義守大學數位論文典藏

Size: px

Start display at page:

Download "義守大學數位論文典藏"

亿培温
4 years ago
Views:

1 理 Graduate School of Management I-Shou University 論立理 To Establish a Life Management Model for SMT Stencil Application to SMT Process. 六年六

2 了兩年理理論理論理論更流理論不理論論都數理論來說益良理論都理論不浪理 SPC 料論論 v

3 SMT 流了 SMT 良率 70-80% 良 SMT 參數 SMT 理 IPC 說都力理量力數行理量力力數 GRR 力量 Minitab NDC = 2 力量不來數量 GO NG 參數理參來良理參 1. 利 DOE 例度力數數聯 2. Weibull 度數立理良 3. 立 SMT 理 DOE Weibull Distribution vi

4 Abstract In the SMT(surface mounting technology) process, solder printing is the most decisive stage for the product's quality yield (around 70~80%). The good quality of solder printing is the target we persuade for SMT process, and variable parameters and printing quality are a key for quality management in SMT process. However, there is no specific specification made on solder printing base on past manufacturing experience and IPC s criteria. As to a proper application of solder paste onto the stencil printing, SMT industry passed along knowledge gained from one generation to generation without further verification and defined it as solder printing spec. We define control limit of solder printing by measuring stencil tension to meet good yield. In general, the method of stencil management is used to measure stencil tension or control frequency of solder printing. We used to tensiomentor to measure stencil tension, and the frequency of solder printing need to be controlled by manual. When the tensile measurement was calculated in Minitab via GR&R function (Software) as data analysis and the result (Number of Distinct Categories) we got was equal to 2, we found it not suitable for the MSA(Measurement System Analysis, MSA) but only a reference data - pass or fail. With the case, there is a need to create another valuable reference indicator for stencil management. We can evaluate stencil life with what it comes out to develop excellent quality and provide suggestion as well as reference for timing of stencil management and process control. There are three stages of the research on the following: 1. The Design of experiment is used to find out main effects factors, for example, thickness of stencil stencil life stencil openings frequency of solder printing and verify that there is no positive correlation between main effects factors and frequency of solder printing. 2. By Regression analysis and Weibull Distribution, we can know correlation between vii

5 the thickness of stencil and frequency of solder printing and establish the optimum management model in order to ensure good quality of solder printing. 3. Define index of stencil life and build up management model of stencil life. Key words: The Stencil life The Design of experiment Weibull Distribution The Regression analysis. viii

6 論流論度理論 Weibull 數 R SMT 料料論論來參錄 A 錄 B 錄 C 錄 D 錄 E 錄 F ix

7 F-test 說 x

8 SMC PTH 量... 力 Gage R&R (ANOVA) for 力... SMT 流年 Pareto Chart 年 Pareto Chart 年 Pareto Chart... 流力不.. 不.. 流.. 切.. 流.. 率度數不度數率數率 Normal Probability Plot Normal Probability Plot-2.. 數.... Overview Plot for STENCIL PRINT. Estimate of Weibull Reliability Model xi

9 論年來路 (Printed circuit board assembly, PCBA) 異量 (Surface Mount Technology, SMT) (Plated Through Hole, PTH) SMT 流 SMT PTH (1) 量 (2) 零度 (3) 度 (4) (5) SMT 40%~60%(Fledmann, K. and Strurn, J., 1994) 1-1 SMC PTH 量 1

10 SMT 說都量零爐零 1206( =0.12, =0.06 ) 0402( =0.04, =0.02 ) 不 Michael R. (Michael. R. Beck, 1997) SMT 60% 都 SMT 不歷鍊不參數參數 SMT 理 IPC 1 都力量理理 SMT 1 In 1999, IPC changed its name from Institute of Interconnecting and Packaging Electronic Circuits to IPC. The name is accompanied by an identity statement, Association Connecting Electronics Industries. IPC is a United States-based trade association dedicated to furthering the competitive excellence and financial success of its members worldwide, who are participants in the electronic interconnect industry. 2

