Match Surface edge and tolerances

[0nuno0]

Hi,

I am trying to match the edges of 2 untrimmed surfaces, to a simple trimmed slab. But i m getting a bit confused.
When i apply the command Match Edge, and ask for G2....It does give me 0.0000 reading from the G2 deviation, on the options of that command, although weirdly it gives me 0,003 for G0 and something for G1 too. To begin with, i feel this is strange. i have G2 without theoretically having G0?...Ugh.
Now even more confusing, if i go to analysys and check the edge to edge deviation, it yealds G2 ok, as i expected, but gives me like 0,6 for G0 and some other value for G1. Why?
If i zoom close on the conctact edges of those surfaces, they look weird, and are not lying on the trimmed edge. I went to preferences and set the tesselation to fine, and i still see artifacts, which i think are actually real and not artifacts, as the G0 shows.

ANother thing that i dont really understand, is why, when matching edges, is it so hard for the software to give me G0=0.0000.... I suppose it should be trivial, even with not so good surfaces.

I send here the file, if anyone can please elucidate me. I m doing something wrong, i think.
P.S. curves are hidden on standard curve layers, in case u wanna use them, or see what i ve done in steps.

Thank you in advance

 

[0nuno0]

Sorry, forgot to mention, it s NX 7.5 with update.

 

[Toost]

I am on unsteady ground here, but, it seems that you are trying to apply Class A techniques on data which isn't class A.
The top slab is degree 3, single patch. The ruled surface is degree 3 but has 93 patches ( non-A), then you try match a thru curve mesh which starts out degree 3x3 but 8x17 patch, and the matching feature changes that to degree 5x5 , 8x17 patch. Looking at feature 23-24:
You are matching a degree 3 edge with a degree 5 surface,
Where the degree 3 math gives one shape and the degree 5 a slightly different shape, = G0 (position)then has a small deviation. G1 is the tangency, I assume that this is due to the G0 deviation. G2 is the curvature, which in Match edge 23 should be 0 ( straight line) which it is.
Note that in a G1, G2 , G3 condition the edges not needs to be connected to each other, they can still be matched.

Divide and clean the "rectangular" splines that are the curves for the "ruled surface(9)", make each 8 splines where the section you have been trying to match is a degree 3 single segment and your matching will be a lot cleaner.

 

[0nuno0]

Thanks for your help Toost.
Yes I am used to model with Class-A kind of techniques (i try), or in other words, i m modelling this with the same approach i use in Alias. Different programs, but since NX has nice surface tools, i hope i can do surfacing with it, with a surfacing thought behind it.

The slab with many patches, is just a draft helper surface from the extracted trimmed edge of the dome shape, so that it could be mirrored after. You can forget that anyway, it s not too important here. And of course u can match different degree surfaces in between them... It s very common u see a degree 7 surface surrounded by degree 5 surfaces... I could even trim convert that big helper surface and make it single span for the blend, but not really important. But of course, nothing like working with simple and clean surfaces, so you are right. Actually, is not the CVs that matter for continuity in the last instance, it s the shape.... For example, mathematically, and even in practice, u can have a degree 2 curve curvature continuous with a degree 5 curve. You only have tangent CV on a degree 2 curve, but it can be G2 to another curve. Many people get bogged down with this statement, but that s actually the reality. Maths.
My problem here is not the theory.... it is actually getting to know how NX likes to work.... I am very used to alias, and i can see that NX doesnt really think the same way... It requires different approaches, which is fair enough.
What i find really weird, is the descripancy between the match edge Gn reading, and then the analysis reading... they dont match.
Zoom closer on the edges where they intersect.... That doesnt seem to look right. I dont know why is it so hard just to align 2 surfaces with G2.
Curves in NX dont really help a lot.... They are not as free as i wanted them to be.... I cant align a curve to a surface with a command apparently like a surface match edge.... i have to tweak the end point of the curve, and is not that flexible how it works. I think my curve layout is what is messing everything. i aligned the curves to a projected line on the surfaces.... BUt this can give some gaps when surface is a trimmed surface.
So, if this part of the surfacing is not done quite well, then to fill the corner between the 2 surfaces will be impossible to G2, and spans galore.

This is another thing in which i am findind Nx different too.... NX likes spans.... It s hard to get 1 span surfaces G2 on all 4 sides.... It wont build them through the advanced options.... It ll always find the spans and adds them for it to work. And then if u want the software to build it explicitly, say with 1 span only, so that u could after tweak and massage hulls with X-form, it simply will not build the surface.... so i m stuck here. Spans in Alias, are like a forbidden thing, if you know how to model to a high standard, but in NX it seems to be different. Actually every span is a discontinuity point. That s the theory, and it is correct. I dont mind this span thing, i just wanted it to work with any spans... It s for visualization purposes, not manufacturing, so it s all good if it looks good. Plus the tolerance in NX seems to be pretty low....0.025, which is nothing compared to 0.001 i work with in alias, and V5 catia has same spec. I never touch the NX tolerance... i work with defaults, it s ok.

So, i am having some problems in dealing with this part. i ll have to get acquainted to NX philosophy, and learn how curves work. They are a bit unflexible. There is a way of getting a curve in NX G2 to a surface (perpendicular to edge), but i forgot how to do it. It s driving me crazy. The constrain thingy of the curve editing is hard to grasp....I m just not used to it. Anyway, being used to catia, i much prefer NX... Blows it away in terms of being more flexible and user friendly. I like NX, that s why i wont give up on it. Modern and sophisticated, for sure.

Anyone who can shed some light and help on this simple issue is very welcome. Looking forward to any help and ideas you can give me.

Thank you .

 

[Toost]

I think i found one of your issues.
You have matched Match edge(24) to the face and not the edge.
Then the matched edge is within the tolerance from the face but not necessary within the tolerance to the edge. If you match to a face,you can drag the edge on that face. ( probably end up within the expected deviation.) Edit Match Edge (24) and switch to Movement = Normal, note the arrowhead on the matched edge, try drag that.
Have a look at the attached part, edit/ unsuppress Match edge (34) and (37). Somehow, but i don't know how to make it happen consistently, one should be able to drag the matched edge on the "reference".