and furthermore, it is possible to allow the detection image to be convex with respect to the detection image. By separately providing a binarization threshold for defect determination and a binarization threshold for concave defect determination, it is possible to detect even a low-contrast, for example, minute defect. In addition, by preparing a judgment pattern for judging a convex defect and a judgment pattern for judging a concave defect separately in advance, flexibility in the defect judgment standard is created. Since the reference pattern and the judgment pattern are compressed and stored, and the necessary portion is read out while being expanded during the inspection, the capacity required for storing the data can be reduced. For this reason, it is possible to accurately and quickly inspect a scattered or bleeding defect with respect to a filler filled in a through hole having a large number of irregularities at an edge portion. Furthermore, even if the substrate has irregularities such as dents, it is possible to accurately and quickly inspect a defect that is insufficient.

Further, since there is no need to separate a detection optical system for detecting a defect such as scattering or bleeding from a detection optical system for detecting an insufficient defect, the inspection can be performed by one detection optical system. Can be With respect to a wiring pad having many edge irregularities, even a fine pinhole defect can be inspected accurately and at high speed regardless of the variation in the size of the wiring pad. Next, a process of manufacturing a circuit board for performing a through hole filling state inspection and a pattern inspection according to the present invention will be described with reference to FIG. First, in a through hole forming step 151, a large number of through holes 2 for transmitting interlayer signals are formed in a green sheet 1 as a substrate material, and in a conductor filling step 152, a conductive paste 3 is filled by printing. At this stage, in the through-hole filling state inspection step 153, the through-hole filling state inspection is performed by using the through-hole filling state inspection apparatus as described above. Eliminate places. As a correction method, defect information is output from a host computer to a monitor or the like and displayed, and an operator corrects the defect information while viewing the displayed defect information. Bleed defect 102
The operator a, the scattering defect 102b, and the foreign matter defect 103b may be removed by using a removal tool or the like. In addition, the bleeding defect 102a and the scattering defect 102b can be removed by irradiating the portion with an energy beam such as a narrow laser beam. In the case of the foreign matter defect 103b, it is desirable to fill a filling at the removed mark. In addition, the insufficient filling defect 10
As for No. 1, the filler may be filled. In the case of a defect without a through-hole, the through-hole is formed by irradiating an energy beam such as a laser and the formed through-hole is filled with a filler, so that the defect can be corrected. However, the through-hole position shift defect cannot be basically corrected.

Next, in a wiring printing step 155, a conductor paste is screen-printed on the sheet surface to form a circuit pattern. At this stage, in the pattern inspection step 156, the wiring pads are inspected by using the above-described wiring pad inspection apparatus. If there is a defective wiring pad, it is corrected in the pattern correction step 157 to eliminate the defective portion. As a correction method, defect information is output from a host computer to a monitor or the like and displayed, and an operator corrects the defect information while viewing the displayed defect information. The protrusion defect 141a and the scattering defect 141b may be removed by an operator using a removal tool or the like. In addition, the projection defect 141a and the scattering defect 141b can be removed by irradiating an energy beam such as a narrow laser beam to this portion. Also,
For the defect defect 142a and the pinhole defect 142b, a filler may be filled.

In the laminating and sintering step 158, only the green sheets having no defective parts thus obtained are laminated and sintered to complete a multilayer ceramic substrate. At this stage, in the continuity inspection step 159,
A continuity test is performed using a continuity test device. In the continuity test, basically, no wiring failure occurs, but if a wiring failure occurs, repair wiring is formed by a thin film pattern formed on the surface of the sintered multilayer ceramic substrate, and wiring is performed. It is also possible to repair defects.

[0054]

According to the present invention, surplus defects (bleeding, scattering, projections) and defective defects (chips, deficiencies, pinholes) in a conductive pattern such as a filling or a wiring pad made of a conductive paste containing fine metal particles. 3) It is possible to reliably and quickly inspect a defect including a foreign substance, a defect without a conductor pattern, and a position shift defect while minimizing a false report.

In addition, according to the present invention, even if the conductive pattern such as the filling or the wiring pad made of the conductive paste containing the fine metal particles varies from sheet to sheet or within the sheet surface, the pattern on the base material is not affected. Even if unevenness occurs in the surface, an effect of being able to reliably and quickly inspect for an agent excess defect, a defect defect, a foreign matter mixed defect, a defect without a conductor pattern, a position shift defect, etc. without erroneous detection. Further, according to the present invention, there is an effect that a multilayer ceramic substrate which does not need to form a thin film for repair can be manufactured with high yield.

