iSixSigma

U chart

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #47596

    Taylor
    Member

    I am attempting to use a U chart to determine control of a process.  I took samples over 25 days.  Sample sizes varied in range of 28 to 32 per day.  There were only 4 issues in that 25 day period.  The out put of the process is either right, or it is wrong, definitely an attribute.  If I am doing the calculation correct, the UCL is around .03 and the LCL is .02.  One mistake and it reflects the process is out of control. Am I missing something?

    0
    #158807

    annon
    Participant

    A couple thoughts:

    When using C or U chart, it is reccomended that your sample size be large enough to capture at least 5 defects per lot.  It appears yours does not. This could account for why you have such narrow control limits (see below) and why your chart appears so overly sensitive:

    UCL/LCL = Cbar +/- 3 (sqrt Cbar)

    C is the # of defects
    It would appear you need larger sample sizes to capture a greater number of defects, providing you with a better estimate of Cbar and less sensitive control limits.
    Another option would be to capture you data in continuous form.  This is usually possible, and will allow you to deal with much smaller sample sizes.
    Verify this and good luck.

    0
    #159128

    Taylor
    Member

    Annon, thanks for the feedback.  This lends credence to my theory that I need more mistakes to show that my process was in control. 

    0
Viewing 3 posts - 1 through 3 (of 3 total)

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