JP3262150B2 - Inspection device for solder joints - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a solder joint inspection apparatus for inspecting whether a solder joint has a defect by detecting the appearance of a solder joint formed on a printed circuit board. It is.

[0002]

2. Description of the Related Art Conventionally, a solder joint inspection apparatus includes a frequency distribution and a correlation of a grayscale image obtained by imaging a solder joint.
Alternatively, the area value of the image after the binary processing is obtained, and these are compared with a preset threshold value to determine whether the solder joint has a defect. But,
In the conventional example, for the image data of the local part of the image necessary to extract the three-dimensional shape of the solder joint, the inspection parameters are increased in order to repeat the determination based on a wide variety of criteria, There is a problem in that it is difficult to uniquely determine the quality criteria. The above-mentioned conventional example is disclosed in, for example, the following gazettes. JP-A-62-215854 JP-A-63-196980 JP-A-63-304109 JP-A-1-219548 JP-A-2-44234

[0003]

[0008] The present invention has been made to solve such a problem, the shading information
A plurality of extraction regions are defined for a captured image serving as angle information of an inclined surface of a solder joint, an average light amount of each extraction region is obtained, and the average light amount is treated as multivariate data. To realize a solder joint inspection apparatus that can globally judge the quality of a solder joint without setting a threshold value by substituting multivariate data into a prepared probability density function and making a judgment. With the goal.

[0004]

The present invention SUMMARY OF THE INVENTION, in the inspection apparatus of the solder joints for inspecting the presence or absence of a defect in a solder joint formed on a printed circuit board from the appearance, expansion forming a cylindrical
Irradiation light from a light source is incident on the end surface of the diffuser plate, and the incident light is totally reflected and diffusely reflected on the inner surface of the diffuser plate.
The irradiation angle and intensity of light emitted from the light source along the length of the diffuser
The brightness of the reflected light emitted from the inner peripheral surface of the diffusion plate , reflected at the solder joint and traveling in the length direction of the diffusion plate is continuously changed according to the inclination angle of the reflection position. The reflected light that travels in the length direction of the illuminating means and the diffusion plate is received, and a reflected light image of the solder joint is taken.
And an area definition for defining a plurality of extraction areas in the image captured by the imaging means that are necessary to extract the three-dimensional characteristics of the solder joint. Means, a light amount calculating means for calculating an average light amount of a captured image in each extraction region, and, for a plurality of samples including a non-defective product and a defective product having a three-dimensional shape of a solder joint, the average light amount of each extraction region. Using multivariate data as variables, a function generation means for generating a probability density function indicating the probability distribution that the solder joint belongs to the sample for each sample, and a probability density function for each sample generated by this function generation means. The function storage means to be stored and the multivariate data obtained by the light quantity calculation means for the solder joint to be inspected are substituted into each probability density function, and the probability that the solder joint to be inspected belongs to each sample is obtained, highest Determination means for determining that the shape of the solder joint to be inspected belongs to the sample taking the ratio, and determining whether there is a defect in the solder joint according to the determined sample. This is a joint inspection device.

[0005]

According to the present invention as described above, the illuminating means obtains a picked-up image in which the luminance changes continuously according to the angle of the inclined surface of the solder joint to be inspected. A plurality of extraction regions are defined for this captured image, the average light amount of each extraction region is determined, these average light amounts are treated as multivariate data, and the multivariate data is substituted into a probability density function provided for each sample. It is determined that the solder joint to be inspected belongs to the sample with the highest probability.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings. FIG.
6 is described in Japanese Patent Application No. 5-1 by the present applicant.
This is the same as the description in the application specification of No. 23992. FIG. 1 is a schematic configuration diagram of one embodiment of the present invention. In FIG.
Reference numeral 1 denotes a printed board, and 2 denotes a mounted component mounted on the printed board 1. The mounted component 2 is joined to the printed circuit board 1 by soldering, and the solder joint is to be inspected. 3
An XY drive table on which the printed circuit board 1 is mounted and which moves and positions the printed circuit board 1 in a two-dimensional direction along the board surface. Reference numeral 4 denotes an illuminating unit disposed above the printed circuit board 1 and includes a ring-shaped light source 41 and a diffusion plate 42. The diffusion plate 42 has a cylindrical shape and is arranged inside the ring-shaped light source 41. The central axes of the ring-shaped light source 41 and the diffusion plate 42 are coincident. Light from the ring-shaped light source 41 enters the diffusion plate 42, is reflected in the diffusion plate 42, and is emitted to the outside world. The emitted light is applied to the solder joint. Reference numeral 5 denotes a camera as an imaging unit, which is disposed above the illumination unit 4 and captures a reflected light image of the inspection target. Reference numeral 6 denotes an inspection processing system control unit.
Then, the three-dimensional shape of the solder joint is detected from the captured image, and the presence or absence of a defect in the solder joint is determined from the detected three-dimensional shape. The inspection processing control unit 6 includes an A / D converter 61, an image memory 62, an image processing unit 63, and an image determination unit 64, and determines the presence or absence of a defect using these components. The judgment unit in the claims corresponds to the inspection processing control unit 6. Reference numeral 7 denotes a drive system control unit, which includes an illumination control unit 71 for controlling the driving of the illumination unit 4, an X-axis drive unit 72 and a Y-axis drive unit 73 for controlling the movement of the XY drive table in the X and Y directions, respectively. .

