Revisited: Sudden Scanning Profile/Cirularity Inaccuracy - Controller Board Issue...?

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

  • Revisited: Sudden Scanning Profile/Cirularity Inaccuracy - Controller Board Issue...?

    Hello folks,

    I decided to start a new topic for this, but the original can be found here (in case I forget any relevant details, or you'd like to watch me whine about my apparent need to delete cookies)

    http://www.pcdmisforum.com/forum/pc-...p600m-2013-mr1

    Long and short of it is as follows:
    • Had sudden trouble getting any probes to calibrate well at A0B0 (Global Advantage 9-12-8, SP600M, PC-DMIS 2013 MR1 SP3)
    • 8x50mm tip that typically gets a StdDev of ~0.00002" was now getting ~0.0004"
    • Cleaned tip, sphere, probe mating balls and re-tried - still no good.
    • Did Low Matrix Calibration, figuring another unreported crash occurred, and then A0B0 - same results.
    • Swapped out the SP600M for another, did LM and Cal'ed A0B0 - better but still not good.
    • Developed a check program to measure a XX ring gage with touch point circles and cylinders, align to it (Leveled/centered to the Auto Cylinder) then measure it with Auto Circle with Adaptive Scan Strategy, and all Tolerances set to 0.0001". Reported the XY location and Diameter of the TTP and scanned circles, as well as the Profile of the Auto Cylinder and scanned circle - all XY and D dimensions look reasonable (within a few tenths), TTP Cylinder profile was ~0.0005" or better, and scanned circle was ~3 - 4 thou
    • Notice that the Graphic Analysis seems to show all of the deviation as a balanced zig-zag at the 'top and bottom' (±Y), while the 'sides' (±X) looked fine.
    • Ran the same tests in the X+ (A90B-90) and Y- (A90B180) orientation, expecting to see the deviated region rotating (X+ would be OK Top and Bottom, with funky Sides) - each test looked roughly the same.
    • Did all sorts of PM cleaning (Ways and Scales, douching out the outboard Y bearing with alcohol, etc),and repeated from Low Matrix through scan tests.
    • Next Swapped out the PH10MQ, keeping the swapped SP600M, squared up, LM and A0B0 Cal, and re-ran all tests - similar results
    • Also tried rotating the ring 90° (all tests up to now were rotated similarly) - same...
    • Sales Rep came out to try and help, brought another SP600M to try, did LM/A0B0) Cal - worse results.
    • Next, Swapped out the Probe Controller, Recalibrated A0B0, and did the Z+ test - similar results.
    • Next, Swapped out the Wrist Controller, Recalibrated A0B0, and did the Z+ test - similar results.
    • Sales Rep recalls a customer with a similar issue (sudden poor Scanned Circularity compared to TTP), and that the solution in the end after exhausting all of the similar testing I did, was to replace a board in the controller, though he wasn't sure which one it was. (also, one board (also not sure which) was replaced on this CMM a few years ago, when it suddenly wouldn't let you get a 'Machine Start' - button just kept blinking after startup.)


    Sales Rep is supposed to be passing our findings along to Hexagon. Have any of you fine folks encountered a similar problem? I'll post a report and/or some Scan Graphics in a moment to provide a thousand more words...



  • #2
    TTP Cylinder Profile:
    https://image.ibb.co/hmX7U5/DMIS_05_Scan_Test_001.png

    Adaptive Scan Circle Profile:
    https://image.ibb.co/bOwO2Q/DMIS_05_Scan_Test_002.png

    Comment


    • #3
      Adaptive Scan in X+:
      https://image.ibb.co/hsaNwk/DMIS_05_Scan_Test_003.png

      Adaptive Scan in Y-:
      https://image.ibb.co/mtoaGk/DMIS_05_Scan_Test_004.png
      Last edited by edrake; 08-24-2017, 09:51 AM. Reason: Accidentally Typed Y+

      Comment


      • #4
        if you rotate the ring 90°, do those waves of variation follow the gage? Yes a XX gage is supposed to be good to .00003", but maybe you really do have some wear present on your ring? Looks to be scratches along the ring's cylinder wall to me.

        Comment


        • louisd
          louisd commented
          Editing a comment
          whoops, missed reading the line "•Also tried rotating the ring 90° (all tests up to now were rotated similarly) - same...".
          basically the erratic wavy deviation vector is always related to z axis quill? or can you get the wavy variation at t1a0b0 as well?

