hexagon logo

Boundary Primer

Does anyone have a primer on types of boundaries and determining the number of boundary crossings? PC-DMIS Help files on the subject suck. I tried searching, but I was only able to find help regarding specific problems that users were having; not anything that explains boundaries and crossings.

Thanks,

Tim
  • This isn't necessarily specific to boundary points but it does explain it a little bit. Not much but all I got.

    https://drive.google.com/file/d/0B9NvKUaV13xjZTlfT1RES3FVcGM/edit?usp=sharing
  • Thanks a bunch! I'll give it a read (oops! watch).
  • There are several options for boundary shapes but, in my experience, plane and sphere will cover almost all situations.

    Rules:
    1) Boundaries become active when the probe touches the part. This means that boundary crossings that happen as the probe approaches the part do not count.
    2) All boundaries, including control point boundaries, are active simultaneously.
    3) A crossing is registered when the ball-center crosses the boundary.
    4) Boundaries are defined based on nominal coordinates. This means that it is quite possible to miss a boundary if the part is too far out of position. Variables can be used to adjust boundary sizes and locations to accomodate highly variable parts.

    Plane Boundaries:
    1) commonly used on closed paths (circles, ellipses, etc.)
    2) Are unbounded (infinite), which means you can't miss them.

    Common application:
    You are scanning a circular path. You set the stop boundary type to "plane" and the number of crossings to "2". You must also pay attention to the planevec. This is the normal vector of the boundary plane and should be ~perpendicular to the endvec(surface vector of your last scan point). Now the scan will proceed around the circle and when it gets to 180 deg from the start point it will cross through the boundary plane for the first time. Since you have your crossings set to "2" nothing will happen, but a crossing will have been registered. The scan will proceed around the circle and when it gets to 360 deg it will cross the boundary plane again. This will be the second crossing so the scan will immediately stop.

    Sphere boundaries:
    1) commonly used as stop boundaries on linear open scans and as control point boundaries.
    2) You must set the sphere radius at least equal to the radius of your stylus + the expected amount of surface deviation in the part at that location. If you set the radius too small the stylus could miss the boundary and the scan will not behave as expected.

    Common application:
    You are scanning an open contour. You set the type to sphere and the crossings to 1. When the probe reaches the end of the scan and the ball center crosses inside of the spherical boundary the scan will stop. This isn't quite as precise as a planar boundary since the exact stopping point might vary a bit depending on exactly where on the sphere surface you cross. One advantage is that you don't need to pay attention to the vector at all. The only information required is the location of the stop point and the radius of the boundary. This can be a bit simpler to handle in parametric programs and is easier for some people to picture in their heads. Another advantage is that the boundary is very localized around a single point so you don't need to worry about accidentally crossing the boundary somewhere else in the scan where it wasn't intended. On simple shapes this isn't a big concern but on very complex or highly variable shapes it can come up.

    Hopefully that is enough to get you started.
  • ^^Excellent post DaSalo.

    Sent from my SPH-L710 using Tapatalk
  • There are several options for boundary shapes but, in my experience, plane and sphere will cover almost all situations.

    Rules:
    1) Boundaries become active when the probe touches the part. This means that boundary crossings that happen as the probe approaches the part do not count.
    2) All boundaries, including control point boundaries, are active simultaneously.
    3) A crossing is registered when the ball-center crosses the boundary.
    4) Boundaries are defined based on nominal coordinates. This means that it is quite possible to miss a boundary if the part is too far out of position. Variables can be used to adjust boundary sizes and locations to accomodate highly variable parts.

    Plane Boundaries:
    1) commonly used on closed paths (circles, ellipses, etc.)
    2) Are unbounded (infinite), which means you can't miss them.

    Common application:
    You are scanning a circular path. You set the stop boundary type to "plane" and the number of crossings to "2". You must also pay attention to the planevec. This is the normal vector of the boundary plane and should be ~perpendicular to the endvec(surface vector of your last scan point). Now the scan will proceed around the circle and when it gets to 180 deg from the start point it will cross through the boundary plane for the first time. Since you have your crossings set to "2" nothing will happen, but a crossing will have been registered. The scan will proceed around the circle and when it gets to 360 deg it will cross the boundary plane again. This will be the second crossing so the scan will immediately stop.

    Sphere boundaries:
    1) commonly used as stop boundaries on linear open scans and as control point boundaries.
    2) You must set the sphere radius at least equal to the radius of your stylus + the expected amount of surface deviation in the part at that location. If you set the radius too small the stylus could miss the boundary and the scan will not behave as expected.

    Common application:
    You are scanning an open contour. You set the type to sphere and the crossings to 1. When the probe reaches the end of the scan and the ball center crosses inside of the spherical boundary the scan will stop. This isn't quite as precise as a planar boundary since the exact stopping point might vary a bit depending on exactly where on the sphere surface you cross. One advantage is that you don't need to pay attention to the vector at all. The only information required is the location of the stop point and the radius of the boundary. This can be a bit simpler to handle in parametric programs and is easier for some people to picture in their heads. Another advantage is that the boundary is very localized around a single point so you don't need to worry about accidentally crossing the boundary somewhere else in the scan where it wasn't intended. On simple shapes this isn't a big concern but on very complex or highly variable shapes it can come up.

    Hopefully that is enough to get you started.



    Wow! Thanks a million! Maybe this should be a sticky.
  • I moved the threqad to the tip and tricks area. That seemed like a good fit for it. Great explanation DaSalo.