• Save
Final report reflow process control
Upcoming SlideShare
Loading in...5
×
 

Final report reflow process control

on

  • 805 views

The basic purpose of the Process Control Project in the Reflow Oven Operation is to determine which are the main variables of the reflow profile that have significant influence on the output (defect ...

The basic purpose of the Process Control Project in the Reflow Oven Operation is to determine which are the main variables of the reflow profile that have significant influence on the output (defect rate, failures, quality of the solder joints, etc.) and determine and implement a process control for them.

Another objective of this project is to find some results or general highlights that can be used as a reference for other EMS factories to have a better process control for reflow.

Finally, this project will help to Engineering team in EMS Monterrey to understand better why process control is required to improve the quality of our products.

Statistics

Views

Total Views
805
Views on SlideShare
805
Embed Views
0

Actions

Likes
1
Downloads
0
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Final report reflow process control Final report reflow process control Document Transcript

  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:1 of 34Table of Contents:1. Objective..................................................................................................................................................22. Introduction............................................................................................................................................23. Project Basic Information......................................................................................................................34. Scope (Defined with EMS Corporate).................................................................................................34.1. Agreements (Defined with corporate) .......................................................................................35. Test vehicle selected ..............................................................................................................................46. Status of the process control at the beginning of the project...........................................................56.1. Controls at the beginning of the project ....................................................................................56.2. Quality levels .................................................................................................................................66.2.1. Quality rate from SMT...........................................................................................................66.2.2. Criteria from IPC-A-610 C Acceptability for Electronic Assemblies ..............................86.2.3. Criteria Standards for Inspection in EMS Monterrey .......................................................86.2.4. Profile used by Production Line...........................................................................................97. Process Control Project Flow..............................................................................................................117.1. Machine settings..........................................................................................................................117.2. Process Flow ................................................................................................................................127.3. Tests Description.........................................................................................................................127.3.1. Experimental Set up of Reflow...........................................................................................127.3.2. Validation of First DOE .......................................................................................................177.3.3. Second DOE...........................................................................................................................177.3.4. Validation of Second DOE ..................................................................................................217.3.5. Conclusion of profile recommended.................................................................................217.3.6. Operation window for each variable.................................................................................228. Other variables of SMT affecting solder beading............................................................................238.1.1. What solder beading is ........................................................................................................238.1.2. How solder beading happens.............................................................................................248.1.3. Why Solder Beading is problematic ..................................................................................248.1.4. Some possible causes of solder beading............................................................................258.1.5. Possible Solutions for solder beading................................................................................259. Control Process Options .....................................................................................................................269.1. Introduction to Process Control Charts ...................................................................................269.2. Control Chart Selection for Reflow Variables.........................................................................309.3. Final variables selected to be monitored .................................................................................329.4. Profiling Frequency: ...................................................................................................................329.5. Monitoring Equipment Options ...............................................................................................329.5.1. Real Time profile monitoring device (Short description)...............................................3310. Project Conclusions ...........................................................................................................................33
