Results Not Making Sense - Concentric Diameters

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

  • Results Not Making Sense - Concentric Diameters

    ***Disclaimer*** - I'm not a machinist

    I'm having a hard time understanding why my CMM is showing these results. I'm measuring the concenticity of 2 circles to 1 cylinder. I use the cylinder as my XY origin then probe 23 points for each circle. 1 circle is a little larger than the cylinder diameter and the other circle is much larger than the cylinder diameter. I then dimension X, Y, and Diameter for the cylinder and 2 circles.

    I get that my cylinder XY is 0 due to the fact that I orgin'd to it and the diameter is in tolerance. My first circle diameter and X is in tolerance but my Y is dev from nominal by -0.008". My much larger circle diameter is undersized and X is in tolerance but my Y dev is -0.030". So my circles are not very concentric to my cylinder.

    It makes sense that the circles are not concentric to my cylinder because a different tool cuts the cylinder and another tool cuts both circles. The problem is that the circles are machined (lathe) with the same tool during the same path/machine operation so production is like no way that these diameters are that far out. I understand that however I've been trying to figure out what I'm doing wrong for over 2 weeks.

    I'm using a 7.10.7 with a touch trigger probe. My probes aren't long enough to reach the circle diameters in A0B0 so I have to use multiple angles and vector points. I used auto circle with to points toggled on and made sure I had clearance to hit each point without hitting the part. So the cmm routine is probe comes above part, changes to an angle, goes in and take 2-4 points, comes up, changes angle and so on until I have made a diameter. I've used different probes and different angles, started the a new program and I can't shake this deviation.

    I have 12 parts to check and I've only checked 2 but the results are similar. Tomorrow production is going to try and put this parts back in the CNC machine and indicate it to see if its truly off.

    My boss told me today that I did something wrong and I know i'm not perfect but it struck a nerve. I've done everything I could think of to shake this deviation and it wont go away.
    Attached Files
    Last edited by acgarcia; 12-09-2019, 11:14 PM.

  • #2
    How am I aligning?

    I use the plane above the cylinder feature as my Level Zplus and Z orgin. There is another cylinder that is in Y- for rotation and then the cylinder in green in the pic that I use for my XY origin.

    Comment


    • #3
      You need to level on the cylinder and origin x,y, then measure the circles

      B&S CHAMELEON/PCDMIS CAD++ V2011

      There are no bugs, only "UNDOCUMENTED ENHANCEMENTS!"

      sigpic

      Comment


      • Douglas
        Douglas commented
        Editing a comment
        or define the cylinder as a datum for the concentricity? that should work regardless what the alignment is even if the cylinder is not at origin.

    • #4
      If your scheme is right, the cylinder is too much shorter to give a good orientation !!!!
      I don't know about the part, but can you measure it in the other side ? As I can see, I would place the green side on the top, level on a plane, origin on the green circle, then measure other circles A0B0, and look at their location (It can't be a concentricity because the cylinder is too short, it could be dimensioned as a location or a runout)

      Comment


      • #5
        Other things to consider is out of balance forces and distortion, if as most machinist do you are using constant surface speed then the RPM will be different for each of the diameters and it can have a large effect on close concentricity limits.

        You could ask the machinist to change to a fix RPM with a compromise between the RPM the two diameters, run a test part and see if it helps.

        Report the form of the diameter establish how much distortion is happening and how it is changing depending on how close it is to the lathe jaws. Differing out of round conditions will impact the concentricity result.
        Is the out of round typical tri-lobe or do you have a lazy jaw resulting in an odd shape.
        Last edited by UKCMM; 12-10-2019, 04:27 AM.

        Comment


        • JEFMAN
          JEFMAN commented
          Editing a comment
          He's measuring with a ttp...
          Do you think you can see those defects with a ttp ? (just a question...)

        • UKCMM
          UKCMM commented
          Editing a comment
          I guess if the errors are big enough then yes so long as the probe build is kept as short as practical.

          Just to clarify I agree this measurement is very difficult given the short reference length but that other factor can have a big influence on results.
          Last edited by UKCMM; 12-10-2019, 04:40 AM.

      • #6
        Okay, so there's a couple of possible causes to this.

        1) Your alignment isn't right. Technically dph51 is right but in practical terms he's not, and Jeffman is. Technically you want to level to your datum cylinder, BUT if you've drawn that half right, them it's too short considering the length of the cylinder and the length of the part. Assuming the small end face is machined at the same time as the datum cylinder you'd be better off doing what Jeffman suggested and turning the part over and leveling off that face. however I'm assuming there's a reason you put the part the way up you did. Is the part raised up at all allowing you to get underneath?

