hexagon logo
Register Sign in
  • Replies 37 replies
  • Subscribers 222 subscribers
  • Views 378 views
  • Users 0 members are here
Options
Related

Running someone else's programs

Smasherofprobes
Hello all, I have recently started running a cmm again after not touching one for 5 years. My manager has told me to run through all the pcdmis programs that already exist and says that none can be trusted as the previous operator was not the best and crashed several times.
What are some good habits for running someone else's programs that cant be trusted?I could rewrite them but i feel i should be able to work through what is already there. I will be using breakpoints before Tip changes in the program and running it super slow. Any other tips for someone jumping back on a machine and dealing with a mess like this Slight smile Thanks.
Parents
  • Often you may need to rewrite, BUT perhaps you may not.

    I'd suggest the following approach.

    1) First run the program offline (check all dimensions show zero deviation, if not it's likely garbage)
    2) Check program structure (mainly alignments), verify all features have correct theoretical values - depending on the part this can either be quite easy (i.e. nice square vectors 0,0,1 / 0,-1,0 etc) or not (complex free form parts)
    3) Check the reporting - if Legacy make sure dimensions are reported in correct alignment and that it's reflect the FCF. If GeoTols/Xact make sure the datums are specified correctly and measured as the correct feature type (i.e. Planes measured as planes where possible, same with holes (Cylinders and not circles).
    4) Check constructions are correct, specifically Primary/Secondary/Tertiary datums where applicable, also makes sure In/Out is specified correctly on constructed diameters.
    5) Check things like a sensible number of hits per feature, touchspeeds, prehit/retract values
    6) If all that's okay then you may have a good program, if so then run collision detection
    7) Execute carefully online



Reply
  • Often you may need to rewrite, BUT perhaps you may not.

    I'd suggest the following approach.

    1) First run the program offline (check all dimensions show zero deviation, if not it's likely garbage)
    2) Check program structure (mainly alignments), verify all features have correct theoretical values - depending on the part this can either be quite easy (i.e. nice square vectors 0,0,1 / 0,-1,0 etc) or not (complex free form parts)
    3) Check the reporting - if Legacy make sure dimensions are reported in correct alignment and that it's reflect the FCF. If GeoTols/Xact make sure the datums are specified correctly and measured as the correct feature type (i.e. Planes measured as planes where possible, same with holes (Cylinders and not circles).
    4) Check constructions are correct, specifically Primary/Secondary/Tertiary datums where applicable, also makes sure In/Out is specified correctly on constructed diameters.
    5) Check things like a sensible number of hits per feature, touchspeeds, prehit/retract values
    6) If all that's okay then you may have a good program, if so then run collision detection
    7) Execute carefully online



Children
No Data
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement