Find radius with Auto Feature circle

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

  • Find radius with Auto Feature circle

    I am not having much luck finding help on this, so I will start a new thread. I am measuring a diesel crankshaft, and need to find the tangency point of the radius to the OD, as well as the size of the radius. The radius is a nominal 2.54mm. The bad part is that the spacing tolerance is 0.75mm. I was using an autocircle to find the center of the hole, but as the crank deviates from nominal spacing, the radius moves away from (or towards) the center of the circle. It will then report the diameter incorrectly, as I believe it is using all of the points. Is there a way in the AutoFeature circle to discard points outside of a particular standard deviation, as you can in a constructed circle?

    I found an older program where the programmer took 10 hits over 60 degrees of the radius, constructed a circle, and excluded the points outside of 3 standard deviations. There are 2 problems with this...
    a) I have not clue how he determined what vectors to use for the points, as they start on the OD, then move into the radius.
    b) 7 bearings and 6 pins x 14 radii x 10 points per radius = 1,820 manual points!! That would be pretty time consuming and prone to error.
    sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

  • #2
    Can you post a sketch or portion of the drawing.... having a little trouble picturing what you are trying to do.
    Lately, it occurs to me
    What a long, strange trip it's been.

    2017 R1 (Offline programming)

    Comment


    • #3
      As John indicated, it s tough to understand what you are after without some sort of sketch.

      If you are trying to work with two circles/arcs that are not concentric, and you expect that their circumferences do intersect at one or more opints, there might be something you should consider.

      If they do intersect at a tangent point, that point lies on a line drawn between the center of each of the circles.

      If there is more than one intersection, it gets more difficult, but lets take this one step at a time.

      Comment


      • #4
        As an options use surface points and tolerance the radii as individual points along the radius. I have been alloud to do that. And you will get better results. Throw that out there to who ever and show him on paper how you will check say three points one at the beginning, middle and end of each radius. And report the T Values.
        sigpicSummer Time. Gotta Love it!

        Comment


        • #5
          Thanks for the responses and sorry for the long delay. I was out yesterday.

          Attached is a picture. I don't want to show too much of the print, but it should give you the idea. It is of a crankshaft, looking at it from the side, across the top of a journal. The sidewall is on the right. The red double arrow shows which way the journal "floats" for spacing (as much as 0.75mm is allowed). I am trying to check the 2.54mm radius (blue arrow). However, when I specify the nominal center of the radius and use 60 degrees of an autocircle, it works well if the journal is right on spacing. If it is off, the autocircle reports a much larger (or smaller) radius than I know to be there AND the center of the circle (which I was using for tangency point) is off as well. I cannot set a subdatum from the right side, as there is only about 70 degrees of the radius present. The drawing shows it going up to vertical, but that is not the actual case, as a deviation was given to only have part of the radius ground. Spacing is to be reported to the tangency points. What is the best way to find it?

          I need to find:
          a) the tangency point
          b) the radius size
          sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

          Comment


          • #6
            I am trying the attachment again.
            Attached Files
            sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

            Comment


            • #7
              Originally posted by Paul Sarrach View Post
              As an options use surface points and tolerance the radii as individual points along the radius. I have been alloud to do that. And you will get better results. Throw that out there to who ever and show him on paper how you will check say three points one at the beginning, middle and end of each radius. And report the T Values.
              How do you determine the vectors of the individual points? The old program used surface points with zero sample hits. Wouldn't that in effect be a vector point?
              sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

              Comment


              • #8
                Given a case like that, where the arc position is in question, and directly leads to the accuracy of the measurement, I would try something like this -

                Measure (or scan) a series of hits from one straight to another, on either side of the arc. Use a spacing such that the arc would include about 8 hits or soo. Many more and it just wastes time, many less and it might not calculate properly, just a guess.

                Loop through the points and work with successive groups of 3. Calculate and arc through each group and store the center and the radius. When both of these repeat within an allowable tolerance, and the radius closely matches your nominal, then you know the approximate center.

                Now take that result and feed it (as variables) in an AutoCircle feature to get an accurate measurement.

                I am running some other measurements on my machine right now, but I will try to cevelop this routine and post it for future reference. . . .

