Skip to main content
T
TomD.inPDX17-Peridot
17-Peridot
July 24, 2014
Question

Another equation curve approximation error

  • July 24, 2014
  • 2 replies
  • 8986 views

I have been fighting some accuracy issues related to equation curves.

 

I was trying to use an even simpler version of a closed end spring when I found that for some reason, a curve from equation doesn't translate well to cylindrical sweep trajectories.

 

You can try this for yourself, but it really is a very distinct problem with using this particular curve type (true cosine segment).

 

Simple part; make a wrap of an equation curve around a cylinder...

 

sweep_artifact-a.PNG

Looks nice, smooth, and properly wrapped. the wrap, of course, is a projection into the wrap-sketch of the original curve.

But, it is free of artifacts!

 

Next we sweep a sketch along its trajectory (<= this becomes important in a minute).

 

Notice the little blue dots along the trajectory (origin)... these are the quilt's "edges".

 

sweep_artifact-b.PNG

 

Close examination of the Origin shows artifacts... precision settings doesn't change this by much.

 

sweep_artifact-c.PNG

 

...which results in:

 

sweep_artifact-d.PNG

 

Now you would think this is a problem for all sweeps like this... but -NO-!

I wrapped a line and an arc along the same cylinder. No issues!

 

Linear:

sweep_artifact-e.PNG

 

exponential (arc)

 

sweep_artifact-f.PNG

 

No artifacts anywhere!

 

So why not create a cylindrical curve to start with?

 

Yea, I thought of that...

 

sweep_artifact-g.PNG


This thread is inactive and closed by the PTC Community Management Team. If you would like to provide a reply and re-open this thread, please notify the moderator and reference the thread. You may also use "Start a topic" button to ask a new question. Please be sure to include what version of the PTC product you are using so another community member knowledgeable about your version may be able to assist.

2 replies

T
TomD.inPDX17-PeridotAuthor
17-Peridot
July 24, 2014

Support case submitted. Stay tuned.

I just wanted to make sure people knew about approximations happening with certain types of geometry.

T
TomD.inPDX17-PeridotAuthor
17-Peridot
July 24, 2014

An SPR is being filed.

Turns out the Wrap feature is also corrupt. A curvature analysis makes it pretty clear that the ends are not merging correctly with the continuation around the cylinder.

The same curvature analysis with the arc wrapped around the cylinder showed no artifacts.

I made the precision absolute at 0.00005 (doesn't change the results much).

sweep_artifact-h.PNG

...and with the Wrapped curve from equation...

sweep_artifact-i.PNG

Lesson learned... don't wrap equation curves around cylinders! Hopefully PTC will fix this in Creo 2.0 and 3.0.

This has some pretty serious implications in areas where equations are being used for turbine features wrapped onto cylinders or cones.

However, even with a single iteration, there are problems that we normally live with. This is back to default accuracy (Relative 0.0012).

This is the curve itself... near perfect:

sweep_artifact-j.PNG

And then the surface created to create the edge needed for the normal control for the spring below...

sweep_artifact-k.PNG

Although for most of us, these small errors are less that catastrophic.

Awareness of the fact that cumulative error will have consequences is something to take note of.

On another note, I would be remiss by not providing a solution.

Attached is a Creo 2.0 spring that uses the cylindrical curve from equation. The Pro|WorkAround^TM to the twisting problem is in a surface sweep.

Sweep has a normal to surface option that is not there when the sweep is not generated from a surface... such as the equation curve.

A 4 family table versions of the accurate version is attached...

Spring_w-equation_curve.png

Driving parameters

Spring_w-equation_curve_ref.png

D
dschenken1-Visitor
1-Visitor
July 24, 2014

Never did use Wrap to make springs. It's too difficult to control what you get. Plus it ruins the extents for the part.

http://communities.ptc.com/docs/DOC-3063

To prevent the weird twist seen in Sweep-artifact-g.png, create a second curve with an constant increment added to the radius and it can be used for X-direction control in a variable section sweep.

Patriot_1776
Patriot_177622-Sapphire II
22-Sapphire II
July 31, 2014

Interesting, either I've never had those issues or never noticed them.

I HAVE noticed an issue with features crossing one of the 3 primary datum planes (way back in '96 on v15), and solved it by creating a new set of planes at 45deg around the Y axis. Try creating the cylinder or cylindrical curve by equation so that ALL geometry is in positive X, Y, and Z space. I feel strongly that Pro/E has rounding errors when it approaches "0" in any plane like that, especially where it wants to split the cylindrical surface.

Edit: I just changed the cross section of my wrapped spring from circular to square, and did not see any of the twisting you're seeing Antonious. Weird......

T
TomD.inPDX17-PeridotAuthor
17-Peridot
July 31, 2014

Interesting observation.

Terms & ConditionsCookie settingsAccessibility statement