hexagon logo

Feature With HUGE Tolerance

I am being asked to confirm the orientation of a round clip, similar to a hose clamp. The dimension is an angle, and the tolerance is ±45°. Since the clip can be anywhere in this 90° range, what is the best way for the CMM to 'find' it? (without taking manual hits on it of course). The polar radius of the clip location is about 35 mm if that helps.
Parents
  • Yeah Matt, if there were 'curved' vectors it'd be a piece of cake. I'll try the auto high point that john suggested if I can figure it out. Thanks guys.


    Make your own "Curved" vector, as it were. I've done it before for exactly this kind of application. Do this by setting up a vector point say, 10 degrees outside of the possible range of this part. The vector point should be searching as if it will touch the side of the clamp. Use the OnError/probemiss command to increment the alignment 5 degrees and redo the vector point. Voila, it searches in a jagged circle until it hits something. An incrementing variable with IF commands would also stop your program from looping the circle forever if the clamp is not there.

    Hope this helps!

    EDIT: Matt got this one first, it seems, but no one commented on his method. Keep it automated! Fewer inputs = fewer errors.
Reply
  • Yeah Matt, if there were 'curved' vectors it'd be a piece of cake. I'll try the auto high point that john suggested if I can figure it out. Thanks guys.


    Make your own "Curved" vector, as it were. I've done it before for exactly this kind of application. Do this by setting up a vector point say, 10 degrees outside of the possible range of this part. The vector point should be searching as if it will touch the side of the clamp. Use the OnError/probemiss command to increment the alignment 5 degrees and redo the vector point. Voila, it searches in a jagged circle until it hits something. An incrementing variable with IF commands would also stop your program from looping the circle forever if the clamp is not there.

    Hope this helps!

    EDIT: Matt got this one first, it seems, but no one commented on his method. Keep it automated! Fewer inputs = fewer errors.
Children
No Data