Point cloud generation on a surface

[Sixmaniac]

We have some SW models with "surface revolved" elliptical surfaces (a reflector). These are to be CNC machined to a high degree of accuracy, but we need to prove that the model itself is accurate. Ideally it would be desirable to generate a point cloud on the surface and extract the numbers to do comparison with the calculated values.
There is a long and tedious way of doing this by generating intersection curves on the surface, measuring points, and manually recording these in a spreadsheet. We would like to get this done this year tho’!
Anyone come up with a smart workaround to achieve this?

 

[SBaugh]

Do you have the points coordinates or are you wanting to export it out from a model?

If you have the data try using Insert\Curve\Curve through XYZ points.

Regards,

Scott Baugh, CSWP

http://www.3dvisiontech.com

http://www.scottjbaugh.com

For all Newbie's - Welcome to the nerd herd!

faq731-376

 

[TheTick]

A programmed solution would not be difficult. A program could read in the points from a delimited data file and create the points in a sketch. Then the distance from each point to the ideal model surface can be measured.

"When everyone is thinking alike, no one is thinking very much." --Eckhard Schwarz (1930--2004)

http://www.EsoxRepublic.com

 

[Sixmaniac]

The surface was generated by a sketched ellipse, which was then revolved. Theoretically the model should be accurate. However we must prove to our customer that the surface generated by the SW software is in fact accurate - to nanometres. Even if I knew of a way to drop a point cloud onto the srface for verification, I need to extract the coordinates of all the points in a simple way rather than measuring and recording them longhand. This method would take days and be open to operator error!

I have tried saving the models as ACIS files and inserting them into AutoCAD where I can chop 'em up and place points onto corners. When it works it's fine if a little time consuming. However, for some reason not all the models will read into AutoCAD, even though they are all just configurations of the same model ???

 

[TheTick]

Hmm...

Your customer's concerns are valid. Especially with lofted surfaces, sometimes a CAD surface does deviate from its mathematical ideal. Fortunately, this is less likely with analytical surfaces like planes, and cylinders, where a modeling kernel can create exact models of those types of surfaces.

A revolved ellipse could go either way. However, I don't think you would need to check too many points to verify whether it is an exact analytical representation or an approximation. Either it's exact or it ain't.

One possibility would be to export the surface to IGES, then examine the text of the IGES file to determine the entity types. If the surface type comes through as revolved surface and the driving curves come through as ellipses, you're in.

"When everyone is thinking alike, no one is thinking very much." --Eckhard Schwarz (1930--2004)

http://www.EsoxRepublic.com

 

[Sixmaniac]

I have been able to verify from the models I could open with AutoCAD, that the surface is very accurate(2xe10-6mm). We need to get "buy off" for each model though, so I need a more robust, expedient, and reliable method of achieving my points.
Typically I need about 120 points on each surface, plus edge definition for approval.

Is it possible to export just the surface from the model as an IGES file?

 

[PlasticFantastic]

Insert>Surface>offset Then hide the solid body and you should be able to save just the surface as an igs I hope that this is what you want

 

[Sixmaniac]

We’re almost there!
I found this query by “APPENG” (

http://www.eng-tips.com/viewthread.cfm?qid=55648

) from 2003.
Using his Macro, we can extract the points to a spreadsheet (thanks Jon).
Now if we can come up with a speedy way of generating the points, I think we have it more or less cracked.
Thanks to all for the suggestions so far. Any more ideas are still very welcome.
Paul

 

[overkill04]

Hi

Wondering if there is a way to do the reverse here. That is to take a table of points and generate a surface?

Will that surface still be linked to the table of points?

Thanks