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:2 of 341. ObjectiveThe basic purpose of the Process Control Project in the Reflow Oven Operation is todetermine which are the main variables of the reflow profile that have significantinfluence on the output (defect rate, failures, quality of the solder joints, etc.) anddetermine and implement a process control for them.Another objective of this project is to find some results or general highlights that canbe used as a reference for other EMS factories to have a better process control forreflow.Finally, this project will help to Engineering team in EMS Monterrey to understandbetter why process control is required to improve the quality of our products.2. IntroductionIn the Industry’s ongoing drive to reach higher levels of quality in surface mountassembly, electronics manufacturers have focused on Process Control. Establishingand maintaining performance limits for each assembly process has been viewed ascritical to achieving output that is consistent, require minimal rework or has few or,ideally zero defects.The process control is vital to obtain a continuous quality performance in ourproducts; the control process of the most important variables must be implementedto assure this performance. Eventually, the Statistical Process Control has taken a bigincreasing in all the companies due to the control that can be approached in the mainselected process variables. It is important to determine and select all the variablesthat must be monitored in the normal process to achieve a continuous qualityperformance. Those variables must be working with the best parameters determinedby tests. DOE’s and statistical analysis should be used to improve analysis andparameters selection.In the SMA process steps, machines are used and they have several processparameters to properly be adjusted. The process parameters are subject to complexinteractions of different factors, determined not by the particular type of process butalso by the design of the products to be processed.Parameters can be categorized to three groups,1) Machine Settings: for example conveyor speed.2) Machine Parameters: conveyor speed (for example) as a measured value.3) Process Parameters: top temperature of the joint during soldering.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:3 of 34To meet the quality challenge, equipment suppliers have designed machines thatperform with greater accuracy and repeatability. Software has been developed todefine and monitor a tight process window, ensuring that output remains withinstated limits. About the Process Control in the Reflow Oven, several kind of softwarecan be used to help on this.3. Project Basic InformationDefine which parameters have more effect on the output (defects), DOE’s and otherstatistical analysis shall be developed to success with the project.We are going to use an existing product (the most problematic on reflow process) toperform the DOE; we will purchase extra material in case scrap is generated.These are the outputs to be considered in the project: testing result, quantity ofdefects in SMT (cold solder, dull joints, poor coalescence, flux residue, dark residues,component cracking, solder balls tombstoning, etc.). Parameters to be evaluated willbe: Ramp-up rate, soaking time, soaking temperature, reflow time, peak temperature,peak time & ramp-down rate.4. Scope (Defined with EMS Corporate)Develop required tests, analysis & DOE’s to:1. Find the most significant input variables.2. Define the process window (lower & upper limits) for each variable.3. Know what is the effect on the output when the parameters move within theprocess window.4. Measure the improvement on the quality (defect rate) once the parameters arecontrolled.5. Reduction of defects due reflow process based on a proposed profile.4.1. Agreements (Defined with corporate)Next comments were discussed with corporate, and was decide to include them in theprojet:• Make the tests with only one product and only one solder paste type. This wouldreduce the DOE matrix size and cost of the project.• Do not use the microstructure analysis to evaluate the process. This does not give arepresentative output due to low samples quantity. Besides the high cost of thetests.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:4 of 34• At the end of the project, be able to give recommendations of the frequency forovens profiling.5. Test vehicle selectedThe test vehicle selected was a wireless product from NMP basically for threereasons:a) This product platform is similar than other products assembled in differentElcoteq plants, so there is more chance to use some of the conclusions fromthis project at least as a reference. Besides, the paste used for this product isthe approved by EMS for Lead Free applications (LF300 from LoctiteHenkel).b) This was one of the products with higher DPU rate.c) The cost of the product would allow us to perform more tests (buying extramaterials) based on the budget assigned by Corporate.The following figures represent the test vehicle chosen for this project:
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:5 of 346. Status of the process control at the beginning of the project6.1. Controls at the beginning of the projectThis control plan explains what kind of controls are used for this product in Reflowprocess:For the first control described in the document above, the visual inspection is madeby using a magnification lens after reflow, 100 % of the product is inspected by thismethod but one problem is that the operator just inspect the areas of the panel thatare not covered by shield cans (Figure 1) and just 10% of the products are inspectedcompletely using the same method (magnification lens).For the second control described, an X-Ray inspection is done for 1 panel every 2hours and a SPC control chart for attribute is used.SizeFrequencyReflowOven Omniflo7 51SolderJointQA014(workmanship)VisualInspection100%eachpanelDefect Report Format(typeof defect,reference, DPUandYieldper hour arecollected)Stopthelineif 3equaldefects or 5differentdefects arefoundwithinanhour52SolderJoint onBGAQA014(workmanship)X-Raymachine1panelevery2hoursAttributeControl chartStopthelineif 3equaldefects or 5differentdefects arefoundwithinanhour110Set UpVerificationFormat withOvenparametersOnceEachshiftCheck List andworkinstructionVerify/ Adjust theovenparameters andsendtorepair thenonconformanceproduct,previuslyidentifiedControl MethodMethodsReactionPlanProduct / Process /Specification/ToleranceEvaluation/MeasurementTechniqueSampleProduct ProcessCharacteristicsSpecialCharClassProcessName/OperationDescriptionMachine,Device, Jig,Tools formanufacturingNo.Figure 1
