hexagon logo
Register Sign in
  • Replies 25 replies
  • Subscribers 226 subscribers
  • Views 403 views
  • Users 0 members are here
Options
Related

Lack of confidence in CMM

QCprecis
I've been using the CMM for a about 4 years now in a job shop. We only use it when we *need* to (when we need to produce an inspection report for the customer). I'm the only CMM operator. Running 2012 MR1 CAD. I have several parts that I need to inspect and report on and have the same issues with, but I'll direct this thread toward one in particular.

I need to measure true position of some threaded holes and straight holes that will have pins pressed in them. I'm using the CMM currently to do an in-process inspection so the machinist knows he can run the parts. These are fairly complex parts with multiple operations. The threaded holes (measured before threads) are okay because they have a tolerance of .2mm to ABC. The pin hole (for 6mm pin) positions look strange and are out of spec. [TP|⌀.08|A|B|C]

I'm pretty sure my alignment is good. I leveled, rotated and set XYZ to Datums A, B and C (all planes). My part is up on 3 points and against a stop (repeatable). I'm taking 6 hits on 3 circles per hole and using the circles to create cylinders, then taking TP on the cylinders. I calibrate the probe every time I change it (manual). (I'm switching between 2 jobs today that both require separate probes.) I made sure my air bearing ways were clean, along with the probe, calibration tool and table.

I can post code or portions of the print if needed, but I'm mostly wondering what the general causes of error are in a CMM program? I've searched through many forum posts to find out what I could be missing in my program. The machinists never like when I have to inspect something because they don't trust my numbers. I try to verify my numbers with the height gauge, but if the numbers aren't exact then everyone has a theory (or 3) as to why things didn't come out right. Often someone will pick whichever they *feel* is better and go with that method (CMM or height gage), even if the other shows numbers out of tolerance.
What else can I do to be confident in my program and my TP numbers?

Any help would be greatly appreciated.
  • When they dont trust the bad dimensions - then how can they trust the good ones ?
  • I've been using the CMM for a about 4 years now in a job shop. We only use it when we *need* to (when we need to produce an inspection report for the customer). I'm the only CMM operator. Running 2012 MR1 CAD. I have several parts that I need to inspect and report on and have the same issues with, but I'll direct this thread toward one in particular.

    I need to measure true position of some threaded holes and straight holes that will have pins pressed in them. I'm using the CMM currently to do an in-process inspection so the machinist knows he can run the parts. These are fairly complex parts with multiple operations. The threaded holes (measured before threads) are okay because they have a tolerance of .2mm to ABC. The pin hole (for 6mm pin) positions look strange and are out of spec. [TP|⌀.08|A|B|C]

    I'm pretty sure my alignment is good. I leveled, rotated and set XYZ to Datums A, B and C (all planes). My part is up on 3 points and against a stop (repeatable). I'm taking 6 hits on 3 circles per hole and using the circles to create cylinders, then taking TP on the cylinders. I calibrate the probe every time I change it (manual). (I'm switching between 2 jobs today that both require separate probes.) I made sure my air bearing ways were clean, along with the probe, calibration tool and table.

    I can post code or portions of the print if needed, but I'm mostly wondering what the general causes of error are in a CMM program? I've searched through many forum posts to find out what I could be missing in my program. The machinists never like when I have to inspect something because they don't trust my numbers. I try to verify my numbers with the height gauge, but if the numbers aren't exact then everyone has a theory (or 3) as to why things didn't come out right. Often someone will pick whichever they *feel* is better and go with that method (CMM or height gage), even if the other shows numbers out of tolerance.
    What else can I do to be confident in my program and my TP numbers?

    Any help would be greatly appreciated.

    The bold section -

    Reporting Position of cylinders can be problematic at times (at least when they report OOT). You also don't say how much OOT the pin holes are, this also matters as far as providing any suggestions.

    Iin any case, here are some things to investigate:

    • Are the cylinders perpendicular to the surface they intersect?
    • Using Legacy or XactMeasure for reporting? XM has additional check boxes that will affect the reporting.
    • When reporting is the Position reported using Worst, Average, or Start/End points? Each one can produce a different result.
    • Do the linear deviations "match up" with the reported Position deviation?
    • Are points taken at each level at the same height?
    • Are separate circles taken and cylinders constructed, or is auto-cylinder used?
    • What algorithm is used to create the cylinders?
  • When they dont trust the bad dimensions - then how can they trust the good ones ?

    I've mentioned this before to them.. Everyone is fine with everything when all the numbers are green/good. I've been asked more than once to "just make it look good."

