There is a caveat to that, Reinhard. If this occurs on the surface of the part being cut, yes it will fail, but if it is beyond the surface it works. So yes, it can intersect itself except under certain circumstances. I know there is a lot more to it than I can explain, but for the most part, the failure is tangency and accuracy rather than a simple rule.
Image may be NSFW.
Clik here to view.
When the edges are very close to tangent, you will fail on accuracy; when the sweep tries to share the centerline, it will likely fail; there is a difference between Section orientation - Through axis of revolution|Normal to trajectory as well. <<== this one gets many people when defining a Helical Sweeps... In springs, for instance, the wire needs to remain round and the orientation should be set to Normal to Trajectory.
In general, it is a bit of extra work to get a successful helical sweep if you want a textbook theoretical thread. Once we get around this, they seem to rarely fail; or at least, we know what to do about it in short order. But one thing I won't except is to have the software tell me what I can't do. One way or the other, I do get exactly what I want.
This will work for the original question where the sweep sketch is on the edge of the cylinder, and the equilateral triangle is normal to the revolve axis... It also controls the thread width the same as the pitch.
