Strange sweep result

James - I agree with Pascal that Grid Snap can cause problems for precision modeling, and is best left off except when needed.

Two checks I used to find the discrepancy with the intended straight lines:

  1. Use the Angle command to measure the angle between opposite straight sections of the curves. If the lines are parallel the angle should be exactly 0.000 or 180.000. On your model the angle between the long straight sections is 0.128 degrees and between the short straight sections is 0.301 degrees.

  2. Use Distance between two points on a straight section. The exact location of the points is not important, just pick points at or near the ends of the straight sections. Look at the results for dx, dy and dz. If the line segment is exactly parallel to the x-axis then dy and dz will be exactly 0.000. Similarly if exactly parallel to the y-axis then dx and dz will exactly 0.000 and if exactly parallel to the z-axis then dx and dzy will exactly 0.000.

Do not assume that because Ortho mode is on that all lines created will be orthogonal. Ortho mode can be over ridden without direct warning. If there is an active Osnap close enough to the cursor in the active view the cursor will jump from to the Osnap rather than being orthogonal to the last point. I don’t know how Gumball behaves but it may be similar.