JP2022060731A - Inspection method for printed board - Google Patents

Abstract

To provide a highly accurate inspection method for a printed board.SOLUTION: In an inspection method for a printed board according to an embodiment, a reference W is determined using a residual of a regression line between a first net resistance (first resistance) and a second net resistance (second resistance) from the second resistance. Then, evaluation of a printed board 100 to be newly manufactured is conducted using the reference W.SELECTED DRAWING: Figure 2

Description

The present invention relates to an inspection method for a printed wiring board.

Patent Document 1 discloses a method for determining the quality of an electronic component. According to paragraph 44 of Patent Document 1, the measured value and the reference range are compared. When the measured value is within the reference range, the electronic component is determined to be good, and when the measured value is outside the reference range, the electronic component is determined to be defective. As an example of the measured value, Patent Document 1 cites resistance.

Japanese Unexamined Patent Publication No. 2013-76633

[Problems of Patent Document 1]
As shown in FIG. 1A, a general printed wiring board 100 includes a conductor circuit 4 and a via conductor 3 connected to the conductor circuit 4. As the density of the printed wiring board 100 increases, the width of the conductor circuit 4 and the diameter of the via conductor 3 become smaller, and the length of the wiring 5 formed by the conductor circuit 4 and the via conductor 3 becomes longer. .. The wiring 5 extends from the front surface F of the printed wiring board 100 to the back surface S. The number of wires 5 is also increasing. Therefore, when a large number of printed wiring boards of the same type are manufactured, an abnormal printed wiring board may be included in the large number of printed wiring boards. An example of the anomaly is shown in FIG. 1 (B). FIG. 1B is a plan view. In FIG. 1B, the conductor circuit 4 includes the missing portion 6. In the missing portion 6, a part of the conductor circuit 4 is missing. In the missing portion 6, the width of the conductor circuit 4 is extremely narrow. However, when the length of the wiring 5 is long or the width of the wiring 5 is narrow, it is predicted that the resistance of the wiring 5 does not change significantly depending on the presence or absence of the missing portion 6. Another example of the anomaly is shown in FIG. 1 (C). FIG. 1C is a cross-sectional view. In FIG. 1C, the foreign matter 7 is present at the connection portion between the conductor circuit 4 and the via conductor 3. An example of the foreign matter 7 is a resin that forms the resin insulating layer 2. In the example of FIG. 1C, the foreign matter 7 reduces the contact area between the conductor circuit 4 and the via conductor 3. However, when the length of the wiring 5 is long or the diameter of the via conductor 3 is small, it is predicted that the resistance of the wiring 5 does not change significantly depending on the presence or absence of the foreign matter 7. Therefore, as the density of the printed wiring board 100 increases, it is expected that it will be difficult to judge the quality of the wiring 5 by comparing the measurement result with the reference range. Then, when a printed wiring board containing the abnormal wiring shown in FIG. 1B is used in the market, it is presumed that the life will be extremely shortened. Similarly, it is estimated that the life of a printed wiring board containing the abnormal wiring shown in FIG. 1C will be extremely shortened when it is used in the market.

The method for inspecting a printed wiring board according to the present invention is to obtain a plurality of printed wiring boards by repeatedly manufacturing a printed wiring board having a first net and a second net, and to obtain the printed wiring board of the first net of the printed wiring board. Measuring the resistance (first resistance), measuring the resistance (second resistance) of the second net of the printed wiring board, and the first set formed by the first resistance and the second resistance. [(X, Y) = (1st resistance, 2nd resistance)], and plotting the first set of data on the XY plane to create a scatter diagram. Using the scatter diagram, the regression line of the first resistance and the second resistance is obtained, and the predicted value of the second resistance is calculated by substituting the first resistance into the regression line. By adding the predicted value of the second resistance to the data of the first set, the data of the second set [(X, Y, predicted value) = (predicted value of the first resistance, the second resistance, the second resistance). ], The difference (residual) between the second resistance and the predicted value of the second resistance is calculated, and the residual is added to the data of the second set to set the third set. Data [(X, Y, predicted value, residual) = (first resistance, second resistance, predicted value of second resistance, residual of second resistance)] and the residual (first). The reference of the second net is determined by using the residual of two resistors), another printed wiring board (inspection target) having the first net and the second net is manufactured, and the inspection target. To measure the resistance (third resistance) of the first net, and to measure the resistance (fourth resistance) of the second net to be inspected.
By substituting the third resistance into the regression line, the predicted value of the fourth resistance is obtained, and the difference between the fourth resistance and the predicted value of the fourth resistance (residual of the fourth resistance) is obtained. And to compare the residual of the fourth resistance with the reference. The correlation coefficient of the regression line is 0.7 or more.

