JPH1040289A - Product quality improvement support system - Google Patents

Abstract

PROBLEM TO BE SOLVED: Let even an inexperienced operator easily know a process or a parameter causative of defect by comparing a standard deviation obtained from each intersection with the statistical evaluation amount of a general intersection, evaluating a correlation between the defect and the parameter and teaching the process and parameter causative of defect. SOLUTION: A process tolerance contribution rate calculating part 70 calculates for example, a rate causative of dispersion on parts dimensions for each process, in view of the values of a worst value deviation, a total standard deviation, etc., in product specification which are obtained by a worst value deviation/total standard deviation acquisition part 60. Then, a defective process teaching part 75 teaches which process is concerned with the generation of defective cause. Then, a parameter design calculating part 80 calculates which is concerned with defect in the parameters concerning fabrication, and a design/ fabrication defective factor improvement condition teaching part 85 teaches a process and a parameter estimated to be concerned with the defective process. Thus, the product quality can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a product quality improvement support system, in which a tolerance is determined for each part, and there is a process of assembling them to produce a product. The present invention relates to a product quality improvement support system suitable for efficiently improving the quality of a product requiring a large number of products and requiring high reliability.

[0002]

2. Description of the Related Art First, a description will be given of a product development procedure using a conventional general method with reference to FIG.

FIG. 21 is a flowchart showing a conventional product development procedure.

In the conventional method, in the product design (S211), the quality and reliability are evaluated (S212) before the production (S215), and the costs are compared (S213). This required a lot of time to correct the product drawing and redesign the product.

[0005] Conventionally, to improve product quality,
There is known a general technique called design review at the time of design, in which an expert in design, manufacture, inspection, and the like determines quality based on experience and points out a part requiring improvement.

[0006]

The work in the product development described in the prior art can be said to be an indispensable work that must be performed at the design stage in order to improve the quality of the product.

[0007] However, the above-mentioned prior art requires a skilled technician to perform inspections, and it is necessary to determine which process is bad and which parameter is the most defective factor in a product. There is a problem that it is difficult to evaluate in terms of time, and it takes a very long time for that. And
When a product contains a new part, quantitative data has not been obtained, and it has been difficult to perform statistical analysis.

[0008] Further, when performing the inspection, there is a problem that it is necessary to inspect all the parts or to perform a sampling inspection, and it is difficult to obtain a clear index.

Further, there is a problem that it is difficult to obtain a clear index of how much debugging must be performed in a factory for quality improvement.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems of the prior art, and its object is to provide a short-time test even if the user is not skilled in inspection and does not have detailed knowledge about products. So, even if the product contains new parts, it is easy to know which process is causing the defect and which parameter contributes to the defect, and how much the product quality can be improved. To provide a support system.

Another object of the present invention is to provide a product quality improvement support system capable of obtaining a clear index of whether a total inspection or a sampling inspection must be performed when performing an inspection. is there.

Another object is to provide a product that can provide a clear index of how much debugging must be performed at a factory to improve quality, even if the product includes a new part. The purpose is to provide a quality improvement support system.

[0013]

In order to achieve the above object, a first aspect of the invention relating to a product quality improvement support system according to the present invention comprises a computer system and a statistical data of a product and parts related thereto. In the product quality improvement support system for improving the product quality based on the result, the system comprises: means for inputting defect data and tolerance of parts related to the product; and A means for calculating a quantity for statistically evaluating one or more of the used tolerances, a process contribution ratio calculation unit, a parameter design calculation unit, and a defect factor improvement condition teaching unit; The calculation unit compares the standard deviation obtained from each of the tolerances and an amount for statistically evaluating the entirety of one or more tolerances,
The parameter design calculation unit evaluates the correlation between the defect and the parameter, and, based on the result, improves the product quality by teaching the process and the parameter that cause the defect by the defect condition improvement condition teaching unit. It is like that.

More specifically, in the above-mentioned product quality improvement support system, the quantity for statistically evaluating the one or more tolerances is a worst value deviation obtained by summing standard deviations obtained from the tolerances, or , The total deviation obtained by taking the positive square root of the sum of the squares of the standard deviations obtained from the respective tolerances.

More specifically, in the above product quality improvement support system, for a part for which statistical data on the part is not sufficiently obtained, the standard deviation of the quantity of the part is calculated by a virtual distribution obtained by computer simulation on the part. It can be obtained by:

More specifically, in the above-mentioned product quality improvement support system, the correlation between the defect of the parameter design calculation unit and the parameter is evaluated by using a direct publication,
Alternatively, the determination is made by comparing the parameter factor contribution rates of the failure rates.

