Probed Points Problems

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

  • Probed Points Problems

    Hello again

    I seem to be having a problem with probing points and getting values much less than if I were to check them using other means. For example, probing points along the profile of parts that have faces perpendicular to one another and verifying them with a micrometer, I will see differences from 0.02mm to about 0.04mm between the two methods. Many of the parts I check are for molds and have square/rectangular bases with some 3-dimensional part shape on the top. I have been reading different threads about running the calibration at the same speed as the programs but it does not seem to make much difference. In the settings all 3 axis are set to 250mm/min and the touch speed is 2%. Typically I will run the calibration and the programs at 2 out of 10 dots on the jog box.

    Thank you in advance for your help.

  • #2
    So much of those differences can be related to your alignments. Are you leveled to the surfaces you are measuring from ? Are you using distance ? Perp to Axis or // to you vector of the point. point to point or point to plane ? See where this list is going ?? Also, you need to be calibrating at full speed. And hopefully someone else can chime in here. JOG BOX OVERRIDE, does this slow TOUCHSPEED also or just MOOVESPEED ? I was instructed to always calibrate at full speed so I always have. As for slowing the machine down during the program, that's what a prove out is for. If its proven why would you need to override ?

    Comment


    • #3
      My jogbox slows both move and touch.

      I recommend getting the speed you want in the code, and leave the box at full during run. You are at 20% of 2% with your box like that, so put the code to .4%. If your move speed is listed in the code at 20%, 20 of 20 is 4% so set your move speed to 4% and get the box back up to 100. If you find this too slow, adjust it with the code on something simple so you can get a speed you like that provides good results, and calibrate at that speed (move and touch).

      The FIRST run, I will slow the box down on moves, if I'm not sure what is going to happen, but I'm a pretty confident (perhaps overly from time to time lol) guy, so even this is rare and I don't crash.

      When I started out, I was camped on that box, with it turned down so far the machine barely moved on a first run.

      As said by Schlag, there are many reasons for differences. How is the part leveled on the surface plate when checking it. Are you bolting it to an angle plate, leveling it off and taking all readings from a single set up or moving it from side to side? If sides are perfectly square this can matter, sometimes a LOT.

      Are the points you are hitting for datum A the same points it sits on the surface plate (or angle plate) with? If it is really out of flat, this can introduce a lot of variation as well to a CMM.

      Maybe some code for the alignments (manual and DCC), lets look at what you are starting with.

      Comment


      • #4
        The parts are usually just places on a magnetic V-block or even just sitting on 1-2-3 blocks, and I know that that isn't ideal but it what I've learned through observation of my co-worker. Not to get to deep into things, he has chosen not to teach/help me so all I really have is the basic level 1 training from a hexagon dealer, the F9 key, and whatever I can figure out on my own. The way I method I learned through studying old programs was to make a plane (usually on the bottom of the part) level it, make a line, rotate, take a point and set up all of my origins. then i take 2 points on opposite end of the part to best fit it in both X and Y axis. Then i just use auto points around the part while in DCC. I am not sure how to post code without maxing out the available characters.

        Comment


        • #5
          Here is the code: http://pastey.org/view/5c49ec06

          This is a bad example, but if you take out that first point, I tend to go with this idea Plane, Line, Point, then best fits.
          Last edited by Jake64; 08-26-2015, 02:09 PM.

          Comment


          • #6
            I, personally, am not a fan of points in alignments, nor am I a fan of minimum number of hits in an alignment.

            I would more likely measure three planes, align Zplus to one, rotate to the second (you are still rotating to the direction of a line, it is the line extending OUT of the plane perpendicular to the plane's surface, not going sideways like the line you are measuring), and then zero on all three planes for Z, Y and X.

            This does not mean I never use lines, and it does not mean I never take a single point. But when it makes even a tiny bit of sense to do more with my alignment, I do more.

            If the alignment is bad, everything after it will be bad.

            You say a plane on the bottom, so I'm presuming it is four hits on the 1-2-3 blocks the part is sitting on, or the surface plate of the CMM the part is sitting on. That is probably ok, they are (usually) square. If it is a plane on a part, and it is bigger than half the 2X3 side of a 1-2-3 block, I would take more than 4 hits. More as it gets bigger/ has more cutter passes across it. If some lunatic cuts a 10X10 datum A with the bottom of a .25 endmill going back and fourth 40 times, I'm taking a LOT of hits.

            Even if you stick with lines and go plane-line-line, get away from 2 hit lines. Space them 20mm apart or so with a 4 hit minimum quantity (if there isn't 80mm of surface length, still do 4 hits in a line).

            Not sure why you are best fitting two points on a block for X and Y.

            If you hit two low points, you will get one answer, two high points different answers. There could be a very valid reason to this, but I don't know what it is from the info I have at the moment.

            If you make three planes with a good sampling of hits, you shouldn't need to go back and best fit two sides of a block, in my opinion.

            If it isn't a block, or there is some tooling point issue, or something similar, this might be required.

            Comment


            • #7
              I'm pretty sure it not related to your issue, but it is good practice to constrain all 6 degrees of freedom in your MAIN DCC alignment.

              Comment


              • #8
                Caemgen, I think what you just described is called "creative alignments". Don't your blue prints have Datum Targets?

