Hi Martin, so for something like that you'll probably need to do the 100 turn sweep and then piece 10 of those together rather than trying to do the 1000 turn sweep all in one go.

The way sweep works is that it starts out by trying to construct the sweep using only a few sample points and then tests the accuracy of the result. If the result is out of tolerance more sections are put in and the process repeats.

With a 1000 turn helix, you're looking quite a high level of iterative refinement needed until it will hit the 0.001 unit accuracy level.

Sweep does not have any way itself to recognize that there is a type of repeating pattern here, but you can take advantage of the repetitive nature of the pattern yourself by constructing a smaller sweep and then arraying it rather than doing one huge sweep.

But anyway because the way sweep works is by refinement of a fitted surface until a high accuracy level is obtained, it just does not work very well to do something with that kind of complexity in just one single surface sweep.

A polygon modeling program would have a much easier time of it because they generally don't care about hitting a specific accuracy level at all, so they'll do things like just move forward by some fixed increment along the curve and not even test accuracy of what that produces.

Sometimes in a case like this the NURBS modeling focus on generating accurate results can have some side effects, like in this case needing to do a smaller section that gets copied rather than doing it all in one go.

- Michael