Create a datum curve based on a csys with a cylindrical coordinate system; use the following equations:
turns = 10
ratio = 25
main_radius = 2
small_radius = .25
theta = t*360*turns
length = 10
z = t*length+sin(t*turns*360*ratio)*small_radius
r = main_radius+cos(t*turns*360*ratio)*small_radius
The ratio is the number of small turns to each big turn. None of the values needs to be an integer.
You should (may) be able to use this as the trajectory for a variable section sweep. If it fails**, try changing the accuracy of the model. If it is not required to be solid, e.g.if this is representing a wire, then save the headache and skip the sweep.
** Under 2000i I tried to create a movie where the turns number would increment using a trail/training file. The sweep feature would fail at seemingly random values. 10.1 turns OK. 10.2, fail. 10.3, OK. 10.4, OK. 10.5, OK. 10.6, fail. That sort of thing. Changes to the accuracy changed which increments it failed on, but it always failed somewhere. Any chance PTC fixed the underlying code?
Attached is file PRT file from WF5 as an example....
features are suppressed to save on file size. resume features
one could now go to the part and add parameters to drive this generic geometry. Pro/E is quite powerful.