Solid and Surfacing Accuracies Not Same 1

[treddie]

Hello.

I noticed that when I built an ellipsoid feature, that if I built it as a solid, the overall dimensions were correct. But if I switched it to surface, then it gained about .001in radius error overall.

Is this a limitation of ProE, or am I doing something stupid?

My relative accuracy is set to the minimum for this model (.001in). It is curious that the accuracy and the radius error coincide, but this does not explain why the problem only occurs when featuring as surface.

Please see attached images.

Thank you very much,
treddie

 

[jeff4136]

It's hard to say without seeing the geometry or, even better, the features so specific creation details can be evaluated. Keep in mind as well; if you are examining unjoined surface edges vs. quilt (or 'solid') edges you are not looking at the same curve entities. Two Sided Edges do not always correspond exactly to the (two) One Sided Edges of the adjacent surfaces (in fact, except for analytic shapes they are rarely coincident to floating point accuracy).

A Relative Accuracy of .001 may be a bit loose but that value, by itself, is meaningless in terms of correlation to the error value. It is one term in an equation that determines an "effective" accuracy at the time a feature is generated / regenerated. The other significant term in the equation is Model Size. If you can't set the accuracy lower* it probably indicates you have a pretty large Model Size. The effective accuracy can be determined by exporting the model and reading a 'minimum resolution' value in the export log. (I never use, unless dictated, Relative Accuracy. I see no benefit and have Very few tolerance related problems working in the .01 mm, .001" Absolute Accuracy range.... And don't believe that 1 E-6 stuff you hear about ACIS. It's BS.)

* Possibly relevent; config option accuracy_lower_bound. Documentation is vague. My interpretation is it limits the lower bound of "effective" accuracy. Could be wrong.

 

[treddie]

I'm sorry, I made a mistake there. That was .0001in accuracy. So there really is no correlation between my error and the accuracy.

If I understand you correctly, I have also made an incorrect assumption about ProE in that I assumed its surfaces were like all typical single description, "one-sided" surfaces in programs like Lightwave or 3DSmax (infinitely thin). Now that I look at the ellipsoid info by right-clicking its edges, I notice two separate descriptions for an infinitely thin surface; one is the quilt, and the other is the surfaces that the quilt is broken up into. Not that the quilt is necessarily made up of surfaces, but that the surfaces are additional information along with the quilt description. Is this the two-sided thing that's going on?

I really need to find a good book that explains what ProE is doing at that level.