                Comment


                • #9
                  Like I have stated, I really do not use scan data for dimensions or constucting. But I have UG. Like Don said, only if you could put it in some CAD Software and analyze it there.
                  sigpicSummer Time. Gotta Love it!

                  Comment


                  • #10
                    Originally posted by Don Ruggieri View Post
                    Given a case like that, where the arc position is in question, and directly leads to the accuracy of the measurement, I would try something like this -

                    Measure (or scan) a series of hits from one straight to another, on either side of the arc. Use a spacing such that the arc would include about 8 hits or soo. Many more and it just wastes time, many less and it might not calculate properly, just a guess.

                    Loop through the points and work with successive groups of 3. Calculate and arc through each group and store the center and the radius. When both of these repeat within an allowable tolerance, and the radius closely matches your nominal, then you know the approximate center.

                    Now take that result and feed it (as variables) in an AutoCircle feature to get an accurate measurement.

                    I am running some other measurements on my machine right now, but I will try to cevelop this routine and post it for future reference. . . .
                    I am not exactly sure I follow you. (Loop through the points and work with successive groups of 3.) I am still pretty new to PC-DIMS, and just returned to CMMs after about a 10 year hiatus. My last exposure was GeoMeasure 2000.

                    I have tried several different methods and each one reports different values. I have radius numbers from ~2.2 up to ~8mm. The Y and Z values shift by several mm as well. I have tried:
                    a) autocircle
                    b) set origin to nominal radius center, take hits every 4 degees through the 60 degrees of arc I have available.
                    c) autocircle, reset origin (YZ) to that circle, remeasure, and the next circle was way off.
                    sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

                    Comment


                    • #11
                      I don't like measuring radiuses like that one, pain in butt. What kinda tollerence do you have? I usualy find a radius like this repeat with in .010-.015 Not great but ok for what I had.

                      Comment


                      • #12
                        a) I have not clue how he determined what vectors to use for the points, as they start on the OD, then move into the radius.
                        I don't know if this will help.

                        You could just move your tip around at different angles and find one close to what you want your approach angle should be, and at the end of that angle will be that IJK vector.
                        sigpichttp://npmichaels.com/ (My sons book)
                        THANKS (CAD IS ALWAYS MASTER)
                        4.3 & 4.2MR2/GLOBAL/PH10MQ/WINDOWS XP

                        Comment


                        • #13
                          A surface point can come handy here with 3 space hits .005 apart, to determain a vector.

                          Comment


                          • #14
                            Originally posted by MIKEY View Post
                            A surface point can come handy here with 3 space hits .005 apart, to determain a vector.
                            \

                            Oddly, the older program I am trying to replicate (which did not have an auto alignment and many other issues) used surface points, but with zero sample hits. In other words, it appeared to be using the surface points like a vector point. I am not sure why.

                            In any case, picked through the advice given, pulled the part off the table, and ran some tests on a part with a hole about the same size as the radius I am trying to check. Set up an auto alignment, and set the nominal of the hole to its actual location.

                            1) Diameter (radius) and location look good when checking 0-360 degrees. It also showed the same with using only 60 degrees of the circle.
                            2) shift alignment 0.5mm, rerun the same hole. 0-360 diameter changes, but very little. Location changes by the 0.5mm. 0-60 degree arc was 0.2mm too small, and 0.1mm off new nominal location!
                            a) did an "iterative" alignment - measure radius, align to it, measure it, align to it 3 times - it won't settle in. Too small of a radius with too little arc.
                            b) checked with auto circle (using only 60 degrees) and manual points with constructed circle - results still terrible.

                            Plan B - Forget Radius size - How can I just get the tangency point (for width and spacing)? I will check the radii on the Mahr.
                            sigpic"Common sense in an uncommon degree is what the world calls widsom." -Samuel Taylor Coleridge

                            Comment


                            • #15
                              Not nessesery, the way i do it (in some cases) the first run I use space hits, to find the vector and after i turn it off bcause vectors already there and no sense to take 3 extra hits since the vectors are known.

                              Comment

                              Related Topics

                              Collapse

                              Working...
                              X