[Brief description of the drawings]

FIG. 1 is a diagram showing a defect to be inspected generated when a through-hole is filled with a filler according to the present invention.

FIG. 2 is a view showing a spectral reflectance of a copper paste according to the present invention.

FIG. 3 is a diagram for explaining a relationship between a detection image of a through-hole filling material and a binarized image thereof according to the present invention.

FIG. 4A is a diagram illustrating an embodiment of a defect determination algorithm for a filling state inspection according to the present invention.

FIG. 4B is a view for explaining an embodiment of a defect judgment algorithm of the filling state inspection according to the present invention.

FIG. 5 is a diagram for explaining an operator for a 3 × 3 pixel in the expansion circuit and the contraction circuit;

FIG. 6 is a diagram for explaining the concept of a defect determination algorithm for a filling state inspection according to the present invention.

FIG. 7 is a configuration diagram showing a first embodiment of a filling state inspection apparatus according to the present invention.

FIG. 8 is a diagram for explaining an isolated point / bright point removal operator.

FIG. 9 is a configuration diagram showing a second embodiment of the filling state inspection apparatus according to the present invention.

FIG. 10 is a view for explaining another embodiment of the defect judgment algorithm of the filling state inspection according to the present invention.

FIG. 11 is a diagram illustrating another embodiment of the defect determination algorithm for the filling state inspection according to the present invention.

FIG. 12 is a diagram for explaining a defect to be inspected of a wiring pad according to the present invention.

FIG. 13 is a diagram illustrating an embodiment of a defect determination algorithm for wiring pad shape inspection according to the present invention.

FIG. 14 is a configuration diagram showing a first embodiment of a wiring pad inspection apparatus according to the present invention.

FIG. 15 is a view illustrating a method for manufacturing a multilayer substrate according to the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Base material (circuit board), 2 ... Through-hole, 3 ... Filling, 4 ... Bleeding, 5 ... Scattering, 6 ... Foreign matter, 7 ... No through-hole, 8 ... Displacement of through-hole, 9 ... Insufficient filling,
11: illumination direction, 12: detection direction, 20a, 20b, 2
0c, 20a ', 20b', 20d ... binarization circuit, 21
... image input ports, 23a, 23a '... minute isolated point removing circuits, 22 ... control PCs, 23b, 23b', 23c ...
Bright spot removal circuit, 24a, 24b, 24c, 24d, 8
3, 84, 85 ... image memory, 25, 25a, 25b ...
PC for data processing, 26, 26a, 26b, 26c ... decompression circuit, 27, 27a, 27b, 27c, 78, 80,
Reference numeral 86: image memory, 28, 28a, 28b: position shift detection circuit, 31a, 31b, 32c: comparison circuit (logic circuit), 29: enlargement processing circuit, 30: reduction processing circuit, 40
... Synthesis circuit, 51 ... Circuit board, 52 ... Stage base,
53: X stage, 54: Y stage, 55: X axis drive unit, 56: Y axis drive unit, 57: X axis length measuring device, 58: Y axis length measuring device, 60: fiber light source, 61: condenser lens , 6
2 ... detection lens, 63 ... line sensor (detector), 6
4, 65: polarizing plate, 66: image detection circuit (A / D converter), 67: wavelength limiting filter, 68: correction circuit, 7
1a to 71d: expansion circuit, 72a to 72d: through circuit, 73a to 73d: contraction circuit, 71a to 71d, 72
a to 72d, 73a to 73d: size matching circuit, 74: image memory (three types of positioning patterns), 75: mismatch counter, 76: through-hole displacement (PC) detection circuit, 82b to 82d: switching circuit, 101 ... Insufficient filling defect, 102: bleeding / scattering defect, 103: no through-hole, sheet dust defect, 106: through-hole displacement defect, 120: wiring pad, 121: pinhole, 1
22: missing, 123: projection, 124: scattering, 141: projection / scattering defect, 142: loss / pinhole defect.

