Stddev

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Stddev

    Can someone please explain, in a succinct way, what STDDEV actually is and how it relates to overall machine accuracy?

    My company follow the 10X rule. If a gage's resolution is 0.001", it can only be used for tolerances of 0.010" or greater. Currently, if we have a probe that "calibrates" and the STDDEV is 0.0003", bossman says "that probe is only accurate within 0.0003" and should only be used for tolerances that are ±0.003" or greater."

    I have my own understanding that differs from the above, would just like some confirmation from others as well as help "wording" my response to management.



    Thanks!
    SF7107(PCD), SF454(PCD), 152614(Quindos), 9159(Quindos), 7107(Quindos), B&S Manual, M&M Gear Checker

  • #2
    Originally posted by DAN_M View Post
    Can someone please explain, in a succinct way, what STDDEV actually is and how it relates to overall machine accuracy?

    My company follow the 10X rule. If a gage's resolution is 0.001", it can only be used for tolerances of 0.010" or greater. Currently, if we have a probe that "calibrates" and the STDDEV is 0.0003", bossman says "that probe is only accurate within 0.0003" and should only be used for tolerances that are ±0.003" or greater."

    I have my own understanding that differs from the above, would just like some confirmation from others as well as help "wording" my response to management.



    Thanks!
    Think of STDDEV as the roundness of the probe.
    sigpic
    Originally posted by AndersI
    I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

    Comment


    • DAN_M
      DAN_M commented
      Editing a comment
      Can you please expand upon what you're saying?

      I need to intelligently explain to my boss exactly what it is and how it effects the machine's accuracy in a way they'll understand/buy into

    • KIRBSTER269
      KIRBSTER269 commented
      Editing a comment
      S-ome
      T-ime
      D-anny
      D-oes
      E-at
      V-aginas

    • DAN_M
      DAN_M commented
      Editing a comment
      Its hard to get my legs all the way behind my head to reach it, though =(

  • #3
    If you have a STDDEV of 0.0001" then basically the probe tip is LESS THAN 0.0001" out of round. If it is 0.0001" out of round, then you can expect 0.00005" of error on any 1 touch (half of the roundness error).
    sigpic
    Originally posted by AndersI
    I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

    Comment


    • #4
      it is the standard deviation of the individual deviations of the tip compared to a perfect tip. How you could break that down.....

      68.27% of the values lie within 1 standard deviation(s) of the mean in normal distribution (+1 stddev to -1 stddev)

      However, that 68.27% is within +/- 1 STDDEV, while Pcdmis is telling 1 standard deviation, so, 68.27" of your hits are within 0.0003" (or is it within +/-0.0003 or within +/-0.00015), now how you would put that into the +/- needed (since it is a total of 2 STD, centered at zero, +1, -1, that 68.27% of the hits are within).....
      Last edited by Matthew D. Hoedeman; 01-24-2020, 11:04 AM.
      sigpic
      Originally posted by AndersI
      I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

      Comment


      • #5
        do 4-hit probe cals, STDDEV will always be zero. Problem solved.
        sigpic
        Originally posted by AndersI
        I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

        Comment


        • #6
          check out Josh's post (#6) https://www.pcdmisforum.com/forum/pc...s-significance

          Comment


          • #7
            Standard Deviation is one of the most common ways of averaging a set of data. If you had a dozen points you could add them together and divide by 12 to get the average. This method uses all the 'raw' data and it will be directly influenced by the min. and max. values.

            Standard Deviation is calculated by...squaring the 'raw' data, adding it together, then finding the square root of the result. It's a method to clean up the raw data and make it more reliable.

            Now...how that relates to the calibration is difficult to say but it is DEFINITELY not a direct value that can be stated.

            The Standard Deviation is a value that is carried forward into more detailed and more complex statistical formulas.

            Comment


            • #8
              Originally posted by Matthew D. Hoedeman View Post
              do 4-hit probe cals, STDDEV will always be zero. Problem solved.
              Truly superior acumen, you have penetrated the deepest recesses of burnout, and have emerged a transcendental, inter-galactic being. In your next life you should emerge as a qualification sphere. Kudos.
              sigpic

              Comment

              Related Topics

              Collapse

              Working...
              X