A statistical method for the electronic manufacturing industry

Abstract

The invention discloses a statistical method for the electronic manufacturing industry. The control limit of SPA (Statistical Dynamic Adjustment) is determined by the following formula: (Formula 1) UCL i = [Xbar i + ABS (Xi-Xi-1)] * + (1-) UCLi-1; (Formula 2) LCL i = [Xbar i- alpha ABS(Xi-Xi-1)]*beta+(1-beta)LCLi-1; (Formula 3) Xbar i = Xi * + Xbar i-1 (1-beta); alpha in the formula I and the formula II is a risk which means that whether the product should be a good product but is rejected is judged according to the specification limit of a client; according to the invention, the good SPA control boundary line can completely realize an on-line monitoring mode. The method does not need to wait for data collection (the formula directly shows that the improved SPA idea is not needed), so thatthe formula can be completely written into a program of test equipment for online monitoring, and the improved SPA control boundary line does not need group data to accord with normal distribution, so that the method is more convenient.

Description

technical field [0001] The invention relates to the technical field of statistical methods, in particular to a statistical method used in the electronic manufacturing industry for detecting early failures of customers in the electronic manufacturing field. Background technique [0002] I have been engaged in on-site quality management for more than ten years. I often find that many products are displayed as "good products" during testing, but they become "defective products" when they reach the client. Generally speaking, the tool used by electronics manufacturing factories to judge good products and defective products is the product specification boundary provided by customers (product specification, the electronics manufacturing factory first inputs the customer's specifications into the equipment program), usually an electronic product test item There are dozens of items at least, and hundreds of items at most. In fact, those products that have passed the factory function...

AI Technical Summary

Problems solved by technology

However, if parts fluctuate in a statistical state but the direction of fluctuation is different between parts, for example: the tolerance boundary of some parts is off the upper limit, and the common line of some parts is off the lower limit, of course, this kind of fluctuation of components will eventually affect the function of the product The fluctuation of characteristics, found in practice, makes the product show the characteristics of "outlier"

[0009] The second "outlier" phenomenon (caused by process operations, such as: solder, plug-in, assembly, etc.): In the electronic assembly process, intermittent failures often occur, and these products often pass during testing. However, the reliability of these products is usually lower than that of normal products, and handling during transportation will further reduce their reliability. Early failures of customers are often related to such products, and conventional SPC control methods cannot Effectively detect this kind of defect (functional inspection can not detect this kind of product, because they are all within the specification range), if this kind of product with intermittent defect can be tested several times, a certain test item can often be found The data has "outlier" characteristics

[0010] The third "outlier" phenomenon (caused by the test equipment): During the production process, the connection performance of the test fixture, the oxidation aging of the test plug and the relaxation of the head of the internal structure will all cause intermittent phenomena. For this phenomenon, they The test data of the product often contains hidden "outliers", which cannot be detected by conventional SPC (nor can the functional inspection detect such products, because they are all within the specification range)

[0012] However, in practical applications, most manufacturing industries use the customer’s specification boundary as a tool for judging good and bad products. Some manufacturing industries will introduce SPC, but most factories use the paper version. If it is a paper version of SPC, The efficiency of data collection is low. Practice has found that the paper-based SPC control method cannot prevent data collection errors well. In addition, this phenomenon often occurs with the paper-based SPC control method: the factory team is looking for the cause of the abnormality, and the product It has been put into storage or even shipped; a few factories will purchase some SPC software and supporting hardware facilities to monitor and test abnormal points online

Even if the SPC software package and supporting equipment are used to monitor the state performance of the test online, it cannot effectively monitor the abnormality

The reasons are: the control boundary used in traditional SPC control is calculated by the mean of the group and the standard deviation of the group. The control boundary is a straight line and cannot be adjusted according to the performance of the adjacent test.

Images