Alignment/Measuring

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

  • Alignment/Measuring

    Hello all,

    I'm back with another rapid-fire set of questions:

    I'm having a lot of troubles getting my alignments down. So, basically where i work, we make exhaust tubes. I am almost always supplied a cad, with a predetermined origin and rotation orientation for my coordinate system. Because of this, I was told in the PC-Dmis 101 class, I would have to almost exclusively use best-fit alignments. Now, to do my alignment, I typically take four separate hits (using my mouse on my CAD) on each end of the tube in a kind-of plus pattern. Then i start a new alignment, click best fit, select all 8 of my hits, click compute, then ok, ok. So that's my initial alignment. Then I select auto-trigger mode and place 4 hits on either end of the tube (The cut part of the tube, by the way; what will eventually be measured as a plane soon), and create a second best fit alignment using the 8 new auto trigger hits.

    After this is done, i shift-left click all my features starting at one plane, then all the cylinders down the part, then the last plane. I construct a pierce point between the first plane and first cylinder, and then the second plane and the last cylinder.

    Here's where i get into trouble:

    By blueprints typically call out distances between these two points across the X, Y, and Z axis, all separately. X normally goes down the length of the part, Y for the width, and Z for height. My X measurement is 'almost' always in spec, and then my Y and Z measurement is be off, sometimes significantly. This leads me to believe that there is a rotational issue present, likely due to how i built my first alignment. On occasion, my x measurement will be near perfect, then Y and Z be off my 8-10mm. Yet when i check the part to the check fixture the issue isn't present, which only further confirms that i am doing something blatantly wrong. In the PC-Dmis class, we only used the training block, so it was incredibly easy (relatively) to visualize my alignments, and make sure i had all degrees of travel and rotation locked.

    Please, someone help me understand what I'm doing wrong, and what i can do to fix this issue. I have been doing this for around 6 months, (did the PC-Dmis 101 course this last December) so I don't know all the lingo or shorthand names for some of these actions i see some people talk about. Please explain how i should be doing it, as step-by-step as you can as though you were instructing an infant, because I'm definitely missing something here.

    With my last post after the guy helping me realized I'm an incompetent newb at this, and he broke down what i needed to do step-by-step, it almost completely solved my issue that i was having.

    Thank you.

  • #2
    What deviations do you get if you best fit to all of your features (planes and all cylinders)? Can you post a drawing of a typical part?
    153010 Global Advantage w/ LSPX1H_T Analog Probe
    7107 Global Classic TP20
    2019R1 SP1 CAD++

    Comment


    • #3
      Well you had AndersI helping you last time, and those are some mighty big shoes to fill.

      Without seeing a CAD model or sketch with X, Y, Z orientation, it's very hard to visualize what you are doing. You say X is good and Y & Z are off, but we have no idea how those axes relate to the part. I know it can be hard to provide all the details needed to troubleshoot a problem when you don't know what is going wrong, but more is usually better in a situation like this. Not simply more wordy explanation, but more actual data, e.g. CAD, sketch, code, etc.
      PC-DMIS 2016.0 SP8

      Jeff

      Comment


      • #4
        Well, here's my issue. I have to shrink down the blueprint so much to get it to fit. It keeps saying i can only post images under 25mb. When i shrink it down to that size, it's literally unreadable. PM me your prefered email, and i will send it via that. Otherwise, I will try posing it from my personal computer once i get home.

        Comment


        • #5
          Don't send the whole print. It usually contains proprietary information and companies can have issues with that. We need to be able to see what points you are taking for the alignment relative to the part to help see why all 6 DOF aren't being constrained.
          PC-DMIS 2016.0 SP8

          Jeff

          Comment


          • #6
            Ok, I finally got an example that would fit. For this part, the X-axis is going vertically, and Y is going horizontally. I have 2 call outs for a measurement across both axis. One is the distance from cylinder to cylinder (plus diameter) in the brackets, and the other is a measurement of the pierce point of each cylinder and their respective plane across both axis.

            I'm almost forced to do best fit here, as I'm required to report XYZ coordinates for those two pierce points, and an intersection point for the two cylinders, labeled 1 2 and 3. The way i was taught my a co-worker to get my alignment constantly allows the part to rotate differently than the CAD model allows. I tried to describe how i usually do my alignment in the OP. Let me know if there is anything else I need to describe.
            Attached Files

            Comment


            • #7
              Tube alignment is usually bestfit of the intersection points - is that what you're finally doing? Are the end planes perpendicular to their resp. cylinder? If not, your points 1 and 3 might be off. What's your variations in the cylinder's size and direction if you run repeatedly on the same part sitting in the same place? Insufficient number of hits, or too short cylinders can be a problem...
              AndersI
              SW support - Hexagon Metrology Nordic AB

              Comment


              • bigblue97mustang
                bigblue97mustang commented
                Editing a comment
                As of right now, I'm having no better luck using the intersection points.

                Will update later.

            • #8
              I will try a program today using bestfit on the intersection points, I've not thought to try using those before. Considering all my features I'm measuring are all 3d, it should be no problem doing my alignment after probing the majority of my part.

              Typically speaking, the majority of the parts i get have end plates perpendicular to the cylinder they are attached to. The end plate, I probe out as a plane, then of course the next cylinder, then i will construct a pierce point out of the two features. For the above example, that's how i would get point 1 and 3. Because of how I'm currently building my alignments, these coordinate values are typically out of tolerance.

              Your question about variance on these cylinders, I'm honestly not sure. I know on occasion, when I examine my cylinders compared to the CAD, I notice the cylinder skewed off to one side, or sometimes has a diameter way outside what the actual part is. In the future, I will more closely inspect the values here, and see if maybe I'm catching too much ovality in the sections I'm probing. It makes sense that this is possibly where some of my issues are coming from.