[Effect of embodiment]
The inspection method of the embodiment utilizes a regression line having a strong correlation. Therefore, the difference (residual) between the predicted value of the second resistance and the measured value of the second resistance obtained by substituting the measured value of the first resistance into the regression line should be small. Further, since the first resistance and the second resistance have a strong correlation, the residual varies within an appropriate range. The fluctuation of the residual is stable. The range of fluctuation should be small. In the inspection method of the embodiment, the standard is determined from the residual obtained by using the regression line having a strong correlation. Therefore, the criteria used in the inspection method of the embodiment can have high sensitivity to outliers. Therefore, according to the inspection method of the embodiment, it is possible to inspect the printed wiring board including the abnormality shown in FIG. 1 (B). The printed wiring board containing the abnormality shown in FIG. 1C can be inspected. By evaluating the printed wiring board by the inspection method of the embodiment, it is possible to provide a printed wiring board that maintains high reliability for a long period of time. For example, the resistance value of the wiring in the printed wiring board is stable for a long period of time.

1 (A) is a cross-sectional view of a printed wiring board according to an embodiment of the present invention, FIGS. 1 (B) and 1 (C) show an example of abnormality, and FIG. 1 (D) shows an example of resistance. show. FIG. 2A is a graph showing a scatter plot and a regression line, FIG. 2B is a graph showing a residual, and FIG. 2C shows a temporary scatter plot.

An outline of the procedure of the inspection method of the embodiment is explained.
First, the printed wiring board 100 having the first net N1 and the second net N2 is repeatedly manufactured. The resistance of the first net N1 (first resistance) and the resistance of the second net N2 (second resistance) are measured. The first resistance is the variable X and the second resistance is the variable Y. The first and second resistances are plotted on the XY plane. A scatter plot is obtained. From the scatter plot, the relationship between the first resistance and the second resistance (regression line 11) can be obtained. The correlation coefficient 11 of the regression line 11 is 0.7 or more. The correlation coefficient 11 is preferably 0.9 or more. Further, the correlation coefficient 11 is preferably 0.95 or more. The residual R of the second resistance is obtained by using the second resistance and the regression line 11. The reference W of the inspection method of the embodiment is determined by using the residual R of the second resistance.
You can add to create a temporary scatter plot between plotting and scatter plotting. It is possible to add to remove outliers from the temporary scatter plot between the temporary scatter plot creation and the scatter plot creation. The reference W can be determined using the standard deviation Σ of the residual R.
A new printed wiring board (another printed wiring board) 1001 is manufactured. The new printed wiring board 1001 is subject to inspection. The resistance of the first net N1 (third resistance) and the resistance of the second net N2 (fourth resistance) of the inspection target 1001 are measured. By substituting the third resistance into the regression line 11, the predicted value of the fourth resistance can be obtained. By comparing the difference R1 between the fourth resistance and the predicted value of the fourth resistance and the reference W, the inspection target 1001 is inspected. When the difference R1 is equal to or less than the reference W, the second net N2 is a non-defective product. Even if the printed wiring board 100 undergoes a heat cycle, the resistance of the second net N2 is stable. The second net N2 does not include the missing portion 6. Starting from the missing portion 6, the second net N2 is not disconnected. The second net N2 does not include a missing portion 6 that causes disconnection.

