Modelling problem - key on shaft

[Klaksac]

Hi!

I have a problem with modeling a special key on a shaft. I need to create a key that has constant wide (6 mm) and constant deep (4 mm) everywhere on diameter. (attachment 3)

I tried two methods of modeling:
1) method - wide of the key is not constant (attachment 1)
2) method - deep of the key is not constant and radii are not tangent (attachment 2)

I will be grateful if someone help me to find the best way of modeling this key.

Klaksac

[StephenW]

If I'm not mistaken, that specific application would require you to be able to sweep a 3d solid and that is not current functionality in Creo, although Creo 5 is supposed to have a 3d helical sweep.

There have been other posts about this:

https://community.ptc.com/t5/Creo-Modeling-Questions/Surface-generated-by-toolpath/m-p/47062

https://community.ptc.com/t5/Additional-Creo-Questions/Remove-interference-material-as-a-result-of-mechansim/m-p/71732

[Kevin]

I've had some success in generating 2d sections for a sweep along a helix in Creo 3. So far I've tried a cylinder and wedge and compared the results of the surfaces generated by sweeping a 2d section and creating a boundary blend to the surfaces generated by patterning 3D cut along a helix.

[Kevin]

Here are the the results if anyone is interested.

[TomD.inPDX]

Nice!  This certainly looks better than anything I've seen from SolidWorks.

This is in Core Creo 4, 5? 

[Kevin]

The models were created in Creo 3.

[TomD.inPDX]

Oh, in that case my understanding has a big hole in it 🙂

Care to help me understand the images better?

[Kevin]

I started by creating bounding curves. Looking at the intersection of the cylinder with the shaft the cut is bounded by two helical curves created by wrapping curves onto the shaft surface. A curve was created by equation for the inner most curve which is used as the trajectory curve for a sweep. Surface contour curves are created by intersecting increments of the shaft diameter with the cylinder and helical curves are creating by wrapping curves on the shaft surface increments. The intersections of the curves created by intersecting the shaft and cylinder with the helical contour curves define points on the cut surface through which a curve through points can be created. The curves can be patterned along a helical path and a boundary blend can be created using the surface contour curves in one direction and the patterned curves in the other. For a sweep I defined points of intersection on the surface contour curves with a plane that passes through the axis of the shaft that define the sweep  profile. The inner most helix is used as the main trajectory curve and the surface contour curves are specified as additional trajectory curves. The sweep is defined with a normal direction specified along the axis. The profile curve is a spline through the intersection points with the trajectory curves. I created quarter pitch surfaces, copied them, patterned them four times, and merged them to create a full pitch surface. The full pitch surface was copied and patterned along the shaft axis, the surfaces were merged, and solidify was used to removed material for the cut.

The wedge was a little different in that the surface contour curves near the wedge inner edge pass through the edge of the intersection of the wedge with the shaft but are not tangent. As you move out the surface contour curves eventually become tangent to the intersection curves.

[TomD.inPDX]

Very nice!

My challenge for this kind of cut is something I started with nearly 30 years ago.

It is a cylinder on a spindle that has paths cut into it with a mill.

The shape was a trench that shifter every 45 degrees and created guiding channels similar to the 1st post.

The channel was just a bit narrower than the ball end mill that created the final cut .

In section, a circle with two parallel lines upward not quite to the tangent.

I've had PTC people try to tell me that a simple sweep is perfect, only to use other techniques to find it is not.

The other real test is a cylindrical tool with a flat bottom in a cylinder.

Question, you are creating the surface with a boundary blend or something?  Or are you simply testing the results.

[Kevin]

I actual made the cut using the three methods of boundary blend, sweep, and pattern of a solid cut. The surface for the cut along the entire length was created the same way for the boundary blend and sweep. I started by looking at the intersections of two horizontal cylinders with one rotated a few degrees to the other. The curve generated by intersecting surface intersection curves with the helix curves isn't planar so I used a boundary blend. The following images show the surface created (I just did the top portion of a cut segment to show the process):

Comparison of the boundary blend and pattern of cylindrical cuts:

[TomD.inPDX]

Thank you.  Yes, those last two are my validation cuts where I find obvious errors in my assumptions.

Fortunately, most of our models don't require that level of accuracy but that day is coming fast.

I think the best "solution" will be when they can create a tool that looks like a hob and actually creates a true involute gear tooth using the straight faces of the hobb.

I don't think we can do this reliably as a routine undertaking. 

It should be that simple to mimic the results of real world tools.

The best we have today is the mechanism "trace" result, but that's just a facet file.

A good test vehicle but not definitive.

[Kevin]

For the sweep the points of intersection are are the helix curves where they intersect a plane that passes through the shaft axis. The sweep uses the inner curve as the main trajectory and the other helix curves as additional trajectories. Since the cylinder is normal to the shaft I made the section normal to projection. The section provides snap points to the helix curves for the top surface of the cut and the bottom is created by placing a construction axis and making the spline curve points symmetric.

Pattern surface and merge to get the following:

Pattern merged surfaces and merge again to get:

[TomD.inPDX]

If you now follow the same idea with a helical sweep with the "Normal to Trajectory" for the References/section orientation, you will find that sweeping a circle is accurate to the 3D tool path of a ball end mill.

In order to confirm this, you might need to dial up the accuracy.

[mbonka]

Hello @Klaksac

 another similar topic is:

https://community.ptc.com/t5/Creo-Modeling-Questions/Slot-on-circle-surface/td-p/16936

Maybe it can helps

Regards

@mbonka