Control chart

1、 companyworkwork instructionsinstructionsdocument number issuepage1 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.1.0purposethe purpose of this specification is

2、to provide the methodology for set-up and operating control charts and guideline whether a system is statistically able to meet a set of specifications or requirements.2.0scopethis document applies to all operations with responsibility for ensuring process control for processes directly affecting pr

3、oduct quality.3.0document information3.1reference documentsdocument numberdocument titlequality system manualcontrol of quality recordsstatistical techniques and analysis of datacontinuous improvement processtotal control methodology (tcm)statistical process control (spc)3.2document classificationth

4、is document is classified as “company general business information”. the information disclosed herein is the property of company, company reserves all proprietary, design, manufacturing, reproduction, use, and sales rights thereto, and to any article or process utilizing such information, except to

5、the extent that rights are expressly granted to others.3.3acronyms, definitions & termsdefinitions & termsdescriptionattributes dataqualitative data that typically shows only the number of articles conforming and the number of articles not conforming.capabilitythe ability of a process to pro

6、duce outputs that meet engineering and/or customer specifications. companyworkwork instructionsinstructionsdocument number issuepage2 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “contr

7、olled copy” in red.center linethe line on a control chart that represents the average value of the statistic being plotted. the center line is not necessarily the target of the process. characteristica distinguishing feature of a process or its output on which variables or attributes data can be col

8、lected.control charta graphic representation of a characteristic of a process, over time, showing plotted values of some statistic gathered from that characteristic, a central line, and one or two control limits. it has two basic uses: as a judgment to determine if a process has been operating in st

9、atistical control, and to aid in maintaining statistical control.control limitscontrol limits are calculated from process data and are not to be confused with product or process specifications. they are represented as line (or lines) on a control chart used as a basis for judging the stability of a

10、process. variation beyond a control limit is evidence that special causes are affecting the process. (ucl: upper control limit /lcl: lower control limit)important processa process is considered important if changes in the process significantly impact the form, fit, function, reliability, or quality

11、of the product.individuala single unit, or a single measurement of a characteristic, often denoted by the symbol x.meanthe average of values in a group of measurements.moving rangethe moving range is the absolute difference between successive individual measurements.normal distributiona continuous,

12、symmetrical, bell-shaped frequency distribution for variables data that is the basis for many control charts.processa combination of people, equipment, methods, materials, and environment that produces output, - a given product or service. a process can involve any aspect of a business.rangethe diff

13、erence between the highest and lowest values in a subgroup, a sample, or a population.rational subgroupsubgroups or samples that are selected so that if assignable causes are present, the chance for differences between subgroups will be maximized, while the chance for differences due to these assign

14、able causes within a subgroup will be minimized.shewhart rules or teststhe eight rules, or tests, used to detect small shifts at a faster rate on x-bar or individuals control charts. companyworkwork instructionsinstructionsdocument number issuepage3 of 26title : control chart electronic versions are

15、 uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.spcspc is the use of statistical methods to analyze the inputs and/or outputs of a process so as to take appropriate actions to achieve and maintain a state of stati

16、stical control of the process. control charts are one of the important spc tools which are mainly used for monitoring process variability, distance from the target, and the results of corrective actions.special causessources of assignable variation that is intermittent, often unpredictable, or unsta

17、ble.stabilitythe absence of special causes of variation; the property of being in statistical control.stable processprocesses that are in statistical control. variation in the output of a stable process arises only from common causes. a stable process is predictable.standard deviationthe scatter or

18、spread in the sample data as defined below: statistical controlthe condition of a process from which all special causes of variation have been eliminated and only common causes remain. statistical control is evidenced on a control chart by the absence of points beyond the control limits and by the a

19、bsence of any non-random patterns or trends.variables datameasurements taken on a continuous scale. they are quantifiable and incremental in nature.variationthe inevitable differences among the measurements of a process.4.0control chartsthe following flowchart may be used to determine the appropriat

20、e control chart for each of the variables being monitored. companyworkwork instructionsinstructionsdocument number issuepage4 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled co

21、py” in red.nooryesnovariableattributen1n=1largesmalllargesmallpercentages /ratiosdefects /eventslargesmalllargesmallyesordata from aprocess to becontrolledwanderingmean?sample size?datatype?shiftsize?x-bar / sx-bar / rcusumewmashiftsize?x(individual)mrcusumewmashiftsize?pnpcusumewmausing pshiftsize?