FIGS. 2 and 3 are diagrams showing specific examples of the configuration of the apparatus shown in FIG. FIG. 3 is an enlarged view of a portion X surrounded by a two-dot chain line in FIG. 2 and 3, the same components as those in FIG. 1 are denoted by the same reference numerals. Hereinafter, the same applies to the drawings. FIG.
3, an optical fiber 412 is drawn inside the ring light guide 411, and light from a light source (not shown) such as a halogen lamp is guided by the optical fiber 412 and transmitted to the ring portion of the ring-shaped light source 41. . In the ring portion, the light emitting surface of the optical fiber 412 is disposed with the light emitting surface facing the center. Optical fibers oriented toward the light emitting surface are arranged along the circumferential direction of the ring. The diffusion plate 42 is made of, for example, an acrylic pipe, and forms a primary diffusion surface 421 by chamfering an inner edge of an end surface. The light emitted from the light emitting surface of the optical fiber 412 is the primary diffusion surface 421
So that the ring-shaped light source 4
One ring portion is inscribed in the ring light guide 411. The central axes of the ring-shaped light source 41, the diffusion plate 42 and the camera 5 are coincident, and the substrate surface of the printed circuit board 1 is orthogonal to these central axes.

[0008] The operation of the device thus constructed will be described. First, the operation will be described with reference to FIG. The positions of the illumination means 4 and the camera 5 are fixed. By moving the XY drive table 3 under the control of the drive system control section 7, the printed circuit board 1 is positioned so that the solder joint to be inspected is located immediately below the illumination means 4. After the positioning is completed, the camera 5 captures an image of the solder joint, the inspection processing controller 6 recognizes the three-dimensional shape of the solder joint from the captured image, and determines the presence or absence of a defect in the solder joint based on predetermined conditions.

Here, the detection principle of the present invention will be described. 4 to 6 are explanatory views of the principle of detection. In FIG. 4, reference numeral 8 denotes a cylindrical illumination, which corresponds to the diffusion plate 42 of the present invention. 9 is a mirror-like spherical surface and corresponds to the solder joint surface 11 of the present invention. Light whose radiation intensity continuously changes in the height direction of the cylindrical surface as shown in FIG. 5 is emitted from the cylindrical illumination 8 to the specular spherical surface 9. At this time, the light traveling in the height direction of the cylindrical surface in the light reflected by the specular spherical surface 9 is:
Different optical paths are taken according to the irradiation angle of the irradiation light that is the source of the reflected light. This is because the reflected light traveling in the height direction of the cylindrical surface has a different reflection position on the spherical surface 9 if the irradiation angle of the irradiation light from which it is based is different. The inclination angle Θ shown in the figure is the angle between the tangent of the spherical surface and the horizontal at the reflection position. Each reflected light traveling in the height direction of the cylindrical surface has an inclination angle of the reflection position Θ
Are different. Accordingly, when the reflected light traveling in the height direction of the cylindrical surface is observed from directly above the specular spherical surface 9, the luminance of the light beam has a value corresponding to the inclination angle Θ. The relationship between the luminance Φ of the light beam and the inclination angle Θ is, for example, as shown in FIG.