    Additionally, a CAD report (pg 1 of 3) is attached. The "pin A" locations are on the back side (Y minus) and I rotate the head manually to get the "Pin B" locations on the front. Would the manual rotation introduce error? I calibrate all the angles before I run the program. Also, the probe is a 3x50. Its been suggested that maybe the length adds error, but I used it to hit enough of the datums to make a suitable plane.

    Attached Files
  • Are the cylinders perpendicular to the surface they intersect?
    Ideally, yes. The tapped holes are drilled. The pin holes were milled (interpolated), but now he's spot drilling, drilling and using a reamer.

    Using Legacy or XactMeasure for reporting? XM has additional check boxes that will affect the reporting.
    I'm using Xact. What boxes are you referring to?

    When reporting is the Position reported using Worst, Average, or Start/End points? Each one can produce a different result.
    I noticed those on the report. I'm not sure how to change this so they are all measured correctly. Sometimes it uses the start point and sometimes the end points. I think it always does Worst.

    Do the linear deviations "match up" with the reported Position deviation?
    I checked the math of the deviations vs the TP dev. Most of the numbers match up. The ones that are off look to be a rounding difference.
    When I used the height gauge to verify I was off by about .03 in one direction.

    Are points taken at each level at the same height?
    Yes (see below)

    Are separate circles taken and cylinders constructed, or is auto-cylinder used?
    The circles are measured at the same nominal depths and then a cylinder is constructed. In a previous program for this part I used auto-cyl.

    What algorithm is used to create the cylinders?
    I'm not sure what this is referring to.. In the cyl edit window it shows Best fit and LEAST_SQR
  • check a ring gauge with the probe see what your results are.
  • check a ring gauge with the probe see what your results are.
    I measured a ring gauge at 3 different depths and they were all spot on. Only one had any deviation (.0001")
  • I would say unless there's something wrong with the program, its hard to disbelieve its results
  • if you can make a simple program, align your part and check one hole if you get the same results then its most likely a bad part. I would use legacy to do this, just to keep it simple
  • I took the part off the CMM to measure the ring gauge, so I would expect a small dev. from the first report. I ran the whole beginning of the program and deleted the other holes (manual alignment, DCC align, measure circ1, measure circ3, measure circ3, construct cyl.).

    I've never used legacy before.. Did I do this right? FCF: [TP|⌀.08|A|B|C] 


    DIM LOC1= POSITION OF CYLINDER PN1A  USE AXIS=AVERAGE  REF LENGTH=0.000  UNITS=MM ,$
GRAPH=OFF  TEXT=OFF  MULT=10.00  OUTPUT=BOTH  FIT TO DATUMS=ON  DEV PERPEN CENTERLINE=OFF  DISPLAY=DIAMETER
AX    NOMINAL       +TOL       -TOL      BONUS       MEAS        DEV     OUTTOL
X      18.050                                      18.095      0.045           
Z    -142.000                                    -141.983      0.017           
DF      6.000      0.000      0.030                 5.978     -0.022      0.000 -----#--------------
TP        RFS      0.080                 0.000      0.097      0.097      0.017 ------------------->
END OF DIMENSION LOC1


    Does the size of the pin matter in the calculation that the software does to find true position?
  • Ideally, yes. The tapped holes are drilled. The pin holes were milled (interpolated), but now he's spot drilling, drilling and using a reamer.

    So the holes are or are not perpendicular? Report each ones' Perp and find out.

    I'm using Xact. What boxes are you referring to?

    "Use Current Alignment" ON or OFF, "Perp to Centerline" ON or OFF. Each check box being ON/OFF can produce different results.

    I noticed those on the report. I'm not sure how to change this so they are all measured correctly. Sometimes it uses the start point and sometimes the end points. I think it always does Worst.

    Change this in the code - hover over "Worst", click when it becomes active, and the other selections show up. For me part function or purpose of measurement determines which I select. I do use "Average" more frequently.


    I checked the math of the deviations vs the TP dev. Most of the numbers match up. The ones that are off look to be a rounding difference.
    When I used the height gauge to verify I was off by about .03 in one direction.

    0.03mm is A LOT when the tolerance is only 0.08mm. Do the linear measurements make sense?

    Yes (see below)


    The circles are measured at the same nominal depths and then a cylinder is constructed. In a previous program for this part I used auto-cyl.


    I'm not sure what this is referring to.. In the cyl edit window it shows Best fit and LEAST_SQR

    Search HELP (Best Fit) to answer this question. The closer the Form of the cylinder gets to being perfectly round the less difference there will be between each algorithm.


    Replies in Blue
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement