CPK and True position...

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

  • CPK and True position...

    Ok Guru's....What is the general take on trying to establish Capability limits while using true position values? I would like to know a good method to show control on a feature's true position. Our higher ups are more confused than I am. I say it breaks down to treating the basic locations as seperate toleranced entities - would anyone agree with this? Or is there a better way?
    sigpic
    Cold Beer MMMmmmm........

    Dell Precision T5400
    Windows 7
    B&S XCEL
    WENZEL LH1512
    PCDMIS 2011 MR1

  • #2
    TP is a BIG no-no for SPC because it is a TOTAL DIAMETER value of a RADIAL DEVIATION and it does NOT take into account DIRECTION of deviaiton.

    Just for example, you have a hole at X0 Y0.

    Measurements:
    Part #1 X1.0 Y0 TP2.0
    Part #2 X-1.0 Y0 TP2.0
    Part #3 X0 Y1.0 TP2.0
    Part #4 X0 Y-1.0 TP2.0

    The stats on the TP would show ZERO range for those 4 parts. Wait, that hole didn't move?
    sigpic
    Originally posted by AndersI
    I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

    Comment


    • #3
      Why no Matt. The hole didn't move. Didn't you meet our new guy? He's the bomb...

      yes, I think you would need to evaluate the position in terms of seperate axis....
      kb
      RFS Means Really Fussy Stuff

      When all you have is a hammer - everything looks like a nail....
      sigpic

      Comment


      • #4
        As Matt sad, this is a no-no but..... If the customer asks for it you can give it to them. I have an XL spreadsheet that was posted on CMMTalk some time ago that will give you Cpk of TP with MMC bonus added into the calculations.

        Duane
        Xcel & MicroVal Pfx & Global 37mr4 thru 2012mr1sp3
        Contura Calypso 5.4

        Lord, keep Your arm around my shoulder and Your hand over my mouth. Amen.

        Comment


        • #5
          Evaluating TP with CPK is like having too many variables and on a one sidded tolerance at that. CPK assumes a centered process with bi-directional variation. TP is a one sidded variation. Not a good application for CPK. You'll need to break it down into individual elements to apply CPK.
          <internet bumper sticker goes here>

          Comment


          • #6
            If you want to do process control, use the individual axes location. TP in itself can NOT be SPC'ed, as said before, because there is no directional information to be learned.

            On top of that, if you have a virtual condition on your datums, it is possible to "wiggle" the hard gauge until you get the best true position. It is therefore possible that the hole me be pretty far off from it's desired location, while still giving you perfect TP. In other words: there are many locations that give you perfect true position. That is NOT something you can use in SPC calculations.


            Jan.
            ***************************
            PC-DMIS/NC 2010MR3; 15 December 2010; running on 18 machine tools.
            Romer Infinite; PC-DMIS 2010 MR3; 15 December 2010.

            Comment


            • #7
              While manufacturing needs directional information, an SQA coming in for PPAP doesn't care about location as long Cpk looks good for position. If you have never been involved with the actual PPAP process you will not understand this. Been directly involved in this with all of the major automotive players and also ag manufacturers and they just want to know how well you can put the hole with in the limits.

              Duane
              Xcel & MicroVal Pfx & Global 37mr4 thru 2012mr1sp3
              Contura Calypso 5.4

              Lord, keep Your arm around my shoulder and Your hand over my mouth. Amen.

              Comment


              • #8
                Help me out Jan true-positon is the only dimension that has a upper and lower spec limit 0= lower limit .2-upper limit if that's your tolerance. They X&Y does not have tolerance they are calculated with TP. We use to run spc on TP only because the X&Y are not individualy tolerance. (H E double toothpicks) I might be wrong on this so help me out.
                sigpic

                Comment


                • #9
                  Originally posted by AuRules View Post
                  Help me out Jan true-positon is the only dimension that has a upper and lower spec limit 0= lower limit .2-upper limit if that's your tolerance. They X&Y does not have tolerance they are calculated with TP. We use to run spc on TP only because the X&Y are not individualy tolerance. (H E double toothpicks) I might be wrong on this so help me out.
                  CPK assumes centeredness as perfect but with TP 0 is perfect. If you have a TP tol of 0.02 then your best CPK would be a TP of 0.01 even though perfect would be 0. A TP that is perfect would show a bad CPK.
                  <internet bumper sticker goes here>

