JPH06259294A - Design aid system - Google Patents

Abstract

PURPOSE:To provide the design aid system for objectively and automatically performing the selection of items to be evaluated and the decision of evaluation order in the case of evaluating a design proposal for product design. CONSTITUTION:Data obtained as an actually measured value or simulated result in the past are stored in a prototype production data base 28. On the other hand, a design knowledge data base 24 stores corresponding relation between the three elements of a defect type, evaluation item and design dimension. When evaluating the design proposal defect cause and effect relation (mutual relation among the defect type, evaluation item and design dimension) concerning the design proposal is generated by using the design knowledge data base 24. Afterwards, the strength of linkage among the elements constituting the caused and effect relation is decided by using the data in the prototype production data base 28. Then, the evaluation items are made by using.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a design support system for selecting an evaluation item for evaluating a design plan and determining the evaluation order thereof without depending on the experience and ability of each designer.

[0002]

2. Description of the Related Art Many analysis methods and analysis systems for evaluating and determining product design specifications have been published.
For example, a system for performing a flow analysis, a cooling analysis, etc. for designing a mold die is disclosed in "Mold Technology-Plastic Injection Mold Design Data Book-" Vol. 2, No. 11, Chapter 2, pages 16 to 19 ( Published by Nikkan Kogyo Shimbun (October 2, 1987)
Day 0)). The system sequentially performs a flow analysis, a heat transfer analysis, and a structural analysis according to a molding process, and confirms whether or not design specifications are satisfied by individual analysis results.

In this case, not all items of each analysis affect the product equally. Therefore, the designer himself has decided the importance of each analysis item in the product and selected evaluation items corresponding to the product. Then, the actual evaluation is performed centering on the selected evaluation item.

[0004]

However, now that the products have become complicated, the number of evaluation items is increasing, and it becomes difficult to select the evaluation items to be actually evaluated. Moreover, since the importance of each evaluation item is different, it is not efficient to simply follow the order without prioritizing. Now that the design period is being further shortened,
This problem is becoming more and more serious.

There are the following two root causes of such problems.

(1) Of many evaluation items, there is no objectivity in selecting an evaluation item that is really necessary for a design proposal.

(2) The relationship between design specifications of design proposals and defects is not clear.

Therefore, without solving these root causes, it is impossible to select an evaluation item that is really necessary for the product and to create an evaluation procedure considering its importance.

A first object of the present invention is to provide a design support system that assists in finding a mutual relationship among a defect type, an evaluation item, and design specifications.

A second object of the present invention is to provide a design support system that supports objective selection of evaluation items.

A third object of the present invention is to provide a design support system for judging the importance of each evaluation item and supporting the decision of the evaluation order.

[0012]

The present invention has been made to achieve the above object, and in one aspect thereof, input means for inputting design data of a design plan to be evaluated, and the above design data. -There is a correlation between the classification criteria for classifying design proposals using data, and the evaluation items defined for each classification that should be analyzed when evaluating design proposals and the evaluation results of the evaluation items. The design knowledge database that stores design knowledge including the correspondence with the confirmed defect type, and the classification to which the design plan input via the input means belongs. The correspondence between the evaluation items and the types of defects defined for the classification determined by the classification means is extracted by searching the classification knowledge determined by the classification criteria and the design knowledge database. Defective causality extraction And stage, actually the fabricated product in the past, at least, the design de - and data, and evaluation results of the respective evaluation items, and the incidence of each type of failure,
Prototype data that stores past data configured by associating
The database and the past data of the products belonging to the same classification as the design plan input from the input means are searched from the trial production database, and the defective data is found from the found past data. An evaluation result of the evaluation item extracted by the causal relationship extracting unit, and an incidence rate of the defect extracted by the defective causal relationship extracting unit as corresponding to the evaluation item,
By using the extracting means for extracting, the evaluation result about the evaluation items extracted by the extracting means, and the incidence of defects extracted by the extracting means, a parameter reflecting the relationship between the evaluation items and the defects is obtained. The parameter calculation means for each combination of the evaluation item and the defect, the correspondence between the evaluation item extracted by the defect causal relationship extraction means and the type of the defect, and the value of each parameter for each correspondence. A design support system is provided, which comprises: an output unit that outputs a weight determined according to

In the above parameters, in the state where the range of values that the evaluation result of the evaluation item can take is standardized among the evaluation items, the defect rate is 0 on the graph plotting the defect occurrence rate and the evaluation result. It may correspond to a wide range. The above parameters correspond to the area in the graph in which the incidence of defects and the evaluation result are plotted in a state in which the range of values that the evaluation result of the evaluation item can take is standardized between the evaluation items. Good. The parameter may be a difference between the maximum value and the minimum value of the defective rate.

As a second aspect of the present invention, a classification standard for classifying design proposals, and evaluation items defined for each classification that should be analyzed when evaluating design proposals and the evaluation items The data supporting the construction of the design knowledge database that stores the design knowledge including the correspondence with the types of defects confirmed to have a correlation with the evaluation result of
In the table construction support device, the classification standard of the design plan obtained by the designer empirically, the types of defects that are considered to occur with respect to the design plan belonging to the classification standard, and the evaluation items that are considered to cause the defect And input means for inputting a correspondence relationship (hereinafter referred to as "knowledge") that is considered to exist between them, and at least design data and evaluation items of products actually manufactured in the past. The trial production database provided with the past data in which the evaluation result and the occurrence rate of each defect type are stored in association with each other, and the past data of the products belonging to the classification criteria of the above knowledge are produced as the trial production data. Search from the database, from the data found,
The occurrence rate of defects contained in the above findings, and the evaluation results of the evaluation items that were considered to be the cause of the defects in the findings, extraction means for extracting, and the evaluation results extracted by the extraction means, Using the incidence of defects extracted by the extraction means, a verification means for testing whether there is a correlation between the defects included in the above knowledge and the evaluation items included in the above knowledge, and the above verification means. And a notifying means for notifying the verification result of 1), and a data base construction supporting device is provided.

