FMEA SPC英文教学PPT.ppt

fmea & (s)pc training module,this training module is based upon and should be studied together with the aiag reference manuals. change authority for this document: quality manager bg mds fons sieben,chapter contents 1 acknowledgement / references 2 potential failure mode and effect analysis 3 process considerations & the need for process control 4 stages in spc and process analysis (qmaps) 5 control charts & out of control conditions 6 from fmea thru control planning to spc - process documentation 7 terminology and abbreviations: crit / sign / key / normal 8 flow diagram statistical process control 9 statistical methods / calculation of control limits 10 capability indices and formulas 11 cpk interpretation and relation to ppm,2000-10-31,updated,2000-07-11 uzw-b0/h100-13,this training module is related to: fmea & (s)pc procedure uzw-b0/h100-13,acknowledgement / references,potential failure mode & effect analysis,an fmea is a systemized group of activities, intended to: recognize and evaluate the potential failure of a product/process and its effects. identify actions which could eliminate or reduce the chance of the potential failure occurring. document the process. it is complementary to the design process of defining positively what a design must do to satisfy the customer,ref: fmea reference manual (aiag),2000-07-11,brainstorming,potential failure mode & effect analysis,ref: fmea reference manual (aiag),2000-07-11,brain storming team effort risk analysis tool preventive / on time = before the event!,potential failure mode & effect analysis: process overview,ref: fmea reference manual (aiag),2001-03-23,describe: function of parts or process steps potential failure mode potential effects of failure,describe and plan recommended actions predict rpn,close the loop / continuous improvement work on next problem in the priority row learn from problem solving / previous designs reflect the learnings in documentation: fmea / control plans / design rules / learning modules etc,working formats design fmea (preferred format),ref: fmea reference manual (aiag) uzw-b0/h100-13,2000-12-29,working formats process fmea (preferred format),ref: fmea reference manual (aiag) uzw-b0/h100-13,2000-12-29,d/pfmea : some practical implications,ref: fmea reference manual (aiag) uzw-b0/h100-13,2001-03-23,steps wrt improvement actions: establish initial fmea, and select the items for actions. identify the recommended actions, and at the same time estimate the predicted rpn (register at the right side under actions results). note: when the recommended actions are - for the time being - an investigation, this new estimated rpn is of course identical to the original, pending the result of the investigation. when actions have been completed, the resulting rpn is again to be established. this new situation becomes now the new current controls, and is to be moved towards these related columns.,living documents & learning: above implies that the fmea list reflects only the best current knowledge, and history is not documented. where it is considered to be necessary to keep the history for learning reasons, a front page should be considered to keep record of the important historical events.,working formats fmea (old format / see previous pages for preferred formats),ref: fmea reference manual (aiag) uzw-b0/h100-13,2001-03-23,old format: do not use anymore!,working formats control plan,ref: fmea reference manual (aiag) uzw-b0/h100-13,2000-12-29,form: process step control plan (preferred version) (for product specific control plans: see proc test specs: uzw-b0/h100-39),working formats control plan,ref: fmea reference manual (aiag) uzw-b0/h100-13,2001-03-23,form: process step control plan: (old format / see previous pages for preferred version),old format: do not use anymore!,considerations around process,processes are important!,almost all quality gurus and almost all management gurus agree that : processes must be defined, the most important processes must be identified and given special attention, all processes must be managed, all processes must be improved.,of course, it may be easier to manage departments or functions.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,considerations around process,deming on processes,william edwards deming suggested that stakeholders will benefit if managers: focus on processes, understand variability, develop a learning culture, value people.,see “the new economics” (1993) w e deming, harvard u p,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,considerations around process,what is a process?,efqm definition: a process is a sequence of steps which adds value by producing required outputs from a variety of inputs.,a better definition: a process is a network of interdependent components that work together to achieve benefits for the stakeholders.,wherever work is being done by one or more people there is a process.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,considerations around process,a useful process model,product,equipment people procedures,materials,management,service,information,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,considerations around process,a very useful process model,product,equipment people procedures,materials,management,service,information,measure,measure,name: xxx yyy aim: to zzzz,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,every process is part of a larger process,a business organisation,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,processes within processes,any process can be broken down into many sub-processes. any process can be regarded as part of a larger process.,to get world class performance from a process you need world class behaviour from your workers.,managers must set up the conditions so that workers can behave in a world class way.