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:6 of 34The third control described is just a verification of the oven parameters, it consistsjust in looking in the monitor of the oven and see if the machine parameters are thecorrect ones.There are some controls out of the documented Control Plan, which showopportunities, and we will recommend include. As an example: process techniciansperform weekly profiles and only Peak Temperature & Reflow Time are evaluated,and there is no history of them.If Peak Temperature & Reflow Time are within specs, the profile is saved, but thereare no monitoring and tendency analysis.The current controls for reflow process are basically based on attribute data by usingSPC charts or just a simple table of defects in a documented format. The rate orquantity of defects are taken as the only process control.The current control method is good to measure the quality of the product and also ishelpful to compare the level of quality with different products, but it is not adequateto measure the ability or behavior of the machine and the process itself. Only outputsor Y’s are measured instead of the inputs or X’s. The measurement of the inputsprocess variables or X’s will lead us to have more revealing information andprevention of the problems or defects.6.2. Quality levels6.2.1. Quality rate from SMTThe quality levels obtained from the production line had not been validated with anAttribute gage R & R for inspectors. Another difficulty is that the areas of the panelsthat are underneath the shields are not inspected 100 % but 10% and the onesinspected are qualified just with a magnification lens (10x), so small solder beadingor solder balling are not easy to detect. However, data from SMT and also fromdiagnostic was taken as a reference for the project as the main issues to solve.As we didn’t find significant data from production line, whether SMT data ordiagnostic data, we inspected a bunch of panels form different lines and differentshifts taking off the shield cans and by using not only magnification glass but alsomagnifiers, we saw a lot of solder beading present on the panels so we increased thesample for inspection without shields and with the same method.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:7 of 34Below is the DPU and some pictures of the defect or process indicator:DPU = Quantity of defects / Units producedDPU = 20Component Size:2.8 mm x 2.2mmComponent Size:1.52 mm x 0.76mm
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:8 of 346.2.2. Criteria from IPC-A-610 C Acceptability for Electronic Assemblies6.2.3. Criteria Standards for Inspection in Elcoteq MonterreyThere is a workmanship standard based on IPC-A-610C used for the inspectors as areference to determine if there are defects on the product.This point can generatecontroversy because howcan we be sure that thesolder ball is not going tobecome dislodged duringnormal service.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:9 of 346.2.4. Profile used by Production LineA linear profile within Abcde specifications has been used so far. A normal productboard with three of four thermocouples attached on different locations along theboard is utilized to set up and optimize the settings until acceptable profile isachieved, once the profile fulfill the requirements and several measurements aredone for validation, a calibration board is used to monitor the profile with a certainfrequency (once a week). Below there are two tables with Abcde specifications, thefirst one is for the product board and the second one is for calibration board.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:10 of 34A profile obtained with the calibration board for Terminator product and themeasured variables are shown below:Calibration board is a squareboard made with FR4, onethermocouple is connected toa screw and the second oneis just on the air
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:11 of 347. Process Control Project Flow7.1. Machine settingsAs a reference, this is the list of the machine settings that affect directly the processconditions or process parameters:• Conveyor speed• Temperature settings of the zones• Convection (Low, Med and High)• Cooling (Low, Med and High)• Separation between consecutives PCB’s going into the oven can affect thevariables in some cases.• Temperature setting of the railsExhaust is not a machine setting but it can influence some of the process parameters.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:12 of 347.2. Process Flow7.3. Tests Description7.3.1. Experimental Set up of ReflowA first DOE was developed to examine several important factors of the reflowprocess, in order to determine which variables had significant effect on the defects(mainly solder beading). The variables examined in the DOE were soak time, reflowtime, peak temperature and cooling slope. The variation on each factors are shown inthe Table 1. All the variations are within the solder paste specifications and customerspecifications except the soaking time which the maximum value in the DOE islonger than the one required by the customer (according to the gradientrecommended from 70°C to 180°C, the soaking would be from 35 to 43 sec).PROCESSINPUT OUTPUTREFLOWPROCESSINPUT OUTPUT- Preheat Slope- Soaking time- SoakingTemperature- PeakTemperature- Peak Time- Reflow Time- Cooling slope- Yield- DPU- Defects- Testing failures- Visual Inspection- X-Ray- Voids- Tombstoning- Solder beading- Poor coalescence- Flux charring- Reliability of thesolder joints- Cold joints- Component cracks- ExcessiveIntermetallics