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Mineo Nomoto, Inventor 292, Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside the Manufacturing Research Laboratory, Hitachi, Ltd. (72) Haruomi Kobayashi Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa 292 Hitachi Manufacturing Co., Ltd.Production Technology Laboratory (72) Inventor Yoko Irie 292 Yoshidacho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Co., Ltd.Production Technology Research Laboratory Hitachi, Ltd. (72) Inventor Hiroto Okuda Totsuka, Yokohama-shi, Kanagawa Prefecture 292 Yoshida-cho, Ward Inside Production Technology Research Laboratory, Hitachi, Ltd. (72) Inventor Norio Chiishi 1-Horiyamashita, Hadano-shi, Kanagawa Prefecture General-purpose Computer Business Division, Hitachi, Ltd. (72) Inventor Emi Emi, Hadano-shi, Kanagawa 1 Horiyamashita F-term in Hitachi Computer Co., Ltd. General-purpose Computer Division (Reference) 2F065 AA03 AA09 AA12 AA18 AA20 AA26 AA56 BB26 CC25 DD06 DD11 EE04 FF04 FF61 GG03 GG05 GG22 HH04 HH09 HH12 HH14 JJ02 JJ25 LL02 LL04 LL22 LL33 LL34 MM02 PP12 QQ05 QQ06 QQ08 QQ24 QQ25 QQ29 QQ32 QQ39 QQ43 RR09 SS04 SS13 UU05 UU07 2G051 AA65 AA90 AB20 BA05 BA10 BA20 BB07 BB17 CA03 CB01 DA07 EA01 EA11 EA12 EA14 EA23 EB01 EB02 EC02 ED01 ED07 ED14 ED15 ED22 FA10

Claims (17)

