Reasonable Error

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

  • Reasonable Error

    I need to nail down a value. I will be measuring a Sphere and setting it to Zero XYZ. Then I will measure this sphere with different tip angles and different modules. I want to get a feel for what you all think is a reasonable deviation (3 Dimensional True Position) to have before it is time to recalibrate.
    I don't want to set the limit too tight as I will end up calibrating constantly. Neither do I want to set it too loose.
    I am thinking 25 microns spatial across all tips all modules? That is basically +/- 12 micorns in any given vector.
    What do you think? Is that too loose? Is that too tight? Just right?
    Bill Jarrells
    A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

  • #2
    Depends on if you are measuring pole barn parts or heart valves. It also depends on the machine. This would be part of your accuracy/uncertainty budget and if you have tolerences to allow that much spatial deviation. Its a fairly application specific number.
    Badges..... We don't need no stinkin badges.

    Comment


    • #3
      Question... will running this comparison program take that much less time than actual tip calibration?

      Comment


      • #4
        Originally posted by Goodluck View Post
        Question... will running this comparison program take that much less time than actual tip calibration?
        Man, your a real spoil-sport....
        Links to my utilities for PCDMIS

        Comment


        • #5
          Originally posted by Goodluck View Post
          Question... will running this comparison program take that much less time than actual tip calibration?
          Absolutely less time. I plan on grabbing 5 tips from each module near the extents (Ie 90/-90 and 90/90) etc. Can determine if calibration is needed in 15 minutes where a full calibration takes 2-3 hours.

          Either way, I want to know what my actual deviation is from tip to tip to module to module.
          Bill Jarrells
          A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

          Comment


          • #6
            Originally posted by Sean Harris View Post
            Depends on if you are measuring pole barn parts or heart valves. It also depends on the machine. This would be part of your accuracy/uncertainty budget and if you have tolerences to allow that much spatial deviation. Its a fairly application specific number.
            I asked for YOUR feel for it. Use your own situation.
            Bill Jarrells
            A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

            Comment


            • #7
              Originally posted by Wingman View Post
              I asked for YOUR feel for it. Use your own situation.
              Your question WAS based on your circumstances, so the questions were valid. You should have measurement range within the value of your uncertainty budget allowed for your machine... Otherwise, expand the uncertainty budget for your machine.
              Links to my utilities for PCDMIS

              Comment


              • #8
                I think a .025mm tolerance on True Position will be extremely tough given the TP formula = 2*sqrt(x^2 + Y^2) you would really be dealing with .006mm

                Right?

                Comment


                • #9
                  Originally posted by Wingman View Post
                  Absolutely less time. I plan on grabbing 5 tips from each module near the extents (Ie 90/-90 and 90/90) etc. Can determine if calibration is needed in 15 minutes where a full calibration takes 2-3 hours.

                  Either way, I want to know what my actual deviation is from tip to tip to module to module.
                  [spoil sport]So you are going to do a sampling of your angles. That will let you know the angles you check are good. What about the angles you don't check? What happens if today you calibrate all your angles on a given probe. Now, you run your "check program" tomorrow and everything looks good. Lets say on Wednesday while you are in a meeting someone comes in and wants to use the machine. Now, lets say they need to use this same probe build but they don't know if the angle they want has been calibrated. So, they calibrate A45B0 answering 'yes' to the cal. sphere moving. Now, all your other angles should be unaffected but A45B0 is going to be all out of whack. If you don't check A45B0 you won't know until???

                  I'm just trying to point out that this isn't a micrometer. With a micrometer you can make a reasonable assumtion of linearity if it is 'on' at 0 and 'on' at 1.0. A CMM is different. With a cmm you can pretty easily adjust 'zero' for one angle without adjusting it for all angles.

