G2 Fillets

[vide]

Hello there,


I'm using the style feature ( isdx) for quite some time now and I think I'm getting the hang on it now. But there are some things that needs more practice. I'm having problems to make small fillets with a G2 curvature on it.


Now I'm producing them by trimming the boundaries of the surfaces that needs to be connected with a fillet. Herefor I'm using COS or projected curves. When the surfaces are trimmed I'm producing planar curves between the surfaces with Curvature Continuity to the surfaces. Now I have four boundaries to create the fillet.


This method is working very good when creating a fillet between 2 surfaces. But when there are more surfaces adjacent to each other to create a fillet between. The continiuity is harder to create. Especially the trimming curves which run over multiple surfaces. Now here is the question:


Is there an easier way to make the trimming curves and is it possible to create G2 fillets outside the Style feature?


The picture below is making it a littlebit clearer:


red: surfaces


blue: trimming curves (COS)


Yellow: Fillet curves ( 2 curve points Surface Curvature)


View attachment 1565


Thanks in advance,


Vide

 

[blockandahole]

Hi Vide,

COS accross multiple quilt boundaries is a known limitation and I'm sure will be addressed soon.

The best workaround is to use the (little known) option of creating a
conventional curve thro' points but setting the 'attributes' option to
on surface.

Not as flexible as in ISDX, but still works.

Also, for these kind of long thin surfaces, it is preferable to enter quite a number of cross curves to ensure a smooth flow.

Alternatively, you can create the trims and boundary surface in
conventional surfacing. This gives more control with control points etc.

Good luck.

 

[proejunkie]

I have the same issues with isdx, sometimes I just resort back to the standard method as stated. Can you make the curve with style on a plane and project it to the surfaces?

 

[blockandahole]

You can project curves, but this may not be desireable depending on how you want the fillet to flow.


Another nice technique is to sweep a circle along the intersection of the two major surfaces. If you use the variable section option and drive the diameter of the circle from a graph feature, you can get a really nice result. Effectively you trim the two major surfaces using the swept form and build the C2 fillet accross the resulting boundaries.

 

[ecirwin]

You can create a normal round outside ISDX using a conic section and change the conic value to some value like .6 or maybe .7. This will not be truly G2, but it will be pretty close.

While creating the round, pick "Sets" and then there is a drop down where you can pick "Circular", "Conic", or "D1 X D2 Conic". Then you can pick the conic value.

Here is a sample of a simple cube with two conic radii (conic value .6) on it and the Reflection Analysis of the surfaces. This reflection anlaysis shows a fairly curvature continuous surface. If the color variations flow smoothly across boundaries, it is curvature continuous. If the color variations simply match up, but are discontinuous at boundaries, it is merely tangent. If neither is the case, then the surfaces just intersect without being continuous (G2) or tangent (G1), they merely intersect (G0).

If you try the same thing with generic circular rounds and then look at the reflection analysis, you will easily see the difference.

Hope this helps,
Eric

View attachment 1580


Edited by: ecirwin

 

[vide]

Hello there,


Sorry for the late reaction but I was gone for a few days. But first thank you all for the comments and tips. But the biggest relief is that I'm not the only one who had some problems with it ( thought my knowledge was not high enough). The problem with the fillet from above was that I constructed two single fillets and then created the fillet section in the corner. But the transitionbetween these two surfaces to the corner couldn't be made Curvature Continous and the arrow pointed in the wrong direction. The problem was in the COS on surface two. The endpoint was curvature to the COS on surface 1. When I switched the connection between these curves and enlarged the curvature (arrow) the problem was solved. Now it has a smooth transition but I'm still not happy with the COS construction. So I will try your tips and of course "'practice makes perfect''!


Thanks again

 

[mgnt8]

ecirwin - what am I missing here regarding your solution:by "conic value" are you referring to the Conic Parameterwhich I believe is the Rho?When Imodify this number, only the sharpness changes not the relation to adjacent surfaces. The smallest I can go is .050which flattens the round and gives me what looks like a chamfer. Curvaturecondition is unaffected - can you explain further?

 

[ecirwin]

mgnt8,

How are you checking the continuity of the surfaces? A conic round may not give perfect continuity, but it will be very close. At least it will be closer than the standard Rolling Ball circular round.

Yes, I would expect a conic value of .050 to give you something that looks like a chamfer. A conic value of 0.6 or 0.7 will be smoother, but as the value approaches 1.0 the curvature dissappears.

Yes, the conic value is Rho, but during the creation of a round it never says anything about setting the Rho value. In the sketcher when you create a conic curve, it shows up as the Rho.

HTH,
Eric

 

[mgnt8]

I see it now - thanks - you are right about the reflection analysis - this is about as c2 as anybody can expect.