[Claims] 1. A circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern is illuminated with polarized illumination light having a wavelength of 550 nm or less, The detector receives an optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the illuminated circuit board to be inspected and converts the optical image signal into an optical image signal. Binarizing with a threshold value and converting it into a binary image signal indicating the conductor pattern; calculating a displacement amount of the conductor pattern based on the binarized image signal indicating the converted conductor pattern; And comparing the binarized image signal indicating the determined conductor pattern with a reference binarized image signal for determining a defect determination criterion of the conductor pattern based on the calculated displacement amount of the conductor pattern. By A method for inspecting a conductor pattern, wherein a defect of the conductor pattern is extracted by using the method. 2. A circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a predetermined conductor pattern on a base material is illuminated with polarized illumination light having a wavelength of 550 nm or less, The detector receives an optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the illuminated circuit board to be inspected and converts the optical image signal into an optical image signal. It is binarized by a threshold value and converted into a binarized image signal indicating the conductor pattern. Based on the binarized image signal indicating the converted conductor pattern, the displacement amount of the conductor pattern, the size of the conductor pattern and Calculating a binarized image signal indicating the converted conductor pattern and a reference binarized image signal for determining a defect determination criterion of the conductor pattern matched to the size of the calculated conductor pattern. Sa A method for inspecting a conductor pattern, comprising extracting a defect of the conductor pattern by performing alignment and comparison based on the amount of misalignment of the conductor pattern. 3. A circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern is illuminated with polarized illumination light. An optical image based on the reflected light obtained by blocking the polarized light component of the polarized illumination light from the circuit board is received by a detector and converted into an optical image signal. The converted optical image signal is binary-coded with a predetermined threshold value. And converting into a binarized image signal indicating the conductor pattern, calculating a displacement amount of the conductor pattern and a size of the conductor pattern based on the binarized image signal indicating the converted conductor pattern; A binarized image signal indicating the determined conductor pattern and a reference binarized image signal for determining a defect determination criterion of the conductor pattern matched to the calculated size of the conductor pattern are converted into positions of the calculated conductor pattern. Deviation A method for inspecting a conductor pattern, comprising extracting a defect of the conductor pattern by performing alignment and comparison based on the pattern. 4. A circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern is illuminated with polarized illumination light. An optical image based on the reflected light obtained by blocking the polarized light component of the polarized illumination light from the circuit board is received by a detector and converted into an optical image signal. A first value indicating the conductor pattern which is binarized by a threshold value
And a second binarized image signal, and calculates a displacement amount of the conductor pattern based on the first or second binarized image signal indicating the converted conductor pattern. A first binarized image signal indicating a pattern is aligned with a first standardized binarized image signal that determines a defect determination criterion of an outer frame with respect to the conductor pattern based on the calculated positional shift amount. A defect of the conductor pattern existing outside the outer frame is extracted by comparison, and a second binarized image signal indicating the converted conductor pattern and a defect determination criterion of the inner frame for the conductor pattern are determined.
Inspecting a conductor pattern existing in the inner frame by aligning the reference binary image signal with the reference binary image signal based on the calculated positional deviation amount and comparing the positions. Method. 5. A circuit board to be inspected, which is formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern, irradiates polarized illumination light to the inspected circuit board. An optical image based on the reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board is received by a detector and converted into an optical image signal, and the converted optical image signal is converted into first and second different optical signals. A first value representing the conductor pattern binarized by a threshold value
And a second binarized image signal, and calculates a displacement amount of the conductor pattern and a size of the conductor pattern based on the first or second binarized image signal indicating the converted conductor pattern. A first binarized image signal indicating the converted conductor pattern and a first reference binarization for determining a defect determination criterion of an outer frame with respect to the conductor pattern adjusted to the size of the calculated conductor pattern; A second binarized image showing the converted conductor pattern by extracting a defect of the conductor pattern existing outside the outer frame by aligning and comparing the image signal with the image signal based on the calculated displacement amount, and comparing the image signal with the image signal. A signal and a binarized image signal as a second reference for determining a defect determination criterion of the inner frame for the conductor pattern matched to the calculated size of the conductor pattern are located based on the calculated positional deviation amount. Inspection method of the conductive pattern and extracting a defect of the conductive patterns that exist inside the inner frame by combined and compared. 6. A circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern is illuminated with polarized illumination light. An optical image based on the reflected light obtained by blocking the polarized light component of the polarized illumination light from the circuit board is received by a detector and converted into an optical image signal, and the converted optical image signal is converted into first and second different optical signals. A first value indicating the conductor pattern which is binarized by a threshold value
And converting it into a second binarized image signal, forming at least three types of standardized binarized image signals that are at least expanded and contracted with respect to the standard binarized image signal indicating the standard conductor pattern, A conductor pattern is obtained by relatively shifting the created at least three types of reference binarized image signals and the first or second binarized image signal indicating the converted conductor pattern to obtain a degree of coincidence. Calculating the position shift amount and the size of the conductor pattern, and the first binarized image signal indicating the converted conductor pattern and the outer frame of the conductor pattern matched to the calculated size of the conductor pattern. The defect of the conductor pattern existing outside the outer frame is extracted by aligning and comparing the binarized image signal of the first criterion for determining the defect criterion based on the calculated positional deviation amount. A second binarized image signal indicating the converted conductor pattern and a binarization of a second criterion for determining a defect determination criterion of an inner frame with respect to the conductor pattern adjusted to the size of the calculated conductor pattern A method for inspecting a conductor pattern, comprising extracting a defect of a conductor pattern present in an inner frame by aligning and comparing an image signal with the image signal based on the calculated displacement amount. 7. An optical image obtained from a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material on a predetermined conductor pattern is received by a detector and converted into an optical image signal. And converting the converted optical image signal into a first value indicating the conductor pattern by binarizing the converted optical image signal with first and second threshold values different from each other.
