Critical hole (not) out of spec

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

  • Critical hole (not) out of spec

    Hi.

    I have this detail that we mostly measure in the old Mistral and we measure about 18-24 pcs a day.
    Lately there´s been a problem with this critical hole. it's measured on 3 depths (30 points each).
    I've been thinking the operators didn't clean the probe before starting the measurement because every time I looked at it, it was dirty.
    but last week I replaced the probe with a new one and calibrated as usual, set the part up and measured it.

    This happened:

    kamk2.png

    So I looked at the probe, still clean. I restarted the program without moving the part and this is what I got:


    kamk1.png

    We haven't had this problem since the Hexagon-guy was here about 6 months ago replacing a bunch of stuff on the machine,
    it's been working great.

    And also this particular hole is measured in the production line and the parts that are out of spec are removed by a robot so this is not out of spec.

    Does anyone have any idea about what's causing this? The program has been the same for years and it's been working great for a long time.

    I appreciate any ideas...


  • #2
    Have you verified the problem by using a bore micrometer or plug gauge.
    Have you verified that the in process measurements are correct.
    Last edited by UKCMM; 09-30-2019, 03:01 AM.

    Comment


    • pernilla
      pernilla commented
      Editing a comment
      As I wrote, it's been checked in the production line and that's done by a three point micrometer. It's supposed to be very accurate, and it's calibrated regularly to ensure proper measuring. I don't believe that's what's causing the CMM to give me two different results on the same part.
      Last edited by pernilla; 09-30-2019, 03:27 AM.

    • UKCMM
      UKCMM commented
      Editing a comment
      Have you verified that the bore gauge is reading correct by using a ring gauge.

      Have you checked the form of the hole a 3 point bore gauge and a CMM taking 30 hits could give very different results if the bore is out of round.
      Last edited by UKCMM; 09-30-2019, 03:38 AM.

    • pernilla
      pernilla commented
      Editing a comment
      The calibration of the in-production measurement equipment is done by our maintenance department.
      But I just realized we've got a tighter tolerance in the CMM to make operators react sooner. That might be causing some of the OOT.
      And I asked the head of quality just now about the measurement uncertainty in the Mistral, he said it's 0.005 mm plus the uncertainty of the head, TP200, magnet and probe. Then there's the temperature and other factors... I'm thinking this could be what's causing it.

  • #3
    Like UKCMM says you need to verify which is correct.

    Of two results you posted the first one is showing smaller so this would definitely point to dirt on the probe.

    A couple of suggestions...

    1) For tight limit features I usually report the form as well even if it's not called for on the print. This can serve as an early flag for a dirty probe, and if you get such a result you can look at it graphically and see where the form error is coming from (i.e. the feature might be tri-lobed in which case the issue is with the part, or theremight be one big spike which would usually signify dirt or a burr.

    2) Feature names (and dimensions and alignment ID's etc) should only use LETTERS_NUMBERS_AND_UNDERSCORES. I saw once where an otherwise perfectly good ring gauge program was reporting 0.0016 small (the feature was called Ø25.4 ring gauge). I looked at all sorts but the program was fine, changed the feature name to CIRC1 and it suddenly measure correctly.

    Automettech - Automated Metrology Technology

    Comment


    • #4
      Originally posted by NinjaBadger View Post
      Like UKCMM says you need to verify which is correct.

      Of two results you posted the first one is showing smaller so this would definitely point to dirt on the probe.

      A couple of suggestions...

      1) For tight limit features I usually report the form as well even if it's not called for on the print. This can serve as an early flag for a dirty probe, and if you get such a result you can look at it graphically and see where the form error is coming from (i.e. the feature might be tri-lobed in which case the issue is with the part, or theremight be one big spike which would usually signify dirt or a burr.

      2) Feature names (and dimensions and alignment ID's etc) should only use LETTERS_NUMBERS_AND_UNDERSCORES. I saw once where an otherwise perfectly good ring gauge program was reporting 0.0016 small (the feature was called Ø25.4 ring gauge). I looked at all sorts but the program was fine, changed the feature name to CIRC1 and it suddenly measure correctly.
      In addition I suggest the following
      Find a plug gage (OD in this case using a mic) or ring gage and match levels, points etc to see how that works, reports and as suggested turn on form or report roundness profile graphics analysis.
      On your report the middle value report is minus to the other 2 reported values, the part is hour glass shape? Not knowing process this is suggesting form error.

      Comment


      • #5
        How well is vibration isolated? On one my machines, which isn't isolated for vibration, I've seen 8 micron swings. (Company won't pay for vibration isolation).

        Comment


        • #6
          Is the probe loose?

          B&S CHAMELEON/PCDMIS CAD++ V2011

          There are no bugs, only "UNDOCUMENTED ENHANCEMENTS!"

          sigpic

          Comment


          • #7
            Thank you for all those great ideas. I'm going to change the ID's of the features because I've read about that before and when I program I always stick to that rule (although I can't understand how the ID's can affect measurement values like that). While I'm at it I'll include evaluation of form.

            Next, I measured another one today (same part) in a complete program that includes all depths (1-12 mm) and many of them were shown too small. Then I measured that same part in our other CMM (Scirocco) and none of the depths were shown out of spec. It differs between 0.000 - 0.017 between the CMM's (same program).

            Vibration has never been an issue before and we don't have any new equipment that might cause vibrations now.

            About the ring gage test, I do that twice a year with many different sizes, just to check. It usually goes well, but this problem has recently occurred so maybe it's time again..

            The probe is not loose, that would be the first thing I check when investigating measurements.

            So my next move is to call Hexagon tomorrow. Maybe this old (21 years) Mistral is just getting too old.

            Comment

            Related Topics

            Collapse

            Working...
            X