iSixSigma

Setting Lower Control Limit and Upper control Limit

Six Sigma – iSixSigma Forums Old Forums General Setting Lower Control Limit and Upper control Limit

  • This topic has 10 replies, 10 voices, and was last updated 13 years ago by Ron.
Viewing 11 posts - 1 through 11 (of 11 total)
  • Author
    Posts
  • #50958

    Long
    Member

    Hi all
    I am currently trying to set a control range for a plastic production company. I currently have a set of historical data at a size of 200 or more. I am contemplating on using X bar chart to allocate my LCL and UCL is this a wise choice?

    0
    #175846

    Frank Deruyck
    Participant

    Hello Roger,
    using an XmR chart is a wise decision if your data are periodically collected from a production stream and show a normal distiribution. In case they’re not normally distributed you still can perform a transformation to normalize them however this makes interpretation difficult..
    In case your data are organized in subgroups with a fixed size an Average and Range Chart is recommended.
    Good luck & regards,
    Frank Deruyck
    ASQ Six Sigma Black Belt
     
     
     
     

    0
    #175879

    Ramil
    Participant

    Hi,
    Do you know what is the wafer fab industry standard in setting control limits for critical parameter control charts? Is it average +/- 3 sigma? We want to know also what is the common practice for measuring wafer samples (e.g. how many sites per wafer? distribution of measurement points across the wafer).
    Regards,
    Ramil

    0
    #175880

    Mikel
    Member

    Wrong – bad advice.Forget the normality and forget the transform nonsense – read
    Wheeler.Nobody is impressed with a CSSBB around here – most are empty
    suites with mediocre knowledge.

    0
    #175884

    anon
    Participant

    “empty suites”????? – unrented hotel rooms?? Don’t get it.

    0
    #175895

    Brian M
    Participant

    Why does everyone want to make distributions normal… and since when does an XbarR chart require a normal distribution?Just a few questions ;)

    0
    #175896

    Ron
    Member

    Shewhart Control charts DO NOT REQUIRE NORMAL DATA SETS…
     

    0
    #175899

    Robert Butler
    Participant

    If you want particulars the post below may be of some value:
    https://www.isixsigma.com/forum/showmessage.asp?messageID=141008

    0
    #176072

    Daniel Van Eeckhoutte
    Participant

    using 200 datapoints is a sufficient number to measure if your system is stable or not, but it does not reflect what your customer expects. a VOC analysis may be helpfull.
    The X-bar R chart will identify if you see variation over time, which is an indication on how good your process is under control. But what you see as unacceptable variation may be of no importance to your customer(s), and if you would conclude that your process is OK (no exeeding limits UCL – LCL) might be unacceptable to your customer. An open discussion with the customers may set target specifications, that you can use to calculate Cpk values. that Cpk value should be >1.33
    Daniel

    0
    #176073

    Clive
    Participant

    I think your confusing Control Limits and Specification Limits. Specification limits are the ones set by the customer and indicate if your process is producing what the customer wants. Control limits are set by the variation of the process and are for your internal control processes.

    0
    #176074

    Ron
    Member

    Roger,
    Control limits are determined by the normal variaiton captuired by using small samples taken over the range of variable the process normally sees. i.e. if you have multiple shifts take samples from all shifts.
    You have not described the process but if it runs 24/7’s take a sample every 15 minutes and subgroup by hour (4 samples per hour) collect these samples for 30 days then put them in a statistical software package and crunch the numbers.
    If you don’t have a minitab type software you can easily do this manually.
    Using a single data set of 200 pieces can be useful only if it was taken from a subset of the process run across all possible source of variaiton.
    Since you have not described how the data you have was collected I cannot give you a responsible answer.

    0
Viewing 11 posts - 1 through 11 (of 11 total)

The forum ‘General’ is closed to new topics and replies.