And converting it into a second binarized image signal, forming at least three types of standardized binarized image signals that are at least expanded and contracted with respect to the standard binarized image signal indicating the standard conductor pattern, A conductor pattern is obtained by relatively shifting the created at least three types of reference binarized image signals and the first or second binarized image signal indicating the converted conductor pattern to obtain a degree of coincidence. Calculating the position shift amount and the size of the conductor pattern, and the first binarized image signal indicating the converted conductor pattern and the outer frame of the conductor pattern matched to the calculated size of the conductor pattern. The defect of the conductor pattern existing outside the outer frame is extracted by aligning and comparing the binarized image signal of the first criterion for determining the defect criterion based on the calculated positional deviation amount. A second binarized image signal indicating the converted conductor pattern and a binarization of a second criterion for determining a defect determination criterion of an inner frame with respect to the conductor pattern adjusted to the size of the calculated conductor pattern A method for inspecting a conductor pattern, comprising extracting a defect of a conductor pattern present in an inner frame by aligning and comparing an image signal with the image signal based on the calculated displacement amount. 8. A circuit board to be inspected, which is formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern, and irradiates polarized illumination light to the inspected circuit board. An optical image based on reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board is received by a detector and converted into an optical image signal, and the converted optical image signal is converted into first, second, and different optical signals. Converting the first and second binarized image signals indicating the conductor pattern into a third binarized image signal indicating a defect in the conductor pattern by binarizing with a third threshold value; At least three types of reference binary image signals that are expanded and contracted as they are with respect to the reference binary image signal indicating the conductor pattern are formed, and the created at least three types of reference binary image signals are created. And the converted conductor putter The position of the conductor pattern and the size of the conductor pattern are calculated by relatively shifting the first or second binarized image signal indicating Calculating the first binarized image signal and the first standardized binarized image signal for determining the defect determination criterion of the outer frame with respect to the conductor pattern adjusted to the calculated size of the conductor pattern; A defect of the conductor pattern existing outside the outer frame is extracted by performing alignment and comparison based on the shift amount, and a second binarized image signal indicating the converted conductor pattern and the calculated conductor pattern are compared. The binarized image signal of the second reference which determines the defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size is aligned and compared based on the calculated positional shift amount. Extracting a defect of the conductive patterns that exist inside the inner frame by a third binary image signal and the first indicating the defect of the converted in the conductor pattern
A third method for determining a defect determination criterion according to the size of the binarized image signal or the calculated conductor pattern of the above.
A method for inspecting a conductor pattern, comprising extracting a defect in the conductor pattern by aligning and comparing the reference binary image signal based on the calculated positional deviation amount. 9. An optical image obtained from a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern is received by a detector and converted into an optical image signal. A first and second binarized image signal indicating the conductor pattern by binarizing the converted optical image signal with first, second, and third thresholds different from each other; Is converted into a third binarized image signal indicating the defect of the reference, and at least the three types of standardized binarized image signals expanded and contracted with respect to the reference binarized image signal indicating the reference conductor pattern at least Forming the at least three types of reference binary image signals and the first or second binary image signal representing the converted conductor pattern to obtain the degree of coincidence. By the conductor pattern Calculating a displacement amount and a size of the conductor pattern; a first binarized image signal indicating the converted conductor pattern; and a defect of an outer frame with respect to the conductor pattern adjusted to the calculated size of the conductor pattern. A defect of the conductor pattern existing outside the outer frame is extracted by aligning and comparing the binarized image signal of the first criterion for determining the judgment criterion based on the calculated positional deviation amount, and the converted A second binarized image signal indicating the conductor pattern obtained and a second criterion binarized image signal for determining a defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size of the calculated conductor pattern. A defect of the conductor pattern existing in the inner frame is extracted by performing alignment and comparison based on the calculated displacement amount, and a defect indicating the defect in the converted conductor pattern is extracted. The binary image signal from the first
A third method for determining a defect determination criterion according to the size of the binarized image signal or the calculated conductor pattern of the above.
A method for inspecting a conductor pattern, comprising extracting a defect in the conductor pattern by aligning and comparing the reference binary image signal based on the calculated positional deviation amount. 10. The method for inspecting a conductor pattern according to claim 1, wherein the conductor pattern is a filler formed by filling and printing a conductive paste in a through hole formed in a base material. A method for inspecting a conductor pattern, the method comprising: 11. A method for inspecting a conductor pattern according to claim 1, wherein the conductor pattern is a wiring pad formed by printing a conductive paste on a base material. Method. 12. A through hole forming step of forming a large number of through holes in a substrate, and a conductor filling step of printing and filling a conductive paste containing fine metal particles in each of the through holes formed in the through hole forming step. A conductor pattern filled in the conductor filling step;