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:13 of 34For this DOE, the printing and placement processes were held constant.The Design of Experiment was defined as follows:• Full Factorial experiment• 4 Factors• 2 Levels• 2 replicates• 2 central pointsOutput of the DOE was Defect quantity or defect rate per panel.The four variables used in the study were set up into a DOE that required 34 runs intotal because of the 2 replicates and the 2 central points. Metrics of visual inspectionunder magnification lens and microscope were used, X-Ray was used for BGA’s todetect any problem (mainly voiding).For the results of the DOE, a 95 % or higher confidence interval was required for afactor to be considered statistically significant.After inspect the 34 panels and test them through functional station, the only defectnoticed was again solder beading. The quantity of solder beading per panel was 8.5.Below you can see the statistical analysis of the DOE, soaking time was the onlystatistically significant factor for solder beading according to the P-value obtained (P-value lower than 0.05).Factor Low HighSoak Time38 - 42seconds98 - 102secondsTime Above Liquidous30 - 32seconds58 - 60 secondsPeak Temperature 235 - 237 C 248 - 250 CCooling Slope 2.4 - 2.6 C/sec 3.4 - 3.6 C/secTable 1 - Variables included in DOE
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:14 of 34Factorial Fit: Defectos SMT versus Peak temp, Ramp Down, ...Estimated Effects and Coefficients for Defectos SMT (coded units)Term Effect Coef SE Coef T PConstant 14.344 0.7276 19.71 0.000Peak temp -0.187 -0.094 0.7276 -0.13 0.899Ramp Down -0.938 -0.469 0.7276 -0.64 0.528Reflow Time 2.937 1.469 0.7276 2.02 0.060Soak Time 4.937 2.469 0.7276 3.39 0.003Peak temp*Ramp Down -1.312 -0.656 0.7276 -0.90 0.380Peak temp*Reflow Time -0.188 -0.094 0.7276 -0.13 0.899Peak temp*Soak Time -0.687 -0.344 0.7276 -0.47 0.643Ramp Down*Reflow Time 2.562 1.281 0.7276 1.76 0.096Ramp Down*Soak Time 0.562 0.281 0.7276 0.39 0.704Reflow Time*Soak Time 1.187 0.594 0.7276 0.82 0.426Peak temp*Ramp Down*Reflow Time 0.437 0.219 0.7276 0.30 0.767Peak temp*Ramp Down*Soak Time -0.562 -0.281 0.7276 -0.39 0.704Peak temp*Reflow Time*Soak Time -1.187 -0.594 0.7276 -0.82 0.426Ramp Down*Reflow Time*Soak Time 2.562 1.281 0.7276 1.76 0.096Peak temp*Ramp Down*Reflow Time* -1.812 -0.906 0.7276 -1.25 0.230Soak TimeCt Pt -5.844 3.0000 -1.95 0.068Analysis of Variance for Defectos SMT (coded units)Source DF Seq SS Adj SS Adj MS F PMain Effects 4 271.375 271.375 67.84 4.00 0.0182-Way Interactions 6 84.188 84.187 14.03 0.83 0.5643-Way Interactions 4 67.875 67.875 16.97 1.00 0.4344-Way Interactions 1 26.281 26.281 26.28 1.55 0.230Curvature 1 64.281 64.281 64.28 3.79 0.068Residual Error 17 288.000 288.000 16.94Pure Error 17 288.000 288.000 16.94Total 33 802.000TermSta nda rdiz e d Effe ctAA CA BCBDA BDA DBC DA C DA BA B C DB C DBCCD3.53.02.52.01.51.00.50.02.110F a ctorS oak T im eN a m eA P e ak te m pB R a m p D ow nC R e flow T im eDP are to Chart of the S tandar dize d Effe cts(response is Defe ctos S M T , A lpha = .05)This chart shows that soaking time has significant effect onthe solder beading
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:15 of 34Standardized EffectPercent43210-1-2-3999590807060504030201051F actorS oak T im eN ameA P eak tem pB Ramp D ow nC Reflow T imeDEffect Ty peNot SignificantSignificantDNormal Probability Plot of the Standardized Effects(response is Defectos SMT, Alpha = .05)This chart shows that only soaking time has significant effecton the solder beading
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:16 of 34The Figure Below represents the combination of the factors and their results inquantity of defects:An optimization of the first DOE was performed and these are the results:The values in red are the ones recommended to minimize the quantity of solderbeading. The quantity of solder beading with this set up is 9. More replicates per setup would be recommended but one big obstacle to do so was that the budget waslimited.1004560303.52.5250235S oak TimeReflow T imeRamp D ow nP eak temp8.518.517.016.024.012.018.017.012.013.513.513.510.58.513.010.512.0CenterpointFactorial PointCube Plot (data means) for Defectos SMT
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:17 of 347.3.2. Validation of First DOEA validation run was run with 15 panels and the quantity of solder beading per panelincreased a little bit (11 solder beading per panel), again just solder beading wasdetected as an issue.7.3.3. Second DOEA second DOE was developed in order to reduce the quantity of solder beading perpanel, we know that there are more possible causes and not only the reflow profilethat can lead to have solder beading but they are going to be mentioned in a separatepoint of this report. The variables examined in this DOE were soaking time andpreheat slope. Peak Temperature, Reflow Time and Cooling slope were held constantwith central points values from first DOE. The variation of the factors for the secondDOE is shown in Table 2. The values of the other reflow variables are shown inTable 3.Best Set Up form DOE: 9 defectsper panelValidation: 11 defects per panelFactor Low HighSoak Time38 - 42seconds78 - 82secondsPreheat Slope(50-140 C)0.7- 0.9C/sec1.1 - 1.3C/secTable 2 - Variables included in second DOE
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:18 of 34Like in the first DOE, the printing and placement processes were held constant.The Design of Experiment was defined as follows:• Full Factorial Experiment• 2 Factors• 2 Levels• 5 ReplicatesOutput of the DOE was Defect quantity or defect rate per panel.Table 4 shows the series of experiments performed in the DOE.StdOrder RunOrderSoaking Time(°C)Preheat slope(°C/sec)19 1 40 1.218 2 80 0.816 3 80 1.214 4 80 0.81 5 40 0.88 6 80 1.24 7 80 1.210 8 80 0.820 9 80 1.217 10 40 0.815 11 40 1.27 12 40 1.23 13 40 1.29 14 40 0.85 15 40 0.812 16 80 1.211 17 40 1.213 18 40 0.86 19 80 0.82 20 80 0.8Table 4 Set Up of Second DOEFactor LowTime Above Liquidus45secondsPeak Temperature 243 CCooling Ramp Down -3.0 C/secTable 3 - Constant Factors for second DOE