11 理理度行行理精神力理理念理來立良益力量力數行理量力力數力力 1-3 3

12 力量理不力行量理力量 GR&R(Gauge Repeatability And Reproducibility) 力量 Minitab (AIAG, 1995) Gage R&R %Contribution Source VarComp (of VarComp) Total Gage R&R Repeatability Reproducibility Part-To-Part Total Variation StdDev Study Var %Study Var Source (SD) (5.15*SD) (%SV) Total Gage R&R Repeatability Reproducibility Part-To-Part Total Variation Number of Distinct Categories = 2 Gage R&R (ANOVA) for 力 Gage name: Date of study: Reported by: Tolerance: Misc: Components of Variation By 100 %Contribution %Study Var 55 Percent Gage R&R Repeat Reprod R Chart by Part-to-Part By Sample Range UCL=9.117 R=3.542 LCL= Sample Mean Xbar Chart by UCL=53.39 Mean=49.76 LCL=46.14 Average * Interaction Gage R&R (ANOVA) for 力 4

13 NDC = 2 力量不來數量 GO/NG 參數理參

14 理良 Jianbiao Pan et al. (Jianbiao Pan, Gregory L. Tonkay, Robert H. Storer, Ronald M. Sallade & David J. Leandri, 1999) Richard S. Clouthier(Richard S. Clouthier, 1997) 都 (Fine Pitch) (Ultra-Fine Pitch) 52% 71% 不論數理切數 SMT 參量數理 YES 量 NO PCB 數錄 1-5 SMT 流 PCB 若爐爐行錄不良狀良率 6

15 歷年 SMD 料年 Pareto Chart 年數量 1 P01_ % 2 I02_ ( ) % 3 L01_ 路 % 4 K01_ % 5 M01_ ( ) % 6 Q02_ % 7 N01_ % 8 N06_ % 9 N07_ % 10 O07_ 金 % 7

16 Count Percent 年 Pareto Chart 年數量 1 I02_ ( ) % 2 K01_ % 3 Q03_ % 4 M01_ ( ) % 5 L01_ 路 % 6 Q02_ % 7 N01_ % 8 N07_ % 9 N06_ % 10 O08_ 金 % 8

17 Count Percent 年 Pareto Chart ) ) 年數量 1 I02_ ( ) % 2 K01_ % 3 N07_ % 4 M01_ ( ) % 5 Q03_ % 6 L01_ 路 % 7 N06_ % 8 O02_ % 9 Q02_ % 列 2006 年類見 2004 年路異來不見 9

18 流流 1-9 度來量參數類數論不理不說 SMT 識 SMT Weibull Weibull 理 1-9 流 10

19 論論說論流 Weibull SMT 行說 Weibull 數論 11

20 度理論 Weibull 數 R 2 SMT 行列說 Bussiness Statistics ( ) 若兩數數了兩數數若了 H 0 : 1 =0 H 0 : =0 連 R 2 見 R 2 r R 2 = r 2 數數 1 r r r -1 1 ( -1 1) 說 -1 r 1 兩數量度 r 不例 x 量易 x 數例 1000 便 r 不 r 不 x y x y r 不 r 量度不來量度 12

21 1 1 數兩數數不量度 1 數 1 1 量度不來量度 r 1 r 1 說兩 0 若行 1 =0 =0 兩數兩說若兩數不兩數例不數若兩數兩數論 2 見若論 x y 不 x y 兩數兩率 1 =0 數 =0 若利若 y 數 y 13

22 度理論 Weibull Weibull 來切度林 1999 參數率行 Weibull Weibull 立度狀力狀 Weibull 度 14

23 理六劉 2005 (design of experiment, DOE) 裡利來說 KPIV(Key Process Input Variable) KPOV(Key Process Output Variable) 率行異數 DOE 數來了度 DOE 參數 2-1 KCD Mapping Fish bone FMEA 1 KPIV CE Matrix 2 益 DOE 2-1 料來劉 (2005) 流了流識量 1 FMEA : Failure Mode and Effects Analysis. 2 CE Matrix : Cause & Effect Matrix 15

24 流參數 1920 年 Rothamsted R.A. Fisher ( ) 理論 (replicates) 精異 (random) 率理度精度 (block) 量降行量論異數六行行 16