              I have a question on this issue as well: For the above example, I would probe out two cylinders, and construct an intersection point between the two. If i noticed one of the cylinders had an issue with how it was constructed, and wanted to re-execute that one particular feature, I know i can. I usually just click on that feature, and use Ctrl-E. However, when i do this, I have not found a way to make my program update where the intersection point is supposed to be. Is there a way to make the program update the construction features, without re-executing the entire program? That is the only solution I have found thus far.


              Lastly, yes, I agree totally. I have noticed that for when i have parts that are just being entirely uncooperative, increasing my number of hits for the feature definitely seems to make a positive impact. I have been using that method a lot here lately. The too short of cylinders is an issue outside of my control. On occasion, i will get a part that has a cylinder only 10-15mm long. In that case, I just have to do as best I can with such a small feature. But with longer cylinders, I try to capture as much length as I can of that feature, so the constructed vector is as correct as I can make it.

              Comment


              • AndersI
                AndersI commented
                Editing a comment
                If the short cylinder is adjacent to an end plane, it might be better to use the direction from the end plane, and just measure a circle (construct line perp to the plane, through the circle). Short cylinder in the middle of the tube can be difficult - if you have many intersection points, it might be better to exclude the two from a problematic cylinder in your bestfit. U-turns can be difficult, as there is a theoretical zero length cylinder in the middle of the U.

            • #9
              We measure a lot of tubes (probe) here and the way we do it is by measuring two 3D-circles on each straight section and then create a 3D-line between them. This method is less prone to throw your cylinder measurements off due to any (large) variation in form. The ends are measured as planes. All lines are then intersected with each other and with the planes of course. Nominals for the intersection points are checked and changed accordingly (if needed). Depending on the datum structure, either 3D or 2D bestfit is used with the intersection points.
              PC-DMIS CAD++ 2o23.1

              Comment


              • bigblue97mustang
                bigblue97mustang commented
                Editing a comment
                Could you explain what you mean by measuring a 3D circle? I though circles were 2D features.

            • #10
              Measure and create a circle (manual probing).
              PC-DMIS will probably nag about you changing to the correct workplane - you can answer NO to that question.
              When the circle has been created, hover with the mouse over the "workplane" section of the circle in the edit window.
              When it becomes selected/boxed, click and choose "3D" in the pop-up menu.
              Check that the type matches that of your circle (INNER or OUTER).
              Doublecheck the diameter of the circle to see if it matches what it "should be", if not, re-measure/re-create it.
              PC-DMIS CAD++ 2o23.1

              Comment


              • #11
                As of right now, I still haven't worked out a method that works for me yet.

                Comment


                • #12
                  Still having issues with this, if anyone want's to take a swing at helping me with this.

                  Comment


                  • vpt.se
                    vpt.se commented
                    Editing a comment
                    Tried my suggestions above?

                  • bigblue97mustang
                    bigblue97mustang commented
                    Editing a comment
                    I still can't get the circles to switch to being 3D. If i could get that far, I could try doing an iterative alignment off the midpoints. As it is, I tried just using the best fit alignment, and it did not work.

                • #13
                  4541.jpg

                  for that pipe Interatvie alignment both ends cir whit 3 sample Point on the top ....
                  align interative cir cir Point,, cir,cir,,,cir
                  after alignment measure cylinder both ends
                  make center Point off circels and intersection Point off the 2 cylinders.
                  Make alignment beastfit if you want ,,,,from the 3 points
                  Last edited by Michael_Swe; 06-28-2019, 09:33 AM.

                  Comment


                  • #14
                    Just curious, when you're constructing the lines to measure the angle between them, what direction are the arrows? As point 2 is the junction where lines 2-3 and 1-2 meet, you want the constructed lines to be consistent. Meaning that if you created a line from 1 to 2, you would want to create the other line from 3 to 2. That will make the direction of the lines coincide.

                    I've found that if you were to create one line from 1 to 2, and then a line 2 to 3, the angle measured between them will yield odd results. But if the line vectors are consistent (both toward or both away), you'll get an accurate dimension.

                    Comment


                    • bigblue97mustang
                      bigblue97mustang commented
                      Editing a comment
                      I'll give that a try. I've noticed that sometimes i really have to play around with the way it get's measured. I'm normally measuring 3d, in relation to eachother, but sometimes I have to switch back to 2d and change the 'relation to' or the 'relation from' selection, then back to 3d. That usually seems to do the trick.

                      Also, normally, I don't measure the angle itself, rather report the XYZ coordinates of all 3 points against the nominal values given by my blueprint.

                      Though, because of part alignment issues, I have though about switching to measuring the length between 1-2, and 2-3, then measuring the angle created there instead of reporting the coordinates, as this would allow me to create my own alignment which wouldn't depend on using the cad's coordinate system/alignment.

                      And as of right now, I still can't get any of the above suggestions to cooperate. It's really burning me up knowing there is obviously some part of this process, that I'm just missing. If someone will tell me how to, I will post some of my programs on here (minus dimension values), and maybe someone will notice what im missing/doing wrong.

                      Thank you all for your help thus far!

                  • #15
                    I'm still looking for guidance on this if anyone else want's to try explaining this to me.

                    Comment


                    • JacobCheverie
                      JacobCheverie commented
                      Editing a comment
                      You can poste code by copying and pasting your PC-DMIS code, surrounded by [..CODE] and [../CODE] (Without the leading two dots).
                      Let's see what you're doing and why you can't choose 3D as your workplane.

                  Related Topics

                  Collapse

                  Working...
                  X