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:19 of 34Metrics of visual inspection under magnification lens and microscope were used, X-Ray was used for BGA’s to detect any problem (mainly voiding).For the results of the DOE, a 95 % or higher confidence interval was required for afactor to be considered statistically significant.After inspect the 34 panels and test them through functional station, the only defectnoticed was again solder beading. The quantity of solder beading per panel was 8.5.Below you can see the statistical analysis of the DOE, soaking time and the 2-wayinteraction between soaking time and preheat slope were statistically significant forsolder beading according to the P-value obtained (P-value lower than 0.05).Factorial Fit: Quantity of defects versus Soaking Time, Preheat slopeEstimated Effects and Coefficients for Quantity of defects (coded units)Term Effect Coef SE Coef T PConstant 11.3000 0.2398 47.12 0.000Soaking Time 5.6000 2.8000 0.2398 11.68 0.000Preheat slope 0.8000 0.4000 0.2398 1.67 0.115Soaking Time*Preheat slope -1.4000 -0.7000 0.2398 -2.92 0.010Analysis of Variance for Quantity of defects (coded units)Source DF Seq SS Adj SS Adj MS F PMain Effects 2 160.000 160.000 80.000 69.57 0.0002-Way Interactions 1 9.800 9.800 9.800 8.52 0.010Residual Error 16 18.400 18.400 1.150Pure Error 16 18.400 18.400 1.150Total 19 188.200TermStandardized EffectBABA1210864202.12F actor N ameA S oaking TimeB P reheat slopePareto Chart of the Standardized Effects(response is Quantity of defects, Alpha = .05)This chart shows that soakingtime and 2-Way interactionbetween factors havesignificant effect on the solder
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:20 of 34An optimization of the first DOE was performed and these are the results:Standardized EffectPercent12.510.07.55.02.50.0-2.5-5.0999590807060504030201051Factor NameA Soaking TimeB Preheat slopeEffect TypeNot SignificantSignificantABANormal Probability Plot of the Standardized Effects(response is Quantity of defects, Alpha = .05)This chart shows that soakingtime and 2-Way interactionbetween the 2 factors havesignificant effect on the solder1.20.88040Preheat slopeSoaking Time13.814.47.49.6Cube Plot (data means) for Quantity of defectsCombination offactors withlower quantityof defects
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:21 of 34An optimization of the first DOE was performed and these are the results:The values in red are the ones recommended to minimize the quantity of solderbeading. The quantity of solder beading with this set up is 7.7.3.4. Validation of Second DOEA validation run was run with 15 panels and the quantity of solder beading per panelwas 9. Again, just solder beading was detected as an issue.7.3.5. Conclusion of profile recommendedAfter the two DOE’s we concluded that soaking time has the more significant effecton solder beading (on this specific product and with this solder paste). Table 5 showsthe recommended values for the profile to reduce the quantity of solder beading.Factor LowTime Above Liquidus 45 secondsPeak Temperature 243 CCooling Ramp Down -3.0 C/secSoaking Time 40 secondsPreheat Slope(50-140 C)0.8 C/secTable 5 - Parameters Recommended after twoDOEs
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:22 of 347.3.6. Operation window for each variableNext table (Table 6) contains the window recommended for each of the processvariables of the reflow profile selected in this project, this parameters arerecommended to have the best results for solder beading and good quality in generalfor the solder joints and assembly. The chart illustration shows the phases of theprofile.0501001502002503000 90 180 270 360Time (sec)Temperature(ºC)Prehat Phase Soaking Time Reflow Phase Cooling PhasePeak TempLiquidus Temp = 217ºC0501001502002503000 90 180 270 360Time (sec)Temperature(ºC)Prehat Phase Soaking Time Reflow Phase Cooling PhasePeak TempLiquidus Temp = 217ºCProfile Feature Large Body Small BodyPreheat Ramp up Rate(From 70 to 140 ºC)Soaking Time(From 130 to 165 ºC)Time Above Liquidus (217 ºC)Peak TemperatureCooling Ramp Down -2.7 to -3.3 ºC/sec0.8 - 0.9 ºC/sec37 - 43 sec42 - 48 seconds240 - 246 ºCTable 6 - Recommended process window for each variableReflow ProfileThe Scale in X & Y axisand the curve are justexamples for a typicalprofile, the curve is notthe necessarily therecommended
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:23 of 34The values of the variables recommended come from the two DOE’s performed andthe validation of them, the values are almost in the center of customer specifications(except the preheat ramp up rate which is in the minimum specification).The operational window recommended for each variable might seem very tight, butis the one that gave better results. The specifications from the customer can be usedas well as an operational window but the quantity of solder beading expected wouldbe higher, other than that the quality of the assembly was acceptable (visualinspection by certified inspector, X-Ray Inspection and testing) according to thepieces that we assembled for the trials in DOE’s.There are some other actions that can lead us to eliminate completely the solderbeading, they are going to be explained in the point 8.Table 7 shows again customer specifications for the variables mentioned in this point(7.3.6)8. Other variables of SMT affecting solder beadingAs we couldn’t avoid completely the problem of the solder beading with the reflowprofile, we analyzed the problem taking in account other possible causes in SMT.8.1.1. What solder beading isProfile Feature Large Body Small BodyPreheat Ramp up Rate(From 70 to 140 ºC)Soaking Time(From 130 to 165 ºC)Time Above Liquidus (217 ºC)Peak TemperatureCooling Ramp Down0.8 - 1.0 ºC/sec35 - 43 sec35 - 60 seconds232 - 250 ºC-2 to -5 ºC/secTable 7 - Customer Specifications