When there is a correlation by the verification means, the contents of the knowledge are the design knowledge data according to claim 1.
It is preferable to have a storage means for storing in a table.

The test means includes a first reference value and a second reference value which are set in advance, and the range of values of the evaluation results of the evaluation items is set to a minute section having a predetermined width. By dividing, the difference between the maximum value and the minimum value of the defect rate in each minute section (hereinafter referred to as "section difference") is calculated, and the average value of the defect rate is calculated for each minute section, and the average value is calculated. The difference between the maximum value and the minimum value (hereinafter referred to as "total difference") is calculated, the section difference is compared with the first reference value, the total difference is compared with the second reference value, As a result, when the section difference is smaller than the first reference value and the total difference is smaller than the second reference value, it is determined that there is a correlation between the defect and the evaluation item. It may be one.

As a third aspect of the present invention, the design data of the design proposal to be evaluated is classified according to a predetermined classification standard, and the analysis is performed when evaluating the design proposal for each classification. The evaluation target is evaluated from the design knowledge database that stores the design knowledge that includes the correspondence between the evaluation items to be performed and the types of defects confirmed to have a correlation with the evaluation results of the evaluation items. Extract the correspondence between the evaluation items for the classification to which the design proposal belongs and the types of defects, and at least for the products actually manufactured in the past, at least their design data and the evaluation results for each evaluation item. , The past data of the products belonging to the same classification as the design plan to be evaluated is searched from the prototype database which stores the past data configured by associating the occurrence rate with each defect type. And discovered the The evaluation results of the evaluation items extracted from the design knowledge database contained in the data and the incidence of the defects extracted from the design knowledge database are retrieved. , A predetermined parameter is calculated by performing a predetermined calculation using the evaluation result extracted from the prototype database and the defect occurrence rate, and based on the parameter value. hand,
There is provided a design support method characterized by evaluating a correlation between an evaluation item extracted from the design knowledge database and a type of defect.

As a fourth aspect of the present invention, a classification standard for classifying design plans, and an evaluation item defined for each classification and to be analyzed when a design plan is evaluated and the evaluation item The data supporting the construction of the design knowledge database that stores the design knowledge including the correspondence with the types of defects confirmed to have a correlation with the evaluation result of
In the table construction support method, the classification standard of the design plan obtained by the designer empirically, and the defects that are considered to occur for the design plan belonging to the classification specified by the classification standard,
Regarding the correspondence relationship (hereinafter referred to as "knowledge") that is considered to exist between the evaluation items that are considered to be the cause of the defect, the design data of products actually manufactured in the past, and the evaluation of the evaluation items. The past data of the products that belong to the above-mentioned classification of the knowledge is searched for and found from the prototype database having the past data in which the results and the occurrence rate of each defect type are stored in association with each other. From the data obtained, the incidence of defects included in the above findings and the evaluation results of the evaluation items that were considered to be the cause of the defects in the above findings were extracted, and the extracted evaluation results And the occurrence rate of defects are used to test whether there is a correlation between the defects included in the above findings and the evaluation items included in the above findings, and a database construction is provided. Support methods are provided.

[0019]

The operation of the first and third modes will be described.

Design data of a design plan to be evaluated is input using the input means. Then, the classification means determines the classification to which the input design plan belongs, according to the classification standard in the design knowledge database. Then, the failure causal relationship extracting means extracts from the design knowledge database the correspondence relationship between the evaluation item and the type of failure, which is defined for the determined classification. Furthermore, the extraction means retrieves the past data belonging to the same classification as the design plan of the evaluation target from the prototype database, and the evaluation result of the evaluation item extracted by the defective causal relationship extraction means, The occurrence rate of the defect extracted by the defect causal relationship extracting means as having correspondence with the evaluation item is extracted.

The parameter calculation means reflects the relationship between the evaluation item and the defect by performing a predetermined calculation by using the evaluation result of the evaluation item extracted by the extraction device and the defect occurrence rate. Predetermined parameters are calculated for each combination of evaluation items and defects. Then, the output means displays the correspondence relationship between the evaluation item extracted by the failure causal relationship extraction means and the type of failure so that the weight determined according to the value of each parameter can be seen.

The operation of the second and fourth aspects of the present invention will be described.

It is considered that the designer exists between the classification standard of the design plan obtained empirically, the defect which seems to occur in the design plan belonging to the classification standard, and the evaluation item which is considered to cause the defect. Correspondence relationship (hereinafter “knowledge”)
Input) using the input means. The extraction means is a trial production of past data of products that belong to the classification criteria of the above knowledge.
From the data found by searching the database, the occurrence rate of defects included in the above findings and the evaluation results of the evaluation items that could be the cause of the defects in the above findings are shown. ,Extract.

The verification means uses the evaluation result extracted by the extraction means and the defect occurrence rate extracted by the extraction means to evaluate the defects included in the knowledge and the evaluation items included in the knowledge. The presence or absence of correlation between and is tested.

Then, the notification means notifies the verification result of the verification means. Further, the storage means uses the knowledge determined to have correlation as a result of the test as new knowledge to design knowledge data.
Store in the table.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment in which the present invention is applied to design support of an LSI package will be described with reference to the drawings.

First, a brief description will be given of the types of defects and evaluation items of the LSI package which is the object of the design support according to this embodiment.