The printed wiring board 100 shown in FIG. 1 (A) is repeatedly manufactured. The number of printed wiring boards manufactured repeatedly is N (number N). The printed wiring board 100 is repeatedly manufactured with the aim of obtaining the same printed wiring board. The printed wiring board 100 is numbered. 1 is attached to the printed wiring board 100 manufactured first. After that, the printed wiring boards 100 are numbered in order. N is an integer of 30 or more. The number M is assigned to the printed wiring board 100 manufactured in the Mth position. M is an integer of 1 or more and N or less. As shown in FIG. 1A, the printed wiring board 100 has a front surface F and a back surface S opposite to the front surface F. Further, the printed wiring board 100 has a plurality of conductor layers 1 and a plurality of resin insulating layers 2. The conductor layer 1 and the resin insulating layer 2 are alternately laminated. The different conductor layers 1 are connected by a via conductor 3 penetrating the resin insulating layer 2. Each conductor layer 1 includes a plurality of conductor circuits 4. The conductor layer 1 formed on the surface F includes a plurality of electrodes 8. Electronic components are mounted on the printed wiring board 100 via the electrodes 8. The conductor layer 1 formed on the back surface S includes a plurality of terminals 9. The printed wiring board 100 is connected to the motherboard via the terminal 9. The electrode 8 includes a first electrode 81 and a second electrode 82. The terminal 9 includes a first terminal 91 and a second terminal 92.

The first electrode 81 and the first terminal 91 are electrically connected to the via conductor 3 in the printed wiring board 100 via the conductor circuit 4. The path formed by the via conductor 3 and the conductor circuit 4 that electrically connect the first electrode 81 and the first terminal 91 is the first net. The first net N1 includes a first electrode 81 and a first terminal 91.

The second electrode 82 and the second terminal 92 are electrically connected to the via conductor 3 in the printed wiring board 100 via the conductor circuit 4. The path formed by the via conductor 3 and the conductor circuit 4 that electrically connect the second electrode 82 and the second terminal 92 is the second net N2. The second net includes a second electrode 82 and a second terminal 92.

The first net N1 and the second net N2 are independent. The first net N1 has the length of the first wiring (first wiring length). The second net N2 has the length of the second wiring (second wiring length). The first wiring length is the sum of the lengths of each conductor circuit 4 in the first net N1. The second wiring length is the sum of the lengths of each conductor circuit 4 in the second net N2. The first wiring length and the second wiring length are almost equal. When the conductor circuit 4 included in the first net N1 and the conductor circuit 4 included in the second net N2 are formed in the same conductor layer 1, the lengths of both are substantially the same. The thickness of both is almost the same. The widths of both are almost the same.

There are many nets N. The first net N1 and the second net N2 can be selected from a large number of nets N. Net N includes net A and net B. Within the same conductor layer 1, it is preferable that the width of the conductor circuit 4 forming the net A and the width of the conductor circuit 4 forming the net B are substantially equal to each other. Within the same conductor layer 1, it is preferable that the thickness of the conductor circuit 4 forming the net A and the thickness of the conductor circuit 4 forming the net B are substantially equal to each other.

(First example of selection)
Each net N has a wiring length (wiring length). The wiring length is the sum of all the conductor circuits 4 forming the net N. The wiring lengths of net A and net B are compared. When the ratio of the wiring length A of the net A to the wiring length B of the net B (wiring length A / wiring length B) is 0.90 or more and 1.1 or less, the net A is the first net N1 and the net. B is the second net N2. When the ratio (wiring length A / wiring length B) is 0.95 or more and 1.05 or less, the first net N1 and the second net N2 can be easily narrowed down from a large number of nets N.

(Second example of selection)
The longest conductor circuit (longest conductor circuit) 4L is extracted from the conductor circuit 4 forming the net N. When the ratio of the length 4LA of the longest conductor circuit 4L of the net A to the length 4LB of the longest conductor circuit 4L of the net B (length 4LA / length 4LB) is 0.97 or more and 1.03 or less. , Net A is the first net N1, and net B is the second net N2.