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:24 of 34Solder beading is a special phenomenon of solder balling when using solder paste incertain SMT applications. In brief, solder beads are large solder balls nearcomponents with very low stand-off.Often confused with solder balling, solder beading is a defect recognized by one or afew larger balls, generally located around chip caps and resistors.8.1.2. How solder beading happensa) Solder paste is printed on the pads of a circuit board.b) During component placement some solder is squeezed underneath the body of thecomponent and broken off from the solder on the pads.c) During reflow, the solder trapped underneath the component does not flow backto the solder pads. Contrarily, its cohesive properties (surface tension) cause it toform a large ball (bead).d) The surface tension of the cooling solder draws the component closer to the pads.As the body of the component is drawn down, the solder bead squeezes out the sideand remains there.8.1.3. Why Solder Beading is problematicBasically, solder beads may form a “bridge” of solder that runs from one componenttermination to another, thus causing an electrical connection that was not designed tobe there. This poses the threat of resulting in a short circuit. This may occur where thebead was originally formed or elsewhere on the assembly if vibration causes the beadto break loose and move around. While the above may not necessarily occur if thesolder beads are present, solder beading obviously remains a defect that should beminimized or eliminated if possible.Solder beading could create an unwanted electrical bridge if they are located betweentwo adjacent parts of the PCB. Additionally, they could be dislodged duringhandling and affect the performance of adjoining assemblies or components.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:25 of 34Therefore, solder beading that are formed need to be detected visually and removedmanually. These steps add to the labor costs associated with the assembly process. Ifa solder micro-ball eludes detection and adversely affects the performance in service,or results in customer returns, then the economic impact can be even greater. Theobvious answer to this problem is to eliminate solder micro-ball formation, whichcan be difficult, because solder micro-balls can be caused by a variety of sources.8.1.4. Some possible causes of solder beadingThere are many causes that can lead to have solder beading but we listed the onesthat we thought are more important in this specific case:- Bad Stencil Aperture 1:1 with the pad- High value of stencil thickness- Incorrect Snap Off: A gap between the stencil and PWB can increase thepaste volume deposited on the assembly- Solder Paste tend to generate solder balling (this feature can be testedfollowing IPC test .- Too slow or too fast preheat ramp rate on Reflow- Incorrect speed and squeegee pressure on printing process- High humidity in the environment.- Oxidation of solder powder.- Incompatibility between board finish (OSP) and solder paste flux.8.1.5. Possible Solutions for solder beadingSome recommendations to prevent solder beading are listed below:- Stencil Aperture Size / Shape - Reduce the aperture size (home plate, bowtie or D shape can be options).- Stencil Thickness – Reduce the stencil thickness to reduce the paste deposit.- Snap Off – Use on contact or zero snap off (top side of PWB on contact withbottom side of stencil).- Solder paste viscosity – Keep the viscosity of the solder paste withinspecifications, the higher viscosity, the better results for solder beading.- Reflow Profile – Improve the preheat ramp rate (too slow or too fast ramprate can lead to solder beading). This optimization was already made in theproject.Solder beading problem appeared mostly in components 0402, 0603 and some midsize components. As with the current stencils the aperture is 1:1 with the pad, werecommended some changes in the aperture for the most problematic components.One stencil was ordered and the changes on the stencil were applied on the morerecurrent references of the product. The results were pretty good, we assembled 30
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:26 of 34panels with this stencil and we didn’t find any solder beading on the referenceswhere the changes on the aperture were applied. Table 6 shows therecommendations of the apertures:Next step is to change the stencil apertures for all 0402 & 0603 references (aperturesfor problematic mid size components were already implemented) on the board.This change is on going because we are waiting for approval from customer.9. Control Process OptionsThis section is intended to determine which controls can be proposed to perform theProcess Control of the selected reflow variables.9.1. Introduction to Process Control ChartsNext section is related to the theoretical structure of the control of the processvariables using Control Chart.Ongoing monitoring is typically managed with a control chart.Uses of control Charts• Determine appropriate managerial action in response to the value of a data point from aparticular processCOMPONENT APERTURE SHAPE DRAWING0402 D-Shape0603 Home PlateMID SIZECOMPONENTReduction in innerpart of thecomponentTable 6 - Recommended aapertures to reduce solder beadingproblemYellow means Landpattern areaArea in White meansrecommendedstencil apertureThe squares in bluerepresent just otherreferences