In order to achieve the above object, a second configuration of the invention according to the product quality improvement support system of the present invention uses a computer system to process statistical data of a product and its related parts, and In a product quality improvement support system for improving a product quality based on a result, the system includes means for inputting defect data and tolerance of a part related to the product, and one or more components used in a manufacturing process of the product. Means for determining a quantity that statistically evaluates the overall tolerance;
And a process capability calculation unit, wherein the process capability calculation unit multiplies the tolerance to be evaluated by a certain constant in consideration of the amount of statistically evaluating one or more tolerances as a whole from the defect rate. A process capability index obtained by dividing the process capability index by a number is compared with a predetermined standard. In the inspection process of this product, when the process capability index is larger than the predetermined standard, the sampling inspection is performed by the process capability. When the index is not larger than the predetermined standard, the product quality is improved by performing a 100% inspection.

More specifically, in the above-mentioned product quality improvement support system, the amount for statistically evaluating the entirety of the one or more tolerances is a worst value deviation obtained by summing standard deviations obtained from the respective tolerances, or , The total deviation obtained by taking the positive square root of the sum of the squares of the standard deviations obtained from the respective tolerances.

More specifically, in the above-mentioned product quality improvement support system, for a part for which statistical data on the part is not sufficiently obtained, the standard deviation of the quantity of the part is calculated by a virtual distribution obtained by computer simulation on the part. Is obtained by the following.

In order to achieve the above object, a third configuration of the invention according to the product quality improvement support system of the present invention is to process statistical data of a product and its related parts using a computer system, and In a product quality improvement support system for improving product quality based on a result, the system includes means for inputting failure data and tolerance of parts related to the product, and means for calculating a failure rate of parts related to the product. And a life and environment condition input unit, and a debug aging time setting unit, wherein the debug aging time setting unit sets a debug aging time for removing an initial failure of the product by the life and environment condition input unit. The life of the entered product or part and the operating environment of the product,
The debug aging time is pointed out in consideration of the failure rate of the component, and the debugging of the product is supported to improve the product quality.

More specifically, in the above product quality improvement support system, for a part for which statistical data on the part is not sufficiently obtained, the failure rate of the part is expressed as
This can be obtained from a virtual distribution obtained by computer simulation on the part.

[0022]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments according to the present invention will be described below with reference to FIGS.

[Configuration of Product Quality Improvement Support System of the Present Invention] First, the configuration of the product quality improvement system according to the present invention will be described with reference to FIG. FIG. 1 is a functional configuration block diagram of a product quality improvement system according to the present invention.

The product quality improvement system according to the present invention is divided into three subsystems: a failure cause indication system, an optimum inspection form indication system, and a debug aging time acquisition system.

A defect cause indicating system is a process in which a worst value deviation and a total standard deviation (described later) are calculated based on statistical data of parts, and based on the calculated values, the cause is assumed to be a cause of design or manufacturing defect. It is a system that points out and parameters, and thereby contributes to the improvement of product quality.

The optimum inspection type indicating system determines a process capability index which is calculated by comparing a tolerance and a virtual distribution, and as a result, whether or not the part must be 100% inspected as a form of inspection or not by sampling inspection. It is a system that points out whether it is good.

Further, the debug aging time acquisition system is a system for indicating an optimum time for debug aging (removing an initial failure in a factory to improve product quality) based on a product failure rate. It is.

Now, on the premise of the above, the role of each function will be described with reference to FIG.

The parts acceptance inspection system 10 is a system for inputting data of parts received from a manufacturer or the like. The component data input here is analyzed and processed by the next failure data / tolerance analysis processing unit 15.

The process equipment data collection system 20 includes:
Gather data during the manufacturing process. The data collected by the process equipment data collection system 20 is also analyzed and processed by the next failure data / tolerance analysis processing unit 15 in the same manner as described above.

The automatic line quality defect collection system 30
This is a system that collects data and tolerances related to component failures already required on the manufacturing process line. The data collected here is also the next failure data / tolerance analysis processing unit 1.
Analyzed at 5 and processed.

The debug aging data collection system collects necessary data such as a failure rate of a part for obtaining a debug aging time. Then, the data collected here is sent to the debug / aging analysis processing unit 25 and analyzed.