(Third example of selection)
The net A has a conductor circuit A in each conductor layer 1, and the net B has a conductor circuit B in each conductor layer 1. Each conductor circuit A has a length (length A), and each conductor circuit B has a length (length B). The length A and the length B are compared for each conductor layer 1. The ratio of length A to length B (length A / length B) is calculated. When the ratio (length A / length B) is 0.9 or more and 1.1 or less for each conductor layer 1, the net A is the first net N1 and the net B is the second net N2. When the ratio (length A / length B) for each conductor layer 1 is 0.95 or more and 1.05 or less, the first net N1 and the second net N2 can be easily selected from among a large number of nets N. You can squeeze it.

(4th example of selection)
The resistance of each net N in the printed wiring board 100 is measured. If N is 30, the number of resistances in each net is 30. The relationship between the resistance A and the resistance B (regression line AB) is obtained by using the resistance of the net A (resistance A) and the resistance of the net B (resistance B) as variables. Candidates for net B are all net N except net A. In that case, the number of combinations (net A and net B (all nets N other than net A)) AB that create the regression line AB is (N1). Therefore, the number of regression lines is (N1). The correlation coefficient AB of each regression line AB is obtained from each regression line AB. The number of correlation coefficients AB is also (N1). Combination AB7 having a correlation coefficient AB of 0.7 or more is extracted from all combinations. In the inspection method of the embodiment, the first net N1 and the second net N2 are combined to form AB7. When there are a plurality of combinations AB7, all combinations AB7 are the targets of the inspection method of the embodiment. Among the combinations AB7, it is preferable to select the combination AB9 having a correlation coefficient AB of 0.9 or more. Further, among the combinations AB7, it is preferable to select the combination AB95 having a correlation coefficient AB of 0.95 or more. The larger the correlation coefficient AB, the higher the accuracy of the inspection method of the embodiment.

(Fifth example of selection)
In the same conductor layer 1, the conductor circuit 4 forming the net A and the conductor circuit 4 forming the net B run substantially in parallel. In that case, net A is the first net N1 and net B is the second net N2.

If the relationship between the first net N1 and the second net N2 satisfies one of the first to fifth examples of selection, the design of the first net N1 and the design of the second net N2 are similar.

The resistance (resistance value) of the first net N1 and the second net N2 of each printed wiring board 100 is measured. An example of resistance measurements is shown in FIG. 1 (D). The resistance (first resistance) of the first net N1 is the resistance between the first electrode 81 and the first terminal 91. The resistance (second resistance) of the second net N2 is the resistance between the second electrode 82 and the second terminal 92.

In the embodiment, the first resistance is the variable X and the second resistance is the variable Y. The value of the first resistance (first resistance) is the X-axis, and the value of the second resistance (second resistance) is the Y-axis. The first resistance and the second resistance form the first set of data [(X, Y) = (first resistance, second resistance)]. The first set of data is referred to as the first data. The first resistance and the second resistance in one first data are the resistances of the first net N1 and the second net N2 belonging to the same printed wiring board 100. The number of first data is N. The first data is plotted on the XY plane including the X-axis formed by the first resistance and the Y-axis formed by the second resistance. A temporary scatter plot is obtained. A temporary scatter plot is shown in FIG. 2 (C). In FIG. 2C, one first data is drawn by a circle (◯). The number of first data plotted on the temporary scatter plot is K (number K). The number K is a number N or less. The number K is 30 or more. When the number K and the number N are the same, all the first data are plotted. When the number K and the number N are different, the first data plotted from all the first data is randomly selected. Alternatively, when a temporary scatter plot is created, if the total first data contains anomalous data, the anomalous first data is not plotted.

Outliers are removed from the temporary scatter plot. If the temporary scatter plot does not contain outliers, the outliers will not be removed. In that case, creating a temporary scatter plot is skipped.

By observing the temporary scatter plot, the first data that deviates from the distribution is selected. The first data away from the distribution are outliers. Such first data is removed from the temporary scatter plot.