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:27 of 34- To see if high or low points are due to special causes.• Understand and predict process capability (expected ranges of future values) forplanning purposes.• Identify Root causes of variations by differentiating between special and commoncauses of variation in the data.• See whether intentional changes in the process had a desired result.• Monitor key processes and identify shifts or changes quickly to help hold the gainsmade from an improvement project.Common-Cause VariationCommon causes are the process inputs and conditions that contribute to the regular,everyday variation in a process.• Common causes are a part of the process.• They contribute to output variation because they themselves vary.• Each common cause contributes a small part of the total variation.• By looking at a process over time, we know how much variation to expect fromcommon causes.• The process is stable, or predictable, when all the variations is due to commoncauses.Special-Cause VariationSpecial causes are factors that are not always present in a process but that appearbecause of some particular circumstance.• Special causes are not usually present.• They may come and go sporadically; may be temporary or long-term.• A special cause is something special or specific that has a pronounced effect on theprocess.• We can’t predict when a special cause will occur or how it will affect the process.• The process is unstable, or unpredictable, when special causes contribute to the variation.Tests for special causes• 8 or more points in a row of the same side of the median indicates a processshift.• If the data are symmetrical, it’s Ok to use the average as the central lineinstead of the median.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:28 of 34• 6 or more points in a row continuously increasing or decreasing indicates atrend.• Start counting at the point where the direction changes.• Too few runs indicates a shift in the process average, a cycle, or a trend.• Too many runs indicates sampling from two sources, overcompensation, or abias.• 14 or more points in a row alternating up and down indicates bias or samplingproblems.• One or more points outside the control limits indicates that something isdifferent about those points.Individual ChartsBecause they can be used with any data that is time-ordered, and in general and veryversatile, individual charts are the most frequently used type of control charts.However with particular kinds of data or situations, they are sometimes slower tosignal special causes than other kind of charts, so it’s best to understand other typesof control charts as well.SPECIFICATION LIMITS Vs CONTROL LIMITSSpecification Limits• Come from engineering or customer requirements.• Represent what someone wants a process to do.• Can sometimes be changed by changing the requirements of the product or service.Control Limits• Come from calculations of the process data.• Represent what a process is actually capable of doing• Can only be changed by changing the process.When to calculate new control limits?You should calculate new control limits when:• You know there was a change in the process based on- Statistical Evidence, such as 8 continuous points above or below the centerline.- You have determined why the change occurred (based on your processknowledge).
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:29 of 34- You are confident the process will stay changed• You are confident the process will stay changed.- The change was not temporary.- The change has become a standard part of the process.Calculate the new limits when you have enough data points to see a change. Call thenew limits temporary until you get at least 24 new data points.Assumptions for individual charts• Data are roughly normal (data might need to be transformed)• Data points are independentWhat to look for when using Control Charts• A good control chart is one that is being used concurrently with the process.- Charts should be posted or be readily at hand.- Charts should be up-to-date.- Charts should look well-used.• Comments should be written on charts- Dates of process changes.- Notes on events that might cause problems later.- Confirmation of verified special causes.- Actions taken to eliminate special causes (only rarely should the chart indicatethat the cause could not be identified).Common mistakes when using control charts- Chart not created correctly- Wrong formula used to calculate “3 sigma” limits (st. dev. used instead of movingranges).- Wrong type of charts used based on type of date collected.- Missing, poor or erroneous measurements.• Chart not regularly updated
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:30 of 34- Data on charts are not current- Process adjustments have not been noted.- Control limits and average have not been updated.• Actions taken are- Rewards given for “good points” or explanations sought for “bad points” eventhough they are not signaled as special.- Special-cause signals ignored.- Non-random patterns or cycles not studied to determine specific causes.- Spec limits or goals are placed on chart instead of control limits9.2. Control Chart Selection for Reflow VariablesAs a part of the project is required to propose an option to measure, monitor and controlthe final variables selected on all the previous analysis. The process to select theappropriate graph was carried based on next flow chart.Selecting an Appropriate Control ChartCan you mistakeproof the defect?Do not use SPC;implement PMOutput or input? Improvement orrun/stop?Continuous data? Pre-controlTracking SystemIndividual and MovingRange ChartsProcess inControl?Past as importantas present?EWMAchartCuSumchartIs subgroup size >8 and sigma easilyCan you measuremore than one defectper unit?Xbar andsigmachartX bar andRangechart% defective(bad parts)Defect/unitSample sizeconstant?Sample sizeconstant?nPchartPchartCchartUchartInputNoYesAutomatic ManualYesNo YesNo YesNo Yes No YesNo YesRun/StopImprovementOutputYesNo
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:31 of 34This is a sample of the “Individual-X & Moving Range Chart” proposed to monitorand control the selected variables in the Reflow Oven profile.E:INDIVIDUAL-X &MOVING RANGES CONTROL CHART.xlsIndividual-X & Moving Range Charts are a set of control charts for variables data(data that is both quantitative and continuous in measurement, such as a measureddimension or time). The Individual-X chart monitors the process location over time,based on the current subgroup, containing a single observation. The Moving Rangechart monitors the variation between consecutive subgroups over time.The Individual chart on top shows each reading and is used to analyze CentralLocation. The Moving Range chart at bottom is used to study system variability.Why it is used?To analyze a subgroup size that cannot be more than one for any reason such ascosts, expense involved, difficulty to collect data or large size collection is notpractical or possible.When it is used?1. When you need to assess the stability of system.2. When the data is variable.3. When the size of subgroups is one.4. When the time order of subgroups preserved.AdvantagesThe charts are sensitive and hence can detect even a small variation in normality.Highly reliable and easy to use.DisadvantagesConstant and thorough study of chart is necessarySlow in detecting sudden jumps in average values