Now, statistical data relating to defect data and tolerances can be prepared for manufacturing and parts by an existing process, but such statistical data cannot be prepared for new processes and new parts. For this purpose, statistical processing is performed by a computer simulation technique.
It is.

The worst value deviation / total standard deviation acquisition unit 60
Based on the data analyzed by the failure data / tolerance analysis processing unit 15, the worst value of the product specification and the total standard deviation are obtained.

The defect cause indicating system which is the first subsystem is shown in FIG. 1A.

The process-tolerance-contribution-ratio calculator 70 calculates, for each process, how much variation, for example, in component dimensions, is caused by the worst value deviation and the total standard deviation. Then, the defective step teaching section 75 teaches which step is involved in the occurrence of the cause of the defect. Further, the parameter design calculation unit 80 calculates which of the parameters related to manufacturing is related to the defect. Finally, the design / manufacturing defect factor improvement condition teaching unit teaches the process and parameters estimated to be related to the defective process.

Next, the system for indicating the optimum inspection form, which is the second subsystem, is the part shown in FIG. 1B.

In this system, a process capability index, which is a value for comparing the statistical distribution of parts with the tolerance of a product, is obtained. In product inspection, the quality of the product is evaluated. Point out whether sampling inspection can be completed.

Next, the debug aging time acquisition system which is the third subsystem is shown in FIG.
Part.

This is because, based on the statistically processed data, the optimum debug aging time is pointed out based on the failure rates of both known parts and unknown parts, and the quality by debugging can be efficiently determined in the minimum time. The goal is to improve it.

First, the hardware / software configuration of the product quality improvement system according to the present invention will be described with reference to FIG. FIG. 2 is a hardware / software configuration diagram of the product quality improvement system according to the present invention.

The processing of this system is mainly performed by the computer main body 200.

Then, in the computer main body 200,
Display device 204, printing device 208, auxiliary storage device 207
The system configuration is such that peripheral devices such as are connected.

First, a description will be given of the peripheral devices. The display device 204 displays the calculation results of the system, statistical data, and the like to the user. While viewing these, the user can give an instruction or input data using the keyboard 205 or the mouse 206.

The output from the system can be printed by the printing device 208 if necessary.

The auxiliary storage device 207 stores various data and programs used in the system. In particular, when a relational database system (hereinafter, also simply referred to as “RDB system”) is used for statistical data processing, data of the database is stored. In many cases, a hard disk or a floppy disk is used as the auxiliary storage device 207. When the hard disk is used, usually, the auxiliary storage device 207 can store data having a larger capacity than the memory 202 of the computer main body.

Next, the computer main body 200 is roughly divided into a CPU (Central Processing Unit) 201, a memory 202, and an input / output control unit 203.

The CPU 201 executes a program stored in the memory 202, reads data stored in the memory, performs calculations, and gives a command to the input / output control unit 203.

The memory 202 is loaded from the auxiliary storage device 207 or temporarily stores data when the program is executed. FIG. 2 schematically shows a hierarchy of software to be executed, and a product quality improvement support system program as an application program includes a statistical library, an RDB access program, and an OS.
(Operating System).

The input / output control unit 203 is
In response to the command of 1, the peripheral device is controlled.

[Operation of Product Quality Improvement Support System of the Present Invention] Next, the operation of the system for each subsystem will be described with reference to FIG. FIG. 3 is a flowchart showing the operation of the product quality improvement support system.

(I) Failure Cause Pointing System The operation of the failure cause pointing system will now be described with reference to FIGS.
4 will be described based on a specific example. In the present embodiment, a QFP (Quad Flat Package) used for an electronic component such as a semiconductor integrated circuit will be described as an example.

FIG. 4 is a perspective view when the QFP package has a solder bridge. FIG.
FIG. 3 is a top view illustrating dimensions of a package lead and a land to which the lead is mounted. FIG. 6 is a pie chart showing a breakdown of defects in the surface mount circuit board. FIG. 7 is a bar graph illustrating the dispersion contribution ratio of each step of the solder bridge.
FIG. 8 is a conceptual diagram showing a part where detailed data of a part is input in the part acceptance inspection system 10. FIG. 9 is a conceptual diagram showing a place where detailed data on process equipment is input in the process equipment data collection system 20.
FIG. 10 is a conceptual diagram showing a state where detailed data is input in the automatic line quality defect collection system 30 and the debug aging data collection system 40. FIG.
FIG. 4 is a diagram showing a virtual distribution generated by normal random numbers by the Monte Carlo method. FIG. 12 is a diagram illustrating a screen when a virtual distribution is generated by the new process / new component / virtual distribution generation unit 50. FIG. 13 is a diagram showing an example of direct publication of L ₁₈ (the control factor is 8, the number of experiments is 18). FIG. 14 shows the result of determining the defect rate and the number of defects for each parameter factor.