Outliers are removed from the temporary scatter plot using statistical techniques.
[Statistical method 1]
The average value X1 of the first resistance and the standard deviation σ1 are obtained. The number of first resistances for obtaining X1 and σ1 is several N. The average value X2 of the second resistance and the standard deviation σ2 are obtained. The number of second resistances for obtaining X2 and σ2 is several N.
If the first resistances are outside the predetermined range, the first data including those first resistances are outliers. The predetermined range is X1 ± 2 × σ1. Alternatively, the predetermined range is X1 ± 3 × σ1.
If the second resistances are outside the predetermined range, the first data including those second resistances are outliers. The predetermined range is X2 ± 2 × σ2. Alternatively, the predetermined range is X2 ± 3 × σ2.
[Statistical method 2]
Outliers are removed by the Smirnov-Grabs test.
[Statistical method 2]
Outliers are removed by the Thompson test.

A scatter plot is obtained by removing outliers from the temporary scatter plot. If the temporary scatter plot does not contain outliers, the temporary scatter plot forms the scatter plot. In that case, the temporary scatter plot and the scatter plot are the same. The first data plotted on the scatter plot is referred to as the data in the scatter plot (fifth data). A scatter plot of the embodiment is shown in FIG. 2 (A). In FIG. 2A, one fifth data is drawn by a circle (◯).

The regression line (Y = aX + b) 11 and the correlation coefficient are obtained using the scatter plot. For example, the regression line 11 and the correlation coefficient can be obtained by using Excel, which is spreadsheet software. An example of the regression line 11 is shown in FIG. 2 (A). The correlation coefficient is 0.7 or more. When the correlation coefficient is 0.7 or more, the inspection method using the regression line 11 is effective. In the inspection method of the embodiment, the correlation coefficient is 0.7 or more. The larger the correlation coefficient, the stronger the correlation between the first resistance and the second resistance. Therefore, the accuracy of the inspection method using the regression line 11 can be improved. When the correlation coefficient is 0.9 or more, the wiring in the printed wiring board 100 does not break for a long period of time. When the correlation coefficient is 0.95 or more, the resistance of the wiring in the printed wiring board 100 is stable for a long period of time. As described above, when the correlation coefficient is 0.9 or more, the outflow of the printed wiring board 100 including the abnormality shown in FIG. 1B can be suppressed. Similarly, the outflow of the printed wiring board 100 including the abnormality shown in FIG. 1C can be suppressed.

In the inspection method of the embodiment, the correlation coefficient affects the accuracy of the inspection. Therefore, it is preferable to control the correlation coefficient in order to improve the accuracy of the inspection. An example of controlling the correlation coefficient is shown below.

According to the fourth example of selection, the first net N1 and the second net N2 are selected by using the magnitude of the correlation coefficient. Therefore, the fourth example of selection can control the correlation coefficient. In the fourth example of selection, the inspectable net N is selected. If the correlation coefficient between net A and net B is 0.7 and the correlation coefficient between net A and net C is 0.9, net A is selected as the first net N1. Net C is selected as the second net N2. This makes it possible to improve the accuracy of the inspection.

By using the first example of selection and the fifth example of selection, the correlation coefficient is 0.9 or more. When a scatter plot is created via a temporary scatter plot, the correlation coefficient is 0.95 or higher.

By using the third example of selection, the correlation coefficient is 0.7 or more. When a scatter plot is created via a temporary scatter plot, the correlation coefficient is 0.9 or higher.

By using the second example of selection, the correlation coefficient is 0.7 or more.

By substituting the first resistance in the fifth data into the regression line 11, the predicted value of the second resistance can be obtained. By adding the predicted value of the second resistance to the first data, the second set of data [(X, Y, predicted value) = (predicted value of the first resistance, the second resistance, the second resistance)] is formed. Will be done. The second set of data is referred to as the second data. Each data (X, Y, predicted value) included in one second data is based on the first net N1 and the second net N2 in one printed wiring board 100. The number of second data is 30 or more. The second data is created for each fifth data.

The difference (residual) R between the measured value of the second resistance and the predicted value of the second resistance is obtained. By adding the residual (residual of the second resistance) R to the data of the second set, the data of the third set [(X, Y, predicted value, residual) = (first resistance, second resistance, The predicted value of the second resistance, the residual of the second resistance)] is formed. The third set of data is called the third data. The third data is formed for each second data. The number of third data is 30 or more.