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,process improvement,we can improve a process by:, improving the performance of the equipment, improving the procedures, improving the behaviour of the workers, improving the flow of information, improving the behaviour of the manager., improving the quality of the inputs,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,process redesign,when redesigning a process we should consider: the quality of the inputs the capability of the equipment the ability of the workers the procedures information needs the ability of the manager.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,performance measurement,you need to measure those things which will help you to:,your metrics should: be measurable, be unbeatable, encourage the right behaviour., assess the current performance, detect changes in performance, find the causes of changes, improve performance, benchmark your performance.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,what should we measure?,what metrics will best help us to: control the process, compare the process, improve the process?,would it be better to measure: yield, first pass yield, or dpmo (defects per million opportunities) ?,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,performance improvement,to improve process performance we might strive to: increase certain performance measurements eg. yield, price, sales volume, strength, decrease certain performance measurements. eg. cycle time, impurity, cost, ovality, keep the average of the performance measurements on target. eg. delivery time, viscosity, diameter, reduce the variability of the performance measurements.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,why are processes difficult to manage?,difficulties result from: complexity, variability, interactions or synergies, feedback, delays, dynamic complexity., wrong metrics,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,dynamic complexity,a process has dynamic complexity if an action taken on the process has more than one outcome, with the outcomes occurring: at different locations, at different times.,the person taking the action may be aware of only one outcome, and may never see the full consequences of the action.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,sub-optimisation,optimising all of the sub processes will not optimise the macro-process.,the danger of sub-optimisation is reduced if people understand how their process fits into the larger process.,it is dangerous to give people only the minimum information you think they need to do their jobs.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,summary,only through process management will you achieve business excellence. process management is difficult. management of departments or functions may be easier. measurement and variability probably cause the greatest difficulties.,ref: pd components foundation training for breakthrough champions / mr caulcutt,2000-12-29,the need for process control,detection - tolerates waste prevention - avoids waste,ref: spc reference manual (aiag),2000-07-11,process control and process capability,lcl,ucl,time,t,t,ref: spc reference manual (aiag),2000-07-11,stages of the continual process improvement cycle,1: analyze the process what should the process be doing? what can go wrong? what is the process doing? achieve a state of statistical control. determine capability,2: maintain the process monitor process performance detect special cause variation and act upon it.,3: improve the process change the process to better understand common cause variation. reduce the common cause variation.,ref: spc reference manual (aiag),use mapping techniques: risk mapping: fmea qmap: visualizing is a great support,2000-07-11,mapping the process : qmap / highest level a0,5 elements input transformation mechanism control output,2000-07-11,mapping the process : qmap, the zooming in principle,2000-07-11,mapping the process : qmap : example,pre-treatment suppliers/inc.insp. unpack cleaning/drying 2.1,mask design plotting cleaning,photolitho coating exposure development 2.2,etching 2.3,post-treatment stripping cleaning drying 2.4,coating spraying pre-bake cooling,exposure,development koh water rins & dry,patterning: level a1,photolitho: level a2,2000-07-11,control charts: 3 phases,upper control limit,control limit,lower control limit,ucl,cl,lcl,1: collection: gather data and plot on a chart,2: control: calculate trial control limits from process data identify special causes of variation and act upon them,3: analysis and improvement quantify common cause variation: take action to reduce it,ref: spc reference manual (aiag),2000-07-11,control charts: out of control conditions (oocc),2000-07-11,from fmea thru control planning to spc,design fmea,critical / significant characteristics,process fmea,control plan process related product related (customer specific),preliminary statistical capability study,spc,steps to fmea (brain storm technique) use flow diagrams / qmaps etc / enhance teamwork 1: establish & analyse product function. 2: establish possible failure modes, cause & effect. 3: define weak points in the design, material and process. 4: define actions and responsibilities to reach the necessary improvements. 5: establish critical/significant characteristics, both on product level as well as on related process level. 6: establish the control plan for the manufacturing phase. 7: apply statistical studies / measurement system analysis etc for at least the crit/sign characteristics. 