Single line sweep succeeds in positive Z, but fails in negative Z with error "Tool body creation failed"

[Anonymous]

Hello, 

I am modelling parametric threads. Because coils do not retain their center point positions (see link) [can this be fixed someday?]  I have taken @HughesTooling advice to model them by hand with a sweep. 

Thread:

https://forums.autodesk.com/t5/fusion-360-design-validate/center-point-of-the-coil-doenst-keep-when-...

However I believe I have hit on what feels a lot like a computation bug. Please see the attached .F3D project. 

Repro:

Start a new design. 

Create a new sketch. 

Draw a line from the origin out in the horizontal plane of any length. 

Enable 3D Sketch in the Sketch Palette. 

Draw a line in the negative Z axis (down) of any length. 

Finish the sketch. 

Change to Surface modelling. 

Create a Sweep. 

Turn off Chain Selection. 

Select the vertical line as the Path. 

Select the Horizontal line as the Profile. 

Click OK. 

Bug:

The Sweep will likely work the first time, but every subsequent edit of the step it will generate an error "Tool body creation failed". If you reload the project, it will also fail, and every subsequent step that depends on it will fail. 

However, if you change the Path to a positive Z line (up) it will succeed every time. If the example F3D I have included both lines so you can switch between them by editing the Sweep. 

After getting one successful Sweep generation with the ZUp line, the ZDown Sweep will work... once. After that, it will fail again. 

Expected: 

Sweep Path direction should not be dependent on Z direction for success. 

I am relatively new to Fusion360, and so it's entirely possible I am doing something incorrectly, but with such a simple example of failure, I an uncertain how that could be. Will need an explanation if so. 

[jhackney1972]

I corrected your sketch.  You had your path at 90 degrees to the sketch plane.  You can do this if you had placed the profile and the path sketches on the same plane but you did not.  I created the second sketch on the same plane you did just to show there is no error.  You can do the profile and path sketch on the same plane as I illustrated in the second attached file Single Plane Sketch.F3D.  Your corrected file is attached also.

[Anonymous]

You missed an important step - selecting the downward Path. 

I loaded your file, edited the Sweep by double clicking, selected the downward line for the Path, and clicked OK. It worked with no error. 

Then I double clicked the Sweep again, and without making any changes at all, clicked OK, and it failed exactly as I describe above. 

[Anonymous]

Nobody?

[TheCADWhisperer]

I did not look at the other "solutions", but I noticed that you used a 3D sketch.

I would use only 2D sketch(es) for this operation.

[Anonymous]

Hi, thanks for your reply. Did you try this personally? It makes no difference to the problem for me. 

I spent the time to reduce this to the simplest case (three lines) so it was clear it wasn't user error. I was hoping to get the attention of one of the Fusion devs, because this really feels like a low level math library error to me (also a dev). 

[Anonymous]

@jeff_strater core modelling engine bug? Something else?

[TrippyLighting]

---

@Anonymous wrote:

I was hoping to get the attention of one of the Fusion devs, because this really feels like a low level math library error to me (also a dev). 

---

Getting a little ahead of your selves arent you 😉

Somehow this is working for me. Bu now I found some strange behavior if I use a single line as the path 😕

Can you demonstrate the behavior in a screencast?

[Anonymous]

> Getting a little ahead of your selves arent you :winking_face:
Why do you say that?

[TrippyLighting]

Don't worry about it. It was irrelevant!

Please create a screencast.

I can create a functioning sweep that behaves OK using your sketch, so seeing exactly what you do might help explain this.

[TrippyLighting]

@jeff_strater you mentioned in another thread that looking at the sweep command was on your list.

I think I have been able to reproduce the behavior @Anonymous has encountered ... and then some:

[jeff_strater]

@Anonymous - yes, this is a bug.  There have been a few similar cases over the years with surface sweep:  vase-with-pattern-that-scales-with-diameter is one of those.  This might be different, but has many of the same characteristics:  simple single line path and profiles.  I created FUS-89561 for the original issue here.  Thanks for reporting it.