The QFP is a component used for a package of a semiconductor integrated circuit or the like as described above. When soldered on a substrate, a lead portion sometimes forms a bridge by soldering as shown in FIG. . In this case, the electronic circuit cannot operate normally. In particular, as shown in FIG. 6, the solder bridge has a low frequency of failure of the surface mount circuit board.
It is bad to fight for one or two.

This subsystem analyzes statistical data and finds a process that causes such a failure.

First, various statistical techniques and concepts used in the present invention will be described.

First, the tolerance which is a central concept of the present invention will be described.

The tolerance is the difference between the maximum value and the minimum value when the object to be processed is quantitatively evaluated. Therefore, in the design stage, it can be evaluated by the following (Equation 1).

[0059]

## EQU1 ## Tolerance = upper limit of design−lower limit of design (Equation 1) This tolerance is used for a virtual distribution generation and an optimum inspection form indicating system described later.

Next, in the present invention, in the statistical processing, it is evaluated how the tolerance contributes to a quantitative value relating to a component such as a component size with a variation.

As shown in FIG. 7, the final dimensions of the component are determined by various factors. Therefore, when it is desired to evaluate the variation due to each factor, the calculation is performed on the assumption that no variation occurs at all for the other factors.

[0062] For example, the determining whether the tolerance T _i for a certain factors may contribute much relative to the size of the parts.

Assuming that the variation width of the component is Range _i , this is expressed by the following (Equation 2).

[0064]

[Number 2] _{Range i = (Ymax) i -} (Ymin) in _i ... (Equation 2) the above equation, (Ymax) _i is, and only the problem tolerance Τ _i of a certain factor, the other tolerance, all 0 Is the maximum value of the part dimensions when fixed to (Ymin) _i
This is the minimum value of the component dimensions in the same case.

Here, in the following processing, 1000
The following description is based on the case where three defects appear among the individual products.

In order to make the above assumption, for the subsequent statistical processing, the standard deviation σ _i is given by the following (Equation 3)
It is convenient to put it like

[0067]

Equation 3] _{σ i = Range i / 6 ...} ( Equation 3) Well, then, the worst value deviation used to evaluate the variation in the components of the present invention, explain the concept of overall standard deviation.

The worst value deviation σ _bad is a statistic for strictly evaluating parts that require particularly high reliability when evaluating products according to the present invention, and is defined by the following (Equation 4). Quantity.

[0069]

(Equation 4)

That is, the worst value deviation σ _bad is the sum of the respective standard deviations σ _i .

On the other hand, the total standard deviation σ _total is a quantity used for evaluating parts and products that do not require much reliability, and is expressed by the following (Equation 5).

[0072]

(Equation 5)

That is, this is the sum of squares σ of each standard deviation.
_{This is} the positive square root of _i .

The following (Equation 6) is provided between the two.
There is a relationship.

[0075]

Σ _total ≦ σ _bad (Equation 6) By using the worst value deviation σ _bad and the total standard deviation σ _total , the contribution ratio γ _i of the tolerance Τ _{i in} the entire dimension is expressed by the following equation. Expression 7)

[0076]

(Equation 7)

Now, the operation of this subsystem will be described with reference to FIG. 3 and considering the above-described statistical concept.

First, system data such as a calculation standard value and a basic design unit are read and input (S00).

Next, the part to be analyzed is determined (S
01).

There are parts to be analyzed for which known statistical data can be obtained. On the other hand, on the other hand, there is no statistical data, and there are parts for which the design tolerance is determined but statistical data is not obtained.

As a system for inputting actual measured value data corresponding to the former case, there are a parts acceptance inspection system 10, a process equipment data collection system 20, and a line quality defect automatic collection system 30 shown in FIG.

The parts acceptance inspection system 10 obtains tolerances and measured value data of the parts from a part handling company or the like, and inputs the data to the system. In this example,
As shown in FIG. 9, QFP and board data are input as components.