The LSI package targeted by this embodiment is of a type in which a semiconductor chip is molded with resin. A lead frame connected to the terminals of the semiconductor chip and a wiring connecting the lead frame and an internal circuit are installed in advance in the package.

As shown in FIG. 1, the outer shape of the LSI package is generally a rectangular parallelepiped, and its size is specified by the length I2, the width w2, the package thickness x, etc. be able to. Further, the internal structure can be specified by the thickness y of the lead frame provided with the tab, the size (I1 × w1) of the semiconductor chip to be stored, and the like.
The defects of the LSI package include voids caused by air remaining as bubbles when the resin is injected. The void can be further classified into an interface void defect in which bubbles are present on the surface of the package, a through void defect existing by connecting the surface of the package to an internal chip or a lead frame, and the like. It is possible (see FIG. 2 (a)). Further, there is a gold wire deformation defect in which the gold wire connecting the lead frame and the chip terminal is deformed as the resin is injected (FIG. 2 (b)).
reference). In addition to this, there are defects such as deformation of the lead frame inside the package.

An important factor in the moldability of the LSI package is, for example, the resin filling speed when the resin is poured. The part that is problematic as the filling speed is shown in FIG.
As shown in Fig. 2, the "Insert" on the upper side of the internal structure such as the chip
"Upstream", "Insert downstream" below internal structures such as chips, and "Chipside flow" lateral to internal structures such as chips. The balance of these filling speeds is important. Can be examined by, for example, evaluating the "tip side flow / insert upstream flow tip position" which is the result of analysis of the imbalance between the chip side flow and the resin flow upstream of the insert.

The design support system 1 of this embodiment includes an input data processing unit 21, a design knowledge data processing unit 22, a design plan determination unit 23, a failure causal relationship data control unit 25, a correlation extraction unit 26, a correlation calculation unit 27, It mainly comprises a design knowledge generating unit 29, a result display processing unit 30, a design knowledge database 24, and a prototype database 28.

In order to make the explanation easy to understand, first, the details of the prototype database 28 and the design knowledge database 24, which are the basis of the design support system 1, will be described, and the input data utilizing these will be described. The data processing unit 21 and the like will be described later.

The prototype database 28 is used for designing specifications of LSI packages actually manufactured in the past.
Information indicating analysis results, defect details, etc. is stored for each prototype (or prototype lot). In the present specification, the "design specifications" refer to dimensions and manufacturing conditions that indicate the shape and size of the product specified in the design proposal.

In the column of design specifications 280, "product name" 2800, "package thickness" 2802, "indicating the type of product
Items such as chip size “2804”, “chip thickness” 2806, “lead frame plate thickness” 2808, etc. are set, and specific data is provided for each prototype (or prototype lot).
Is listed. In addition to these, the design specifications in the present embodiment are tab suspension lead width, chip top / bottom balance,
Items such as the type of resin used can be considered (see FIG. 1).

The column 282 of the analysis result includes evaluation items such as "insert upper flow tip position" 2820, "insert lower flow tip position" 2822, and "read frame deformation amount" 2824. The specific evaluation results are listed for each prototype (or prototype lot). For example, in the case of the item of "read frame deformation amount", the maximum value and the minimum value of the lead frame deformation amount of the LSI package included in the prototype lot are described. The data described in the analysis result is the measured value when the measured data is obtained, and the simulation result of the evaluation item when the measured data is difficult.

In the defect content column 284, there is "interface void".
Items are set for each defect type such as 2840 and "Gold wire deformation" 2842, and each prototype (or prototype lot)
The occurrence rate is described for each.

"Prototype No." 2801 is an identification number given to each individual prototype (or prototype lot collectively manufactured under the same conditions).

The contents of the prototype database 28 are not fixed. It is possible to make the data more effective by adding data obtained by trial production and the like and increasing the "experience". Data is added to the prototype database 28 by displaying a table as shown in FIG. 5 on the display 7 and inputting the data obtained by an experiment or the like to the table. It can be done easily.

The design knowledge database 24 is a design know-how (for example, a design know-how obtained by a designer based on past experience).
Correspondence between design specifications, evaluation items, types of defects, etc.)
It is a collection of.

FIG. 6 shows an example of the design knowledge database 24. The design knowledge database 24 classifies and stores product specifications according to design specifications and contents (values) of characteristic items (Note: these design specifications and contents of characteristic items are claimed. It is the "classification criteria" in the range.). Then, for each category, the evaluation items to be considered when evaluating the design proposal belonging to the category are shown. In the figure, the area indicated by reference numeral 240 is an area in which the classification and the evaluation items corresponding to the classification are stored. In the example of this figure, as design rule classification rule 1, package thickness ≤ 2 mm, and
It is shown that the insert up / down flow balance should be an evaluation item when evaluating a design proposal that satisfies the condition of chip area / package area ≧ 0.8. Design Proposal Classification Rule 2 shows that when evaluating design proposals that satisfy the conditions of package thickness ≤ 2 mm and chip area / package area <0.8, Not only lower flow balance, but also insert upper / side flow balance, insert
It is shown that the lower / side flow balance should also be an evaluation item.

The design knowledge database 24 stores the types of defects and the evaluation items to be examined in order to detect the defects in association with each other. In the figure, an area indicated by reference numeral 242 is a portion storing these. In this example, as the evaluation items for detecting a defect called "interface void", the lead frame deformation amount, the insert
Lower flow balance is mentioned.

Further, the design knowledge database 24 is
Each evaluation item and the design specifications that influence the contents of the evaluation item are stored in association with each other. In the figure, an area indicated by reference numeral 244 is a portion storing these. In the example of this figure, the "lead frame deformation amount" is the tip position,
It is shown that it is influenced by design specifications such as tab area and lead frame thickness.

The "characteristic item" is an item used for classification of design specifications, and these items can be calculated based on the design specifications themselves or the design specifications. The characteristic items include, for example, the volume and surface area of the LSI package, and the area ratio between the chip and the package. In the following description, the characteristic item may be referred to as a “characteristic variable”.

In the example of FIG. 6, the relationship between the classification and the evaluation items is rule-based knowledge (area 240), and the relationship between the evaluation items and defects and the relationship between the evaluation items and the design specifications are factual knowledge. Although it is stored as (areas 242, 244), the form of storing (or expressing) knowledge is not limited to this.

As described above, the design knowledge database 24 that accumulates past experiences tends to be low in reliability depending on the ability of the designer involved in the construction of the database. However, in this embodiment, in order to avoid such a situation, the knowledge acquired based on the intuition of the designer is not taken in as it is, but the prototype database 28 is used.
Is used to sufficiently test the validity of the knowledge. Then, only those satisfying the conditions of the test are accepted as new knowledge. The introduction and verification of new design knowledge will be explained in detail later.

Next, description will be given of each part which functions by utilizing the above-mentioned databases 24 and 28.

The input data processing unit 21 receives the data of the design plan (that is, design specifications, etc.) input from the keyboard 9 described later, and checks the appropriateness of the input data. Have a function. For example, if an alphabet or the like is input where a number should be input, the input is rejected. When the user's purpose is design support, the input data processing unit 21 passes the received data to the design knowledge data processing unit 22. On the other hand, when the purpose is to store new knowledge in the knowledge database 24, the input data is passed to the design knowledge generator 29.

The design knowledge data processing section 22 calculates the characteristics of the product (the LSI package in this embodiment) using the input design specifications. The "feature" referred to here corresponds to the "feature item" set in the design knowledge database 24 described above.

The design plan judging section 23 is a design knowledge database.
By referring to the design knowledge stored in the program 24, the evaluation items to be evaluated when selecting the design plan are selected. Furthermore, it has a function of extracting the types of defects that may occur depending on the content (value) of the evaluation item and design specifications that may affect the evaluation item. Finally, as shown in FIG. 7, it has a function of obtaining a mutual correspondence relationship between them (hereinafter, such a relationship is referred to as a "defective causal relationship"). In the example of this figure, the occurrence of "void" defects can be examined by analyzing evaluation items such as insert up / down flow balance, insert up / side flow balance, and lead frame deformation. Is shown.

Furthermore, the insert up / down flow balance depends on design values such as chip thickness and chip position, and the insert up / side flow balance depends on design values such as tab area and chip position. It shows that it changes with.

"Classifying means" in the claims
The "defective causal relationship extracting means" is realized mainly by the design knowledge data processing unit 22 and the design plan determining unit 23 operating in cooperation with each other.

The defect-causal relationship data control unit 25 activates the prototype database 28 to extract the information of the classification to which the design proposal belongs from the database and output it to the correlation extraction unit 26. It has a function. In addition, it also has a function of passing the characteristic variables included in the defective causal relationship passed from the design plan determination unit 23 to the correlation extraction unit 26. The "extracting means" referred to in claim 1 is mainly realized by the defective causal relationship data control unit 25.

The design knowledge generation unit 29 activates the trial production database 28 when storing new knowledge in the design knowledge database 24, and acquires the data necessary for performing verification of the new knowledge. Data is output to the correlation extraction unit 26. The extraction means referred to in claim 5 is mainly realized by the design knowledge generation unit 29. The “storing means” is also mainly realized by the design knowledge generating unit 29.

The correlation extraction unit 26 and the correlation calculation unit 27 are
When evaluating design proposals and storing new knowledge, prototype data
It has a function for analyzing data and the like read out from the database 28. The details of these functions will be described later in the description of the operation. In the claims, "parameter calculation means", "
The “verification means” is mainly realized by the correlation extracting unit 26 and the correlation calculating unit 27 operating in cooperation with each other.

The result display processing section 30 is for displaying the evaluation result of the design plan and the like. The “output means” and the “notification means” referred to in the claims are mainly realized by the result display processing unit 30.

Next, the hardware configuration of the design support system 1 will be described.

The hardware of the design support system 1 is
As shown in FIG. 2, the central processing unit 2, the multi-bus 3,
It comprises an external storage control device 4, an external storage device 5, a main storage device 6, a display 7, a bus control device 8 and a keyboard 9.

The central processing unit 2 is constructed by including electronic circuits such as a microprocessor and a memory. The central processing unit 2 executes the software stored in the internal memory or the main storage unit 6 to execute the above-mentioned input data processing unit 21 and design knowledge data processing unit 2
2, design plan determination unit 23, defect causal relationship data control unit 2
5, at least part of the functions of the correlation extraction unit 26, the correlation calculation unit 27, and the design knowledge generation unit 29 are realized. However,
These functions are not realized by the central processing unit 2 alone, but are realized by operating in close cooperation with other units.

The main memory 6 is mainly composed of a semiconductor memory, and the central processing unit 2 updates / reads the stored contents. The main storage device 6 stores various data programs read from the external storage device 5, the processing results of the central processing unit 2, and the like as needed.

The external storage device 5 is mainly for storing the design knowledge database 24 and the prototype database 28. In addition, when performing design support, the central processing unit 2
The software executed by is also stored in the external storage device 5.

The external storage control device 4 controls the operation of the external storage device 5 in response to a command from the central processing unit 2.

The display 7 realizes a part of the function of the result display processing section 30. Input data and various analysis results are displayed on the display 7.

Further, each of these parts is configured to send and receive data, control signals and the like via the multi-bus 3 controlled by the bus control device 8.