The third data is associated with the number of the printed wiring board 100. By adding the number M of the printed wiring board 100 to the data of the third set, the data of the fourth set [(X, Y, predicted value, residual, number of the printed wiring board) = (first resistance, second). Resistance, predicted value of second resistance, residual of second resistance, M)] is formed. The fourth set of data is referred to as the fourth data. Each data (X, Y, predicted value, residual of the second resistance, M) included in one fourth data is based on the first net N1 and the second net N2 in one printed wiring board 100. .. The relationship between the number M of the printed wiring board 100 and the residual of the second resistance is plotted. The relationship is shown in FIG. 2 (B). FIG. 2B is an XY graph. In FIG. 2B, the X-axis is the number M of the printed wiring board 100, and the Y-axis is the residual of the second resistance. In FIG. 2B, the numbers M of the printed wiring boards 100 are arranged in ascending order.

FIG. 2B is used to determine the reference W of the inspection method of the embodiment. In order to determine the reference W, the standard deviation Σ of the residual of the second resistance is obtained.
Criteria W satisfy any of the following relationships 1, 2 and 3.
Relationship 1:10 x standard deviation Σ ≤ reference W ≤ 30 x standard deviation Σ
Relationship 2: 15 x standard deviation Σ ≤ reference W ≤ 25 x standard deviation Σ
Relationship 3: 17 x standard deviation Σ ≤ reference W ≤ 23 x standard deviation Σ
For example, the reference W is 20 times the standard deviation Σ.

A new printed wiring board (inspection target) 1001 is manufactured. At that time, the goal is that the printed wiring board 100 and the inspection target 1001 are the same. The inspection target 1001 is manufactured aiming at the same printed wiring board as the printed wiring board 100 for determining the standard.

The resistance of the first net N1 and the resistance of the second net N2 of the inspection target 1001 are measured. The resistance value of the first net N1 is the third resistance (measured value of the third resistance), and the resistance value of the second net N2 is the fourth resistance (measured value of the fourth resistance). By substituting the third resistance into the regression line 11 shown in FIG. 2A, the predicted value of the fourth resistance can be obtained. The residual R1 of the 4th resistance is obtained from the measured value of the 4th resistance and the predicted value of the 4th resistance.

The residual of the fourth resistance and the reference W are compared. When the residual of the fourth resistance is equal to or less than the reference W, the second net N2 in the inspection target 1001 is determined to be a non-defective product.

The inspection method of the embodiment utilizes a regression line 11 having a strong correlation. Then, a reference is made using the residual R. Therefore, by evaluating the printed wiring board using the inspection method of the embodiment, it is possible to provide a printed wiring board that maintains high reliability for a long period of time. For example, the resistance value of the wiring in the printed wiring board is stable for a long period of time.

[Example 1]
The printed wiring board 100 including the first net N1 and the second net N2 is repeatedly manufactured. In the first embodiment, the first net N1 and the second net N2 satisfy the fourth example of selection. The number N of the printed wiring board 100 is 30. The resistances (first resistance and second resistance) of the first net N1 and the second net N2 in each printed wiring board 100 are measured. The first and second resistances are plotted on the XY plane. A temporary scatter plot is created. Temporary scatter plots do not include outliers. Therefore, the temporary scatter plot is a scatter plot. The correlation coefficient between the first resistance and the second resistance is 0.95 or more. The reference W is determined using the scatter plot. The reference W is 20 times the standard deviation Σ of the residual R of the second resistance. The inspection target 1001 is manufactured. The resistance of the first net (third resistance) and the resistance of the second net (fourth resistance) of the inspection target 1001 are measured. By substituting the third resistance into the regression line 11, the predicted value of the fourth resistance can be obtained. The difference (residual of the 4th resistance) R1 between the 4th resistance and the predicted value of the 4th resistance is obtained. The residual R1 of the fourth resistance is equal to or less than the reference W.

