Wide Spectra of Quality Control Part 18 pdf

In other words, the deviations between the measured and estimated center distance variations should proportionally influence the accuracy of the estimated profile errors.. Therefore, th

square value of 2.75 μm Considering that Δfmea ranged between −264.59 and 5.41 μm and had a root-mean-square value of 80.85 μm, the statistically relative deviation between Δfmea and Δfest was evaluated as 3.4% [= (2.75/80.85) ×100%] Such results implied well agreement

between the measured results and the estimated ones

Fig 11 Evaluated center distance variations for the experiment

Cam angle Extreme value Cam angle Extreme value

θ = 61.7° (ΔrA,est − ΔrA,mea)max = 13.95 μm θ = 219.3° (ΔfB(l),est)min = −375.73 μm

θ = 61.7° (ΔrB,est − ΔrB,mea)max = 13.75 μm θ = 219.9° (Δf rf)min = −17.6 μm

θ = 179.9° (Δfest − Δfmea)min = −7.7 μm θ = 308.6° (Δfest − Δfmea)max = 6.91 μm

θ = 188.6° (Δf r)max = 187.17 μm θ = 308.6° (ΔrA,est − ΔrA,mea)min = −11.29 μm

θ = 215.5° (ΔrB,mea)min = −100.46 μm θ = 308.6° (ΔrB,est − ΔrB,mea)min = −12.36 μm

θ = 215.5° (ΔrB,est)min = −103.14 μm θ = 315.9° (ΔfA(u),est)max = 42.38 μm

Table 3 Extreme values of the experiment

A Convenient and Inexpensive Quality Control Method for

Fig 12 Measured and estimated results of the experiment

As shown in Figs 12(b) and 12(c), the trends and magnitudes of the estimated profile errors were well consistent with those of the measured ones The differences between the estimated and measured profile errors are once again shown in Figs 13(b) and 13(c) for

clarity of illustration The difference (ΔrA,est − ΔrA,mea) ranged between −11.29 and 13.95 μm and had a root-mean-square value of 4.68 μm Considering that ΔrA,mea ranged between

66.95 and 268.89 μm and had a root-mean-square value of 146.13 μm, the statistically relative

deviation between ΔrA,est and ΔrA,mea was evaluated as 3.2% [= (4.68/146.13) ×100%] Also,

the difference (ΔrB,est − ΔrB,mea) ranged between −12.36 and 13.75 μm and had a square value of 4.69 μm Considering that ΔrB,mea ranged between −100.46 and 185.12 μm

root-mean-Fig 13 Differences between the measured and estimated results

A Convenient and Inexpensive Quality Control Method for

and had a root-mean-square value of 109.5 μm, the statistically relative deviation between

ΔrB,est and ΔrB,mea was evaluated as 4.28% [= (4.69/109.5) ×100%] Thus, from a statistical

viewpoint, the differences and relative deviations in root-mean-square forms between the

estimated and measured profile errors were less than 5 μm or 4.3% Such results showed the

effectiveness of the presented method for the profile error examination

Center distance variations Profile errors of cam A Profile errors of cam B

(Δfmea)rms = 80.85 μm (ΔrA,mea)rms = 146.13 μm (ΔrB,mea)rms = 109.5 μm

(Δfest)rms = 81.48 μm (ΔrA,est)rms = 146.42 μm (ΔrB,est)rms = 109.8 μm

(Δfest − Δfmea)rms = 2.75 μm (ΔrA,est − ΔrA,mea)rms = 4.68 μm (ΔrB,est − ΔrB,mea)rms = 4.69 μm

Table 4 Root-mean-square values of the experiment

In Fig 13, it is found that without considering the scale, the wave of difference (Δfest − Δfmea)

was upside down to the waves of their corresponding differences (ΔrA,est − ΔrA,mea) and