The operation of the present invention to which such a detection principle is applied will be described with reference to FIGS. The light emitted from the optical fiber 412 in the ring light guide 411 enters the diffuser plate 42 inscribed therein, and is further irradiated on the primary diffusion surface 421. The light reflected by the diffusion surface is diffusely reflected by the secondary diffusion surface 422 using the diffusion plate 42 as a medium, and a part of the light is emitted from the boundary between the inner surface of the diffusion plate 42 and the outside to the outside. Here, the light that has not been emitted to the outside is totally reflected inside the diffusion plate 42, is again diffused and reflected on the secondary diffusion surface 422, and a part of the light is emitted to the outside. By repeating this, the brightness of the light emitted to the outside world is continuously attenuated as the distance from the light incident position of the diffusion plate 42 increases. Thus, light having different irradiation angles and intensities is emitted from the inner peripheral surface of the diffusion plate 42 along the length direction of the diffusion plate 42. In FIG. 2, the light emitted from the optical path A to the solder joint 11 has an inclination angle α.
The light emitted from the optical path B is reflected vertically upward on the slope having the inclination angle β. Since the reflection position of each light is different, the optical path of each reflected light is different. That is, the reflected light traveling vertically upward has a luminance corresponding to the inclination angle of the reflection position. Therefore, the luminance information of the reflected light is the tilt angle information. By capturing such reflected light by the camera 5, the three-dimensional shape of the solder joint 11 is detected.

FIGS. 7 to 12 show examples of the shape of the fillet of the solder. 7, 9 and 11 show the camera 5
FIG. 3 is a diagram showing a captured image of FIG. 8, 10, and 12
(A) and (b) are sectional views, and (c) is a perspective view. 7 and 8 show shapes when the fillet is insufficient, FIGS. 9 and 10 show the shape when the fillet is normal, and FIGS. 11 and 12 show the shape when the fillet is semi-cylindrical.

FIG. 13 is a diagram showing the correspondence between fillet shapes and judgment codes. As shown in FIG. 13, judgment codes “10” to “10”
"90" (judgment code "60" is a missing number).
Various fillet shapes are referred to as categories.

FIG. 14 is a view showing a window to be arranged in a picked-up image of the solder joint. W1 to W7 in FIG. 14 are windows. The window is arranged in an extraction area, which is an area necessary for extracting a three-dimensional shape feature of the solder joint. The lead portion and the pad portion in FIG. 14 correspond to the lead portion and the pad portion shown in FIG. 7, taking FIG. 7 as an example. In the quality judgment of the solder joint, each window W1 to W1
An average light amount is obtained for W7, and these average light amounts become variables 1 to 7.

FIG. 15 shows seven types of windows W1 to W for categories of "insufficient", "normal", and "kamaboko".
7 is a graph showing values of variables 1 to 7 extracted from FIG. In FIG. 15, (a) is a graph showing a category of “insufficient”, (b) is a category of “normal”, and (c) is a graph showing a category of “kamaboko”. In each category, a line graph is drawn for each of 20 samples.

The operation of inspecting a solder joint using the apparatus of the present invention will be described. For the image captured by the camera 5, the first
The average light amounts of the windows W1 to W7 shown in FIG. 4 are respectively calculated. The calculated seven average light amounts are information that gives the angle of the inclined surface of the solder joint while the original image data is a two-dimensional gray image. Therefore, the seventh-order vector constituted by the seven average light amounts is a vector according to the three-dimensional shape of the fillet.

According to the graph of FIG. 15, the graph of FIG. 15A which is a category of "insufficient" has small values of seven variables as a whole, and the graph of FIG. , All variables have relatively large values. This is because the “normal” category has more inclined surfaces for guiding the irradiation light near the optical path A in FIG. 2 than the “insufficient” category. This is apparent from the images shown in FIGS. 7 and 9 and the perspective views shown in FIGS. 8C and 10C. In the graph of FIG. 15C, which is a category of “Kamaboko”, the variables 5 and 6 take small values. This is because the area (windows W5 and W6) in which variables 5 and 6 are defined as shown in FIG.
This is because the tip of the lead is not wet as a characteristic of this category as shown in FIG.

