Hi Marco,
not sure if this is the result your looking for but this was produced by using half of the helix (outer diameter of ring).
I extruding it to max depth then trimmed as Michael as shown.
I then shelled it and filleted.
It was necessary to trim the solid where it was going to join in the middle as this was not flat due to the shelling.
I then copied the half revolved it 180 and joined.
Hope this helps.
Barry

Barry!!!
That is exactly what I expected to achieve!!!!!!

Thanks a lot

Barry,

I follow Michael advises and yours, but... how did you solve the issue in the point where it has to be join?
Mhhh...
In all this job, a profile is missing I think!

Thanks a lot!
Marco

Hi Marco,
did you use the shell command (see photo's).
There is no need for a profile.
Make sure to do the filleting on just the half as I had problem trying to fillet after joining.
Cheers
Barry

Yes,

exactly Barry... that face is not squared and it's rotated, so when I copy-rotate the solid, the result is what I've attahced to my post before...
I can't figure out how to solve...
I followed the same yours steps!

Thanks
Marco

Hi Marco,
it looks like you have extruded the half helix square to the angle of the helix and it should be square to the axis.
If this isn't the case please send me your file.
Cheers
Barry

Ok understood...
Thanks.
What about the fillets, now?
I've cutted and reconstructed the surfaces to avoid crossing points and then launched the fillet... but you can see the results... any idea on how to apply correctly it to the model?

Thanks a lot.
Marco

Post your file before fillets!

Normallly you must have the same fillet along all the volume (don't accept the fillet if you want the following possibility)
then free to you after to make any new "variable fillets" during the same session of fillet on some edges!

Pilou... or maybe Frenchy? ;-)
The bottom small fillet is ok... The top one, instead, is just a suface that moi has used as filletting surface, so its another surface! Just delete it and you have the original solid. I left it just to make you know where is the issue.
Here attached you can find the solid without fillet

Thanks.
Marco

Hi Marco,
not sure what your problem with filleting is ?
Do all large fillets first then the small ones.
Attached your test with fillets.
Cheers
Barry

Ps: just found a problem with your large fillet where you have tapered it from front to back you have 3 edges created by the fillets converging at the front so when trying to add a fillet on the front edge no go.
I selected the 3 edges of the top ran 0.7 fillet and second fillet points for a 0.4 fillet at the back and that is what I have posted.
Are you trying to get the side fillets to match the front fillet where they meet at the front ?

Pilou is sufficient! ;)

The file that you send has yet many fillets!
As it's not so speed to erase them will be fine if you have the original without any fillets :)

Remember that you can make very tiny fillets on some places if needing!

Ok Pilou!
Thanks.
This evening I'll see the video... it seems very complete!
So do you think it's fillet-able?

Ok, let me post the original one with no fillets!

Thanks

Hi Marco, so some of the fillet problems are due to low surface quality in the end pieces:

For example here it's pretty subtle but I noticed a wobbly shape here:



The surface control points kind of fold back on themselves here:



This surface does not look like one that would result from a fillet, are you probably making these by using the Network command? If so then that's the problem, you would want to instead use the fillet command to make these big rounded areas too, with a larger fillet radius.

Also this one along the side is problematic:



It's not good for a NURBS surface to have 2 of its natural edges meet up smoothly like that, it makes the surface compressed and messy there. It is easy for the surface normal to wobble around in a small area with stuff like that. Instead things like that should be formed from a larger surface that has distinct corners and then it's trimmed rather than the surface directly constructed to an irregular boundary like that.

Putting in Network patches like that will also result in your surfaces not meeting up smoothly to each other which is then a lot more difficult to fillet. Like here for example I don't think your model is smooth:



It's a lot more difficult for a fillet to be constructed across a sharp edge like that especially if it is somewhat close to being smooth. Fillet surfaces only naturally meet up with with each other end-to-end when the surfaces that are being filleted are smooth to each other. When they have a sharp corner the fillets do not naturally meet up instead they have to be extended and intersected with each other and when things meet at a shallow angle that is nearly smooth it means the extensions have more like a zone of intersection rather than a crisp intersection.