The process equipment data collection system 20
Process data is also an important factor in quality, so if you know that,
The user inputs specific data as shown in FIG.

The automatic line quality defect collection system 30
This is a system for inputting data on parts already handled in the factory using an RDB access program as shown in FIG.

On the other hand, since there is no existing statistical data for a new part or the like, a virtual distribution is generated and statistically processed by a computer simulation technique such as the Monte Carlo method.

For example, when normal random numbers are generated by the Monte Carlo method, the distribution of frequencies becomes a normal distribution as shown in FIG. In this example, the random number is generated 5 million times. At this time, the average value and the standard deviation σ _i ′ are
Assume beforehand before simulation. By this simulation, the frequency for each dimension shown in FIG. 11, the upper limit value and the lower limit value, and more accurate standard deviation σs _i can be obtained. Then, in the present system, this result can be displayed as a graph on the display device 204 as shown in FIG.

The distribution is not limited to the normal distribution. For example, a binomial distribution is used in the initial trial production, and a Poisson distribution is used in the distribution before mass production.
What is necessary is just to select an appropriate distribution and generate a virtual distribution.

As described above, the actually measured value data is input (S
02), statistical processing is performed (S03).

In this way, the known statistical data is processed, and the standard deviations appearing are σ ₁ , σ ₂ ,..., Σ _n (S04). Further, the standard deviation obtained by generating a virtual distribution _{(S06), σs 1, σs} 2, ..., σs m
(S05). Then, these standard deviations are registered in the system (S07).

Next, whether to use the worst value deviation σ _bad and the total standard deviation σ _total is determined depending on the reliability of the object whose quality is to be evaluated (S08).

As described above, when high reliability is required for a product, the worst value deviation σ _bad is used (S0
9). The worst value deviation σ _bad is represented by the following (Equation 8).

[0092]

[Equation 8] _{σ bad = σ 1 + σ 2} + ... + σ n + σs 1 + σs 2 + ... + σs m ... ( Equation 8) On the other hand, when it is not required so much high reliability products, a comprehensive standard deviation σ _{to tal} Used (S10). The total standard deviation σ _total is represented by the following (Equation 9).

[0093]

(Equation 9)

Next, in this subsystem, which process is the cause of the failure is determined based on the standard deviation of each process (S12). To do this, it can be obtained by the following process steps variance contribution rate gamma _i. Here, the process-dependent variance contribution ratio γ _i can be obtained by the following (Equation 10).

[0095]

[Number 10] _{^{γ i 2 = σ i 2 /}} σ 2, or, γs i 2 = σs i 2 / σ 2 ... ( Equation 10) _{where, σ = σ bad, or,} σ total γ i ≧ 0, γs i ≧ 0 From the γ _i and γs _i obtained in this way, it is possible to point out which process is highly likely to be the cause of the defect (S13). That, gamma _i, it is not likely to be the cause of the causative step bad enough that the standard deviation sigma _i, [sigma] s _i is greater gamma] s _i. For example, as shown in Table 1 below, assuming that the dispersion-by-step contribution ratio to the lead pitch is large, it is concluded that the lead manufacturing process must be reviewed.

[0096]

[Table 1]

Next, in order to determine the cause of the manufacturing defect in more detail (S14), the direct publication shown in FIG. 13 used in the experiment design method can be performed.

Here, a case regarding an automatic soldering failure will be considered. And the number of parameters (control factors) is 8
Where one parameter is of three levels and 18 experiments are to be performed. In this case, to experiment the spot published L ₁₈ shown in FIG. 13, it is possible to obtain the results shown in Figure 14. The parameters are: flux type, flux density, conveyor speed, as shown in FIG.
Solder residual heat, flux flow + solder spill height, solder temperature, flux viscosity, and solder flow rate. In this example, read from the graph, the type of flux is a ₂
In the conveyor speed at C _1, it is possible to specifically analyze the value of the parameter which is a factor of poor and so ....

Such a result can be instructed to the user by displaying it on the display device 204 shown in FIG.

(II) Optimal Inspection Type Pointing System Next, the operation of the optimal inspection type pointing system will be described based on a specific example with reference to FIGS. FIG. 15 is a diagram illustrating a display example of a result of performing a statistical analysis on QFP. FIG. 16 is a graph showing the process capability index for each distribution.

As can be seen from FIGS. 1 and 3, statistical data of parts is input and processed (S00 to S03).
In the case of a new part, the generation of the virtual distribution is the same as in the above-described failure cause indicating system (S06).