                  I'm just trying to urge caution and make sure you think this through completely before you discover you've shipped bad parts.[/spoil sport]

                  Comment


                  • #10
                    Wingman,
                    Do a search of posts by Hilton Roberts. He has written a very good paper on his "golden part". It covers just this concept. Basically, you have a part that you measure an a set frequency. Using statistics, you track the variation of the machine.
                    Hilton is a great source of measurement knowledge.

                    John
                    When in doubt, post code. A second set of eyes might see something you missed.
                    sigpic

                    Comment


                    • #11
                      Originally posted by cmmguy View Post
                      Your question WAS based on your circumstances, so the questions were valid. You should have measurement range within the value of your uncertainty budget allowed for your machine... Otherwise, expand the uncertainty budget for your machine.
                      OK, what is it going to take to get some numbers instead of terms like uncertainty budget?
                      Bill Jarrells
                      A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

                      Comment


                      • #12
                        I wasn't picking a fight with this one. I was simply suggesting that this value isn't really a value based on our opinion. The uncertainty budget will dictate what your number should be. Its not an arbitrary value but an actual science project that needs to be done to determine what that value is, based on your shops specific conditions. For my current workplace the 25 micron number is too tight, for the last place I was at the 25 micron number was way to loose. Where I am now we assemble sheet-metal frames, last place made pacemakers.....Different worlds have different requirements.
                        Badges..... We don't need no stinkin badges.

                        Comment


                        • #13
                          Originally posted by Goodluck View Post
                          [spoil sport]So you are going to do a sampling of your angles. That will let you know the angles you check are good. What about the angles you don't check? What happens if today you calibrate all your angles on a given probe. Now, you run your "check program" tomorrow and everything looks good. Lets say on Wednesday while you are in a meeting someone comes in and wants to use the machine. Now, lets say they need to use this same probe build but they don't know if the angle they want has been calibrated. So, they calibrate A45B0 answering 'yes' to the cal. sphere moving. Now, all your other angles should be unaffected but A45B0 is going to be all out of whack. If you don't check A45B0 you won't know until???

                          I'm just trying to point out that this isn't a micrometer. With a micrometer you can make a reasonable assumtion of linearity if it is 'on' at 0 and 'on' at 1.0. A CMM is different. With a cmm you can pretty easily adjust 'zero' for one angle without adjusting it for all angles.

                          I'm just trying to urge caution and make sure you think this through completely before you discover you've shipped bad parts.[/spoil sport]

                          1) I would kill them. Nobody will EVER calibrate a single probe or Probe set with Move Sphere on unless they want to die.

                          2) This is what I want to avoid. I do not want anybody calibrating any single probe angle for any reason other than to add one to a program. In fact, I don't want anybody's fingers in the calibration utility unless you are a programmer. Period.

                          That is the reason for this.

                          Now, if I can get a number please.
                          Bill Jarrells
                          A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

                          Comment


                          • #14
                            Originally posted by Sean Harris View Post
                            I wasn't picking a fight with this one. I was simply suggesting that this value isn't really a value based on our opinion. The uncertainty budget will dictate what your number should be. Its not an arbitrary value but an actual science project that needs to be done to determine what that value is, based on your shops specific conditions. For my current workplace the 25 micron number is too tight, for the last place I was at the 25 micron number was way to loose. Where I am now we assemble sheet-metal frames, last place made pacemakers.....Different worlds have different requirements.
                            Now THAT is what I am talking about.
                            Thanks.
                            Bill Jarrells
                            A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

                            Comment


                            • #15
                              Originally posted by John Kingston View Post
                              Wingman,
                              Do a search of posts by Hilton Roberts. He has written a very good paper on his "golden part". It covers just this concept. Basically, you have a part that you measure an a set frequency. Using statistics, you track the variation of the machine.
                              Hilton is a great source of measurement knowledge.

                              John
                              Thanks!
                              Bill Jarrells
                              A lie can travel half way around the world while the truth is putting on its shoes. - Mark Twain

                              Comment

                              Related Topics

                              Collapse

                              Working...
                              X