22、cucusumewmausing c,u;time between eventsmodifiedcusum,ewmafit armamodel; applycontrol chartsto modelresidualsuse ewmawith controllimits basedon predictionerror variancearedataautocorrelated?variables orattributes?fit arimamodel; applycontrol chartsto modelresiduals companyworkwork instructionsinstru

23、ctionsdocument number issuepage5 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.4.1the control chart is a graph used to analyze variation from a processes or equi

24、pment. they provide real-time feedback on the stability and predictability of processes and equipment. by comparing current data to historically determined lines, one can make conclusions about whether the process is stable or is being affected by special causes of variation. note on terminology: if

25、 a process is stable over time, it is said to be in-control. if it is unstable, it is said to be out-of-control. the terms (stable, in-control, and unstable, out-of-control) are used interchangeably in spc literature and applications. (figure 1) illustrates a typical control chart. .u uc cl ll lc cl

26、 lc cl lt ti imme e v va ar ri ia ab bl le es su ummmma ar ry y s st ta at ti is st ti ic ca a c co on nt tr ro ol l c ch ha ar rt t(figure 1. example of control chart)4.2control charts are trend charts that are used to monitor and control processes or equipment. they provide real-time feedback on t

27、he stability and predictability of processes and equipment.each point on the chart is an outcome of a subgroup summary statistic of the process measurements, such as a mean, a range or a standard deviation, or an individual value, plotted at some given point of time. thus, the control chart plots th

28、e trend of the particular summary statistic over time. the horizontal axis of a control chart is a time variable, such as hour, shift, day or week. companyworkwork instructionsinstructionsdocument number issuepage6 of 26title : control chart electronic versions are uncontrolled except when accessed

29、directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.4.3control limits and center linethe features of a control chart that distinguish it from a basic trend chart are the presence of statistically based control limits and a center line. these features a

30、llow for the objective evaluation of the stability of a process.4.3.1 the center line (cl) the center line of a control chart (denoted by cl in figure 1) is the mean value of the summary statistic when the process is stable (i.e., in-control). note: in general, the center line is not the target of t

31、he process variable.4.3.2 control limits (ucl and lcl) upper and lower control limits (ucl and lcl, respectively in figure 1) are constructed so that a high percentage of the time (e.g., 99.73%) the process summary statistic will fall within the control limits, if the process is stable (i.e., in-con

32、trol). note: control limits are not specification limits.the state of the process determines the control limits, which will change based upon elimination of causes of variability and/or process changes (including procedural or equipment changes).if the control limits are constructed correctly, and t

33、he process remains stable, it is very unlikely that a value of the summary statistic will fall outside these limits. if there has been a change in the process, equipment or data collection procedures, a point falling outside of the control limits is a signal that some corrective action is required.

34、5.0how to make control charts control charts exist to distinguish special cause variation from common cause variation, so that the process perturbations that generated the special cause variation can be effectively diagnosed and eliminated from the process. companyworkwork instructionsinstructionsdo

35、cument number issuepage7 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.the control chart limits must be calculated appropriately to optimize the performance of t

36、he control system. limits that are too wide for the process data will fail to detect special cause variation, and result in missed opportunities to improve the process. limits those are too tight for the process data will mistakenly identify an inordinate amount of common cause variation as special

37、cause, thwarting effective trouble shooting of the process. note: the use of the shewhart formulas for some processes can lead to incorrect control limits, thus resulting in false signals of out-of-control events, or an inability to detect actual out-of-control conditions. 5.1determine new control l

38、imits and/or change in existing limits5.1.1 a new control chart will need limits calculated before it can be put to use in the manufacturing area. it is important to determine the appropriate control limits for each control chart. incorrect control limits will allow for data points to be in control

39、when the process is out of control, or for data points to be out of control when the process is in control.5.1.2 listed below are reasons that control limits need to be changed:recalculation and revision control of control limit should be performed in spc system. periodic recalculation which should

40、be performed in spc system automatically, is based on set-up criteria of scheduled recalculation and should be reviewed and approved by the pertinent process engineer or spc responsible engineer. for process change or equipment, the recalculation and acceptance of new control limit should be perform

41、ed by the pertinent process engineer. process change scheduled recalculation (recalculation per 30 subgroups) companyworkwork instructionsinstructionsdocument number issuepage8 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.pri