Further, the processing for the worst value deviation and the total standard deviation thereafter (S09, S10) is the same as that of the above-mentioned defect cause indicating system.

First, a process capability index, which is an important concept in this subsystem, will be described.

The process capability index Cp is represented by the following (Equation 11).

[0105]

Cp = tolerance / 6σ (Equation 11) where σ = σ _bad , or, σ _total This process capability index Cp indicates how wide the tolerance has with respect to the actual distribution. Quantity. That is, when the tolerance is large, it is rare that a component related to the quantity causes a defect. Conversely, when the tolerance is small, it is easy to cause defects. Therefore, in this subsystem, by evaluating this process capability index Cp,
This indicates whether the inspection must be performed by 100% inspection or sampling inspection, depending on the likelihood of failure.

In this example, it is assumed that the rate of occurrence of defects must be within 3 out of 1000. Then, according to the statistical theorem, Cp expressed by (Equation 9)
Is known to be compared with 8σ / 6σ ≒ 1.33.

Accordingly, in this subsystem, the process capability index Cp of (Equation 9) is calculated (S20), and compared with the constant 1.33 determined by the defect rate, if Cp is smaller, the 100% inspection is performed. (S23), Cp
If is larger, sampling inspection can be completed (S
22) Teach the user that:

After being statistically processed, each data is displayed on the display device of the system together with the distribution graph as shown in FIG. This example shows that the tolerance and the distribution of the lead width are compared. It is known that this distribution approaches a gentle normal curve as the statistical parameter increases, in the theory of probability.

FIG. 16 shows the relationship between the distribution and the process capability index. The width of the distribution (the distance where the curve intersects the x-axis) should fall within the tolerance,
When there is a portion protruding from the diagonal line in (c), it indicates that the process capability is insufficient and the process itself needs to be reviewed. The graph of (b) is just Cp
= 1.33, which is a boundary value of whether to perform a 100% inspection or a sampling inspection. If the tolerance is larger than this, C
When p becomes large and sampling inspection can be performed, conversely, when the tolerance is small, 100% inspection must be performed.

The results are summarized in a table as shown in Table 2 below.

[0111]

[Table 2]

(III) Debug Aging Time Acquisition System Next, the operation of the debug aging time acquisition system will be described.
This will be described based on a specific example with reference to FIGS. First, in order to understand this system, a general relationship between time and failure rate will be described with reference to FIG. FIG. 17 shows a bathtub curve.

Generally, the failure rate of a product or part is calculated as shown in FIG.
It is known to draw a curve as shown in FIG. First, when the product is started to be used, the quality is not stable and the number of failures increases. A failure that occurs at the beginning of use is an initial failure, and a period during which this failure occurs is referred to as an initial failure period. After that, the quality becomes stable and relatively few failures occur. The failure that occurs at this time is called a random failure in the sense that it is a random failure, and this period is called a random failure period.
After that period, deterioration of products and parts starts. The failure that occurs during the final stage of the product life is called a wear failure, and this period is called a wear failure period.

Now, when products and parts are manufactured in a factory, those that cause initial failure by inspection are as few as possible.
I want to remove it. Debugging is performed for this purpose, and the product use period required for quality improvement is the debug aging time.

The debug aging time is preferably long from the viewpoint of improving the product quality, but taking this time too long is not efficient and does not match the cost. Therefore, it is desirable that the debug aging time is set as an optimal time in consideration of a balance between improvement in product quality and cost. It is this subsystem that sets the optimal time for this.

Next, in order to understand this subsystem, the relationship between distribution and failure will be described with reference to FIG. 18 taking the lead width as an example. FIG. 18 is a diagram showing the distribution of the lead width.

Suppose that the average value of the lead width is 0.2 mm, and the distribution of the actual lead width is as shown in FIG. At this time, it is considered that the portions -3σ and 3σ at the ends far from the center are defective and cause a solder bridge. In other words, the actual distribution of the component dimensions generally draws a chevron-shaped curve with the standard value at the center, but the distribution of the dimensions at the ends is a defective product far from the standard.

Next, a method of obtaining a defect rate by generating a virtual distribution using the above concept will be described with reference to FIG. FIG.
FIG. 9 is a diagram showing a virtual distribution generated by normal random numbers by the Monte Carlo method.