A conductor pattern inspection step of inspecting by any one of the conductor pattern inspection methods according to any one of 9 to 9; A printed wiring forming step for forming a wiring conductor having the wiring pad; A lamination and sintering step of laminating and sintering circuit boards determined to be non-defective in the wiring pad inspection step to obtain a multi-layer substrate. 13. Illumination for illuminating polarized illumination light having a wavelength of 550 nm or less on a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductive pattern. An optical system, and a detection optical system that receives an optical image based on reflected light obtained by shielding the polarized component of the polarized illumination light from the circuit board to be inspected illuminated by the illumination optical system, and converts the optical image into an optical image signal. A binarization circuit that binarizes the optical image signal converted by the detector of the detection optical system with a predetermined threshold value and converts the binarized image signal into a binarized image signal indicating the conductor pattern; Calculating means for calculating the amount of displacement of the conductor pattern based on the binarized image signal indicating the conductor pattern converted by the circuit; and a binarized image signal indicating the conductor pattern converted by the binarization circuit; Conductor pattern defect criteria And comparing means for extracting a defect of the conductor pattern by aligning and comparing the binarized image signal with a reference binary image signal based on the displacement amount of the conductor pattern calculated by the calculation means. Inspecting device for conductor pattern. 14. An illumination optical system for illuminating polarized illumination light on a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern. A detection optical system for receiving, by a detector, an optical image based on reflected light obtained by blocking a polarization component of the polarized illumination light from a circuit board to be inspected illuminated by an optical system and converting the optical image into an optical image signal; Binarizing circuit for binarizing an optical image signal converted by the detector with first and second thresholds different from each other to convert the optical image signal into first and second binary image signals indicating the conductor pattern Calculating means for calculating a displacement amount of the conductor pattern and a size of the conductor pattern based on a first or second binarized image signal indicating the conductor pattern converted by the binarization circuit; Conductor pattern converted by the digitizing circuit And a first reference binary image signal for determining a defect determination criterion of the outer frame for the conductor pattern adjusted to the size of the conductor pattern calculated by the calculation means. A second defect indicating a conductor pattern converted by the binarization circuit is extracted by extracting a defect of the conductor pattern existing outside the outer frame by performing alignment and comparison based on the displacement amount calculated by the calculation means. The calculation means calculates a binarized image signal and a second reference binary image signal for determining a defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size of the conductor pattern calculated by the calculation means. A conductor pattern inspection apparatus, comprising: comparing means for extracting a defect of the conductor pattern existing in the inner frame by performing alignment and comparison based on the determined positional deviation amount. 15. An illumination optical system for illuminating polarized illumination light on a circuit board to be inspected formed by printing a conductive paste containing metal fine particles on a base material in a predetermined conductor pattern. A detection optical system for receiving, by a detector, an optical image based on reflected light obtained by blocking the polarization component of the polarized illumination light from the circuit board to be inspected illuminated by the optical system with a detector, and converting the optical image signal into an optical image signal; First and second binarized image signals indicating the conductor pattern by binarizing the optical image signal converted by the detector with first, second, and third thresholds different from each other and the conductor pattern And a binarization circuit for converting into a third binarized image signal indicating a defect within the three-dimensional reference image signal which is at least expanded, unchanged, and contracted with respect to the reference binary image signal indicating the reference conductor pattern Forming a binary image signal; The degree of coincidence is determined by relatively shifting the created at least three types of reference binarized image signals and the first or second binarized image signal representing the conductor pattern converted by the binarization circuit. Calculating means for calculating the amount of displacement of the conductor pattern and the size of the conductor pattern, a first binarized image signal indicating the conductor pattern converted by the binarization circuit, and a value calculated by the calculation means A first reference binary image signal for determining a defect determination criterion of an outer frame with respect to a conductor pattern adjusted to the size of the conductor pattern is compared with a binary image signal based on the displacement amount calculated by the calculation means. The second binarized image signal indicating the conductor pattern converted by the binarization circuit is extracted by extracting the defect of the conductor pattern existing outside the outer frame by the operation, and the conductor pattern calculated by the calculation unit. By comparing and aligning the binarized image signal of the second reference which determines the defect determination criterion of the inner frame with respect to the conductor pattern adjusted to the size based on the positional shift amount calculated by the calculating means, The defect of the conductor pattern existing in the inner frame is extracted,
Third indicating a defect in the conductor pattern converted by the digitizing circuit
And a third standardized binary image signal for determining a defect determination criterion according to the size of the conductor pattern calculated by the calculating means or the first standardized binary image signal. And a comparing means for extracting a defect in the conductive pattern by aligning and comparing the positions based on the displacement amount calculated by the calculating means. 16. An optical image of a circuit board having a large number of through-holes filled with an electrically conductive substance is detected, and the detected image is binarized to obtain a detected binary pattern. The reference binary pattern of the entire inspection area is compressed and stored, and the reference binary pattern of the portion corresponding to the detected binary pattern during inspection is read out while being expanded, and the detected binary pattern is aligned with the reference binary pattern. After that, criterion 2
Value pattern expanded to size determined by inspection criteria,
A through hole filling state inspection method, wherein a defect is extracted by comparing the shapes of a reduced expansion pattern, contraction pattern and a detected binary pattern. 17. An optical image of a circuit board on which a circuit pattern made of an electrically conductive substance is formed, and the detected image is binarized to obtain a detected binary pattern. The reference binary pattern is compressed and stored, and read while expanding the reference binary pattern in a portion corresponding to the detected binary pattern during inspection. After the detected binary pattern is aligned with the reference binary pattern, A pattern inspection method characterized in that a defect is extracted by comparing a shape of a detected binary pattern with an expanded pattern, a contracted pattern that has been enlarged or reduced to a size determined by an inspection determination standard.