As these hardware can be used widely as they are, the detailed description thereof will be omitted.

The operation at the time of design support is shown in the flowchart of FIG.
The description will be made with reference to the above.

The package designer activates the design support system. That is, the central processing unit 2 loads the system program from the external storage device 5 to the main storage device 6 via the external storage control device 4 and the multibus 3.

Subsequently, the keyboard 1 is used to input design specifications such as the outer dimensions of the LSI package, the frame shape, the material characteristics, and the gold wire position. These data are checked by the data input processing unit 21 and then taken into the design knowledge data processing unit 22 (step 901). If the result of the check is that the input is ineligible, the input is rejected and the user is warned by an alarm sound or the like.

The design knowledge data processing unit 22 determines the characteristic variables (for example, the values of the design parameters such as the package thickness and the package structure, or the package volume and the chip area calculated from the design parameters) based on the input data. , Chip area / package volume, etc.) to calculate the design plan determination unit 23.
(Step 902).

The design plan judging section 23 calculates the value of the characteristic variable calculated by the design knowledge data processing section 22 and the contents of the area 240 of the design knowledge database 24 (that is, the range of the characteristic variable).
By comparing and, it is determined which category the design proposal belongs to. Then, regarding the evaluation items stored in association with the classification, the types of defects caused by the evaluation items and the design specifications that may affect the evaluation items are defined in the area of the design knowledge database 24. 242, 244 to obtain.
Finally, a defective causal relationship (see FIG. 7) is constructed and passed to the defective causal relationship data control unit 25. The value of the characteristic variable of the design plan is also passed to the failure-causal relationship data control unit 25 (step 903).

The defect-causal relationship data control unit 25 activates the prototype database 28 to output the data belonging to the classification to which the design proposal belongs to the correlation extraction unit 26. For example, if the design proposal to be evaluated conforms to the design proposal classification rule 1 in FIG. 6, the prototype database 2
8 is retrieved to read the data of a prototype satisfying the conditions of package thickness ≦ 2 mm and chip area / package area ≧ 0.8.

In addition to this, the defect-causal relationship data control unit 2
5 passes the characteristic variable (characteristic item) included in the defective causal relationship for the design plan to the correlation extraction unit 26 (step 904).

The correlation extraction unit 26 analyzes these pieces of information, and determines the degree of correlation between the defect included in the defect causal relationship and the analysis result of the evaluation item associated with the defect. A graph is created as a material for the determination (step 905).

For example, in the defect causal relationship, when the defect "interface void" is related to the evaluation item "insert upper / lower flow balance", the defect rate of interface void defect and the evaluation item Generate a graph with "Insert Up / Down Flow Balance" values as variables. Then, each plot point is complemented by, for example, the method of least squares to obtain a single curve (or straight line) (see FIG. 10).
The values for the insert up / down flow balance are the prototype data.
Are included in the data, insert upstream flow front distance at the tip near the rear end and (mm), the insert under flowing tip distance at the tip near the rear end of the (mm), and calculating the average of the balance (See FIG. 11)). Incidentally, FIG. 11, the horizontal axis of the insert upstream flow front distance (mm) at the tip near the rear end, placed vertically insert under flowing tip distance (mm) at the tip near the rear end, at the time of injection of the resin The state of the position movement is shown. Therefore, the slopes at the respective points forming the curves (11 to 14) correspond to the "balance" here. On the other hand, the defect rate can be used as it is in the prototype data.

Further, the correlation extraction unit 26 normalizes the graph to enable comparison for each evaluation item and outputs it to the correlation calculation unit 27. For example, the minimum value of the evaluation result values of the evaluation items is set to 0, and the maximum value is set to 1, so that the graph of FIG. 10 becomes as shown in FIG. However, the standardization method is not limited to this.

The correlation extraction unit 26 creates similar graphs for other evaluation items and outputs the graphs to the correlation calculation unit 27. However, the graph here (Fig. 10, Fig. 12)
(Or creation) means that it is created inside the system, and does not necessarily mean that it is drawn on the screen of the display 7.

The correlation calculator 27 ranks the evaluation items based on the information from the correlation extractor 26 (step 90).
6). In this example, the following three methods are provided as the ordering method (however, the ordering method is not limited to these three methods).

In the first method, the difference ΔM between the maximum value Max and the minimum value Min is parameterized in the graph of FIG.
It is a method of arranging the data according to the size of the parameter. In this case, the larger the parameter (difference ΔM), the higher the priority. When the difference ΔM is substantially equal, the evaluation item with the smaller minimum value Min is given higher priority. It can be said that the method is arranged in the order in which the defect rate that can be reduced is large, that is, in the order in which the effect of the defect countermeasure is large.

In the second method, in the graph of FIG. 13, the range of the range e where the defect rate is 0 is set as a parameter,
This is a method of ordering based on the size of the parameter (Note: the graph of FIG. 13 is a graph having the same meaning as in FIG. 12, which is standardized by the correlation extraction unit 26). . In this case, an evaluation item having a smaller parameter (that is, an evaluation item having a narrower range e) has a higher priority. It can be said that the method first evaluates the order in which it is difficult to take measures so that the evaluation result falls within the allowable range, that is, the evaluation items that have large restrictions when taking measures.

The third method is a method in which the size of the area s in the graph of FIG. 13 is set as a parameter, and the rank is set based on the size of the parameter. In this case, the larger the parameter (that is, the larger the area s is)
Increase the priority. This method can be said to be a ranking for evaluating the evaluation items that are likely to cause defects first.