25 數 R Bussiness Statistics (Triola, Mario F. & Franklin, Leroy A., 1994) 兩數料料兩料 SSError 例料 SSError 10 SSError 20 異料若理料 SSError SSError 利數 (coefficient of determination) R 2 量料數 R 2 R 2 = SS Re gression SSy R 2 利 SSy SSRegression SSError R 2 = SS Re gression SS Re gression + SSError SSError 0 R 2 1 SSError 數離 R 2 0 說 R 2 1 良若 R 2 0 料不利數 R 2 數列 0 R 2 1 SSError SSRregression 兩來量料利 R 2 易 17

26 SMT 參數例流度率度度力 Snap-off PCB 了更精諸 2003 參數葉 2005 參數度力更度不 Aspect Ratio Area Ratio 落量 2004 SMT 量 Weibull 讀 18

27 兩理流料說不不度數數 (Surface Mount Device, SMD) (Surface Mount Technology, SMT) 良率量精度爐 profile 量度參 IPC-7525A 1 精度爐度不論度 1 IPC-7525A: Stencil Design Guideline. 19

28 3-1 力 SMT 了量力力不異異狀狀不路不良狀力狀量力數行度數理 20

29 3-2 力不料來 AIM(2000) 3-3 不料來 AIM(2000) 21

30 利 ANOVA Weibull 理利數料 Weibull 度數立理良參理若率行料六 (Montgomery, 2001) 說行論 3-4 流 22

31 2k Design 數 k 數度若識不略數 2k Franklin 1984 年兩度量不量量 2D 3D 量量狀參數狀了狀料切 4MM*12MM 行度量度落量參數度 3-5 切 23

32 料料 ANOVA 理論兩了數料 Weibull 理立數 Y 數 response variables, dependent variables 數 X 數 predictors, independent variables 若數利數料 (bivariate-data) (Correlation) (Prediction) 兩聯度料數 1989 理 1. 料數 2. 數數數參數列 1. (Method of Moment, MOM) 2. (ordinary least square estimation, OLSE) 3. (maximum likelihood estimation, MLE) 4. 率 (Probability Plot Method) 立 t-test F-test R 2 流 24

33 ( ) 料不參數立 3-6 流料來 (2004) t-test 率 ( 數 ) Y Xi F-test 3-1 F-test 說 Source SS df MS F R 2 數 SSR k SSR/K MSR/MSE SSR/SSTl SSE N-k-1 SSE/(N-k-1) SST N-1 F-test, Y x 立 25

34 數 R 2 量 Y 異 i (i=1,2,3 ) 度都 R 2 R 2 2 R adjusted = 1 (1 R 2 N 1 ) N K 1 R 2 度理論 Weibull 度率度數 (3.3.1) β f ( t) = t γ θ θ t λ exp θ β 1 β ( > 0, θ > 0, γ 0) β 狀參數率度參數 γ 參數若 γ 0 數 F() t 率數 λ() t β t γ F( t) = 1 exp (3.3.2) θ β 1 β t γ λ( t ) = (3.3.3) θ θ 26

35 <1 λ( t) DFR(Decreasing Failure Rate Distribution) =1 λ( t) CFR(Constant Failure Rate Distribution) >1 λ( t) 増 IFR(Increasing Failure Rate Distribution) 率度數不度數率數率度數數理靖 2006 率 (3.3.1) 兩兩數理 (3.3.4) 1 ln ln = β ln( γ ) β ln( θ γ ) 1 ( ) t F t (3.3.4) (3.3.4) 若 Y 令 ln t-γ X ln -γ 數 C 率 (3.3.1) 說 27

36 數料行參數參數 Model : R t Thus ( t / ) () = e ln( R() t ) = ( t / θ ) 1/ β ( ln( R() t )) = t ln( ln( R() t )) = Let Yˆ = a + bx where Yˆ = ln θ = e a / b θ β β / θ ( ln( R() t )) & β = b β lnθ + β ln t, X = ln t, a = β lnθ, b = β 說 Y ˆ = a + bx 參數 R 2 來便 R 2 28

37 料率 4-1 率不度數數度力料度 29

38 料不 Probability Stencil Average: StDev: N: 56 Anderson-Darling Normality Test A-Squared: P-Value: Normal Probability Plot-1 料不料度異行 Probability Average: StDev: N: stencil data Anderson-Darling Normality Test A-Squared: P-Value: Normal Probability Plot-2 30