                My experience is that you should have a clearly defined way to control the 6 Degrees of Freedom. Your Primary should
                be clearly defined. Your secondary should be defined clearly as either 2 points or a line, or 2 lines that create 2 points,
                and not open for guesswork.

                If your "part" is something like a cube, for example, with intended parallel/perp sides, and the print calls for a
                Secondary using 2 Datum Target points, and you instead measure the whole plane, and then rotate to that plane,
                while you might get some good numbers . . . they aren't to print. If the inspector checks the same part on the plate,
                using Print defined Datum Targets, he's goingt to get different results than you.

                The 6 Datum Targets should be clearly defined, with coordinates for each point. And in many cases, include the size of
                the Datum Target. The Secondaries might indeed be 2 points. But how does the print define these points? As in, are
                they actual "points", or a 1/4" diameter "area", or are they 2 points defined as high points, simulating 2 "bar" locator pins?

                If the CMM alignment is done to print (requiring more than one iteration) and the plate set up aligns the part the same
                exact way, then your results should be the same.

                Comment


                • #9
                  I haven't seen datum targets on a print in over a decade.

                  Mounting surfaces for manifolds or bores on hydraulic cylinders.

                  No castings or forgings, all hog-out.

                  If I did in fact have target points, then I would use them, with itterative alignment, so that anyone else in the supply chain who measured the part gets the same alignment I do. This is why I put in my post that I wasn't sure if he had a reason to be hitting that way, and there are very valid reasons to do so.

                  The parts I check I need to get enough of a surface to replicate what the datum simulator would be doing if I built the hard box to put the part in.

                  Comment


                  • #10
                    So from reading different threads and posts, I think my issue might be the speed at which I run things. Looking up the settings, the max speed of the machine is 250mm/min. The calibration settings (which were setup by my co-worker I think) are max speed 85%, touch speed 7%. I think he used to run the calibrations at 3 out of 10 dots on the jog box. The problem still occurs after I run the calibration and program at 4 out of 10 dots (didn't feel comfortable running it at 100% as some of the posts suggested), and it looked like the probe was just getting deflected by the artifact during the calibration. I still get the same results from running the program and looking at the results. I looked at more saved programs and all of the alignments use a 3-2-1 (or 4-2-1) plane line and point setup, or in the case of iterative 3 or 4 points to level and z-axis, 2 points to rotate and y-axis, and 1 point for the origin and x-axis, and he seems to get very similar results to measuring parts with micrometers which has me questioning if my speeds are correct.

                    Comment


                    • #11
                      Please note:

                      ADJUST the program speed THEN run at 100 on the jog box.

                      Do NOT just turn the knob up to 100 and go.

                      7% of 250 is too fast. That is 17.5mm/min.

                      The knob is down to 3 clicks or 30% of that. So 5.25mm/min.

                      Set the touch speed to that, 5.25mm/min (or 2%) and ONLY THEN put the knob on the jog box up to full.

                      If you are doing an itterative alignment, then 3-2-1 or 4-2-1 is perfectly fine. If you have tooling points, 3-2-1 is mandatory. If you are trying to match a datum simulator, it won't work... sometimes. Datum simulators only touch where they tough, if that is the same as your 3-2-1 you are golden.

                      With the same 250 max, I run at 85% move and 2% touch and get matching data with the manual cmm's here, surface plate, customers on Zeiss and Mitutoyo, and customers not using CMM's. The jog box, with THOSE settings, it always at full speed.

                      This is when I RUN. Realize that "Run" means the program has been proven out on 10-30 parts with no crashes or wild data points. Moving the move speed up to a high rate will cause it's own crashes if you don't pull off the part a long way. I don't. I run about .100 off the surface so I'm not wasting time with move points. They sometimes need to be adjusted.

                      If seeing the machine move that fast scares you, IN THE PROGRAM, drop the move speed to 40% or 20%, whatever floats your boat and keeps you employed. The knob on the box is for safety when debugging and proving out a program to make it run. If it is already down at 30% for normal, how do you slow it down to prove anything out?

                      Comment


                      • #12
                        Run your calibration at 10 of 10 dots, run the program the same way with the same touch speed. If you need to slow the speed down do it with a movespeed and touchspeed command. The jogbox speed dial is only to help you control the joystick or perhaps to observe a dcc move/ touch slow enough to see what you need to while programming or troubleshooting. Do not use it for calibration or running your programs.

                        .02,

                        TK
                        sigpicHave a homebrew

                        Comment


                        • #13
                          Thank you all for your input, no one has explained it to me like that, will keep you posted with my results.

                          Comment


                          • #14
                            Because your brain is not full enough and we can cram more in. Consider using ABSOLUTE speeds. This would make all your moves in MM\SEC and not based on a %. Like the post earlier said "7% of 250 is too fast. That is 17.5mm/min"
                            Wouldn't using 17.5mm\sec ( example only ) be simpler to understand ?

                            Comment


                            • #15
                              PROBLEM SOLVED!!!! For some reason the value for my calibration sphere was altered, so after changing the value to the correct diameter and double checking it by probing a gage block it seems to have solved the problem. Thank you everyone for you help, I definitely learned alot from all of your input.
                              Last edited by Jake64; 08-28-2015, 03:17 PM.

                              Comment

                              Related Topics

                              Collapse

                              Working...
                              X