[Example 2]
The number N of Example 2 is 35. The first net N1 and the second net N2 satisfy the first example of selection. The temporary scatter plot of Example 2 includes outliers. The correlation coefficient of the temporary scatter plot is 0.9. Outliers are removed using statistical method 1. The predetermined ranges adopted in Example 2 are shown below.
Predetermined range of first resistance: X1 ± 2 × σ1
Predetermined range of second resistance: X2 ± 2 × σ2
Outliers are removed from the temporary scatter plot. The number K of the third data is 30. A scatter plot is created using the third data. The correlation coefficient of the scatter plot is 0.95. The reference W is determined using the scatter plot. The reference W is 25 times the standard deviation Σ of the second residual. The inspection target 1001 is manufactured. The resistance of the first net N1 (third resistance) and the resistance of the second net N2 (fourth resistance) of the inspection target 1001 are measured. Similar to Example 1, the inspection target 1001 is evaluated.

[Example 3]
The number N of Example 2 is 35. The first net N1 and the second net N2 satisfy the second example of selection. The temporary scatter plot of Example 2 includes outliers. The correlation coefficient of the temporary scatter plot is 0.6. Outliers are removed using statistical method 2. Outliers are removed from the temporary scatter plot. The number K of the third data is 30. A scatter plot is created using the third data. The correlation coefficient of the scatter plot is 0.7. The reference W is determined using the scatter plot. The reference W is 15 times the standard deviation Σ of the second residual. The inspection target 1001 is manufactured. The resistance of the first net (third resistance) and the resistance of the second net (fourth resistance) of the inspection target 1001 are measured. Similar to Example 1, the inspection target 1001 is evaluated.

[Example 4]
The method of removing outliers differs between Example 3 and Example 4. Other than that, Example 4 conforms to Example 3. Outliers are removed using statistical method 3

[Example 5]
The number N of Example 5 is 50. The first net N1 and the second net N2 satisfy the third example of selection. The temporary scatter plot of Example 5 includes outliers in the temporary scatter plot. The correlation coefficient of the temporary scatter plot is 0.9. Outliers are removed using statistical method 1. The predetermined ranges adopted in Example 5 are shown below.
Predetermined range of first resistance: X1 ± 3 × σ1
Predetermined range of second resistance: X2 ± 3 × σ2
Outliers are removed from the temporary scatter plot. The number K of the third data is 45. A scatter plot is created using the third data. The correlation coefficient of the scatter plot is 0.95. The reference W is determined using the scatter plot. The reference W is 20 times the standard deviation Σ of the second residual. The inspection target 1001 is manufactured. The resistance of the first net (third resistance) and the resistance of the second net (fourth resistance) of the inspection target 1001 are measured. Similar to Example 1, the inspection target 1001 is evaluated.

[Example 6]
The first net N1 and the second net N2 satisfy the fifth example of selection. Other than that, Example 6 conforms to Example 5.

1: Conductor layer 2: Resin insulating layer 3: Via conductor 4: Conductor circuit 5: Wiring 8: Electrode 9: Terminal 100: Printed wiring board

The method for inspecting a printed wiring board according to the present invention is to obtain a plurality of printed wiring boards by repeatedly manufacturing a printed wiring board having a first net and a second net, and to obtain the printed wiring board of the first net of the printed wiring board. Measuring the first resistance , which is a resistance, measuring the second resistance , which is the resistance of the second net of the printed wiring board, and the second resistance formed by the first resistance and the second resistance. Creating a scatter diagram by preparing a set of data [(X, Y) = (first resistance, second resistance)] and plotting the first set of data on the XY plane. The predicted value of the second resistance is calculated by obtaining the regression line of the first resistance and the second resistance using the scatter diagram and substituting the first resistance into the regression line. By adding the predicted value of the second resistance to the data of the first set, the data of the second set [(X, Y, predicted value) = (prediction of the first resistance, the second resistance, the second resistance). (Value)] is prepared, the residual of the second resistance, which is the difference between the second resistance and the predicted value of the second resistance, is calculated, and the standard deviation of the residual of the second resistance is obtained . That, the reference of the second net is determined by using the residual of the second resistance, and another printed wiring board (inspection target) having the first net and the second net is manufactured. , Measuring the third resistance , which is the resistance of the first net of the inspection target, measuring the fourth resistance , which is the resistance of the second net of the inspection target, and the regression line. By substituting the third resistance, the predicted value of the fourth resistance is obtained, the residual of the fourth resistance, which is the difference between the fourth resistance and the predicted value of the fourth resistance, is obtained, and the first. It has 4 resistance residuals and comparing the criteria. The correlation coefficient of the regression line is 0.7 or more, and the reference and the standard deviation satisfy the following relationship.
Relationship: 10 x standard deviation ≤ 30 x standard deviation

