Sweep deviates from specified path

tony_459

I'm trying to sweep the rails of a roller coaster track along a spline curve. The rails need to pass through a series of connection plates so that the plates are centered on the rail cross section. To ensure this, I used the same spline curve to pattern the connection plates so that they fall on the rail path. Trouble is, when I sweep the rails, they deviate slightly from the specified path.



After some length, the connection plates are no longer centered on the rails as intended.  By the end of the sweep, the connection plates don't even pass through the rails at all---this despite the rails and connection plates being based on precisely the same spline. Part of that spline is shown in the figure in orange. You can see that the bottom of the upright tubes fall neatly on that line while the spline veers upward sightly but noticeably. 

The sweep profile is on a plane normal to the sweep path and located at the start of that path. The profile does not intersect the path but it doesn't seem unreasonable to expect that the sweep would try to respect the path nevertheless. Any thoughts on what could cause this?

tony_459

Another look at the sweep. It looks as if the sweep is following a polyline instead of the smooth spline it is supposed to follow. It touches the spline but only periodically. I'm very confused. Why does the sweep not follow the smooth spline I specify and which it already successfully followed when patterning the connection plates? 

tony_459

Also, the cross section seems far from uniform. The rails are supposed to be circular in cross section with a constant radius, but the result is more what I would expect if the rails were hollow and I'd mangled them severely by beating them against a wall. Is this normal for sweeps?

MBartlett21

I think that this would be due to tessellation.
Try editing the appearance of the part and setting the tessellation to very fine

tony_459

Thanks for the tip! Is the tesselation part of the appearance settings? Does this mean that the deviation is just a matter of appearance---that numerically the geometry is fine but that the appearance settings make it show as if it had been mangled and twisted?

MBartlett21

Yes, tessellation is part of the appearance settings.
Yes, I think the tessellation is making it appear different to what it should be. 

tony_459

Thanks!!! It's okay if the tesselation is coarse for display purposes to ensure that performance is good. My worry was that the geometry itself was inaccurate---which I now see is not. I have been having severe performance issues, so the coarse tesselation makes sense, though a warning about tesselation settings would have been nice! I just need to know that behind my coarse display the model is fine