As described above, one of the features of the present invention is to obtain the distribution of new parts from which unknown data cannot be obtained by the Monte Carlo method. For example, the number of trials n = 5
It is assumed that the distribution shown in FIG.
Here, σs _j in the figure is the standard deviation of the distribution related to the part dimensions obtained by generating a virtual distribution and performing statistical processing. As shown in the figure, it is assumed that a component having a dimension at a position apart from the average value by ± 3σ (hatched portion) is defective.

Here, assuming that the total number of the hatched parts is N and the number of trials is n, the defect rate q _j is expressed by the following (Equation 12).

[0121]

Equation 12] q _j = N / n ... (Equation 12) Next, step-by-step in Figure 20, the operation of the subsystem. FIG. 20 is a flowchart illustrating a processing procedure of the debug aging time acquisition system.

Based on the above-described concept, a virtual distribution is generated for a part whose data is unknown, and the defect rate is calculated (S201 to S206). This is because inputting the design value, design upper limit value, and design lower limit value of the component first is necessary for generating the virtual distribution. The description of the figure shows an example in which normal random numbers are generated by the Box-Mulano method.

From parts for which existing data is known,
The failure rate is input (S01 to S03), and for a new component or the like whose existing data is unknown, a virtual distribution based on the tolerance of the component is generated (S06), and the failure rate is calculated (S3).
0). This is similar to the two subsystems described above.

Here, the failure rate obtained from the known data is represented by λ _i (S 207) and a virtual distribution of a new part or the like using
The failure rate obtained by (Equation 10) is represented by λ _j (S20
8), the total failure rate λ _c is obtained by the following (Equation 13).

[0125]

(Equation 13)

Here, a _i and a _j are the number of parts.

When necessary, the life and environment are input (S31).

Finally, the debug aging time td is
It is determined by the following (Equation 14) (S32).

[0129]

[Number 14] _{td = 25ln (βN / 25λ c} ) (Hr) ... ( Equation 14) where, β is, initial defect rate in the initial assembly (S20
9) N is the total number of parts ln is the natural logarithm The system teaches the result obtained above to the user on the display device 204 or the like, and the user determines the debug aging time of the product by referring to the value. become.

[0130]

According to the present invention, even if a product is not skillful in inspection and does not have detailed knowledge about a product, even if a product contains a new component in a short time, In addition, it is possible to provide a product quality improvement support system that can easily know which process is the cause of the defect and which parameter contributes to the defect and how much.

Further, according to the present invention, there is provided a product quality improvement support system capable of obtaining a clear index of whether all inspections must be performed or sampling inspections must be performed when performing inspections. Can be.

Further, according to the present invention, even if a product includes a new part, a clear index can be obtained as to how much debugging must be performed at a factory for quality improvement. Product quality improvement support system can be provided.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a product quality improvement system according to the present invention.

FIG. 2 is a hardware / software configuration diagram of a product quality improvement system according to the present invention.

FIG. 3 is a flowchart showing the operation of the product quality improvement support system.

FIG. 4 is a perspective view when a QFP package has a solder bridge.

FIG. 5 is a top view illustrating the dimensions of the leads of the QFP package and the lands for mounting the leads.

FIG. 6 is a pie chart showing a breakdown of defects in the surface mount circuit board.

FIG. 7 is a bar graph illustrating the dispersion contribution ratio of each process of the solder bridge.

FIG. 8 is a conceptual diagram showing a part receiving detailed data of a part in the part acceptance inspection system 10;

FIG. 9 is a conceptual diagram showing a place where detailed data on process equipment is input in the process equipment data collection system 20.

FIG. 10 is a conceptual diagram showing a place where detailed data is input in the automatic line quality defect collection system 30 and the debug aging data collection system 40.

FIG. 11 is a diagram illustrating a virtual distribution generated by normal random numbers according to the Monte Carlo method.

FIG. 12 shows a new process new part / virtual distribution generation unit 50,
It is a figure showing the screen at the time of having produced the virtual distribution.

FIG. 13 is a diagram showing an example of direct publication of L ₁₈ (the control factor is 8, the number of experiments is 18).

FIG. 14 shows the result of determining the defect rate and the number of defects for each parameter factor.

FIG. 15 is a diagram illustrating a display example of a result of performing a statistical analysis on QFP.

FIG. 16 is a graph showing a process capability index for each distribution.

FIG. 17 is a diagram showing a bathtub curve.

FIG. 18 is a diagram showing a distribution of a lead width.