Claims (14)

By repeatedly manufacturing a printed wiring board having a first net and a second net, a plurality of printed wiring boards can be obtained.
Measuring the resistance (first resistance) of the first net of the printed wiring board and
Measuring the resistance (second resistance) of the second net of the printed wiring board and
Preparing the first set of data [(X, Y) = (first resistance, second resistance)] formed by the first resistance and the second resistance, and
By plotting the first set of data on the XY plane, a scatter plot can be created.
Using the scatter plot, the regression line of the first resistance and the second resistance can be obtained.
By substituting the first resistance into the regression line, the predicted value of the second resistance can be calculated.
By adding the predicted value of the second resistance to the data of the first set, the data of the second set [(X, Y, predicted value) = (predicted value of the first resistance, the second resistance, the second resistance). ] And to prepare
To calculate the difference (residual) between the second resistance and the predicted value of the second resistance,
By adding the residual to the data of the second set, the data of the third set [(X, Y, predicted value, residual) = (first resistance, second resistance, predicted value of the second resistance, first 2 Residual of resistance)] and
Using the residual (residual of the second resistance) to determine the reference for the second net,
To manufacture another printed wiring board (inspection target) having the first net and the second net, and
Measuring the resistance (third resistance) of the first net to be inspected, and
Measuring the resistance (fourth resistance) of the second net to be inspected, and
By substituting the third resistance into the regression line, the predicted value of the fourth resistance can be obtained.
Finding the difference between the 4th resistance and the predicted value of the 4th resistance (residual of the 4th resistance)
A method for inspecting a printed wiring board having the residual of the fourth resistance and the comparison of the reference.
The correlation coefficient of the regression line is 0.7 or more. The method for inspecting a printed wiring board according to claim 1, further comprising obtaining the standard deviation of the residual. The method for inspecting a printed wiring board according to claim 2, wherein the reference and the standard deviation satisfy the following relationship.
Relationship: 10 x standard deviation ≤ 30 x standard deviation In the method for inspecting a printed wiring board according to claim 3, when the residual of the fourth resistance is equal to or less than the reference, the second net to be inspected is a good product. The method for inspecting a printed wiring board according to claim 1, wherein the number of the plurality of printed wiring boards is 30 or more. The method for inspecting a printed wiring board according to claim 1, wherein the repeated manufacturing aims at obtaining the same printed wiring board. The method for inspecting a printed wiring board according to claim 1, further, creating a temporary scatter plot between preparing the first set of data and creating the scatter plot, and the temporary. Includes removing outliers from a typical scatter plot. In the method for inspecting a printed wiring board according to claim 1, the length of the wiring forming the first net and the length of the wiring forming the second net are substantially equal to each other. In the method for inspecting a printed wiring board according to claim 1, two nets having a correlation coefficient of 0.7 or more are selected as the first net and the second net. In the method for inspecting a printed wiring board according to claim 1, two nets having a correlation coefficient of 0.9 or more are selected as the first net and the second net. In the method for inspecting a printed wiring board according to claim 1, two nets having a correlation coefficient of 0.95 or more are selected as the first net and the second net. The method for inspecting a printed wiring board according to claim 1, wherein the design of the second net and the design of the first net are similar. The method for inspecting a printed wiring board according to claim 1, wherein the number of data in the first set is 30 or more. The method for inspecting a printed wiring board according to claim 2, wherein the number of residuals is 30 or more.

Images