The correlation calculating section 27 applies any one of the above three methods to rank the evaluation items and passes the result to the result display processing section 30. The user selects the ordering method by using the keyboard 9. The selection is performed when the design plan is input (that is, in step 901), but once the evaluation order is shown (that is, step 907 described later).
Even in the state after performing the process (1), the selection can be performed as necessary.

The result display processing unit 30 displays the evaluation items on the display 7 in the order in which they are arranged (step 90).
7). Defects and evaluation items are not necessarily associated in a one-to-one relationship, and one evaluation item may be associated with multiple defects. In such a case, the display order is determined by judging for each correspondence between each defect and the evaluation item. For example, the example of FIG. 14 is a state in which the result of ordering by the first method is displayed. Regarding the insert up / down flow balance, although the parameter value between the deformation of the gold wire and 0.1 is small, the parameter value between the interface void defect is 0.4, which is shown in this figure. Is the largest. Therefore, the insert up / down flow balance is displayed at the top as the highest priority to be evaluated. Inner / side flow balance, lead-free
Both the amount of deformation is the parameter between the interface void and
Data value is equal to 0.3. However, a parameter value of 0.1 is given for the insert upside flow balance even with deformation of the gold wire. On the contrary, the amount of deformation of the lead frame is not associated with defects other than the interfacial voids. Therefore, it is displayed on the upper side to prioritize the evaluation on the insert top / side flow balance. The "weight" in the claims corresponds to the magnitude of the values of these parameters. However, the larger the parameter value, the higher the weight. What kind of value each parameter value takes and the "weight" is made large depends on the type of parameter.

Although not shown in FIG. 14, other than this, the name of the solver that performs numerical analysis and the like when analyzing the evaluation item, the type of defect that is considered to occur, and the like are also displayed. By looking at this, the designer can know the types of evaluation items to be examined and their order. As described above, even in this state, the ordering method can be selected. Therefore, it is also possible to compare and confirm the difference in the evaluation order due to the difference in the ordering method.

It should be noted that the display mode in FIG. 14 is not limited to simply changing the display position of the evaluation item, but also the parameter.
The values of data are also displayed as numbers.

Next, the verification of whether or not the designer incorporates new knowledge into the design knowledge generating unit 29 as new knowledge based on his own experience and intuition will be described with reference to FIGS. 15 and 16.

When the designer obtains new knowledge based on his experience and intuition, the input data processing unit 21 from the keyboard 1
The new knowledge is input to (step 1401). For example, when it is found that the LSI package outer shape dimension value causes a defect in a certain range, the classification of the target LSI package, the type of the defect, and the evaluation items that are considered to cause the defect are described. input.

The input data processing section 21 checks the suitability of these input data and then passes them to the design knowledge generating section 29. Then, the design knowledge generation unit 29 receives the input LSI
Prototype data applicable to package classification is retrieved from the prototype database 28, and together with the above new findings,
This is output to the correlation extraction unit 26 (step 140).
2).

The correlation extraction unit 26 analyzes these pieces of information and tests the validity of this new finding. for that reason,
The correlation extraction unit 26 first creates a graph that is a material for determining the degree of correlation between a defect and the analysis result of the evaluation item associated with the defect (step 140).
3).

For example, in the defect causal relationship, when the defect "interface void" is related to the evaluation item "upper / lower flow balance of insert", the defect rate of interface void defect and the evaluation item Generate a graph with the values of "upper and lower flow balance of insert" as variables (Fig. 1
6). Note that the method of generating the graph is described in FIG.
The same as 1. However, the graph here (Fig. 16)
Is generated inside the system, and is not necessarily the display 7
Does not mean that it is drawn on the screen.

Subsequently, the axis of the evaluation item of the generated graph (see FIG. 16) is divided into minute sections (hereinafter, the section thus divided is referred to as "minute section Δi"). Then, for each minute section Δi, the minute section Δi
Of the maximum defective rate and the minimum defective rate, and further, the variation of the defective rate (= the difference between the maximum defective rate and the minimum defective rate in the minute section Δi), and the correlation calculation unit 27.
Output to. The "minute section" referred to here is a section having a size such that the minimum number of data required for obtaining the variation in data is included. However,
In addition to this, in determining the size of the minute section, the analysis accuracy of the numerical analysis means used in the analysis is also taken into consideration.

The reason why the processing is divided into the minute sections Δi in this way is that it is not appropriate to simply discuss the presence or absence of correlation in the entire area. That is,
This is because there are cases where there is a correlation in one area but no correlation in another area. Also, in the prototype database 28, the data for all the areas are not always available. Also, it is to confirm that the value of the defect rate is set to a substantially constant value with respect to a certain value of the evaluation item. The plurality of vertically elongated ellipses drawn in FIG. 16 are drawn in order to give an image of the variation of the data in each minute section Δi, and the processing performed here. Has nothing to do with

The correlation calculator 27 analyzes the received data of the variation of the defective rate. First, the variation of the defective rate within each minute section Δi is first set to a reference value (for example, 1
Confirm that it is less than 0% (step 140)
4). For example, when the maximum value of the defective rate of the plot points included in the minute section Δ1 is 45% and the minimum value thereof is 20%, the variation is 25% (= 45-20). . Therefore, when the reference value is set to 10%, the reference value is exceeded.

If there are many minute sections Δi in which the variation of the defective rate exceeds the reference value, it is determined that the defective rate cannot be determined for the evaluation item, and the processing is terminated (step 1407). The reference value referred to here is the "first reference value" in claim 7 of the invention.