(ΔrB,est − ΔrB,mea), respectively In other words, the deviations between the measured and

estimated center distance variations should proportionally influence the accuracy of the

estimated profile errors Figure 14 shows the uncertainty of the measured center distance

variations, u f, which is evaluated from the 10 data sets of the interpolated center distance

variations through using the three-standard-deviation-band approach (Beckwith et al., 2004)

with respect to each corresponding cam rotation angle The evaluated uncertainty u f ranged

between 0.43 and 3.7 μm and had a root-mean-square value of 1.97 μm The statistical

representatives of the measured center distance variations, Δfmea,SR, can be expressed as

mea,SR mea f

Thus, the upper and lower bounds of Δfmea,SR(θ), Δfmea,SR(u)(θ) and Δfmea,SR(l)(θ), are defined as

terms [Δfmea(θ) + u f (θ)] and [Δfmea(θ) − u f (θ)], respectively Considering one of the worst cases,

when data of Δfmea,SR(u)(θ), ΔrA,mea(θ) and ΔrB,mea(θ) were adopted to calculate ΔrA,est(θ) and

ΔrB,est(θ) by using Eqs (20) and (21), respectively, the evaluated difference (ΔrA,est − ΔrA,mea)

as shown in Fig 15(a) ranged between −6.8 and 17.57 μm and had a root-mean-square value

Fig 14 Uncertainty of the measured center distance variations

of 5.97 μm, and the evaluated difference (ΔrB,est − ΔrB,mea) as shown in Fig 15(b) ranged

between −7.44 and 17.32 μm and had a root-mean-square value of 5.93 μm The statistically

relative deviation between ΔrA,est and ΔrA,mea was evaluated as 4.09% [= (5.97/146.13)

×100%], and that between ΔrB,est and ΔrB,mea was evaluated as 5.42% [= (5.93/109.5) ×100%] Likewise, considering the other of the worst cases, when data of Δfmea,SR(l)(θ), ΔrA,mea(θ) and ΔrB,mea(θ) were adopted to calculate ΔrA,est(θ) and ΔrB,est(θ) by using Eqs (20) and (21), respectively, the evaluated difference (ΔrA,est − ΔrA,mea) as shown in Fig 16(a) ranged

between −15.78 and 10.32 μm and had a root-mean-square value of 5.55 μm, and the

evaluated difference (ΔrB,est − ΔrB,mea) as shown in Fig 16(b) ranged between −17.27 and

10.18 μm and had a root-mean-square value of 5.64 μm The statistically relative deviation

between ΔrA,est and ΔrA,mea was evaluated as 3.8% [= (5.55/146.13) ×100%], and that between ΔrB,est and ΔrB,mea was evaluated as 5.15% [= (5.64/109.5) ×100%] In other words, when considering the worst cases, the differences and relative deviations in root-mean-square forms between the estimated and measured profile errors were still less than 6 μm or 5.5% Therefore, the uncertainty of the measured center distance variations in this experiment should have merely slight effect on influencing the accuracy of the estimated profile errors

Fig 15 Differences between the measured and estimated results evaluated by considering the upper bounds of the statistical representatives of the measured center distance

variations

A Convenient and Inexpensive Quality Control Method for

Fig 16 Differences between the measured and estimated results evaluated by considering the lower bounds of the statistical representatives of the measured center distance variations

In addition, by applying the criteria established in Sub-section 3.1, the allowable upper and lower limits of the measured center distance variations are shown in Fig 17, and whose extreme values are also listed in Table 3 As shown in the figure, the measured values of

Δfmea exceeded their allowable upper bound, ΔfA(u),est, when θ = 80° ~ 110° but totally fell within the range of ΔfB(l),est ~ ΔfB(u),est Recall from Fig 10 that the magnitude of ΔrA,mea exceeded the specified tolerance of ±220 μm at about θ = 80° ~ 110°, while the magnitude of ΔrB,mea fell within the range of its specified tolerance Obviously, the profile error evaluating

results by using the established criteria agreed with the measuring results by using a CMM

As a result, the method presented in this study has been verified a feasible means for examining profile errors of assembled conjugate disk cams

As compared with the use of a CMM to examine profile errors of conjugate disk cams that had taken 3 hours for measuring each cam, the presented method that took 15 minutes for examining each cam through the rotation of the assembled conjugate cams for 1 revolution could provide acceptable results with efficiency Although the presented method cannot completely replace the use of CMMs, but in certain aspects it should be a more convenient and inexpensive means for the quality control in mass production of assembled conjugate disk cams