Some other areas that kind of show a "patch by patch" fill in construction method are stuff like this:



What I would expect for that surface is for it to be a longer piece that is trimmed instead of the surface being directly contained inside these end edges.

I'd guess that maybe the rounded end was sort of "tacked on" ? Instead what you want is to have the entire base model extend further and then use filleting to make those large rounds instead of trying to manually grow them on to the end.

- Michael

Hi Marco, so you'd want to do it more like this - starting with the main object which is already extended as far as you need (so no "tacking on" needed), do the fillets in order of largest to smallest, so starting with this one:





Then next these:





Then select the whole model and do the smallest radius:



Don't try to simulate a fillet by filling in areas with Network, just in general you want to be working more by carving away pieces rather than filling in pieces patch by patch.

- Michael

Hi Michael,

thanks for the explanation and... you are right. I used network and sweep to simulate fillets just because, I suppose, I'm not skilled enough in filleting :-(
I use solidworks daily ant it makes fillets of all king in a click, no matter of the complexity and geometry errors... It automatically adapts fillet radius at the model: it's pretty different from moi, so I've to learn more about filleting!
So this night I've dealt with it a bit to undestand... no solved very much, but I've learnt some new stuffs.

Thanks for your effort. Let me study your last posts.
Marco

Method shown of Michael "successive fillets" is cool but "variable fillets" will be more tricky
because you will can change easily any one(s) on the same race untill all is not like you want! ;)

In the successive method you can't because you can't (for the moment) kill a fillet for change it(s)!
So you must all remake! :)

Hi Pilou,

re:
> In the successive method you can't because you can't (for the moment) kill a fillet for change it(s)!
> So you must all remake! :)

There is a very simple solution for that, just save off a version of your model before you put the fillets on. If you decide you don't like one of the previous fillet steps then just load in your previous model and do new fillets on it. It isn't necessary to rebuild everything then.

In the future though I do want to have "edge sets" in the fillet command in addition to "point sets", so you could set different fillet radius values on different edges and adjust them before leaving the fillet command.

- Michael

Hi Marco, so the part I don't quite understand is why you decided to make this part with a sweep or network instead of a fillet:



Was it something having to do with the side pieces? :



I mean something like wanting the side pieces to connect up with the larger round edge-to-edge?

The thing is you can't really make a good quality surface that is constructed directly to a boundary like that. To have a good shape that isn't pinched or skewed weirdly it needs to be a surface that is extended and trimmed like how a fillet is made.

If you put in a bunch of isocurves on that surface you can get a better sense of how that surface is "stressed":



It's trying to transition between 2 different orientations:



While at the same time it's trying to slide directly between these different length sides which is where you'll get a type of skew problem:



That's too many things it's trying to do all at the same time so it's going to have bad shaping. Like in this spot here it's going to have stuff like bumps in it but in a very small area so you would have to zoom in a lot to see them:



That's what can happen pretty easily with a "fill in areas patch by patch" method. Even though yes there is a surface that fills in that area if it's an irregular boundary your surface can be a lot messier than what it first appears. To relieve the stress the surface would need to be extended in a more natural way something more like this:



and then it would have a trimming boundary on it to cut it (created by intersecting it with the other neighboring surfaces) while the "underlying surface" remains with that regular shaping. Hope that helps explain some why that is not good.

So the thing is filleting will do all that for you so that's why you would want to use fillets to do all those areas instead.

- Michael

<< There is a very simple solution for that, just save off a version of your model before you put the fillets on. If you decide you don't like one of the previous fillet steps then just load in your previous model and do new fillets on it. It isn't necessary to rebuild everything then.

Yep but not so practical than the variable fillets when you have tones of "segments" so means save a version of each segments fillets project yet defined if you want change this and not this one... ;)

Hi Pilou,

> Yep but not so practical than the variable fillets when you have tones of "segments" so means
> save a version of each segments fillets project yet defined if you want change this and not
> this one... ;)

From what I could tell it does not seem like this model would have tons of different radius zones like that or need to use variable radius filleting at all.

- Michael