If the above conditions are satisfied in step 1404, then the maximum value (hereinafter referred to as the "maximum average value") and the minimum value are selected from the average values of the defect rates calculated for each minute section Δi. A value (hereinafter referred to as "minimum average value") is extracted, and the difference is calculated / determined (step 14).
05). The fact that the difference is large means that the defective rate greatly changes with the change of the evaluation item (hereinafter, the difference between the maximum average value and the minimum average value is referred to as “correlation degree”). That is, the larger the value of the degree of correlation, the stronger the correlation. Therefore, if the degree of correlation is a certain value or more, it can be finally determined that there is a correlation between the two. The constant value used to determine the degree of correlation here is the "second reference value" referred to in claim 7.

The reference value (see step 1404) for judging the variation of the defective rate is set in advance according to the required accuracy and is basically not changed. To do. However, if data with higher accuracy and reproducibility is obtained due to improvements in manufacturing technology and laboratory equipment, it is possible to change this to improve the accuracy of the design support system. it can.
However, in this case, all the knowledge stored in the design knowledge database 24 must be retested using the new value. This is because the objectivity cannot be secured if the test conditions differ for each knowledge.

The value of the criterion of the degree of correlation (step 1405
In principle, the reference value is the same value as the reference value for determining the variation of the defective rate. This is because it may be considered that there is a correlation if the reference value, that is, the allowable range of variation is exceeded.

The designer inputs these values from the keyboard 9 or the like in advance and stores them in the correlation calculating section 27.

When it is determined that there is a correlation as a result of the above-described test, the design knowledge generating unit 29 displays on the display 7 that the test has passed, and at the same time, the new knowledge is acquired from the design knowledge database. 24 (step 1406). On the other hand, when a sufficient correlation is not recognized, the fact is displayed and the process is terminated (step 1).
407).

In addition to this, the knowledge in the design knowledge database 24 is appropriately confirmed by the same method in response to the increase in the trial production data. As a result, when it is determined that there is no correlation, the knowledge is deleted. This makes the prototype data
As the number of data increases, the quality (validity) of design knowledge can be improved.

In this way, the design knowledge database can be maintained using the prototype database 28.

Although the above-mentioned embodiment is directed to the design proposal of the LSI package, the present invention is not limited to this, and can be applied to the evaluation of design proposals of all other products. Is.

[0101]

According to the present invention, it is possible to objectively select the evaluation items and determine the evaluation order, which are required when evaluating the design proposal. In this case, the past experience gained by self and others can be effectively used. Therefore, it is possible to reduce the burden on the designer and improve the accuracy of design proposal evaluation. Also, the time required for this can be shortened.

[Brief description of drawings]

FIG. 1 is a diagram showing an outline of an LSI package.

FIG. 2 is a diagram showing a specific example of a defect in an LSI package.

FIG. 3 is a diagram showing an example of evaluation items of an LSI package.

FIG. 4 is a functional block diagram showing a configuration of an LSI package design support system which is an embodiment of the present invention.

FIG. 5 is a diagram showing an example of an internal structure of a prototype database 28.

FIG. 6 is a diagram showing an example of the contents of a design knowledge database 24.

FIG. 7 is a diagram showing an example of a causal relationship among defects, evaluation items, and design specifications.

FIG. 8 is a diagram showing a hardware configuration of the present embodiment.

FIG. 9 is a flowchart showing a processing procedure for evaluating a design plan.

FIG. 10 is a graph showing the correlation between defects (interfacial void defects) and evaluation items (upper / lower flow balance of insert), which is one of the materials for determining the ordering among evaluation items.

FIG. 11 is a graph of analysis results of upper / lower insert flow balance.

FIG. 12 is a graph showing a first method of ranking evaluation items.

FIG. 13 is a graph showing second and third methods of ranking evaluation items.

FIG. 14 is a diagram showing how a display 7 is displayed.

FIG. 15 is a flow chart showing a verification process when new knowledge is incorporated into the design value formula database 24.

FIG. 16 is a graph showing a method of verification processing.

[Explanation of symbols]

1 ... Keyboard, 2 ... Central processing unit, 3 ... Multibus, 4 ... External storage control device, 5 ... External storage device, 6 ... Main storage device, 7 ... Display,
8 ... Bus control device, 9 ... keyboard, 21 ...
Input data processing unit, 22 ... Design knowledge data processing unit,
23 ... Design proposal determination unit, 24 ... Design knowledge database, 25 ... Defect-causal relationship data control unit, 26 ...
... Correlation extraction unit, 27 ... Correlation calculation unit, 28 ... Prototype database, 29 ... Design knowledge generation unit, 30 ...
Result display processing unit, 240 ... Area for storing design proposal classification and corresponding evaluation items, 242 ... Area for storing evaluation items and defects corresponding thereto, 244 ...
An area that stores evaluation items and design specifications corresponding to them,
280 ... Design specification column, 2822 ... Analysis result column, 284 ... Defect content column.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Toshihiro Yasuhara 5-20-1 Kamimizumoto-cho, Kodaira-shi, Tokyo Inside the semiconductor design and development center, Hitachi, Ltd.