          Have you tried t1a0b90? Maybe your probe stylus or extension has wear/damage, rotating PH10 about B 90 degrees should make the variation rotate 90 degrees if it's stylus or module related.

      • #5
        Yeah, not likely a board problem with the controller (LOL). This usually means someone ran out of ideas ...

        So the problem is in the Y axis of the machine. If you run a scan test on the left side of the machine and compare this to data from the right side do you see a significant difference? If you do I can suggest two things you should look at.

        Was there any recent changes to the machine (machine calibration, collision, new software, ...). So it is not a probe problem (the replacement SP600 should be a good indicator of a probe problem). Why it was worse it a bit of a mystery. If you did update your software recently then go back to something you know works. If the replacement probe was worse then, maybe, the replacement probe was bad which would cloud the issue at best.

        Comment


        • #6
          Did you check all parameters (F10 / option probe) and setting editor scan parameters ?
          Here an interesting old doc to down load, just for information...
          ftp://ftp.wilcoxassoc.com/docs/How%2...Docs/Sp600.zip
          or look at ftp://ftp.wilcoxassoc.com/docs/How%20To%20Docs/ and download Sp600.zip...

          Comment


          • #7
            These errors seem like they begin when the CMM starts to transition in to the Y axis. I'd suspect a drive belt or drive motor bearing is the culprit. A 3 pass LDA on that axis will give you a better idea of what's going on. Maybe a stethoscope on the bearings to give you a better idea if there is something starting to give up. It could be a hundred different things but without seeing the machine myself I'd have to put my money on a failing bearing.
            Systems Integrator
            Hexagon Manufacturing Intelligence

            Comment


            • #8
              Thank you all for your input, I know I threw a lot out there to digest.

              I'm not sure if I was clear, but these tests have been done normal to each axis. The Z+ test was using A0B0, the X+ test using A90B-90, and the Y- test using A90B180. Like some of you have suggested, I was expecting it to be a Y axis issue, and expected the deviations to rotate depending on the workplane. They did not. The Y- test effectively locked the Y axis out of the equation, outside of whatever minor squareness issues there may be, and still shows a very similar deviation pattern. But to that end, these tests were all done with a very repeatable fixture setup on a Rayco plate, with a raised vee block for the X+ and Y- tests.

              Comment


              • #9
                I initially suspected the Z axis, because of the strange deviations I was seeing - it looked like the sun from above (was coming in about 0.014" oversized with a 30X multiplier - not shown above) but the arrows pointed 'out and down' from a side view (and the hit targets were all over the place in Z...?). But that went away when I more precisely leveled to the ring's cylinder - size and location were decent, but profile was crap. At that point I suspected the Y axis, but I think my testing rules that out, doesn't it? It also seems unlikely that each axis would exhibit the same issue independently of each other (if each axis had the same issue, I should be seeing the same jagged profile all the way around). And again, this is very 'all of a sudden'. I have swapped out everything I could (including three different SP600M's, and two PH10MQ's, and controllers for each), cleaned all ways and scales very well, and get very similar results no matter which axis I scan in.

                Comment


                • #10
                  Also worth noting is the issue that got me to this point, which was not being able to get a decent A0B0 calibration with any tips - even the 8x50mm tip that sits in a box until I take it out for A Low Matrix Calibration, or some other testing, and typically comes in with a StdDev of ~0.00002".

                  OH! That reminds me of another bit of strangeness. While the StdDev's I was getting at one point seemed poor for the 8x50mm tip (0.0002") they would be 'acceptable' for normal use (tolerance trigger is set to 0.0003") but the PrbRdv values were pretty erratic, fluctuating by several tenths. But then that suddenly stopped, with the same set of swappery that was just erratic, they from then on gave PrbRdv's of 0.00000", but still with crappy StdDev's...

