Unexpected sweep error when cutting a flute embouchure

[r.cameronstahl]

I'm having some difficulty performing what seems like a pretty simple operation. Flute embouchures start from an ellipse and then have an undercut corresponding to different angles at the N, S, E, and W points. I've marked those out in cross-section and extended them to a constant depth in the bore, which gives the footprint of the shape that needs to be subtracted.

I thought the most straightforward way to do this would be to draw a 'blade' and perform a sweep, using the curvature path of the embouchure and the footprint as a guide rail. This ideally would blend the different angles together in one smooth shape.

However, I keep getting an "illegal surface" error when I try to do this.

 I've tried pretty much every combination of settings the sweep+guide rail function offers, and none are working. I know the embouchure path itself is valid because it WILL sweep a basic shape around it when I tested it.

 

I'm sure plenty of you are thinking "why not use a loft" but here's the rub - a loft can only be done between two planar shapes. If I just projected the embouchure curve, I would get excess material removed from the E and W sides of the actual hole, because the angles no longer correspond to the same reference points. This is actually what happened to my first batch of embouchure inserts.

I also tried extending the angle lines to a known distance above the embouchure, and lofting to an ellipse from THAT projected shape, but I'm then getting portions of the embouchure that are not cut at all, at the midpoints between each cardinal direction.

I think I could fix this by using a fit-point spline on the upper projection instead of an ellipse, but I have NO idea how to mathematically determine the handle placements, nor how to compute the geometry for a shape that is correct at every conic section between the N/S and E/W depths of the embouchure.

Any ideas how to coax Sweep in to behaving?

Thanks,

Cameron

 

[jhackney1972]

I do not have your model so I just guessed at the shapes so what I did may not be what you need.  Model is attached I used in the Screencast.

[TrippyLighting]

I am not entirely sure what surface/geometry you envision for this workflow to create but it looks needlessly complicated.

Why not loft between the two profiles, extend the surface if needed and just use that surface to split away from the body what's not needed ?

[r.cameronstahl]

---

@TrippyLighting wrote:

I am not entirely sure what surface/geometry you envision

---

That's probably why it seems overly-complicated... All kidding aside, this is a high-precision task where even a few thou off can wreck the tone, response, or both.

A flute is essentially just a hollow cylinder, and the embouchure starts as an oval carved into the upper surface. However, each cardinal direction of that oval has an undercut angle - rather than going straight down, each wall slopes away from the original footprint of the hole.

The reason I'd like to use a sweep if possible is because I can set the undercut angles for each cardinal direction, and then draw a fit point spline connecting all of them, which smoothly transitions one angle into the adjacent ones.

Imagine a hot wire anchored at two points, the first point following the oval of the top surface, and the second following the undercut profile, moving outward or inward from the original top-surface silhouette.

The reason a loft won't work is because the top surface oval shape is curved with the body of the flute; trying to loft to a flat projection of it tangent to the body will actually change the angle on two sides which will either be above or below the plane being lofted to.

Lofting to this projection will both change the undercut angle and widen the top surface oval. Can you elaborate on where extending the surface would be utilized in this situation?

[r.cameronstahl]

Thanks for giving this a try, John! It works for two centered ellipses that are essentially concentric, but where I'm getting the error is for the irregular shape corresponding to the undercuts down below. See how the lower profile gets closer or further away from the top surface oval?

I've attached a file with some of the different things I've described. The closest solution is to extend the angle measurements until they reach the same plane as the initial ellipse profile, and then draw a new oval shape with splines to preserve the undercut angle at the East and West sides.

However, the transition areas between the angles have areas where the cut did not get made all the way to the top oval profile.

[r.cameronstahl]

The reason I think this might be a bug is because it will actually display about half of the path if I manually set the extent to 0.5. Like, it KNOWS what the geometry should be but can't seal the deal. If I try to do it a half at a time, it still says it can't complete the sweep, even though it looks like it's going to.

Instead of the illegal surface error during the operation, I get an after-the-fact warning:

[r.cameronstahl]

Interestingly, I'm also getting an incorrect projection of the maximum width of the ellipse, where the projection line stops short of the left side. What the heck?!

[TrippyLighting · Accepted solution]

---

@r.cameronstahl wrote:

---

@TrippyLighting wrote:

I am not entirely sure what surface/geometry you envision

---

That's probably why it seems overly-complicated...

---

No, it IS complicated 😉

If you create a surface loft (with the loft in the Surface tab) between the upper edge on the flute body and the "angle path" that will create a surface body.

You can then use the Split body tool and use that surface to split away the unneeded material from the flute body.

[r.cameronstahl]

That worked a treat! Thank you so much for your help - I really have not delved into what the surface tool can do.

With that missing piece solved, I've completed my "Dial-an-Embouchure" system where I can now enter key dimensions of any historical or experimental cut I want, and the model builds itself! This is awesome!