        2) Probe correlation - you're obviously indexing the probe to get around the part, if your probes don't agree with one another you will get duff numbers. Easy enough to check, just check the circularity of the features in question, it should jump out at you if there's something amiss.

        Automettech - Automated Metrology Technology

        Comment


        • #7
          Also, to consider, the deviations are in the same axis and same direction. So, all you have is a simple skew. Likely it's due to the face being A and it's slightly off perpendicular to the cylinder. I would check it with the cylinder primary, but even then, be prepared for bad news because it is so short.
          "This is my word... and as such is beyond contestation."

          Comment


          • #8
            I agree, the cylinder length is too short. The amount of error over your distance is skewing it. Wouldn't he be able to measure concentricity of a circle to a circle?

            Comment


            • #9
              Calibrate all your probe angles.

              Level to an axis line constructed from the largest ID circle, and to the smallest diameter that you are currently trying to use as a datum.
              Your datum cylinder is too short to project a measurement along that axis length.

              If your concentricity error in the part is real, you will see ~0.015" for the middle diameter.
              Get ready for disappointment. :-D

              Comment


              • #10
                A way to *test* if the alignment is the main culprit is to align Z to the line through top and bottom circles, and dimension the middle one.

                Edit: louisd was slightly faster than me...
                AndersI
                SW support - Hexagon Metrology Nordic AB

                Comment


                • #11
                  If you are measuring circles, make sure the toleranced feature shares the same height as the datum (at the same level/in the same plane/surface), otherwise the distance between them will be included in the concentricity result. At least it used to.

                  Just a heads-up.
                  PC-DMIS CAD++ 2o19 R1 SP9

                  Comment


                  • #12
                    so as nearly everyone said, the datum is to short.
                    Do you have to measure it like it's drawn?
                    If you can measure function over print you could measure as many circles as possible on all your cylinders, then create a line out of the midpoints (axis) and measure the straightness of this line. This is a possibility if the short cylinder has to be the datum and no other geometrical element will be the one which does the alignment of the part. Mostly this is just possible if you know the function, but maybe you could give it a try. Just imagine what a concentricity is and what you get if you do a straightness. Maybe draw if for yourself with really bad form, so you see what happens if the datum is this short

                    Comment


                    • #13
                      Thank you all for your posts. A lot of this makes sense on how the short cylinder skews the measurement. The part is a large valve body used in oil and gas. I'm checking the pocket so it has to be orientation like my drawing. Its about 500 lbs.

                      From what was indicated, the machinist used the large circle as his zero and found the small circle to be +0.010" from it and the short cylinder to by -0.020" from it.

                      Let me try to make a line using my 2 circles as the XY origin and then probing the short cylinder.
                      Last edited by acgarcia; 12-10-2019, 11:56 AM.

                      Comment


                      • louisd
                        louisd commented
                        Editing a comment
                        Did you re-calibrate all probe angles?

                    • #14
                      Ok, so here's what I did. I measured the large circle and small circle, constructed a line in Z, level to line, rotate Y- to my perp cylinder for rotation, the XYZ Orgin to the leveled line. Then I probed the short cylinder and dimensioned XY to find X in tolerance and Y out of tolerance by 0.0157" which is a lot better than 0.030 and my alignment is better too.

                      The circles are done with the same tool and operation so production was adamant that they have to be concentric and therefore should be my XY origin feature.

                      I'm not sure if this is the end of my adventure.

                      Comment


                      • #15
                        Originally posted by acgarcia View Post
                        Ok, so here's what I did. I measured the large circle and small circle, constructed a line in Z, level to line, rotate Y- to my perp cylinder for rotation, the XYZ Orgin to the leveled line. Then I probed the short cylinder and dimensioned XY to find X in tolerance and Y out of tolerance by 0.0157" which is a lot better than 0.030 and my alignment is better too.

                        The circles are done with the same tool and operation so production was adamant that they have to be concentric and therefore should be my XY origin feature.

                        I'm not sure if this is the end of my adventure.

                        I'm not having a go here, but please try be a bit more consistent with your descriptions.

                        'Large circle' - fair enough we can assume you mean what you called Large ID

                        'Small circle' - eh? which one, the one you've described as Small ID, or the one that was 'cylinder xy zero'.

                        It makes is difficult to follow what's going on.


                        Anyway, what I was going to dd here was this...

                        1) Make sure that it's a 3D line you've constructed from the Large/Small circles

                        2) If you CBA then re-jig your program so the part is rotated round 90° and you should see the deviation is now in X instead of Y (this is just a nice way or verifying what you're seeing is your part and not your machine/prone that causing the deviation)
                        Automettech - Automated Metrology Technology

                        Comment

                        Related Topics

                        Collapse

                        Working...
                        X