                  Comment


                  • #11
                    Originally posted by edrake View Post
                    I initially suspected the Z axis, because of the strange deviations I was seeing - it looked like the sun from above (was coming in about 0.014" oversized with a 30X multiplier - not shown above) but the arrows pointed 'out and down' from a side view (and the hit targets were all over the place in Z...?). But that went away when I more precisely leveled to the ring's cylinder - size and location were decent, but profile was crap. At that point I suspected the Y axis, but I think my testing rules that out, doesn't it? It also seems unlikely that each axis would exhibit the same issue independently of each other (if each axis had the same issue, I should be seeing the same jagged profile all the way around). And again, this is very 'all of a sudden'. I have swapped out everything I could (including three different SP600M's, and two PH10MQ's, and controllers for each), cleaned all ways and scales very well, and get very similar results no matter which axis I scan in.
                    I'm still going to think it's mechanical, could be a failing air bearing? Clogged from poor quality air. It could still be a board, I've seen the UMP controller have failing PWM boards exhibit similar variation in scans. If I were servicing this machine, I'd probably put a laser on it and check the linear accuracy for repeat issues or any type of cyclical static. You should be able to narrow down or eliminate any type of hardware issues at that point. For the most part I think you're at the limits of what you can do with the equipment you have on hand.

                    Systems Integrator
                    Hexagon Manufacturing Intelligence

                    Comment


                    • #12
                      We do have a Hexagonian coming in to calibrate an almost identical CMM (the one we just got, that I have been using for my 'swaptronics' testing) on Monday. They will also be looking at the one in question. Apparently we are ordering the board to be delivered for him to test out.

                      I thank you all for your input, and for dealing with my long-winded nature! I will be sure to post an update after Monday's visit.

                      Comment


                      • #13
                        Originally posted by Peter Fuller View Post

                        I'm still going to think it's mechanical, could be a failing air bearing?
                        +1 for this point.
                        I would open the gussets (not sure of this word - which looks like to an accordion) and check if air bearings can turn easily around their axis.
                        If not, the air bearing is going bad...

                        Comment


                        • #14
                          edrake , did you run ring gage untouched at A0B0 then again with probe rotated 90 degrees at A0B90? This will help dictate if the components below the head are suspect.
                          Measuring in X and Y workplanes (Ran the same tests in the X+ (A90B-90) and Y- (A90B180) orientation) produced data still aligned with the same orientation of head and module.
                          If you rotate head 90 degrees about A0, without touching ring gage, the pattern could rotate 90 degrees, showing the error coming from head/body/module, and not drives/scales/motors.
                          I will also confirm variation isn't caused by your ring gage fixturing/setup.

                          Comment


                          • edrake
                            edrake commented
                            Editing a comment
                            No, I didn't try it at A0B90, but thanks, that is a good idea and I'll give it a stab (I think when I first read that suggestion, I thought it read A90B90, or something, which was close to what I was doing). I didn't see why at first, but with what I've seen thus far, I suspect that will be the one way I will get the pattern to rotate. Update to come...

                        • #15
                          OK, so I got around to trying louisd's suggestion, using A0B0, and A0B-90 (figured I'd use B-90 since my X+ Scans were using A90B-90...) Here's what that got me.

                          Adaptive Scan, Z+, A0B0
                          https://image.ibb.co/bLOdP5/DMIS_05_...st_ZPA0_B0.png

                          Adaptive Scan, Z+, A0B-90
                          https://image.ibb.co/cgOpWk/DMIS_05_..._ZPA0_BM90.png

                          While this 'looks like' the probe is the problem, as the deviation pattern follows the probe orientation (can't really duplicate this in X+ or Y- without a star probe), I have had three different SP600m's and two PH10MQ's (and two interfaces for each) swapped in and out with the same results (although I didn't and am not about to try this A0B-90 test with the other ones, now that everything is swapped back...).

                          I know 'Board Swapping' is probably usually a last grasp at some straws when ideas are scarce. We recently were talked into swapping a logic board on another CMM with A Tesa/Lietz Wrist/Probe that was losing it's mind in the middle of scans and wandering off in the wilderness (Free-form, defined path scans, no control point/crossings issues). Turned out swapping the Wrist was the solution, as the next Hexatech proved (the CMM that we swapped the wrist to suddenly exhibited the same symptoms that the problem CMM was now free of). The way this is presenting though, especially the out of nowhere, all of a sudden nature of it makes me think it may actually be the solution this time, but then those straws are just out of reach...

                          Comment

                          widgetinstance 190 (Related Topics) skipped due to lack of content & hide_module_if_empty option.
                          Working...
                          X