Fig 17 Allowable upper and lower limits of the measured center distance variations

of assembled conjugate cams consisting of one master cam and the other being the inspected cam, then the profile errors of the inspected cam can be estimated with the use of the analytical equations derived in this study Then, the accuracy of the inspected cam can be examined through the information of the measured center distance variations with the use

of the criteria established in this study An experiment meant to examine the profile errors

of a pair of machined conjugate cams had been conducted The machined conjugate cams had been examined by the presented method to compare with the measuring results obtained by using a CMM The experimental results showed that the estimated profile errors were well consistent with those of the measured ones by using a CMM From a statistical viewpoint, the differences and relative deviations in root-mean-square forms between the estimated and measured results of the cam profile errors were less than 6 μm

A Convenient and Inexpensive Quality Control Method for

and 5.5%, respectively, even though the machined cams had been intentionally specified to have a large tolerance grade of IT11 In conclusion, the method presented in this study has been verified a feasible and efficient alternative means for examining profile errors of assembled conjugate disk cams Therefore, the presented method could be useful for the quality control in mass production of assembled conjugate disk cams and may replace the use of expensive CMMs in certain aspects Integrating the presented method with machine

system design to develop a specialized quality control system could be possible future work

7 Acknowledgment

The authors are grateful to the National Science Council of Taiwan for supporting this research under Grant No NSC-95-2221-E-007-012-MY2 and Grant No NSC-98-2221-E-007-015-MY2

8 References

Beckwith, T.G.; Marangoni, R.D & Lienhard V, J.H (2004) Mechanical Measurements (5th

edition), pp 45-125, Pearson Education Taiwan, ISBN 986-154-022-9, Taipei, Taiwan Chang, W.T & Wu, L.I (2006) Mechanical Error Analysis of Disk Cam Mechanisms with a

Flat-Faced Follower Journal of Mechanical Science and Technology, Vol.20, No.3,

(March 2006), pp 345-357, ISSN 1738-494X

Chang, W.T.; Wu, L.I.; Fuh, K.H & Lin, C.C (2008) Inspecting Profile Errors of Conjugate

Disk Cams with Coordinate Measurement Transactions of the ASME, Journal of

Manufacturing Science and Engineering, Vol.130, No.1, (February 2008), 011009, ISSN

1087-1357

Chang, W.T & Wu, L.I (2008) A Simplified Method for Examining Profile Deviations of

Conjugate Disk Cams Transactions of the ASME, Journal of Mechanical Design,

Vol.130, No.5, (May 2008), 052601, ISSN 1050-0472

Chang, W.T.; Wu, L.I & Liu, C.H (2009) Inspecting Profile Deviations of Conjugate Disk

Cams by a Rapid Indirect Method Mechanism and Machine Theory, Vol.44, No.8,

(August 2009), pp 1580-1594, ISSN 0094-114X

Hsieh, J.F & Lin, P.D (2007) Application of Homogenous Transformation Matrix to

Measurement of Cam Profiles on Coordinate Measuring Machines International

Journal of Machine Tools and Manufacture, Vol.47, No.10, (August 2007), pp

1593-1606, ISSN 0890-6955

Koloc, Z & Václavík, M (1993) Cam Mechanisms, pp 411-413, Elsevier, ISBN 0-444-98664-2,

New York, USA

Lin, P.D & Hsieh, J.F (2000) Dimension Inspection of Spatial Cams by CNC Coordinate

Measuring Machines Transactions of the ASME, Journal of Manufacturing Science and

Engineering, Vol.122, No.1, (February 2000), pp 149-157, ISSN 1087-1357

Norton, R.L (2009) Cam Design and Manufacturing Handbook (2nd edition), pp 27-30, pp

433-440, Industrial Press, ISBN 978-0-8311-3367-2, New York, USA

Qiu, H.; Li, Y.; Cheng, K & Li, Y (2000) A Practical Evaluation Approach towards Form

Deviation for Two-Dimensional Contours Based on Coordinate Measurement Data

International Journal of Machine Tools and Manufacture, Vol.40, No.2, (January 2000),

pp 259-275, ISSN 0890-6955

Qiu, H.; Li, Y.B.; Cheng, K.; Li, Y & Wang, J (2000) A Study on an Evaluation Method for