                  Comment


                  • #10
                    One thing I forgot to mention is that you need some SPC software that will you set your limits to have a unilateral curve (I think it's called a Pearson curve) applied and then it will give a true Cpk.
                    Xcel & MicroVal Pfx & Global 37mr4 thru 2012mr1sp3
                    Contura Calypso 5.4

                    Lord, keep Your arm around my shoulder and Your hand over my mouth. Amen.

                    Comment


                    • #11
                      Originally posted by dwade View Post
                      While manufacturing needs directional information, an SQA coming in for PPAP doesn't care about location as long Cpk looks good for position. If you have never been involved with the actual PPAP process you will not understand this. Been directly involved in this with all of the major automotive players and also ag manufacturers and they just want to know how well you can put the hole with in the limits.

                      Duane
                      Cp/Cpk and Pp/Ppk are ALL dependant on VARIATION, TP does NOT correctly take into account the actual variation of the hole. The example I gave proves that TP is not acceptable BY ITSELF for determining any statistical information because TP hides the direction of variation, it ONLY gives the 'amount' of variation, not the direction, and in the case of LOCATION, direction is VERY important. As a VERY simple answer to the XY tolerance issue, you can do it one of 2 ways, either use +/- 1/2 of the TP tolerance for each axis, thus giving a 'box' zone around the TP zone or you can use 0.7071 times 1/2 of the TP tolerance to give you a square zone that has the corners touching the circle zone of the TP.

                      If I passed a hole, using ONLY TP for stats and the hole WAS staying on an extreme 'circle' from nominal (say 80% of the TP tolerance, every time, but all around the nominal location), then the assembly crew, who are putting together the assembly fixtures, can not place the 'bolt' for the hole because the hole is bouncing around and never in the same place twice. However, if I STAT the XY, then we KNOW that the hole is NOT moving and bouncing around and that it DOES repeat.

                      TP, by itself, does not give any good answers for statistical summary.
                      sigpic
                      Originally posted by AndersI
                      I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

                      Comment


                      • #12
                        Originally posted by Matthew D. Hoedeman View Post
                        TP, by itself, does not give any good answers for statistical summary.
                        I do agree completely with that statement. BUt I have never had a SQE ask for Cpk on the coordinate location of the feature. I will also add that the companies that I worked for pushed Cp if the Cpk was not within acceptable limits of the SQE. If your Cpk was below 1.66/1.33 but your Cp was like 3 or 4 we knew we could produce a part that was within tolerance and acceptable to the customer.
                        Xcel & MicroVal Pfx & Global 37mr4 thru 2012mr1sp3
                        Contura Calypso 5.4

                        Lord, keep Your arm around my shoulder and Your hand over my mouth. Amen.

                        Comment


                        • #13
                          That is where cp and cpk worked with our customers even though the onesided cpk wasn't as good as it should be the cp was, showing that our cnc's were capable of running consitant. So then we could adjust our target where it needed to be....
                          sigpic

                          Comment


                          • #14
                            But the CP is nothing more than the range NOT compared to the tolerance (while Cpk is the range vs tolerance zone), it still does not take into account the actual rage of the movement of the hole. You HAVE to look at the axis movement as well or you can not say for certain that the hole is only moving 0.2mm (range of TP) instead of 2.2mm (range of the X axis).
                            sigpic
                            Originally posted by AndersI
                            I've got one from September 2006 (bug ticket) which has finally been fixed in 2013.

                            Comment


                            • #15
                              When arguing with an SQE that just will not get it I tell him that before you can appropriately apply a capability index to a process you must prove that the process in question is in control. Then ask him to demonstrate how to control chart TP.
                              <internet bumper sticker goes here>

                              Comment

                              Related Topics

                              Collapse

                              Working...
                              X