8: holding the gains (control) and continuous improvements: feedback loops / learn / apply pdca!,2000-07-11 uzw-b0/h100-13,control plan model.,these lead to the philips test specification (with optional: customer specific control plan),customer and product specific requirements,general product requirements fmeas (risk analysis) preventive maintenance process requirements,product axis,these lead to the process-(cluster) step related manufacturing instructions & control plans (re: process manual),product flow,2000-07-11 uzw-b0/h100-13,process documentation: multi site vs local,2000-07-11 uzw-b0/h100-13 uzw-b0/h500-01,process control plan pfmea for each control plan & mi ocap/spc instr/ major technology class minor gager&r / lwi photolitho batch s/b f/c etc,1,2,2,1,1,2,1,2,3,routing for a given type,local control,multi site,procedure uzw-b0/h100-39 & test specification deal with product (customer) specific control plan,technology class examples photolitho gap 30 s/b: glass thickness acf: pitch / aspect ratio,the qmaps (process mapping) process elucidation training documentation.,3,re: change procedure uzw-b0/h500-01,control plan & manufacturing instructions,pm_control plans process parameters: critical & significant parameters,process manual control plan, multi site controlled,mi_control plan key and normal parameters,for definitions of crit/sign key and normal, see spc procedure: uzw-b0/h100-13,local documentation, single site controlled, unless otherwise indicated by validity list.,manufacturing instruction contents: ref to: pm_control plan ref to: mi_control plan mi chapters,calibration instruction,local work instruction,maintenance instruction,2000-07-11 uzw-b0/h100-13 uzw-b0/h500-01,re: change procedure uzw-b0/h500-01,terminology and abbreviations: crit / sign / key / normal,finished product characteristic: fpc area product parameter: app,process parameter: pp,crit: critical product characteristic,sign: significant product characteristic,kpc : key product characteristic npc : normal product characteristic,crit: critical process parameter,sign: significant process parameter,kpc : key process parameter npc : normal process parameter,resulting from reasonably anticipated variation: severity likely to significantly affect safety / compliance with laws etc. critical likely to significantly affect fit / function significant unlikely to affect customer satisfaction / above areas key / normal,2000-07-11 uzw-b0/h100-13,parameter definitions,2000-07-11 uzw-b0/h100-13,overview of the critical & significant characteristics,2000-07-11 uzw-b0/h100-13,production flow,critical char. (legal & safety): no critical characteristics for lcds,dimensions a, b & dcg for products with: narrow tolerance (+- 0.2mm) 0.7 & 1.1mm glass types,gop: pin length gop: pin bending (for narrow tolerance types: with full swing 4deg),for stn only / at oqc: color reproducibility within one cell,short circuits opens / missing segments inoperatives,significant characteristics (fit & function):,unpack / check glass plates / clean / dry,incoming goods quality assurance,patterning a printing layers & curing rubbing (dot printing: through contacting),patterning b printing layers & curing rubbing (conductive) seal line printing,batch inspection,batch,mep b,mep a,couple pamp,scribe / break into cells fill liquid crystal / close cells,visual / functional test,affix polarizers,finished cell,final inspection (100%),packed finished product,cog: bond ic,functional test,interconnection foil / pins etc.,module assembly,lot release - oqc (sampling),pcb- sub- assy.,example: screenprint process,flow diagram statistical process control,2000-07-11 uzw-b0/h100-13,statistical methods / calculation of control limits,2000-07-11 uzw-b0/h100-13,moving range: introduction / instead of x-r chart,2000-07-11 uzw-b0/h100-13,moving range: weakness of x-r chart,2000-07-11 uzw-b0/h100-13,moving range: weakness of x-r chart: the essential weakness of the x-r chart concerns the estimation of the process inherent (or common cause) variation. in the standard procedure this estimate is based on the variability that exists within the subsequent samples, using the within sample range (note that the same holds true when the within sample standard deviation is used). for many processes however, the within sample variability is not equal to the common-cause variation. the two most frequent reasons for this are: a. the existence of between sample variation that is part of the common-cause system of the process and thus has to be treated as process inherent. examples are samples per batch or situation where both long and short term common cause sources of variation exist. in these cases the use of within sample variability underestimates the process inherent variation and thus leads to control limits in the x-chart that are too tight. to justify the x-chart in such situations asks for a very careful composition of the samples, that is difficult to implement in practical situations. use of the moving range chart is preferred. b. the existence of systematic effects within a sample. examples are position effects on plates or in printing processes or differences between multiple heads, tools or jigs. in this case the x-chart, methodology includes these systematic differences in the estimate of process inherent variation, which leads to control limits in both x and r chart that are too far from the central line.again the use of the moving chart is preferred. c. interpreting patterns of a moving range spc chart should be done with extra care, because of the possible correlation between 2 consecutive sample results. such correlation may induce a particular pattern of runs. verification could be done