Form Deviations of 2D Contours from Coordinate Measurement The International

Journal of Advanced Manufacturing Technology, Vol.16, No.6, (May 2000), pp 413-423,

ISSN 0268-3768

Qiu, H.; Cheng, K.; Li, Y.; Li, Y & Wang, J (2000) An Approach to Form Deviation

Evaluation for CMM Measurement of 2D Curve Contours Journal of Materials

Processing Technology, Vol.107, No.1-3, (November 2000), pp 119-126, ISSN

0924-0136

Rothbart, H.A (Ed.) (2004) Cam Design Handbook, pp 8-9, pp 44-46, McGraw-Hill, ISBN

0-07-137757-3, New York, USA

Wu, L.I (2003) Calculating Conjugate Cam Profiles by Vector Equations Proceedings of the

Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science,

Vol.217, No.10, (October 2003), pp 1117-1123, ISSN 0954-4062

Wu, L.I & Chang, W.T (2005) Analysis of Mechanical Errors in Disc Cam Mechanisms

Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, Vol.219, No.2, (February 2005), pp 209-224, ISSN 0954-4062

26

Material Characterization and

Failure Analysis for Microelectronics Assembly Processes

Chien-Yi Huang1,2, Ming-Shu Li1, Shan-Yu Huang1, Cheng-I Chang1 and Min-Hui Huang1

1Process Technology Enabling & Materials Characterization Div Operations,

Wistron Corporation, Hsinchu 300, Taiwan,

2Department of Industrial Engineering and Management, National Taipei University of Technology, Taipei 106, Taiwan,

R.O.C

1 Introduction

In recent decades, the electronic industry has shown a clear trend towards miniaturization with increasing functionality In the context of essential competition within the market, the reliability of long term operations has become a popular issue This study examines the properties of printed circuit board (PCB) and its failure phenomena PCB reliability is characterized through verifications taken from various process conditions Notably, results can be used as selection criteria for PCB materials, helping to reduce PCB delamination during the assembly process In addition, surface finish is a key factor seen to affect a product’s durability, as the microstructure between solder and the metallized layer varies between surface finish types and has been shown to affect overall solder joint strength Notably, the black pad phenomenon will reduce the strength of solder joint significantly and affect product’s durability

Characterization of failures and materials related to chemical and soldering processes used

in microelectronics assembly are also discussed in this study Analytical techniques used for chemical structures, compositions, and soldering properties including Fourier transform infrared spectrometer (FTIR), scanning electron microscopy/energy-dispersive x-ray spectroscopy (SEM/EDX), and dye staining are conducted The jumper pillow speaker connector (JPSPK) connector pins show an obvious difference in color between the clean and contaminated areas The contaminants on the connector pin were identified as the flux used

in the assembly processes following a comparison of the FTIR spectra database Additionally, the incoming plastic housings showed different bright and dark surfaces whose chemical structures were shown to be polycarbonate (PC) and acrylic ester, respectively It indicates that varied surface treatments for the incoming housings

To determine whether any cracks in the solder joints occurred in the CPU BGAs, a dye staining analysis was carried out The crack size percentage is classified according to the crack’s (dyed) area The establishment of the infrared spectra database for fluxes and

process materials helps determine the root cause of the contaminants in order to reduce the chance of a re-occurrence of similar problems thereby enhancing the manufacturing capability The infrared spectrophotometry technique can be used by professional design manufacturers and/or electronics manufacturing service (ODM and/or EMS) providers to investigate board/component defects during product pilot run stages as well as during full-volume production

2 PCB evaluation of lead free soldering process

The performance and heat resist-ability of PCB are investigated This study focuses on the inner layers of the boards, looking at critical properties such as the glass transition (Tg) temperature and curing agents Two factors, each with two levels, are used in the experimental design and are shown in Table 1 Other factors which remained constant are also shown in Table 1 Tg refers to the temperature at which material changes from glass-like to rubber-like, and where the coefficient of expansion (CTE) increases dramatically Excessive time durations for processing temperatures above the Tg may lead to cracking at the plated-through-hole (PTH) This study considers PCBs with Tg’s in the range of 110~150°C (normal Tg) and above 170°C (high Tg) The curing agent is critical to the polymerization of epoxy resins, and includes two categories: Dicy and Phenolic Based on the experimental matrix, we select PCB types that are available to the current market Four types of PCBs are used: ND (normal Tg / Dicy), NP (normal Tg / Phenolic), HD (high Tg / Dicy), and HP (high Tg / Phenolic) The material properties are shown in Table 2 All types

of PCBs meet the IPC-4101B industry standard requirements

Table 1 Experimental design

Table 2 Material properties

Material Characterization and Failure Analysis for Microelectronics Assembly Processes 511