Claims (9)

[Claims] 1. An input means for inputting design data of a design plan to be evaluated, a classification standard for classifying the design plan using the design data, and a definition for each classification. , A design knowledge that includes a correspondence between an evaluation item to be analyzed when evaluating a design plan and a type of defect confirmed to have a correlation with the evaluation result of the evaluation item is stored. The design knowledge database, the classification means for determining the classification to which the design proposal inputted via the input means belongs according to the classification criteria, and the design knowledge database by searching the design knowledge database Defect-causal relationship extraction means for extracting the correspondence between the evaluation items and the types of defects defined for the classification determined by the classification means, and at least the design data of products actually manufactured in the past. When, An evaluation result for each evaluation item and an occurrence rate for each defect type are stored in association with a prototype data base that stores past data, and a design plan input from the input means. Search past data of products belonging to the same classification from the above prototype database,
Out of the found past data, the evaluation result of the evaluation item extracted by the defective causal relationship extracting unit and the occurrence of the defect extracted by the defective causal relationship extracting unit as corresponding to the evaluation item. The rate is used to extract the relationship between the evaluation item and the defect by using the extraction unit, the evaluation result of the evaluation item extracted by the extraction unit, and the defect occurrence rate extracted by the extraction unit. The parameters,
Parameter calculating means for each combination of the evaluation item and the defect, the correspondence between the evaluation item extracted by the defect causal relationship extracting means and the type of defect, and the value of each parameter for each correspondence. A design support system comprising: an output unit that outputs a weight determined in accordance with the above. 2. The above parameter is a defect rate on a graph plotting a defect occurrence rate and an evaluation result in a state in which the range of values that the evaluation result of the evaluation item can take is standardized among the evaluation items. The design support system according to claim 1, wherein the design support system corresponds to a range of a range in which is zero. 3. The above parameter corresponds to the area in a graph in which the failure occurrence rate and the evaluation result are plotted in a state where the range of values that the evaluation result of the evaluation item can take is standardized between the evaluation items. The design support system according to claim 1, wherein: 4. The design support system according to claim 1, wherein the parameter is a difference between the maximum value and the minimum value of the defective rate. 5. Correlation between a classification standard for classifying design plans, and evaluation items defined for each classification to be analyzed when evaluating design plans and evaluation results of the evaluation items. The design knowledge database that stores the design knowledge including the correspondence relationship with the confirmed defect type and
In a database construction support device for supporting the construction of a database, a classification standard of a design plan obtained by a designer empirically, a type of defect that is considered to occur for a design plan belonging to the classification standard, and the defect The evaluation items that are considered to be the cause of the above, the input means for inputting the correspondence relationship (hereinafter referred to as “knowledge”) that is considered to exist between them, and at least the design data of the products actually manufactured in the past. -Prototype data base with past data that stores the data, the evaluation results of the evaluation items, and the occurrence rate of each defect type in association with each other, and the products that belong to the classification criteria of the above knowledge. Of the past data from the above-mentioned prototype database, and from the found data, the occurrence rate of defects included in the above findings and what may cause the defects in the above findings Evaluation items The evaluation results of the extraction means for extracting, the evaluation results extracted by the extraction means, and the incidence of defects extracted by the extraction means are used to include the defects included in the findings and the findings. A data base construction supporting device, comprising: a verification means for verifying the presence or absence of a correlation with the evaluation items that have been previously set up, and a notification means for reporting the verification result of the verification means. 6. The design knowledge data according to claim 1, wherein the contents of the knowledge are correlated when the verification means determines that there is a correlation.
6. The database construction support device according to claim 5, further comprising a storage means for storing in a database. 7. The test means comprises a first reference value and a second reference value which are determined in advance, and the range of values of the evaluation results of the evaluation items is within a predetermined width. It is divided into sections, the difference between the maximum value and the minimum value of the defect rate in each minute section (hereinafter referred to as "section difference") is calculated, and the average value of the defect rate is calculated for each minute section. The difference between the maximum value and the minimum value of the average value (hereinafter referred to as "total difference") is obtained, the section difference is compared with the first reference value, and the total difference is compared with the second reference value. Then, as a result, when the section difference is smaller than the first reference value and the total difference is smaller than the second reference value, it is determined that there is a correlation between the defect and the evaluation item. A database construction support according to claim 5, characterized in that it is a judgment. Location. 8. The design data of the design proposal to be evaluated is classified according to a predetermined classification standard, and the evaluation items and the evaluation items to be analyzed when the design proposal is evaluated for each classification. From the design knowledge database that stores the design knowledge that includes the correspondence relationship with the types of defects that are confirmed to have correlation with the evaluation result, the classification to which the design proposal to be evaluated belongs Correspondence between the evaluation items and the types of defects is extracted, and at least the design data of products actually manufactured in the past, the evaluation results for each evaluation item, and the occurrence rate for each defect type , And the past data of the product belonging to the same classification as the design plan to be evaluated is searched from the prototype database that stores the past data configured by The above included in And the results of the evaluation of the evaluation item that has been extracted from the scan, - a total of knowledge de - eat
The occurrence rate of the defect extracted from the design knowledge database is taken out, and a predetermined calculation is performed using the evaluation result and the defect occurrence rate extracted from the prototype database. By calculating a predetermined parameter by performing
A design support method comprising: evaluating a correlation between an evaluation item extracted from the design knowledge database and a defect type based on a data value. 9. Correlation between a classification standard for classifying design proposals, and evaluation items defined for each classification that should be analyzed when evaluating design proposals and evaluation results of the evaluation items. The design knowledge database that stores the design knowledge including the correspondence relationship with the confirmed defect type and
In the database construction support method for supporting the construction of the database, the classification criteria of the design plan obtained by the designer empirically and the defects considered to occur with respect to the design plan belonging to the classification specified by the classification standard are identified. , The correspondence that is considered to exist between the evaluation items that are considered to be the cause of the defect (hereinafter “
"Knowledge"), the past data that stores the design data of products actually manufactured in the past, the evaluation results of the evaluation items, and the occurrence rate of each defect type in association with each other. From the prototype data base equipped with the above, the past data of the products belonging to the above-mentioned classification of the knowledge is searched, and from the found data,
The incidence of defects included in the above findings and the evaluation results of the evaluation items that were considered to be the cause of the defects in the findings were extracted, and the extracted evaluation results and the incidence of defects were used. A method for supporting the construction of a database, wherein the existence of a correlation between the defect included in the above knowledge and the evaluation item included in the above knowledge is tested.