>>"the sweep results in self intersecting surfaces."

Usually that means that the sweep path has to tight of curvature for the given profile. You may need to post the file.

I've tried various sweep options when i hit this problem. Sometimes, i've had to segment the curve, create the surfaces i needed, then joined the surfaces. By the way when doing a sweep, how many curves are you selecting? For example, when sweeping a surface for a rectangle, are you selecting all 4 curves/lines? (Maybe that gives another error message...)

Yesssss - I have been playing this afternoon... I have also found that smaller diameters will sweep successfully. I need to dig a little into the mathematics to see what the limits of the conic constructions can be (related to the diameter of the sweep profile circle) before the bend radius becomes critical. I don't know if I remember that much about conic sections, but I have a book somewhere when I have the time.

When I sweep, I generally "shift-select" the entire continuous path. My belief is that if this works, the part is generally manufacturable. Sometime, I too have found that segmenting the path builds the swept element successfully when the entire path failed, but that may not be manufacturable...

My simplest fix has been to "crank up the rho" of the conic construction components and resolve links to reconstruct the model I am building - that process is trial and error. It would be really elegant if the error message was a dialog that asked to adjust the rho to the minimum successful value!

Thanks for the remarks, fellows!

Brian

I wish all programs did this. But they won't do the extra effort to find such values for you. I just do the "higher or lower" shuffle until a program doesn't nag me anymore, while also trying to keep a shape manufacturable.

Try lathing a closed profile. You'll only be successful if the lathe axis doesn't cross the boundary. Selfintersection occurs in sweeps to the same criteria, ie if the instantaneous radius of the sweep path places an instantaneous axis of rotation within the profile boundary, but it's not immediately obvious because of the 2D or 3D path. You can guarantee success with the profile and path of your choosing when the reference point is offset from the geometric centre of the profile, as in the picture. How much offset is pretty much obliged to be an empirical test, because it also depends on the orientation of the profile, if it's anything other than circular.

murray