2.1 Test vehicle

The surface finish of the PCB is organic solderability preservatives (OSP) Sample boards are composed of eight layers with a thickness of 1.6 mm and length and width of 224 mm x 114

mm, respectively There are eight samples on each test board, and each sample has a total of

1922 PTHs The PTH is 10 mils in diameter, with 40 mil spacing; the width of the annular ring is 6 mils (Figure 1) The large dimensionality combined with fine spacing simulates the worst-case scenario The thickness of copper in the PTH is in the range of 0.8 mil and 1.0 mil

to ensure that no damage occurs during the PCB fabrication process

10mil Annular Ring = 6mil

Fig 1 PTH on the test board

Fig 2 Surface of the HD samples

Fig 3 Resistance measurements for the test vehicles

The test boards are subjected to the Standard Inspection Process (SIP) which includes the incoming quality control (IQC) for inspection of trace, solder mask, labelling, board thickness, warpage, electrical resistance, and PTH dimension The HD (high Tg / Dicy, HD) sample shows blisters on the board surface (Fig 2)

The meter used for measuring the resistance is correct within 0.02% (Fig 3) The average resistances are: ND= 5.77 Ω, NP= 5.66 Ω, HD= 5.55 Ω, HP= 5.50 Ω The normal Tg (ND, NP) material shows a higher resistance compared to the high Tg material (HD, HP) Failure is defined as: the value of resistance increased by 15% Failed samples are then cross-sectioned for failure analysis Samples are then subject to two tests In the first test, heat resist-abilities of different PCB materials are evaluated The second test involves the simulation of potential environments through thermal shock testing followed by a failure analysis

2.2 Assembly process verification

The melting temperatures of lead-free solder alloys are usually higher than that of traditional tin-lead solder Therefore, the reflow and wave soldering temperatures for PCB assembly processes are also higher This may have a negative impact on the solder joint, the electronic components, and the board In this study, we verify the performances and effectiveness of various types of PCBs used in lead-free applications The heat resistibility and corresponding failures are investigated and are followed by an analysis of failure modes Sample size of 48 arranged in six panels is prepared The initial resistance

is measured when the PCB is taken from the dry package Sample boards are then processed through two reflow cycles with one wave soldering in a random sequence Resistances are again measured Samples are cross-sectioned if the resistance increases by 15% or more

2.2.1 Assembly process

The assembly process includes two reflow cycles (for two sided PCBs) and one wave soldering The reflow oven has nine temperature zones The conveyor velocity is 65 cm/min TAL (time above liquid) is 90 s while process spec is between 40 s and 90 s (Fig 4) The oxygen level is 350 ppm During the wave soldering process, the conveyor velocity is 50 cm/min The dwell time is 10 s while process spec is between 4 s and 6 s The soldering temperature is 265°C The preheat temperatures are 170°C, 190°C and 210°C (Fig 5)

A profile board is made to ensure that the desired temperature profile is achieved (Fig 6) Thermal couples are attached at locations near the conveyor edge, the center of the oven, the PCB top side, and the PCB bottom side The temperature variation across the board is within 10°C

2.2.2 Results

The crack of solder mask is observed in samples of the above mentioned four types of PCB materials A black attached substance is also observed at the edge of the samples The analysis through scanning electronic microscope (SEM) equipped with the energy dispersive spectrometer (EDS) indicates that the substance is carbide (Fig 7) The results of resistance measurement show that the variation in resistance increases as the assembly process progresses, but all within the 15% criterion

Material Characterization and Failure Analysis for Microelectronics Assembly Processes 513

Fig 4 Reflow temperature profile

Fig 5 Wave soldering profile

Fig 6 Profile board and temperature recorder