42、nted versions are uncontrolled except when stamped “controlled copy” in red. equipment:- new piece of equipment from existing equipment set- equipment modification- equipment move or re-qualification- identified and accepted equipment wear5.1.3 extract data over proper timeframe and reference

43、.there must be adequate data to encompass the natural variation of the process. 30 plot points is a rule of thumb (rot) that, historically, delineates the progression from a small sample to a large sample. a plot point is the point to be plotted on the chart. it may represent an individual da

44、ta point or the summary statistic for a subgroup of data points. in reality, 30 plot points rarely encompasses the natural variation of a process. for instance, some operations can generate 30 plot points in two days, while batches of material in the same operations are changed once a month.

45、if limits are calculated on the two-days-worth of data, it is probable that those limits will not encompass the monthly material change, ensuring that the limits need to be recalculated once every month. conversely, some operations may take days to complete. waiting several months to institute proce

46、ss control is probably not prudent for these operations. 5.1.4 determine whether change or not control limits if the range of control limit (current or new) is tighter than below 30% of spec limit, and each control limit must be parted from each spec limit as 20% of spec range, then additiona

47、l change is not needed. if new control range is tighter than current control range as over 20%, new control limit should be changed. if below 20%, change is not needed. companyworkwork instructionsinstructionsdocument number issuepage9 of 26title : control chart electronic versions are uncon

48、trolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red. if new control range is wider than current control range as below 20%, change is not needed. if over 20%, the process of interest is probably too unstable to imple

49、ment control charts upon, and should be reviewed by spc responsible engineer. if new control limit is shifted from current control limit as below 20%, new control limit can be accepted. if over 20%, the process of interest is probably too unstable to implement control charts upon, and should

50、be reviewed by spc responsible engineer.5.1.5 assess the control chart with limits and /or new limits if no points are out of control, continue. if less than 10% of the plot points are out of control based upon the newly-calculated control limits, cull the out-of-control plot points fr

51、om the reference dataset, recalculate the limits, reattach them to the chart, and reassess the new limits. if 10% or greater of the plot points are out of control based upon the newly-calculated control limits, the process of interest is probably too unstable to implement control charts upon,

52、 and shall follow continuous improvement procedure. note： company follow quarterly review control limit5.2 data type5.2.1 variable datavariables type data consist of continuous (i.e. quantitative) type data. such data are measured on a continuous scale, where all values on this scale are possible ou

53、tcomes of a measurement. the actual observed outcome is limited only by the discrimination capability of the measuring tool.examples of variables data include measurements of length, thickness, temperature, resistance, voltage, time, etc., as measured in such continuous scales as millimeters, degree

54、s, ohms, etc. companyworkwork instructionsinstructionsdocument number issuepage10 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.printed versions are uncontrolled except when stamped “controlled copy” in red.5.2.2 attribute dataattribute dat

55、a include categorical data (e.g., good or bad, machines 1, 2, or 3), count data (e.g., number of defects, particles added, number of failed wire bonds) and percentages or proportions (e.g. fraction defective, yields). attribute data are measured on a discrete rather than a continuous scale. that is,

56、 only certain outcomes are possible. 5.3 control charts by data typethere are two types of data collected in the manufacturing line; variables data and attribute data. the type of control chart used depends upon the data type being monitored. 5.3.1 variable control chartsvariable control charts are

57、used to monitor either the summary statistics or individual data points that measure the location (e.g., the mean) and the variability of the process data. examples of variable control charts include: - individual x / moving range charts - x-bar chart - r chart (range chart) - s chart (standard devi

58、ation chart) - multivariate chart - cusum chart - ewma chart note: it is strongly recommended that variables data always is monitored using a combination of a location chart (e.g. a mean chart) and a variability chart (e.g. a range chart).5.3.2 attribute control chartsattribute control charts are us

59、ed to monitor attribute type data. common summary statistics include counts, average counts and percentages or proportions. companyworkwork instructionsinstructionsdocument number issuepage11 of 26title : control chart electronic versions are uncontrolled except when accessed directly from server.pr

60、inted versions are uncontrolled except when stamped “controlled copy” in red.examples of attribute control charts include: - c charts - u charts - p charts - np charts5.4 calculate limits and/or new limitslimits for variable charts will be calculated using the grand mean and the standard deviation of all th