FIG. 16 is a diagram showing a specific configuration example of a portion for performing the inspection process. 16, means 631 to 637 are provided in the image processing unit 63, and means 641 is provided in the image determination unit 64. Each means will be described. 631
Is an image capturing means for capturing image data obtained from the camera 5, 632 is a lead tip extracting means for extracting image data of a lead tip from the captured image data, 6
33 is a window definition means for defining a window in the image, and 634 is a window average brightness extraction means for obtaining the average brightness of the image data in the window among the image data at the leading end of the lead. Here, the average luminance corresponds to the above-described average light amount. Reference numeral 635 denotes a teacher data editing unit that generates teacher multivariate data by editing teacher data prepared in advance for each category. Here, the teacher data is, for example, data of each variable provided for each category as shown in the graph of FIG. 636 is a statistical value / matrix calculating means for calculating a mean value vector and a variance-covariance matrix as a normal population with respect to the teacher multivariate data to obtain a base value of a quadratic form discriminant function. This is a category / parameter defining means for defining and defining a secondary format discriminant function in a discriminant function list. Numeral 641 substitutes the multivariate data into the quadratic form discriminant function defined in the discriminant function list, finds the probability that the solder joint to be inspected belongs to all categories, and determines the solder joint to be inspected in the category having the highest probability. This is a determination unit for determining that a unit belongs. Reference numeral 65 denotes a storage unit, which is provided in the inspection processing system control unit 6 and stores window definition data, teacher multivariate data, a secondary format discriminant function, a list of discriminant functions, multivariate data, and the like. Image data, read tip data, and the like are stored in the image memory 62. Here, the area defining means referred to in the claims is the window defining means 633, the light amount calculating means is the window average luminance extracting means 634, the function generating means is the statistical value / matrix calculating means 636, and the function storing means is the storing means 65. And the judging means correspond to the judging means 641 respectively.

Here, how to determine the category to which the solder joint to be inspected belongs will be described. In the present invention, discrimination is performed using multivariate analysis. Multivariate analysis is a method of determining which category a sample belongs to based on multivariate data obtained from a sample whose affiliation is unknown. The specific analysis procedure will be described below. (1) Characteristics of data Independent variables (multivariate data): quantitative data (numerical values, etc.) Dependent variables (categories): qualitative data (names, etc.) (2) Formulation and conditions Multivariate data x = (x ₁ , x ₂ ,... x _m ) (m is an integer), when discriminating an individual into g (g is an integer) categories, the total loss due to erroneous classification of unknown individual x into the j-th category Is given by the following equation. The multivariate data x is _m having elements x ₁ , x ₂ ,.
The next vector. However, it is assumed that the conditions of (a) to (c) are satisfied, and the category determination that minimizes the expression minimizes the loss due to erroneous determination. (A) Multivariate data x for each of g categories
= Probability density function f _i (x) for (x ₁ , x ₂ ,..., X _m )
Is known. (B) A priori probability π _i (i
= 1, 2,... G) are known. (C) The loss C _ij when the category i is erroneously determined from the category i to the category j is known. Here, since it is difficult to express the loss C _ij due to erroneous discrimination with a concrete numerical value, by adopting a criterion for minimizing the erroneous discrimination rate based on the assumption of (d), the category that maximizes the following equation is obtained. Determine. Probability of belonging to category = π _i f _i (x) (d) By setting C _ij = 0 when i = j and C _ij = 1 when i ≠ j, the total loss due to erroneous classification becomes the erroneous classification rate. Is the same as the criterion for minimizing. (3) Discriminant function of quadratic form If the probability density function of each category is a normal distribution function of m variables, the probability density function f _i (x) in the i-th category
Is as follows. Here, μ _i : population mean vector, Σ _i : population variance covariance matrix The basis values obtained by the statistical value / matrix calculation means 636 are the population mean vector μ _i and the population variance covariance matrix Σ _i . (X-
μ _i ) ′ is a transposed matrix of a matrix formed by each element of the vector (x−μ _i ). Here, by substituting the equation into the equation, taking the logarithm, and removing the terms common to all the categories, the following equation is obtained as a discriminant function of the quadratic form. When the multivariate data x = (x ₁ , x ₂ ,..., X _m ) is substituted into the equation, the value of the equation is not the value of the probability belonging to the i-th category, but a part that determines the magnitude of the probability. .
Σ _i ^-1 is the inverse of matrix Σ _i . The belonging probabilities f _i (x) (i = 1, 2,... g) are obtained for g categories, and it is determined that the solder joint to be inspected belongs to the category having the largest belonging probability. The quality of the solder joint to be inspected is determined according to whether the determined category is a category of a good product or a category of a defective product.

The above determination will be described with reference to the drawings. FIG. 17 is a graph of a probability density function for a first-order random variable. The case where there are two categories will be described. In the figure, f ₁ (x) and f ₂ (x) are categories E, respectively.
1, E2. If random variable given by multivariate data to be inspected is a x _a is inspected becomes each probability P ₁ and P ₂ belonging to the category E1 and E2. At this time, the inspection target than P _1> P ₂ is judged to belong to the category E1. If random variable given by multivariate data to be inspected is a x _b is the probability that the inspection target belongs to the category E1 and E2 becomes P ₃ and P _4, respectively. At this time, it is determined that the inspection target belongs to the category E2 from P ₃ <P ₄ .