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:32 of 349.3. Final variables selected to be monitoredBased on results, soak time was the only significant variable obtained.Every time the profile is evaluated, the Data pack can be programmed and all datacan be obtained and graphed. Even though only soaking time was the main variableobtained in all results, there is no restriction to increase the control of all the list ofvariables mentioned:• Soak Time• Peak Temp• Ramp Down• Reflow Time• Cooling slope9.4. Profiling Frequency:Actually there is a process instruction which states that the profile must be evaluatedeach week. This can give enough data of how the oven is working. Gauge R&R to thereflow oven shows that repeatability and reproducibility are good, due to this, it isnot recommendable to decrease the profiling period (i.e. from one week, to twoweeks or one month), until we obtain enough data in a period of at least 6 months.After that, we would be able to decrease frequency.9.5. Monitoring Equipment OptionsData logger Real Time Reflow ProfilerGold M.O.L.E Oven watchData PackSlim KIC 2000 KIC 247The proposed graph in section 9.2, is based on the use of the “Gold M.O.L.E.” or“Data Pack” data loggers, used at the moment in the Monterrey Plant. A “Real TimeReflow Profiler” device is recommended as a better option to control the variables.This is an option online, which can give the profile for each PCB and result on abetter monitoring.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:33 of 349.5.1. Real Time profile monitoring device (Short description)The automated real-time thermal process monitoring system, is an equipment thatprovides and records real-time thermal process data for every product, as opposed tothe conventional practice of only periodically checking oven performance. Thisallows this equipment to automatically catch potential defects before they happen,rather than discovering actual defects during “Inspection”. It utilizes thermocouplesensors permanently embedded in the oven at the process level. The key feature isthe creation and recording of a Virtual Profile. By associating the dynamics withinthe oven with the profiled data collected using profile data logger, the real-timemonitoring device, automatically calculates a unique product profile including allpertinent process data for every product that exits the oven.It utilizes real-time process data for real-time SPC charting. Process data isautomatically charted for all critical process specs: peak temperature, soak time, timeabove liquidus, etc. The data is plotted on real-time control charts and ProcessCapability (Cpk) is calculated for each spec. Any the process drift outside of controllimits will bring an immediate alarm. The process engineer also has the option ofsetting a warning limit on the Cpk. Real-time Cpk tracking enables the system to flagan out of control process before the oven has produced a single defect.10. Project Conclusions• Control process is required to assure the continuous quality performance in ourproduct, a Control Chart is proposed to control main variables on the oven profile.Eventually, changing to 100% monitoring of the profile or “real time” isrecommended.• Individual values and Moving Range (I-MR) charts are recommended as SPC forsoaking time, preheat ramp up, reflow time, peak temperature and cooling slope.• The use of control chart applies not only for the variables mentioned above but alsofor all significant variables at any kind of process. The more variables monitored,the better process control.• Solder Beading defect can be categorized as a process indicator or defect based onIPC criteria. This was found as an opportunity as an output of the DOE’s. Severalrecommendations were done related to this.• Requesting for theoretical information, solder beading is mainly caused due tosoaking time, same as our results, this is a validation of the results.• Historically, the profiling period has been 1 profile per week. Give a frequency forprofiling depends pretty much on the repeatability of the machine (oven); again, a100 % or real time monitoring of the profile is more recommended but Budget forinvestment has to be analyzed.
  • MTDREFLOW PROCESS CONTROLDocument code:NOT IN USEClassification:ConfidentialDate:14-Feb-2005Prepared by:(Also subject responsible if other)AlejandroRodriguezReviewed by:(Also subject responsible if other)Guadalupe VillarrealApproved by:(Also subject responsible if other)Version:0.0Page/pages:34 of 34Having a 100% or real time monitoring of the profile is recommended but notmandatory to have a good process control, we can have a good process control bydetermining the right frequency of profiling and analyzing the data properly.We have seen that the repeatability among ovens varies; Analyze how repeatableyour oven is and how small your process window is will lead you to determine theright frequency for profiling. Of course SPC and good analysis of it is going to behelpful.• During DOE’s tests, it was shown that profile specification limits are correct, basedon visual inspection (IPC certified operator), X-ray inspection, and electrical-testingresults.• The final proposed profile is basically in the center of all the customer specificationlimits. This is giving us a confirmation that we are running under the best profile,considering only few changes at the beginning of the profile.