FIG. 19 is a diagram showing a virtual distribution generated by normal random numbers according to the Monte Carlo method.

FIG. 20 is a flowchart illustrating a processing procedure of the debug aging time acquisition system.

FIG. 21 is a flowchart showing a conventional product development procedure.

[Explanation of symbols]

50: New part / new process virtual distribution generation unit, 60: Worst deviation, total standard deviation acquisition unit, 70: Process tolerance contribution ratio calculation unit, 75: Teaching unit of defective process, 80: Parameter design calculation unit, 85: Design / Manufacturing failure factor improvement condition teaching section, 9
0: process capability calculation unit, 95: life and environmental condition input unit, 9
8: debug aging time setting unit, 100: product quality improvement support system.

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Akihiro Matsuda 2880 Kozu, Odawara-shi, Kanagawa Prefecture Storage Systems Division, Hitachi, Ltd.

Claims (9)

[Claims] 1. A product quality improvement support system for processing statistical data of a product and its related parts by using a computer system and improving the product quality based on the processing result. Means for inputting defect data and tolerances of related parts, means for calculating a quantity for statistically evaluating one or more tolerances used in the manufacturing process of the product, a process contribution ratio calculation unit, and parameter design A calculation unit, comprising a defect factor improvement condition teaching unit, wherein the process contribution ratio calculation unit compares a standard deviation obtained from each of the tolerances and an amount for statistically evaluating the entirety of one or more tolerances. The parameter design calculation unit evaluates the correlation between the defect and the parameter. Based on the result, the defect condition improvement condition teaching unit teaches the process and the parameter that cause the defect. Product quality improvement support system for improving product quality. 2. The amount for statistically evaluating the entirety of the one or more tolerances may be a worst-case deviation obtained by summing standard deviations obtained from the respective tolerances, or a square of a standard deviation obtained from the respective tolerances. 2. The product quality improvement support system according to claim 1, wherein the total deviation is a total deviation obtained by taking a positive square root. 3. For a part for which statistical data on the part is not sufficiently obtained, a standard deviation of an amount of the part can be obtained from a virtual distribution obtained by a computer simulation on the part. The product quality improvement support system according to claim 1 or 2. 4. The method according to claim 1, wherein the parameter design calculation unit evaluates the correlation between the failure and the parameter by comparing the number of failures or the parameter factor contribution rate of the failure rate using direct publication. The product quality improvement support system according to any one of claims 1 to 3, wherein: 5. A product quality improvement support system for processing statistical data of a product and its related parts using a computer system and improving the product quality based on the processing result. Means for inputting defect data and tolerances of related parts, means for calculating an amount for statistically evaluating one or more tolerances used in the manufacturing process of the product, and a process capability calculation unit, The process capability calculation unit obtains a process capability index obtained by dividing a tolerance to be evaluated by a number obtained by multiplying a certain constant by considering a quantity for statistically evaluating one or more tolerances in consideration of a defect rate. By comparing this with a certain standard, in the inspection process of this product, when the process capability index is larger than the certain standard, the sampling inspection is performed, and the process capability index is higher than the certain standard. When no hear, the product quality improvement support system to improve product quality by performing a total inspection. 6. The statistically evaluated amount of the one or more tolerances may be a worst-value deviation obtained by summing standard deviations obtained from the respective tolerances, or a square of a standard deviation obtained from the respective tolerances. 6. The product quality improvement support system according to claim 5, wherein the total deviation is a total deviation obtained by taking a positive square root. 7. In a part for which statistical data on the part is not sufficiently obtained, a standard deviation of an amount of the part can be obtained by a virtual distribution obtained by a computer simulation on the part. The product quality improvement support system according to any one of claims 5 and 6. 8. A product quality improvement support system for processing statistical data of a product and its related parts using a computer system and improving the product quality based on the processing result. Means for inputting failure data and tolerance of related parts, means for determining a failure rate of parts related to the product, a life and environmental condition input unit, and a debug aging time setting unit, wherein the debug aging time The setting unit takes into account the debug aging time for removing the initial failure of the product, the life and environmental condition input unit, and the life of the product or component, the operating environment of the product, and the failure rate of the component. By pointing out the debug aging time, the product debugging work is supported to improve the product quality. Quality improvement support system. 9. The method according to claim 8, wherein, for a part for which statistical data on the part is not sufficiently obtained, a failure rate of the part can be obtained by a virtual distribution obtained by computer simulation on the part. Product quality improvement support system described.