39 度異數 Estimated Effects and Coefficients for stencil (coded units) Term Effect Coef SE Coef T P Constant 度力見度力不行 Estimated Effects and Coefficients for 數 (coded units) Term Effect Coef SE Coef T P Constant 力度力 * 度力 * 度 * 度力不數度不類不率異立行料數 ( ) 數 31

40 數率數 H 0 β = 0 Minitab Predictor Coef SE Coef T P Constant 數 p<0.05 數數行 The regression equation is = 數 ( ) Predictor Coef SE Coef T P Constant 數 S = R-Sq = 84.1% R-Sq(adj) = 83.8% R-Sq = 84.1% 不數異 32

41 Residuals Versus the Order of the Data (response is 度異 ) Residual Observation Order 立數 = * 數 ( ) mm 數論 SMT 理更數 CPK µ LSL C PK = σ 力 CPK µ LSL C MK = σ 33

42 C MK =1.33 異數 LSL 1.33 = * LSL = 若都量度不行力力浪 4.2 = * 數 LSL 4.4 數 = / = Weibull C PK 念 LSL 數行 Weibull MTTF Table of Percentiles Standard 95% Normal CI Distribution Percent Percentile Error Lower Upper Weibull Weibull Weibull Weibull Table of MTTF Standard 95% Normal CI Distribution Mean Error Lower Upper Weibull

43 Probability Density Function Overview Plot for 數 ML Estimates - Complete Data Percent Weibull Probability Shape Scale MTTF Failure Censor Goodness of Fit AD* Survival Function Hazard Function Probability Rate Overview Plot for STENCIL PRINT 4-7 Estimate of Weibull Reliability Model 35

44 4-1 數 R 2 Regression % Weibull % 兩數異不 R 2 Weibull Regression Weibull 理數 36

45 論論立參數理理立良不理 1. 力量力量良良 2. 理 Weibull 不立了更 R % 度度量 MTTF(Mean time to failure) 來力量落不良若行兩行力理立狀數理兩行理來論不理 37

46 參 (1989) 78 北林 (1999) 切度立 (2004) 落量理論靖 (2006) 理 Vol.13 No 1 21 (2004) 北葉 (2005) 參數理 (1999) 量理 (2003) 劉 (2005) 理六北 AIAG(1995). Measurement Systems Analysis Reference Manual(2 nd ed.). Automotive industry action group. AIM(2000). The Quick Pocket Reference For Solder Assembly.. Fledmann, K. and Strurn, J.(1994). Closed loop quality control in printed circuit assembly., IEEE Transactions on components, hybrids and, manufacturing technology Part A, Vol. 17, No. 2, Jianbiao Pan, Gregory L. Tonkay, Robert H. Storer, Ronald M. Sallade, David J. Leandri(1999). Critical variables of solder paste stencil printing for micro BGA and Pitch QFP., IEEE/CPMT Int l Electronics Manufacturing Technology 38

47 Symposium, Michael R. Beck(1997) Characterizing and controlling the smt screen printing process, General Scanning Inc., Montgomery (2001). Design and Analysis of Experiments, Washington, DC: John Wiley & Sons, Inc. Richard S. Clouthier(1997). Smt Printing Process for Fine and Ultra Fine Pitch, Canandaigua, New York : AMTX, Inc. Triola, Mario F. & Franklin, Leroy A.(1994) Business Statistics., Addison Wesley, Inc. 39

48 錄 A 錄 A 力量 GRR 錄 I 40

49 錄 B 錄 B 力量 GRR 錄 II 41

50 錄 C NO 錄 C 度力 R/C

51 錄 D 錄 D-1 錄度力錄 D-2 異錄

52 錄 E Distribution ID Plot Variable: 數 Goodness of Fit Distribution Anderson-Darling (adj) Weibull Table of Percentiles Standard 95% Normal CI Distribution Percent Percentile Error Lower Upper Weibull Weibull Weibull Weibull Table of MTTF Standard 95% Normal CI Distribution Mean Error Lower Upper Weibull

53 錄 F 錄 F Weibull Minitab 錄數數異數 3 y = x = x 類 45

1

論諸零易量零路車行不年行行年行金行年率流率行論識不易年行年行兩不兩兩行便了識易度行流識年金量更不良不便良不不行度參度度參臨數益數來行行流識率若例量度立行參行識不易料料類料論年流率益行料來度度利度度年料料