Although the embodiment has been described mainly for the case where the inspection is performed after soldering the surface mount components by reflow, the extraction area and the discriminant function are prepared exclusively by using the same hardware configuration. Accordingly, the present invention can also be applied to inspection of the application state of cream solder, appearance inspection of flow solder for discrete components, and the like. Also,
By the uniformity in the circumferential direction by the cylindrical diffuser and the lighting means weighted to an arbitrary irradiation angle, cream solder targets spherical solder particles, and flow solder targets fillet portions. The category can be determined with high accuracy by recognizing a three-dimensional feature from the gray image.

[0021]

According to the present invention, the shading information is stored in the solder joint.
A plurality of extraction regions are defined for the captured image serving as the angle information of the inclined surface, and the average light amount of each extraction region is obtained. The average light amount is treated as multivariate data, and the multivariate data is substituted into a probability density function provided for each category, and it is determined that the solder joint to be inspected belongs to the category having the highest probability. Thus, the quality of the solder joint can be globally determined without binarizing the image and providing a specific threshold value.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of an embodiment of the present invention.

FIG. 2 is a diagram showing a specific configuration example of the device of FIG. 1;

FIG. 3 is a diagram showing a specific configuration example of the device of FIG. 1;

FIG. 4 is an explanatory diagram of a detection principle.

FIG. 5 is an explanatory diagram of a detection principle.

FIG. 6 is an explanatory diagram of a detection principle.

FIG. 7 is a diagram illustrating an example of a fillet shape of solder;

FIG. 8 is a diagram illustrating an example of a fillet shape of solder;

FIG. 9 is a diagram showing an example of a fillet shape of solder.

FIG. 10 is a diagram showing an example of a solder fillet shape.

FIG. 11 is a diagram showing an example of a fillet shape of solder.

FIG. 12 is a diagram showing an example of a fillet shape of solder.

FIG. 13 is a diagram showing the correspondence between fillet shapes and determination codes.

FIG. 14 is a diagram showing a window arranged in a captured image of a solder joint.

FIG. 15 is a graph showing variable values for each category.

FIG. 16 is a diagram illustrating a specific configuration example of a portion that performs an inspection process.

FIG. 17 is a graph of a probability density function.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Printed board 11 Solder joint part 2 Mounting part 4 Illumination means 41 Ring-shaped light source 42 Diffusion plate 5 Camera 6 Inspection processing system control part 633 Window definition means 634 Window average luminance extraction means 636 Statistical value / matrix calculation means 641 Judgment means 65 Storage means

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01B 11/00-11/30

Claims (1)

(57) [Claims] In a solder joint inspection apparatus for inspecting the appearance of a solder joint formed on a printed circuit board for defects, the irradiation light of a light source is incident on an end face of a cylindrical diffusion plate. , total reflection and diffuse reflection of the inner surface of the diffuser plate to light entering, expanding
The irradiation angle and intensity of light emitted from the inner peripheral surface of the
Change continuously along the length direction,
Illumination means that continuously changes the brightness of the reflected light emitted from the solder joint and traveling in the length direction of the diffuser according to the inclination angle of the reflection position, and the reflected light traveling in the length direction of the diffuser capturing a reflection image of the solder joint receives, shading information solder
An imaging unit for obtaining a captured image serving as angle information of the inclined surface of the joint; and a plurality of extraction regions in the captured image of the imaging unit, which are regions required for extracting a feature of the three-dimensional shape of the solder joint. Area defining means for individually defining, light quantity calculating means for calculating an average light quantity of a picked-up image in each extraction area, and each extraction area for a plurality of samples including non-defective and defective three-dimensionally shaped solder joints. Function generating means for generating, for each sample, a probability density function indicating the distribution of the probability that a solder joint belongs to a sample, using multivariate data consisting of the average light amount as a variable; and for each sample generated by this function generating means. Function storage means for storing the probability density function of the above, and multivariate data obtained by the light quantity calculating means for the solder joint to be inspected are substituted into each probability density function, and the solder joint to be inspected belongs to each sample. Sure Determining that the shape of the solder joint to be inspected belongs to the sample having the highest probability, and determining the presence or absence of a defect in the solder joint according to the determined sample. Inspection device for solder joints.