hexagon logo
Register Sign in
  • Replies 60 replies
  • Subscribers 222 subscribers
  • Views 1159 views
  • Users 0 members are here
Options
Related

Measuring the same part on 2 CMMs, Different Results

acgarcia
We have 2 4.5.4 SFs. I ran a part 5 times on CMM A and then ran the exact same part on CMM B. Got different and same results.

CMM A runs 2016 and CMM B runs 2017. Both use the same star probe build. Both checked the same features with the same parameter... number hits, prehit retract, etc..

One CMM seems to read a little larger than the other. CMM B measures one feature 0.002" smaller than CMM A and CMM A measure a feature 0.0006" smaller than CMM B. One feature was exact on both CMMS, exact nominal and deviation, 0.000.

Essentially I copied the program from CMM A to CMM B and just adjusted some movespeeds.

Is this something normal? Operators at night sometimes will fail parts on one cmm and then take them to the other cmm and it pass so they pass the part through. I know they CMM will not match perfectly but they should be closer than 0.002". I'm trying to build confidence in the CMMs for the operators as its only be a few years of having CMMS.

Any advice or ideas on this?
Parents
  • Hurricane went more North of my location, I'm considering myself lucky. Never lost power or any utilities. Others not so much.

    Ok, this is what I did.
    • Created Test Program in MM
    • Created Standard Force 4x20 (A0B0) (don't have a light force or 5x20)
    • Created Cal Sphere in mm units (29.0504, all 6 digits entered in menu)
    • Calibrated 25 hits at 4 levels at touchspeed of 5mm
    • Used a 117.5mm master ring for measurement with touchspeed of 5mm throughout
    • Used same probe, sphere, and ring for both machines.
    • Calibration results on both machines were 0.002
    • Created a model for autofeatures to be used during programming.
    • Manual and DCC alignment, Z+ plane, XY Orgin cylinder
    • Looped an auto circle with 23 pnts 5 times.
    RESULTS
    CMM_A
    117.513
    117.514
    117.515
    117.515
    117.514

    CMM_B
    117.501
    117.502
    117.502
    117.502
    117.502


    These results indicate a difference of about 5 tenths between the two machines when measuring a ring gauge.
    Considering the machines in question: Looks to me like this is about what you could expect. Certainly a lot better than the .002" you were reporting earlier.
    When you add the other variables that come with actual part measurement (as opposed to a controlled test on a master ring standard) I think you need to look at reducing those variables rather than assuming that one of the machines is significantly less accurate than the other.
Reply
  • Hurricane went more North of my location, I'm considering myself lucky. Never lost power or any utilities. Others not so much.

    Ok, this is what I did.
    • Created Test Program in MM
    • Created Standard Force 4x20 (A0B0) (don't have a light force or 5x20)
    • Created Cal Sphere in mm units (29.0504, all 6 digits entered in menu)
    • Calibrated 25 hits at 4 levels at touchspeed of 5mm
    • Used a 117.5mm master ring for measurement with touchspeed of 5mm throughout
    • Used same probe, sphere, and ring for both machines.
    • Calibration results on both machines were 0.002
    • Created a model for autofeatures to be used during programming.
    • Manual and DCC alignment, Z+ plane, XY Orgin cylinder
    • Looped an auto circle with 23 pnts 5 times.
    RESULTS
    CMM_A
    117.513
    117.514
    117.515
    117.515
    117.514

    CMM_B
    117.501
    117.502
    117.502
    117.502
    117.502


    These results indicate a difference of about 5 tenths between the two machines when measuring a ring gauge.
    Considering the machines in question: Looks to me like this is about what you could expect. Certainly a lot better than the .002" you were reporting earlier.
    When you add the other variables that come with actual part measurement (as opposed to a controlled test on a master ring standard) I think you need to look at reducing those variables rather than assuming that one of the machines is significantly less accurate than the other.
Children
No Data
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement