Parametric and constant surface (section+path)

[davidemin]

Hi experts,
I’m getting crazy these days trying to solve this “little” problem, hope someone could help me.
I just want to extrude a section along a path with constant dimension (the section dimension), simple isn’t it..? Obviously I’d like to have a simple-to-edit surface, the best would be one section and one path.
I’m using:

- Alias design 2010
- Construction options: Pro-engineer

I made this surface in hundreds of different ways with different combinations of curves, lines or blend curves, but the distance is never constant (I’m using curve to curve deviation tool or surface to curve tool to check).
Some (bad)attempts, please see the attached image:

1-section +path > extrude
Section: line or strait curve 3° degree
Path: curve 3° degree or blend 5°

2- section +path > mono rail
Section: line or strait curve 3° degree
Path: curve 3° degree or blend 5°

3-section +offset section > skin
Path: curve 3° degree or blend 5°

Using the offset tool and unchecking Explicit control I almost reach the desired surface, but it’s very heavy (tons of isoparm).
I know Alias can do this!
Thank you in advance to all.

[ravenzep]

With so many tries, why didnt you use the most obvious method?

SQUAREEEEEEEEEEEEEEEEEEEE !


Bye

[davidemin]

Ravenzep thank’s for your replay,

But why Square?

I really don’t think is a good solution, first because you need 4 curves instead of 2, and second because one of these 4 curve will be an offset right? But the offset tool, as shown in the image on my previous post, it is definitely not precise.
I’ve also tried with flange/draft, it seems the “most” precise solution (max deviation 1mm found ), the only thing the surfaces appear full of isoparm…

Any other suggestion?

[ravenzep]

Check your construction options.....Make them tighter.
Since nurbs are rectangular patches, it makes sense to work with 4 sides.

[davidemin]

>Check your construction options.....Make them tighter.

Can you be more precise, please? I use Pro-engineer construction option, do you think is enough or do you suggest something else?

>Since nurbs are rectangular patches, it makes sense to work with 4 sides."

You're right but I’m trying to achieve something more flexible, if possible, and 2 curves to adjust is better then 4. Especially in this case, where I’m working with surfaces quite regular, and not really organic. Using just one section it means, for me, having a constant section in all the points of my rail. And having just one rail should guarantee that my section will not stench or change. This is just my opinion, but i already had this kind of trouble with other programs.... Edited by: david m on Jan 12, 2010 5:06 PM

[ravenzep]

To be honest, i dont understand why u are having trouble at such an apparently easy thing.
Try this: Have one curve....dont offset it, instead, duplicate it and move it along normal, or however u find it more adequate.
Measure it. Dont u get what u want now?

Spounds weird to me u cant achieve that....Did u check the curve combs?....is it a really good quality curve?
Feel free to msn me... no probs.......ravenzep@hotmail.com

Regards

[JohnDixon8251]

Fortunately the solution is simple, unfortunately, you may not like it. You need to increase the spans on your rail curve. For the amount of curvature you have, I would recommend around 10 spans to get the offset accuracy where you want it. You can try different numbers of spans to get the exact tolerance you are after. For construction, I would recommend your attempt number 2, using monorail construction. Skin is fine if you don't need any section to the surface. Another option would be to use a birail, 1 gen curve and 2 rails, but since the offset curve would have history, you could adjust only one curve, and the other would follow along. Of course the only problem with adding the spans is that you need to move a lot more points to make adjustments.

Another thing to consider is how tight you need the tolerance over such a long skinny object. Even though you show a variance of about a bit less then 2%, which seems like a lot, if that is 150 units wide it must be about 5000 long, so the width changing subtly 3 units over a 5000 unit length is not so bad or noticeable but it depends what it is for.

Good luck.

[ravenzep]

Well.... Bad answer..... In case you dont know the mathematics, let me tell you that even in a single curve, introducing a span is introducing a discontinuity. Yeah, the edit point , in a single curve is a discontinuity point. So the solution, following your line of thought, would be more like DIVIDING the curve, into more individual curves, all of them being 1 SPAN, obviously.

NOBODY wants spans! SPANS are enemy.
Amen

P.S. A curve with 10 spans??????....was that a joke dude?.....LEGO is your friend.

Edited by: ravenzep on Jan 28, 2010 2:51 PM

Edited by: ravenzep on Jan 28, 2010 2:52 PM Edited by: ravenzep on Jan 28, 2010 2:54 PM

[JohnDixon8251]

Well, now I know why no one is on here answering questions.

It is not a bad answer, it is the correct answer for the question. To achieve the tolerance he is looking for will require more information. Whether that comes as more spans in a single curve or more curves is irrelevant. Although I would not personally put 10 spans on that curve, for Alias to achieve what he has requested (controlling the resultant surface by manipulating just one curve with very tight tolerance) would require more spans. 10 may be overkill as that will get the tolerance down into the thousandths, but that amount of precision is available if that is what he needs. As mentioned in my post, I question the need for the accuracy he is looking for over such a long skinny surface, but I don't know what he is using it for. Just adding 1 or 2 spans would increase the accuracy quite a bit.

Also, if you create a smooth curve, you can increase the spans and the curvature will remain smooth. An edit point is a potential discontinuity point.


Have a nice day.

[davidemin]

Ok I found method n°4….
It is easy, you just need 1 rail curve and then you can make a draft surface, with 90° angle to have the smaller edges perpendicular to their rail curves (don’t pick the cross manipulator but type 90° in the draft tool options).
The deviation is more acceptable than the other solutions I mention before and I think it’s a good idea to use this draft tool whenever you need to offset a curve. In fact, and I still don’t understand why, the offset tool is just too approximate (see image 2 attached in the next post), that’s why I suggest to use the draft edge created as an offset of your curve.
>Have one curve....dont offset it, instead, duplicate it and move it along normal, or however u find it more adequate.
Ravenzep, I’m sorry but the operation you suggest is quite different than making a real offset.
JohnDixon8251: it is right, increasing numbers of spans make my offset more precise, as you can see in the attached image.

[davidemin]

In this image you can see what i mean when I say the draft tool produce a more accurate offset (the edge of the draft surface) than the offcial offest tools...but why??

[JohnDixon8251]

That is an interesting result.

The draft tools better tolerance comes from the fact that it added isoparms to the surface where as the offset curve did not add any spans.

You can get tighter tolerance from the offset tool. Offset in the default setting, keeps the curve construction the same as your original curve. This is nice as it does not add a bunch of spans and makes the curves nicely related. If you go in the option box, you can change the setting for "Max. Span Factor" from 1 to something bigger. This allows offset to add more spans based on the number you input. The only problem with this is the added spans will build up quickly and the resultant curve will not have it's points where you may have liked them to be.